Synthesis and reactivity of cyclo-tetra(stibinophosphonium) tetracations: redox and coordination chemistry of phosphine–antimony complexes† †Electronic supplementary information (ESI) available. See DOI: 10.1039/c4sc03939d Click here for additional data file. Click here for additional data file.

Reactions of trialkylphosphines with antimony(iii) triflates yield catena-antimony(i) cations revealing a new reductive elimination/oxidative coupling reaction for P–Sb coordination complexes.


Introduction
Phosphines are prototypical ligands in the coordination chemistry of d-block metals. While the chemistry of p-block elements is primarily dened by covalent bonding as typied by organic frameworks, an array of phosphine adducts has also been characterized for main group element acceptors. [1][2][3][4][5] Beyond their versatile ligand properties as neutral, two-electron donors (L-type), 6 phosphines also exhibit redox reactivity within the coordination sphere of an acceptor. For example, reductive elimination of tetraorgano-or halotriorganophosphonium cations (Scheme 1a), 7 and oxidative addition of PR-X bonds (Scheme 1b), 8 or P-R bonds (Scheme 1c) 9 are all known pathways of tertiary phosphine activation in transition metal chemistry. One report 10 hints at the reductive elimination of a diphosphonium dication from a phosphine-metal complex (Scheme 1d). In this instance, spectroscopic studies indicate that the reaction of excess PMe 3 6 ], respectively. 10 However, neither the high oxidation state reactants nor the reduced products have been structurally veried and three different 31 P NMR chemical shis were ascribed to [Me 3 PPMe 3 ] 2+ (depending upon the counterion: +65.0 ppm, +46.3 ppm, or +27.8 ppm). As reductive elimination is observed for both a transition metal (Cu II ) and a main group metal (Tl III ) acceptor, phosphine activation may be broadly applicable to complexes exhibiting a mismatch between hard (high oxidation state/charge) acceptors and so phosphine donors. Indeed phosphines are considered poor donors for hard acceptors and coordination to such centres generally requires enforcement by chelate or pincer ligands. [11][12][13] As part of a systematic evolution of p-block element phosphine complexes, we have sought derivatives featuring multiplycharged, hard acceptors and now report evidence of a new phosphine ligand activation pathway in the coordination sphere of polycationic Sb III centres. Specically, reductive elimination of diphosphonium dications (Scheme 1e) from trialkylphosphine complexes of Sb III has been demonstrated, comprehensively dening a fundamental P-P bond forming redox process. The reduction products are the unusual cyclotetra(stibinophosphonium) tetracations [10(R)] 4+ , representing a new catena-homocyclic framework. 14 Examples of cationic homocycles for p-block metalloids are limited to unsupported selenium and tellurium dications 15 and heavily substituted silicon 16 or germanium 17 monocations. For antimony, a number of acyclic catenated monocations ([1] 1+ and [2] 1+ ) [18][19][20] and dications ([3] 2+ , [4] 2+ , [5] 2+ and [6] 2+ ) 19,[21][22][23] have recently been isolated (Chart 1), but generally on small scales, precluding further reactivity studies of these interesting species. Enabled by a rational and large scale synthetic protocol for cations [10(R)] 4+ , we now report the reaction chemistry of the prototypical derivative, [10(Me)] 4+ , debuting the coordination chemistry of a new catenaelement framework.

Results and discussion
Reactions of PR 3 with FSb(OTf) 2 and Sb(OTf) 3 Combinations of FSb(OTf) 2 or Sb(OTf) 3 with PR 3 (R ¼ Me, Et, Pr, or Bu) in MeCN solvent at the optimized stoichiometries given in Scheme 2 have been investigated. The 31 P, 13 C, 19 4+ . Given the high molecular charge, these values are expectedly smaller than the sum of the van der Waals radii ( P r,vdW ¼ 3.61Å) 24 but nevertheless signicantly longer than the sum of the single bond covalent radii ( P r,cov ¼ 2.05Å) 25 for the two elements. For [10(Me)] 4+ , a gas-phase optimization 26 of the cation at the MP2 level in the absence of the triate anions produced a geometry that is essentially identically to that observed experimentally, and we therefore infer that the anion contacts do not distort the structural features to a measurable extent.
The reactions in Scheme 2a represent a two electron reduction of each antimony(III) center and collectively, an eight electron reductive coupling of four antimony centers to form derivatives of [10(R)] 4+ . In Scheme 2a, eight of the twelve equivalents of phosphine are involved in the redox process, being oxidatively coupled to give four diphosphonium cations, [11(R)] 2+ , 27,28,31 and the remaining four equivalents represent ligands on the reduced antimony(I) centers of [10(R)] 4+ . Scheme 2b describes a similar redox process that involves formation of [11(R)] 2+ as transients, which are converted to the corresponding uorophosphonium cations, [12(R)] 1+ , in the presence of the uoride ion, as envisaged in the mechanism outlined in Scheme 3 (le). The key feature in both processes is reductive elimination of a diphosphonium unit from a hard, tricationic Sb III centre to give a so, monocationic Sb I centre, representing a novel mode of phosphine ligand activation in the coordination sphere of metals (Scheme 1e). 31 P NMR spectra (Fig. 2) of reaction mixtures containing PR 3 and FSb(OTf) 2 in a 2 : 1 stoichiometry show a broad doublet in the +20 to +40 ppm range and the signal due to the free phosphine (À60 to À20 ppm) is not observed.  32 suggesting that the free chloride ion is sequestered in an  4+ does not occur catalytically in the chloride system because the reaction is arrested upon formation of [11(Me)] 2+ , which is the spectroscopically detected oxidation product. Generation of free phosphine from diphosphonium, the turnover limiting step, does not take place (Scheme 3, right). In contrast, no diphosphonium is detected in reactions involving the uoroantimony complexes [7(R)] 2+ (Scheme 3, le), where, due to nucleophilic attack by uoride anions on [11(R)] 2+ , only the uorophosphoniums [12(R)] 1+ are detected as the oxidation product and the formation of [10(R)] 4+ occurs catalytically in the presence of free PR 3 . Differences in the reactivity of homologous Sb-X (X ¼ Cl, F) complexes towards Lewis acids have been noted previously. 33 Solution  Table 2, with evidence for the assignments discussed below. It has not been possible to detect or isolate derivatives of [9(R)] 1+ . Attempts to trap these cations, or radical intermediates arising from one-electron processes, in the presence of a twenty-fold excess of 2,3-dimethyl-1,3-butadiene were unsuccessful.   Table 2 2 have both been isolated as analytically pure substances and spectroscopically characterized, we were unable to obtain X-ray quality crystals. Moreover, to the best of our knowledge, there are no known examples of 2 J PF coupling constants through an antimony centre for direct comparison with our assigned NMR data. For this reason, we prepared and isolated the analogous [(dmpe) SbF][OTf] 2 , [13][OTf] 2 , from an equimolar mixture of 1,2-bis-(dimethylphosphino)ethane (dmpe) and FSb(OTf) 2 in MeCN. The solid state structure of [13][OTf] 2 , as determined by X-ray crystallography, shows a dimeric arrangement with the cations bridged by O-S-O contacts from the triate anions, and additional interactions with two non-bridging triate anions, as shown in Fig. S3 (ESI). † The pyramidal geometry at Sb in the cation is retained in solution, as demonstrated by the two nonequivalent methyl group resonances in the 13 C (6.1 and 7.2 ppm) and 1 H NMR (1.86 and 2.10 ppm) spectra. Crucially, the expected 2 J PF coupling was unambiguously observed (Fig. S4, ESI †) in signals due to [13] 2+ , and the chemical shi and coupling constants are comparable to those assigned to derivatives of [7(R)] 2+ (Table 2).
It was not possible to isolate salts of [8(R)] 3+ due to their high reactivity, consistent with their disproportionation to [11(R)] 2+ and [10(R)] 4+ in solution as proposed above. The 31 P NMR signals assigned to derivatives of [8(R)] 3+ are singlets and broadened (Dn 1/2 ¼ 90-500 Hz), presumably due to a combination of the quadrupolar antimony nuclides 40 24 with each contact appearing trans to a P-Sb bond, illustrating a triple displacement of triate anions from Sb(OTf) 3 partially eclipsed conformation is observed between the six ethyl groups. In contrast to a previously assigned 31 (Fig. 4) represents the rst structural characterization of a trialkyluorophosphonium salt, and involves three hydrogen bonds with the triate anion in addition to one weak contact [3.301(2)Å] between a triate oxygen atom and the phosphorus atom, which is marginally shorter than S r,vdw for the two elements (3.320Å). 24 The O-P-F angle generated by this contact is 177.34 (8) , representing adjustment of the D 3h structure of Me 3 PF 2 . 43 The 31 P NMR chemical shis for species in Table 2 over a 150 ppm range but within each class of cations, generally decrease in the order d( Notably, the chemical shis of the free phosphines (range of 30 ppm) also show the same order, providing additional support for the proposed assignments (Fig. S6, ESI †).
Attempts to isolate cations [7(R)] 2+ or [10(R)] 4+ with bulky phosphines such as P i Pr 3 were unsuccessful. A 31 P NMR assay of the reaction mixture containing P i Pr 3 and FSb(OTf) 2 in a 2 : 1 ratio displayed numerous uorine containing products as indicated by the observation of spin system with P-F couplings but no pure compounds could be isolated. A 3 : 1 mixture of P i Pr 3 with Sb(OTf) 3 also gave a complex mixture of products at room temperature which could not be separated. Deprotonation of MeCN solvent was observed upon reuxing the reaction mixture for short periods or stirring at room temperature for 16 hours. We conclude that steric bulk at the a-carbon of the phosphine hinders the coordination required for clean transformation of bis-phosphine cations [ 4 on a scale up to 10 g. Consistent with the exquisite sensitivity of these compounds towards hydrolysis and oxidation, particularly in solution, the key determinant of purity and reactions yields is the rigorous drying and deoxygenation of the solvent and careful application of dynamic vacuum (ca. 10 À1 mbar) in the latter stages of the reaction to avoid free phosphine-catalyzed decomposition (vide infra).

Thermolysis and photolysis of [10(Me)][OTf] 4
The four-membered ring of [10(Me)] 4+ contains four of the six Sb-Sb bonds required to make neutral, tetrahedral Sb 4 , which is directly analogous to P 4 and As 4 . Moreover, [10(Me)] 4+ also contains four phosphine ligands which may be susceptible to further reductive elimination of two diphosphonium dications, [11(Me)] 2+ , to yield neutral Sb 4 . While P 4 and As 4 are well characterized, Sb 4 has not been isolated as a bulk solid, and only one solid-state structural determination has been made using a scanning tunnelling microscope to characterize a thin lm of Sb 4 under ultrahigh-vacuum conditions. 44 In this context, we envisioned the thermal or photochemical decomposition of [10(Me)][OTf] 4 as a route to bulk solid Sb 4 .
A sample of solid [10(Me)][OTf] 4 (yellow-colored) heated under argon at 120 C for 16 hours turned black, consistent with the formation of elemental antimony (Scheme 5). A CD 3 CN extract of the black product showed 31 P, 1 H and 13 C NMR signals corresponding exclusively to [11(Me)] 2+ as the sole oxidation product. A Raman spectrum of the black solid (Fig. S7, ESI †) matched that of the amorphous a-phase (110 cm À1 , 150 cm À1 ) 45 of antimony rather than the reported Raman spectrum of tetrahedral Sb 4 in argon matrix (138 cm À1 , 179 cm À1 , 242 cm À1 ). 46 Identical results were obtained when heating was carried out in the dark, under vacuum, or in solution (toluene). Irradiating solid [10(Me)][OTf] 4 or as a solution in MeCN at 256 nm for 3 hours at room temperature had no measurable effect. It should be noted that in the gas phase tetrahedral Sb 4 is the preferred allotrope of the element up to 1050 K. 47 It is possible that despite its gaseous stability, tetrahedral Sb 4 is thermodynamically unstable with respect to its amorphous phases in the condensed state, preventing its isolation as a solid and is, in this context, analogous to tetrahedral As 4 (yellow arsenic) which spontaneously decomposes to a hexagonal allotrope (grey arsenic, a-As) at room temperature. 48 The thermolysis described above must be carried out in rigorously dried glassware, the surface of which has been treated with Me 3 SiCl to silanize terminal -OH groups. Samples heated without prior passivation of glassware produced elemental antimony and [11(Me)][OTf] 2 , but also showed resonances due to [Me 3 PH] 1+ and a singlet at +115.6 ppm in the 31 P { 1 H} NMR spectrum (CD 3 CN) of the reaction mixture, consistent with formation of [Me 3 POPMe 3 ] 2+ . This assignment is supported by an independent synthesis from a 2 : 1 mixture of Me 3 PO and triic anhydride, using a well-established protocol for these reagents. 49 We interpret the formation of these byproducts as being due to the reaction of the extremely moisture sensitive [10(Me)][OTf] 4 with surface hydroxyl groups in nonsilanized glassware. . Consistent with this formulation, the 31 P-1 H coupled NMR spectrum of the reaction involving CyPH 2 shows (Fig. 5a) both 1 J PP and 1 J HP couplings for the phosphinic signal centered at À83.6 ppm. The P a -P b (H b )Cy connectivity is also conrmed in the 1 H NMR spectrum of the reaction mixture (Fig. S8, ESI †), where H b resonates at +3.65 ppm exhibiting 1 J HbPb , 2 J HbPa , and 3 J HbHg couplings, the last of these arising from coupling to the ipso proton (H g ) of the cyclohexyl ring. The methyl protons (H a ) around P a also show the expected 2 J HaPa and 3 J HaPb couplings, indicating a P-P bond. Finally, a two-dimensional 31 P/ 1 H HSQC (Fig. 5b) spectrum, which was optimized to show one-bond couplings, shows coupling between H b and P b but no coupling involving H b and P a . The corollary two-dimensional HMBC experiment (Fig. 5c), optimized to exclude one-bond couplings, shows coupling between H b and P a , but no coupling involving H b and P b . Despite numerous attempts, it was not possible to separate [Me 3

PP(H)Cy][OTf] from [Me 3 PH][OTf]
, precluding elemental analysis or structural determination by X-ray diffraction. Nevertheless, to the best of our knowledge this is the rst spectroscopic detection of an H-phosphinophosphonium cation.
The formation of [Me 3 PH] 1+ and the phosphinophosphonium salts is understood in broad terms as a metathesis step followed by a reductive elimination step as outlined in Scheme 6. We speculate that coordination of Cy 2 3+ . This trication can undergo rapid intramolecular reductive elimination of the rst equivalent of the phosphinophosphonium cation to give dication [B] 2+ . A second round of coordination, deprotonation and reductive elimination completes the reduction of antimony to its elemental form and furnishes the observed distribution of products. Unfortunately, the partially reduced species were not observed and appear to be eeting intermediates. Nevertheless, formation of [Me 3 PP(H)R] 1+ from reactions involving primary phosphines (Scheme 7) is consistent with the proposed mechanism, although it is unclear why the second deprotonation does not occur to yield the corresponding dication [(Me 3 P) 2 PR] 2+ . As before, Raman analysis of the black precipitate matches the amorphous a-phase of metallic antimony rather than pyramidal Sb 4 Table 3.

Reaction of [10(Me)][OTf] 4 with PMe 3
The 31 P NMR spectrum of a reaction mixture containing 15 mol% of PMe 3 and [10(Me)][OTf] 4 shows slow disappearance of the signal due to the latter and evolution of broadened signals due to [11(Me)] 2+ and free PMe 3 . Concomitantly, a mirror of antimony is deposited in the reaction vessel. Within 12 hours at 298 K, there is no evidence of [10(Me)] 4+ , while signals due to [11(Me)] 2+ and free PMe 3 persist, consistent with complete decomposition of the tetracation, catalyzed by PMe 3 . The proposed mechanisms (Scheme 9) involve nucleophilic attack by the added phosphine at either the antimony or the to FSb(OTf) 2 is too high, a dark orange solution is obtained which rapidly deposits elemental antimony (see note in Experimental section). However, if a dynamic vacuum is applied to the dark orange solution to remove the volatile PMe 3 (b.p. ¼ 38 C), the solution maintains a yellow colour, leading to the formation of [10(Me)][OTf] 4 . Moreover reactions with Lewis bases that displace PMe 3 must be carried out with explicit steps to remove the liberated phosphine in order to avoid decomposition (vide infra).

Reaction of [10(Me)][OTf] 4 with [Li][nacnac (dipp) ]
In contrast to the sterically unhindered and neutral base PMe 3 , a bulky and anionic base is expected to yield products arising from ligand substitution rather than from addition. Consistently, the 31 (Fig. 7) shows three phosphine ligands and the rare g-coordination mode for the nacnac substituent, 57 which, to the best of our knowledge, has not been observed for haloantimony centres bound to this substituent. 58 Heteroleptic substitution is very rare in antimony homocycles 59    To assess whether or not bonding via the g carbon of nacnac is a general feature of antimony compounds and because nacnac functionalized antimony centers are rare in the literature, we also prepared (nacnac)Sb(OTf) 2 by salt metathesis between an equimolar mixture of in situ generated Sb(OTf) 3 and [Li][nacnac (dipp) ]. Upon removal of LiOTf, the compound was isolated as a pure substance and comprehensively characterized. The molecular structure of (nacnac)Sb(OTf) 2 , determined by X-ray diffraction, shows a see-saw geometry around antimony with two strongly-interacting triate anions in axial positions (Fig. S10, ESI †). In contrast to g-coordination observed for [15(Me)] 3+ , N,N 0 -chelation is observed for (nacnac) Sb(OTf) 2 , and we attribute the difference in bonding modes to the different steric environments around antimony in the two compounds, rather than intrinsic features of the nacnac-Sb interaction.
Interestingly, the 19 F resonances for the two triate CF 3 groups in (nacnac)Sb(OTf) 2 are different (À78.3 and À78.4  ppm), implying a rigid ring system with non-equivalent positions above and below the plane of the ring. Consistently, the isopropyl substituents show two unique resonances for the C ipso protons. Furthermore, there is restricted rotation around the C ipso -C phenyl bond giving rise to four unique signals for the methyl groups in the 1 H NMR spectrum of the compound. We speculate that this is due to solution-phase persistence of the weak hydrogen bonding interactions between the nitrogen atoms and the isopropyl C ipso protons, detected as short contacts in the solid state molecular structure (Fig. S10, ESI †).

Reaction of [10(Me)][OTf] 4 with dmap
The reaction of [10(Me)][OTf] 4 with 4-dimethylaminopyridine (dmap) has been examined by 31 P NMR (Fig. 8) and shows displacement of one phosphine ligand by dmap (Scheme 11). It was not possible to isolate the resulting products. Following ltration of the reaction mixture (black suspension), the yellow-green ltrate shows the expected AX 2 spin system (triplet at +66.9 ppm, doublet at +42.5 ppm, 3   Neutral catena-antimony rings are known to participate in ringring equilibria unless bulky substituents or dilute solutions are employed. For instance, solutions of hexaphenylcyclohexastibine (Ph 6 Sb 6 ) equilibrate to give a mixture of four-, ve-, and six-membered rings suggesting labile Sb-Sb bonds. 61 To assess the possibility of preparing heteroleptic derivatives of [ 4 were combined in a 1 : 1 stoichiometry. The 31 P{ 1 H} NMR spectrum (Fig. 9c) (24 Hz). Collectively, these data enable a tentative assignment of the spectral features observed in Fig. 9.

Conclusions
The reductive elimination of diphosphonium dications [11(R)] 2+ from trialkylphosphine complexes of highly electrophilic antimony(III) centres is reported. The reduced antimony(I) fragments cyclize into frameworks identied as cyclo-tetra-(stibinophosphonium) tetracations, [10(R)] 4+ . As outlined in Scheme 3, a phosphine catalyzed mechanism is proposed for uoroantimony complexes, and isolation or spectroscopic characterization of key mechanistic intermediates is presented. The scope of this reductive assembly is dependent upon the steric bulk of the phosphine employed as demonstrated by nonproductive reactions involving P i Pr 3 . Formation of cyclic (R-Pn) n or [L-Pn] n (n+) species (R ¼ aryl group, L ¼ alkylphosphine ligand, E ¼ heavy pnictogen) appears to be the general fate of low-valent (R-Pn) or [L-Pn] 1+ monomers, respectively. A multi-gram scale synthesis for the triate salt of a prototypical cyclo-tetra-(stibinophosphonium) tetracation, [10(Me)][OTf] 4 , has enabled reactivity studies that are summarized in Scheme 14.
In broad terms, the reactivity of catena-antimony(I) cation [10(Me)] 4+ is directed by two features: (i) high charge concentration, and (ii) the presence of strongly polarized P-Sb bonds. The former explains the electrophilicity of cation [10(Me)] 4+ , its thermolysis to extrude [11(Me)] 2+ , and the observed facility for reductive elimination to yield elemental antimony (Scheme 14, reactions a-f). The signicant polarization of the P-Sb bonds enables activation of a wide spectrum of bonds with the unusual outcome of yielding the same products via reaction with oppositely polarized substrates (e.g. P-Cl and P-H containing reagents) (Scheme 14, reactions c-f). This unique feature has led to the spectroscopic detection of the an Hphosphino-phosphonium cation, [Me 3 PP(H)Cy] 1+ , examples of which have not been reported previously. The high P-Sb bond polarization also supports a coordinate bonding model, consistent with ligand displacement reactivity demonstrated for cation [10(Me)] 4+ (Scheme 14, reactions g-i). Ligand displacement has permitted functionalization of the fourmembered Sb ring with substituents such as [nacnac] 1À or dmap (transiently). A heteroleptic phosphine substitution pattern around the Sb 4 is feasible, but multiple isomers are observed on a relatively shallow potential energy surface hindering the isolation of a single derivative.
Within the broader context of phosphines as ubiquitous ligands in coordination chemistry, evidence of a novel ligand activation pathway has been presented and the associated reactants and products characterized. Taken together with previous, albeit less denitive, detection of such reactivity, 10,42 the observation of this reductive elimination pathway conrms that these prototypical ligands can behave simultaneously as reducing agents and stabilizing ligands, a feature that may be generally applicable for phosphine complexes of highly electrophilic acceptors across the periodic table. Diversication of this synthetic protocol may therefore provide access to more extensively catenated systems for antimony as well as other elements. As demonstrated for [10(Me)] 4+ , a unique and rich reaction chemistry can be expected, in addition to the potential for valuable emergent properties such as s-bond conjugation and cooperative catalysis due to metal catenation.