Antarafacial Mediation of Oxygen Delivery By a Phenylsulfinyl Group in the Epoxidation of Proximal Double Bonds: Intramolecular Trapping of an Early Pummerer Intermediate with Stereoelectronic Control

Abstract

Stereospecific intramolecular antarafacial epoxidation of a double bond via an early Pummerer reaction intermediate has been demonstrated. The intermediate is presumably generated via trifluoroacetylation of a sulfoxide precursor. Ionization of trifluoroacetate would generate, formally, a di-positive “sulfenium” equivalent. This specie attacks an otherwise unactivated, proximal olefinic linkage in an anti-periplanar fashion, with trifluoroacetate serving as the nucleophile. Proposed mechanistic intermediates have been characterized structurally (in several cases by crystallographic means) and shown to serve as precursors en route to the final antarafacial epoxides. The sense of the cyclization seems to be driven by principles inherent in Markovnikov’s Rule.

Recently, we described a simple two-step sequence for generating what we have termed “trans-Diels-Alder” (trans-DA) motifs, containing angular functionality.¹ Of course, many issues remain to be addressed before this capability can be fully exploited. We have already come upon a remarkable observation in the course of an attempted Pummerer rearrangement² of sulfoxide 2 (derived from “trans-DA” product, 1, Scheme 1). Treatment of 2 with trifluoroacetic anhydride (TFAA)³ led, not unexpectedly, to disappearance of the starting material. However, subsequent treatment with aqueous sodium bicarbonate failed to produce the anticipated aldehyde 3. Instead, there was obtained a compound in which the erstwhile angular sulfoxide function in 2 had given way to a sulfide. Correspondingly, the double bond had become an epoxide. It is clear that the oxygen of the oxido linkage had not arisen by direct transfer from the sulfoxide since the epoxide (see compound 4)⁴ had emerged at the α-face of the double bond, i.e. anti to the β-face phenylsulfinyl precursor. In this paper, we establish an interesting mechanistic pathway to account for the formation of 4. Follow-up studies served to generalize the scope of this novel intramolecular, yet antarafacial, sulfoxide–induced overall oxidation of an otherwise unactivated double bond, and to identify its limitations.

Scheme 1.

To initiate our study, we began by trying to understand the conversion of 2→4 at a more rigorous level. Compound 2 was subjected to the action of TFAA in dichloromethane from 0 °C to room temperature over one hour. The resultant product was treated with aqueous sodium bicarbonate (in a two phase setting). Compound 4 was produced in 94% yield. In another experiment, following the TFAA step, but before workup with aqueous sodium bicarbonate, evaporation of the solvent gave rise to a semi-solid residue. However, exchange of trifluoroacetate by tetrafluoroborate⁵ gave rise to a new salt, which fortunately could be coaxed into a state of crystallinity. X-ray analysis of this new salt revealed it to be structure 9. As shown, the double bond had apparently been attacked, in an electrophilic sense, by the TFAA-activated sulfoxide, giving rise to the bridged sulfonium trifluoroacetate 8 and, subsequently, to the corresponding sulfenium tetrafluoroborate 9. Moreover, the formal carbenium-like entity at C₈ in 7, had been discharged by trifluoroacetate. The precise timing of these steps, as to level of concertedness, is not established from our data set. However, since the covalent trifluoroacetate in 9 is trans-periplanar with respect to the “C-S⁺” bond, the two bond formations (C₇-S⁺ and C₈-O) may well correspond to orchestrated trans-diaxial attacks upon the resident C₇-C₈ olefin in precursor 5.

Treatment of 9 with aqueous sodium bicarbonate indeed afforded epoxide 4. In summary, then, the activated sulfonium salt⁶ had attacked C₇ of the double bond, giving rise to a stable but unprecedented bicyclic [2.2.2] sulfonium-containing substructure with neutralization by trifluoroacetate having occurred at C₈. Treatment of this salt with sodium bicarbonate apparently accomplishes de-acylation of the covalent trifluoroacetate at C₈, leading to formation of the oxido linkage with concurrent release of the neutral thiophenyl function as shown in 4.^6,7,8

Of course, the di-positive “sulfenium” (early Pummerer⁹) specie formalized as 6, arising from heterolysis of the presumed trifluoroacetoxy sulfenium specie (of intermediate 5), could, in principle, have attacked the proximal double bond at either C₇ or C₈ (pre-steroid numbering), generating, in either case, tertiary carbenium ion character at the alternate carbon. That the reaction produces apparently only the bicyclo [2.2.2] “sulfonium” substructure, as in 7, may reflect either kinetic or thermodynamic preferences as to the optimal size of the sulfonium-containing ring system and the optimal point for neutralization of carbenium ion character by the weakly nucleophilic trifluoroacetate counterion.

Given these results and uncertainties, it was of interest to examine this novel type of epoxidation sequence, in the context of sulfoxide 11, obtained from sulfide 10 (Scheme 2). Here, too, we were asking several questions. First, at this stage there was an unaddressed issue as to whether the less substituted olefin would even be sufficiently nucleophilic to interdict what must be a highly reactive formal “di-positive” sulfenium specie 12 before it progresses to the normal Pummerer rearrangement. Moreover, in the case at hand, the sense of trapping by cyclization raised the question as to whether trapping would occur, as above, at C₇, which would generate secondary carbenium ion character at C₈. Alternatively, cyclization might now occur at C₈, thus delivering tertiary carbenium ion character at C₇.

Scheme 2.

In the event, reaction was conducted as before. TFAA treatment left a sulfonium trifluoroacetate intermediate (now known to be 13), which was, as above, converted to the crystalline tetrafluoroborate. The structure of this compound was shown by X-ray crystallographic analysis to be 14. Apparently, the formal “di-positive” sulfenium equivalent arising from heterolysis of 12 (in sharp contrast to 5) attacks the proximal C₇–C₈ double bond in a trans-diaxial Markovnikov¹⁰ sense, with trifluoroacetate acting as the nucleophile at C₇. Treatment of 12 with aqueous sodium bicarbonate afforded epoxide 15 (70% overall yield from 11). Thus, in the case of trisubstituted olefin 11, cyclization, either kinetically or thermodynamically, leads to the bridged [3.2.1] Markovnikov substructure shown in 14, in contrast to the bridged [2.2.2] sulfonium-containing substructure of 9 en route to 4.

We studied the scope and limitations of this stereospecific overall, “anti-orchestrated” delivery of oxygen to the proximal double bond. Entry 1 shows that the reaction is applicable even to a disubstituted double bond, albeit in sharply diminished yield. Entry 2 shows the reaction to be operative, even with a particularly congested tetrasubstituted double bond. Entry 3 demonstrates the capacity to incorporate functionality in each of the rings of the Diels-Alder-derived bicycle. Entry 4 shows the extendability of the epoxidation to the trans-hydrindanoid series, while entry 5 demonstrates its applicability to the corresponding cis junction, albeit in somewhat diminished yield.¹¹ Entry 6 shows the reaction sequence to also be applicable to a more conformationally flexible monocyclic system.

It will be noted that the reaction conditions employed in Schemes 1 and 2, as well as Table 1 do not include pyridine, which could be used in effecting Pummerer rearrangement.¹² Interestingly, under Pummerer rearrangement conditions where pyridine was added, aldehyde products were observed, along with the epoxides. In some instances the Pummerer-derived aldehydes appear to be the more predominant products, but the formation of mixtures of epoxides and aldehydes compromises the value of such reactions from the standpoint of synthesis.

Table 1.

entry	sulfoxide	product	Temp.	yield
1			−78°C→rtb	35%
2			−0°C→rta	59%
3			−0°C→rta	86%
4			−0°C→rta	93%
5			−78°C→rtb	40%
6			−0°C→rta	77%

^aGeneral procedure: The mixture of sulfoxide (0.1mmol) and TFAA (0.3mmol) in CH₂Cl₂ was stirred at 0°C (0.5h) and then warmed to rt (0.5h). Aqueous NaHCO₃ (sat.) was then added and the biphasic mixture was stirred for 4h prior to standard workup.

^bThe reaction was run at −78°C and warmed to rt for up to 12 hours prior to addition of aq. NaHCO₃.

We also studied the possible applications of this reaction to olefins in acyclic contexts. Accordingly, the reactions of sulfoxides 28 and 31 were examined as substrates. In the case of 28, the reaction did, indeed, produce compound 29, bearing the the thiophenyl and oxido linkages. However, the yield was only 22%. The major product of the reaction was 30, containing the allylic trifluoroacetate as well as the phenylthio functions. Since we do not know the precise structure of the intermediate sulfonium species arising from cyclization, we cannot rigorously assert the mechanism of formation of either 29 or 30 in detail. However, the formation of 30 as the major product can clearly be accommodated in various obvious ways via the chemistry described above.

Attempts to apply the reaction to sulfoxide 31 led, at best, to very low levels of epoxide 32. This compound was produced in less than 5% yield. While its presence could be inferred from analysis of the NMR spectrum of the crude reaction, the presumed 32 could not be obtained in homogeneous form. Hence, the reaction seems to have broken down when the resident olefin is housed in a setting of an acyclic terminal vinyl group, which is presumably a less activated intramolecular nucleophilic trapping agent.

Finally, we were able to demonstrate that the chemistry described above could be used to generate an angular aldehyde function. For this purpose, we returned to epoxide 4. Treatment of this compound with m-CPBA gave rise to sulfoxide 33. Exposure of 33 to standard Pummerer conditions¹³ gave, as shown, the hitherto unknown compound 34 containing keto, aldehydo and oxido functions.

In summary, we have discovered, admittedly with considerable happenstance, an unexpected but interesting and potentially valuable line of chemistry, arising from the ability of a properly placed olefinic linkage to interdict the course of the normal Pummerer reaction at an early stage. Aside from the novel structural chemistry associated with the now characterized complex sulfonium salts, 9 and 14, the reaction raises interesting issues of mechanism, which were addressed. In addition to its potential value in building diversity libraries, this chemistry raises possibilities for application toward the synthesis of complex target systems of natural origin. Such studies are, in fact, underway

Scheme 3.