Cationic Phosphinidene as a Versatile P₁ Building Block: [L_C–P]⁺ Transfer from Phosphonio–Phosphanides [L_C–P–PR₃]⁺ and Subsequent L_C Replacement Reactions (L_C = N-Heterocyclic Carbene)

Abstract

Cationic imidazoliumyl(phosphonio)-phosphanides [L_C–P–PR₃]⁺ (1a–e⁺, L_C = 4,5-dimethyl-1,3-diisopropylimidazolium-2-yl; R = alkyl, aryl) are obtained via the nucleophilic fragmentation of tetracationic tetraphosphetane [(L_C–P)₄][OTf]₄ (2[OTf]₄) with tertiary phosphanes. They act as [L_C–P]⁺ transfer reagents in phospha-Wittig-type reactions, when converted with various thiocarbonyls, giving unprecedented cationic phosphaalkenes [L_C–P=CR₂]⁺ (5a-f[OTf]) or phosphanides [L_C–P–CR(NR₂^′)]⁺ (6a-d[OTf]). Theoretical calculations suggest that three-membered cyclic thiophosphiranes are crucial intermediates of this reaction. To test this hypothesis, treatment of [L_C–P–PPh₃]⁺ with phosphaalkenes, that are isolobal to thioketones, permits the isolation of diphosphirane salts 11a,b[OTf]. Furthermore, preliminary studies suggest that the cationic phosphaalkene [L_C–P=CPh₂]⁺ may be employed to access rare examples of η²–P=C π-complexes with Pd⁰ and Pt⁰ when treated with [Pd(PPh₃)₄] and [Pt(PPh₃)₃] for which analogous complexes of neutral phosphaalkenes are scarce. The versatility of [L_C–P]⁺ as a valuable P₁ building block was showcased in substitution reactions of the transferred L_C-substituent using nucleophiles. This is demonstrated through the reactions of 5a[OTf] and 6c[OTf] with Grignard reagents and KNPh₂, providing a convenient, high-yielding access to MesP=CPh₂ (16) and otherwise difficult-to-synthesize 1,3-diphosphetane 17 and P-aminophosphaalkenes.

Introduction

Simple phosphorus-containing functionalities can represent important tools for the construction of novel structural architectures with application in areas such as catalysis, polymers, and materials. For instance, phosphinidenes [R–P] are considered valuable and simple P₁ building blocks for the synthesis of organophosphorus substrates or as diverse ligands in transition-metal complexes.¹ Despite their synthetic utility and fundamental curiosity, phosphinidenes display exceedingly high reactivity, and the first isolable “free” phosphinidene, reported in 2016, remains the only example.² Thus, researchers have designed a variety of more applicable precursors, phosphinidenoids, which can be employed in phosphinidene transfer reactions. Of particular importance were the early investigations of transition-metal-supported phosphinidenoid reagents as [R–P] building blocks, thereby affording otherwise difficult to access organophosphorus compounds.^1,3 In a few cases, the release of [R–P] from metal-free precursors has also been described, for example, from so-called inversely polarized phosphaalkenes R–P^δ−=C^δ+R₂.⁴

More recently, a new generation of phosphinidene chemistry has evolved with exciting breakthroughs involving isolable, neutral, and metal-free singlet [R–P] transfer reagents that are tolerable of a variety of substituents (Figure 1a).

Figure 1.

(a) Examples of neutral phosphinidene [R–P] transfer reagents; (b) synthesis of cationic [L_C–P]⁺ transfer reagents reported here. Mes* = 2,4,6-^tBuC₆H₂, Dipp = 2,6-ⁱPr₂C₆H₃.

For instance, amino-phosphinidene, [R₂N–P], transfer has been enabled from precursor I with concomitant formation of anthracene.⁵ Importantly, the first transfers of the parent phosphinidene, [H–P], were observed from II.⁶ Carbene-phosphinidene adducts III and phosphanylidenephosphoranes IV (or “phospha-Wittig reagents”⁷) have been shown to transfer aryl- and alkyl-substituted phosphinidenes to a wide range of substrates [e.g., organic electrophiles,⁸ aldehydes,⁹ NHCs (N-heterocyclic carbenes),¹⁰ isonitriles,¹¹ ammonia,¹² and Al^I species.¹³ Diphosphadiboretane V has been utilized as a [Mes*–P] transfer agent to ketones, amides, and esters in the unprecedented phospha-bora-Wittig reaction.¹⁴ Despite advances in the field, the development of a single phosphinidene transfer reagent capable of transferring phosphinidenes [R–P] with a multitude of different substituents R is still desired. In a recent study, we demonstrated the versatility of cationically substituted phosphorus compounds for the formation of P–C, P–N, and P–O bonds by easily replacing the cationic substituent using commercially available reagents.¹⁵ This inspired us to explore the potential of employing cationic substituents for phosphinidenes to create unprecedented cationic phosphinidene transfer reagents, namely, [L_C–P]⁺ (Figure 1b). This could enable further functionalization at the P atom after the transfer reaction and render [L_C–P]⁺ a versatile P₁ building block.

We recently discovered that the tetracationic tetraphosphetane 2⁴⁺ (Figure 1), formally a tetramer of [L_C–P]⁺, may be conveniently obtained in good yields (86%) as its triflate salt from the reduction of 3[OTf] with 1,4-bis(trimethylsilyl)-1,4-dihydropyrazine (4, Scheme 1).¹⁶ Computational studies on 2⁴⁺ suggest a high electrophilicity due to the four imidazoliumyl substituents. We therefore hypothesized that nucleophilic cleavage with tertiary phosphanes R₃P might provide suitable access to phosphonio–phosphanides 1⁺ as potential [L_C–P]⁺ transfer reagents.

Scheme 1. Synthesis of 2[OTf]₄ and Its Nucleophilic Fragmentation with Tertiary Phosphanes R₃P (R = Me, Et, Cy, and Ph).

Reagents and conditions: (i) +4 Ph₃P, CD₃CN, rt, 16 h; (ii) +4 R₃P, CH₃CN, rt, 4–16 h, 88–93%.

We now report a straightforward route to simple [L_C–P]⁺ transfer agents (1a-d⁺) from readily available starting reagents. Their utility is demonstrated by cationic phosphinidene transfer to substrates, including thioketones, thioamides, thiourea, thioesters, and phosphaalkenes R–P^δ+=C^δ−R₂. Unprecedented phosphonio–phosphanides, 1b-d⁺, have been characterized crystallographically as triflate salts, along with a series of hitherto unknown cationic phosphaalkenes, phosphanides, diphosphiranes, and metal complexes, including very rare η²–P=C–Pd⁰ and Pt⁰ complexes. In addition, we demonstrate the ability to perform substitution reactions of the transferred L_C-substituent in selected substrates using widely applied nucleophilic aryl and alkyl Grignard reagents RMgBr (R = Mes, Me), as well as amide KNPh₂. This results in the formation of differently P-functionalized organophosphorus compounds.

Results and Discussion

Preparation of Phosphonio–Phosphanides

Upon adapting our published synthesis of 2[OTf]₄ to a larger scale (ca. 50 g, see Supporting Information S2.1), we opted to investigate the reaction of 2[OTf]₄ with Ph₃P (Scheme 1). Thus, isolated 2[OTf]₄ was treated with Ph₃P (4 equiv.) in CD₃CN. Subsequent analysis of an aliquot removed from the reaction mixture revealed a new AX spin system [δ(³¹P_A) = −168.6 ppm, δ(³¹P_X) = 31.3 ppm, ¹J(PP) = −519 Hz] in its ³¹P NMR spectrum assigned to phosphonio-phosphanide 1a⁺ (Figure 2). In addition, the spectrum showed signals assigned to the starting materials suggestive of equilibrium. In comparison with phosphanylidenephosphorane DmpP–PPh₃ (Dmp = 2,6-Mes₂C₆H₃, Table 1),¹⁷ the phosphanide (P_A) moiety in 1a⁺ is further upfield, and the magnitude of ¹J(PP) is significantly lower. In related triphosphenium cations {e.g., [(Ph₃P)₂P]⁺}¹⁸ the high field chemical shift and smaller coupling constant have been attributed to the dominance of the bis(ylidic) canonical structure.¹⁹ Further investigation of the reaction mixture by means of ³¹P–³¹P EXSY NMR experiments confirmed the underlying thermodynamic equilibrium (Figure 2). Notably, nucleophilic fragmentation of pentaphospholane (PhP)₅^4a,20 and the more electrophilic tetraphosphetane [(CF₃)P)₄]²¹ has been described previously, although stronger nucleophiles, that is, NHCs or Me₃P, respectively, are required.

Figure 2.

³¹P NMR spectrum of an aliquot of the reaction mixture of 2[OTf]₄ with four equivalents of Ph₃P in CD₃CN after 16 h (top, CD₃CN, 300 K) and zoom in of a ³¹P–³¹P-EXSY NMR spectrum (bottom, CD₃CN, 300 K) displaying spin polarization exchange between the phosphonio moiety in 1a⁺ and Ph₃P.

Table 1. Comparison of ³¹P NMR Chemical Shifts and Coupling Constants in 1a-e⁺ and Selected Related Compounds^9,17,23.

compound	PA (in ppm)	PX (in ppm)	1J(PP) (in Hz)
1a[OTf] (R = Ph)	–168.6	31.3	–519.0
1b[OTf] (R = Me)	–167.0	12.0	–472.0
1c[OTf] (R = Et)	–202.0	36.0	–492.0
1d[OTf] (R = Cy)	–208.8	38.1	–545.0
1e[OTf] (R = Ph2(CH2PPh2))	–164.3	38.1	–519.0
DmpP–PPh317	–138.8	25.2	–639.0
DmpP–PMe39	–114.7	–2.8	–582.0
[(Ph3P)2P][AlCl4]23	–174.0	30.0	–502.0

In an effort to prepare isolable phosphonio–phosphanides, the reaction of 2[OTf]₄ with more nucleophilic trialkyl-substituted tertiary phosphanes (R₃P: R = Me, Et, and Cy) was conducted in CH₃CN solution. The complete formation of the corresponding phosphonio–phosphanides 1b-d⁺ was observed after 4–16 h, and they could be isolated by precipitation with Et₂O in excellent yields as their triflate salts (88–92%, Scheme 1). Their respective ³¹P NMR spectra show the expected characteristic AX spin systems (see Table 1), in accordance with reported values for the related phosphanylidenephosphoranes ArP–PMe₃^9,10 and with the expected group contribution effects.^18,19,22 Alternatively, 1b-d[OTf] can be synthesized directly from the reduction of 3[OTf] using an excess of R₃P (Scheme S2, Figure S6); however, isolation is best achieved using the procedure described above. Vapor diffusion of Et₂O into saturated CH₃CN solutions of 1b-d[OTf] at −30 °C afforded colorless crystals suitable for single crystal X-ray analysis. The molecular structures of 1b[OTf] and 1d[OTf] are shown in Figure 3 and that of 1c[OTf] is shown in the Figure S14.

Figure 3.

Molecular structures of phosphonio–phosphanides 1b,d⁺ in 1b,d[OTf]; hydrogen atoms and anions are omitted for clarity, and thermal ellipsoids are displayed at 50% probability; selected bond lengths (Å) and angles(°): for 1b⁺: P1–P2 2.1162(4), P1–C1 1.8306(13), C1–P1–P2 97.98(4); 1c⁺ (Supporting Information, Figure S14): P1–P2 2.1195(4), P1–C1 1.8299(11), C1–P1–P2 99.29(4); 1d⁺: P1–P2 2.1446(4), P1–C1 1.8280(11), C1–P1–P2 105.93(4).

The observed P–P bond lengths [for 1b⁺: P1–P2 2.1162(4) Å, 1c⁺: P1–P2 2.1195(4) Å, and 1d⁺: P1–P2 2.1446(4) Å] match values for related triphosphenium cations^19,23 and range between a typical P–P single²⁴ and P=P double bond.²⁵ This shortening has previously been attributed to result from ylidic-type negative hyperconjugation between the lone pairs at the phosphanide moiety and the σ*(P–R) orbitals.¹⁹ Phosphonio–phosphanides 1b-d[OTf] can be stored indefinitely under an inert atmosphere at ambient temperature, whereas neutral derivatives of phosphanylidenephosphoranes ArP–PMe₃ have a tendency to decompose with respect to the formation of (ArP)_n (n = 2,3) under concomitant release of PMe₃.^9,26 In an effort to rationalize this apparent high stability, density functional theory (DFT) calculations were performed on the 1a⁺, 1b⁺, and 1c⁺ cations using CH₃CN as solvent (details are provided in the Supporting Information). As the energy of the HOMO–LUMO gap in 1a⁺ (E_gap = 2.472 eV; 1b⁺: R = Me: E_gap = 2.791 eV; 1c⁺: R = Et: E_gap = 2.760 eV) is still slightly higher than in DmpP–PMe₃ (E_gap = 2.443 eV), even larger HOMO–LUMO gaps can be achieved through the introduction of alkyl substituents at the phosphonio moiety (Table S9, Figure S124).

Notably, reaction of 2[OTf]₄ with ditopic phosphane 1,1-bis(diphenylphosphino)methane (dppm) leads to the formation of 1e[OTf] (Table 1) instead of the corresponding bis(phosphonio-phosphanide). Changing the ditopic phosphane to bis(diphenylphosphino)ethane (dppe) gives rise to a mixture of previously reported cyclic triphosphenium cation [(Ph₂PC₂H₄PPh₂)P]⁺ and di(imidazoliumyl)phosphanide [(L_C)₂P]⁺ as evidenced by means of ³¹P NMR spectroscopy (Scheme S3 and Figure S8).^18,23,27

Phosphonio–Phosphanides as Cationic Phosphinidene Transfer Agents

We continued to investigated the ability of compounds 1a-d[OTf] to transfer [L_C–P]⁺ in phospha-Wittig-type reactions. An initial effort to treat 1b[OTf] with 4-methoxybenzaldehyde resulted in encouraging ³¹P NMR spectra (see Supporting Information S2.14) of the reaction mixture, showing a small downfield signal at 178.3 ppm along with resonances assigned to free Me₃P (δ = −61.5 ppm) and Me₃PO (δ = 36.2 ppm). We speculated that the downfield signal observed was consistent with that anticipated for an unprecedented cationic phosphaalkene {i.e., [L_CP=CH(C₆H₄OMe)]⁺}. However, the conversion to phosphaalkene was very low (<5%), thus we turned our attention to more reactive thiocarbonyls. The latter can typically be accessed directly by thionation of the respective ketone, for example, via conversion with H₂S, P₄S₁₀, or Lawesson’s reagent.²⁸

For our following studies, compound 1a⁺ was selected to investigate its [L_C–P]⁺ transfer capability, as it holds the greatest synthetic value compared with 1b-d[OTf], owing to its ease of handling and the comparatively low cost of its starting material PPh₃ compared with the other alkyl-substituted tertiary phosphines. When treated with equimolar amounts of selected thioketones in CH₃CN, in situ generated 1a⁺ completely converts into the respective phosphane sulfide R₃PS and cationic phosphaalkenes 5a-e⁺ within 16 h at room temperature (Scheme 2), as evidenced by ³¹P NMR spectroscopy. The resonances of 5a-e[OTf] in CD₃CN (Figure 4) are significantly upfield shifted relative to neutral phosphaalkenes [e.g., MesP=CPh₂: δ(³¹P) = 233 ppm].²⁹ For heteroleptic 5d[OTf], both configurational diastereomers (E/Z) are observed in a near 1:1 ratio. The formation of (+)-camphor-derived 5e[OTf] requires heating of the reaction mixture to 80 °C for 3 h in a microwave reactor. The title compounds can be isolated as analytically pure solids as their triflate salts in very good to excellent yields by precipitation from the respective reaction mixture by addition of Et₂O (63–91%, Figure 4).

Scheme 2. Reactions of 1a,d⁺ with Thiocarbonyls Yield Phosphaalkenes 5a-f[OTf] (R = Aryl, Alkyl; R′ = Aryl, Alkyl, OMe) or Phosphanides 6a-d[OTf] (R = H, Aryl, NR₂^″; R′ = NR₂^″).

Reagents and conditions: (i) for 5a-d[OTf] and 6a,b[OTf]: −R₃PS, CH₃CN, rt, 16 h, 63–91%; for 5e[OTf] and 6c,d[OTf]: −R₃PS, CH₃CN, 80 °C, 3 h, 76–77%; 5a,d,e[OTf] and 6a-d[OTf] were prepared using in situ generated 1a⁺, 5b-c[OTf] were prepared using isolated 1d[OTf].

Figure 4.

Synthesized phosphaalkenes 5a-f[OTf] and phosphanides 6a-d[OTf] (left); molecular structure of phosphaalkenes 5a⁺ in 5a[OTf] and phosphanide 6a⁺ in 6a[OTf] (right); hydrogen atoms and anions are omitted for clarity, and thermal ellipsoids are displayed at 50% probability; selected bond lengths (Å) and angles (°): for 5a⁺: P1–C1 1.707(3), P1–C2 1.834(33), C1–P1–C2 104.55(16); 6a⁺: P1–C1 1.7417(13), P1–C2 1.8382(12), C1–P1–C2 93.77(6).

A second set of cationic phosphinidene transfer reactions were explored by treating in situ generated 1a⁺ with thioamides [R(NMe₂)C=S (R = H, Ph, NMe₂)] and L_C=S. In each case, analysis of the reaction mixtures by means of ³¹P NMR spectroscopy showed only a signal assigned to Ph₃PS (δ = 42.4 ppm) along with a new singlet resonance {δ(³¹P) = −8.8 ppm (br), R = H; 7.9 ppm, R = Ph; −60.2 ppm (br), R = NMe₂; −124.6 ppm, cf. known [(L_C)₂P]⁺³⁰}. Remarkably, each were shifted considerably upfield compared to those of 5a-e⁺. A similar trend to higher field shifts is observed in the ³¹P NMR spectra of inversely polarized phosphaalkenes bearing C-amino substituents when compared to conventional phosphaalkenes.^4e,31 Given this apparent higher shielding/increased electron density at the P atoms, the products were formulated with the cationic phosphanide canonical form (i.e., 6a-d⁺ in Figure 4) rather than cationic phosphaalkene (cf. 5⁺).

A supporting trend for this observation was found in the molecular structures of 5a-f[OTf] and 6a-c[OTf] (5a[OTf] and 6a[OTf] in Figure 4; 5b-f[OTf] and 6b,c[OTf] in Supporting Information). The P=C bond length in cationic phosphaalkene 5a[OTf] [P=C 1.707(3) Å, Figure 4] is only slightly elongated compared to the related MesP=CPh₂ [P=C 1.692(3) Å].³² Likewise, the P=C bonds of 5b,d⁺ [1.700(4), 1.702(4) Å, respectively] are in the range typical of phosphaalkenes. In contrast, the camphor-substituted 5e⁺ has a shorter P=C bond length [1.675(4) Å], presumably due to reduced delocalization of the P=C bond. The introduction of donating amino groups leads to significant elongation of the P–C bond [5c⁺: 1.7247(19) Å, 6a⁺: 1.7417(13) Å, 6c⁺: 1.7838(11) Å], consistent with increased contribution of the phosphanide canonical form, and comparable to values for reported inversely polarized phosphaalkenes.^4e,8a,20,33 Generally, the L_C–P=C bond angles are more acute for those formulated as phosphanides [C1–P1–C2 (°) = 93.77(6), 6a⁺; 99.30(15), 6b⁺; 100.81(5), 6c⁺] when compared to those formulated a phosphaalkenes [C1–P1–C2 (°) = 104.55(16), 5a⁺; 104.68(19), 5b⁺; 103.86(9), 5c⁺; 104.56(16), 5d⁺; 99.3(3), 5e⁺]. The reaction with O-methyl benzothioate proceeds likewise but results in multiple products, as evidenced by the ³¹P NMR spectra of the reaction mixture (Figure S74). Single crystal analysis of some crystalline material obtained by vapor diffusion of Et₂O into the reaction mixture, however, confirms the formation of methoxy-substituted phosphaalkene 5f[OTf] [δ(³¹P) = 53.0 ppm, see Supporting Information S2.25].

In general, Wittig-type conversions are thought to proceed via open-chain betaine-type structures or four-membered oxaphosphetanes as intermediates.^14,34 One report proposes an oxadiphosphetane intermediate for the related phospha-Wittig-Horner reaction.³⁵ Next to this, detailed investigations on the mechanism of the phospha-Wittig reaction specifically are scarce. Therefore, we employed DFT calculations [RI-BP86-D3/def2-TZVP (acetonitrile)] to gain further insight into the reaction of 1a⁺ with thiobenzophenone (Figure 5). The reaction profile reveals an overall exergonic transformation (−13.0 kcal/mol) that is initiated by the formation of a supramolecular complex INT-1 (see Figure S126), which is 13.3 kcal/mol more stable than the isolated reactants and possibly results from interaction of the lone pairs at the phosphanide moiety of 1a⁺ with the π*-orbital of the C=S double bond. This pre-organization is followed by a [L_C–P]⁺ transfer onto the C=S double bond via TS-1 with concomitant release of Ph₃P to give the three-membered thiophosphirane³⁶INT-2, which is more stable than TS-1 by 5.8 kcal/mol. In an attempt to identify the formation of INT-2 during the course of the reaction, we monitored the reaction between isolated 1d[OTf] (in favor of 1a⁺ to rule out possible [L_C–P]⁺ transfer from 2[OTf]₄ present in the reaction mixture of in situ generated 1a⁺) and thiobenzophenone by means of time-resolved ³¹P NMR spectroscopy. The spectrum of an aliquot of the reaction mixture after 10 min at room temperature reveals the formation of a new singlet resonance at δ(³¹P) = −89.3 ppm (Figure S1), which is within the margin of error³⁷ for the calculated ³¹P NMR shift of INT-2 [δ(³¹P) = −78.4 ppm]. We therefore assign this resonance to thiophosphirane 7a⁺. Notably, the formation of thiophosphirane 7a⁺ can also be observed in other transformations (see Figures S2, S4, and S7), as evidenced by its characteristic chemical shift in the ³¹P NMR spectra.

Figure 5.

Reaction profile for the conversion of 1a⁺ to 5a⁺ at the RI-BP86-D3/def2-TZVP (acetonitrile) level of theory; optimized geometries of the transition states (TS) with distances (red) in Å. Y axis shows ΔG in kcal/mol.

The formation of phosphanides 6a-d⁺, exemplified for the conversion of 1a⁺ with tetramethylthiourea, can be described by the same mechanism, although higher energy barriers are calculated (Figure S125). However, the corresponding thiophosphiranes 8c⁺ could not be observed spectroscopically.

Following the initially reported protocols for the phospha-Wittig reaction,^9,38 phosphaalkenes 5a-e⁺ and phosphanides 6a-c⁺ can also be synthesized using one-pot reactions of dichlorophosphane 3[OTf] with thiocarbonyls in the presence of Ph₃P and Zn, yet isolation cannot always be achieved satisfyingly (see Supporting Information S2.15).

Cationic Disphosphiranes from [L_C–P]⁺ Transfer to Phosphaalkenes

We further explored the [L_C–P]⁺ transfer reactivity of 1a⁺ toward phosphaalkenes,³⁹ which are isolobal to thioketones. Indeed, reacting 2[OTf]₄ with a phosphaalkene 9(29) (4 equiv) or 1,2-diphosphetane 10(40) (2 equiv) in the presence of catalytic Ph₃P (0.1 equiv) afforded diphosphiranes 11a,b[OTf] in very good or excellent yield (83 and 90%, respectively, Scheme 3). ³¹P NMR spectroscopic investigations on the isolated compounds showed the expected AB spin systems (11a[OTf]: δ(³¹P_A) = −127.7 ppm, δ(³¹P_B) = −100.8 ppm, ¹J(PP) = 146 Hz, 11b[OTf]: δ(³¹P_A) = −139.5 ppm, δ(³¹P_B) = −107.9 ppm, ¹J(PP) = 133 Hz).

Scheme 3. Synthesis of Diphosphiranes 11a,b [OTf] (R = Mes, Ph) via [L_C–P]⁺ Transfer from 1a⁺ onto Phosphaalkenes.

Reagents and conditions: (i) +0.1 Ph₃P, +0.25 2[OTf]₄, CH₃CN, rt, 2 h, 83%; (ii) +0.1 Ph₃P, +0.5 2[OTf]₄, CH₃CN, 60 °C, 2 h, 90%.

The observed high field-shifted resonances for both phosphorus nuclei are characteristic for phosphorus-containing three-membered ring {e.g., [(L_C)₃P₃][OTf]₃, (tBuP)₃}.^16,41 The molecular structures of 11a-b⁺ in 11a[OTf] (Figure 6) and 11b[OTf]·o-C₆H₄F₂ (Figure S86) reveal shortened P–P bond lengths [11a⁺: 2.1905(7) Å, 11b⁺: 2.1817(4) Å] and acute P1–C1–P2 bond angles [11a⁺: 70.50(9)°, 11b⁺: 70.82(15)°] in the range of other diphosphiranes⁴² and diphosphiranium cations [RP(R(CH₂^tBu)P)((^tBu)HC)]⁺ (R = ^tBu, Ad).⁴³

Figure 6.

Molecular structure of 11a⁺ in 11a[OTf]; hydrogen atoms and the anion are omitted for clarity, and thermal ellipsoids are displayed at 50% probability selected bond lengths (Å) and angles (°): P1–C1 1.908(2), P1–P2 2.1905(7), P2–C1 1.887(2), P2–C2 1.852(2), P1–C1–P2 70.50(9), C1–P1–P2 54.30(7), P1–P2–C2 109.68(7), C1–P2–C2 106.65(9).

Reactivity of Cationic Phosphaalkenes

Motivated by the successful isolation of 5a-e[OTf], we further probed the reactivity of 5a[OTf], as a model compound for cationic phosphaalkenes, due to its structural similarity to the well-studied MesP=CPh₂, toward typical conversions of phosphaalkenes (Scheme 4). Treating 5a[OTf] with [Pd(PPh₃)₄] or [Pt(PPh₃)₃] gave the metallaphosphiranes 12a[OTf] and 12b[OTf], respectively, under concomitant release of Ph₃P as evidenced by ³¹P NMR spectroscopy. The η²-coordination of the phosphaalkenes was indicated by a strong high field shift [12a⁺: δ(³¹P_A) = −5.6 ppm (br), 12b⁺: δ(³¹P_A) = −55.3 ppm] compared to the resonance of 5a[OTf] and the modest coupling constant to the Pt atom in 12b[OTf] [¹J(P_APt) = 564 Hz].⁴⁴ The ¹⁹⁵Pt NMR spectrum of 12b[OTf] showed the expected doublet of doublet of doublet resonance at δ(¹⁹⁵Pt) = −4822 ppm [¹J(PtP) = 3571 Hz, ¹J(PtP) = 3207 Hz, and ¹J(PtP) = 563 Hz, Figure S102].

Scheme 4. Reactions of Phosphaalkenes 5a[OTf] with Low Oxidation State Transition Metal Complexes [M(Ph₃P)_n] (M = Pd: n = 4, M = Pt: n = 3) toward Metallaphosphiranes 12a,b[OTf], with [Fe₂(CO)₉] toward Iron Complex 13[OTf] and with 3,4,5,6-Tetrachloro-1,2-benzoquinone (14) toward 15[OTf].

Reagents and conditions: (i) +[M(PPh₃)_n], −(n – 2) Ph₃P, C₆H₅F (12a[OTf]), toluene (12b[OTf]), rt, 1–4 h, 90–97%; (ii) +[Fe₂(CO)₉], −[Fe(CO)₅], THF, rt, 16 h, 86%; (iii) +14, C₆H₅F, rt, 1 h, 85%.

In general, the chemical shifts for 12a,b⁺ are comparable to the reported values for some related η²-diphosphene complexes^16,45 and the η²-phosphaalkene complex [Pt(Ph₃P)₂(η²-MesP=CPh₂)].⁴⁶ Notably, the latter phosphaalkene complex was structurally characterized as the η¹-complex with the η²-complex only being observed by ³¹P NMR spectroscopy at −70 °C in solution. Remarkably, the ³¹P NMR spectra of solutions 12a-b[OTf] in toluene-d₈ did not show evidence for the formation of η¹-complex upon heating to 100 °C. The latter might be a result of an increased π-acceptor ability in 12a,b⁺ due to the presence of the imidazoliumyl-substituent, thereby favoring the η²-coordination mode.⁴⁷ After workup, analytically pure red 12a[OTf] and yellow 12b[OTf] were obtained in excellent yields (90 and 97%, respectively).

Subsequently, single crystals of each were obtained and characterized by X-ray crystallography (Figures 7 and S96). Both complexes show the expected trigonal core, including the P–C bond of the [L_CP=CPh₂]⁺ ligand and the metal centers. Therein, the P–M–C bond angles are relatively acute [12a⁺ (M = Pd): 45.89(12)–46.46(9)°, 12b⁺ (M = Pt): 47.93(5)°]. In line with the η²-coordination mode, the P=C bond in both metallaphosphiranes is elongated to the extent of a P–C single bond [12a⁺: 1.782(4)–1.787(5) Å, 12b⁺: 1.8260(18) Å] presumably due to electron-backdonation from the metal centers into the π*-orbitals of the P=C bond.⁴⁸ Similar structural features have been observed for other transition-metal complexes involving η²-phosphaalkene or η⁴-phosphabutadiene ligands.⁴⁹

Figure 7.

Molecular structures of metallaphosphirane 12b⁺ in 12b[OTf], iron complex 13⁺ in 13[OTf] (left) and of 15⁺ in 15[OTf]·0.5C₆H₅F·n-pentane (right); hydrogen atoms and anions are omitted for clarity and thermal ellipsoids are displayed at 50% probability; selected bond lengths (Å) and angles(°): 12b⁺ (Pt): P1–C1 1.8260(18), Pt–P1 2.3127(5), Pt–C1 2.1718(16), Pt···P1–C1 2.0418(10), P1–Pt–C1 47.93(5); 13⁺: P–C1 1.6846(18), P–Fe 2.1697(5), C–P–C 106.51(8), ⌀C≡Oax.: 1.1405, ⌀C≡Oeq.: 1.1435; 15⁺: P–C1 1.897(3), P–O1 1.657(2), C1–O2 1.458(3); C1–P–C2 104.01(12)°.

We further reacted phosphaalkene 5a[OTf] (1 equiv) with [Fe₂(CO)₉] (1 equiv) in THF, which led to a red precipitate after 16 h at room temperature. Analysis of a CD₃CN solution of the red product by means of ³¹P NMR spectroscopy showed only one singlet resonance at δ(³¹P) = 154.5 ppm. The slight downfield shift compared to that for 5a[OTf] [δ(³¹P) = 152.8 ppm] suggests η¹-phosphaalkene complex 13[OTf] (Scheme 4). Crystallographic analysis of single crystals confirmed that the phosphaalkene ligand binds in a η¹-fashion in the equatorial position of the Fe⁰ center (Figure 7). The P–Fe–CO_eq [119.62(6)° and 122.44(7)°] and P–Fe–CO_ax angles [88.29(6)° and 89.69(6)°] are typical of [FeL(CO)₄] complexes, where L is a π-acceptor.⁵⁰ As a result, the bonding parameters of the phosphaalkenes ligand are only slightly affected compared to uncoordinated 5a⁺.⁵¹

For instance, the P=C bond in 13[OTf] is marginally shortened [1.6846(18) Å versus 5a⁺: 1.703(3) Å] and the C–P–C bond angle is widened [106.51(8)° versus 5a⁺: 104.55(16)°], which may indicate a strengthening of the double bond character. Compound 13[OTf] could be isolated in 86% yield. The IR stretching frequencies of the CO ligands are found at 2073, 2013, 1993, and 1965 cm^–1, which renders the ligand properties of 5a⁺ similar to those observed in phosphites, according to Tolman analysis.⁵²

To evaluate the potential of the P=C double bond in 5a[OTf] to be involved in cycloaddition reactions, we performed its conversion with 3,4,5,6-tetrachloro-1,2-benzoquinone (14). Upon dropwise addition of the red solution of 14 in C₆H₅F to yellow 5a[OTf], a colorless reaction mixture is obtained immediately. X-ray analysis of single crystals obtained by vapor diffusion of n-pentane into the reaction mixture confirms 15[OTf] as the product resulting from a [4 + 2] cycloaddition reaction (Figure 7). While the C1–P–C2 bond angle is barely affected [104.01(12)°], the P–C bond length is significantly elongated [1.897(3) Å] as a result of the conversion, and the P atom takes on a trigonal pyramidal geometry. The loss of the double bond character is consistent with the high field shift in the ³¹P NMR spectrum of 15⁺ [δ(³¹P) = 110.3 ppm], which has been observed in the reaction of related phosphaalkenes with the same and other ortho-quinones.⁵³

Replacement Reactions of Imidazoliumyl Substituent L_C

We further assumed that, due to the cationic charge in synthesized 5a-f⁺ and 6a-c⁺, their interaction with nucleophiles may lead to substitution at the P atom. Based on this, we hypothesized that the reaction of 5a[OTf] with MesMgBr could provide a practical route to obtaining MesP=CPh₂ (16), a significant monomer for the creation of P-containing polymers⁵⁴ that is typically synthesized via the phospha-Peterson reaction.⁵⁵ As confirmed by ³¹P NMR spectroscopy, treating the cationic phosphaalkene 5a[OTf] with 1 equiv of MesMgBr in THF at −78 °C resulted in its complete conversion to MesP=CPh₂ (16) within 30 min (Figure 8a).

Figure 8.

(a) Substitution reactions of the cationic imidazoliumyl substituent in 5a[OTf] using Grignard reagents; reagents and conditions: (i) +MesMgBr (1 M in THF), −0.5 [Mg(OTf)₂(THF)₄], −0.5 [MgBr₂(L_C)₂], THF, −78 °C, 30 min, 93%; (ii) +MeMgBr (1 M in ⁿBu₂O), THF, −78 °C, 30 min, quenched with 2 M HCl in Et₂O, 49%; (b) reactions of 5a[OTf] and 6c[OTf] with KNPh₂; reagents and conditions: (iii) +KNPh₂, THF, rt, 30 min, quenched with Me₃SiOTf, 90% (for 18), 80% (for 19); (c) molecular structures of 1,3-diphosphetane 17 and aminophosphaalkene 18; hydrogen atoms are omitted for clarity, and thermal ellipsoids are displayed at 50% probability; selected bond lengths (Å) and angles(°): for 17: P1–C1 1.8373(12), P1–C2 1.9085(11), P1–C4 1.9096(12), C1–P1–C2 103.44(5), C2–P1–C4 88.30(5), P1–C2–P2 91.70(5), C1–P1–P2–C3 180.00(9); 18: P–C1 1.726(2), P–N 1.7132(17), C1–P–N 103.28(9).

The product was conveniently obtained in 93% yield after extraction with n-hexane and subsequent recrystallization. One of the side products of the reaction was identified crystallographically as [MgBr₂(L_C)₂].⁵⁶

The successful conversion of 5a[OTf] to MesP=CPh₂ prompted us to expand our investigation to screening reactions with other nucleophiles that would result in the substitution of the imidazoliumyl substituent. Without adequate steric protection at the P atom, phosphaalkenes have a tendency to dimerize into either head-to-head (1,2-diphosphetanes) or head-to-tail (1,3-diphosphetane) dimers.^40,55,57 Consequently, the ³¹P NMR spectrum of an aliquot removed from the reaction mixture of 5a[OTf] with MeMgBr at −78 °C after quenching with HCl and warming to room temperature shows two new major resonances at δ(³¹P) = 35.6 ppm (70% integral ratio) and δ(³¹P) = −34.8 ppm (18% integral ratio), which we assigned to the head-to-tail and head-to-head dimers of MeP=CPh₂, respectively. 1,3-Diphosphetane (17) was isolated from the reaction mixture in 49% yield (Figure 8a) and structurally characterized by single crystal X-ray analysis (Figure 8c). The nearly square planar P₂C₂ core of 17 features P–C bond distances of approximately 1.909 Å and bond angles C2–P1–C4 and P1–C2–P2 of 88.30(5)° and 91.70(5)°, respectively. The two methyl substituents are arranged in a trans position, forming a dihedral angle C1–P1–P2–C3 of 180.00(9)°.⁵⁷

In some cases, thermodynamic equilibria between phosphaalkenes and their dimers have been reported.^{40,57a,57c,57e} Remarkably, no meaningful change within the ¹H and ³¹P NMR spectra of isolated 17 in toluene-d₈ was observed upon stepwise heating to 80 °C.

In a separate experiment, we investigated the reaction of two model compounds, 5a[OTf] and 6c[OTf], which are representative of cationic phosphaalkenes and phosphanides, respectively, toward KNPh₂. The reaction was carried out by adding 1 equiv of KNPh₂ to solutions of either 5a[OTf] or 6c[OTf] in THF at ambient temperature. After 30 min, the formation of free ^Me/iPrNHC was observed, as confirmed by ¹H and ¹³C NMR spectroscopic analysis of a sample removed from the reaction mixture. The ³¹P NMR spectra displayed low-field shifted resonances in both cases, indicating the exchange of the L_C-substituent with an amino group, resulting in the formation of two new examples of aminophosphaalkenes [18: δ(³¹P) = 231.0 ppm; and 19: δ(³¹P) = 106.1 ppm; Figure 8b]. Since both resulting compounds had similar solubility to free ^Me/iPrNHC, 1 equiv of Me₃SiOTf was added to the respective reaction mixture, resulting in the formation of [L_CSiMe₃][OTf], which is a precursor in the synthesis of 3[OTf].⁵⁸ Subsequently, 18 and 19 were isolated via extraction with n-hexane, with yields of 90 and 80%, respectively. Single crystals of both compounds were obtained through recrystallization from n-hexane or n-pentane and subjected to X-ray analysis.

In the molecular structures of 18 [P–C1 1.7132(17) Å, Figure 8 bottom], the P–C1 bond length is slightly elongated compared to a typical P=C double bond. This elongation is attributed to the donating effect of the P-amino substituent. The P–C1 bond length in C-amino substituted 19 [P–C1 1.754(1) Å, Figure S123] is even further elongated. The P–N bonds in both compounds [18: P–N 1.726(2) Å; 19: P–N 1.7501(12) Å] are comparable with other structurally related compounds.⁵⁹

Conclusions

In summary, we showed the synthesis of cationic imidazoliumyl(phosphonio)-phosphanides 1a-d⁺ via the nucleophilic fragmentation of tetracationic tetraphosphetane 2[OTf]₄ with tertiary phosphanes R₃P (R = Ph, Me, Et, Cy). We tested their ability to undergo the hitherto unknown transfer of a cationic phosphinidene, that is, [L_C–P]⁺. Employing in situ generated 1a⁺ or isolated 1d[OTf] in phospha-Wittig-type reactions with thiocarbonyls allowed the isolation and characterization of a series of novel cationic phosphaalkenes 5a-f⁺ as well as phosphanides 6a-d⁺ bearing a wide variety of substituents as their triflate salts. As evidenced spectroscopically and by DFT calculations [RI-BP86-D3/def2-TZVP (acetonitrile) level of theory], the mechanism of the formation of phosphaalkenes proceeds via the intermediary three-membered thiophosphiranes as a result of a [L_C–P]⁺ transfer from 1a⁺ onto the C=S double bond. Although calculations show a similar pathway for the formation of phosphanides, energy barriers are found to be significantly higher. Furthermore, when in situ generated 1a⁺ is reacted with phosphaalkenes that are isolobal to thioketones, [L_C–P]⁺ transfer is also observed, leading to the isolation of heteroleptic diphosphiranes 11a,b[OTf].

In order to evaluate the reactivity of the formed cationic phosphaalkenes, we subjected 5a[OTf] to reactions with low oxidation state transition-metal complexes [Pd(PPh₃)₄], [Pt(PPh₃)₃], and [Fe₂(CO)₉]. While the conversion with the latter gave iron complex 13[OTf], in which the phosphaalkenes are in an equatorial position and have a η¹-coordination mode, metallaphosphiranes 12a,b[OTf] are formed in the reaction with the former two complexes, including 5a⁺ in a η²-coordination mode. Lastly, we showed the potential to use the P=C double bond in 5a[OTf] for cycloaddition reactions by its conversion with ortho-quinone 14 giving 15[OTf].

We furthermore exemplified the possibility of exchanging the transferred L_C-substituent in 5a[OTf] by reacting it with MesMgBr. This reaction allowed for the convenient and high-yield synthesis of MesP=CPh₂, a compound that is typically obtained through reactions involving malodorous primary phosphines and silylphosphines. Moreover, our work enables access to unprecedented 1,3-diphosphetane 17 as well as aminophosphaalkenes 18 and 19 through the conversions of 5a[OTf] and 6c[OTf] with MeMgBr or KNPh₂. These conversions demonstrate the versatility of [L_C–P]⁺ as a P₁ building block.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacs.3c02256.

• Experimental and characterization details for all new compounds, including spectroscopic data and NMR spectra, molecular structures, crystallographic data, and computational details (PDF)

The authors declare no competing financial interest.