Dehydropolymerization of H3B·NMeH2 Using a [Rh(DPEphos)]+ Catalyst: The Promoting Effect of NMeH2

[Rh(κ2-PP-DPEphos){η2η2-H2B(NMe3)(CH2)2tBu}][BArF4] acts as an effective precatalyst for the dehydropolymerization of H3B·NMeH2 to form N-methylpolyaminoborane (H2BNMeH)n. Control of polymer molecular weight is achieved by variation of precatalyst loading (0.1–1 mol %, an inverse relationship) and use of the chain-modifying agent H2: with Mn ranging between 5 500 and 34 900 g/mol and Đ between 1.5 and 1.8. H2 evolution studies (1,2-F2C6H4 solvent) reveal an induction period that gets longer with higher precatalyst loading and complex kinetics with a noninteger order in [Rh]TOTAL. Speciation studies at 10 mol % indicate the initial formation of the amino–borane bridged dimer, [Rh2(κ2-PP-DPEphos)2(μ-H)(μ-H2BN=HMe)][BArF4], followed by the crystallographically characterized amidodiboryl complex [Rh2(cis-κ2-PP-DPEphos)2(σ,μ-(H2B)2NHMe)][BArF4]. Adding ∼2 equiv of NMeH2 in tetrahydrofuran (THF) solution to the precatalyst removes this induction period, pseudo-first-order kinetics are observed, a half-order relationship to [Rh]TOTAL is revealed with regard to dehydrogenation, and polymer molecular weights are increased (e.g., Mn = 40 000 g/mol). Speciation studies suggest that NMeH2 acts to form the precatalysts [Rh(κ2-DPEphos)(NMeH2)2][BArF4] and [Rh(κ2-DPEphos)(H)2(NMeH2)2][BArF4], which were independently synthesized and shown to follow very similar dehydrogenation kinetics, and produce polymers of molecular weight comparable with [Rh(κ2-PP-DPEphos){η2-H2B(NMe3)(CH2)2tBu}][BArF4], which has been doped with amine. This promoting effect of added amine in situ is shown to be general in other cationic Rh-based systems, and possible mechanistic scenarios are discussed.


INTRODUCTION
Polyaminoboranes, 1−4 exemplified by N-methylpolyaminoborane (H 2 BNMeH) n , have alternating main-chain B−N units and are of interest as precursors to BN-based ceramics or as new unexplored materials that are isosteres of polyolefins. Since the original report of the synthesis of (H 2 BNMeH) n by the dehydropolymerization of H 3 B·NMeH 2 using an Ir(POCOP)-H 2 catalyst (POCOP = κ 3 -C 6 H 3 -2,6-(OP t Bu 2 ) 2 ), 4−6 there has been significant progress in developing catalytic methodologies, 7−13 as well as noncatalyzed routes. 14 The accepted overarching catalytic mechanism operates via initial dehydrogenation of H 3 B·NMeH 2 to form a transient free, or metalbound amino−borane, which then undergoes a head-to-tail BN coupling (Scheme 1). A number of different propagation scenarios have been proposed for this latter step that show elements of chain-growth, 4,10 step-growth, 15 or hybrid mechanisms. 16 Particularly interesting would be systems that demonstrate the potential for control 17 over the polymerization process, holistically defined by degree of polymerization (as measured by M n ), dispersity (Đ), initiation/termination events, and catalyst lifetime (i.e., TON). While aspects of these performance criteria have been noted, [7][8][9][10]15 there is no general approach to their optimization.
We have reported cationic dehydropolymerization precatalysts based upon {Rh(Xantphos-R)} + motifs, 18,19 in which the identity of the PR 2 group is changed (Scheme 2). 9,10,20 When R = Ph (A), medium 2 molecular weight polymer is formed (M n = 22 700 g/mol, Đ = 2.1), a higher catalyst loading promotes lower M n , and H 2 acts to modify the polymer chain length (M n = 2 800 g/mol, Đ = 1.8). Although detailed kinetics for H 3 B· NMeH 2 dehydropolymerization were not reported, these observations were interpreted as signaling a coordination/ insertion/chain-growth mechanism in concert with more extensive studies on H 3 B·NMe 2 H. 9 There is also a significant induction period observed (∼10 min). In contrast, when R = i Pr (B), H 2 and catalyst loading do not significantly change M n (9 500 g/mol, Đ ≈ 2.8), there is a negligible induction period, and a dual role 11,12 for the organometallic species was proposed in which dehydrogenation/propagation occurs from different metal centers. This mechanistic switch may be influenced by the preferred ligand-coordination modes: 21 Xantphos-Ph is a hemilabile ligand preferring to coordinate cis-κ 2 -PP and mer-κ 3 -POP, while Xantphos-i Pr prefers mer-κ 3 -POP ( Figure S1 compares coordination modes for crystallographically characterized Xantphos-R complexes).
We now report a detailed and systematic study on the dehydropolymerization of H 3 B·NMeH 2 using a different Rh-POP-based system: {Rh(DPEphos)} + [DPEphos = bis(2-(diphenylphosphino)phenyl)ether]. Using this ligand, which favors cis-κ 2 -PP coordination ( Figure S1), significant control over M n by both catalyst loading and H 2 is achieved, with M n ranging from 5 500 to 40 000 g/mol and Đ = 1.5−1.8. These studies also reveal the formation of dimeric species, and the key role of added amine, NMeH 2 , in both promoting catalysis and increasing M n /lowering Đ of the isolated polymer. Finally, combining these observations, the synthesis and evaluation in catalysis of a simple [Rh(κ 2 -PP-DPEphos)(NMeH 2 ) 2 ] + precatalyst is reported. This positive influence of added amine is also shown to be general for other previously reported cationic Rh-based systems. The role of added amine has been recently noted with regard to increasing catalyst lifetime of Ru-based catalysts for the dehydropolymerization of H 3 B·NH 3 by trapping BH 3 formed from B−N bond cleavage, 8 although the influence of amine on the characteristics of the polymer produced were not commented upon.  Figure  1A). The resulting 11 B NMR spectra of the reaction mixtures and isolated polymer show the characteristic 2,12 broad signal at δ −6 ppm for (H 2 BNMeH) n and only trace (HBNMe) 3 ( Figure S18). The 13 C{ 1 H} NMR spectra (H 8 -THF) show a relatively sharp peak at δ 35.5 ppm (NMe). In contrast, at 0.1 mol % catalyst loading, M n does not increase compared to 0.2 mol %, and there is significant 1,2-F 2 C 6 H 4 insoluble polymer that is tetrahydrofuran (THF)-soluble. NMR spectroscopic analysis of this material ( Figure S19) showed additional signals at δ( 11 B) ∼1 ppm and δ( 13 C{ 1 H}) ∼35.7 ppm (br, NMe) that may signal tertiary or quaternary main-chain centers, suggesting cross-linking/chain branching. 10,11,19,25 While we currently have no explanation for this change in polymer characteristics, at these very low loadings trace impurities (or products of B−N bond cleavage, vide infra) may have a disproportionate effect on the polymerization process, leading to a different product being formed. When dehydropolymerization was conducted under H 2 measurement conditions (eudiometer, H 2 established in the head space), or in a closed system that allows for H 2 buildup, H 2 likely acts as a chaintransfer/termination agent and significantly shorter polymer is isolated, for which a significantly larger signal at δ( 11 B) ∼−18 ppm is observed, which could be assigned to BH 3 end groups 15 ( Figure 1B; Figure S20 shows a representative 11 B NMR spectrum). Similar Đ are retained compared with the open system, as is the inverse relationship between M n and catalyst loading ( Table 1, entries 5−8). Interestingly, there is now a significant difference in M n between 0.1 and 0.2 mol %, suggesting that H 2 modifies the influence of the very low catalyst loading. A conversion versus M n study (0.2 mol %, open system, Figure 1C) indicates that a chain-growth mechanism is operating, because at low (10%) conversions long polymer chains are observed (M n = 24 800 g/mol, Đ = 1.2) and H 3 B·NMeH 2 monomer dominates ( Figure S21).

RESULTS
We have previously, but briefly, reported similar control of molecular weight by catalyst loading and H 2 for catalyst A and suggested a coordination/dehydrogenation/insertion/chaingrowth mechanism for the dehydropolymerization, in which the same metal center both dehydrogenates an amine−borane and promotes propagation. 9 This more comprehensive data with 2a supports a similar mechanism in the {Rh(DPEphos)} + system. That H 2 acts to modify the polymer chain may arise

ACS Catalysis
Research Article from chain-termination/transfer by hydrogenolysis of a Rh− BH 2 (polymeryl) or Rh−NMeH(polymeryl) bond. The use of H 2 as a chain-termination agent in olefin polymerization is well-established, operating through sigma-bond metathesis of [M]-CH 2 -polymeryl with H 2 to form a metal hydride and free polymer. 26 The inverse relationship between M n and catalyst loading suggests dehydropolymerization at a single metal center, as lower catalyst loadings lead to less propagating sites for the concomitantly formed H 2 BNMeH. Interestingly, this relationship between M n and initiating sites is also reminiscent of a classical radical polymerization mechanism where the net order in initiator is negative, 27 as has been recently noted. 3 2.3. Speciation Experiments: The Formation of Dimeric Rh 2 Species. With the polymer growth kinetics in hand, we turned to identifying the species that formed during catalysis using NMR spectroscopy. The low catalyst loadings used for polymerization (0.1−1 mol %) meant that these speciation studies were performed instead at 10 mol % 2a to obtain good signal/noise (sealed NMR tube, 1,2-F 2 C 6 H 4 ). Under these in situ conditions, 11 B NMR spectroscopy showed the formation of a mixture of (H 2 BNMeH) n , (HBNMe) 3 , and (H 2 B) 2 (μ-H)(NMeH) [td, δ −22.3 ppm 28 ], with the latter potentially signaling free BH 3 by loss of amine. 31 P{ 1 H} NMR spectroscopy under these conditions showed the initial formation, after 5 min, of two new dimeric complexes: a bridging hydrido-aminoborane 3a, [Rh 2 (DPEphos) 2 (μ-H)(μ-H 2 BNHMe)][BAr F 4 ], and an amidodiboryl 4a, [Rh 2 (κ 2 -P,P- Figure 2A). After 2 h 4a is dominant (80%), but the mixture slowly returns to favoring 3a after 5 h ( Figure S22). Complex 3a can be prepared as the only organometallic species by addition of H 2 / 2 equiv of H 3 B·NMeH 2 to 1a. Boronium [BH 2 (NMeH 2 ) 2 ] + [δ −7.1 ppm, J(BH) = 110 Hz, cf. authentic sample δ −7.4 ppm, J(BH) = 117 Hz, 1,2-F 2 C 6 H 4 10 ] is also observed under these conditions, 29 in line with the reported mechanism for the formation of analogous complexes with [Rh 2 (R 2 P-(CH 2 ) n PR 2 ) 2 (μ-H)(μ-H 2 BNR′ 2 )] + motifs. 30,31 Here, attack of free amine (from B−N bond cleavage 32 ) at a precursor σamine−borane complex generates a neutral dimeric Rh− hydride and [BH 2 (NMeH 2 ) 2 ] + , for which subsequent proton transfer and NMeH 2 loss result in the bridging amino−borane motif. NMR and ESI−MS data for 3a are fully consistent with its formulation (Supporting Materials) and are very closely related to previously reported [Rh 2 ( i Pr 2 P(CH 2 ) 3 P i Pr 2 ) 2 (μ-H)(μ-H 2 BNH 2 )][BAr F 4 ]. 30 Attempts to characterize these products using single-crystal X-ray diffraction were frustrated by the formation of oily materials. The identity of 4 was only revealed using the [Al(OC(CF 3 ) 3 ) 4 ] − anion, 33 by a singlecrystal study of 4b, [Rh 2 (κ 2 -P,P-DPEphos) 2 (σ,μ- 4 ], which comes from a slow (days) recrystallization of 3b, formed in situ from [Rh(κ 2 -P,P-

ACS Catalysis
Research Article Figure 2B). 4b is not isolated pure, formed alongside 3b (∼5% by 31 P{ 1 H} NMR spectroscopy) and (H 2 BNMeH) n . The NMR data for 4b, aside from the signals due to the anion, are the same as for 4a, as are the ESI−MS data.
The structure of the cation in 4b has a Rh 2 core [Rh−Rh 2.6421(4) Å] with a bridging amido−bisboryl ligand that has two α-BH···Rh agostic interactions with the proximal Rh centers [e.g., Rh2−B1 2.107(5), Rh1···B1 2.326(5) Å]. Such a description results in formally Rh(II) centers with a Rh−Rh bond accounting for the diamagnetism. An alternative description of the bonding in 4b is a diborylmethylammonium complex that would result in the Rh centers being formally Rh(0). The DPEphos ligand adopts a κ 2 -PP motif, with two of the phosphines (P2, P3) trans to the BH agostic interaction and cis to the Rh−Rh bond, while P1 and P4 lie trans to the Rh−Rh bond and couple to both Rh centers in the 31 P{ 1 H} NMR spectrum [e.g., J(RhP) = 139, 102 Hz]. The four 31 P environments are chemically inequivalent. There is no evidence for a Rh−H−Rh bridging hydride (NMR, ESI− MS), and the α-BH···Rh are observed as two broad doublets at δ −8.86 and −9.44 ppm [J(PH) ≈ 70 Hz] in the 1 H{ 11 B} NMR spectrum. 34 The 11 B NMR spectrum shows a broad signal at δ 9.4 ppm. These data show that the solid-state structure is retained in solution. As the NMeH group forces C 1 symmetry in the molecule, this also shows that the amido− bisboryl ligand is not undergoing rapid and reversible Rh center are also observed in the QTAIM analysis and also observed experimentally, e.g., Rh1···C38, 2.997(5) Å. Consistent with such interactions, a broad asymmetric signal is observed at δ 3.94 ppm (2 H) in the 1 H NMR spectrum of 4b that is attributed to agostic Rh···HC phenyl interactions, similar to that observed in [Ru(P i Pr 3 ) 2 (H)(H 2 )(C 6 H 5 C 5 H 4 N)]-[BAr F 4 ] (δ 4.14 ppm). 35 4b is a rare example of a complex with both C−H and B−H agostic interactions. 36,37 Related structures to 4b that show bridging "BNB", 20,38 α-BH···Rh agostic, 39 or amino−boryl motifs 9,40 have been reported before. However, as far as we are aware, the amido−bisboryl structure is a new motif in metalloborane chemistry. Perhaps most closely related to 4b is a Rh-dimer with P−C activated Xantphos-Ph ligands and a bridging N,Ndimethylaminodiboranate unit ([H 3 BNMe 2 BH 3 ] − ) that is isolated at the end of dehydrocoupling of H 3 B·NMe 2 H when using catalyst A. Interestingly, this is also a competent catalyst for H 3 B·NMeH 2 dehydropolymerization. 20 While we currently can only speculate on the mechanism of formation of 4, it is connected to 3 by simple addition of BH 3 and loss of H 2 . Under catalytic conditions 3 likely forms first, while the role of 4 is less clear. To help resolve the identity of the active species in catalysis, kinetic studies were undertaken, taking 2a, 3a, and 4b as precatalysts.
2.4. Kinetic Studies of Dehydropolymerization As Followed by H 2 Evolution. The kinetics of dehydropolymerization were followed by volumetric studies of H 2 generation using a eudiometer. In all cases ∼1.1 equiv of H 2 was measured and very little N-trimethylborazine was observed by 11 B NMR spectroscopy (<5%, Figure S23), indicating that evolved H 2 is a good proxy for transient 41

ACS Catalysis
Research Article and subsequent polymer chain growth. A significant induction period was observed prior to faster turnover (e.g., ∼60 min, 0.4 mol %), that gets longer with increase in [2a] 0 (Figures 4A and S24; e.g., 0.1 mol %, t ind = 33 min; 1 mol %, t ind = 110 min). An induction period has also been noted for catalyst A in H 3 B· NMeH 2 dehydropolymerization 9 as well as for [Rh(Ph 2 P-(CH 2 ) 3 PPh 2 )(FC 6 H 5 )][BAr F 4 ], C, in H 3 B·NMe 2 H dehydrocoupling (10 and 5 min, respectively, at 0.2 mol %). 42 For this latter system, increased [Rh] TOTAL also led to longer induction periods, and a subsequent study showed the initial formation of an amino−borane-bridged dimer analogous to 3a. 30 While the observation of an induction period might suggest a heterogeneous system here, 43−45 addition of excess Hg or substoichiometric PPh 3 during productive turnover did not significantly reduce reaction rate, and no darkening of the reaction was noted, pointing toward homogeneous catalysis ( Figure S25). Overall, the kinetics evolve in a sinusoidal manner, with a rate maximum reached approximately at the midpoint (e.g., 0.4 mol %, ν max = 4.1(2) × 10 −5 M s −1 ). This behavior is suggestive of a long induction period coupled to rate-attenuation as the substrate is depleted. There is a noninteger dependence of the maximum rate on the initial catalyst concentration ( Figure S28), which hints at more complex kinetics. Using 0.223 M D 3 B·NMeH 2 or H 3 B·NMeD 2 at 0.4 mol % 2a, kinetic isotope effects (KIEs) determined from ν max were k(BH)/k(BD) = 1.1 ± 0.1 and k(NH)/k(ND) = 2.2 ± 0.1, which suggests that N−H bond cleavage is involved in the turnover-limiting step. These data are very similar to those measured for A. 9 The polymerization is not living as recharging 2a gives approximately the same M n , at a similar rate for second recharge ( Figure S31). A short induction period was noted for each recharge, which reflects the reformation of 3a at the end of catalysis (vide infra).
Use of in situ generated dimeric 3a leads to a shorter, but still significant, induction period (∼30 min, Figure 4B) and a similar profile and rate maximum as for 2a. In contrast, reaction of crude 4b resulted in no detectable induction period. Furthermore, H 2 evolution (a proxy for H 2 BNMeH formation) followed a first-order profile ( Figure 4B, k obs = 3.2(1) × 10 −4 s −1 ), and this allowed for a half-order dependency on initial catalyst concentration, i.e., [Rh] TOTAL , to be estimated ( Figures 4C and S30).
The polymers isolated from these H 2 evolution studies using 3a and 4b are similar by GPC analysis but slightly longer compared to that from 2a at equivalent [Rh] TOTAL (Table 1, entries 7, 9, and 10). Speciation studies at 1 mol % 2a return only 3a at the end, which suggests that, if formed, 4a must be consumed under the conditions of catalysis. Overall these data show the following: a change in H 2 -evolution kinetics on moving from 2a (complex) to 4 (pseudo first-order), that 4 likely sits close to the actual catalyst, and that 3 still requires an induction process to bring it on-cycle. The approximately halforder dependence in [Rh] TOTAL when using 4a as a precatalyst suggests a lower-order (ligation or nuclearity) active catalyst that is in a rapid equilibrium with a higher-order inactive species, as is discussed later.
2.5. Kinetic Studies: Doping Experiments and the Promoting Effect of NMeH 2 . Seeking to understand the observed kinetics, and in particular the underlying reason for the induction period, the influence of various species that may be present, or formed, during catalysis was examined. Addition of 1 equiv of H 3 B·THF (in 50 μL of THF) to 0.4 mol % [2a]/ H 3 B·NMeH 2 /1,2-F 2 C 6 H 4 solvent increased the induction period significantly ( Figure 4A) and gave significantly shorter polymer (Table 1, entry 11), while 10 equiv halts catalysis, possibly by the formation of inactive boron-rich species (see Supporting Information). 32 Added [H 2 B(NMeH 2 ) 2 ][BAr F 4 ] (10 equiv) significantly slows catalysis, now taking 24 h for completion to produce very short polymer (M n = 2 800 g/mol, Đ = 2.3). This argues against its role in productive catalysis, in contrast with other systems, 10,29,46 in particular the [Rh-(Xantphos-i Pr)] + system, where it promotes catalysis. 10 At low relative concentrations, H 3 B·THF presumably acts to titrate out NMeH 2 , while we propose that excess [H 2 B(NMeH 2 ) 2 ] + acts to poison catalysis, possibly sequestering NMeH 2 via N− H···NMeH 2 hydrogen bonding, as noted for related bis-(phosphine)boronium salts. 47 The control experiment of THF addition (50 μL) reduced the induction period to 30 min and produced polymer comparable to nondoped experiments ( Table 1, entry 13). The most dramatic change came from addition of ∼2 equiv of NMeH 2 (in 50 μL of THF) to 0.4 mol % [2a]/H 3 B·NMeH 2 . This resulted in a kinetic profile for H 2 evolution that now showed no induction period and pseudofirst-order kinetics for hydrogen evolution (k obs = 3.7(1) × 10 −4 s −1 ), similar to that of 4b at the same [Rh] TOTAL . Isolated polymer, however, was considerably longer (M n = 27 400 g/mol, Đ = 1.9) than for when just 2a was used. As expected, under open conditions M n increases (M n = 32 100 g/mol, Đ = 1.6), albeit to a lesser extent than compared with the analogous nondoped experiments (cf. entries 14/15 and 2/ 7, Table 1). These observations, alongside the speciation data at 10 mol %, which demonstrate that 3a is likely the first formed species, show that free NMeH 2 formed from B−N bond cleavage is key to not only bringing the catalyst on-cycle but also promoting propagation or attenuating chain-transfer/ termination, leading to higher molecular weights of isolated polymer. Given these observations, the role of NMeH 2 was next investigated.
2.6. Rh−Amine Adducts As Effective Precatalysts. We first sought to understand the likely species generated in situ by

ACS Catalysis
Research Article addition of amine to the precatalyst, 2a. Addition of ∼2 equiv of NMeH 2 (in THF) to 2a gave the simple bisamine complex [Rh(κ 2 -P,P-DPEphos)(NMeH 2 ) 2 ][BAr F 4 ], 6, which reacts rapidly (on time of mixing) with H 2 in situ to form the corresponding dihydride [Rh( κ 2 -P,P-DPEphos)-(H) 2 (NMeH 2 ) 2 ][BAr F 4 ], 5 (Scheme 4). Complex 5 reversibly, but slowly, loses H 2 under extended degassing to reform complex 6, and thus we suggest that, under the conditions of dehydropolymerization, 5 would be persistent. NMR spectroscopic data are fully consistent with the proposed structures (see later), but under these conditions of synthesis isolating pure samples of 5 and 6 in bulk has proved difficult; and a 1:1 mixture of 5/6 is conveniently prepared from 1a/∼2 × NMeH 2 /H 2 /degas and used directly in catalysis (see Supporting Information). Complex 5 is the sole organometallic product on addition of ∼2 equiv of NMeH 2 to a 1:3 mixture of 3a/4a, alongside HB(NMeH) 2 [δ( 11 B) 28.6 ppm, J(BH) = 127 Hz], demonstrating the role of NMeH 2 in both generating 3, via boronium formation, 29,30 and bringing dimeric 3 and 4 back to monometallic species. Complex 6 (and 5 on subsequent addition of H 2 in solution) can be prepared as a free-flowing pure solid in bulk via an alternative route, from addition of NMeH 2 to [Rh(κ 2 -P,P-DPEphos)(η 6 -o-Me 2 C 6 H 4 )][BAr F 4 ], 7, 48 which enables definitive characterization by NMR spectroscopy. However, this involves laborious multiple triturations with cold pentane, and thus, the in situ prepared mixture is more convenient to use. Notable NMR spectroscopic data for 6 are the observation of equivalent NMeH 2 groups in the 1 H NMR spectrum, while for 5 addition of H 2 makes these groups inequivalent and diastereotopic; two Rh−H environments are observed, one of which shows a large trans coupling to 31 P [J(HP) = 182 Hz], and inequivalent phosphorus environments are observed in the 31 P{ 1 H} NMR spectrum (Supporting Information). Data from H 2 -evolution kinetics and isolated polymer using isolated 6 fit well with the trends apparent from using the 5/6 in situ mixture (Table 1 and Figure 5).
Using in situ generated 5/6 gave pseudo first-order plots for H 2 evolution (e.g., 0.4 mol %, k obs = 4.1(1) × 10 −4 s −1 ) with no induction period observed. These were also half-order in [Rh] TOTAL ( Figure 5A). Half-order behavior is indicative of either a rapid equilibrium between species of different nuclearity, e.g., monomer−dimer, prior to the turnoverlimiting step, in which the higher nuclearity species is inactive but dominant, 49 or the rapid and reversible dissociation of a ligand that reveals a low concentration of an active species. 50 Monomer/dimer equilibria have been proposed in polymerization systems previously, 51−53 and in amine−borane dehydrocoupling specifically. 49,54,55 While addition of 10 equiv of NMeH 2 caused no significant change in rate (k obs = 4.2(1) × 10 −4 s −1 ), suggesting that NMeH 2 dissociation is not occurring, the polymer isolated from this experiment was insoluble in THF. We thus cannot rule out a change in mechanism. We discount rapid and reversible H 2 loss as the reason for the observed half-order kinetics because under conditions of measurement H 2 effectively becomes saturated and constant. Speciation studies with excess NMeH 2 (10 equiv, [Rh] TOTAL = 5 mol %) revealed 5 to be the only observed organometallic species. No significant change in kinetics was observed on addition of excess Hg, or 0.2 equiv of PPh 3 , during catalysissuggesting a homogeneous system. 56 The use of these in situ prepared amine complexes 5/6 leads to polymer with greater M n (but still inverse with regard to [Rh] TOTAL ), while Đ is kept relatively low ( Figure 5B, e.g., 1 mol %, M n = 20 600 g/mol, Đ = 1.5). Thus, the added aminewhether bound or freenot only brings the catalyst onto cycle but also promotes greater apparent degrees of polymerization. Whether this is by faster propagation or attenuation of termination is not currently known.
Following catalysis by 31 P{ 1 H} NMR spectroscopy using pure 5 (1 mol %) showed that during productive catalysis a single organometallic species is observed (albeit with low signal-to-noise) as a doublet at δ 41 ppm [J(RhP) = 150 Hz], which slowly resolves to complex 3 at the end of catalysis. Importantly, the same species is observed when starting with precatalyst 4b (0.5 mol %, 1 mol % [Rh] TOTAL ). This strongly suggests that both precatalysts evolve to a common species the identity of which remains to be resolved.
Interestingly, the promoting effect of NMeH 2 is not operative in the [Rh(Xantphos-i Pr)(H) 2 ] + system, 10  . This is probably due to the relatively strongly bound amine blocking access of H 3 B· NMeH 2 to the metal center, at which the Xantphos-i Pr is also not hemilabile ( Figure S1), so that σ-complex formation by coordination of amine−borane, and subsequent dehydrogenation by BH/NH activation, does not take place. The broader promoting effects of NMeH 2 are, however, evident in other cationic {Rh(chelating phosphine)} + systems that are suggested to undergo a coordination/dehydrogenation/chaingrowth mechanism. Under the specific conditions reported here, both [Rh(Xantphos-Ph)] + , A, 9 and [Rh(Ph 2 P-
2.7. Discussion of Proposed Mechanistic Landscape. Bringing these observations together, we propose an overall mechanism shown in Scheme 5, in which the induction period that gets longer with increased [2a] can also now be explained. NMeH 2 , generated by slow B−N bond cleavage of H 3 B· NMeH 2 , at a rate that is independent of [2a], first promotes the formation of 3a and then more slowly the active precatalyst 5. In this model, higher concentrations of 2a result in more 3a needing to be first formed, via hydride abstraction and boronium formation, and then converted to the active catalyst with an unchanged amount of NMeH 2 , thus leading to a longer induction period. The active catalyst is closely related to both 5/6 and 4a, but we suggest both of these sit outside of the productive cycle, as their structures and reactivity are incompatible with the observed kinetics. The insensitivity in rate to added NMeH 2 suggests this does not reversibly dissociate, while a sensible model in which dimeric 4a, with its Rh−Rh bond and bridging amido−bisboryl ligand, undergoes rapid and reversible dissociation (vide supra) or loss of ligand is not obvious. Moreover, 4b reacts rapidly with NMeH 2 to form 5, suggesting that if formed in catalysis it is not persistent. In addition, the fact that both 5 and 4b evolved to the same, currently unresolved, organometallic species under catalytic conditions suggests that both sit just outside of the productive catalytic cycle. While we cannot currently confidently comment on the nature of the actual catalyst for dehydrogenation, chain growth, or the termination process, the half-order relationship in [Rh] TOTAL and the observation of dimeric species (3 and 4) suggest that such Rh 2 motifs may be intimately involved. The strong, and persistent, inverse relationship between M n and [Rh] TOTAL , coupled with the sensitivity to H 2 , suggests a coordination/insertion/chaingrowth mechanism for which NMeH 2 also modifies chain lengthpossibly by attenuating chain termination. On the basis of the half-order kinetics observed from the dehydrogenation studies, we suggest three possible general motifs for the active catalyst (Scheme 6): one which invokes a monomer− dimer equilibrium in which one of the monomers is the active catalyst (A), and one in which a persistent dimer reversibly loses a bound ligand (B). Scenario A is reminiscent of the unsymmetrical Rh 2 hydride dimers that can form in Rhcatalyzed alkene hydrogenations, 57 while scenario B is supported by the recent report that dimeric early transitionmetal complexes have been shown to act as competent catalysts for H 3 B·NMeH 2 dehydropolymerization. 7 A third possibility is that deprotonation of bound NMeH 2 provides an active Rh−NMeH amido motif, similar to the bifunctional catalysts developed by Schneider and co-workers (C). 11

CONCLUSIONS
We have shown that a combination of catalyst loading, H 2 , and NMeH 2 can be used to control the dehydropolymerization of H 3 B·NMeH 2 in a {Rh(DPEphos} + -based catalyst. We proposed this to be an important observation and one that may show some generality, building upon the already demonstrated improvement in catalyst lifetimes on addition of amine. 8 The ability to control polymerization by catalyst loading, NMeH 2 addition, and H 2 in {Rh(DPEphos)} + and {Rh(Xantphos-Ph)} + systems is markedly different from that found for the {Rh(Xantphos-i Pr)} + catalyst and further supports that a different mechanism operates between the two sets, which may be related to the preferred coordination geometry of the ligands: DPEphos and Xantphos-Ph prefer cisκ 2 -P,P while Xantphos-i Pr generally adopts mer-κ 2 -P,O,P motifs. The amine systems we describe thus provide a tractable platform for further detailed mechanistic studies, and efforts Scheme 6. Generalized Possible Active Species in Catalysis a a P = phosphine, L = ligand (e.g., NMeH 2 , or amine−borane-derived fragment). All structures shown are representative, and the actual number of hydrides/coordination geometry is undetermined.

ACS Catalysis
Research Article are directed to determining the details of the propagating species and termination events so that fine control of the overall process, and thus the polymer produced, can be realized. It will be interesting to see if this effect of added amine is a more general observation across the now numerous 2,3 dehydropolymerization catalysts from across the transition metals.
■ ASSOCIATED CONTENT

* S Supporting Information
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acscatal.9b00081.
Full experimental section, characterization details, kinetic data and details of the DFT calculated structure, and QTAIM analysis of 4b (PDF) Crystallographic data (CIF)