Synthesis and Characterization of Keteniminyl Anion Salts.
Initially, we synthesized the monophosphino-substituted diazomethane 1 (31P NMR: 112.1 ppm) via the methanolysis of its precursor, (phosphino)(silyl)diazomethane.57 Subsequent in situ deprotonation of 1 and the addition of equimolar quantities of 2,6-xylyl isonitrile (XylNC) at ambient temperature yielded the potassium keteniminyl anion salts 2a (31P NMR: 75.4 ppm), 2b (complexed with 18-C-6, 31P NMR: 81.7 ppm) and 2c (complexed with [2,2,2]-cryptand, 31P NMR: 83.2 ppm) (Fig. 2).
Further explorations were directed towards thiophosphino keteniminyl anions to probe the influence of substituents on the structural attributes of keteniminyl anions. The oxidation of 1 to 3 (31P NMR: 65.6 ppm) was achieved using 1/8 equivalents of S8 in benzene (Fig. 2). This was followed by in situ deprotonation and subsequent addition of equimolar quantities of XylNC at ambient temperature, leading to the synthesis of potassium keteniminyl anion salts 4a (31P NMR: 68.4 ppm) and 4b (complexed with 18-C-6, 31P NMR: 45.8 ppm).
The infrared (IR) spectra of 2a-c and 4a-b display CCN stretching frequencies ranging from 1997 to 2043 cm− 1 (Table 1). Except for 4a, which exhibits a stretching frequency at 2043 cm− 1, these frequencies are generally lower than those observed in B (2032 cm− 1).45 The 13C NMR spectra of 2 and 4 show that C(1) manifests as a doublet across the series, with specific chemical shifts and coupling constants recorded as 56.8 ppm (1JP−C = 59.6 Hz) for 2a, 55.8 ppm (1JP−C = 46.0 Hz) for 2b, 56.9 ppm (1JP−C = 32.3 Hz) for 2c, 47.0 ppm (1JP−C = 228.5 Hz) for 4a, and 46.7 ppm (1JP−C = 193.9 Hz) for 4b. Additionally, C(2) also appears as a doublet with shifts of 140.9 ppm (2JP−C = 12.5 Hz) for 2a, 142.4 ppm (2JP−C = 12.4 Hz) for 2b, 143.2 ppm (2JP−C = 11.7 Hz) for 2c, 136.8 ppm (2JP−C = 4.5 Hz) for 4a, and 138.6 ppm (2JP−C = 4.6 Hz) for 4b. When compared to structurally similar compounds such as B (47.9 ppm),45 Ph3P = C = C = NPh (5.8 ppm),58 Ph3PC(AuCl) = C = NPh (33.4 ppm),59 and M(Ph3PC − C ≡ N) (M = Li, Na, K) (− 5.4–5.4 ppm),60 the C(1) resonances exhibit a significant high-frequency shift. Although the C(2) signal for B was not assigned,45 preventing a direct comparison, the C(2) signals in our systems are comparable to those found in Ph3P = C = C = NPh (136.4 ppm) and M(Ph3PC − C ≡ N) (127.7–142.5 ppm), albeit displaying a lower frequency shift relative to Ph3PC(AuCl) = C = NPh (170.1 ppm).
X-ray crystallographic analysis unambiguously established the molecular structures of 2a-c and 4a-b (Fig. 3). Compound 2a crystallizes as a tetrameric arrangement (Fig. 3a), consisting of two dimeric subunits linked through K(2)−η6-Xyl interactions. Within each dimeric subunit, K(1) is coordinated by P(1), C(1), C(2), C(3), C(4), N(2), and the Dipp group from the diazaphosphinyl moiety, while the coordination sphere of K(2) includes C(2), N(1), C(3), C(4), and N(2). In 2b, the [K(18-C-6)] cation preferentially interacts with N(1) rather than C(1) (Fig. 3b), likely influenced by steric hindrance from the diazaphosphino group. For 2c, the potassium ions are fully encapsulated by cryptands, resulting in no direct contact with the “naked” keteniminyl anions (Fig. 3c). Similar to 2a, compound 4a crystallizes as a tetrameric structure (Fig. 2d). Most notably, in 4b, the [K(18-C-6)] cation forms a monomeric complex by chelating C(1) and S(1) atoms (Fig. 3e), closely resembling the structure of Gessner’s ketenyl anion potassium salt [Ph2P(S)C = C = O][K(18-C-6)].5 It is noteworthy that 4b represents the first crystallographically characterized α-alkali-metallated keteiminyl anion.
Table 1
Selected bond lengths and angles as well as IR frequencies of the C − C−N vibration of 2a-c, 4a-b and B.
| 2a | 2b | 2c | 4a | 4b | B |
C − C (Å) | 1.231(2) 1.247(3) | 1.242(2) | 1.232(6) | 1.249(2) 1.245(3) | 1.284(4) | 1.221(5) |
C − N (Å) | 1.306(2) 1.291(2) | 1.2927(18) | 1.295(5) | 1.265(2) 1.272(3) | 1.275(4) | 1.316(4) |
C − C−N (°) | 177.1(2) 167.8(2) | 175.25(14) | 172.5(4) | 168.3(2) 170.9(2) | 170.4(3) | 175.3(3) |
P − C (Å) | 1.749(2) 1.747(2) | 1.7395(14) | 1.731(5) | 1.693(1) 1.698(2) | 1.692(3) | 1.796(3)[a] |
P − C−C (°) | 164.3(2) 146.0(2) | 154.24(12) | 162.7(4) | 154.6(2) 151.6(2) | 152.2(2) | 171.6(3)[b] |
C − C−N (cm− 1) | 2016 | 2006 | 2010 | 2043 | 1997 | 2032 |
[a] Si − C bond length. [b] Si − C−C bond angle.
The P − CCN units in these compounds demonstrate notably distinctive bond angles and lengths compared to related compounds (Table 1). Specifically, the P − C−C bond angles in 2a-c range from 146.0(2) to 164.3(2)°, and in 4a, they are 151.6(2) and 154.6(2)°, all of which are significantly more acute than the Si − C−C in B (171.6(3)°).45 The C − C bond lengths in 2a-c (1.231(2) to 1.247(3) Å) and 4a (1.245(3) and 1.249(2) Å) sit between typical double and triple bond lengths,61 and are slightly longer than those in B (1.221(5) Å). Conversely, the C − N bond lengths in 2a-c (1.291(2) to 1.306(2) Å) and 4a (1.265(2) and 1.272(3) Å) are intermediate between typical single and double bond lengths,61 and slightly shorter than those in B (1.316(4) Å). These structural features demonstrate an intermediate bonding situation that leans more towards a keteniminyl character than that observed in B, likely due to the weaker negative hyperconjugation effect arising from a phosphino substituent compared to a silyl substituent. It is particularly noteworthy that the acute P − C−C bond angle of 152.2(2)°, extended C − C bond length of 1.284(3) Å, and shortened C − N bond length of 1.275(3) Å in 4b distinctly showcase the most pronounced keteniminyl character among the series.
The phosphorus atoms in 2a-c exhibit pyramidalized geometries, akin to the structural features observed in the phosphinoketenyl anion6 and the rhodadiphosphinocarbene.62 This geometry suggests that the phosphino groups in 2a-c do not act as π-donors; typically, when a phosphino group serves as a π-donor, a planar configuration of the phosphorus atom is observed (Fig. 4a), as evidenced in monophosphino-substituted carbenes.63 However, the P − C bond lengths in 2a-c—measured at 1.749(2) and 1.747(2) Å for 2a, 1.7395(14) Å for 2b, and 1.731(5) Å for 2c—lie between typical single (1.86 Å) and double (1.69 Å) bond lengths,61 indicative of P − C multiple bond character. This indicates that π-interactions, specifically involving the π electrons on C(1) and the low-lying N − P σ* antibonding orbitals, contribute to the molecular structure of 2a-c (Fig. 4a) (vide infra). Indeed, the P − C bond lengths in 4a-c—recorded at 1.693(1) and 1.698(2) Å for 4a, and 1.692(3) Å for 4b—are much closer to typical P = C double bond lengths, pointing to a greater π-accepting capability of the diazathiophosphinoyl substituent compared to the diazaphosphinyl substituent.
In our pursuit to unravel the electronic intricacies of the keteniminyl anions of 2, we have employed analyses based on intrinsic bond orbitals (IBOs)64,65 (Figure S67) and natural bond orbital (NBO) methods, utilizing the geometry optimized at the BP86/def2-TZVP theoretical level. The highest occupied molecular orbital (HOMO) of 2 comprises an out-of-plane P(1) − C(1) − C(2) three-center-two-electron (3c-2e) π bonding orbital, which interacts with the out-of-plane p-orbital at N(1) (Fig. 4b). The HOMO − 1 is characterized by an in-plane C(1) − C(2) π bonding orbital, accompanied by a lone pair at N and a lone pair at P. Moreover, electron localization function (ELF)66 calculations indicate a more localized electron density at C(1) in 2 than in B (Figs. 4c and 4d). This observation explains the more acute P − C−C bond angle relative to the Si − C−C bond angle in B, thereby underscoring the pronounced keteniminyl character of 2.
Natural population analysis (NPA) of 2 elucidates negative charges localized at C(1) (− 0.78 a.u.) and N(1) (− 0.54 a.u.) (Figure S68). The Wiberg bond indices (WBIs) for the bonds P(1) − C(1), C(1) − C(2), and C(2) − N(1) register at 1.11, 2.19, and 1.56, respectively. Particularly notable is the relatively short P(1) − C(1) bond in 2, which measures between 1.731(4) and 1.749(2) Å and exhibits a high WBI. This, along with a 3c-2e P(1) − C(1) − C(2) π bond, robustly underscores the π-acceptor nature of the diazaphosphinyl group, especially when the phosphorus atom adopts a pyramidalized configuration.
Reactivity of the Keteniminyl Anion Salt 2b.
The notable contribution of C(1) to both the HOMO and HOMO − 1 in 2, combined with the substantial negative charge localized at C(1), leads us to hypothesize that keteniminyl anions preferentially undergo selective reactions at the carbon atom, rather than at the nitrogen end. Building on this, we examined the reactivity of derivative 2b. Introduction of 2b to a mild Brønsted acid Et3N•HCl, rapidly yielded ketenimine 5 (Fig. 5), which was characterized by a 31P NMR resonance at 105.1 ppm. The 1H NMR spectrum of 5 displayed a doublet at 4.30 ppm with a two-bond phosphorus-proton coupling constant of 10.9 Hz, integrative of a single proton. X-ray diffraction techniques later confirmed the structure of 5 as a hydrogen-substituted ketenimine (Fig. 6a).
Further investigations of 2b with electrophiles, namely MeI, TMSCl, and IDippCuI (IDipp = 1,3-bis(2,6-diisopropylphenyl)imidazol-2-ylidene), smoothly led to the formation of ketenimine 6 (31P NMR: 121.1 ppm), silyl ketenimine 7 (31P NMR: 120.7 ppm), and copper-keteniminyl complex 8 (31P NMR: 103.5 ppm), respectively (Fig. 5). X-ray crystallography confirmed the solid-state structures of 6–8 (Figs. 6b-d), verifying the bond formations at the C(1) position.
Notably, the C(1) − C(2) bond lengthens from 1.242(2) Å in 2b to 1.319(3) Å in 5, 1.307(5) Å in 6, 1.311(3) Å in 7, and averages 1.294(3) Å in 8, while the C(2) − N(1) distances decrease (from 1.2927(18) Å in 2b to 1.216(2) Å in 5, 1.227(5) Å in 6, 1.224(3) Å in 7, and an average of 1.239(4) Å in 8). This alteration in bond lengths can be attributed to the disappearance of the ynamide anion resonance form upon formation of the neutral ketenimines 5–8. The straightforward formation of these derivatives emphasizes the crucial role of the keteniminyl resonance in 2 and its remarkable potential for synthesizing diverse ketenimine derivatives.
Upon stirring a THF solution of 2b with diphenylamine, a new compound 9 was formed (31P NMR: 119.0 ppm) (Fig. 5). Mirroring the characteristics of 5, the 1H NMR spectrum of 8 displayed an alkene resonance at 4.13 ppm, indicative of a single proton. The solid-state structure of 9 (Fig. 6e), is identified as an amino enamide salt, which is the result of a 1,2-addition of diphenylamine to the C(1) − C(2) bond in 2b. This discovery implies that keteniminyl anions can serve as powerful synthons for a variety of functionalized compounds. Furthermore, chemical modification of the phosphino group in 2b has been achieved. In particular, selective oxidation of the phosphino group with Ph2CN2 leads to the formation of phosphazene 10 (31P NMR: 15.3 ppm), while maintaining the structural integrity of the keteniminyl moiety (Figs. 5 and 6f).
Reactivity of the Copper Keteniminyl Complex 8.
It is important to note that previous studies had indicated the transient nature of copper-keteniminyl complexes, elusive in isolation but implicated in various organic transformations.29–39 Chang’s group previously reported the synthesis of ketenimines through the copper-catalyzed azide-alkyne cycloaddition, followed by ring-opening rearrangement and nitrogen release.34,35 The in situ generated copper-keteniminyl species are theorized to couple with diverse nucleophiles, yielding functionalized carbonyl derivatives. Yet, these proposed copper-keteniminyl intermediates have not been directly observed.
Remarkably, we now have 8, the first isolable copper-keteniminyl complex. This spurred our investigation into whether 8 acts as a key intermediate in such transformations. Intriguingly, the addition of excess Nu − H (MeOH or EtNH2) to a THF solution of 8 led to the formation of imidate 11 (31P NMR: 112.0 ppm) and amidine 12 (31P NMR: 117.1 ppm), respectively. We presume these reactions initially undergo a protodecupration to yield the neutral ketenimine 5, which is then followed by a 1,2-addition of Nu − H across the C = C double bond.67 However, our experiments showed that 5 did not react with MeOH at room temperature. The reaction of 5 with an equimolar amount of IDippCuOMe resulted in a mixture of 11 (31P NMR: 112.0 ppm), 8 (31P NMR: 103.4 ppm), and two additional species (31P NMR: 106.4 ppm and 99.6 ppm, Figure S42). Adding an excess amount of MeOH into such a mixture led to the clean formation of 11 over 13 hours. Under the influence of a catalytic amount of IDippCuOMe (30 mol%), the reaction between 5 and MeOH slowly yielded 11 at room temperature (Figure S43-45). Indeed, IDippCuOMe, released from the reaction of 8 with methanol, serves a catalytic function, facilitating the transformation of 5 into 11, a mechanism supported by our computational studies (Figure S69). These findings substantiate the hypothesis that copper-keteniminyl complexes play a crucial role in Chang’s copper-catalyzed multi-component cascade reactions, with copper acting as a key mediator in enabling the interaction between ketenimine and nucleophiles.