Ylide‐Stabilized Phosphenium Cations: Impact of the Substitution Pattern on the Coordination Chemistry

Abstract Although N‐heterocyclic phosphenium (NHP) cations have received considerable research interest due to their application in organocatalysis, including asymmetric synthesis, phosphenium cations with other substitution patterns have hardly been explored. Herein, the preparation of a series of ylide‐substituted cations of type [YPR]+ (with Y=Ph3PC(Ph), R=Ph, Cy or Y) and their structural and coordination properties are reported. Although the diylide‐substituted cation forms spontaneous from the chlorophosphine precursor, the monoylidylphosphenium ions required the addition of a halide‐abstraction reagent. The molecular structures of the cations reflected the different degrees of electron donation from the ylide to the phosphorus center depending on the second substituent. Molecular orbital analysis confirmed the stronger donor properties of the ylide systems compared to NHPs with the mono‐ylide substituted cations featuring a more pronounced electrophilicity. This was mirrored by the reaction of the cations towards gold chloride, in which only the diylide‐substituted cation [Y2P]+ formed the expected LAuCl]+ complex, while the monoylide‐substituted compounds reacted to the chlorophosphine ligands by transfer of the chloride from gold to the phosphorus center. These results demonstrate the tunability of ylide‐functionalized phosphorus cations, which should allow for further applications in coordination chemistry in the future.


Introduction
Phosphenium cations are compounds of type R 2 P + and hence valence isoelectronic to carbenes. [1] However, owing to their positive charge phosphenium ions usually exhibit a weaker σdonor strength but increased π-acceptor abilities. [2] These electronic properties make phosphenium cations attractive ligands for metal complexation which might be used complementarily to stable singlet carbenes, such as N-heterocyclic carbenes (NHCs), which are in general strong donor ligands. Despite many advances made in the past years in phosphenium cation chemistry, their properties and coordination chemistry are far less explored than those of their carbon analogues. The limitation is mostly due to the higher reactivity connected with their cationic nature. Thus, only few phosphenium cations have been isolated and thoroughly explored. The most extensively studied systems are -analogous to carbenes [3] -N-heterocyclic phosphenium cations (NHPs). Since the first reports on NHPs in 1970s [4] various derivatives [5] and metal complexes have been reported [6,7] and more recently also applied in catalysis including asymmetric organocatalysis. [8] Apart from amino groups also other π-donor substituents have been employed to stabilize these carbene-like species. In the 1990s Schmidpeter pioneered the use of ylide groups in phosphenium cation chemistry. He realized that depending on the second substituent R, ylide-substituted chlorophosphines of type Y(R)PCl (Y = ylide) can spontaneously dissociate into the phosphenium cations. [9,10] Diylide-substituted cations are particularly stable due to the strong π-donation from both ylide groups resulting in an efficient charge delocalization in the CÀ PÀ C linkage. In case of the cyclic cation A this delocalization even results in a partial phosphide character at the central P atom (A 3 , Figure 1). [11] Nonetheless, these systems exhibit phosphenium cation reactivity and for example react with nucleophiles such as organolithium compounds at the central phosphorus atom. [12] However, the coordination chemistry of ylidephosphenium cations remains almost unexplored. To the best of our knowledge, only the coordination chemistry of bisphosphonio-isophosphindolides of type A has been explored. [13] In the course of our research program on ylidic compounds we became interested in the stabilization of reactive main group compounds. [14] With the isolation of metalated ylides as versatile reagents for ylide-functionalization we were able to generate the first diylide-substituted boron cations [15] and heavier tetrylenes. [16] Furthermore, the introduction of ylide substituents in phosphines led to a remarkable increase of their donor capacity and hence to excellent performance in gold [17] and palladium catalysis. [18] This led us to re-investigate the chemistry of ylide-substituted phosphenium cations to evaluate their potential as ligands in transition metal chemistry.

Synthesis of the phosphenium cations
We began our studies with the synthesis of the diylidylphosphenium cation 2 a reported by Schmidpeter. [11a] To further explore the impact of the substitution pattern also the isolation of the monoylide cations 4 was targeted. With only one πdonating ylide substituent, we expected these phosphenium cations to be more reactive and stronger acceptor ligands. All three ligands were prepared via a chlorosilane elimination protocol by reaction of the silyl-substituted ylide 1 with PCl 3 and the dichlorophosphines (PhPCl 2 or CyPCl 2 ), respectively (Scheme 1, see the Supporting Information). Whereas the reaction with PCl 3 to form the diylide-substituted compound directly led to the formation of cation 2 by chloride elimination, the monoylidyl compounds formed stable chlorophosphines 3a and 3b. In case of the diylide-substituted compound chloride exchange to the BF 4 salt was conducted to exclude any weak interaction of the counter-anion with the phosphorus cation. The chlorophosphines 3 were isolated as colorless solids in 85 and 81 % yield and are characterized by two doublets in the 31 P { 1 H} NMR spectrum appearing at approx. 24 ppm for the PPh 3 group and at 132 ppm (for 3a) and at 160 ppm (for 3b), respectively. In the molecular structures ( Figure 2), both chlor-ophosphines feature remarkably different PÀ Cl bond lengths. As a consequence of the stronger donor properties of the cyclohexyl substituent, 3b (2.256(1) A) shows a 55 pm longer PÀ Cl bond than 3a (2.191(1) A). Such varying PÀ X distances have been discussed in the past as snapshots of the pending dissociation. [19] With the chlorophosphines in hand, we next addressed their conversion into the corresponding phosphenium cations. Chloride abstraction from 3 was accomplished with AlCl 3 , thus delivering the cations 4 as colorless solids in good yields (72 for 4a, 73 for 4b). [20] The cations 2 and 4 were characterized by multi-nuclear NMR spectroscopy, elemental and single-crystal X-ray diffraction analysis. 31 P{ 1 H} NMR spectroscopy shows that the cyclohexyl-substituted cation 4b features the most deshielded signal ( Figure 3, Table 1), which resonates at 403.1 ppm and hence is considerably downfield shifted compared to 4a, 2, NHPs and acyclic aminophosphenium cations. [21] This suggests that 4b exhibits the smallest HOMO-LUMO gap in this series of compounds (see below for discussion) and presumably is the most nucleophilic species with the strongest acceptor properties. Interestingly, the 2 J PP coupling constants Scheme 1. Synthesis of the phosphenium cations 2 and 4 from the silylsubstituted ylide 1.  increase from 4b to 4a and 2 and become significantly smaller compared to the chlorophosphine precursors. Overall, the mono-ylidylphosphenium cations exhibited are considerably higher reactivities and more readily decomposed in solution. However, in the solid state they could be stored for weeks at À 20°C.
Single crystals for 2b, 4a and 4b with BF 4 À or AlCl 4 À as counter-anions were obtained by layering DCM solutions of the cations with cyclohexane. [22] All structures are depicted in Figure 4, important structure parameters are listed in Table 1. The molecular structures clearly reflect the different stabilizing abilities of the second substituent at P2 and its impact on the π-donation from the ylide substituent. Due to the competing πinteraction by both ylide groups in 2b, the diylide cation features the longest C1À P2 distance of approx. 1.72 Å. The P2À C1 bonds in the monoylide phosphenium cations are considerably shorter, with the phenyl compound featuring the shortest distance of 1.673(5) Å. The reverse trend is found for the P1À C1 distances to the phosphonium group. With increasing π-bonding between the ylide and the phosphenium center the electrostatic attraction to the phosphonium group decreases. Consistent with the C1-P2 bonds, this results in the shortest C1-P1 bond for 2 and the longest in the phenyl-substituted compound, suggesting that the strongest ylide-P bond is found in cation 4a. It is noteworthy that the phenyl substituent in 4a is not or only little involved in π-interaction with the central phosphorus atom. The same is true for the phenyl groups in the ylide-backbone. While the two ylide groups in 2b form a fully delocalized π-system over the planar PÀ CÀ PÀ C-P linkage, the phenyl group in 4 a is oriented out of the P1À C1À P2 plane with dihedral angles of approx. 20°. Consequently, no changes in the CÀ C bond lengths are observed within the phenyl groups of 4a compared to that in 3a, which contrasts for example the quinoidal structure reported for the borinium cation Mes 2 B + . [23] It is also noteworthy, that in contrast to some ylide-substituted phosphines only one conformational isomer of 2 and 4 are observed in the solid-state and in solution. [17b] In these isomers, the lone pair at phosphorus is perpendicularly arranged to the lone pair at the ylidic carbon atom in order to minimize electronic repulsion and enable π-stabilisation of the empty p orbital at P2. Due to additional steric effects an anti-/zigzag arrangement is favored, which in case of 2 leads to the conformer with the smaller phenyl groups facing each other.
To shed light on the electronic structure and the observed spectroscopic properties of the synthesized cations we per-  (14) 1.7382 (14) 114.00 (8) (14), P2À C1 1.7216 (14), C26À P2 1.7141 (14), C26À P3 1.7566 (14) formed computational studies on the PW6B95D3/def2tzvp level of theory (see the Supporting Information for details). Calculations of the molecular orbitals show that the ylide-substituted phosphenium cations feature energetically high-lying LUMOs (LUMO = lowest occupied molecular orbital) and hence are less electrophilic/weaker acceptors than NHPs ( Figure 5). This agrees well with the stronger π-donor ability of the ylide compared to the amino substituent. Accordingly, the diylidyl cation 2 + is the by far weakest acceptor in this series of compounds. While the LUMO in all cations represents the empty p orbital at P2 and thus is always predominantly localized at the phosphenium center, the HOMO (except for 4b) displays a π-symmetric orbital delocalized over the substituents with a strong contribution by the ylidic carbon centers. In general, the energy of the molecular orbitals representing the lone pair at the phosphenium center (shown in blue in Figure 5) only vary slightly in the ylide-substituted cations but are again significantly higher than those of NHPs. This clearly demonstrates that ylide-substituted phosphenium ions are particularly strong donor, but weaker acceptor ligands. Overall, the spectroscopic and computational data clearly indicate that the diylide-substituted phosphenium cation 2b is electronically quite different from the monoylide congeners. It is considerably more nucleophilic as expressed by its resonance structure shown in Scheme 1 with a negative formal charge at phosphorus, whereas cations 4a and 4b retain a higher acceptor ability. Thus, the monoylide cations are best described by the two canonical structures 4' and 4'' (Scheme 1). Nonetheless, also these monoylide cations are considerably stronger donors than NHPs and thus should exhibit different donor properties.
Next, we turned our attention toward the coordination chemistry of the phosphenium cations. We first focused on the synthesis of gold complexes since complexes with cationic phosphorus ligands have shown catalytic activity. [24] Treatment of 2b with one equivalent of (tht)AuCl (tht = tetrahydrothiophene) led to the clean formation of the expected gold chloride complex [2b-AuCl]BF 4 (5b; Scheme 2). Using an excess of gold precursor and the chloride salt of the phosphenium ligand 2a led to the same complex, yet with AuCl 2 as counter anion (5a), which was found to weakly bind to the ylidic PÀ C bond in the solid state. A similar observation was made by Gudat with the cyclic diylide phosphenium cation A (see the Supporting Information for details). [13a] 5b is characterized by a highfield shift in the 31 P{ 1 H} NMR spectrum for P2 from 290.4 in 2b to 229.2 ppm in 5b. The molecular structure ( Figure 6) features a linear PÀ AuÀ Cl linkage with a PÀ Au bond length of 2.2127(6) A. This bond length is comparable to other phosphido-gold complexes. [25,13a] However, in contrast to previously reported complexes, the central phosphorus atom adopts a completely planar geometry with a sum of angles of 360°. The P2-C1 bond shortens from 1.718(1) Å to 1.692 (3) Å upon coordination to gold as a consequence of the further shift of electron density from the ylide groups to the phosphenium center.
Interestingly, the monoylide-substituted phosphenium cations 4 showed a different coordination behavior towards gold chloride as suggested by the different orbital energies. This was already indicated by the NMR spectroscopic data. Upon treatment of 4a and 4b, respectively, with one equivalent (tht)AuCl, the signal of the central phosphorus atom experienced a distinct high field shift by more than 250 ppm in the 31 P{ 1 H} NMR spectrum, giving rise to a signal at 93.3 ppm for 4a and 73.2 ppm for 4b, respectively. This shift is much more pronounced than in case of 2 b, suggesting considerable changes in the ligand structure. This was confirmed by XRD analysis of the complex formed with 4a, which revealed the formation of complex 6a (Scheme 3), in which a cationic gold center is coordinated by the chlorophosphine ligand 3a and tht. This structure matches well with the NMR spectroscopic observations. 6a is presumably formed from the expected AuCl complex by chloride shift from gold to phosphorus. Thus, this gold complex should also be accessible from a simple AuCl complex of the chlorophosphines precursors 3. Indeed, treatment of 3a and 3b, respectively, with (tht)AuCl delivered the corresponding gold chloride complexes 7, which could both be isolated in moderate yields of approx. 40 % and unambiguously characterized, including XRD analysis ( Figure 6 and the Supporting Information). Treatment of the gold chloride complexes with aluminium trichloride finally yielded the cationic complexes as judged by 31 P NMR spectroscopy (Scheme 3). Using an excess of AlCl 3 (or NaBAr F ) did not lead to the abstraction of the chloride at the phosphorus center in 6 to form the corresponding dication. The same holds true for the cationic gold complex 5b. Single crystals for 6a could be obtained by slow vapor diffusion of pentane into a saturated dichloromethane solution. In the crystal (Figure 6), the gold center adopts a linear geometry between the phosphine and tetrahydrothiophene ligand with an AuÀ P distance of 2.260(2) A. This bond length is slightly longer than the one found in 5b and also longer than the one in the AuCl precursor 7a. The P2À C1 bond shortens in the compound series 3a > 7a > 6a from 1.742(2) to 1.700(7) Å, thus accounting for an increasing shift of electron density from the ylidic carbon atom to the phosphorus (and metal) center. The C1-P1 distance increases in the same sequence of compounds.
An interesting combining structural feature of all gold complexes shown in Figure 6 are short distances between the gold center and one phenyl group of the phosphonium moiety. A survey (Table 2) of all structures revealed that the Au-C ipso distance is shorter in the cationic gold complex 6a (3.160 Å) than in the neutral AuCl complexes 7 which exhibits an AuÀ C ipso distance of 3.220 Å. Also, the cationic Au complex 5b with the diylidylphosphenium ligand shows a short distance of 3.132 Å. Such short distances have been attributed to attractive arene gold interactions and have often been discussed in the context of gold catalysis and their importance for stabilizing catalytically active cationic gold species. [26] Prominent examples are the Echavarren's catalysts with Buchwald's biarylphosphine JohnPhos. [27] Here, an arene Au distance of 3.04 Å was reported in the gold cation which is slightly shorter than the one in 6a thus indicating a weak attractive interaction. It is important to note that such interactions have also been suggested for gold complexes of ylide-substituted phosphines (YPhos) and thought to be one reason for their high catalytic efficiency. [17,18d] However, no cationic Au-YPhos complex has been isolated until today. Thus, 6a constitutes the first complex of its kind.

Scheme 3.
Synthesis of the cationic gold complexes 6 from the chlorophosphine ligands 3 or cations 4. Table 2. Bond length of the shortest Au-arene interaction in the discussed gold complexes.

Chemistry-A European Journal
Research Article doi.org /10.1002/chem.202104074 their reaction towards a series of late transition metal complexes. Although, the isolation of any reaction product revealed to be difficult due to the formation of further side products or the sensitivity of the formed complexes, NMR spectroscopic analyses of the reaction mixtures with [Pd-(cod)Cl 2 ] clearly indicated the formation of the corresponding chlorophosphines 3, for example with two doublets at δ = 78.4 and 17.4 ppm with a coupling constant of 9.2 Hz for 3a. In the case of [Rh(cod)Cl] 2 however, the formation of a phosphenium complex was observed as evidenced by a considerably downfield-shifted signal in the 31 P{ 1 H} NMR spectrum for the P(III) center (δ = 371.5 ppm, 1 J RhP = 163.5 Hz and 2 J PP = 52.5 Hz for 4b).
Although all attempts to isolate these complexes failed, single crystals were obtained from the crude reaction mixture of 4b in CD 2 Cl 2 , which confirmed the still intact RhÀ Cl bond (Figure 7). The RhÀ P bond amounts to 2.217(4) A and is thus longer compared with the rhodium phosphorus bond of 2.089(2) Å reported for an NHP-RhCl(PPh 3 ) 2 complex with a bis(alkylamido)naphthalene framework, [28] and in the range of RhÀ P bonds in phosphine complexes. [29] Overall, the marked difference between the coordination chemistry of [Y 2 P] + and the monoylide cations clearly shows that a certain stabilization of the cation is required for selective applications in transition metal chemistry. One ylide moiety as sole π-donating substituent seems to be insufficient so that future research endeavors should focus on the diylidyl or monoylide systems with a further π-donor substituent.

Conclusion
The synthesis of three phosphorus cations stabilized by two or one ylide moiety is reported. Due to the strong electrondonating ability of the ylide group, these formal phosphenium cations are considerably more electron-rich than Nheterocyclic phosphenium cations but can be tuned by variation of the second substituent bound to the phosphorus center. Thus, the bis(ylidyl)phosphenium cation Y 2 P + forms spontaneously from the corresponding chlorophosphine in solution, whereas the mono(ylidyl)phosphenium ions are only accessible by halide abstraction. The stronger electrophilic nature of the monoylide substituted compounds was also confirmed by computational studies and led to a different coordination chemistry compared to [Y 2 P] + . Although the bis(ylidyl) cation formed the expected cationic LAuCl complex, the monoylide systems reacted with (THT)AuCl by chloride transfer from gold to phosphorus, thus resulting in the formation of a cationic chlorophosphine gold complex. Crystallographic studies on the phosphenium cations and their metal complexes reflected the electron donation from the ylide to the phosphorus center. As such, the PÀ C distances in the ligand backbone changed depending on the second substituent at phosphorus and the presence of a metal center.
Overall, these studies clearly demonstrate that ylide groups efficiently stabilize cationic phosphorus species and allow a tuning of their properties by changing the molecular design. Although systematic studies on the impact of the substituents in the ylide backbone on the stability and reactivity of phosphenium cations are missing, it can be assumed that a further control of reactivity should be possible.

Crystallographic Details
Deposition Number