Carbopalladation of bromoene-alkynylsilanes: Mechanistic insights and synthesis of fused-ring bicyclic silanes and phenols

The palladium-catalyzed cascade cyclization of silylated bromoenynes and alkenylstannanes provides a straightforward route to a range of bicyclic silylated cyclohexadienes. Mechanistic insights into aspects of carbopalladation and unusual palladium-mediated isomerizations have been obtained through the detection of reaction intermediates, the isolation of byproducts, and reaction monitoring by VT NMR spectroscopy. The utility of the bicyclic products is illustrated through oxidation to bicyclic enones and phenols.


Introduction
Phenols represent one of the most important aromatic functional groups, featuring in numerous natural products and drugs. 1 Polycyclic phenols are of particular interest, as illustrated by highly bioactive compounds such as the anthracycline antibiotics, estrone, and morphine ( Figure 1). 2 As traditional methods for phenol preparation require harsh conditions that may not be compatible with sensitive functional groups, 3 phenols are usually incorporated at an early stage of a synthesis, thus restricting synthetic planning. Late-stage phenolation is therefore an attractive prospect that has driven the development of transition metal-catalyzed aryl halide hydroxylation and aryl C-H activation / oxidation methods. 4 An alternative that avoids the prolonged heating often required by these processes is the use of a phenol surrogate, which could be revealed at a late stage of a synthetic route, but is stable to intermediate transformations. Here, arylboron derivatives have met with some success, particularly due to Molander's elegant work on the oxidation of aryl trifluoroborate salts, 5 although the synthetic processability of these remains to be proven. In contrast, organosilanesstalwarts of the protecting group field -display unrivalled tolerance towards multistep synthesis, and in recent work we have reported the use of arylsilanes as a source of phenols through 7 We have also disclosed a palladium-catalyzed cascade cyclization which prepares bi-or tricyclic products (1, Scheme 1) from the coupling of bromoenynes 2 with alkenyl-and dienylstannanes 3, 8 a reaction pioneered and impressively explored by Suffert et al. for the synthesis of various polycycles. 9 A feature of this work is the requirement for an internal alkyne (i.e. R 1 ≠ H, Scheme 1) to avoid the formation of undesired triene isomers 4, which cannot undergo electrocyclization. One solution is to employ trimethylsilyl alkynes, 10 and although this functionality indeed enables the desired cascade, it was clear that the resultant TMS-substituted cyclohexadienes would be of limited synthetic utility.
We realized that the use of a more functional silane would lead to a silicon-containing product capable of undergoing a range of further transformations, such as the aforementioned Tamao oxidation or Hiyama coupling. 11  illustrated this principle in approaches to the CDE rings of rubriflordilactone A, where arylsilanes formed from the cyclizations of bromoendiynes could be oxidized to give specific tricyclic phenols. 8b, 12 In this paper we wish to report methods for the synthesis of a range of silylated bromoenynes, and their intermolecular coupling with alkenylstannanes to provide bicyclic silylcyclohexadienes 5. The oxidation of these dienes and silane substituents leads to bicyclic arylsilanes (6) and phenols (7) that would be challenging to prepare by other routes, and offers a de novo approach to ring systems of this type. 13 Alongside this synthetic work, we expand on the mechanistic pathways operative in the cascade cyclization, including two unusual palladium-mediated isomerization processes. The unexpected observation of an unprecedented formal 4-endo-trig cyclization to afford fused ring cyclobutenes is also described.

Synthesis of Cascade Cyclization Substrates
We selected a range of linkers to connect the bromoalkene and silylalkyne, such that bicyclic products containing 5-to 7membered rings, and heterocycles, could be obtained. Synthesis of carbon-tethered bromoenynes began with dimethylmalonate (Scheme 2), which following monoallylation with 2,3dibromopropene (70%) was alkylated with alkynyl electrophiles 8a-c to give bromoenynes 9a-c (69-90%). The silylation of these alkynes was initially attempted using conditions employed in our arylsilane work. 6a Thus, deprotonation of 9a with n-butyllithium followed by trapping of the resultant alkynyllithium species with the moisture-sensitive (diethylamino)chlorodimethylsilane (Et 2 NMe 2 SiCl) gave an intermediate aminosilane which was not isolated, but immediately converted to the dimethylisopropoxysilane 10a (Route A, Scheme 2). 14 Perhaps unsurprisingly, this silane proved to be rather unstable towards chromatography, however through rapid chromatographic purification using a solvent system buffered with triethylamine, 10a could be isolated in 75% yield. In order to address the chromatographic instability of this dimethylisopropoxysilane, we employed the analogous Scheme 2. Synthesis of malonate--derived silylbromoenynes. diethylsilane reagent (Et 2 N)SiEt 2 SiCl, 15 which gave the alkynyl isopropoxydiethylsilanes 11a-c in much improved yields following chromatographic purification (Route B). An alternative solution to the synthesis of 10a proved to be the use of dimethylchlorosilane (HMe 2 SiCl) as the acetylide trapping agent (Route C); the ensuing hydrolytically stable intermediate alkynylhydrosilane could be easily purified, and then smoothly oxidized to 10a using Lee and Chang's elegant rutheniumcatalyzed dehydrogenative silylation methodology, 16 which improved the yield of 10a to 93%.
The syntheses of heteroatom-tethered silylbromoenynes could be achieved uneventfully using equivalent chemistry, with sulfonamide-tethered enyne 12 and ether-linked enyne 13 being prepared in reasonable yields over two steps from known sulfonamide 14 and propargyl alcohol, respectively (Scheme 3).

Cascade Cyclizations of Silylated Bromoenynes
With a selection of silylated alkynes in hand, their palladiumcatalyzed cascade cyclizations were examined. These reactions were carried out according to our previously optimized conditions, 8a using PdCl 2 (PPh 3 ) 2 as precatalyst in refluxing  toluene (Table 1). For the purposes of this work, the propenyl and styrenyl stannanes 3a and 3b were selected as generic cross-coupling partners. The reaction of dimethylsilane 10a was first investigated, which in spite of the relative instability of the alkynylsilane cyclized in excellent yields with both stannanes (92-97%, Entries 1, 2). Notably, the hydrolytic stability of the silane increases markedly upon cyclization owing to the increased steric bulk of the cyclohexadienyl substituent in 15a compared to the alkyne substitutent of the starting material. The cyclizations of the (more robust) diethylsilane substrates proved more challenging (Entries 3-5), with these reactions requiring longer reaction times, likely due to steric hindrance of the key transmetallation step following carbopalladation. This resulted in lower isolated yields, as exemplified by the malonate and sulfonamide tethered enynes 11a and 12 which gave the corresponding bicycles 15c and 15d in 77 and 63% yields respectively. The oxygen-tethered substrate 13 proved particularly troublesome, delivering the dihydrofuran product 15e in low yield, possibly due to instability of the allyl propargyl ether. Reaction of substrates featuring a longer tether also proved effective, giving the 6,6-and 7,6-bicyclic products 15f and 15g in good yields (Entries 6, 7). In the latter two cases, the desired bicyclic products were contaminated with significant quantities of partially separable isomeric byproducts 16a,b and 17a,b (Figure 2), which were identified by careful analysis of 1 H-1 H COSY, and 1 H-13 C HSQC and HMBC 2D NMR spectra; 17 the assignment of one of these sets of byproducts (16a and 17a) is discussed here. Byproduct 16a contains an exo-methylene unit (δ H 5.01 and 4.85 ppm), and two vicinal alkene protons (δ H 6.00 ppm, dd, J = 15. 5,1.5 Hz;and 5.32 ppm,dq,J = 15.5 and 6.5 Hz), the latter being characteristic of the trans-alkene of the propenyl unit. These data, together with key HMBC correlations, are strongly suggestive of the formal anti-carbopalladation / crosscoupling product 16a, the formation of which is entirely consistent with our earlier results. 8a The identification of the second byproduct proved less straightforward. Key signals in the 1 H NMR spectrum at δ H 6.36 (d, J = 10.6 Hz), 6.31 (ddd, J = 16.5, 10.6 and 9.7 Hz), 5.26 (dd, J = 16.5 and 2.0 Hz) and 5.25 ppm (dd, J = 9.8 and 2.0 Hz) revealed a connectivity between four alkene protons and thus the surprising presence of a 1,3-butadienyl unit. A further significant piece of evidence was the detection of a methyl singlet at 1.50 ppm, to which the proton at 6.33 ppm showed an nOe enhancement. These combined observations, and 1 H-13 C correlations, led us to propose structure 17a, which features an (E)-alkenylsilane as part of a conjugated triene.
A mechanistic hypothesis for the formation of these byproducts is depicted in Scheme 4. The formation of byproduct 16 is consistent with our earlier observation 8a that such undesired anti-trienes form in the course of the cyclization reaction, potentially via isomerization of the intermediate dienylpalladium complex 18 to its isomer 19. 18 The rate of transmetallation of complex 19 is likely to exceed that of 18, so even small amounts of 19 may lead to significant quantities of anti-triene 16 (i.e., a Curtin-Hammett situation). The formation of increased quantities of this triene for substrates 11b and 11c may reflect the increased flexibility of the larger tethering ring, which can distort to alleviate 1,3-allylic strain between the silyl substituent and exo-methylene in 19 -and therefore reduce the steric cost of placing a silane in this more hindered position. The formation of 17 may be explained by our second previous observation 8a that anti-trienes can isomerize to bicyclic products on prolonged exposure to the reaction conditions. We  suggest that this process may proceed via oxidative addition of Pd(0) with diene 16 to give a palladacyclopentene 20. 19 This could undergo a 1,3-allylic migration of the palladium atom to the 7-membered palladacycle 21, reductive elimination from which would afford the cyclohexadiene product 15. 20 The formation of byproduct 17 could be rationalised by β-hydride elimination from this common palladacycle intermediate 21, followed by reductive elimination of the resultant palladium(II) hydride species. We presume that the steric hindrance imposed by the methyl group in 17 prevents a 6π-electrocyclization of this compound. As [4+1] oxidative additions of Pd(0) to dienes are rare, 19 more concrete evidence to support these pathways was sought. Firstly, the formation of 16 via the intermediacy of dienylpalladium complex 19 was probed through the exposure of alkynylsilane 22 to one equivalent of Pd(PPh 3 ) 4 in d 8 -toluene at 110 °C ( Figure 2). The characteristic methylene signals of 22 at δ H 5.41 and 5.71 ppm were rapidly converted (10 min) to two new sets of peaks: a prominent (apparent) singlet at 4.60 ppm, and two smaller singlets at 4.58 and 4.78 ppm. These were tentatively assigned as the exo-methylene peaks of the syn-and anti-dienylpalladium complexes 23 and 24 respectively, based on analysis by COSY and HSQC experiments. Support for this assignment was gained through the addition of 1.5 equivalents of tributylvinyltin to the NMR tube; further heating for 10 minutes led to exclusive formation of cyclohexadiene 25 and triene 26 in a ratio mirroring that of these intermediate species (84 : 16).
To explore the isomerization of 26 to 25, a 1:0.32 mixture of these compounds, formed from reaction of silane 22 with vinyltributyltin for two hours, was purified by silica gel chromatography (Scheme 5); the triene 26 in this mixture would be expected to isomerize to give the 25 upon resubmission to the reaction conditions. Heating the mixture in toluene overnight at 110 °C in the absence of catalyst led to no conversion of 26 to the bicyclic product 25, with both compounds recovered unchanged after this period (thus ruling out a purely thermal process). Palladium(II) salts have previously been shown to promote alkene isomerisation, 21 but subjection of the anti-triene to PdCl 2 (PPh 3 ) 2 at 110 °C also led to no reaction. However, heating the mixture in the presence of Pd(PPh 3 ) 4 for 20 h led to complete consumption of anti-triene, and by performing this isomerization in d 8 -toluene with monitoring by 1 H NMR spectroscopy in the presence of an internal standard (1,4-dimethoxybenzene), a smooth conversion of 26 to 25 was observed ( Figure 3). This clearly demonstrates that the isomerization of anti-triene to product is not a thermal process, and in fact requires a Pd(0) catalyst, thus offering some support to our proposed mechanism. At no point do we detect the formation of syn-triene, which lends some weight to our mechanistic hypothesis for the direct conversion of anti-trienes to bicyclic products (Scheme 4, although we recognise that any syn-triene formed could electrocyclize rapidly).

Cyclization to 7,4-fused ring cyclobutenes
In the course of cyclization reactions to form seven-membered rings (including 11c), we had noticed the occasional formation of a different byproduct to those discussed thus far, the production of which seemed highly dependent on the reaction concentration, and quantity of stannane coupling partner. Although this byproduct was not observed under our optimized conditions, the use of <1.5 equivalents of stannane, or more dilute reaction conditions (i.e. such that transmetallation would be slowed) increased its formation. In fact, we had first  observed such a species in the attempted 8π-electrocyclic coupling of bromoenyne 27 with stannane 28 (Figure 4), which resulted in a surprising degree of apparent protodestannylation of 28 (leading to the known diene 29). 22 We assigned the product formed in this reaction as the 7,4-fused cyclobutene 30 based on detailed analysis by 2D NMR experiments. Specifically, a complete set of HMBC correlations (Figure 4) was observed between protons H1, H3 and H7, and carbons C2, C8 and C9, together with long-range coupling between H1 and H7 ( 5 J = 2.7 Hz; this coupling was also observed in a 1 H-1 H COSY spectrum). The formation of this product is not unreasonable if potential mechanisms for its formation are considered (Scheme 6). Following carbopalladation (31), one possibility would involve a 4-endo-trig carbopalladation (Path A), leading to cyclobutene 32 -a pathway that might be favoured, in spite of ring strain, due to the formation of an allylpalladium intermediate. β-Hydride elimination would afford the observed product 30, and liberate a palladium(II) hydride species that could be reduced to palladium(0) by transmetallation with 28, thus leading to the 'protodestannylated' product 29. However, due to a lack of precedent for this mode of carbopalladation, we also consider an electrophilic palladation route feasible (Path B), in which attack by the exo-methylene on the proximal palladium(II) atom leads to palladacycle 33 -which may again be rendered feasible by the formation of an allyl cation in this process. Now, loss of a proton generates palladacyclopentene 35, reductive elimination from which leads to 30. 23 The resemblance of intermediate 35 to those proposed in enyne cycloisomerization processes is clear; 24 the proton lost from this patheway could then effect protodestannylation of the coupling partner 28 to afford 29.
Whatever mechanism is operational, it was clear to us that this process overall corresponds to a Heck reaction in which regeneration of the palladium(0) catalyst is mediated by the stannane reagent. This suggested that an amine base might perform a similar role, and to our delight, the use of common Heck conditions (toluene, triethylamine) at just 5 mol% catalyst loading indeed led to a high-yielding cyclization of 27 to 30 (90%, Scheme 7). The more functional bromoenyne 11c was also tested in this chemistry, which gave the corresponding 7,4fused silylcyclobutene 36 in excellent yield (86%). This efficient process offers an alternative entry to this type of fused cyclobutene ring system. 25

Oxidations of Silylcyclohexadienes
To illustrate the potential utility of this chemistry, we subjected the product dienylsilanes to a selection of oxidative transformations. Firstly, a direct Tamao oxidation of the dienylsilane was carried out, 26 which we hoped would deliver a bicyclic enone. In the case of the isopropoxydimethylsilane 15a, this met with some success (Scheme 8), delivering 37 in moderate yield (43%). The successful formation of this product, which lacks any olefinic protons, could be confirmed by 13 C NMR (δ C 194.9,158.7,134.8 ppm for the enone region) and IR spectroscopy (ν max 1667 cm -1 ). Attempted oxidation of the equivalent diethylisopropoxysilanes proved unsuccessful, potentially due to competing nucleophilic epoxidation of the product enone, which highlights a reactivity benefit of the lesshindered dimethylalkoxysilane group.
The silylcyclohexadiene frameworks could also be readily oxidized to the corresponding arylsilanes using manganese Scheme 8. Bicyclic enone formation dioxide, conditions that we had successfully employed in other work 27 and which proved superior to the use of other oxidants such as DDQ. The resultant arylsilanes could generally be isolated in excellent yield; two examples are shown in Scheme 9 (38: 88%; 39: 93%). These arylsilanes show potential for a range of transformations -but here, in keeping with our interests in the synthesis of phenols from arylsilanes, 6 we chose to investigate Tamao oxidation. Under Tamao conditions (TBAF, KHCO 3 , H 2 O 2 , 60 °C), 28 good yields of the corresponding phenols 40 and 41 were obtained, thus validating this approach to the synthesis of bi/polycyclic phenols. Scheme 9. Oxidation of the silylcyclohexadiene framework.

Conclusions
In summary, we have developed efficient routes for the synthesis of functional alkynylsilanes and demonstrated their application in bicyclization cascade reactions. The resultant silylated bicyclic cyclohexadienes are substrates for oxidation to bicyclic enones, arylsilanes, and phenols; the latter process thus affords bicyclic phenols from acyclic precursors in just three steps. Investigations into the mechanism of the cascade cyclization have revealed some unusual palladium-mediated isomerization pathways. Finally, 7,4-fused cyclobutene ring systems, arising from cyclization of a 7-membered exocyclic dienylpalladium complex, could be formed under standard 'Heck' type conditions. Together, these processes underline the rich reactivity -both expected and unexpected -that can be harvested from carbopalladation chemistry.

Experimental
General: Reagents were used as purchased, or purfied by standard laboratory techniques. Reaction solvents were purified using an alumina column drying system, and reactions were performed under inert atmosphere unless otherwise stated.
The residue was purified by flash chromatography to afford the silylated bromoenyne.