An Oxy-anion Accelerated [1,5]-o-Quinone Methide Shift During the Nucleophilic Epoxidation of Salicylfulvene

Abstract

A new facet of nucleophilic fulvene epoxidations has been uncovered. 6-Arylfulvenes containing an ortho or para hydroxyl group react with basic hydrogen peroxide in an unusual manner; the epoxidation of the fulvene exocyclic double bond is followed by a phenoxide ion initiated epoxide ring opening to form an o-quinone methide (o-QM) intermediate. The resulting cyclopentadienolate undergoes an unusual oxy-anion accelerated [1,5]-sigmatropic o-QM shift. Computational studies reveal that the activation energy for the [1,5]-QM-shift in the cyclopentadienolate intermediate is quite low, signifying the acceleration caused by the oxy-anion group. Placement of a second hydroxyl group in the 6-aryl ring at C5 epoxidation via electron donation to the o-QM carbon; instead, an intramolecular oxa-6-π-electrocyclization of the o-QM intermediate onto the cyclopentadiene is observed.

Graphical Abstract

Nucleophilic epoxidation of o- or p-hydroxy substituted fulvenes proceed with epoxide ring opening giving o- or p-quinone methide (QM) intermediates. These intermediates undergo an unusual oxy-anion accelerated [1.5]-QM shift followed by intramolecular Michael addition and subsequent nucleophilic epoxidation. The [1.5]-shift is computed to be exothermic by over 28 kcal/mol and has a barrier of 3.2 kcal/mol.

The polarization of the π-electrons in the exocyclic double bond toward the cyclopentadiene ring in fulvenes imparts aromatic character to the molecule.¹ As a result, C6 in fulvenes is quite electrophilic and susceptible to nucleophilic attack. Hard nucleophiles such as Grignard and organolithium reagents,² hydride reductions,³ nucleophilic cyclopropanations with sulfur ylids,⁴ and in particular nucleophilic epoxidations⁵ are well documented. A survey of literature in this area reveals that except for the parent fulvene,⁶ nucleophilic epoxidations have been confined to 6,6-disubtsituted derivatives. In none of the cases studied, the monomeric 1-oxaspiro[2.4]hepta-4,6-dienes were isolated since they are too reactive at room temperature and undergo Diels-Alder dimerization, as was first reported by Alder et al.⁵ We were intrigued by the fact that the latter epoxidation has not previously been applied to (or reported on) 6-monosubstituted fulvenes. Thus, our first objective was to apply Alder’s epoxidation conditions to 6-p-tolylfulvene. The reaction appeared to be capricious and highly particular to the reactions conditions. We applied several different nucleophilic epoxidation conditions to our fulvenes since these oxidative transformations turned out to be quite capricious and highly particular to the nature of base and temperature. After attempts using H₂O₂/t-BuNH₂ at room temperature,⁷ t-BuO₂H/KOtBu in THF at −10 °C,⁸ t-BuO₂H/Triton-B in THF at room temperature,⁹ only the original Alder procedure with KOH/H₂O₂ in MeOH at −5 °C⁵ temperature (3h) was successful to give the epoxide dimer 2 in 38% yield along with the fulvene-cyclopentadiene Diels-Alder cycloadduct 3¹⁰ isolated as a minor product (7%). Compound 3 is probably formed under the alkaline conditions by a retro-condensation of the starting fulvene since a small amount of p-tolualdeyde was also isolated (Scheme 1). Cycloadduct 3 was independently prepared by stirring a 1:1 mixture of cyclopentadiene and p-tolylyfulvene at 0 °C for 3 hours. Under similar epoxidation conditions, fulvenes derived from benzaldehyde and o-anisaldehyde gave rise to intractable mixtures.

Scheme 1.

Nucleophilic epoxidation of a 6-arylfulvene at different temperatures

After having shown that 6-monosubstituted fulvenes are more difficult to epoxidize with alkaline H₂O₂ than the 6,6-disubstituted counterparts, we decided to explore the effect of o- and p-hydroxy groups on the epoxidation process. The impetus for this study was provided by our previous report on the unusual hydroxyl effect on fulvene endoperoxide decompositions.¹¹ It was expected that a similar hydroxyl effect might alter the course of fulvene epoxidations, in particular, the 6-(2-and 4-hydroxyphenyl) substituted derivatives. We are pleased to disclose here our results from oxidative transformation reactions of salicylfulvene and analogs.

Under slightly different conditions (0-15 °C, 48h) as used for 1, fulvene 4,¹² derived from salicylaldehyde, gave a major product¹³ that did not exhibit the olefinic protons expected from a Diels-Alder dimer of the corresponding epoxide in the ¹H NMR spectrum and appeared to be monomeric. The HRMS data confirmed a structure with the exact mass of 202 and chemical formula of C₁₂H₁₀O₃, indicating that two additional oxygen atoms had formally been added to the starting fulvene. Based on NMR spectroscopic analysis, assisted by 2D techniques, the unusual structure of 5 was assigned to this product (Scheme 2).

Scheme 2.

Results from attempted nucleophilic epoxidation of 6-salicylfulvene

At first glance, it is apparent that a major skeletal rearrangement must have taken place during the oxidation. A reasonable mechanism would commence with the nucleophilic epoxidation of the exocyclic double bond in 4 (Scheme 3).

Scheme 3.

Mechanism for the tandem epoxidation-oxyanion accelerated 1,5-o-QM shift

Then the phenoxide ion causes epoxide ring opening to form an o-quinone methide (o-QM) intermediate. The resulting cyclopentadienolate undergoes an oxy-anion accelerated 1,5 shift (ordinarily, [1,5]-alkyl shifts in 5,5-dialkylcyclopentadienes occur at temperatures 300 °C and above¹⁴) of the o-QM group with concomitant intramolecular Michael addition to the methide carbon in 8, resulting in cyclopropane formation, yielding 9. The enone unit in 9 subsequently undergoes a conventional nucleophilic epoxidation¹⁵ with the hydroperoxide ion to give 5 as the final product. The anti-stereochemistry of the cyclopropane and epoxide rings in 5 and the other products (see Table 1) are assigned based on the results from nucleophilic epoxidations of bicyclo[3.1.0]hex-3-ones similar to 9.¹⁶

Table 1.

Reactions of 6-(hydroxypenyl)-substituted fulvenes with alkaline H₂O₂

^[a]Fulvene-cyclopentadiene Diels_Alder side-products of the type 6, stemming from retrofulvenation not shown

^[b]Isolated yields, not optimized.

To confirm the low barrier to the unusual oxy-anion accelerated 1,5-shift, the transition state was located at the M06-2x/6-311+G** level.^{17, 18} The reaction is computed to be exothermic by over 28 kcal/mol and has a barrier of 3.2 kcal/mol.

It passes through an early transition state (Figure 1) with a very acute angle on the transferring carbon (~49°). As expected, it is an early transition state with the forming C-C bond (1.92 Å) being much longer than the breaking one (1.67 Å) (Figure 1). To the best of our knowledge, the only other oxy-anion accelerated [1,5]-alkyl or vinyl shift in a similar system was observed on 2,3,4,5-tetraphenylcyclopenta-2,4-dienolates.¹⁹ Other anionic oxy-Cope rearrangements and related sigmatropic shifts are well known.²⁰

Figure 1.

Transition state structure of the oxy-anion accelerated [1,5]-shift in 8

To test whether a similar mechanism with a potential p-quinone methide could be realized, the corresponding fulvene from p-hydroxybenzaldehyde was first synthesized. Fulvene 10 was subjected to the same reaction conditions as 4. In this case the corresponding tricyclic epoxide 13 was isolated, obviously by a similar p-quinone methide 1,5-shift in 12 (Scheme 4).

Scheme 4.

Tandem epoxidation-1,5-p-QM shift

The fulvene derived from vanillin, 15, likewise gave the corresponding cyclopropanated compound 16. The somewhat lower yields in these reactions than with 6,6-disbstituted fulvenes²¹ can be ascribed to the fact that in the less unsuccessful attempts mentioned earlier there were considerable amounts of unreacted starting material in the crude mixture, some cyclopentadiene-fulvene cycloadducts, as well as dicyclopentadiene, along with polymeric material. The fulvene cleavage (retrofulvenation²²) is a major problem in these types of reactions that are carried out in the presence of strong base. An additional problem may arise with 6-monosubstituted fulvenes: the first intermediate in the nucleophilic epoxidation after ⁻O₂H attack at C6 is a secondary hydroperoxide that is prone to a base-catalyzed dehydration (Kornblum-DeLaMare reaction²³), competing with the epoxide formation. The resulting acylcyclopentadienes are exceptionally unstable compounds and tend to polymerize.²⁴

It is noteworthy that fulvene 16, derived from 2,5-dihydroxybenzaldehyde, did not unergo the expected tandem epoxidation-rearrangement-epoxidation, instead, the cyclopenta[b]chromene²⁵ derivative 17 was isolated in 19% yield whose structure was readily assigned based on NMR spectroscopy and HRMS data (Table 1, entry 4).

We figured that the strongly basic epoxidation conditions were responsible for the facile oxa-6π-electrocyclization of the intermediate o-quinone methide indermediate (Scheme 6).²⁶ At first, it is difficult to account for the lack of reactivity of 16 toward the hydroperoxide ion, and the base-promoted retrofulvenation was the major pathway in this case along with a small amount of 17. The mechanism leading to 17 must involve the intermediacy of an o-quinone methide intermediate that undergoes an oxa-6π-electrocyclization. ²⁷

Scheme 6.

Alternative mechanism via conjugate hydroperoxide attack at the o-QM carbon for the epoxidation of fulvene 4

Two questions arose in view of the result mentioned above: a) why does a similar oxa-6π-electrocyclization not occur in the case of 4?; b) why is 16 not reactive toward HOO⁻ in a nucleophilic epoxidation? The answer to the first question is presumably the fact that the o-quione methide that would form initially would be the more stable E-isomer which cannot undergo the electrocyclization. In regards to the second question, if indeed an o-quinone methide forms initially, before the attack by the hydroperoxide ion, then the methide carbon would be rendered electron-rich and less prone to attack by HOO⁻ due to the p-benzoquinone formation (19a) by the aryloxide ion in the para position (Scheme 5).

Scheme 5.

6π-Electrocyclization of the o-quinone methide intermediate 19b to 17

This consideration also lends some credence to an alternative epoxidation mechanism in which conjugate hydroperoxide attack at the methide carbon of the o-QM (or p-QM) intermediate might be responsible for the epoxidation. Intermediate 19a also accounts for the isomerization of 18 to 19b, the configuration requisite for the electrocyclization. The conversion of 16 to 17 was also accomplished in the absence of H₂O₂ with KOH/MeOH or t-BuOK/DMSO, but best results were obtained with pyrrolidine/AcOH (2:1) which helped avoid retrofulvenation observed otherwise with stronger base (Scheme 5).

If our hypothesis presented above is correct, then replacing the p-OH group in 17 by a methoxy group should impede 19a formation and E→Z isomerization, and the conjugate attack at the o-QM carbon should result in the epoxidation, and eventually leading to the oxy-anion accelerated 1,5-o-QM shift. Gratifyingly, 22 underwent the tandem epoxidation-1,5-shift to give 23 (Table 1, entry 5) with no traces of the possible electrocyclization product.

Based on the results for 16 and 22, it appears likely that the nucleophilic epoxidation involves conjugate attack of the hydroperoxide ion at the methide carbon, rather than the fulvene C6. Thus, the alternative epoxidation mechanism involving conjugate attack at the methide carbon of the o-QM in the case of 4 (p-QM for 11) that also accounts for the lack of reactivity of 16 toward nucleophilic epoxidation is shown in Scheme 6. Moreover, under the basic conditions 16 was subjected to, 22 did not undergo a similar oxa-6π-electrocyclization, supporting the mechanism outlined in Scheme 5.

In conclusion, we have uncovered a novel facet of Alder’s nucleophilic fulvene epoxidations, namely, the effect of an ortho or para hydroxyl group at the 6-aryl position in the fulvene. Under basic conditions, abstraction of the phenolic proton precedes hydroperoxide attack at the electrophilic fulvene C6, leading to an o-QM intermediate that- by way of a conjugate HO₂⁻ attack- gives rise to the hydroperoxy intermediate before the cyclopentadienide ion displaces the hydroxide ion to yield the initial epoxide. The phenoxide ion in the ortho or para position then causes epoxide ring opening via a QM formation (no Diels-Alder product from the epoxide intermediate 7 was observed), before the facile oxyanion-accelerated 1,5-QM shift occurs. The resulting enolate intramolecularly attacks the QM carbon resulting in a cyclopropanation. The cyclopent-2-enone then undergoes another nucleophilic epoxidation with the hydroperoxide ion to give the 3,4-epoxy-2-oxobicyclo[3.1.0.]hex-3-ene derivative. Studies are under way to explore the effects of groups other than hydroxyl on the nucleophilic epoxidations of fulvenes, as well as trapping a fulvene epoxide monomer via a cycloaddition.

Experimental Section

1. General procedure for the nucleophilic epoxidation of 4 and related fulvenes.

The mixture of salicylaldehyde fulvene (1) (1.71 g, 10.0 mmol) and potassium hydroxide (0.625 g, 3,7 mmol, 1 eq) was stirred with methanol (25 mL) at 0°C. Hydrogen peroxide (10.02mL, 30%, 10 eq) was added dropwise. The mixture was stirred at 15°C for 48 h.. The mixture was poured with water (20 mL) and extracted with diethyl ether (3 × 20 mL), washed with brine (20 mL), dried (MgSO₄), and filtered. Evaporation of the residue gave the 0.22 g (29%) of 7-(2-hydroxyphenyl)-3-oxa-tricyclo[4.1.0.0^2,4]heptan-5-one (5). The product was purified by flash chromatography on silica gel eluting with hexane-ethyl acetate (3:1).

2. Base-promoted conversion of 16 to 17.

The mixture of 2-(cyclopenta-2,4-dien-1-ylidenemethyl)benzene-1,4-diol (16) (0.484 g, 2.6 mmol) and acetic acid (0.149 mL, 1 eq) was stirred with methanol (20 mL) at 0°C. Pyrrolidine (0.427 mL, 2 eq) in methanol (5 mL) was added dropwise. Molecular sieve (3 A, 5.00 g) was added. The mixture was stirred at 15°C for 72 h. The mixture was neutralized with acetic acid (0.149 mL, 1 eq) and filtered, then poured with water (20 mL), extracted with diethyl ether (3 × 20 mL), washed with brine (20 mL), dried (MgSO₄), and filtered. Evaporation of the solvent and flash chromatography of the residue gave the compound 17 (0.92 g, 19%).