A Divergent Mechanistic Course of Pd(0)-Catalyzed Aza-Claisen Rearrangement and Aza-Rautenstrauch-Type Cyclization of N-Allyl-Ynamides

Abstract

A fascinating mechanistic study of ynamido-palladium-π-allyl complexes is described that features isolation of a unique silyl-ketenimine via aza-Claisen rearrangement, which can be accompanied by an unusual thermal N-to-C 1,3-Ts shift in the formation of tertiary nitriles, and a novel cyclopentenimine formation via a palladium catalyzed aza-Rautenstrauch-type cyclization pathway.

We recently reported an account on the synthesis of pharmacologically useful amidines^1–4 from ynamides^5,6 via a Pd(0)-catalyzed N-to-C allyl transfer.⁷ As shown in Scheme 1, the formation of amidine 3 was proposed to proceed through ynamido-π-allyl complexes 2a or 2b after the initial oxidative addition [O.A.] of N-allyl-ynamides 1. Subsequently, after the addition of an amine, the reaction pathway would diverge from 4 or 5 depending upon the concentration of the amine HNR₂ and the nature of the palladium catalyst and ligands. Excess amount of amines or more nucleophilic secondary amines⁸ tend to attack the ynamido-π-allyl complexes 4 [or 5], leading to deallylated-amidines 6, whereas Pd(0) catalysts such as Pd₂(dba)₃ [instead of starting from Pd(II) species], and more bulky ligands such as X-phos⁹ and/or bidentate ligands with unique bite angles such as xantphos^10,11 that presumably promote reductive elimination [R.E.] favored the formation of allyl transferred amidines 3. Given the novelty of these ynamido-metal complexes and the potential of harvesting new reactivities, we examined this reaction in greater details mechanistically and uncovered a unique ketenimine intermediate, a rare 1,3-Ts shift, and an unusual and formally a Nazarov-type pathway leading to cyclopentenimine formation. We report here these findings.

Scheme 1.

Ynamido-Pd-π-Allyl Complexes.

Our initial experiments involved removing the amine nucleophile to suppress amidine formation in an attempt to isolate and/or observe key intermediates. As shown in Scheme 2, in the presence of 1 mol % of Pd₂(dba)₃ and 2 mol % of xantphos, heating of N-allyl ynamide 7 at 70 °C afforded two interesting products: Cyclopentenimine 8 and silyl-ketenimine 9 in ~ 5% and 88% yield, respectively.¹² The yield of 9 was improved with the formation of 8 completely impeded when the reaction was run at lower temperatures. While characterizations of 9 were unambiguous given its stability, the formation of amidine 10 in 95% yield via treatment of 9 with c-hex-NH₂ solidifies the identification of this novel intermediate.¹³

Scheme 2.

Isolation of a Stable Silyl-Ketenimine.

Despite the potential reactivity of N-sulfonyl-ketenimines, the surprising stability of ketenimine 9 is likely unique to the silyl substitution.^14,15 Under similar reaction conditions, ynamide 11 containing a Ph substituent led to a very different product, although in low yields. The product was initially assigned based on literature report¹⁶ as cyclobutane bis-imine 13, presumably attained through a facile dimerization or [2 + 2] cycloaddition of the less stable ketenimine 12.

The formation of ketenimines from N-allyl ynamides invokes an aza-Claisen rearrangement,¹⁷ specifically 3-az-Claisen, although those involving a C1-C2 acetylenic motif are very rare if not unprecedented.^17–19 However, the involvement of the palladium metal in the formation of 9 is distinctly clear, as a non-palladium involved aza-Claisen pathway required higher temperatures and longer reaction times [see 14→ 15→ 16 in Scheme 3]. When the aza-Claisen rearrangement was carried out at 70 °C, the reaction was sluggish as evident in yield of the trapping of the ketenimine 15 with pyrrolidine, and the usage of Ts or p-Ns group was not critical to the reactivity.

Scheme 3.

An Unusual N-to-C 1,3-Ts Shift.

When heating ynamide 17 at 110 °C led to again the dimer 19 and not the intended intramolecular cycloadduct 20 [Scheme 3], we sensed a possible mis-assignment because an intermolecular process had just dominated over an intramolecular one. We attained X-ray structure of the aza-Claisen product from 11 and were surprised that it was not the dimer 12 but a tertiary nitrile 22. Aza-Claisen rearrangements of ynamides in fact can occur in tandem with a rare N-to-C 1,3-Ts shift at 110 °C,²⁰ leading to nitriles with a quaternary carbon formation.

It is noteworthy that these results further accentuate the stability of silyl-ketenimine 9. While thermal rearrangement of 7 took place at 110 °C, 1,3-Ts shift to give TIPS-less nitrile 23 only proceeded after desilylation [9→9′], thereby suggesting sterics could also be at play. While this unusual 1,3-Ts shift in the formation of tertiary nitriles holds significant merit in synthesis,²¹ and that our finding cautions the assignment of possible homo-dimeric products from ketenimines, details of this shift will be examined in another study.

Having established the significance of the palladium metal in this aza-Claisen rearrangement, we believe ynamido-π-allyl complexes 2 derived from the oxidative addition of ynamide 7 should be responsible for the formation of ketenimines [Scheme 4]. The question is whether it proceeds through a reductive elimination process via an intramolecular pathway, or a Tsuji-Trost pathway. To addressing this question, we pursued two experiments.

Scheme 4.

A Pd-Model for the Ketenimine Formation.

The first experiment involved the use of ynamide 25 containing a crotyl group as shown in Scheme 4. While the reaction temperature had to be higher, new ketenimine 27 was isolated in 90% yield with a regiochemical ratio of 5:1. While this outcome concisely suggests that the crotyl group is scrambled through the oxidative addition, the resulting regiochemical ratio reflects that both pathways are possible with the major isomer 27a being derived from either reductive elimination of 2a [or 2a′: a ketenimine palladium complex], or a favored addition of ynamido anion [N^o−] to 26 at the less hindered site [blue arrow].

The second study is a revealing crossover experiment as shown in Scheme 5. With a 1:1 mixture of ynamides 25 and 28 containing a crotyl and allyl group, respectively, we were able to concisely assign 4 sets of ketenimines: 27a/b, 9, 29a/b, and 30. While the regiochemical ratio of 27a/b or 29a/b is the same at 5:1, the ratios of crossover of 10:1 even when the reaction was run at 0.10 M suggest that the ketenimine formation is likely an intramolecular process, and that these are likely tightly bound ynamido-π-allyl complexes shown as 2a and 2b [or 2a′/2b′].

Scheme 5.

A Crossover Experiment.

While literature precedents¹⁵ suggest that ketenimines represent highly reactive entities for developing useful synthetic methods, the observation of trace amounts of cyclopentenimine 8 and its mechanistic course captured our attention. Although it took much effort to optimize the formation of 8,²² it was found that conditions which relatively disfavored reductive elimination [5 mol % of Pd(PPh₃)₄] as well as the use of a phenolic substrate as an additive led to cyclopentenimine 8 in 70% yield. Preparation of 31 led to an X-ray structure, allowing an unambiguous assignment.

The difference in conditions implied that cyclopentenimine 8 is likely derived from a different mechanistic course for which silyl-ketenimine 9 is not an intermediate.²³ This assertion is consistent with the fact that treatment of 9 with either 5 mol % of Pd(PPh₃)₄, or by using thermal conditions, did not provide any identifiable amount of 8, thereby ruling out the possibility of a formal imino-Nazarov type cyclization^24,25 involving 32a–c [Scheme 6].

Scheme 6.

An Effective Cyclopentenimine Formation.

Instead, a likely mechanistic course would involve an aza-variant of a Rautenstrauch-type cyclization [or also formally an aza-Nazarov-type cyclization]^26,27 as shown in Scheme 7. While the Pd-complex 2b could readily reductive eliminate to give ketenimine 9, under conditions in which the reductive elimination is slowed, a Pd-[3,3] sigmatropic rearrangement could occur to give α-imino palladium carbenoid 33a. While a number of possibilities could take place from thereon, one possibility that is consistent with the use of PhOH would entail the formation of enamido-Pd-Complex 33c could undergo migratory insertion [M.I.] followed by β-elimination to afford cyclopentenimine 8 after tautomerization of cyclopentadienamide 33e. An alternative pathway would proceed through dienyl palladium carbenoid 34a derived from tautomerization of 33a.

Scheme 7.

An Aza-Rautenstrauch-Type Cyclization Pathway.

We have uncovered here a fascinating divergent mechanistic pathway consisting of a Pd(0)-catalyzed aza-Claisen rearrangement of N-allyl ynamides, which can also be accompanied with an N-to-C 1,3-Ts shift through the ketenimine intermediate, and an aza-Rautenstrauch cyclization. These studies provide insight into the nature of ynamido-π-allyl complexes as well as new reactivities with synthetic potential. Efforts are underway in pursuing synthetic methods involving ketenimines and the N-to-C 1,3-Ts shift as well as applications using cyclopentenimines.