Trapping Reactions of Benzynes Initiated by Intramolecular Nucleophilic Addition of a Carbonyl Oxygen to the Electrophilic Aryne

Abstract

We describe here reactions in which a carbonyl oxygen atom initiates cascade reactions by nucleophilic attack on a covalently attached benzyne. The benzynes are produced by thermal cyclization of triynes via hexadehydro-Diels-Alder reaction. The initially produced oxocarbenium/aryl carbanionic zwitterion is protonated in situ by an external protic nucleophile (NuH) of appropriate acidity. The resulting ion pair (oxocarbenium⁺/Nu⁻) collapses through several different mechanistic manifolds, adding to the diversity of structure classes that can be generated.

Graphical Abstract

Because of their highly electrophilic character, arynes will engage even weakly nucleophilic species that typically would not be thought of as sufficiently reactive to add readily to more classical electrophiles.¹ Having seen hints of reaction between thermally generated benzynes [from a tethered 1,3-diyne/diynophile pair via a hexadehydro-Diels-Alder (HDDA) cycloisomerization reaction²] and ester functionality when the benzynes were produced in hot ethyl acetate, we designed a substrate in which an ester group was poised to intramolecularly trap the proximal benzyne carbon atom. Specifically and as outlined in Fig. 1a, we asked what chemistry might ensue when the acetate ester I was heated in the presence of various additives, the need for which was predicated on the expectation that the benzyne II would initially cyclize to the zwitterion III. This reactive zwitterion would need to be “charge quenched” by further reaction with an appropriate external capture agent. Protic nucleophiles (NuH) have served this role in other settings of HDDA trapping reactions.^3,4 When sufficiently acidic, they can protonate the carbanionic aryl carbon to produce a transient ion pair (cf. IV), which then was envisioned to collapse to a stable adduct (cf. V).⁵ Here we describe observations consistent with this line of thinking that result from our exploration of reactions of benzynes bearing suitably disposed carbonyl groups.

Figure 1.

(a) Hypothesis: a pendant carbonyl oxygen atom (red) traps the proximal benzyne carbon in II to produce zwitterion III, which, in the presence of an external protic nucleophile NuH, is protonated to give the ion pair IV that collapses to product V. (b) The first example.

Our first experiment proved fruitful (Fig. 1b). The triyne 1a was heated in chloroform solution and the tetracyclic fluorenone derivative 2a was the major product of the reaction. The presence of carbon resonances at 103.4 and 102.6 ppm in the ¹³C NMR spectrum of this compound, the latter with characteristically low intensity, were attributed to the C2 ketal carbon of a 1,3-dioxane and a CCl₃ group, respectively. Those, as well as the pair of AB doublets for the geminal methylene protons in the ¹H NMR spectrum, quickly revealed the structure that had arisen from this new type of benzyne trapping process. An analogous experiment performed in CDCl₃ clearly established that the hydrogen atom on the new arene ring, now a deuterium (97% deuteration), had originated from a molecule of the solvent.

Having seen that chloroform trapping⁶ of substrate 1a was a valid platform for the transformation hypothesized in Fig. 1a, we explored a variety of other carbonyl-containing functional group analogs of this acetate ester. The results are summarized in Fig. 2a. Perhaps unsurprising, the benzoate ester 1b smoothly gave 2b. More interestingly, an array of nitrogen-containing functional groups was also shown to be competent. Products arising from trapping by the carbonyl oxygen of pendant carbamate (2c and 2d), isatin (2e), N-acylcarbamate (2f), N-sulfonylamide (2g), and imide (2h and 2h’) groups in substrates were all formed as the major isolable material. In the last instance, the phenol 2h’ was also isolated, likely resulting from premature intervention by water.

Figure 2.

(a) Various carbonyl-containing functional groups can trap the intermediate HDDA-benzyne. (b) Benzynes derived from the HDDA substrates 3 and 5 are also compatible with the carbonyl-trapping event.

To demonstrate that the reaction is not unique to the benzophenone-like linker present in triyne substrates 1a-h, we examined the substrates 3 and 5, each again containing an acetate ester as the initial carbonyl capturing moiety (Fig. 2b). These reactions were carried out at 130 °C, given the known slower rate of the HDDA cycloisomerization for substrates i) that contain linkers such as that in 3^7a or ii) are triynes (cf. 5) rather than tetraynes.^7b The analogous adduct 4 or 6a was again observed as the major product, respectively. Use of bromoform as the reaction solvent gave that tribromomethyl analog 6b, albeit with reduced efficiency.

We next explored the behavior of a simple ketone carbonyl group in this transformation (Fig. 3). Both aryl (7a and 7b) and alkyl (7c) ketones were examined. In the first experiment, cyclization of 7a in chloroform alone produced two products, the expected CCl3-adduct 8a accompanied by the furanofluorenone 9a. The intermediate oxocarbenium ion was now able to undergo proton loss to produce the aromatic furan ring, a process not accessible to the dioxocarbenium ions III (X = heteroatom). When the same substrate was heated in a 1:1 mixture of CHCl₃/AcOH, only the furano-adduct 9a was isolated, suggesting either that AcO⁻ did not trap the carbenium ion or that, if so, the resulting acyloxy adduct was labile toward elimination of AcOH to proceed on to 9a. The triynes 7b and 7c underwent the same process, producing the benzofurans 9b and 9c, respectively.

Figure 3.

Ketone carbonyl groups participate enroute to CCl3 adducts 8a or, after intervening proton loss, to benzofurans 9a-c.

The results shown in Fig. 4 indicate that protic nucleophiles other than CHCl₃ will also participate in this chemistry. In the presence of malonate, chloromalonate, propiolate, p-hydroxybenzoate esters or maleimide, the acetoxy triyne substrate 1a gave rise to the trapped esters 10a-e, respectively (Fig 4b). These were the anticipated outcomes, given the expectations set by the chloroform trapping reactions. To our surprise, however, the malonate-related nucleophiles shown in Fig. 4c gave, instead, products of net ring-opening of the dioxocarbenium ion IIIa. This could be explained by the competitive events suggested by arrows ii vs. i in the intermediate IIIa in Fig. 4a. The origin of preference for these two competing events is not obvious, and it is curious that we did not identify any examples of a protic nucleophile that gave a mixture of products from both the ion pair ring-opening (ii) and collapse (i) pathways.

Figure 4.

(a) Competing events i vs. ii leading to products 10 vs. 11, depending upon the nature of the protic nucleophile NuH. (b) NuHs other than chloroform that trap at the carbenium ion carbon, producing products 10a-e. (c) Protic nucleophiles that result in ring opening events, producing products 11a–11c.

^aIn each of these cases, along with the major adduct incorporating the NuH, the dimer 16 (see Fig. 6 and discussion there) was isolated as a minor product (21% with 10c and 35% each with 11b and 11c).

We then explored the behavior of the series of β-dicarbonyl compounds 12, each a methyl ketone-containing protic nucleophile (Fig. 5). Yet another new reaction profile emerged. Namely, when 1a was cyclized in the presence acetylacetone (12a), ethyl acetoacetate (12b), or methyl acetopyruvate (12c), the diastereomeric mixture of hemiketals 13a-c was generated. Each of these could be easily dehydrated by treatment with a Brønsted acid to its respective 4H-chromene derivative 14a-c. Not surprisingly, entirely analogous reactivity was observed when the triyne 15 was heated with 12a or 12b.

Figure 5.

(a) 1,3-Dicarbonyl compounds involving a ketone make a new C–C bond in analogy to the nitriles (cf. Fig. 4c); the initial hemiketals can be dehydrated by treatment with CH₃SO₃H. (b) Rationale for conversion of hemiketal 13c to the pyruvate ester 14c.

The somewhat different behavior of the pyruvate series is notable. At the hemiketal stage, it was the α-ketoester carbonyl that was engaged, as shown in structure 13c (Fig. 5b). However, the dehydration occurred through the alternative, less stable, hemiketal VI (formed via the ring-opened diketone, not shown), likely a thermodynamic expression of the greater preference of the α-ketocarbonyl group to enolize (cf. VII).

Finally, we observed one additional new type of reaction product: the dimeric structure 16, a spirocyclic orthoester. This compound was first detected as a byproduct in three of the reactions described in Fig. 4 (see footnote to Fig. 4). Heating the triyne 1a alone in 1,2-dichloroethane (i.e., in the absence of any protic nucleophile) also led to the formation of this dimeric product (Fig. 6a), albeit always in modest yield. Although we do not have full mechanistic understanding to account for this unusual reaction (see Fig. 6b), it seems most reasonable to propose that it arises by encounter of the ketene acetal VIII, which is an isomer of substrate 1a, with the o-quinone methide⁸ IX. It is worth noting that this reactive intermediate could also be involved in the formation of the adducts 11a-c (Fig. 4c) and 13a-c (Fig. 5a), via conjugate addition, as an alternative to the direct ring-opening suggested by arrow ii in Fig. 4a.

Figure 6.

(a) Formation of mixed dimer 16 when 1a is heated in the absence of a NuH. (b) Mechanistic proposal for formation of the key ketene acetal VIII and quinone methide IX, guided by results using the trideuteroacetate 1a-d₃.

We considered that an o-quinone methide such as IX might arise by thermal ejection of ketene from VIII. However, a DFT computation [M06-2X/6-311+G(d,p), see SI] of that process (lower left, Fig. 6a), done using the conformationally simpler lactone analogs VIII’ and IX’, suggested that the barrier for that concerted cycloreversion is excessively high.

In an attempt to learn more about the mechanism of formation of 16, we prepared the trideuteroacetate analog 1a-d3 and examined the distribution of deuterium in the dimer product 16. It showed nearly equal levels of incorporation at C11 and C11’ (each 30–40%, see SI). Although only partial deuteration was observed, likely due to small amounts of H₂O in the reaction medium, the fact that both aromatic sites were nearly equal in their extent of deuteration implies that a common intermediate is involved in the reaction pathway. The mechanistic formulation outlined in Fig. 6b is consistent with that observation. More specifically, the deuterated intermediate benzyne II-d₃ proceeds to the dioxocarbenium ion III-d₃, which ring opens to the intermediate Xa when there is not a sufficiently reactive NuH (or enough of it) to protonate the carbanion in III. It is notable that rotamers Xa and Xb are examples of a (highly substituted) α,3-dehydrotolene, a relatively rare type of reactive intermediate.⁹ Rotamer Xb is poised to deliver a deuteron to the arene ring, producing the enolate zwitterion XIa and, subsequently, rotamers XIb and XIc. Any of the XI species could eject ketene, producing the quinone methide IX concomitantly. Rotamer XIb can collapse to the ketene acetal VIII. Finally, rotamer XIc can cyclize forming a C–C bond and the lactone 17. Indeed, a small amount of 17, also bearing partial deuteration at the indicated sites (¹H NMR analysis), was isolated as a closely eluting counterpart of 16 in this experiment. Although we do not have comprehensive evidence to support this mechanism in full, we offer it as a proposal that is consistent with all of our observations.

In summary, we have described a collection of reactions in which carbonyl-containing functionality, tethered to (HDDA-generated) benzynes, initiates nucleophilic attack on the electrophilic aryne. We are unaware of other examples of arynes being engaged by carbonyl oxygen nucleophiles in either a bimolecular or unimolecular sense. The initially formed zwitterion (cf. II to III) then encounters a sufficiently acidic external protic nucleophile (i.e., NuH) that protonates the aryl carbanion to give an ion pair (cf. IV). This collapses, producing the observed array of product types (cf. V). Overall, this strategy appears to be general and leads to compounds with novel structure motifs that may be of use to researchers interested in, e.g., structure-activity relationships.