Ortho-aryl substituted DPEphos ligands: rhodium complexes featuring C–H anagostic interactions and B–H agostic bonds†

The synthesis of new Schrock–Osborn Rh(i) pre-catalysts with ortho-substituted DPEphos ligands, [Rh(DPEphos-R)(NBD)][BArF4] [R = Me, OMe, iPr; ArF = 3,5-(CF3)2C6H3], is described. Along with the previously reported R = H variant, variable temperature 1H NMR spectroscopic and single-crystal X-ray diffraction studies show that these all have axial (C–H)⋯Rh anagostic interactions relative to the d8 pseudo square planar metal centres, that also result in corresponding downfield chemical shifts. Analysis by NBO, QTAIM and NCI methods shows these to be only very weak C–H⋯Rh bonding interactions, the magnitudes of which do not correlate with the observed chemical shifts. Instead, as informed by Scherer's approach, it is the topological positioning of the C–H bond with regard to the metal centre that is important. For [Rh(DPEphos–iPr)(NBD)][BArF4] addition of H2 results in a Rh(iii) iPr–C–H activated product, [Rh(κ3,σ-P,O,P-DPEphos-iPr′)(H)][BArF4]. This undergoes H/D exchange with D2 at the iPr groups, reacts with CO or NBD to return Rh(i) products, and reaction with H3B·NMe3/tert-butylethene results in a dehydrogenative borylation to form a complex that shows both a non-classical B–H⋯Rh 3c-2e agostic bond and a C–H⋯Rh anagostic interaction at the same metal centre.


Introduction
Diphosphine chelates that contain an ether linkage in their backbone, such as DPEphos and Xantphos, are an important and popular class of ligand that are used in synthesis and catalysis ( Figure 1A). Initially developed as wide bite-angle, k 2 -P,P-ciscoordinating, ligands for Rh-based hydroformylation catalysis, 1, 2 such ligands also have the ability to act in k 3 -P,O,P binding modes often leading to hemilabile 3 behaviour through reversible coordination of the ether linkage in response to changes in the metal coordination sphere or oxidation state. DPEphos is now widely used in a variety of catalytic settings, [4][5][6][7] and the vast majority of applications make use of the commercially available phenyl phosphine derivative. Modification of aryl phosphine ligands, more generally, by introducing steric bulk using ortho-substitution has been shown to promote enantioselectivity; 8 regioselectivity; 9 overall efficiency and catalyst stability; [10][11][12][13] as well as aryl-group restricted rotation. 14 Despite these potential advantages, orthosubstituted variants of DPEphos (or Xantphos) are rare, Figure  1B, and their use limited to a handful of examples. 11,[15][16][17][18][19] The cationic Schrock-Osborn [Rh(chelating-phosphine)] + system is a widely used one in catalysis and synthesis, 20, 21 and the active species are often accessed via hydrogenation of a suitable diene precursor, such as [Rh(chelatingphosphine)(NBD)][anion] (NBD = norbornadiene), in a coordinating solvent such as acetone. We have particular interest in such systems with the DPEphos ligand, with regard to their use as pre-catalysts for amine-borane dehydropolymerisation, 22 Figure 2). A detailed structural, variabletemperature spectroscopic, and computational study reveals these to show well-defined examples of anagostic C-H···Rh interactions, 28,29 even for the previously-reported 24 parent DPEphos complex; while a reactivity study demonstrates intramolecular C-H activation can occur after hydrogenation of the NBD ligand, that is dependent on the identity of the Rgroup. Reaction of such a cyclometallated complex with H3B·NMe3 leads to a dehydrogenative borylation and a complex that features both non-classical B-H 3c-2e agostic 28 and anagostic C-H structural and spectroscopic features, Figure 2B. This serves to highlight the key differences between anagostic and agostic motifs of X-H bonds with d 8 -metal centres in a single complex.
In describing the anagostic interactions in these systems we borrow from the analysis of Scherer 30 who showed that axial positioning of a C-H bond at a square-planar d 8 metal centre orientates it over a region of charge concentration. When the complex is then placed in a magnetic field (i.e., the NMR experiment) induced current density at the metal results in magnetic field effects that cause the signature downfield chemical shift of the anagostic proton. In our analysis we find that descriptors that define the bonding between the Rh centres and C-H bonds show no correlation with either the observed or computed chemical shifts, supporting Scherer's topological, induced current, description for anagostic interactions.

Synthesis and Solid-State Structures of the NBD-Complexes.
The ortho-substituted DPEphos-R ligands used in this study are shown in Figure 3: R = H, 1-H; Me, 1-Me; OMe, 1-OMe; and i Pr, 1-i Pr. Ligands 1-H and 1-Me are commercially available, 1-OMe was prepared using the reported procedure. 17 DPEphosi Pr, 1i Pr, is a new ligand and was prepared as an analytically pure white solid from reaction of the corresponding dichlorophosphine with ortho-isopropyl phenyl lithium (ESI). The solid-state structure is shown in Figure 3. In the room temperature 31 P{ 1 H} NMR spectrum a single 31 P environment is observed at d -37.6.  Interestingly, the room temperature 1 H NMR spectrum is rather simple with only a single (integral 24 H) environment observed for the i Pr-methyl groups -despite their diastereotopic nature in the solid-state structure. This suggests inversion at P is a low energy process for free 1-i Pr, 31 which has been shown to be the case for other bulky i Pr-substituted tris-aryl phosphines. 32 The  F 4] was used to make 2-i Pr. The new complexes were isolated in moderate to good yield (65 to 85%), as crystalline, solids. Figure 4 shows the solid-state structures of the cations in these new complexes as determined by single-crystal X-ray diffraction. While 2-H is known, 24 the solid-state structure had not been reported, and so is included here. All the cations have pseudo square planar Rh(I) centres, with the NBD ligands binding h 2 h 2 , and cis-k 2 -P,P DPEphos-R ligands. Bond lengths and angles are generally unremarkable (ESI). The closet Rh···O distance in 2-OMe is 3.081(3) Å from an axially-orientated methoxyl group -which is clearly non-bonding.
Notable differences, however, come from the relative orientation of the DPEphos-R diphenylether backbone, Figure  4B. For 2-H, 2-Me and 2-OMe this lies above the P-Rh-P plane sitting in an asymmetric envelope-like conformation. 33 If retained in solution this would give the cation C1 symmetry (i.e. none). The i Pr groups in 2-i Pr force a, non-crystallographic, C2axis. Reflecting the increase in steric bulk, the Rh-P distances are ~0.1 Å longer and the P-Rh-P bite angle ~3º wider in 2-i Pr compared with the other complexes (Table S2). In all cases the ether oxygen atom sits distant from the Rh-centre [3.498(8)-3.5545(18) Å]. For all, there are aryl or methyl C-H bonds in the ortho-aryl groups that are axially positioned above the Rhsquare plane, i.e. potential anagostic interactions. These are discussed in detail after the solution NMR spectroscopic data have been presented that signal this orientation.   (Table S2).

Variable Temperature Solution NMR Spectroscopy and the Identification of Anagostic Interactions in Solution and Solid-State
Room temperature NMR spectra of the Rh-NBD complexes indicate fluxional behaviour in solution that is dependent on the identity of the phosphine ancillary group.  Figure 5A. There is no evidence for Rh-H coupling. While such downfield shifted signals are not observed in the room temperature 1 H NMR spectra of the other complexes, progressive cooling to much lower temperatures reveals similarly shifted peaks and corresponding changes in the 31 P NMR spectra. For 2-H cooling to 183 K (acetone-d6) results in very broad signals in the 1 H NMR spectrum, suggesting the low temperature limit had not been reached. By using CDCl2F 34 as a solvent a 1 H NMR spectrum could be obtained at 140 K in which a low-field shifted, albeit broad, signal (2 H) is observed at   with two mutually coupled signals in the corresponding 31 P{ 1 H} NMR spectra [e.g. J(PP)= 28 Hz 2-Me] that also couple to 103 Rh. For 2-H these signals are broader even at 140 K (fhwm = 80 Hz) and the 31 P-31 P coupling is not resolved, Figure 5B. These data point to fluxional processes that are arrested at low temperature to give structures that are similar to those determined in the solid-state, i.e. an envelope-like conformation of the DPEphos-R ligand. On increasing the temperature, conversion between enantiomeric C1 forms via a C2 intermediate is proposed, Figure 5C. This has been modelled for 2-Me using line-shape analysis (see ESI). Related ringflipping processes in POP-type ligands have been reported previously. 36,37 For 2-i Pr there is no change on cooling ( Figure  5B), the ~C2-symmetric solid-state structure is retained in solution at room temperature. It is thus not fluxional. These observations are consistent with relative steric bulk of the osubstituents: 1-H < 1-Me ~ 1-OMe << 1-i Pr. Downfield chemical shifts in the 1 H NMR spectrum can be diagnostic of anagostic C-H interactions, which are located above a region of charge concentration at a d 8 metal centre, i.e. an occupied dz 2 orbital. 30,[38][39][40] These are distinct from agostic, 28 3c-2e, bonds that are characterised by donation from a C-H bond into an unoccupied metal orbital and upfield chemical shifts in the 1 H NMR spectrum. The fluxional processes operating at room temperature mean these characteristic signals are only resolved on cooling, apart from for 2-i Pr in which the static structure makes them persistent. We next turn to inspecting the solidstate structures of the NBD adducts more closely to identify such anagostic interactions: Figure 6 and Table 1.
All four complexes show relatively close C-H···Rh approaches from an ortho C(aryl)-H group in the phenyl phosphine (H atoms in calculated positions, see Table 2 for computational analysis). For 2-H there are two, albeit long (~2.9 Å); for 2-i Pr there are also two, but these are considerably shorter (~2.5 Å); while 2-OMe has a single close C(aryl)-H···Rh distance (~2.9 Å). 2-Me shows two different types: C(aryl)-H···Rh (2.57 Å), and C(Me)-H···Rh (2.63 Å). The phenyl rings associated with these C(aryl)-H···Rh contacts generally align with the associated Rh-P vector (C-H/Rh-P torsion angles, Ψ, 8.2 to 1.3º) and the C-H···Rh angle (θ) is rather open (122.6-144.2º). Although 2-H has one phenyl ring twisted away from this (Ψ = 42.0, θ = 114.7º), the Rh···H distance is similar. The number of these close C-H···Rh distances correlates well with relative integrals of the downfield shifted signals observed in the 1 H NMR spectra: 2-H, 2H; 2-Me, 1H (aryl), 3H (methyl); 2-OMe, 1H; and 2-i Pr, 2H. As there is no crystallographically imposed symmetry in the solid-state we assume any equivalent environments observed in solution arise from very low energy fluxional processes. The changes in chemical shifts of these C-H protons due to the presence of the Rh(I) centre have been experimentally determined by comparison with the free ligands, as aided by 1 H/ 1 H COSY, HMBC and HSQC experiments. While all shift downfield, the variation observed shows no strong correlation with any of the structural descriptors discussed, as detailed in Table 1. However, in a more general sense, for all the complexes the angle formed between the Rh!H vector and the RhP2 plane (φ) shows the C-H proton is orientated towards the apical position (which at the limit φ = 90º). Thus, following Scherer's analysis, 30 the positioning of the C-H bond over a region of charge concentration (occupied d orbitals, φ approaching 90º) induces the downfield chemical shift in the NMR spectrum that is diagnostic of an anagostic interaction. In contrast, orientation of a C-H bond toward a charge depleted region (a vacant orbital in the metal coordination plane, φ approaching 0º) results in upfield-shifted signals that are characteristic of agostic, 3c-2e, bonding. Such demarcations are not always clear-cut, however, as axial sites can also display Lewis-acidic character. 29,41 While with hindsight it is not surprising that the most sterically bulky ligand, DPEphosi Pr, enforces an anagostic interaction at room temperature, the presence of both aryl  and, rarer, 42, 43 alkyl anagostic interactions in 2-Me is perhaps more notable. What was unanticipated is that in the parent DPEphos-H complex such interactions are also present -albeit only observed at very low temperature in solution. Similar properties (C-H···M, 2.23-3.01 Å, low-field chemical shifts and apical approaches of C-H groups to d 8 metal centres), have been discussed by others, including: Bergman, 44 Dyker, 42 Fairlamb, 45 and Sabo-Etienne, 38 Figure 7.
So, while the presence of anagostic C-H···Rh(I) interactions has been demonstrated here experimentally by both structural and spectroscopic studies, the correlation between the observed chemical shifts and measured structural descriptors is less obvious. We thus turned to a computational analysis to examine the nature of these anagostic C-H···Rh(I) interactions more closely.

Computational Studies: Structures, Bonding and Chemical Shifts.
Computed metrics for the Rh!H-C moieties in the isolated cations of all four DPEphos-R complexes are provided in Table 2. Geometries for these analyses are based on the experimental structures with the heavy atoms fixed at their observed positions and the H atoms optimised. The calculated Rh!H distances are therefore ca. 0.1 Å shorter (and the C-H bonds ca. 0.15 Å longer) than those determined experimentally. Figure    Interestingly, although there is a clear relationship between the Rh!H-C distance and the computed bonding metrics, no such correlation is seen with the computed chemical shifts of the anagostic hydrogens (Table 2). Thus, the nature of the Rh!H-C interaction does not relate to the extent of the downfield chemical shift, suggesting the orbitals involved are not responsible for the chemical shift. Instead the situation is more consistent with Scherer's observations 30 that it is the spatial positioning of the anagostic H above the d 8 square-planar metal coordination plane (i.e. φ) together with the complex interplay of induced current densities that are responsible for the precise chemical shift observed. Thus while the computation of a weak M!H bond path and weak Rh®s*C-H donation are usually features that are associated with an anagostic interaction, 40 they are not in themselves responsible for the signature downfield chemical shifts observed in NMR spectra that signal the positioning of the C-H bond relative to the metal centre.   53 These data, alongside selective decoupling experiments (ESI), allow a structure to be assigned for 4-Me as shown in Scheme 2, that is similar to [Rh(k 3 -P,O,P Xantphos)(H)2(acetone)][BAr F 4]. 51 The two different species observed at low temperature are assigned to conformers arising from different orientations of the ortho-Me substituted phenyl groups that undergo restricted P-C rotation. 10,14 For the DPEphosi Pr ligand the product of hydrogenation in acetone is different, and a Rh(III)-hydride i Pr-cyclometallated product is formed, Collectively these NMR data suggest complex 4-i Pr is formed as a mixture of at least three i Pr-cyclometallated species, that interconvert on the NMR timescale at room temperature by a process that does not break and exchange the Rh-H bond. Reversible reductive elimination and exchange with other C-H groups in the ligand would be expected to result in loss of the hydride signal and associated coupling if it occurred on the NMR timescale. [56][57][58][59][60] We thus propose that this fluxional process is associated with a restricted P-C rotation 10, 14 of the bulky i Praryl groups that leads to different, but exchanging, rotamers 61 of the same ortho-metalled isomer. In the absence of a singlecrystal X-ray structure we cannot definitively assign a structure to 4-i Pr as one where the i Pr methine or methyl group has undergone C-H activation, and both motifs are known. 62 While we cannot unequivocally rule out a ground-state structure arising from methinei Pr activation, we favour methyl activation as the hydride peaks correlate to methyl, aromatic and methine signals in the low temperature NOESY spectrum (ESI). Very similar spectra are obtained on hydrogenation in 1,2-F2C6H4 or o-xylene solvent (ESI), meaning there is no evidence for significant solvent coordination at the Rh(III) centre, or agostic interactions, the latter albeit expected to be weak. 63,64 The hydride is located trans to the coordinated oxygen on the basis of the observed chemical shift. 65 While reversible cyclometallation of 4-i Pr is not observed on the NMR timescale, it does occur on the laboratory timescale as probed by a variety of experiments, Schemes 3 and 4: (i) Addition of NBD quantitatively reforms 2-i Pr on time of mixing.    Particularly noteworthy in the 1 H NMR spectrum of 6-i Pr are two downfield shifted signals (1H relative integral each) at d 4.92 and 4.75 (dcalc = 5. 2 and 4.7), which are comparable to the signals assigned to anagostic C-H hydrogens in 5-i Pr. Closer inspection of the solid-state structure shows that the methine C-H protons H49 and H46 are in close approach to the Rh(I) centre and orientated above and below the RhP2B1 plane, Figure 8, (φ = 66.2º and 64.8º). In comparison, the upfield shifted signal, at d -7.54, is due to the agostic 3c-2e Rh···H-B motif that sits squarely in the RhP2 plane (φ = 6.6º). 6-i Pr thus highlights, in a single complex, the relationship between the orientation of the approaching E-H bond to the metal centre: the C-H anagostic interaction lying above the metal coordination plane and the 3c-2e B-H®Rh agostic bond sitting within the coordination plane.    72,81 (ii) hydroboration of the alkene using H3B·NMe3; 82 (iii) followed by dehydrogenation, via C-H activation/b-elimination. 75,76 Throughout tbe acts as a sacrificial hydrogen acceptor. While this scheme captures the gross transformations, the precise order of events currently remains unresolved.

Conclusions
We have shown that aryl-group ortho-substitution in [Rh(NBD)(DPEphos-R)] + leads to differences in structures, fluxional processes and reactivities -which can be related to the steric bulk of the ortho-group. Broadly speaking, OMe and Me substituents lead to solid-state and solution structures that are not too dissimilar to parent DPEphos. With the i Pr group fluxional processes in solution are retarded, and C-H activation processes occur. DPEphosi Pr thus cannot be considered an innocent ligand, this being related -more broadly -to the decomposition pathways of parent DPEphos that occur via C-O bond cleavage. 27,83 Common to all the Rh(I) DPEPhos-R complexes structurally described herein (with their associated NBD, CO or vinylborane co-ligands) is the observation of downfield-shifted signals in their 1 H NMR spectra that signal an anagostic M···H-C interaction, 28 for which the steric bulk of the ligand determines the temperature at which they are observed. As discussed previously, 30,38,40,45 while such anagostic interactions are associated with weak Rh®s*C-H and minimal sC-H ® Rh orbital donations, the driver for the downfield chemical shift of the C-H protons observed in the 1 H NMR spectrum does not come from these. Instead, the positioning of the anagostic hydrogen with reference to different regions of valence shell for the d 8 metal centre is important, as Scherer 30 has previously elegantly described for Rh(CAAC)(CO)Cl systems (CAAC = cyclic alkyl-aminocarbene). Our observations here, on a consistent set of complexes, reinforce this analysis. Thus, when the hydrogen atoms are forced, through steric constraints, to sit in an axial position (φ approaching 90º) that places them above a region of charge concentration, the associated magnetically-induced current density results in a downfield shift in the NMR spectrum, Figure 9A. This analysis differentiates anagostic interactions from 3c-2e agostic bonds, the latter being characterised by upfield shifts in their 1 H NMR spectra due to the associated hydrogen atoms being located in a region of charge depletion in the ligand plane of a d 8 ML3 type fragment ( Figure 9B, φ approaching 0º). Complex 6-i Pr offers E-H bonds (E = C, B) in both these topologies, and thus show both upfield and downfield chemical shifts in the 1 H NMR spectrum. While, as for 6-i Pr, any agostic bond will likely show a significantly stronger 3c-2e sX-H ® Rh interaction compared to the weak Rh®s*H-C donation associated with the anagostic interactions, the relationship, if any, between these bonding descriptors and the observed chemical shift has yet to be demonstrated.
These observations reinforce the analysis that the chemical shift changes observed by 1 H NMR spectroscopy in d 8 square planar complexes with anagostic C-H bonds located above the ligand plane result from topologically enforced ring current effects, rather than signalling an interaction that has a considerable orbital contribution. In this regard they are perhaps more related to the chemical shift changes that are