Karan
Goyal
a,
Garrett A.
Kukier
b,
Xiangyang
Chen
b,
Aneta
Turlik
b,
K. N.
Houk
*b and
Richmond
Sarpong
*a
aDepartment of Chemistry, University of California, Berkeley, CA 94720, USA. E-mail: rsarpong@berkeley.edu
bDepartment of Chemistry and Biochemistry, University of California, Los Angeles, CA 90095, USA. E-mail: houk@chem.ucla.edu
First published on 4th October 2023
A novel synthesis of aryl-substituted, enantioenriched fulvenes from an oxidative Heck cascade and rearrangement of a carboxy-substituted spiro[4.4]nonatriene is disclosed. Mechanistic investigations with density functional theory (DFT) calculations and empirical results support the net transformation occurring through a novel Pd(II)-mediated 1,5-vinyl shift from a vinyl–palladium intermediate that terminates with protodepalladation. This spiro-to-fused bicycle conversion tolerates a range of electron-rich and deficient arylboronic acids to give a range of mono- and diaryl substituted annulated fulvenes in moderate to good yields and enantiomeric ratios. Overall, this work connects two classes of molecules with a rich history in physical organic chemistry.
Fulvenes, such as the pentafulvenes, are another class of molecules that possess unique π orbital systems (Fig. 2B, left).10 The arrangement of p-orbitals in these systems confers some aromatic character, as well as versatile reactivity—both of which can be affected by substituent changes.11 Unlike spiro[4.4]nonenes, however, pentafulvenes have found much broader application in synthesis. These structural motifs are not only featured in polymers12 and dyes13,14 due to their tunable electronic properties, but are also found in some natural products such as the illudin family (e.g., see acylfulvene and irofulven in Fig. 2B, middle).15 Furthermore, the reactivity of fulvenes with nucleophiles and with other π systems as diverse cycloaddition partners has made them synthetically attractive (Fig. 2B, right), as demonstrated by Carreira and coworkers in the total synthesis of the pallambin natural products.16
A subset of fulvenes, known as annulated fulvenes, were first synthesized by Paquette and coworkers in 1985 and used as intermediates toward conformationally restricted methylenenorbornadiene derivatives (Fig. 2C).17 While the Paquette approach yielded the desired molecules, it proceeded in low yield. Specifically, treatment of a gem-dibromocyclopropane with four equivalents of an organolithium reagent resulted in a Skattebøl rearrangement18 followed by nucleophilic attack to give the substituted annulated fulvenes in low yields, alongside indene and ring-opened aryl alkyne side products. Notably, this synthesis only allowed for substitution at a single position and afforded racemic products.
Herein, we report the development of a method which results in the synthesis of enantioenriched, aryl-substituted annulated fulvenes in moderate to good yields and enantiomeric ratios (er's) through an intercepted oxidative Heck reaction of a carboxy-substituted spiro[4.4]nonatriene (Fig. 2A). This discovery points to an intriguing mechanism in which a C–C cleavage occurs at low temperature and under weakly oxidative conditions. Recently, Zhu and coworkers demonstrated a cleavage of unstrained C–C bonds in related five-membered ring systems through an oxidative Heck reaction followed by a dyotropic shift (Fig. 2D).19 We reasoned that a similar pathway might be operative in this spirononatriene system. Our mechanistic investigations using a combination of DFT and empirical insights have led to a mechanistic hypothesis involving C–C cleavage from a key vinyl palladium species. The net coupling/rearrangement transformation of 1 to 2a (Fig. 2A) occurs with a range of commercially available arylboronic acids. Overall, this work expands the chemistry of spirononenes by showcasing their transformation to annulated fulvenes and provides evidence for a novel mechanism of C–C cleavage and migration—a Pd(II)-mediated 1,5 vinyl shift.
Entry | Deviation | Resultsa |
---|---|---|
a Yields were determined by 1H NMR integration using pyrazine as an internal standard; isolated yields are shown in parentheses. | ||
1 | None | 76% (59%) 2a, 87:13 er |
2 | 5% AcOH (v/v) instead of base | 88% (46%) 2a, 84:16 er |
3 | No base | 65% (53%) 2a, 84:16 er |
4 | KOAc instead of Na2CO3 | 43% 2a, 44% 4 |
5 | Pd(OAc)2 instead of Pd(TFA)2; no base | 45% 2a, 34% 4 |
6 | 5% AcOH (v/v), no base; 0.4 equiv. Ag2O | 28% 2a, 24% 4, 21% 1 |
7 | BQ instead of 2,5-dimethyl BQ | 49% 2a, 39% 4 |
8 | 1 equiv. PhB(OH)2 | 46% 2a, 43% 1 |
9 | 0.1 M instead of 0.01 M | 47% 2a, 12% 4 |
While the reaction was shown to work well under neutral, basic, or acidic conditions (Table 1, entries 1–3), basic conditions proved most suitable for a range of electron-rich and -deficient arylboronic acids (see ESI, Table S6†). Protodeboronation became more pronounced under more acidic conditions.21 Additionally, the presence of acetate, either from the added base or in the form of Pd(OAc)2, favored formation of Heck product 4 (Table 1, entries 4 and 5). The use of silver salts to promote decarboxylation22 and favor fulvene product 2a had no such effect and led to lower overall conversion and nearly equal amounts of the two products (Table 1, compare entries 2 and 6). The choice of oxidant was found to be crucial with 2,5-dimethyl-para-benzoquinone outperforming p-benzoquinone (Table 1, entry 7). The use of two equivalents of boronic acid was found to be necessary to obtain high yields. Using only one equivalent of the boronic acid coupling partner still led primarily to products that had incorporated two of the boronic acid units, albeit in low overall conversion (Table 1, entry 8). Finally, the reaction yield was highest at a concentration of 0.01 M (Table 1, entry 9).
Arylboronic acids with extended aromatic systems also gave fulvene products such as 2o–2q in moderate yields, albeit with diminished enantiomeric ratios. Finally, heterocyclic arylboronic acids performed poorly as coupling partners under the optimal conditions that we have identified. Nonetheless, 3-furanyl-boronic acid gave a 15% yield isolated yield of fulvene product 2n. While low-yielding, this result suggests the potential for optimization of this method for heteroarylboronic acid coupling partners. Vinyl boronic acids and arylboronic pinacol esters did not participate in the coupling/rearrangement transformation, presumably due to inefficient transmetalation to the organo-palladium intermediate as well as competing decomposition of the vinyl boronic acid coupling partner (see ESI, Table S7†).23,24
Scheme 2 Proposed pathways for C–C cleavage: (A) 1,5-vinyl shift from 4 (B) β-carbon elimination from alkyl palladium species Int1 (C) dyotropic shift from vinyl palladium species 7. |
A second proposed pathway involves a β-carbon elimination from the intermediate alkyl palladium species generated after migratory insertion into the spirononatriene olefin (see Int1, Scheme 2B). β-Carbon eliminations in relatively unstrained systems of this type are rare due to the higher kinetic barriers and thermodynamic considerations.27 However, Lautens and coworkers have reported a Heck cascade in which a β-carbon elimination is driven by the relief of strain arising from bulky ortho-aryl substituents.28 We hypothesized that in our system, analogous relief of developing strain between the newly introduced phenyl group and bulky substituent on the ligand may serve as a driving force (see Int1, Scheme 2B). Additionally, enthalpic gain through formation of a chelate with the pendant carboxylic acid could further facilitate this rearrangement to 6 (Scheme 2B).
Finally, a dyotropic shift pathway could be operative. Here, concerted migration of two bonds across a stationary scaffold could occur from vinyl palladium species 7 (Scheme 2C).29 In this case, such a rearrangement would furnish the observed 5,6-fused ring system of the annulated fulvene. The dyotropic shifts that have been invoked for transition metal intermediates generally involve high oxidation state metal centers such as putative Cu(III) or Pd(IV) intermediates.18,30 In a related report, a possible dyotropic rearrangement in a Heck reaction with Pd(II) is proposed, albeit without sufficient supporting data.31 Under our reaction conditions, accessing a Pd(IV) intermediate appears unlikely and so a Pd(II)-mediated dyotropic shift was considered.
Entry | Conditions | Results |
---|---|---|
a Fulvene products (2a and 3a) do not readily form from Heck product (4). | ||
1 | 5% AcOH/DCE, 50 °C | No reaction |
2 | 5% AcOH/DCE, 50 °C, Pd2(dba)3/(S)-tBu-quinox | No reaction |
3 | 5% AcOH/DCE, 50 °C, Pd(TFA)2/(S)-tBu-quinox | No reaction |
4 | Optimized reaction conditions | Only trace 2a/3a, largely SM |
These empirical observations are supported by DFT calculations. For all calculations involving Pd, alkyl palladium species Int1 was used as the starting point (i.e., ΔG = 0 kcal mol−1). The energies of all cationic structures include the energy of a trifluoroacetate (TFA) anion at an infinite distance from the Pd-complex. Pathway A was initially considered with coordination of palladium (Fig. 3A). In this scenario, 9, which may be accessed through vinyl shift of the Pd-bound alkene (Int2), was determined to be uphill by 15.9 kcal mol−1 (ΔG) and has a transition state (TS) energy, ΔG‡, of 44.8 kcal mol−1 (see the ESI for the structure of TS1). This value is unlikely for a transformation that occurs at 50 °C (approximate maximum of 28 kcal mol−1 accessible).33 In considering Pathway A without the involvement of Pd(II), 4 was used as the thermodynamic starting point (Fig. 3B). Overall, while this transformation was found to be more kinetically and thermodynamically accessible, it was still unlikely to occur under the reaction conditions that had been employed. This reaction was endergonic by 1.0 kcal mol−1 with an activation barrier of 29.4 kcal mol−1 (Fig. 3B, see the ESI† for the structure of TS2). Altogether, these computational and experimental results do not support Pathway A as a viable reaction path for the formation of the annulated fulvene products from the precursor spirononatriene.
Next, the β-carbon elimination pathway (Scheme 2B) was investigated using DFT calculations. A transition state for β-carbon elimination was found with a TS energy of 38.7 kcal mol−1 (see ESI, Scheme S1†). The reaction itself was also found to be endergonic, giving a product that lies 11.6 kcal mol−1 higher. On the basis of these initial results, we concluded that the β-carbon elimination pathway was unlikely and additional mechanistic studies and calculations to support or refute this possibility were not pursued.
Finally, Pathway C (Scheme 2C), which involves a dyotropic shift, was investigated computationally. Following an extensive search, a dyotropic shift transition state was not found. However, searches for this transition state converged to one where the C–C bond migrates but the C–Pd bond remains stationary, constituting a 1,5-vinyl shift from vinyl–Pd species Int5 (Fig. 3C). Since transition state optimizations of geometries constituting a dyotropic rearrangement converged to this Pd(II)-mediated 1,5-vinyl shift transition state, we conclude that this pathway must be lower in energy than the proposed dyotropic rearrangement. Initial searches furnished a 1,5-vinyl shift transition state involving a cationic Pd(II) intermediate. It was then found that coordination of TFA reduces the energy of this transition state, resulting in a sufficiently low transition state energy of 25.5 kcal mol−1 and an exergonic reaction which places Int6 at −5.6 kcal mol−1 (Fig. 3C). On the basis of these calculations, the 1,5-vinyl shift pathway from vinyl palladium species Int5 was then deemed as a plausible pathway and the energies of intermediates leading to Int5 were calculated (Scheme 3).
The immediate step following formation of alkyl Pd-species Int1 (Scheme 3) is expected to be a β-hydride elimination to give the spirononatriene-bound Pd–H species Int2 at 10.0 kcal mol−1.
Subsequently, a TFA anion coordinates to the cationic Pd(II)-complex to give a neutral trigonal bipyramidal complex (Int3 at 11.9 kcal mol−1). In Int3, the Pd–H is positioned such that the hydride is proximal to the pendant carboxylic acid. At this stage, we propose a deprotonation-palladation sequence which results in the loss of H2 and the formation of a palladated carboxylate. Pd(II)-hydrides have been reported to be basic enough to deprotonate phenols.34 As such, we posit that the deprotonation of a more acidic carboxylic acid should be possible under these conditions. Calculations show this transformation from Int3 to Int4 to be thermodynamically feasible with the palladated carboxylate at 14.2 kcal mol−1. Furthermore, it has been shown in the context of organolithium systems that complex-induced proximity effects (CIPE) can be leveraged for selective deprotonation.35 Due to the observation that 4 does not form any rearranged products when subjected to the reaction conditions (Table 2, entry 4), we hypothesize that CIPE may be operative in this intramolecular deprotonation to form Int4, and that Int4 does not form from intermolecular palladation of 4. This direct palladation of 4 is likely impeded by steric hindrance between the newly installed aryl group and bulky palladium complex.
From Int4, we propose that a Pd(II)-mediated decarboxylation,36–38 which is calculated to be exergonic by 12.3 kcal mol−1, furnishes an initial vinyl Pd species at 5.8 kcal mol−1 (see ESI, Int-S1).39 A slight conformational change to pre-reaction complex Int5 at 2.5 kcal mol−1 then leads to C–C cleavage through a kinetically feasible and exergonic Pd(II)-mediated 1,5-vinyl shift (with a TS energy of 25.5 kcal mol−1) to give cyclopentadienyl Pd-complex Int6 at −5.6 kcal mol−1. The η-1 bound palladium isomers Int7 and Int8 were calculated to have energies of −10.7 and −5.2 kcal mol−1, respectively (Scheme 3). At this stage, the catalytic cycle could terminate in protodemetalation from any of these intermediates (Int6–Int8). Direct protodemetalation from Int8, or protodemetalation from Int6 and Int7 followed by 1,5-hydride shifts, could furnish monoaryl fulvene 3a (Scheme 3, red box). A subsequent oxidative Heck reaction would then yield diaryl fulvene 2a.
Our hypotheses regarding protodemetalation and second oxidative Heck reaction were both investigated experimentally. The protodemetalation step was tested through a series of deuterium-labeling studies (Fig. 4). First, spirononatriene 1 was subjected to the standard reaction conditions in the absence of base with an added 5% (v/v) of deuterated acetic acid (AcOD). For reactions with both 4-NO2 and 4-CF3 phenylboronic acids, the mono- and diaryl fulvene products were obtained with partial deuterium incorporation at multiple positions (Fig. 4A). This outcome not only provides evidence for protodemetalation but the observation of deuteration at multiple sites provides additional mechanistic insight. Specifically, it is likely that alkyl palladium intermediates that result from the Pd(II)-mediated vinyl shift (e.g., Int6–Int8, Scheme 3) equilibrate prior to protodemetalation. Suprafacial 1,5-hydride shifts may ultimately give the fulvenyl π system. This rationale is in line with Semmelhack's proposal of rapid 1,5-hydride shifts following the vinyl shift in spiro[4.4]nonenes to the 5,6-fused bicycles under high temperature conditions.25 To support the proposal that deuteration results from protodemetallation en route to the fulvene, and not after fulvene formation, diaryl fulvene 2h was subjected to the conditions with added deuterated acetic acid (Fig. 4B). In this case, deuterium incorporation was not observed, supporting our initial hypothesis.
Because the standard reaction conditions use base, we sought to determine the source of the proton that terminates the transformation. We hypothesized that adventitious water might be the source of the proton. Addition of D2O to the reaction conditions did, indeed, lead to partial deuterium incorporation (Fig. 4C). Additionally, on the basis of the pKa's of aryl boronic acids (pKa = 4.2–9.0),40 which tend to be lower than the pKa of H2O (pKa = 14),41 the proton involved in protodemetalation might arise from the boronic acid coupling partner or a by-product of boronic acid transmetalation. These observations provide strong support for a terminating protodemetalation and 1,5-hydride shifts in the formation of the initial monoaryl fulvene product (3a). On the basis of these results, we hypothesize that 1,5-hydride shifts of the benzylic hydrogen in intermediates similar to 12 (Scheme 3) may also be responsible for the lower er's observed for diaryl fulvenes as compared to the corresponding monoaryl fulvenes (i.e.2hvs.3h).
Subjecting 4-NO2 monoaryl fulvene 3h to the reaction conditions cleanly gives diaryl fulvene 2h in 73% yield, consistent with the second aryl group arising through a second oxidative Heck reaction.42 Measuring the product distribution of 2hvs.3h over the course of the reaction provides further support for this second oxidative Heck reaction. An initial increase in monoaryl fulvene 3h is observed, followed by a decrease in 3h as a simultaneous increase in the amount of diarylfulvene 2h is observed (see the ESI, Section 3.5†). Overall, these experiments provide support for the hypothesis for conversion of alkyl palladium intermediates Int6–8 to fulvenes 2a and 3a (Scheme 3, red box).
Collectively, these findings support the mechanistic picture illustrated in Scheme 3, which is consistent with our empirical observations and calculations. Additional insight may be provided by some observations made during the process of reaction optimization. DFT calculations indicate that the key transition state (TS3) bearing acetate instead of trifluoroacetate as the X-type ligand is lower in energy (18.2 kcal mol−1, see ESI, Scheme S2†) despite the observation that addition of acetate results in reduction in the amount of fulvene product (2a) and greater amounts of the Heck product (4; see ESI, Table S8†). In this case, it may be that acetate competitively binds the Pd(II) center, displacing arylated spirononatriene 4 from either the Pd(II)-bound alkene (Int3) or palladated carboxylate (Int4). Silver salts might serve an analogous role due to their ability to bind carboxylates.43
Footnote |
† Electronic supplementary information (ESI) available. CCDC 2245692 (for 1), 2245684 (for 2a), and 2245677 (for 4). For ESI and crystallographic data in CIF or other electronic format see DOI: https://doi.org/10.1039/d3sc03222a |
This journal is © The Royal Society of Chemistry 2023 |