Effect of the Phosphine Steric and Electronic Profile on the Rh-Promoted Dehydrocoupling of Phosphine–Boranes

The electronic and steric effects in the stoichiometric dehydrocoupling of secondary and primary phosphine–boranes H3B·PR2H [R = 3,5-(CF3)2C6H3; p-(CF3)C6H4; p-(OMe)C6H4; adamantyl, Ad] and H3B·PCyH2 to form the metal-bound linear diboraphosphines H3B·PR2BH2·PR2H and H3B·PRHBH2·PRH2, respectively, are reported. Reaction of [Rh(L)(η6-FC6H5)][BArF4] [L = Ph2P(CH2)3PPh2, ArF = 3,5-(CF3)2C6H3] with 2 equiv of H3B·PR2H affords [Rh(L)(H)(σ,η-PR2BH3)(η1-H3B·PR2H)][BArF4]. These complexes undergo dehydrocoupling to give the diboraphosphine complexes [Rh(L)(H)(σ,η2-PR2·BH2PR2·BH3)][BArF4]. With electron-withdrawing groups on the phosphine–borane there is the parallel formation of the products of B–P cleavage, [Rh(L)(PR2H)2][BArF4], while with electron-donating groups no parallel product is formed. For the bulky, electron rich, H3B·P(Ad)2H no dehydrocoupling is observed, but an intermediate Rh(I) σ phosphine–borane complex is formed, [Rh(L){η2-H3B·P(Ad)2H}][BArF4], that undergoes B–P bond cleavage to give [Rh(L){η1-H3B·P(Ad)2H}{P(Ad)2H}][BArF4]. The relative rates of dehydrocoupling of H3B·PR2H (R = aryl) show that increasingly electron-withdrawing substituents result in faster dehydrocoupling, but also suffer from the formation of the parallel product resulting from P–B bond cleavage. H3B·PCyH2 undergoes a similar dehydrocoupling process, and gives a mixture of stereoisomers of the resulting metal-bound diboraphosphine that arise from activation of the prochiral P–H bonds, with one stereoisomer favored. This diastereomeric mixture may also be biased by use of a chiral phosphine ligand. The selectivity and efficiencies of resulting catalytic dehydrocoupling processes are also briefly discussed.


■ INTRODUCTION
The development of efficient catalytic methods for the formation of bonds between main group elements is of considerable interest for the continued development of main group chemistry. Such processes enable new discoveries to be made in the promising application areas that main group species are now occupying, such as high performance polymers, emissive materials, etch resists for lithography, and precursors to ceramic thin films or devices. 1−6 However, the development of this field lags substantially behind the advances made in catalytic C−C and C−X bond formation, for which there are now a myriad of efficient ways to promote such unions that are important for the construction of new molecules. Catalytic dehydrocoupling 5,7,8 of amine− and phosphine−boranes is one method that has emerged for the formation of B−N and B−P bonds, and development in the area has been spurred on by the potential for ammonia−borane to act as a hydrogen carrying vector. 9−11 In addition, polymeric materials that can arise from dehydropolymerization of primary analogues are also of significant interest as they are valence isoelectronic with technologically ubiquitous polyolefins. Although the metal catalyzed formation of polyaminoboranes has attracted recent attention, 12−18 catalytic routes to polyphosphinoboranes have also been known since 1999. 19 Perhaps the best example is that of the [Rh(COD) 2 ][OTf] catalyzed dehydrocoupling of secondary, H 3 B·PR 2 H, and primary, H 3 B·PRH 2 , phosphine− boranes to give oligomeric and polymeric materials (Scheme 1). 19 −21 In contrast to amine−borane dehydrocoupling, 8,10,15,22−24 the mechanism of catalytic dehydrocoupling of phosphine− boranes has received less attention. Although initial reports demonstrated that catalysis using [Rh(COD) 2 ][OTf] was a homogeneous process (i.e., not colloidal), 25 there has been only sporadic further work on elucidating the mechanistic details. 26−29 Progress has no doubt been slowed due to the fact that the reaction conditions reported for phosphine−borane dehydrocoupling often require melt conditions, thus making interrogation of the catalytic cycle problematic. Recently, we have reported that the Rh(I) complexes [Rh(P t Bu 2 H) 2 (η 6 -FC 6 H 5 )][BAr F 4 ], 30 and [Rh(L)(η 6 -FC 6 H 5 )][BAr F 4 ], 31 [L = Ph 2 P(CH 2 ) 3 PPh 2 ] are particularly well-suited to the study of the dehydrocoupling mechanism of secondary phosphine− boranes in solvents such as fluorobenzene; and on the basis of the observation of intermediates, kinetic studies, and H/D exchange experiments we have proposed a catalytic cycle for the dehydrocoupling of H 3 B·PR 2 H (R= Ph, t Bu; Scheme 2). For this cycle, intermediate species were isolated, but their structures could not be confirmed by X-ray crystallography. In particular for R = Ph, a β-B-agostic σ complex B, and the product of dehydrocoupling F, that is proposed to sit off cycle, could be isolated and spectroscopically characterized. Under stoichiometric conditions the observation that B transforms into F on gentle heating allowed for kinetic parameters to be determined that suggested that the rate-determining step(s) for dehydrocoupling were located within the transformations B to D. In solution phase the turnover limiting step for catalysis is proposed to be the displacement of the linear diboraphosphine product (i.e., F to A), although under the melt conditions used for efficient catalysis this may well be different. Further insight comes from the observations that for R = t Bu the barrier to dehydrocoupling is higher (70°C versus 25°C for reaction), P−H activation appears also to be a higher energy process, different intermediates (A and E) are observed, and the turnover limiting process in catalysis is now suggested to be the P−H activation/dehydrocoupling steps. Prior work has demonstrated a similar difference in relative rates of dehydrocoupling of secondary H 3 B·PR 2 H [R = p-(CF 3 )C 6 H 4 , Ph, t Bu, i Bu] and primary H 3 B·PRH 2 [R = Ph, t Bu, i Bu] phosphine−boranes using the [Rh(COD) 2 ][OTf] catalyst, and this was suggested to be due to a combination of steric and electronic (relative P−H bond strengths) factors, 21,32,33 although the mechanism of dehydrocoupling of phosphine− boranes using this catalyst is currently not known. 20,25,30 Interestingly, the related dehydrogenation of aryl amine− boranes shows that the activity of the N−H bond is such that spontaneous dehydrocoupling occurs in the absence of catalyst, with electron-withdrawing aryl groups [p-(CF 3 )C 6 H 4 ] undergoing faster reaction than electron-donating [p-(OMe)C 6 H 4 ]. 34 Very recent work has shown that paramagnetic Ti(III) centers might also be involved in dehydrocoupling of phosphine− and amine−boranes when using Cp 2 Ti-based catalysts, 35 while oligomerization of base-stabilized phosphino−boranes at Cp 2 Ti centers has been described. 29 Likely decomposition routes in Rh-systems for phosphine−borane dehydrocoupling to form bis(phosphine)boronium salts have also recently been discussed. 36 In this Article, we report an extension of our investigations into the mechanism of phosphine−borane dehydrocoupling using the {Rh(Ph 2 P(CH 2 ) 3 PPh 2 )} + fragment, by varying the electronic and steric profile of the secondary phosphine− boranes H 3  These studies provide insight into the determining role of the electronics and sterics of the phosphine−borane in the dehydrocoupling process, as well as providing as yet unreported examples of the solid-state structures of the intermediates related to the catalytic cycle. We also report for the first time the partial control of diastereoselectivity in dehydrocoupling of primary phosphine−boranes, that can additionally be biased by use of a chiral chelating phosphine on the rhodium center.
■ RESULTS AND DISCUSSION Phosphine−Borane and Diboraphosphine Starting Materials. A range of secondary phosphine−boranes with differing electronic and steric properties have been used in this study (1,2,3, and 4, Figure 1), which also provide comparison with the previously reported Ph, 6, and t Bu, 7, analogues. 31 The primary phosphine−borane 5 has also been used. 37 Compounds 2 33 and 3 38 are known adducts and offer electronwithdrawing and donating aryl groups, respectively. Bis-CF 3substituted 1 is a new complex and offers an alternative to 2. The synthesis of adamantyl-substituted phosphine, 4, an analogue of 7, has been reported in the patent literature. 39 Compared with the t Butyl group, adamantyl has a greater steric bulk due to its larger volume and rigid structure. 40,41 The new linear diboraphosphines, 10−13, have also been synthesized to aid in the identification of final dehydrocoupling products. Complexes 10−12 are synthesized by a Rh-catalyzed process from the corresponding phosphine−boranes, while primary phosphine containing 13 has been synthesized in good isolated yield (85%) by addition of [NBu 4 ][BH 4 ] to the bis-(phopshine)boronium [(CyH 2 P) 2 BH 2 ]Br. 36 Stoichiometric Dehydrocoupling of Secondary Phosphine−Boranes. Addition of 2 equiv of 1 to [Rh(L)(η 6 -FC 6 35 Hz] is assigned to the phosphorus atom trans to the coordinated phosphido ligand. In the 1 H NMR spectrum of 14 one broad, relative integral 3H, signal is observed at δ −0.78, indicative of a Rh···H 3 B σ interaction in which the B−H bonds are undergoing rapid site exchange on the NMR spectroscopic time scale between terminal and bridging sites. 42 A broad, relative integral 1H, resonance at δ −6.12 is assigned to a static β-B-agostic B−H interaction. Cooling of the solution to 0°C led to the resolution of this signal as doublet [J(PH) = 65 Hz], fully consistent with its trans disposition to a phosphine. The remaining BH(terminal) signals are not observed, and it is likely they are coincident with the {CH 2 } 3 signals. A sharper signal at δ −16.21, relative integral 1H, is assigned to a metal− hydride resonance, in which the coupling to both 103 Rh and 31 P   is clearly small and unresolved. The PH group is observed at δ 5.81 that collapses into a singlet in the 1 H{ 31 P} NMR spectrum.
The 11 B NMR spectrum shows a broad signal centered at δ −39.8, which is not shifted significantly from that of free phosphine−borane 1 (δ −42.0). This is assigned to a coincidence of the η 1 β-B−H···Rh agostic and σ Rh···H 3 B signals, as has been noted previously. 31,43 Complexes 15 and 16 have similar 1 H, 11 B, and 31 P NMR spectra, and thus we assign very similar structures. Crystals of complex 14 of suitable quality for analysis by Xray diffraction were obtained by layering of a 1,2-F 2 C 6 H 4 solution with pentane at −26°C. The structure of 14 in the solid-state ( Figure 2) is fully consistent with the structure deduced from the solution NMR spectroscopic data. The formally Rh(III) center adopts a pseudo-octahedral geometry, with the chelating phosphine ligand and the hydride located on one of the faces of the octahedron. Two of the three remaining coordination sites are occupied by a phosphine−borane unit that has undergone P−H activation, and is bound to the metal via a phosphido bond [Rh1−P3, 2.3045(10) Å] and a β-Bagostic bond [Rh1−B1, 2.515(4) Å]. The other phosphine− borane unit occupies the last coordination site via a σ η 1 -Rh··· H−B interaction. 42 All the hydrides (B−H and Rh−H) were located in the final difference map. The structure is in full accord with the solution NMR spectroscopic data, confirming the spectroscopic assignments that have been made previously. 31 β-B-agostic interactions are known, 35,44,45 and we have recently reported [Rh(κ 1 ,η-PPh 2 BH 2 ·PPh 3 )(PPh 3 ) 2 ]-[BAr F 4 ] in which a base-stabilized phosphine−borane adopts a β-B-agostic interaction with the Rh-center. 36 σ phosphine− boranes are also known, 42,46,47 and bimetallic complexes showing both B-agostic and σ borane coordination modes have been reported. 48 Compared to a Rh(I) complex that shows a bidentate η 2 -coordination mode for the σ borane, [Rh(P t Bu 2 H) 2 (η 2 -H 3 B·P t Bu 2 H)][BAr F 4 ], 30 the Rh···B distance for the η 1 -interaction in 14 is considerably longer [2.188(3) Å versus 2.740(4) Å, respectively], consistent with this different binding motif. Similar changes in M···B distance have been noted on moving between η 1 and η 2 coordination modes in chelating phosphine−boranes. 43 Complexes 14−16 undergo spontaneous dehydrocoupling (25°C) to form products of the general formula [Rh(L)H-(σ,η 2 -PR 2 ·BH 2 PR 2 ·BH 3 )][BAr F 4 ]: 17, R = 3,5-(CF 3 ) 2 C 6 H 3 ; 18, R = p-(CF 3 )C 6 F 4 ], 21 and 22, respectively, on the basis of NMR spectroscopic data. These complexes are formed in parallel to 17 and 18, as preformed 17 (vide inf ra) does not proceed to form 21. Complex 21 has been independently prepared by addition of two equivalents of HP((CF 3 ) 2 C 6 H 3 ) 2 to [Rh(L)(η 6 -FC 6 H 5 )][BAr F 4 ]. This mixture of products observed for the electronwithdrawing phosphine substituents (i.e., 1 and 2) contrasts with that found for when R = Ph 31 and p-(OMe)C 6 H 4 , which yield the dehydrocoupled (e.g., 19 and F, Scheme 2) product in essentially quantitative form (∼95% by 31 P{ 1 H} NMR spectroscopy). Complex 17 has been synthesized cleanly from direct addition of the preformed dehydrocoupled diboraphosphine product, 10, to [Rh(L)(η 6 -FC 6 H 5 )][BAr F 4 ], Scheme 5. It was from this reaction that material of 17 suitable for single crystal X-ray diffraction was obtained.    The NMR spectroscopic data for 17 are fully consistent with the solid-state structure being retained in solution and are also very similar to that reported for the analogous complex formed from the deydrocoupling of 6 (R = Ph). 31 The 31 P{ 1 H} NMR spectrum shows four different phosphorus environments. Two of these signals are well-resolved and show coupling to 103 Rh, δ 46.6 [J(RhP) 111 Hz] and δ 12.8 [J(RhP) 91 Hz], and are attributed to the chelating phosphine ligand. One of these signals (δ 12.8) also shows large 31 P− 31 P coupling [J(PP) 260 Hz] suggesting a trans position relative to the phosphido center. The other two environments are broad, typical of those observed when coupling to a quadrupolar boron center. For one of these trans J(PP) coupling is also observed. The 1 H NMR spectrum shows three different broad, relative integral 1H, environments assigned to the BH 3 moiety [δ −4.54, −1.20, and 4.37]. This indicates that the BH 3 unit is not undergoing exchange on the NMR spectroscopic time scale, as noted previously for similar η 2 -M···H 3 B systems. 31,43,50,52 The Rh−H signal is observed at δ −13.98 as a sharper signal, although this also shows unresolved coupling. The 11 B NMR spectrum shows two different environments [δ −27.1 and 0.21] for the two boron atoms present in the diboraphosphine, with the latter assigned to the η 2 -H 3 B unit on the basis of the large downfield shift from free ligand (Δδ = +36.8). 43 Spectroscopic data for complexes 18 and 19, that are produced by the direct dehydrocoupling route are similar, although for 18 this is also formed as a mixture with 22.
The dehydrocoupling reaction (i.e., 14 to 17) shows a dependence on the substituents on the phosphine. For electron-withdrawing aryl groups (e.g., p-CF 3 ), it is faster when compared with electron rich groups (i.e., p-OMe). Following these processes in situ using NMR spectroscopy demonstrated that these dehydrocoupling reactions follow a first order rate profile for the consumption of the starting material over at least three half-lives (see Supporting . That the parallel products 21 and 22 are formed in approximately equal ratio to the dehydrocoupled product (17, 18, respectively)) suggests that k 1 ≈ k 2 (Scheme 4). In addition to this parallel process, direct comparison of the rate constants is further complicated by the fact that 16 → 19 required heating to 35°C to make the reaction run over a convenient time scale for analysis by NMR spectroscopy. Nevertheless these relative rates reflect previous observations on the rate of catalytic dehydrocoupling when the electronics of a system are changed, in as much as electronwithdrawing groups promote the reaction. 21 Interestingly, for all the aryl complexes initial P−H activation to form a phosphido hydride complex (i.e., 14) is very rapid, occurring on time of mixing. This suggests that for aryl-substituted phosphine−boranes it is not initial P−H activation that is rate-determining for the dehydrocoupling event, as we have commented on for R = Ph. 31 In this study we suggested that B−H activation/reorganization in intermediates such as B (Scheme 2) prior to P−B bond formation might be the rate limiting process. 31 This might well be promoted by a weaker B−H bond, and calculations on analogous H 3 B·L (L = Lewis base) systems show that the B−H bond is considerably weaker when there are electron-withdrawing groups on the Lewis base. 53 However, we cannot rule out that the relative P−H bond strengths in intermediates such as 14 also might play a role, or that there is a change in the rate determining step on changing the phosphine−borane ligand, as the intimate details of the mechanism leading to P−B formation still remain to be resolved. The observation that for an electron-withdrawing phosphine there is a significant proportion of parallel product formed that results from P−B bond cleavage is consistent with the weakening of the P−B bond with increasingly electronwithdrawing aryl substiutents. 8,54 P−B bond cleavage has been noted previously in σ phosphine−borane complexes to give either simple adducts 47 or further reaction to yield bis-(phosphine)boronium salts. 30 Prior to the formation of the parallel product 21 (R = 3,5-(CF 3 ) 2 C 6 H 3 ) an intermediate is observed that has been characterized by 1 F 4 ] 20, i.e., a complex that sits directly between 14 and 21 by loss of one "BH 3 " fragment (Scheme 6). Complex 20 results from P−B bond cleavage, formally of the σ-H 3 B·PR 2 H ligand, to afford a complex with a β-B-agostic interaction from a phosphide borane ligand (as for 14) and a simple PR 2 H ligand trans to a hydride. Complex 20 was not isolated in pure form, being observed alongside 14 and the final products 17/21. However, after 2 h reaction a significant proportion of 20 is present (∼20% by 31 P NMR spectroscopy), allowing for its identification aided by comparison with the NMR spectroscopic data for 14 (Supporting Information). In particular four environments are observed in the 31 P NMR spectrum, with only one of these broadened significantly by coupling to quadrupolar boron. This signal also shows a large, mutual, trans J(PP) coupling with another phosphine environment. In the high-field region of the 1 F 4 ], 23, which was characterized in situ by NMR spectroscopy. This complex could not be isolated as it undergoes further reaction, by P−B bond cleavage at room temperature, to form 24 (Scheme 7). Addition of 1 equiv of 4 resulted in a final mixture of 24 and [Rh(L)(η 6 -FC 6 H 5 )][BAr F 4 ].
The 1 H NMR spectrum of complex 23 immediately after preparation shows a broad, relative integral 3H, signal at δ −1.36 characteristic of a σ-bound phosphine−borane that is undergoing site exchange between the coordinated and uncoordinated B−H environments. 42 Two signals are observed in the 31 P{ 1 H} NMR spectrum, in a 2:1 ratio at δ 35.1 [J(RhP) 167 Hz] and δ 30.1 (br). Over time (1 h), complex 23 disappears to be replaced by a new complex that has been characterized by NMR spectroscopy and a solid-state X-ray diffraction experiment as [Rh(L)(PHR 2 )(η 1 −H 3 B·PHR 2 )]-[BAr F 4 ] (24, R = adamantyl). Figure 4 shows the structure of the cation present in 24 in the solid-state. A Rh(I) center is in a pseudo-square-planar geometry with a chelating ligand, and the other two coordination sites are occupied by P(adamantyl) 2 H and a η 1 -H 3 B·P(adamantyl) 2 H [Rh···B, 2.457(7) Å] ligands, respectively. The BH and PH hydrogen atoms were located in the final difference map. The solution NMR spectroscopic data for 24 are fully consistent with the solid-state structure, and in particular the trans disposition of P1 and P3, and the η 1 -H 3 B· PR 2 H ligand.
A significant amount of P−B bond cleavage product is thus observed for both electron poor aryl phosphine−boranes (e.g., 14) and very bulky electron rich phosphine−boranes (e.g., 24), but not the electron rich aryl phosphine 3 or H 3 B·PPh 2 H (6). 31 Interestingly we have recently reported that for H 3 B·P t Bu 2 H P−B bond cleavage is also observed during dehydrocoupling catalysis being accompanied by a further dehydrocoupling step, through which bis(phosphine)boronium salts are ultimately formed. 30,36 Similar complexes can be prepared on rhodium using H 3 B·PPh 2 H and PPh 3 under stoichiometeric conditions. 36 One suggested mechanism for this process is the reaction of a short-lived phosphino−borane (or its masked equivalent) with coordinated phosphine, not dissimilar to the mechanism suggested for the formation of diaminoboranes from amine−boranes and amines catalyzed by alkaline earth catalysts. 55 Complexes 20 and 24 serve as models for intermediates in this process [Rh(III) and Rh(I), respectively], although we do not observe the formation of corresponding bis(phosphine)boronium salts in this case.
Stoichiometric Dehydrocoupling of Primary Phosphine−Boranes. The dehydrocoupling of primary phosphine−boranes can yield polyphosphinoboranes, rather than the simple oligomers observed with secondary phosphine− boranes (Scheme 1). With an appreciation of the intermediate metal complexes formed with secondary phosphine−boranes from this and previous work, 30,31,36 it was of interest to explore whether the proposed dehydrocoupling mechanism for secondary phosphine−boranes using [Rh(L)(η 6 -FC 6 F 4 ] could be applied to primary analogues. Such insight into the mechanism of dehydropolymerization of phosphine− boranes is important, as these processes currently remain unresolved due to the melt conditions employed that make following intermediates or kinetics problematic. 20,28,33 In situ investigations using stoichiometric quantities of primary phosphine−boranes H 3 B·PPhH 2 resulted in immediate reaction when combined with [Rh(L)(η 6 -FC 6 H 5 )][BAr F 4 ], but a number of products were formed which we have not been able to convincingly characterize. This mixture of species observed is in contrast with H 3 B·PPh 2 H where single products    F 4 ], as a proposed diastereomeric pair (Scheme 8). This stereoisomerism comes from P−H activation at the prochiral primary phosphine. These new products are directly analogous to those formed with secondary phosphine−boranes (i.e., 14), and the NMR spectroscopic data match closely. The 31 P{ 1 H} NMR spectrum from this reaction shows 8 resonances, in addition to a broad peak at δ −35.5 due to excess phosphine−borane, as each diastereomer contains four distinct phosphorus environments. Signals centered at δ 31.7 and 30.5 are assigned to one of the chelating phosphine ligand 31  Complexes 25a/b cannot be isolated in pure form, and characterization by NMR spectroscopy is best performed on freshly prepared samples, as after 1 h (25°C) they have undergone dehydrocoupling to give a mixture of two resolvable diastereomers 26a and 26b, with one of the diastereomers present in a significantly larger amount ∼6:1 (Scheme 8),
Resonances in the 31  . For these hydride signals the separate signals are not resolved for each diasteroisomer, although each resonance is rather asymmetric suggesting two overlapping environments.
A 31 P{ 1 H} NMR spectrum taken of this mixture after 18 h at 25°C showed a significant change in the ratios of the diastereomers 26a/26b (Scheme 9). The peaks for one isomer at [δ 34.5, 16.2, 10.7, and −14.9] have reduced relative area, giving an approximate ratio of 6:1 for the two diastereoisomers. This ratio is similar to that found from direct dehydrocoupling in 25a/25b after 1 h (vide supra), underscoring the stereocontrol occurring in the P−B bond forming process. Leaving this solution for one week resulted in no significant change to this ratio, suggesting equilibrium had been reached. We suggest that the mechanism for equilibration involves reductive elimination of the phosphido and hydride ligands to form a Rh(I) σ phosphine−borane complex, 30 similar to E in Scheme 2, which then undergoes rapid oxidative addition of the other P−H bond. This must be a reversible process, leading to a thermodynamic ratio of the diastereoisomers and the resulting   selectivity. Unfortunately we were unable to deduce the stereochemistry of the preferred isomer using ROESY experiments or a solid-state structure. However, inspection of models leads us to propose that the thermodynamic product is likely to have the cyclohexyl group pointing away from the chelating phosphine ligand's phenyl groups, i.e., 26b. That these diastereoisomers are a result of the metal activation of the prochiral terminal P−H bonds in 13 is shown by addition of an excess of dppe to 26a/b. 56 This affords [Rh(dppe)(L)]-[BAr F 4 ] 31 with the concomitant formation of free 13 (Scheme 9).
We are unable to comment in more detail on the conformation of these isomers, although the observation of stereocontrol in the direct dehydrocoupling is similar to that observed for the achiral system. Addition of excess dppe to this mixture forms a product identified by ESI-MS as [Rh(BDPP)(dppe)] + and free 13 (by 31 P and 11 B NMR spectroscopy). We have not explored whether there is enantiocontrol at the central {PCyH unit} arising from this PB coupling event on release from the metal.
For these experiments with H 3 B·PCyH 2 it is interesting to note that P−H activation is rapid and reversible with the Rh(I) precursor. This is in contrast to results obtained with secondary phosphine−boranes H 3 B·P t Bu 2 H and H 3 B·P t Bu 2 BH 2 ·P t Bu 2 H, which on addition to [Rh(L)(η 6 -FC 6 H 5 )][BAr F 4 ] gave the corresponding Rh(I) σ-phosphine−borane complexes with no P−H activation. 31 Such selectivity for primary over secondary phosphines in P−H activation at a metal center has been described previously for both phosphine 57 and phosphine− borane ligands. 27 In particular it has been shown that addition of H 3 B·PPhH 2 to Pt(PEt 3 ) 2 H(PPh 2 ·BH 3 ) results in exchange of the metal bound phosphide complex to give the primary phosphido−borane complex. 26 Here it was suggested that the greater thermodynamic driving force for formation of the primary phosphido−borane complex comes from steric effects, as M−P bonds to smaller primary phosphido ligands are likely to be stronger.
Catalytic Dehydrocoupling of Secondary Phosphine− Boranes. Under the standard catalytic melt conditions (90°C, 5 mol %), 20 [Rh(L)(η 6 -FC 6 H 5 )][BAr F 4 ] will dehydrocouple the secondary aryl phosphine−boranes used in this study to form the corresponding linear diboraphosphines 10−12, although we have not explored in detail the temporal evolution of these systems due to the problems associated with directly interrogating the melt. However, trends can be observed. For electron-withdrawing groups (1 and 2), complete consumption of starting material occurs in 4 h ( Table 1). The reaction at this temperature is not selective, and although the main product is the linear diboraphosphine, there are products that we tentatively identify as the cyclic oligomers (BH 2 PR 2 ) n (n = 3, 4). 20,33 Our results are broadly in line with the previously reported catalyzed dehydrocoupling of 2 using [Rh(COD)Cl] 2 , which, at a slightly lower temperature (60°C, 16 h, melt), affords the linear diboraphosphine product in 69% isolated yield, while at 100°C only the cyclic oligomers are isolated. The mechanism of formation of the higher cyclic oligomers, (BH 2 PR 2 ) n , remains to be resolved. 20 For electron-donating 3 the reaction is slower using the [Rh(L)(η 6 -FC 6 F 4 ] catalyzes dehydocoupling to give the corresponding linear diboraphosphine in greater than 95% conversion after 4 h. 31 For secondary phosphine−boranes, H 3 B·PPh 2 H thus offers balance between overall rate and selectivity. Given the product distributions and likely decomposition pathways in the melt it is inappropriate to comment in detail on the nature of the rate-determining steps during catalysis under these conditions. However, on the basis of the solution studies, P−B bond formation, (dehydrocoupling) is faster with electron-withdrawing groups. The temporal differences in observed product conversion in the melt could reflect a difference in the rate of the P−B bond forming event, or alternatively, they could reflect the ease at which the bound product is substituted on the metal center, i.e., a turnover limiting step. To probe this latter scenario, reaction between 19 (aryl-OMe) and diboraphosphine 11 (aryl-CF 3 ) to form 18 and free 12 demonstrates that an equilibrium is established slightly in favor of 18 (Scheme 11). This suggests that there is not a strong inherent difference in binding strengths between the two products, with the implication being that the observed rate differences in the melt arise from the dehydrocoupling step. Although this is different from what is observed in solution at room temperature, in which release of the product is likely the turnover limiting step, it is consistent with the high local concentration of H 3  ]. There were also other species observed ∼δ −55, which could be reduced in relative concentration (to ∼10%) by precipitation into hexanes. Such species have been previously suggested to be short-chain oligomers. 20 Interestingly, these proposed shorter chain oligomers are present in a greater proportion at shorter reaction times, which might suggest that polycondensation is occurring to give higher molecular weight polymer. Under non-melt conditions 20 (toluene heated to reflux, 0.5 mol %, 16 h) these shorter oligomers are by far the dominant species (Supporting Information). It thus appears that a high local concentration of phosphine−borane is necessary for productive dehydropolymerization. Positive mode ESI-MS (electrospray mass spectrometry) of the melt reaction product demonstrated p o l y m e r i z a t i o n , s h o w i n g r e p e a t u n i t s o f [ H -(PPhHBH 2 ) n PPhH 2 ] + up to n = 10 (Supporting Information). Similar analyses have been reported for amine−borane dehydropolymerization. 12,14,58 That these polymers are terminated by {PPhH 2 } groups rather than {BH 3 } is confirmed by inspection of the corresponding isotopomer patterns. This formulation also argues against cyclic oligomers being observed by ESI-MS, and presumably the additional phosphine arises from P−B bond cleavage. Use of H 3 B·PCyH 2 under these conditions afforded significantly more complex mixtures that we were unable to resolve.

■ CONCLUSIONS
The solid-state structures of the intermediates in the dehydrocoupling of secondary phosphine−boranes using the {Rh(Ph 2 PCH 2 CH 2 CH 2 PPh 2 )} + fragment have been determined. This demonstrates that the complex that precedes dehydrocoupling to form a linear diboraphosphine has σ bound and P−H activated phosphine−borane ligands, while the product has a linear diboraphosphine bound to the metal center. For aryl phosphine−boranes, electron-withdrawing groups (CF 3 ) promote stoichiometric dehydrocoupling faster than for more electron-donating (OMe) groups. This increase in rate is accompanied by a significant degree of parallel and competitive P−B bond cleavage to afford metal complexes with two monodentate phosphine ligands, which we suggest is due to a weakening of the P−B bond with electron-withdrawing aryl groups. These systems also turnover catalytically under melt conditions, with the overall rate of conversion broadly following the relative dehydrocoupling rates observed in the stoichiometric studies, suggesting that the dehydrocoupling step under melt conditions might also be the turnover limiting step. P−B bond cleavage also occurs for very bulky electron rich adamantyl phosphine−boranes, to such an extent that stoichiometric dehydrocoupling is not observed. For this phosphine−borane we suggest that sterics play a role in this process.
A significant observation is that, for primary phosphine− boranes, which are precursors to polyphosphinoboranes, use of the {Rh(Ph 2 PCH 2 CH 2 CH 2 PPh 2 )} + fragment results in some apparent diastereoselectivity in the dehydrocoupling step, at least in the stoichiometric reactions that produce metal-bound diboraphosphines. Such selectivity could well have implications in the control of the stereochemistry of polymer that would result from further insertion events. A significant future challenge is to harness any inherent bias in each dehydrocoupling insertion step productively while also developing the necessary spectroscopic and physical characterization markers to interrogate the oligomer and polymer stereochemistry.

■ EXPERIMENTAL SECTION
All manipulations, unless otherwise stated, were performed under an atmosphere of argon, using standard Schlenk and glovebox techniques. Glassware was oven-dried at 130°C overnight and flamed under vacuum prior to use. Hexane and pentane were dried using a Grubbs type solvent purification system (MBraun SPS-800) and degassed by successive freeze−pump−thaw cycles. 59 (4), and CyH 2 P·BH 3 (5) were prepared by the same method as Me 3 P·BH 3 61 but with the phosphine changed. (4-Trifluoromethylphenyl) 2 PH·BH 3 (2) and (3,5-bis(trifluoromethyl)phenyl) 2 PCl were prepared according to literature procedures reported by Clark et al. 33 NMR spectra were recorded on a Bruker AVD 500 MHz spectrometer at room temperature unless otherwise stated. In 1,2-C 6 H 4 F 2 , 1 H NMR spectra were referenced to the center of the downfield solvent multiplet (δ 7.07), and 31 P and 11 B NMR spectra were referenced against 85% H 3 PO 4 (external) and BF 3 ·OEt 2 (external), respectively. The spectrometer was prelocked and preshimmed using a C 6 D 6 (0.1 mL) and 1,2-C 6 H 4 F 2 (0.3 mL) sample. Chemical shifts are quoted in ppm and coupling constants in Hz. ESI-MS were recorded on a Bruker micrOTOF instrument. 62 In all ESI-MS spectra there was a good fit to both the principal molecular ion and the overall isotopic distribution. Signals in the 31 P{ 1 H} NMR spectra were integrated relative to those in similar environments (i.e., Rh−P or B−P) to obtain the relative ratios of products, and data was acquired with a pulse repetition time of 1 s. This avoids potential problems with different relaxation times for different phosphorus environments. Nevertheless, the quoted relative ratios based upon this data should be treated as qualitative rather than quantitative.
Details follow for 17. Slow diffusion of pentane (10 mL) over a solution of 17 in 1,2-F 2 C 6 H 4 at −24°C afforded yellow crystals (one of which was employed for an X-ray diffraction study). 1   Further experimental and characterization details, including selected NMR data and X-ray crystallographic data (including data in CIF format). This material is available free of charge via the Internet at http://pubs.acs.org. Crystallographic data have been deposited with the Cambridge Crystallographic Data Center (CCDC) and can be obtained via www.ccdc.cam.ac.uk/ data_request/cif.

Author Contributions
The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.