A Borane Lewis Acid in the Secondary Coordination Sphere of a Ni(II) Imido Imparts Distinct C-H Activation Selectivity

Abstract

Two borane-functionalized bidentate phosphine ligands that vary in tether length have been prepared to examine cooperative metal-substrate interactions. Ni(0) complexes react with aryl azides at low temperature to form structurally unusual κ²-(N,N)-N₃Ar adducts. Annealing these adducts affords products of N₂ extrusion and in one case, a Ni-imido compound that is capped by the appended borane. Reactions with 1-azidoadamantane (AdN₃) provide a distinct outcome, where a proposed nickel imido intermediate activates the sp² C-H bonds of arenes, even in the presence of benzylic C-H sites. Combined experimental and computational mechanistic studies demonstrate that the unique reactivity is a consequence of Lewis-acid induced polarization of the Ni–NR bond, potentially providing a synthetic strategy for chemoselective reaction engineering.

Graphical Abstract

INTRODUCTION

In addition to steric/electronic perturbations that can be made to a metal’s primary coordination sphere, introduction of acidic/basic groups within a metal’s secondary coordination sphere can be used to augment metal-based reactivity preferences.^1–3 Secondary coordination sphere engineering is receiving increased attention from the synthetic community, largely to facilitate small molecule activation (H₂, O₂, N₂, CO₂, CO, oxyanions, etc.).^4–9 Within the context of small molecule functionalization, secondary sphere acidic groups can modify the electronic structure of metal-coordinated intermediates that ultimately can introduce divergent reactivity for complexes with/without secondary sphere acids.^{6–7, 10–22} Exogenous borane Lewis acids have been shown to influence substrate binding/activation, although strong Lewis acids are typically required.^23–24 One strategy to lower the acidity required for substrate binding is to decrease the entropic penalty of Lewis acid/substrate binding. A highlight of this approach was work by Miller and Bercaw, who demonstrated that a borane-tethered monodentate phosphine ligand enabled Re-mediated CO reductive coupling.^8,25 Related examples with bidentate phosphines are rare, and multi-borylated bis(phosphino)ethanes were recently reported by Drover and co-workers^26–29 as a strategy to promote hydride transfer.²⁷

Transition metal imido complexes are important intermediates in nitrogen fixation,^30–31 nitrene transfer,^32–34 and C-H bond amination.^35–38 Although metal imidos are less common among the late metals such as nickel, Hillhouse and coworkers made pioneering contributions using sterically encumbered bidentate phosphine (dtbpe: ^tBu₂PCH₂CH₂P^tBu₂)^39–40 and N-heterocyclic carbene (NHC) ligands⁴¹. These imides can activate the C-H bonds of phenylacetylene⁴² or ethylene,⁴¹ while they are stable in aromatic solvents (benzene or toluene). Related work by groups of Warren⁴³ as well as Betley⁴⁴ reported several formally Ni(III) imides using β-diketiminate or dipyrrin ligands whose electronic structures can also be described as Ni(II) iminyls. These compounds have been applied in stoichiometric⁴⁵ and catalytic^46–47 C-H bond aminations, and importantly, both systems activate toluene and related substrates at the homolytically weak benzylic position. By contrast, heterolytic C-H activation pathways are most commonly observed in early metal imido complexes, which are more polarized than late metal analogues, and less prone to react through homolytic pathways.^48–50 We hypothesized that the addition of a suitable borane Lewis acid to a nickel imido unit might similarly polarize the Ni-NR bond, providing an avenue to access unique reactivity.^51–52

Our group has recently explored a series of ligands containing tunable secondary coordination sphere environments with nitrogen-based and NHC ligands^{7, 15, 19, 53–56} and our previous studies showed that the appended 9-BBN (BBN = borabicyclo[3.3.1]nonyl) group provides moderate Lewis acidity, enabling unique reactivity that is facilitated by reversible borane-substrate bond formation.¹⁹ Our prior efforts demonstrated that, when coordinated to low valent metals, depe (depe = 1,2-bis(diethylphosphino)ethane) provides sufficient electron density to activate nitrogenous substrates, without precluding interactions to a Lewis acid.^7,15 Herein, we establish the synthesis and reactivity of depe derivatives containing a single appended Lewis acidic borane (Figure 1) to evaluate the extent that acidic groups modulate the reactivity of late transition metal (Ni) imidos.^57–58

Figure 1.

C-H bond activation reactivity triggered by previously reported Ni imido complexes as well as that reported in this work.

RESULTS AND DISCUSSION

Synthesis of borane-functionalized bisphosphine ligands.

Key to our synthetic approach was access to Et₂PCH₂CH₂P(Et)(Cl). Because this compound was not cleanly formed from reactions between Cl₂PCH₂CH₂PCl₂ (A) and 3 equiv. EtMgBr, we adapted a method previously used to prepare polymer-supported phosphines.⁵⁹ As shown in Figure 2, the reaction of A with 1 equiv. ⁱPr₂NH in the presence of 1 equiv. Et₃N generated a mixture of Cl₂PCH₂CH₂P(Cl)(NⁱPr₂) (B) and (ⁱPr₂N)(Cl)PCH₂CH₂P(Cl)(NⁱPr₂), from which B was isolated in 41% yield (³¹P NMR: 191.7 for -PCl₂ and 130.7 ppm for -P(Cl)(NⁱPr₂) group). The reaction of B with 3 equiv. EtMgBr afforded C (³¹P NMR: 35.2 ppm for -P(Et)NⁱPr₂ and −18.9 ppm for -PEt₂). Subsequent deprotection using HCl (2 eq)⁶⁰ revealed a new ³¹P NMR resonance at 116.7 ppm, consistent with a -P(Et)(Cl) group, and after workup, afforded D as a white solid in 53 % yield.^61–62

Figure 2.

Synthesis of 9-BBN functionalized bisphosphine ligands 1a and 1b.

To install terminal olefins amenable to hydroboration, D was treated with vinylmagnesium bromide, affording E in 66% yield (allylmagnesium bromide afforded F in 75% yield). The ³¹P NMR spectrum of E contained two resonances at −19.4 and −20.0 ppm^63–64 (−19.0 and −23.9 ppm for F). Hydroboration of E and F with 9-H-BBN in benzene at 130 °C afforded 1a and 1b as colorless oils. In contrast to the precursor, ³¹P NMR spectra of 1a exhibited resonances far removed from each other at −25.9 ppm and −2.3 ppm. The ¹¹B NMR spectrum exhibited a resonance at −8.1 ppm (−21.8/8.5 ppm for ³¹P NMR and −2.0 ppm for ¹¹B NMR of 1b), consistent with the coexistence of a free phosphine group and a quenched P→B interaction (note that 1a may exist as a higher order aggregate, but for clarity, we have depicted one structure that satisfies NMR data).^65–66

Metalation with nickel.

Metalation of 1a with NiCl₂(dme) afforded 2a as a yellow solid in 85% yield (Figure 3). The ³¹P NMR spectrum of 2a exhibited two resonances at 78.2 and 76.1 ppm, similar to that of (depe)NiCl₂ (76.9 ppm), and consistent with bidentate coordination.⁶⁷ The ¹¹B NMR spectrum exhibited a resonance at 85.5 ppm, consistent with boron in a trigonal environment. Metalation of 1b with NiCl₂(dme) gave similar results, with two ³¹P NMR resonances at 76.4 and 73.2 ppm as well as a ¹¹B NMR resonance at 87.6 ppm. Reduction of 2a (or 2b) with Mg in the presence of 10 equiv. of cyclooctadiene (COD) afforded the Ni(0) complex 3a as an oil in 95% yield (or 3b in 91% yield) as assessed by NMR spectroscopy (Figure 3). The ³¹P NMR spectrum of 3a featured resonances at 52.3 and 49.2 ppm (49.1 and 46.9 ppm for 3b), while the ¹¹B NMR spectrum contained a broad resonance at 87.7 ppm (88.0 ppm for 3b), consistent with boron in a trigonal environment.

Figure 3.

Synthesis of Ni(0)-COD complexes 3a and 3b based on ligands 1a and 1b.

Group transfer from azides to low valent metal centers is a common strategy to prepare metal imido complexes, including related Ni=NR compounds reported by Hillhouse.⁴⁰ Prior to N₂ extrusion, several metal azide complexes, either in η¹-N_γ or η²-N_β/N_γ coordination modes, have been isolated as intermediates.^68–69 To access nickel imido complexes with an appended borane and study their subsequent reactivity, we evaluated reactions between 3 and representative aryl and alkyl azides.

Synthesis of monomeric Ni(0)-azide complexes.

We evaluated reactions between 3 and two different aryl azides (MesN₃ or DmpN₃, Mes = 2,4,6-Me₃C₆H₂, Dmp = 2,6-Mes₂C₆H₃). Solutions of aryl azide were slowly added to frozen solutions of 3 in toluene-d₈ and the reaction progress was assessed by ³¹P NMR spectroscopy. Upon warming to −30 °C, 3 transformed into a new compound, 4, with concurrent generation of free COD (based on ¹H and ³¹P NMR spectroscopy), and was tentatively assigned as a Ni(0) azide species. We found that the tether length to the appended borane dramatically influenced stability. While 4a and 4c, with two methylene units (–CH₂CH₂–), were stable at −35 °C for >12 h, 4b, with three methylene units (–CH₂CH₂CH₂–), started to decompose as it was generated (ca. −30 °C). The ³¹P NMR spectrum of 4 exhibits two major resonances in a 1:1 ratio (4a: 76.5 and 47.6 ppm, J_P-P = 21.9 Hz; 4b, 57.5 and 52.9 ppm, J_P-P = 21.8 Hz; 4c, 62.3 and 39.3 ppm, J_P-P = 26.3 Hz). Although the thermal instability of 4 hampered isolation in bulk quantities, crystals suitable for X-ray diffraction studies were obtained from diethyl ether solutions at −35 °C.

The solid-state structures of 4b and 4c revealed Ni-(N₃Ar) complexes, where the azide unit is κ²-coordinated through the α- and γ-N atoms. This coordination mode is supported by an interaction between the γ-N atom and the appended boron (Figure 4. B-N3 bond length = 1.591(3) Å and ∑B_α= 325.17(19)° for 4b). In 4b, the N-N bond lengths lie between single and double bonds, with the N1-N2 bond (1.378(3) Å) longer than the N2-N3 bond (1.276(3) Å). The Ni-N1 and Ni-N3 distances range between 1.87 - 1.94 Å, which is consistent with single bonds. 4c exhibited a similar structure, although the data quality did not allow reliable discussion of metrical parameters (see SI section 6.2 for more details). Complexes 4b and 4c represent the first structurally characterized examples of monometallic complexes with a κ²(N,N) Ni(II) triazenido binding mode. Importantly, this bonding mode has been proposed by Hillhouse and co-workers as transition state en route to Ni(II) imidos from Ni(0) azide complexes.^40,70–72 Azide coordination modes through a single nitrogen (N_γ), or two nitrogen atoms (either η²-N_β/N_γ or κ²-N_α/N_γ) have previously been calculated to be within 15 kcal/mol for several (PP)Ni(RN₃) species (PP = bisphosphine ligands) and η²-N_β/N_γ coordination is the most common and stable.^73–74 The secondary sphere borane modulates the binding mode to prefer κ²-N_α/N_γ over η²-N_β/N_γ by providing significant stabilization through a B-N interaction (24.1 kcal/mol for the 2 carbon tether and 20.7 kcal/mol for the 3 carbon tether based on DFT calculations, see xyz file in SI). These results demonstrate that the Lewis acid serves an integral role to favor this unusual azide binding mode

Figure 4.

Generation of Ni(0)-azide monomer complexes, and the X-ray structures of 4b with 50% probability ellipsoids and 4c with 30%. The solvent (diethyl ether) and all H atoms were omitted and the 9-BBN is displayed in wireframe for clarity. Note that compounds 4 are only stable at temperatures <−35 °C.

Generation of Ni(II) imido complexes.

Although monomeric Ni(0) azide complexes, such as 4a and 4b can be generated at low temperature, they underwent subsequent thermally-induced N₂ extrusion at higher temperatures. For example, the thermal decomposition of 4a generated a major product 5a in ca. 75% chemical conversion (vs. free COD; see SI section 2.15). The ³¹P NMR spectrum exhibited resonances at 85.0 and 45.6 ppm, while the ¹¹B NMR spectrum exhibited a single sharp resonance at −8.4 ppm, consistent with boron in a tetrahedral environment. The Ni=NR unit of 5a was tentatively confirmed by its reactions with CO or ^tBuNC, which formed MesNCO or ^tBuNCNMes, as assessed by GCMS analysis and ¹³C NMR spectroscopy (see SI section 2.16). Although solutions of 5a decomposed in less than 1 day to give unidentified mixtures, single crystals of 5a were grown by slow vapor diffusion of hexane into a concentrated solution of 5a in toluene at −35 °C. Analysis of single crystal X-Ray Diffraction (scXRD) of 5a revealed a Ni imido species with a B–NAr interaction (Figure 5). The Ni-N bond length (1.897(2) Å) is longer than that in Hillhouse’s imido complex [(dtbpe)Ni(NMes)] (1.703(4) Å) but comparable to those in a tetranuclear Ni imido cluster [(ⁱPr₃P)₄Ni₄(μ₄-NCH₂Ph)] (1.87-1.90 Å).⁷⁵ Similar to redistributing charge over more than one metal site, we propose that the B-N interaction (B-N bond length 1.567(3) Å) elongates the Ni–N bond in a manner that resembles a bridging mode. The Ni-N bond length may reflect the continuum between nickel imido resonance structures.^76–77 The structure of 5a also features a π-arene interaction (Ni-C40 and Ni-C41 bond lengths = 2.038(2) Å and 2.114(2) Å, respectively). This interaction imposes a bent Ni-N1-C40 bond (bond angle 75.40(14)°), which is distinct from that in Hillhouse’s imido complex (bond angle = 180°).⁴⁰ Another notable difference between complex 5a and Hillhouse’s (dtbpe)Ni=NAr complex is the coordination number at Ni: the latter is 3-coordinate. In contrast to the geometry imposed by steric interactions, we attribute the different local geometry in 5a primarily to the tridentate chelate formed by the B-N bond, which positions the imido N linearly to one phosphine (P2-Ni-N bond angle: 164.61(7)°).

Figure 5.

Generation of Ni(II) imido complex 5a and Ni(II) borylamide complex 5b, and X-ray structures (50% probability ellipsoids, selected H atoms omitted).

We found that the tether length to the appended borane has large effects on reactions with aryl azides. Although 4b exhibited a similar thermal stability profile to 4a, the reaction products following N₂ extrusion were unique. ScXRD experiments of 5b indicated that rather than a nickel imido complex, B–C cleavage occurred, forming a nickel alkyl borylamide complex (Figure 5). In 5b, the boron and nitrogen are both trigonal planar (∑B_α = 359.99°, ∑N_α = 359.84°), which is similar to a previous reported Ni(II) borylamide complex (∑N1_α = 358.61° and ∑N2_α = 358.40°).⁷⁸ Formation of 5b is consistent with a Ni intermediate that is unstable with respect to the B-C bond. Thus, we propose that ligand 1a provides better stabilization of a Ni imido species compared with ligand 1b. Following these observations, subsequent reactions with azides were only pursued using 3a.

In order to provide a sterically encumbered environment surrounding the imido moiety, we evaluated reactions with dimesitylphenyl (Dmp) azide. When a toluene-d₈ solution of 4c was warmed above 0 °C, a complex mixture formed that slowly transformed into major product 5c (42% isolated yield) after 16 h at room temperature, as assessed by ³¹P NMR spectroscopy (62.7 and 60.5 ppm). An X-ray diffraction study revealed a Ni(II) amido, rather than imido, with a cyclometalated mesityl ligand (Figure 6). Although a Ni(II) imido [(PPB)Ni=NDmp] may be implicated as an intermediate during the transformation of 4c to 5c, it was not isolable. Similar Ni-mediated benzylic C-H activation of the Dmp group has been reported by Hillhouse and co-workers following single-electron oxidation of an NHC-supported nickel amide.⁷⁹ In the solid-state structure of 5c, the Ni-amido unit is coordinated to the appended borane (B-N1 bond length: 1.669(4) Å; ∑B_α = 327.6(3)°). The Ni-N bond length (2.032(3) Å) is ca. 0.135 Å longer than that in complex 5a, and similar to those in related Ni-amido complexes that feature Lewis acid interactions to the amido nitrogen atoms.⁸⁰

Figure 6.

Generation of Ni(II) benzyl amide complex 5c as well as its structure (50% probability ellipsoids, selected H atoms omitted).

Studies toward the generation and reactivity of a Ni(II) imido complex with an adamantyl substituent.

To provide a Ni=NR unit that is able to react with exogenous substrates, we targeted an organoazide that cannot interact with nickel through a π system, and is unlikely to undergo intramolecular H-atom transfer. We selected AdN₃ for subsequent reactivity studies because it fulfills both of these criteria. When we allowed 3a and AdN₃ to react in benzene, the ³¹P NMR spectrum of the product exhibited resonances at 46.8 and 42.7 ppm (J_P-P = 25.8 Hz). Following isolation as a yellow solid in 65% yield, a scXRD experiment was performed, enabling assignment as a Ni phenyl amido complex 6a (Figure 7). Similar to 5c, the amido unit features a B-N interaction (1.658(4) Å) with the appended borane Lewis acid. To interrogate the intermediate formed with AdN₃, we quenched the reaction of 3a and AdN₃ with ^tBuNC in THF solvent and observed ^tBuNCNAd (see SI, section 2.25). This reactivity profile is consistent with imide character of the intermediate derived from AdN₃. Collectively, the formation of 6a may indicate that a transiently formed nickel imido species undergoes subsequent C-H bond activation of a benzene solvent molecule.

Figure 7.

Arene activations by proposed Ni(II) imido complexes as well as the structures of 6a and 6b (only one isomer was shown) with 50% probability ellipsoids, with selected H atoms omitted.

The reactivity observed with benzene is atypical for late metal imidos, which most commonly activate C-H bonds that have lower BDE values.^{37–38, 44–45, 81} We hypothesized that we might capitalize on this result, perhaps by favoring distinct regioselectivity to activate sp²− rather than sp³ C-H bonds. Toluene and m-xylene contain both weak benzylic sp³ (C-H BDE: ca. 90 kcal/mol) and strong sp² C-H bonds (~20 kcal/mol higher).⁸² When complex 3a was allowed to react with AdN₃ in toluene at room temperature for 16 h, we observed a mixture of products, as assessed by ³¹P NMR spectroscopy. To clarify the composition, we quenched the reaction with CH₃COOD and analyzed the deuterium position by ¹H NMR spectroscopy. From this experiment, we found that the reaction afforded a 1.7 : 1 ratio of m- and p-CH₃C₆H₄D (See SI section 2.22), with no ortho-deuterated products. Importantly, we did not observe any C-H activation products of the sp³ C-H bonds. Based on these results, we assign the ³¹P NMR spectrum as two isomers of meta and one of para with resonances between 40-45 ppm.⁸³ The structural assignment was validated by scXRD (Figure 7). The bond metrics of 6b are similar to those of 6a.⁸⁴ Similar to toluene, we found that reactions with m-xylene, 3a, and AdN₃ generated the meta-activation product, which was isolated in 51% yield. Distinct from toluene, we observed only one pair of doublets in the ³¹P NMR spectrum. Complex 6c was characterized by NMR (¹H, ¹³C, ³¹P, ¹¹B) spectroscopy and scXRD (see SI sections 2.24 and 6.8). The regioselectivity profile of these reactions is distinct from known monometallic nickel imide complexes.^{44–47,85–86} The high BDE values of arene sp² C-H bonds likely preclude a homolytic mechanism, therefore we experimentally investigated the C-H bond activation step.

To provide a mechanistic understanding for the observed C-H bond regioselectivity, we performed a series of experimental and computational studies. Most late transition metal imido complexes operate through H atom transfer (HAT) that exhibit large primary KIE values (>5-13^87–88), while a 1,2-addition mechanism has been proposed by a dinickel imido complex which exhibits a similar sp² C-H regioselectivity (KIE = 4.3⁸⁹ and 5.1⁹⁰). We performed parallel kinetic experiments with C₆H₆ and C₆D₆ and product formation was tracked at 10°C using ³¹P{¹H} inverse-gated NMR spectroscopy to measure the initial rates. These experiments afforded a k_H/k_D value of 3.22, which is within a reasonable range for a classical primary KIE and suggestive of little tunneling.

To examine the charge transfer properties of the transition state of C-H activation, we performed a Hammett-type study. To simplify analysis, we used 1,3-disubstituted arenes because they: a) form a single C-H activation product (vide supra), and b) ensure that primarily inductive substituent effects contribute to the reaction. Solutions of 3a in THF were mixed with 1 equiv. of AdN₃ and a 1,3-disubstituted arene in 1:1 volumetric ratio (see SI section 3.1). Spanning 2σ_m-values from 0.86 to −0.42, we observed a linear correlation (R²=0.94) with positive slope, where arenes with electron-withdrawing substituents proceed with faster rates (Figure 8). These results are similar to those from studies that proposed a σ-bond metathesis or oxidative hydrogen migration mechanism in aromatic C-H activation.^91–92

Figure 8.

Hammett-type plot of observed initial rate of C-H activation on 1,3-disubstituted arenes against 2σ_m.

We assessed whether an intramolecular, compared to intermolecular, Lewis acid is required to mediate the C-H bond activation pathway. Through a series of control reactions, we repeated conditions noted above that afforded C-H activation of toluene, then assessed deuterium incorporation after quenching with CD₃COOD. To probe the requirement of the borane Lewis acid, we independently prepared a borane-free variant, (depe)Ni(COD) (3c), as a suitable control complex because of the similar steric and electronic properties to 3a (³¹P NMR spectra of 3c and 3a are similar; 49.8 ppm vs. 49.1 and 46.9 ppm, respectively). To probe the requirement of the tethered Lewis acid, we prepared a borane with a similar steric and electronic profile as that found in 3a. 9-octyl-9-BBN exhibits similar steric and electronic environments (¹¹B NMR: 88.1 ppm vs. 87.7 ppm for 3a). When each control reaction was subjected to identical conditions, we observed distinct results. Compound 3a afforded aryl C-H activation (Figure 9). In contrast, we observed no deuterium incorporation when (depe)Ni(COD) was used in place of 3a, either with or without exogenous 9-octyl-9-BBN Lewis acid (Figure 9). In both cases, we found that ³¹P NMR resonances attributable to (depe)Ni(COD) decrease after introducing AdN₃. These results are consistent with either: 1) no formation of a Ni=NAd intermediate with a depe ligand, or 2) a Ni=NAd intermediate that does not react with toluene. The latter point is consistent with prior reports where a related compound, (dtbpe)Ni=NAd, is stable to benzene-d₆.^40,93 Importantly, these control experiments demonstrate that the unique C-H bond reactivity of 3a is derived from the intramolecular borane within the ligand scaffold.

Figure 9.

²H NMR spectra of toluene-d₈, C-H activation experiments with 3a, 3c, 3c in conjunction with 9-octyl-9-BBN.

To further clarify the role(s) that the appended Lewis acid serves to modify the electronic properties of the intermediates across the C-H activation pathway, we performed DFT computational analysis (PBE-D3/6-31G(d,p)//SDD(Ni)). Geometry optimization of a truncated nickel imido species (tBu instead of Ad) converged to structure 7a (Figure 10, bottom center). In the absence of an added solvent molecule, the optimized structure has a T-shaped geometry (P1-Ni-N angle = 164.7°).^94–96 The Ni-N bond length (1.776Å) is longer than prior reported Ni-imido complexes (1.673(4) Å in (dtbpe)Ni(NAd)),⁴⁰ and closer to that of a cationic nickel amido complex reported by Hillhouse (Ni-N = 1.771(4)Å).⁹⁷ Furthermore, the optimized structure has a Ni-N-C bond angle of 117.4°. Strongly bent imido ligands have been attributed to: 1) the chelate effect, 2) lone pair character, or 3) triplet nitrene contribution on the nitrogen atom.^76,98–99 We sought to investigate the contribution(s) of the appended borane to influence both the geometry and the electronic structure of the nickel imido moiety.

Figure 10.

NBO analysis of (dmpe)NiNtBu (7b) and 7a with a comparison in structural parameters such as natural charge difference between Ni and N (Δq_(Ni-N)), Ni-N bond length and order (BO_(Ni-N)).

Given the atypical primary sphere environment of the nickel imido unit in 7a, we probed whether the appended Lewis acid primarily serves a structural role to perturb the primary sphere geometry by forming a chelate to the imido unit. We undertook a Walsh-type analysis to examine the relationship between P1-Ni-N/Ni-N-C angles and the structural and electronic parameters of the nickel imido unit, separately from the influence of Lewis acid. We performed relaxed potential energy surface scans with constraints on P1-Ni-N (7b to 7c; from 134° to 165°) and Ni-N-C (7b to 7d from 171° to 117°) angles. Finally, we optimized a structure with identical P1-Ni-N and Ni-N-C angles to 7a (7e). During these angular perturbations, we observed modest polarization and lengthening of Ni-N bond (see Table S16–S18). 7e exhibits a larger natural charge difference between nickel and nitrogen (Δq_(Ni-N) of 0.430) than 7b (Δq_(Ni-N) = 0.172). Similarly, the Ni-N bond elongates from 1.660 Å to 1.728 Å while the Wiberg bond order decreases from 1.436 to 1.214. Collectively, these results show that modest changes on the Ni imido unit are imposed as a result of purely structural changes to the Ni-primary sphere geometry.

We found that the electronic and structural parameters of 7a were influenced by the addition of a borane Lewis acid to a greater extent than could be imposed by the primary sphere modifications above. We compared the computed structures of the T-shaped Ni imido unit where an appended BBN Lewis acid is either present (7a) or absent (7e). When the BBN Lewis acid is present, the Δq_(Ni-N) increases to 0.622 (larger than 7e by 0.192). Concomitant with this polarization are changes to the Ni-N bond (lengthening from 1.728 Å (7e) to 1.782 Å) and the Wiberg bond order (decreasing from 1.214 (7e) to 0.776). We also calculated a solvent-coordinated intermediate to provide an approximation of 7a in benzene solution (7a-C₆H₆). The solvent adduct optimized as a σ complex slightly higher in energy than 7a (6.8 kcal/mol) with a more polarized (Δq_(Ni-N = 0.716) and slightly lengthened Ni-N bond (1.786 Å). Both 7a and 7a-C₆H₆ exhibit the longest Ni-N bonds in this series, which is consistent with an increase in N-basicity.¹⁰⁰ Analysis on (dmpe)NiNMes and a truncated model of 5a show a consistent pattern in bond polarization (see Table S19). Overall, results of the calculations support our proposal that the Lewis acid redistributes charge at the nickel imido unit, rendering the nitrogen more basic. This proposal is further bolstered by related work where alkali metal additives were shown to both lengthen and polarize the Fe-N bond of an iron bis(NHC) imido complex.⁵²

Following C-H bond activation, we evaluated whether a subsequent C-N bond could be induced to form via reductive elimination, using 6a as a model compound. In contrast to previous related reports that showed thermal- and oxidation-induced reductive elimination, we did not observe any reductive elimination products when 6a was: 1) heated to 110 °C, 2) treated with I₂ and FcPF₄ oxidants, or 3) combined with a π-accepting ligand such as CO. We propose that the B–N interaction between the BBN group and amide prevents reductive elimination in this system. To overcome this challenge, we hypothesized that a strong Lewis base might competitively bind the BBN Lewis acid, thus rendering the Ad-NH unit available for C-N reductive elimination. When 6a was treated with a combination of 9-azajulolidine and sodium t-pentoxide¹⁰¹ we observed the reductive elimination product, N-phenyl-N-adamantyl amine in 26% yield (Figure 9).^102–103 These results represent a proof of principle that reductive elimination can be induced using a suitable set of reagents.

CONCLUSION

In summary, we have reported the preparation of a bidentate phosphine containing a single appended borane Lewis acid. This ligand complements the suite of commonly used bidentate phosphine ligands, such as depe, and it’s fully borylated counterparts.^26–28 The single secondary sphere borane provides access to unique reactivity for reactions between Ni(0) and organoazides. Although κ²(N,N)-coordinated Ni(0) azide complexes were stabilized by both ligands 1a and 1b, only the scaffold of 1a was stable during subsequent reactions, illustrating a clear effect of Lewis acid tether length. Nickel azide complexes supported by 1a afford Ni imido complexes that exhibit diffeent reactivity profiles when the N-substituent is either aryl or alkyl. The aryl products were isolable with mesityl, while bulkier substituents underwent intramolecular C-H activation. The intermediate derived from AdN₃ underwent 1,2-addition reactions with exogenous arenes: a reaction pathway atypical of late metal imidos. For all compounds featuring a Ni-NR unit, we observed Lewis acid/base interactions to the nitrogen atom. Through experimental and computational analyses, these secondary B–N interactions impart three important consequences during reactions of Ni(0) with organoazides, they: 1) improve stability of reaction intermediates, 2) influence the geometry by forming chelates, and 3) polarize the Ni–NR bond, which enables a rare example of monometallic late metal sp² C-H activation favored over benzylic sp³ C-H bonds. These conclusions contribute several important principles that can be used to provide regioselectivity control over C-H bond activation pathways at first row, late-metals. Efforts to apply these principles to related organometallic reactions and ultimately develop catalytic versions, including C-H bond amination are currently underway in our lab.

Figure 11.

Base-induced reductive elimination of 6a.