Reactivity of (bi-Oxazoline)organonickel Complexes and Revision of a Catalytic Mechanism

Abstract

Bi-Oxazoline (biOx) has emerged as an effective ligand framework for promoting nickel-catalyzed cross-coupling, cross-electrophile coupling, and photoredox-nickel dual catalytic reactions. This report fills the knowledge gap of the organometallic reactivity of (biOx)Ni complexes, including catalyst reduction, oxidative electrophile activation, radical capture, and reductive elimination. The biOx ligand displays no redox activity in (biOx)Ni(I) complexes, in contrast to other chelating imine and oxazoline ligands. The lack of ligand redox activity results in more negative reduction potentials of (biOx)Ni(II) complexes and accounts for the inability of zinc and manganese to reduce (biOx)Ni(II) species. On the basis of these results, we revise the formerly proposed “sequential reduction” mechanism of a (biOx)Ni-catalyzed cross-electrophile coupling reaction by excluding catalyst reduction steps.

Since their emergence in nickel-catalyzed reactions, 2,2′-linked bioxazoline (biOx) ligands have achieved extensive success in promoting Lewis acid-catalyzed Diels–Alder,¹ asymmetric cross-coupling,^2,3 cross-electrophile coupling,^4-6 and photoredox-nickel dual catalytic reactions.^7-11 The reactivity of biOx appears complementary to bis-oxazoline (box).^12,13 For instance, an asymmetric diarylation of vinylarenes employs an isoleucine-derived biOx ligand, whereas box ligands gave no desired product (Scheme 1A).¹⁴ The elucidation of the organometallic steps comprising the catalytic cycle would facilitate the understanding of the overall mechanism and the integral effect of the ligand. In contrast to the recent organometallic characterization of (box)Ni complexes by Fu and coworkers,¹⁵ the fundamental reactivity of (biOx)organonickel complexes remains ill-defined.¹⁶ The only precedent is has not been related to modern cross-coupling reactions.¹⁷

Scheme 1.

Application of biOx Ligands in Nickel-Catalyzed Cross-Coupling Reactions

A seminal application of (biOx)Ni catalysts is the enantioconvergent coupling of secondary benzyl chlorides with iodoheteroarenes developed by Reisman and co-workers (Scheme 1B).⁴ On the basis of a recent mechanistic study on (phen)Ni (phen = 1,10-phenanthroline) catalyzed cross-electrophile coupling,¹⁸ we postulated that this reaction could follow a “sequential reduction” pathway, involving the reduction of Ni(II) intermediates to Ni(0) and Ni(I) species that subsequently activate aryl and alkyl electrophiles (Scheme 1B).¹⁹ It remains indeterminate if biOx would alter the reaction mechanisms compared to phen and other common N-ligands. In this report, we establish the fundamental organometallic reactions of (biOx)Ni complexes. The results verify the plausibility of some proposed elementary steps but contradict others. These findings form the basis for a revision of the catalytic cycle of reductive coupling reactions involving biOx ligands.

This report also addresses the redox activity of biOx in stabilizing Ni(I) intermediates, which are prevalent in Ni-catalyzed cross-coupling reactions.^20-24 Ni(I) intermediates are commonly supported by imine, oxazoline, and pyridine ligands.^25,26 As strong σ-donors and π-acceptors, bis-(oxazolinyl)pyridine (pybox),²⁷ α-diimine,^28-30 and pyridyl oxazoline (pyrox)³¹ proved redox active when stabilizing low-valent nickel radical species by delocalizing the unpaired electron into the ligand π* orbital (Figure 1).^32,33 A (^Ph box)Ni(Br) complex that lacks π-conjugation between the two oxazoline rings accommodates a Ni-centered radical.¹⁵ It remains unclear whether the conjugation between two oxazoline motifs in biOx would enable redox activity. In this study, characterization of a monovalent (biOx)Ni complex elucidates the redox activity of biOx, which contributes to the unusually negative redox potentials of (biOx)Ni complexes and leads to alternative mechanistic pathways in catalysis.

Figure 1.

Low-valent nickel radical complexes bearing chelating oxazoline and imine ligands and their electronic structures.

We selected ^iPrbiOx as the model ligand for the synthetic study of biOx-ligated organonickel complexes (Scheme 2). Addition of neosilylmagnesium chloride (TMSCH₂MgCl) to a mixture of Ni(acac)₂ (acac = acetylacetonate) and ^iPrbiOx produced (^iPrbiOx)Ni(CH₂TMS)₂ 1 in 93% yield. Following a literature protocol,¹⁷ we prepared (^iPrbiOx)Ni(Dipp)Br (Dipp = 2,6-diisopropylphenyl) 2, (^iPrbiOx)Ni(Mes)Br (Mes = 2,4,6-mesityl) 6, and (^diMebiOx)Ni(Dipp)Br 7 by transmetalation of Grignard reagents. The diamagnetic ¹H NMR spectra of 1, 2, and 6 (Figures S6-S8, and S12), in conjunction with their square planar geometry, evident from single crystal X-ray structures (Figures S36-S37), suggest that the d⁸ Ni(II) complexes adopt a low-spin configuration. Significant line broadening in the ¹H NMR spectrum of 7 implies some geometry distortion, likely due to the steric strain (Figure S13). Mixing biOx ligands with NiBr₂(DME) afforded (^iPrbiOx)NiBr₂ 3 and (^diBnbiOx)NiBr₂ 8. Complex 8 shows a dimeric structure of [(^diBnbiOx)NiBr(μ-Br)]₂ in the solid state (Figure S40). Addition of LiN(TMS)(Dipp) to 3 forms 4, whereas the bulkier ^diMebiOx gave 5 instead. The paramagnetic ¹H NMR spectra (Figures S10-S11) and tetrahedral geometries for 4 and 5 , established by X-ray structures (Figures S38-S39), indicate high-spin d⁸ Ni(II) configurations caused by weak-field amido ligands.

Scheme 2.

Syntheses of (biOx)Ni(II) Complexes

Access to complexes 1-8 provided a platform for investigating elementary reactions that possibly comprise the catalytic cycle in Scheme 1B. Reduction of 8 by KC₈ in THF resulted in formation of a teal solution, displaying no signals in the ¹H NMR spectrum (Scheme 3A). X-ray crystallography establishes a dimeric [(^diBnbiOx)Ni(μ-Br)]₂ 9’ structure with an imine bond (C(1)─N(1)) length of 1.285(6) Å (Figure 2) almost identical to that of 8 (1.273(4) Å) (Figure S40). The EPR spectrum of 9 at 10 K reveals a rhombic signal with g values of [2.486, 2.175, 2.062] (Figure 3A).^34,35 Data simulation ascribed the hyperfine couplings of the radical to two nitrogen atoms, assigned to the two imine nitrogen atoms of biOx. The double integration of the EPR signals against a Cu external standard established monomeric Ni(I) as the predominant species (99%) in solution. The minimal imine bond elongation in the crystal structure and the observation of a nickel-centered radical by EPR spectroscopy together suggests the lack of redox activity of biOx in 9.

Scheme 3.

Organometallic Transformations of (biOx)Ni(I), (II), and (III) Complexes

Figure 2.

X-ray structures of 9’ at 50% probability thermal ellipsoids. Hydrogen atoms are omitted and phenyl groups are truncated for clarity. Selected bond distances (Å): C(1)─N(1) = 1.285(6), C(4)─N(2) = 1.273(6), Ni─Br = 2.4560(8), Ni─Ni = 3.4620(12).

Figure 3.

X-Band EPR spectra (black) of 9 (A) and 10 (B) at10 K. The simulated spectra (red) use following parameters: 9, g = [2.486, 2.175, 2.062], A_N,N = [107, 210, 0; 96, 148, 0] MHz; 10, g = [2.492, 2.179, 2.070], A_N,N = [218, 31, 54; 218, 29, 50] MHz.

Reduction of 7 by KC₈ formed a golden-brown species 10 (Scheme 3A). The EPR spectrum of 10 indicates a nickel-centered radical with g values of [2.492, 2.179, 2.070] and hyperfine features corresponding to the interaction of the radical to the two nitrogen atoms on the ligand (Figure 3B). We performed X-ray absorption spectroscopy (XAS) experiments on 10 and a series of Ni(I) and Ni(II) reference complexes. The pre-edge energies in the X-ray absorption near edge structure (XANES) span 8.3313 to 8.3346 keV across ten reference samples (Table S1) but do not show apparent correlation to the oxidation states.³⁶ We then transformed the data to k² weighted χ and analyzed by curve fitting of the extended X-ray absorption fine structure (EXAFS) spectra. The position of the features in the Fourier transform magnitude along the coordination shell radius (R) indicates the scattering pairs present, and the relative intensity corresponds to the coordination number. The comparison of 10 with 7 clearly indicates the dissociation of the bromide upon reduction, whereas the number of Ni─N bonds remains the same (Figure 4). Based on the above data, we assign this reduced species to (^diMebiOx)Ni^I(Dipp) 10. DFT calculations suggest that (^iPrbiOx)Ni^I(Dipp) is a nickel-centered radical (Figure S41).

Figure 4.

Fourier transforms of k² χ (k) EXAFS functions of 7 (red) and 10 (blue)measured at the Ni─K edge.

Both 9 and 10 are nickel-centered radicals, evinced by EPR spectroscopy and DFT calculations. The lack of redox activity of biOx, in contrast to α-diimine,²⁹ stems from the mesomeric effect of the oxygen atoms of oxazoline that increase the electron-density in the π orbitals. In the absence of the redox-active ligand stabilization, E[Ni(II/I)] of (biOx)Ni is noticeably more negative than complexes stabilized with redox-active bpy and phen ligands (Table 1).³⁷ Consequently, low-valent (biOx)Ni species cannot be generated from the reduction of 7 and 8 by manganese or zinc powder, although Rieke’s manganese and zinc can reduce 7, due to an inner-sphere mechanism instead of an outer-sphere electron reduction (cf. SI).³⁸

Table 1.

Comparison of E[Ni(II/I)]

Complexes	Redox-activity	E1/2[Ni(II/I)] (V vs Fc+/Fc)
(Phen)Ni(Mes)Br39	Yes	−1.50
(bpy)Ni(Mes)Br40	Yes	−1.79
7	No	−2.12a
(6,6′-Mebpy)NiBr241	Yes	−1.26
8	No	−1.55a
3	No	−1.46a

^aE_P/2.

We then explored the activation of electrophiles by monovalent nickel complexes. (^diBnbiOx)NiBr 9 proceeded to activate iodomesitylene to afford (^diBnbiOx)Ni(Mes)Br 11 in 21% yield (Scheme 3B). Two possible pathways could account for this reactivity: iodomesitylene may oxidatively add to 9, followed by Mn reduction of Ni(III) or iodide radical ejection to afford 11; alternatively two molecules of 9 could engage in bimolecular oxidative addition.⁴² These pathways are not discerned at this point. While 9 can also activate 1-chloroethylbenzene 12, the rate is slower than that of iodomesitylene (cf. SI), consistent with a previous report that distinguishes the “radical chain” mechanism from the “sequential reduction” mechanism.¹⁸ (^diMebiOx)Ni(Dipp) 10 reacted with 12 to give 13 and 14 in 26% and 50% yields, respectively. The inability of manganese to reduce 7 prompted us to postulate that alternative mechanisms were involved in the activation of 12. Control experiments reveal that 12 can readily dimerize in the presence of manganese and TMSCl,⁴³ while the rate is increased by Ni(II) complexes, including 3 and 7 (Scheme 3B). Previous studies reveal that Lewis acids, such as nickel salts⁴⁴ and ZnCl₂,⁴⁵ can promote the activation of allylic bromides and chlorides by manganese and TMSCl. Therefore, we attribute the formation of benzyl radical from benzyl chloride to the cooperative reactivity of manganese and TMSCl, catalyzed by Ni(II) species. This mechanism, involving manganese and TMSCl, complements the reactivity of Ni(I) complexes in C(sp³) electrophile activation⁴⁶ but is restricted to benzyl halides, evident by the inactivity of chlorocyclohexane (cf. SI).

In order to probe the interaction of radicals with nickel species, we prepared 4-benzyl-1,4-dihydropyridine 15, a known radical precursor upon irradiation.⁴⁷ While 1 was stable under 390 nm light irradiation, addition of 15 led to C─C bond formation to furnish 16 and 17 (Scheme 3C). The concomitant formation of 16 and 17 supports the emergence of Ni(III) 18 by radical capture, in which both neosilyl and benzyl groups are coordinated to nickel, followed by rapid reductive elimination.

Finally, we conducted the oxidation of 1 by FcPF₆ (Fc = ferrocenium) (Scheme 3D). The blue solution of 1 immediately reacted to afford 17 in 97% yield (Scheme 3D). We propose that the oxidation of 1 resulted in Ni(III) 19 as an intermediate, followed by rapid reductive elimination.

The above data prompted us to reassess the initial mechanistic proposal shown in Scheme 1B (Scheme 4). Stoichiometric experiments of 9 establish that (biOx)Ni(X) can activate iodoarenes to afford (biOx)Ni(Ar)(X) intermediates (Scheme 3B). In contrast to reactions catalyzed by (phen)Ni complexes,¹⁸ (biOx)Ni(Ar)(X) is not reduced by zinc or manganese powder, due to its unusually negative reduction potential (Scheme 3A). Instead, benzyl chloride is activated by manganese and TMSCl, promoted by Ni(II) species. The resulting benzyl radical can be captured by (biOx)Ni(Ar)(X) to form a Ni(III) intermediate, followed by rapid reductive elimination to afford the cross-coupling product (Scheme 3C,D). Compared to the original proposal, the new mechanism excludes any catalyst reduction step. The radical capture step by Ni(II) species resonates with its participation in photoredox catalytic reactions by trapping photochemically generated radicals.⁷

Scheme 4.

Revised Mechanism of (biOx)Ni-Catalyzed Cross-Electrophile Coupling Reaction

In summary, we investigated the fundamental reactivity of a series of organometallic (biOx)Ni complexes, including reduction, electrophile activation, radical capture, and reductive elimination. BiOx lacks redox-activity in stabilizing Ni(I) complexes, in contrast to other common oxazoline, imine, and pyridine chelating ligands. This attribute results in more negative reduction potentials of Ni(II) species and prevents their reduction by zinc or manganese. As a result, we revise the mechanism of (biOx)Ni-catalyzed cross-electrophile coupling reactions, by invoking direct radical formation from the activation of alkyl chlorides by TMSCl and manganese. In this scenario, nickel catalysts serve to activate iodoarenes, capture the radical generated from alkyl chlorides, and mediate C─C bond formation, but are not reduced by manganese.