Modulation of Nickel Pyridinedicarboxamidate Complexes to Explore the Properties of High-valent Oxidants

Graphical Abstract

High-valent nickel: The synthesis of a family of Ni^II and Ni^III complexes supported by a common 2,6-pyridinedicarboxamidate ligand is described. The complexes display coordination numbers of 4, 5, and 6 and different sets of N- and O-atom donors. The coordination environment allows for tuning of the magnetic and spectroscopic (UV/Vis, NMR, EPR) properties of the complexes and provides insights into the oxidising power of high-valent Ni.

Abstract

High-valent Ni oxidants have been implicated in hydrocarbon oxidation catalysis, however, little is understood about the properties of these oxidants. Herein, a family of Ni^II complexes supported by a pyridinedicarboxamidate ligand and different ancillary ligands was synthesized. The series spans coordination numbers 4, 5, and 6, and contains neutral, mono- and di-anionic donor types. X-ray crystallography and magnetic measurements showed that the 4-coordinate complexes were square planar and low spin (S=0) and the 5- and 6-coordinate were intermediate spin (S=1). The Ni^II complexes could be oxidized by one electron to form a series of metastable Ni^III species. EPR analysis confirmed their description as S= Ni^III compounds with signal shape and hyperfine coupling dependent on the coordination environment. The oxidation of phenols by the Ni^III species was probed, providing evidence for a correlation between oxidizing power and electron-donating properties of the supporting ligands. Critically, we found that the pyridinedicarboxamidate ligand may be a non-innocent proton acceptor in the oxidation reactions.

Introduction

High-valent transition metal species are important intermediates in a multitude of biological and synthetic oxidations. Numerous homogeneous and heterogeneous oxidation catalysts based on Ni have been developed for reactions such as alkene epoxidation,1 alkane2 and arene3 hydroxylation and chlorination,4 and water oxidation.5 Terminal Ni^III=O and Ni^IV=O are frequently invoked as intermediates in these reactions, and their existence is supported by mechanistic experimental6 and computational studies,7 as well as by model complexes.8

While terminal M=O species have attracted the most focus, a slew of other nickel oxidants have been proposed, and in some cases demonstrated to be effective oxidants.9 These include Ni^II(O₂^.),10 Ni^III(O₂),10d, 11 Ni^II(OOR),11c, 12 Ni^IV−OH,13 Ni^II−OOCOR,14 Ni^IV−OCl,9 [Ni^III₂(μ-O)₂],10c, 15 and [Ni^III(μ-O)₂Cu^III].16 Ni^III−OX species have been shown to mediate the formation of highly reactive, metal-free Cl^. radicals in chlorination reactions.4 Complexes with imides, Ni^III=NR, also constitute powerful oxidants.17 This large family of nickel-based oxidants display a variety of coordination numbers and geometries, without any clear understanding of their structure/function relationships.

In light of this structural variety of nickel-centered oxidants, we aimed to explore the systematic variation of the coordination environment around the metal center. Pyridinedicarboxamidate ligands constitute a versatile support, which is effective in the stabilization of high valent nickel: hexacoordinate [Ni^II/III/IV(pyN₂^H2)₂]^−2/−1/0 complexes have been reported (pyN₂^H2=N,N′-diphenyl-2,6-pyridinedicarboxamidate).18 We have previously prepared square planar [Ni^II/III(OX)(pyN₂^Me2)] complexes (OX=OCO₂H, O₂CCH₃, ONO₂; pyN₂^Me2=N,N′-bis(2,6-dimethylphenyl)-2,6-pyridinedicarboxamidate)19 and demonstrated their reactivity as hydrogen atom transfer (HAT) oxidants towards O−H and C−H bonds. As a means of exploring structure/function relationships, we were interested in the role the ancillary ligands play in modulating the HAT reactivity of the high-valent Ni species. Furthermore, we were interested to understand the role the pyridinedicarboxamidate ligand plays in oxidation reactions. As the ligand has multiple potential proton-acceptor sites, we wanted to understand its role in HAT reactivity. We report herein the preparation of an extended and unique family of Ni^II and Ni^III complexes derived from the [Ni(pyN₂^Me2)] core and explore the reactivity of the Ni^III entity (with varying neutral and anionic donors, in different coordination environs) in phenol oxidation.

Results and Discussion

A family of [Ni^II(L)(pyN₂^Me2)] complexes (1 a-8 a, Scheme 1) was prepared (see the Supporting Information for details). To form previously reported 2 a,22 substitution of the neutral CH₃CN ligand in the previously reported 1 a (Scheme 1) was performed using Et₄NCl.19a, 20 For 3 a-5 a, 2 a was initially generated in a reaction vessel without being isolated, and subsequently reacted with KO₂CH (1.2 equiv), NaOC₆H₅ (1.2 equiv), or Na(acac) (1.05 equiv, acac=acetylacetonate), respectively. Yields of 76 % (3 a), 68 % (4 a), and 55 % (5 a) were obtained after work-up and re-crystallization. These synthetic methods improved the previously reported synthesis of 3 a, whose preparation from [Ni(OH)(pyN₂^Me2)]Et₄N required longer reactions time and displayed modest yield.21

Scheme 1

Synthetic routes to the Ni^II complexes 2 a–8 a. L*=N,N′-bis(2,6-dimethylphenyl)-2,6-pyridinedicarboxamidate. (a) Et₄NCl.H₂O (Et=C₂H₅), CH₃CN; (b) KO₂CH, CH₃OH; (c) NaOC₆H₅, CH₃OH; (d) Na(acac), DMF (DMF=dimethylformamide); (e) bpy, CH₃CN; (f) terpy, CH₃OH; (g) Li₂dipic, CH₃OH. See Supporting Information for details.

It is important to note that the synthesis of 5 a was conducted under strictly anhydrous conditions. Early synthesis attempts under an air atmosphere yielded an impure product, which after re-crystallization was identified, by ¹H NMR, to contain the acetate complex [Ni^II(OAc)(pyN₂^Me2)]^− [19a] and 5 a. Furthermore, an X-ray diffraction (XRD) measurement on crystals obtained from these impure reaction conditions showed the presence of 1:1 co-crystals of 5 a and [Ni^II(OAc)(pyN₂^Me2)]⁻ (Figure S1). We believe [Ni^II(OAc)(pyN₂^Me2)]⁻ originated either from the oxidation of the acac ligand by atmospheric O₂, or through its hydrolysis, through a retro-Claisen mechanism (Scheme S1). Previous reports on Ni^II complexes supported by β-diketonate ligands were spontaneously oxidized, on exposure to O₂, to yield two carboxylates and CO.23 Furthermore, Ni-dependent acireductone dioxygenase,24 promotes the oxidation of its β-diketonate substrate in the same manner. However, we noted that a [D₆]DMSO solution of pure 5 a did not appreciably decay over the course of two days, when exposed to oxygen. In contrast, 12 hours after the addition of 0.1 mL of D₂O to the same solution, [Ni^II(OAc)(pyN₂^Me2)]⁻ was identified by ¹H NMR spectroscopy. For this reason, we propose hydrolysis of 5 a via the retro-Claisen mechanism yields acetone and acetate (Scheme S1).

Bipyridine (6 a, bpy=2,2′-bipyridine) and terpyridine (7 a, terpy=2,2′;6′,2“-terpyridine) complexes were prepared by a simple exchange reaction of bpy or terpy (1 equiv) with 1 a in CH₃CN (for 6 a) or CH₃OH (for 7 a), in 78 and 75 % yields, respectively. Finally, the 2,6-pyridinedicarboxylate (2,6-dipic) complex 8 a was generated by the reaction of 1 a with Li₂(2,6-dipic) (1.2 equiv.) in CH₃OH, giving the complex in 64 % yield.

Electrospray ionization mass spectrometry (ESI-MS) confirmed the elemental composition of 2 a-8 a. Ions corresponding to the Ni-containing fragment of the compounds were identified (in negative mode for the anionic 2 a–5 a and 8 a, and in positive mode for the neutral 6 a–7 a). 4 a–8 a were further characterized via single crystal XRD measurements, NMR, electronic absorption, Fourier transform infrared (FT-IR) spectroscopies, and cyclic voltammetry, as described in detail below.

Solid-state structures, of previously reported compounds 2 a,21 3 a,21 and [Ni^II(OX)(pyN₂^Me2)]⁻ (OX=OH, OCH₃, OCO₂H, O₂CCH₃, ONO₂)19a, 21 were all four coordinate species with a square planar geometry. The XRD crystal structure of 4 a displayed a similar square planar geometry (Figure 1), with the geometry index25 τ₄=0.14, in line with the previously reported square planar [Ni(L)(pyN₂^Me2)] complexes.19a, 21

Figure 1

ORTEP plots (at 50 % probability level) of the X-ray structures of 4 a–8 a. Hydrogen atoms, the counterions Et₄N⁺ for 4 a–5 a and Li⁺ for 8 a, and solvents of crystallization have been omitted for clarity; disorder is not shown.

The metal–ligand bond distances in the [Ni^II(pyN₂^Me2)] core were essentially unvaried in the square planar complexes 2 a–4 a and [Ni^II(OX)(pyN₂^Me2)]⁻,19a, 21 whereas for 5 a–8 a, with higher coordination numbers, there was an average elongation of 0.1–0.2 Å of both the Ni−N_amide and the Ni−N_pyridine bonds (Table S3).

In the square pyramidal structures, the apical ligands showed significantly longer Ni−O/N bonds than the equatorial bond lengths. Even in the pseudo-octahedral compounds 7 a and 8 a, there was a marked lengthening of the axial (apical) bond distances of the ligands above and below the [Ni(pyN₂^Me2)] plane.

Compounds 5 a and 6 a displayed a square pyramidal geometry in the solid state (Figure 1), with the three N donors from the pyN₂^Me2 ligand and one N/O donor from the acac or bpy ligands in the basal plane, and the other binding apically. τ₅ Values of 0.22 and 0.35 respectively,26 demonstrated the environment to be closer to square pyramidal rather than trigonal bipyramidal. In one of the two independent molecules of 6 a in the asymmetric unit cell, the molecule was refined with a CH₃CN ligand bound in the sixth coordination position, with 10 % occupancy. 7 a and 8 a both displayed distorted octahedral geometries (Figure 1), with double axial bending,27 that is, two of the L-Ni-L axes had angles smaller than 180°. This distortion was caused by the tridentate, meridional ligands pyN₂^Me2, terpy, and 2,6-dipic having bite angles smaller than 90° for each cis pair of donor atoms. In 8 a there was an interaction between one of the carboxylate C=O groups and one of the lithium counterions, whose coordination is completed by H₂O and CH₃OH molecules.

The ¹H NMR spectrum of 4 a was comparable to those of previously reported square planar 2 a, 3 a, and [Ni(OX)(pyN₂^Me2)] complexes,19a, 21 and indicative of a diamagnetic, low-spin (S=0) d⁸ Ni^II electronic structure (Figures S2–S5). The C−H resonances attributed to the phenoxide group of 4 a displayed five distinct signals (one of which is masked by the aniline resonances), indicating that they are neither exchanged by symmetry (consistent with the solid-state structure, Figure 1), nor by rotation about the Ni−OC₆H₅ bond on the NMR timescale. In contrast, the five- and six-coordinate complexes 5 a–8 a displayed peak broadening and peak shifts typical of paramagnetic species (Figure 2). With the exception of 7 a (Figure S10), the peaks were sufficiently sharp and well-resolved, that integration and cross-comparison allowed for the partial assignments of the peaks. This is expected among intermediate-spin Ni^II complexes–four-coordinate tetrahedral and five-coordinate complexes have relaxation times that are short enough for reasonable ¹H NMR data acquisition, while this is more difficult for the six-coordinate octahedral geometry.28

Figure 2

¹H NMR spectra of 5 a ([D₆]DMSO), 6 a (CD₃CN), 8 a (CD₃OD) and 9 (CD₃OD), with tentative assignment of the paramagnetically shifted peaks.

Assignment of the signals was facilitated by analysis of the spectrum of [Ni(2,6-dipic)₂]Li₂ (929), a few crystals of which were serendipitously obtained from a failed synthesis of 8 a (the solid-state structure was also investigated; see Supporting Information). 9 had a simple two-resonance ¹H NMR spectrum. On the basis of signal intensity and integration (Figure 2), the peak at 63 ppm was assigned to the meta protons of the pyridine ring, and the peak at 22 ppm was assigned to the para protons in 9. Similar resonances were identified in 8 a (which contains the 2,6-dipic ligand, Figure 2). A signal at ca. 60–80 ppm was present in complexes 5 a, 6 a, and 8 a, which we assigned to the meta-pyridine protons of the pyN₂^Me2 ligand. The para-proton signal of the pyN₂^Me2 ligand was more mobile, at 7 ppm in 4 a, 10 ppm in 6 a, and 20 ppm in 8 a. Integration of the signals at −6 to −10, 6 to 10 ppm, and 10 to 14 assisted in identifying the aniline carboxamide para- and meta-phenyl, and 12 -CH₃ signals, respectively, for all complexes.

The number of signals in the spectra of 5 a and 6 a suggested a high symmetry. For 5 a one additional resonance, integrating to one proton, at −20 ppm can be assigned to the methine proton of acac, and a broad signal at 1.7 ppm, corresponding to six protons, is attributed to the methyl protons of the acac ligand. Four additional signals that integrated to two protons each were present in the spectrum of 6 a, corresponding to the bpy ligand. In the solid state XRD structures of 5 a and 6 a a square pyramidal geometry was observed, which would suggest that the CH₃ resonances of acac, and the pyridine ring ¹H NMR signals of bpy would be non-equivalent as a result of apical and equatorial coordination. Even at lower temperatures, similarly symmetric ¹H NMR spectra were obtained (Figures S6b–S8b). The solution ¹H NMR spectra were indicative of either a change to trigonal bipyramidal coordination, or fast dynamic exchange between the two alternative square pyramidal conformations.

The magnetic moments of 5 a-8 a in solution were measured via the Evans method (Table 1, see supporting info).30 The μ_eff values (5 a: 2.99, 6 a: 2.50, 7 a: 3.07, 8 a: 3.04 μ_B) were within the typical range for five- and six-coordinate S=1 Ni^II complexes,31 with the exception of 6 a, which was slightly lower but nevertheless indicative of a intermediate spin d⁸ ion, possessing two unpaired electrons. The different magnetic properties of the square planar four-coordinate Ni^II complexes and their five- and six-coordinate counterparts are a consequence of their different ligand fields. A second effect was evident in their electronic absorption spectra (Table 1, Figures S27–S33). The square planar 1 a–4 a and [Ni^II(OX)(pyN₂^Me2)]⁻,19a, 21 all displayed orange or red colours arising from moderately intense (ϵ∼100 m⁻¹ s⁻¹) bands at 450–550 nm, as is common for Ni^II complexes of this geometry.32 5 a–8 a, conversely, displayed weak d–d bands (ϵ≤100) at lower energy (800–1000 nm), which is again typical for five- and six-coordinate Ni^II.32a

Table 1. Spectral, magnetic, electrochemical and reactivity properties of 2 a/b-8 a/b.

	λmax [nm] (ϵ/m^−1  cm^−1)	μeff. (μB)	E1/2 vs. Fc^+/Fc^V		λmax [nm]	gx, gy, gz (gav.)	t	k2 [m^−1s−1] ^[a]
3 a	455 (1200)	0	0.48	3 b	530, 830	2.25, 2.02 (2.17)	8 h ^[e] at −80 °C	na
4 a	390 (4400), 460sh (850), 530sh (380)	0	0.26
5 a	380sh (1800), 770 (110), 980 (130)	2.99	−0.14	5 b	530, 720sh	2.32, 2.23, 2.01 (2.19)	8 h ^[e] at −40 °C; 3 h at 25 °C	no reaction
6 a	380sh (2400), 470sh (450), 850 (40)	2.50	0.26	6 b	580, 790sh	2.19, 2.16, 2.02 (2.12)	h at −40 °C	0.087 ^[c]
7 a	440sh (70), 860 (100)	3.07	0.33	7 b	420, 500, 680	2.17, 2.14, 2.02 (2.11)	8 h ^[e] at −40 °C; 30 s at 25 °C	1.29 ^[d]
8 a	400sh (40), 900 (20)	3.04	−0.07	8 b	560, 700sh	2.22, 2.19, 2.01 (2.14)	8 h ^[e] at −40 °C; 260 s at 25 °C	no reaction

• [a] Reaction with 2,6-DTBP at −40 °C. [b] 2 b prepared from 2 a+CAN in acetone. [c] 6 b prepared from 6 a plus 2 equiv. CAN in 9:1 acetone/CH₃CN. [d] 7 b prepared from 7 a plus 2 equiv. CAN in CH₃OH. [e] 8 h was the maximum allowable measurement time, lifetimes likely to be considerably longer.

Cyclic voltammetry (CV) measurements were performed on 1 a–8 a in order to gain insight into their redox behaviors. All CV experiments were done in acetone (with 20 % CH₃CN for 6 a and 50 % CH₃OH for 7 a–8 a, in order to improve solubility), with 0.1 m Bu₄NPF₆ and scan rates of 0.05 V s⁻¹. Almost all compounds displayed one (quasi)-reversible redox wave at relatively low potentials vs. Fc⁺/Fc (Table 1, Figure 3), with peak separations of 0.08–0.10 V, which we attributed to the Ni^II/III couple. The exceptions were 1 a, with no redox events at E<1 V, and 4 a, which only had an irreversible oxidation wave at 0.26 V vs. Fc⁺/Fc. The redox behavior of 4 a was most likely complicated by the redox-active nature of the phenoxide ligand. The variation in E⁰_Ni(II/III) can be explained on the basis of the charge of the complex, and the electron-donation from the ligands, estimated from their basicity (pK_a of their conjugate acids). The lowest values belonged to 5 a (E⁰_Ni(II/III)=−0.14 V), which is an anionic molecule and has the most basic donor ligand (acac, aqueous scale pK_a of Hacac=9), and 8 a (E⁰_Ni(II/III)=−0.07 V), which has three donor atoms in its ancillary ligand, two of which are anionic carboxylates. The highest E⁰_Ni(II/III) values corresponded to chloride complex 2 a (0.43 V) and formate complex 3 a (0.48 V). It is interesting to compare the three neutral compounds 1 a and 6 a–7 a. Clearly, the lack of a negative charge makes oxidation more arduous, as demonstrated by 1 a, which was only oxidised (irreversible wave) at 1.03 V. The bpy and terpy complexes 6 a–7 a, however, had relatively low E⁰_Ni(II/III) of 0.26 and 0.33 V respectively, having multiple pyridine donors which are more electron-rich than CH₃CN; nevertheless, the six-coordinate, dianionic 8 a had a potential that was lower by 0.4 V than the terpy containing 7 a.

Figure 3

Cyclic voltammograms of 2 a–8 a vs. Fc⁺/Fc (orange trace: 2 a, solid blue: 3 a, dashed blue: 4 a, purple: 5 a, pink: 6 a, green: 7 a, red: 8 a); conditions: 0.1 mm, 0.1 nBu₄NPF₆ (nBu=C₄H₉), scan rate 0.05 V s⁻¹, room temperature. Cyclic voltammetry experiments were conducted with a CH Instruments 600E electrochemical analyser, using a glassy carbon working electrode, a platinum wire counter electrode and an Ag/AgNO₃ 0.01 m reference electrode.

2 a–8 a all displayed further oxidation waves beginning at E=0.8 V (Figures S34–41), however these were all irreversible and likely involve ligand oxidation processes. In conclusion, 2 a–8 a constitute a range of compounds, based on the same [Ni^II(pyN₂^Me2)] core, with varied structural and electrochemical properties, suitable as starting materials for the study of Ni^III species; the presence of reversible redox behavior was encouraging in suggesting that a simple one-electron oxidation without successive reaction is plausible.

Preparation of Ni^III complexes

We attempted to generate Ni^III species by the chemical oxidation of the 1 a–8 a precursors (see Supporting Information for details). The Ni^II complexes were oxidized with the one-electron oxidants tris(4-bromophenyl)ammoniumyl hexachloroantimonate (magic blue) and cerium ammonium nitrate, ((NH₄)₂[Ce^IV(NO₃)₆], CAN). The reactions were performed in thermostated cuvettes and monitored by electronic absorption spectroscopy. For the acetonitrile complex 1 a and the phenoxide 4 a, we could not find suitable conditions to generate a stable oxidized species. The preparation, characterization, and reactivity properties of 2 b, are described in a separate report.20 3 a and 5 a (0.4 mm acetone solutions) reacted with one equivalent of magic blue at −80 °C. New species with intense visible absorption bands (3 b: 530 and 830 nm, 5 b: 530 and 720 nm, Figure 4) that were comparable to previously reported [Ni^III(OX)(pyN₂^Me2)] complexes,19 were obtained. Because of solubility limitations associated with 6 a–8 a, the latter three Ni^III compounds could not be prepared in pure acetone solutions: 6 a was dissolved in a 9:1 acetone/acetonitrile mixture; 7 a in pure CH₃OH; and 8 a in 1:1 acetone/CH₃OH. For 6 a-8 a, the addition of 2 equivalents of CAN (as an acetonitrile solution) yielded the maximum yields of the corresponding Ni^III species 6 b–8 b.

Figure 4

Electronic absorption spectra of 2 b–3 b and 5 b–8 b (orange trace: 2 b; blue: 3 b; purple: 5 b; pink: 6 b; green: 7 b; red: 8 b). Conditions: see experimental section in the Supporting Information.

The products showed bands at λ_max=580 and 790 nm (6 b), 680 nm (7 b), and 560 and 700 nm (8 b). High yields for 6 b could only be obtained at −80 °C, resulting in long preparation times (50 minutes), while 7 b–8 b were stable and prepared in high yields at higher temperature (−40 °C). We have previously observed that, in oxidation reactions mediated by CAN, the ^-ONO₂ anion can compete with the ancillary ligand for binding to the metal centre.19a Indeed, reacting more than 2 equivalents of CAN with 6 a caused the formation of [Ni^III(ONO₂)(pyN₂^Me2)], as evident from its absorption bands at λ_max=560 and 890 nm.19 We did not encounter this problem with 7 a–8 a. Complexes 5 b–8 b represent the first examples of non-square planar high-valent oxidants supported by the 2,6-pyridinecarboxamidate ligand. Furthermore, complexes 6 b and 7 b are the only cationic Ni^III species supported by the 2,6-pyridinedicarboxamidate ligand (in contrast to the previously described neutral (2 b, 3 b, [Ni^III(OX)(pyN₂^Me2)]) and anionic (8 b) complexes).

ESI MS analysis of the freshly prepared, cold solutions revealed ions corresponding to [3 b+CH₃CN+H]⁺ (m/z=516, Figure S52), [5 b+Na]⁺ (m/z=552, Figure S53) and [8 b−Li⁺]⁻ (m/z=594, Figure S54), confirming the identity of the oxidized species. Ions corresponding to the other complexes were not identified, presumably due to the thermal instability of these compounds. The solutions of 2 b–8 b generated through the aforementioned methods were analyzed by EPR spectroscopy (Figure 5). All showed the presence of S= species, with average g-values of 2.11–2.22, which are consistent with the unpaired spin density localized mainly on a low-spin, d⁷, Ni^III ion.33 Spin quantification against a standard ((2,2,6,6-tetramethylpiperidin-1-yl)oxyl (TEMPO)) provided the following yields of Ni^III species: 3 b=90±20 %; 5 b=80±20 %=80; 6 b=70±20 %, 7 b=80±20 %, 8 b=80±20 %.

Figure 5

X-Band EPR spectra of 2 b–3 b and 5 b–8 b (orange trace: 2 b; blue: 3 b; purple: 5 b; pink: 6 b; green: 7 b; red: 8 b), measured as frozen solutions at 77 K, 2 mW microwave power.

The formate complex, 3 b, displayed an axial signal (g_⊥=2.25, g_∥=2.02), and was very similar to previously reported square planar bicarbonate, acetate, and nitrate [Ni^III(OX)(pyN₂^Me2)] complexes. However, the broadness of the features in 3 b showed it was either not entirely pure or there was some fluxional behaviour of the potentially bidentate ligand. The spectrum of the ^-Cl complex (2 b) was considerably more rhombic, with g=2.32, 2.23, 2.00, and hyperfine coupling to the ^35/37Cl nucleus (I=3/2), evident from the four-line splitting of the z component of the signal (A_zz=110 MHz).20

The spectrum of the acac complex, 5 b, displayed a mixture of a major, markedly rhombic (g=2.32, 2.23, 2.01) species, plus a relatively small contamination of a second species (g_x=g_y=2.21, g_z=2.02; ∼15 % of total Ni^III, evaluated on the basis of the height of the signals at g∼2 and of simulation of the spectrum). The contamination appears similar to 3 b and previously prepared [Ni^III(OX)(pyN₂^Me2)] complexes. We identified that 5 a was readily converted to [Ni^II(O₂CCH₃)(pyN₂^Me2)]⁻ and thus surmise the contaminant is [Ni^III(O₂CCH₃)(pyN₂^Me2)]. The remaining three compounds, 6 b–8 b, displayed EPR spectra with g_x>g_y≫g_z. Interestingly, 6 b showed a three-line signal shape corresponding to hyperfine coupling to one I=1 ¹⁴N nucleus in the z direction (A_zz=65 MHz), whereas 7 b had coupling to two ¹⁴N nuclei (five-line splitting, A_zz=55 MHz), whereas 8 b had no visible hyperfine coupling.

In principle, a complex with unpaired spin density will give rise to an axial EPR signal if it has a four-fold symmetry axis, at least locally around the unpaired spin density. Hence, in the putative square planar complexes (2 b–3 b and [Ni(OX)(pyN₂^Me2)]19) by virtue of having a pseudo-D_4h symmetry, we expect an axial signal, as we indeed observed for 3 b and [Ni(OX)(pyN₂^Me2)].19 For 2 b, because chlorine is considerably larger and has different donor properties than the O-atoms of the other complexes, we suggest the markedly different first coordination sphere induces a more rhombic EPR signal. Similar arguments are valid for the square pyramidal structures, which 5 b and 6 b potentially are, that is, there would be a plane with four similar N/O ligands, plus one apical ligand, to give a pseudo-C_4v symmetry, and hence an axial EPR signal would be expected. The alternative for 5 b and 6 b, is a trigonal bipyramidal geometry, which we postulate is more likely to give a more rhombic EPR spectrum. For 6 b by virtue of the fact that g_x≈g_y, suggesting a near-axial symmetry of the compound, a square pyramidal geometry is more likely than trigonal bipyramidal. In contrast, 5 b's greater separation between g_x and g_y signal was suggestive of either a trigonal bipyramidal geometry, or a bigger distortion in the coordination plane. Finally, the pseudo-octahedral complexes 7 b–8 b were comparable to the square planar complexes, with approximate fourfold rotational symmetry in at least one direction, and thus more axial EPR signals.

We observed resolved hyperfine coupling in the three complexes 2 b and 6 b–7 b, in all cases only in the g_z component. Coupling in the other directions cannot be excluded, as the g_x/y parts of the signal display considerably larger line broadening that could mask a fine structure. The bpy and terpy compounds 6 b–7 b coupled to one and two (equivalent) ¹⁴N nuclei, respectively. In the 2,6-dipic complex 8 b, which still contains a pyridine donor, no coupling was visible. The three complexes have, respectively 5, 6, and 4 N donor ligands. We can infer that, as a result of the particular electronic structure of the complexes, only donor atoms above and below the molecular plane formed by the [Ni(pyN₂^Me2)] core have large hyperfine coupling. On the basis of this hypothesis, we can further deduce that the two N donor atoms of bpy in 6 b are non-equivalent, and support the conclusion that a symmetrical trigonal bipyramidal structure is not present in solution.

The thermal stability of 3 b–8 b was quite varied (Table 1), however, all Ni^III species were reasonably stable at −40 °C to −80 °C. Generally, the complexes supported by more electron-rich ancillary ligands displayed longer lifetimes. 2 b and [Ni^III(OX)(pyN₂^Me2)] complexes were stable at −40 °C,19, 20 whereas 3 b, conversely, could only be prepared and remained stable at −80 °C. 5 b, with the particularly basic bidentate acac ligand, was the most stable of the prepared complexes, being stable at −40 °C, and even displaying a t =3 h at 25 °C. 6 b, was prepared in highest yield at −80 °C, and had a t =30 minutes after warming to −40 °C. 7 b and 8 b were indefinitely stable at −40 °C, and were even moderately stable (t =30 s and 260 s, respectively) after warming to 25 °C. A general trend in stability was that the complexes supported by multiple donors (e.g. 8 b), or highly basic anionic donors (e.g. 5 b) were the most stable, while those with low coordination numbers and/or single anionic donors were less stable. This would suggest that by simply tuning the donor strength of the ancillary ligand one could prepare highly stable (less reactive) high-valent oxidants.

In order to understand the oxidation reactivity properties of 5 b–8 b, and the role the supporting ligands play in modulating the reactivity of these high-valent oxidants, we reacted 5 b–8 b with phenols. 3 b was insufficiently stable to explore its reactivity. We explored the reactivity of 5 b–8 b towards 2,6-di-tert-butylphenol (2,6-DTBP). At −40 °C, neither 5 b nor 8 b reacted with a large excess of 2,6-DTBP (100 equiv.), while both 6 b and 7 b reacted readily, as evidenced by the disappearance of their characteristic visible absorption bands (Figures S55–S56). Mass spectrometric analysis of the post-reaction mixtures showed the formation of 3,3′,5,5′-tetra-tert-butyl-[1,1′-bi(cyclohexylidene)]-2,2′,5,5′-tetraene-4,4′-dione (Figures S59–S60), a homo-coupling product likely derived from the phenoxyl radical (from HAT). These observations correlate well with the thermal stability observations that the complexes supported by highly basic ancillary donors are unreactive oxidants, whereas those with neutral donors are reactive towards phenols.

The reactions of 6 b and 7 b with 2,6-DTPB were pseudo-first order in the presence of excess 2,6-DTPB, and we measured their rate constants (k_obs) by exponential fitting of the decay plot. Second order rate constants (k₂) were determined by plotting k_obs (determined at varying substrate concentrations) against substrate concentration and calculating the slope of the resulting linear plots (Figures S57–S58). 6 b reacted with a k₂=0.087 m⁻¹ s⁻¹ in a solvent mixture of 9:1 acetone/acetonitrile. 7 b (E⁰_Ni(II/III)=0.33 V) reacted an order of magnitude faster, with k₂=1.29 m⁻¹ s⁻¹, however in pure CH₃OH. Importantly, none of 5 b–8 b reacted with hydrocarbons containing weak C−H bonds (xanthene and 9,10-dihydroanthracene). These rate constants compare favourably with those determined for [Ni^III(OX)(pyN₂^Me2)] suggesting these cationic complexes (6 b and 7 b) with neutral ancillary ligands are active HAT oxidants. Unfortunately we cannot draw much insight into structural effects on reactivity from these studies, because only 6 b and 7 b displayed reactivity towards 2,6-DTBP. Given that these 5- and 6-coordinate complexes reacted at rates similar to 4-coordinate [Ni^III(OX)(pyN₂^Me2)] complexes, we surmise that the geometry and coordination number of these oxidants has minimal effect on the reactivity properties, and that the charge on the metal ion appears to govern its relative reactivity in phenol oxidation.

We previously found that 2 b and [Ni^III(OX)(pyN₂^Me2)] (OX=OCO₂H, O₂CCH₃, ONO₂ reacted with phenols through a HAT mechanism.19, 20 Unfortunately, a consistent comparison with the extended series presented here is hampered by the different conditions necessary to prepare [Ni^III(L)(pyN₂^Me2)] (L=Cl, OCO₂H, O₂CCH₃, ONO₂) and 5 b–8 b. We previously found that the most electron-poor [Ni^III(ONO₂)(pyN₂^Me2)] complex reacted at the highest rates.19 The observations that only 6 b and 7 b were capable of oxidizing 2,6-DTBP is thus not surprising, because both are cationic complexes, with Ni^II/III redox potentials in the same range as [Ni^III(OX)(pyN₂^Me2)].19a 7 b constituted somewhat of an exception, being considerably more reactive than 6 b in spite of there being little-to-no difference in their redox potentials. The presence of a protic solvent medium (CH₃OH) for 7 b likely influenced the reaction rate.

The oxidation of 2,6-DTBP by high-valent metal-based oxidants comes under the class of proton coupled electron transfer (PCET) reactions.34 We previously demonstrated that 2 b and [Ni^III(OX)(pyN₂^Me2)] complexes reacted with 2,6-DTBP and hydrocarbons through a HAT (concerted PCET) mechanism. We proposed the OX ligand acted as a proton acceptor in these reactions, yielding [Ni^II(L)(HOX)] products. The discovery that the bpy and terpy complexes 6 b and 7 b can perform phenol oxidation suggests that the 2,6-pyridinedicarboxamidate ligand may also act as a proton acceptor. All of the [Ni^III(OX)(pyN₂^Me2)] complexes contain an anionic ancillary donor ligand, unlike 6 b–7 b. If the reactions between 2,6-DTBP and 6 b–7 b (containing neutral ancillary ligands) proceeded through a HAT mechanism, it may indicate that the 2,6-pyridinecarboxamidate (pyN₂^Me2) ligand played a non-innocent role in the HAT reaction. This is because the bpy and terpy ligands 6 b–7 b are neutral donors, and thus in principle may not act as proton acceptors, whereas the pyN₂^Me2 could potentially act as a proton acceptor on either the carboxamidate O- or N-atoms. The argument can also be made that a pyridine donor of the bpy and terpy ligands may act as a proton acceptor site in 6 b and 7 b.

Unfortunately, we cannot assume that the same PCET (thus HAT) mechanism is at play in the case of 6 b–7 b, particularly by virtue of the fact that they were analysed in different solvent media (7 b could only be analysed in CH₃OH). Indeed the use of a protic solvent can potentially have large effects on the mechanisms of PCET reactions. Work is underway in our laboratory to probe the mechanism of phenol oxidation by 6 b–7 b. That is, do they oxidise via HAT or by another non-concerted form of PCET? We also intend to identify conditions where a fair comparison of 6 b–7 b and [Ni^III(OX)(pyN₂^Me2)] complexes can be made. This will be a critical piece of future work, because the observation that these complexes will oxidise phenols with weak O−H bonds is critically important in the future development and studies of oxidants supported by carboxamidate ligands. Our results demonstrate that such functional groups maybe non-innocent proton acceptors in PCET reactivity.

Conclusion

We synthesised and characterised a series of Ni^II complexes based on the [Ni^II(L)(pyN₂^Me2)] core, with different coordination numbers (4–6) and geometries, and ligands with a variety of donor groups. The binding mode of the supporting ligand pyN₂^Me2 was essentially unvaried, as a meridional N3 ligand, with the exception of minor bond lengthening caused by increased coordination numbers. The other (ancillary) ligand determined the coordination geometry, with the mono-dentate -O₂CH and -OC₆H₅ donors forming square planar complexes 3 a–4 a, the bidentate acac and bpy yielding square pyramidal complexes 5 a–6 a, and the tridentate terpy and 2,6-dipic forming pseudo-octahedral complexes 7 a–8 a. While the four-coordinate square planar complexes were low spin (S=0), higher coordination numbers caused a switch to the intermediate spin state (S=1).

Most of the complexes could be oxidized at low temperature to generate putative Ni^III species 3 b–8 b. Electronic absorption and EPR spectra typical of metal-centred spin density confirmed the formation of Ni^III species. The oxidation reactivity of the Ni^III species was examined by their reactivity towards 2,6-DTBP. Diverse reactivity properties were evident, with higher basicity of the ligands contributing to higher stability and lower reactivity levels. The major influence on reactivity derived from the relative charge at the Ni ion, and not from the geometry or coordination number. Fascinatingly, complexes with neutral ancillary donors were still capable of oxidizing 2,6-DTBP, suggesting the supporting 2,6-pyridinedicarboxamidate ligand may not be innocent in PCET phenol oxidation reactions.

Experimental Section

See Supporting Information for experimental details.

Acknowledgements

We are grateful to COST Action CM1305 (ECOSTBio) for networking support. This publication has emanated from research supported by the European Union (FP7-333948, ERC-2015-STG-678202). Research in the McDonald lab is supported in part by research grants from Science Foundation Ireland (SFI/12/RC/2278, SFI/15/RS-URF/3307). We are grateful to Prof. Robert Barkley for training on and use of an EPR spectrometer.

Conflict of interest

The authors declare no conflict of interest.