Ambidentate bonding and electrochemical implications of pincer-type pyridylidene amide ligands in complexes of nickel, cobalt and zinc †

Pincer-type tridentate pyridyl bis(pyridylidene amide) (pyPYA 2 ) ligand systems were coordinated to the Earth-abundant ﬁ rst row transition metals nickel, cobalt and zinc. A one-pot synthesis in water/ethanol a ﬀ orded octahedral homoleptic bis-PYA complexes, [M(pyPYA 2 ) 2 ](PF 6 ) 2 , whereas ﬁ ve-coordinate mono-PYA dichloride complexes, [M(pyPYA 2 )Cl 2 ], were obtained upon slow addition of the ligand to the metal chlorides in DMF. Electrochemical measurements further revealed a facile oxidation of the metal centers from Ni 2+ to Ni 4+ and Co 2+ to Co 3+ , respectively, while the Zn 2+ system was redox inactive. These experiments further allowed for quanti ﬁ cation of the much stronger electron donor properties of neutral N , N , N -tridentate pyPYA 2 pincer ligands as compared to terpy. Remarkably, ortho -PYA pincer ligands feature amide coordination to the metal center via oxygen or nitrogen. This ambidentate ligand binding constitutes another mode of donor ﬂ exibility of the PYA ligand system, complementing the resonance structure dynamics established previously. NMR spectroscopic and MS analysis reveal that the meta -PYA ligand undergoes selective deuteration when coordinated to cobalt. This reactivity suggests the potential of this ligand as a transient proton reservoir for HX bond activation and, moreover, indicates the relevance of several resonance structures and therefore supports the notion that meta -PYA ligands are mesoionic.


Introduction
The large-scale production of bulk chemicals requires the utilization of inexpensive materials, and many industrially relevant processes rely on heterogeneous catalysts based on Earthabundant metals. Prominent examples are the iron oxide catalysts used in the Haber-Bosch process, 1-3 nickel catalysts in steam reforming for the production of syngas [4][5][6] or RANEY® nickel catalysts utilized in hydrogenation reactions. 7,8 In contrast, the development of homogeneous catalysts has long focused on complexes based on platinum group and coinage metals. 9 These noble metals feature excellent reactivity and stability properties, though their high price 10 and scarcity 11,12 as well as the high toxicity 13 demand for more sustainable solutions and have led to extensive research activities aiming for the replacement of noble metals by first row transition metals. 14 As a consequence, iron catalysis has made major progress over the last years, 15 including for example the develop-ment of highly efficient hydrogenation catalysts that are competitive with the best ruthenium catalysts. 16 Likewise manganese(I) pincer complexes have been explored as attractive alternative to established ruthenium-and iridium-based catalysts in (de)hydrogenation as well as in C-C and C-X bondforming reactions. 17 Also nickel complexes have gained much attention recently, e.g. as low-cost replacement for palladium in cross-coupling reactions. 18 We have recently developed a new pincer platform based on pyridylidene amide (PYA) ligands (A, Fig. 1), which imparted high catalytic activity to ruthenium centers in transfer hydrogenation catalysis. 19,20 Based on the relatively hard yet formally neutral amide donor site of these PYA ligands, we sought to expand the application of this ligand system towards first row transition metals in view of providing a more sustainable alternative to noble metals. PYA ligands have been shown to induce attractive catalytic properties to iridium, ruthenium [21][22][23][24][25][26] and palladium 27 complexes, in parts by stabilizing high-valent oxidation states and in other parts due to their unique electron donor flexibility. [28][29][30] Despite these attractive features, no first row transition metal complexes with PYA ligands have been reported, but analogous pyridylidene amine (PYE) ligands have been coordinated to nickel centers by the groups of Johnson 31 and Douthwaite 32 (B and C, Fig. 1). Here we report nickel, cobalt, and zinc complexes con-systems. [34][35][36] Such a linkage isomerism of PYA ligands has been proposed, 37 yet never established so far 38 and presents a further type of flexibility of the PYA system. Notably, the same ortho-pyPYA 2 ligand displayed the classic κ 3 -N,N,N-coordination mode when bound to ruthenium(II) in [Ru(ortho-pyPYA 2 )(MeCN) 3 ](PF 6 ) 2 (Fig. S2 †). Likewise, pyridine dicarboxamide systems with ortho-dimethylated aryl substituents show exclusively κ 3 -N,N,N-bonding. 35,39 The κ 3 -O,N,O coordination mode is unprecedented and presumably a consequence of the Lewis acidity of the first row transition metal centers and their ensuing oxophilicity, combined with the unique donor flexi-bility of the formally neutral and sterically constrained ortho-PYA donor site, and distinguishes these ligands from the more commonly used pyridine carboxamide systems.

Synthesis and characterization of mono(PYA-pincer) complexes
The synthesis of cobalt and nickel mono-PYA complexes, [M( pyPYA 2 )Cl 2 ] complexes 4a-c (M = Co) and 5a-c (M = Ni; Scheme 2), was accomplished by slow addition of a filtered DMF solution containing the ligand and Na 2 CO 3 to a DMF solution containing the metal chloride. The slow addition of ligand as well as the choice of solvent played a crucial role. Initial attempts to control the reaction of the PYA ligands 1a-c with MCl 2 by quenching the reaction at early reaction times or by using a large excess of the metal salt (up to 10 eq.) invariably resulted in the formation of the homoleptic [M( pyPYA 2 ) 2 ] 2+ compounds 2 and 3.
Formation of complexes 4 and 5 was established by CHN microanalysis as well as HR ESI-MS, which showed the monocationic complex fragments [M − Cl] + , e.g. at 441.0409 amu for 4a (calcd 441.0403). The set of cobalt complexes 4a-c was furthermore characterized by single crystal X-ray diffraction analysis (Fig. 2). The complexes feature a strongly distorted trigonal bipyramidal arrangement (τ 5 = 0.45-0.49), 40 which is in good agreement with the reported analogous structure of [Co(terpy)Cl 2 ] (terpy = terpyridine). 41 The para-and meta-PYA pincer ligands in complexes 4a,b bind to the metal center in a κ 3 -N,N,N coordination mode whereas the ortho-PYA pincer ligand in 4c is again κ 3 -O,N,O coordinated as observed for the bis-PYA pincer complexes. Notably, steric constraints should be less relevant when the PYA pincer ligand is bound to the CoCl 2 fragment than in bis-PYA pincer complexes, indicating an electronic stabilization of the PYA coordination through oxygen.
The para-and meta-PYA pincer ligands in complexes 4a,b are almost planar, i.e. the dihedral angle between the PYA heterocycles and central pyridine ring are only 8.5°, 9.8°in 4a and Scheme 2 Synthesis of Ni(II) and Co(II) [M( pyPYA 2 )Cl 2 ] complexes 4a-c and 5a-c by slow addition of deprotonated ligand at room temperature. 10.4°in 4b respectively (Table 1), which is distinctly different when compared to the correspondent homoleptic complex variation [Co(meta-pyPYA 2 ) 2 ](PF 6 ) 2 which show a 55°twist between the PYA and central pyridyl planes. This high planarity may hint to a significant π-contribution of the N amide -C PYA bonds (N1-C3/4) in both complexes. In such a model, the amide nitrogen is acting predominantly as neutral L-type N-donor ligand, presumably as a consequence of the higher electron density at the formally neutral metal center in complexes 4 and 5 compared to the dicationic center in the bis ( pincer) complexes 2 and 3. In agreement with a more pronounced L-type and π-acidic binding mode of the amide nitrogen, the para-PYA heterocycles of complex 4a reveal a pronounced diene-imine resonance form with localized double and single bonds (average C2-C3 and C5-C6 1.365(4) Å, average C3-C4 and C4-C5 bonds (1.415(7) Å; Table 1). This bond alteration implies considerable contribution of the imine resonance structure of the PYA unit (cf. bottom structure of A in Fig. 1). The

Dynamic linkage isomerism of the ortho-PYA-pincer ligand system
For better spectroscopic analysis in solution, the analogous zinc(II) complexes [Zn( pyPYA 2 ) 2 ](PF 6 ) 2 6a-c were prepared by reacting ligands 1a-c with 0.5 eq. ZnCl 2 under mildly basic conditions (Scheme 3). In contrast to the paramagnetic and NMR-silent octahedral Co(II) and Ni(II) complexes 2 and 3, complexes 6a-c are diamagnetic and allow for NMR spectroscopic characterization. Ligand coordination to the Zn II center resulted in the loss of the NH amide proton resonances and in a marked upfield shift of the aromatic H PYA proton resonances, e.g. from δ H = 9.76 (H 6 PYA ) in meta-PYA ligand 1b to δ = 8.43 in complex 6b.
All zinc complexes 6a-c feature one distinct set of PYA ligand proton resonances, including N-CH 3 (δ H around 4), pyridyl (δ H 8. 25-8.70) and pyridylidene (δ H 7.08-8.43) signals. This symmetric arrangement indicates that both pyPYA 2 ligands in each zinc complex are coordinated in the same mode, most presumably the typical κ 3 -N,N,N-coordination based on the structures of the cobalt and nickel analogues and previously described structures. 19,[42][43][44][45] The symmetric ligand bonding in 6c containing ortho-pyPYA 2 ligands contrasts the solid state bonding established for the analogous nickel and cobalt complexes 2c and 3c comprised of a mixture of κ 3 -N,N,N and κ 3 -O,N,O pincer coordination. NMR spectroscopic analysis of complex 6c at low temperature indicated decoalescence of the signals into two sets in a 9 : 1 integral ratio, indicating the presence of two species (Fig. 3). Based on the six inequivalent aromatic resonances, the major species was assigned to the homoleptic bis-κ 3 -N,N,Ncoordinated pincer isomer. The minor isomer features at least ten inequivalent proton signals, which were therefore attributed to mixed κ 3 -N,N,N/κ 3 -O,N,O-coordination isomer (cf. structures of 2c, 3c). These measurements therefore suggest that at room temperature, complex 6c undergoes rapid κN-/κO-amide isomerization.
The dynamic process of this complex was investigated by variable temperature 1 H NMR spectroscopic experiments in CD 2 Cl 2 in the +20 to −80°C temperature range (Fig. S3 †). Decoalescence of most of the pyridyl and PYA proton resonances was observed around −40°C, apart from the PYA H 6   (Table S2 †). 46,47 A similar behavior was noted in CD 3 CN (Fig. S4 †), though due to the higher melting point of this solvent, the decoalescence was not resolved. Nonetheless, these experiments suggest that the dynamic bonding is not significantly depending on the solvent. The bonding mode in the bis-( pyPYA 2 ) complexes was further analyzed by solid state infrared (IR) spectroscopy of the pure compounds. The ligand precursor 1a features three characteristic absorption bands for the amide carbonyl and PYA-imine units in the 1580-1700 cm −1 region (Table 2, Fig. S5-S11 †). 48 Upon metalation, the two lower bands around 1590 and 1645 cm −1 remain essentially unchanged, while the high energy band at 1706 cm −1 shifts characteristically to 1617 (±1) cm −1 , presumably due to metal coordination to the amide unit (Fig. 4a). Likewise, the strong absorption above 1700 cm −1 for ligand precursors 1b and 1c shifts to lower energy upon metalation, though assignments are more complicated as all bands shift considerably ( Fig. 4b and Fig. S5 †). It is worth noting that the shift is independent of the metal center and that the IR spectra are essentially identical for Co(II), Ni(II), and Zn(II) complexes for the para-and meta-pyPYA 2 ligand even in the low frequency area. This high similarity suggests an identical coordination mode for all three metal complexes. For the ortho-pyPYA 2 series, however, subtle differences are noted in the 1550-1650 cm −1 range which distinguish the Zn from the Ni and Co systems (Fig. 4c). These differences may point to a different ligand behavior, which may be associated with the κ 3 -N,N,N vs.   for the [Zn(ortho-pyPYA 2 ) 2 ](PF 6 ) 2 complex 6c by NMR spectroscopy. IR spectroscopy therefore does not conclusively indicate whether this linkage isomerism is exclusively occurring at Zn 2+ , or whether it is more general for ortho-pyPYA 2 ligands and also takes place at Co 2+ and Ni 2+ . The exclusive κ 3 -O,N,O bonding in the solid state of the latter two complexes (cf. structures of 2c-5c) might be induced by lower solubility or a preferential crystal packing of the mixed coordination isomer, leading to a dynamic kinetic resolution of the two isomers through crystallization.

Redox properties of homoleptic bis-PYA pincer complexes
The orange meta-pyPYA 2 cobalt(II) complex 2b oxidized to the green Co(III) analogue 2b + under aerobic conditions in a saturated aqueous solution (Scheme 4). The oxidation state of the cobalt center in 2b + was unambiguously identified by crystallographic analysis due to the 0.2 Å shorter Co-N bond distances compared to 2b as well as the presence of three PF 6 − counter ions. This oxidation under mild conditions indicates strong donor properties of the PYA pincer ligand in these cobalt and nickel complexes. Therefore, the electrochemical properties of the complexes were investigated by cyclic voltammetry.
Cobalt complexes 2a-c show reversible redox processes in the 0 to +0.5 V potential range that were assigned to metal-centered Co 2+ to Co 3+ oxidations (all potentials vs. Ag/AgCl; Fig. 5). The octahedral complex [Co(terpy) 2 ](PF 6 ) 2 49 was also included to compare the donor strength of the formally neutral triden-tate PYA pincer ligands with a well-studied reference compound. The reversibility of the M II/III oxidation processes in complexes 2 and 3 is supported by the linear relationship of anodic (I pa ) and cathodic peak currents (I pc ) vs. the square root of the scan rate ( Fig. S12 †). This linearity together with the small peak potential difference ΔE p < 105 mV ( Table 2) indicates a diffusion limited electron transfer process. [50][51][52] The oxidation potential is dependent on the PYA substitution pattern and is lowest for cobalt coordinated to the meta-PYA pincer ligand (E 1/2 = 0.01 V in 2b) and increases with para-PYA (E 1/2 = 0.20 V in 2a) 53 and even more with ortho-PYA pincer ligands while the para and even more so the meta analogue are stronger donors. The metal-centered nature of the oxidation process was confirmed by redox analyses of the Zn analogues 6a-c. These complexes do not show any redox activity in the 0-2.2 V potential window (Fig. S14 †), indicating that the coordinated PYA pincer ligands are redox inactive within this range. Likewise, the three pyridinium salt ligand precursors 1a-c do not show any oxidation process up to 2 V and feature an irreversible reduction processes around −1 V (Fig. S13 †). This reduction is dependent on the PYA substitution pattern and easiest for ortho-PYA system 1c (E pc = −0.98 V), intermediate for meta-PYA 1b (E pc = −1.22 V) and at lowest potential for the para analogue 1a (E pc = −1.33 V).
Electrochemical characterization of the nickel complexes supports the trend in donor strength of the PYA pincer ligands as observed for the Co series. Thus, the Ni 3+ oxidation state was stabilized better by the meta-PYA pincer ligand (E 1/2 = 1.03 V in 3b) than the para analogue (E 1/2 = 1.20 V in 3a) or the ortho-pyPYA 2 system (E 1/2 = 1.34 V in 3c; Fig. 6a-c). In addition, the stronger donor sets also provided access to Ni IV complexes via a second reversible oxidation at higher potential (E 1/2 = 1.72 V and 1.54 V for complexes 3a and 3b, respectively). Notably, [Ni(terpy) 2 ](PF 6 ) 2 as reference compound is oxidized at considerably higher potential (E 1/2 = 1.65 V), 55 suggesting weaker donation of the terpy ligand than even the least basic ortho-pyPYA 2 ligand.

H/D exchange in meta-PYA sites
A closer investigation of the homoleptic cobalt complex 2b with meta-PYA pincer ligands revealed intriguing ligand-based reactivity. Crystallization of this complex in D 2 O under aerobic conditions resulted in a mixture of orange and green crystals. The latter are a product of spontaneous oxidation of a the metal center and furthermore revealed selective deuterium incorporation into the PYA heterocycle, as demonstrated by MS and NMR analyses. Material crystallized from D 2 O showed the m/z signals for the mono-, di-, and tri-cationic species at 8, 4 and 2.66 amu higher, respectively, than samples of 2b + crystallized from H 2 O (Table S3 †). Formation of 2b + -d 8 was further supported by the absence of the PYA resonances at δ H 8.11 and 8.25 in the 1 H NMR spectrum of the complex generated in D 2 O (Fig. S15 †). Based on NOE spectroscopy as well as the pertinent coupling constants, these resonances were assigned to H 2 PYA and H 6 PYA , respectively, i.e. the positions ortho to the PYA nitrogen. These data indicate a highly selective H/D exchange. Interestingly, the isomeric complexes 2a and 2c containing para-and ortho-PYA pincer ligands, respectively do not undergo any detectable deuteration or oxidation (MS, NMR) when exposed to D 2 O, presumably because of the higher oxidation potential of these complexes (vide supra). The selective deuterium incorporation in the PYA 2-and 6-position of 2b may be rationalized by considering resonance forms B and C of the meta-PYA ligand system, which imply a formally neutral imine donor site and opposite charges within the pyridylidene heterocycle (Scheme 5). Reversible deuteration/protonation at the partially negatively charged positions would account for the observed selectivity and indicates that resonance forms B and C are contributing in addition to resonane form A. This contribution of several resonance structures therefore validates meta-PYA systems as mesoionic compounds. 56

Conclusions
Here, we have shown that PYA pincer ligands form stable complexes with nickel, cobalt and zinc, which expands the application potential of this ligand beyond platinum group metals to first row metals. The substitution pattern of the PYA heterocycle influences the binding mode of the pincer ligand to these Earthabundant transition metals. meta-and para-pyPYA 2 ligand systems are exclusively This deuteration suggests partial delocalization of the negative charge and therefore mesoionic properties of the meta-PYA unit. Moreover, this reactivity may become potentially useful for metal-ligand cooperative HX bond activation catalysis.

General
All reagents were commercially available and used as received. Ligands 1a and 1b have been prepared according to previously

Dalton Transactions Paper
This journal is © The Royal Society of Chemistry 2020  57 ATR-FTIR spectra were recorded with a Jasco FT/ IR-4700 spectrometer with a Jasco ATR Pro One unit containing a diamond crystal.

Crystal-structure determination
Suitable single crystals of 2b-c, 3b-c, and 4a-c were mounted in air at ambient conditions and measured on an Oxford Diffraction SuperNova area-detector diffractometer 58 using mirror optics monochromated Mo Kα radiation (λ = 0.71073 Å) and Al filtered. 59 The unit cell constants and an orientation matrix for data collection were obtained from a least-squares refinement of the setting angles of reflections. Data reduction was performed using the CrysAlisPro 58 program. The intensities were corrected for Lorentz and polarization effects, and a numerical absorption correction based on Gaussian inte-gration over a multifaceted crystal model was applied. The structure was solved by direct methods using SHELXT, 60 which revealed the positions of all non-hydrogen atoms of the title compounds. The non-hydrogen atoms were refined anisotropically. All H-atoms were placed in geometrically calculated positions and refined using a riding model where each H-atom was assigned a fixed isotropic displacement parameter with a value equal to 1.2U eq of its parent atom (1.5U eq for methyl groups and water). Refinement of the structure was carried out on F 2 using full-matrix least-squares procedures, which minimized the function ∑w(F o 2 − F c 2 ) 2 . The weighting scheme was based on counting statistics and included a factor to downweight the intense reflections. All calculations were performed using the SHELXL-2014/7 61 program in OLEX2. 62 The PF 6 − anions in the crystals of complex 2b, 2c, and 3b are disordered as indicated by the high residual electron density peak near its P-atom. No conclusive disorder model was found for 2b and 3b, while for 2c, disorders about two sites were restrained by SHELX SIMU and RIGU instructions. Further details are given in Tables S4-S6

Conflicts of interest
The authors declare no conflict of interest.