Generation [via Tetradehydro-Diels-Alder (TDDA) Reactions] and Reactivity of 6-Fluorocyclohexa-1,2,4-triene Intermediates

Abstract

We report the engagement of 2,6-difluorophenyl substituent in tetradehydro Diels–Alder (TDDA) reactions. The use of fluorine was guided by DFT computations. Acceleration of the TDDA cycloisomerization by an anhydride linker in the substrates was also key. Trapping of the strained cyclic allene intermediate by nucleophilic carbons in external heterocycles, an enol, and alkenes led to unusual reactivities (e.g., one rationalized by rearrangement to a carbene intermediate).

Graphical Abstract

Among cycloaddition reactions, the Diels–Alder [4+2] reaction between a diene and a dienophile is the most extensively studied and applied. Analogous reactions in which one, two, or three of the alkenes are replaced by the more highly oxidized analogs (i.e., alkynes) are known collectively as dehydro-Diels–Alder reactions.^1,2 One specific variant, known as the tetradehydro-Diels–Alder (TDDA) reaction, involves the net cycloaddition between a conjugated enyne with an (usually tethered) alkyne – the enynophile (Figure 1a). The intermediate cycloadduct is a transient species containing a strained 1,2,4-cyclohexatriene (cf. 2), which frequently rearranges to an aromatized, benzenoid product via a net prototropic shift (cf. 2 to 3).^3,4,5,6 The parent 1,2,4-cyclohexatriene (4, Figure 1b) was first generated as an intermediate by Christl and coworkers as deduced from its trapping reactions with furan (5 [4 + 2]) and styrene (6 [2 + 2]).⁷

Figure 1.

a) Prototypical example of a TDDA cyclization; b) First preparation of 1,2,4-cyclohexatriene (4);⁷ c) TDDA reactions typically convert arylalkynes into naphthalene derivatives via prototropic aromatization; d) This study: TDDA reactions of 2,6-fluorophenylated alkynes 10 and reactivity of 6-fluoro-1,2,4-cyclohexatriene intermediates 11.

Most commonly, the ene component in TDDA substrates is embedded within an aryl substituent (7, Figure 1c). The strained allene intermediate 8 typically rearranges to naphthalenoid products 9. Notably, we know of only two examples of a TDDA reaction in which neither of the ortho carbons of the aryl group in the generic substrate 7 bear a hydrogen atom. Staab and Ipaktschi reported engagement of the ortho carbon of a mesityl group in a (slow) reaction to give 10% of a TDDA product (that, incidentally, had lost a methyl group).⁸ Wessig and coworkers have observed photochemical versions of this type of reaction that also give (minor) products in which a mesityl ring has participated in the cyclization.^9,10

We were intrigued by the challenge of finding a substrate containing a 2,6-disubstituted aryl substituent (cf. R in 7 in Figure 1c) that would undergo a TDDA cycloisomerization to a 1,2,4-cyclohexatriene intermediate. If so, what would the fate of that strained allene be, since the prototropic reorganization to a benzenoid product would no longer be available? In this preliminary report, we describe our success in answering some of these questions. More specifically, we have found that several 6-fluorocyclohexa-1,2,4-triene species 11 can be formed by TDDA cyclization of precursors 10 bearing a 2,6-difluorophenyl substituent. Additionally, these strained species undergo several different types of novel trapping reactions.

In view of the rarity of TDDA reactions in which a 2,6-disubstituted arene has engaged in the cyclization,^8,9 we felt that the choice of the ortho substituent as well as the nature of the tether between the enyne and enynophile would likely be important factors if we were to succeed. To guide our choice of the substituent to explore experimentally in precursors such as 7, we carried out a preliminary DFT computation. Presuming that the TDDA reaction involving alkynyl arenes would proceed through a stepwise diradical (or, if concerted and asynchronous, would have some of the character of a diradical),^11,12 we examined substituent effects on the relative ease of C–C bond formation in the cyclization of a very simple model: closure of the alkenyl radical 12 onto the pendant arene to give the delocalized pentadienyl radical 13 (Figure 2). Of the several analogs we examined, the E_act for closure of the difluorinated case 12b was closest to that of the unsubstituted parent species 12a: namely, 8.2 kcal mol⁻¹ for 12b vs. 5.2 kcal mol⁻¹ for 12a. This preliminary evaluation led us to focus our initial efforts, those reported here, on 2,6-difluorophenyl-containing TDDA precursors. More recently, we have become aware of a 2024 report by Wessig and coworkers in which alkynes bearing a 2-fluoro-6-hydroaryl moiety underwent photo TDDA reactions involving competitive cyclizations to the 2- and 6-positions.¹³ In one instance, cyclization to the fluorinated carbon was observed exclusively. That study clearly indicates that closure of a radical center to a fluorine-bearing carbon can be quite competitive with that to an unsubstituted ortho carbon.

Figure 2.

DFTⁱ calculated activation energies (E_act) for cyclization of the simple model mono-radicals 12 to an ortho-carbon in the pendant aromatic ring to produce the more stable, delocalized radicals 13.

ⁱSMD(dichloromethane)/M06-2X/6-311+G(d,p)

The structure of the linking group between the reacting π-components in dehydro-Diels–Alder substrates can significantly impact the rates of their intramolecular cyclizations. We recently reported that anhydride-tethered triynes 14, generated in situ by the dehydration of conjugated diynoic acids, show dramatically increased rates of hexadehydro-Diels–Alder (HDDA) cyclizations (cf. 14 to 15 at room temperature; Figure 3a).^14,15 The anhydride linker involved in the first TDDA reaction (phenylpropiolic acid in refluxing acetic anhydride)¹⁶ was later shown to have had a dramatic accelerating effect in the cyclization of the in situ-formed phenylpropiolic anhydride.¹⁷ Taken together, these facts suggested that the use of an anhydride linker strategy could also be advantageous in overcoming the potential challenge of effecting the conversion of 10 to 11 (Figure 1d).

Figure 3.

a) Rate enhancement of diyne-yne cyclization with an anhydride tether to produce HDDA benzyne derivatives;14 b) Now: cyclization of the 2,6-difluorophenylated acid 16 to 17 via the TDDA cyclization of the hindered anhydride 18 to the strained allene 19 and trapping by furan.

In the first experiment (Figure 3b), we used methanesulfonyl chloride (MsCl) and pyridine to effect the dehydrative dimerization¹⁸ of 2,6-difluorophenylpropiolic acid (16) in the presence of furan as a potential trapping agent. The choice of furan was guided by its precedented use in capturing the parent 1,2,4-cyclohexatriene to produce a [4+2] cycloadduct (Figure 1b).⁷ The reaction was carried out at room temperature, and we were pleased to observe formation of an adduct that had incorporated elements from two molecules of 16 and one of furan. In particular, GC-MS data of the reaction mixture showed a significant peak having a nominal mass of 394 amu. This indicated that one molecule each of water as well as hydrogen fluoride had been lost. That is, furan had not trapped the intermediate to give simply a 1:1 adduct. Isolation (in 30% yield) and NMR characterization of the major product led to its identification as the naphthalic anhydride derivative 17 – the product of a formal S_N2’ trapping reaction of the intermediate 6-fluorocyclohexa-1,2,4-triene species 19. Although the mode of trapping was unexpected, we were nonetheless encouraged by the evidence this gave for the facile (room temperature) TDDA cycloisomerization of the anhydride 18 to the cyclohexadiene derivative 19 – the first instance of participation of a 2,6-disubstituted aryl ring other than a mesityl in a TDDA reaction. There was no sign of formation of a cycloadduct such as 20 in either NMR or MS analysis of the crude product mixture.

Building on this initial result, we explored variation of reaction conditions to learn if we could improve the yield of product 17. Changing the amounts of pyridine or furan or elevating the reaction temperature had a slight detrimental effect. Use of acetonitrile, acetone, or ether as solvent returned only the starting acid 16. Replacement of pyridine with Proton-sponge^® (PS), a more hindered weak base that we had previously seen to be effective in the formation of reactive anhydrides from diynoic acids,⁷ significantly improved the yield. We then explored the use of methanesulfonic anhydride instead of MsCl as the dehydrating agent; again, PS proved superior to pyridine as the base and gave a 48% yield (NMR). Changing the dehydrating reagent again, this time to trifluoroacetic anhydride (TFAA), showed a further improvement in yield. Finally, adjustment of the molar ratios of TFAA and PS led to the highest isolated yield (80% in DCM at ambient temperature) of 17 using acid 16 (1.0) to TFAA (1.0) to PS (1.0) to furan (3.0 equiv). These are the conditions used in the subsequent experiments designed to learn about the reactivity of the strained allene 19.

We explored a variety of different nucleophilic trapping agents (Table 1). The heterocycles indole, N-methylpyrrole, and 2,3-dimethylthiophene trapped the allene in moderate to good yields, affording in each case a mixture of regioisomeric products resulting from engagement at one of two adjacent carbons on the five-membered heteroaromatic ring (21a,b, 22a,b, and 23a,b).

Table 1.

Capture of 19 by electron-rich C=C nucleophiles.

^aConditions: [16]₀ = 0.14 M; 1.0 molar equivalent of Proton-sponge^®, 1.0 molar (= 2.0 stoichiometric) equivalent of TFAA; 3.0 molar equivalents of the nucleophile; DCM as solvent; overnight at ambient temperature.

We then identified two trapping agents in addition to furan that would yield a single product rather than a mixture of regioisomers. Traps containing a single nucleophilic carbon atom, such as (the enol of) 2,4-pentanedione and 3-methylindole, provided 24 and 25 as the sole allene-derived products. Several attempts to trap the allene 19 with heteroatomic nucleophiles such as alcohols and amines were unsuccessful, likely because of the inherent reactivity of the anhydrides present in the reaction mixture.

With the success of trapping of the strained allene with species containing nucleophilic C=C π-bonds, we asked whether simple alkenes might be effective trapping agents. Styrene (cf. 4 to 6, Figure 1b⁷) and cyclohexene were selected as prototypical examples. Each led to unexpected and interesting products.

In the case of styrene, two products were isolated and identified (Figure 4a): i) the benzylic fluoride 26 (31% yield), formally a fluoro-ene product, and ii) the stilbene derivative 27-Z (10% yield). These can be viewed to arise by loss of fluoride ion from allene 19 to generate the naphthyl carbenium ion 28a followed by addition of styrene to produce the benzylic cation 28b. This ion pair could then proceed to the products 26 or 27-Z.

Figure 4.

a) Styrene and b) cyclohexene trap the strained allene 19 to give unusual 1:1 adducts.

In the case of cyclohexene trapping of the allene again afforded unusual products: the isomeric cyclohexenyl adducts 29 and 30. It is tempting to envision these as arising via competitive proton loss from the zwitterion 31 formed by attack of the alkene upon 19. However, invoking the intermediacy of a secondary carbocation in a mechanistic pathway has been discouraged when alternatives can be considered.^19,20 Another option in the current setting involves formation of an initial [2 + 2] adduct such as the cyclobutane intermediate 32 (cf. Figure 1b). Admittedly, there are eight isomeric cyclobutane analogs of 32, but the intramolecular extrusion of HF from the isomer shown (a formal α-elimination) is enticing. That strain-relieving event leads to 33, a fully aromatized naphthalene (+ HF) bearing a carbenic cyclohexyl substituent. DFT studies (see discussion in the SI) indicate that the Gibbs energy of reaction to form carbene 33 from 32 is exergonic by 7.6 kcal mol⁻¹. Moreover, the computed relative energies of activation for insertion into the vicinal tertiary (to 29) vs. secondary (to 30) C–H bonds in the carbene 33 are in accord with the observed product ratio.

We also examined the fate of the strained allene 19 formed in the absence of any exogenous trapping agent (Figure 5). Following chromatographic purification on silica gel of the crude product mixture arising using PS and TFAA alone, the phenolic compound 34 was the major isolated product. The GC-MS data of the crude mixture prior to purification indicated that the dominant material had a nominal mass of 440 amu, indicative of the (labile) trifluoroacetate derivative 35. This outcome represents the only instance in which we have observed a heteroatomic nucleophile to engage the strained allene.

Figure 5.

Evidence for trifluoroacetate trapping of the allene 19 when no other trapping agent was purposely present. TFFA = trifluoroacetic anhydride.

Finally, to learn if the TDDA engagement of a 2,6-difluorophenyl ring could be extended to substrates containing linkers other than the anhydride in 18, we studied the diynyl ester 36 and enone 38 (Figure 6). These proved to be far less reactive than 18. The ester 36 required heating to 220 °C, and in the presence of 2,4-pentanedione (only minimal conversion at 180 °C overnight) produced the acac-derivative 37. A cycloheptenone linker was earlier seen to significantly enhance the rate of HDDA cycloisomerizations of tetrayne precursors.²¹ The diynyl enone 38 reacted at a rate intermediate to those of the ester 36 and anhydride 18, cyclizing at 120 °C to give the furan trapped product 39.

Figure 6.

1,6-Difluorophenylated substrates having linkers other than the anhydride will also undergo the TDDA reaction, albeit much more slowly.

In summary, we have described here our preliminary observations of the tetradehydro-Diels–Alder reaction of substrates where the ene portion of the enyne component is a phenyl group lacking an ortho hydrogen atom. Fluorine was chosen as the hydrogen atom replacement based on a model DFT study of the ease of radical cyclization by addition to a substituted carbon atom of a pendant aryl group. An anhydride linker, while not a strict requirement, greatly enhances the rate of the TDDA cyclization. Neutral, weakly nucleophilic, C=C moieties end up engaged to the central carbon atom (C2) of the strained 1,2,4-cyclohexatriene intermediates. Alkenes lead to unusual and mechanistically intriguing outcomes. Studies of further aspects (e.g., change of R = F to R = Hal, OR, alkyl, etc. and extension to alkenyl enynes ~C≡CCH=CR₂) are in progress.

Supplementary Material

“The Supporting Information is available free of charge on the ACS Publications website.”

Details for the preparation and spectroscopic characterization (including copies of the ¹H and ¹³C NMR spectra) for each new compound and details and energies of DFT computations (PDF).

FAIR data: Gaussian .log files for all DFT optimized structures (ZIP).

FAIR data: a master .mnova file of manipulatable NMR spectra for compounds S3, S4,16, 17, 21–23a&b, 24–27, 29&30, 34, and 36–39 has been uploaded as a ZIP file to figshare: (Private link: https://figshare.com/s/da9f3b4ec610a0f9fa8f)

Data Availability Statement

The data underlying this study are available in the published article, in its Supporting Information, and openly available in figshare at Public link: DOI 10.6084/m9.figshare.30172291. Private link: https://figshare.com/s/da9f3b4ec610a0f9fa8f