Synthesis and ligand-based reduction chemistry of boron difluoride complexes with redox-active formazanate ligands†

Mono(formazanate) boron difluoride complexes (LBF2), which show remarkably facile and reversible ligand-based redox-chemistry, were synthesized by transmetallation of bis(formazanate) zinc complexes with boron trifluoride. The one-electron reduction product [LBF2] [Cp2Co] + and a key intermediate for the transmetallation reaction, the six-coordinate zinc complex (L(BF3))2Zn were isolated and fully characterized.

The chemistry of coordination complexes bearing redox-active (or non-innocent) ligands has recently received increasing attention due to its potential application in small molecule activation and catalysis, 1 and its relevance to biological (enzymatic) transformations. 2 The most studied ligands in this class are dithiolenes/ dioxolenes, 3 a-diimines 4 and bis(imino)pyridines. 5 Formazanates (1,2,4,5-tetraazapentadienyls), which are nitrogen-rich analogues of the well-know b-diketiminates, 6 have received comparatively little attention as ligands in coordination chemistry. 7 Unlike b-diketiminates, which have a NCCCN backbone, formazanates feature a NNCNN backbone; the two additional nitrogen atoms provide formazanates with redox-active properties. Specifically, the reduced form of formazanates should be relatively accessible and stable due to delocalization of the SOMO over all 4 N atoms. This is in a way related to the stability of the analogous organic verdazyl radicals, which may be obtained from formazan precursors. 8 In previous studies, we found that bis(formazanate) zinc complexes such as [PhNNC(p-tol)NNPh] 2 Zn (1a) are capable of storing one or two electrons in the ligand frameworks to form stable singly-and doubly-reduced zinc complexes. 9 The crystal structures of the reduced complexes show (weak) interactions between the internal nitrogen atoms of formazanate ligands and sodium counter cations. Based on this observation we anticipated that the reduction potential of bis(formazanate) zinc compounds could be altered by coordination to neutral Lewis acids. In the course of testing a range of Lewis acids for binding to 1a, we found that BF 3 reacts cleanly via salt metathesis, opening a high-yield synthetic route to obtain mono(formazanate) boron difluoride (LBF 2 ) complexes which are difficult to access otherwise, either from the parent formazan or formazanate salts (LK or LNa). A related formazanate diacetate compound LB(OAc) 2 was described by Hicks and co-workers, 7b and although spectroscopic data suggested ligand-based 1-electron redoxchemistry to occur in these compounds, the 'borataverdazyl' radical anions obtained were too unstable to allow full characterization. Here we report the synthesis and X-ray crystallographic characterization of LBF 2 and the radical anion LBF 2 À , with the formation of a relatively stable 2-electron reduction product LBF 2 2À confirmed by cyclic voltammetry. In addition, we provide evidence for the transmetallation pathway by characterization of a likely intermediate. Mono(formazanate) boron difluoride complexes are readily accessible by transmetallation of bis(formazanate) zinc complexes with BF 3 ÁEt 2 O in hot toluene. In the case of (PhNNC( p-tol)NNPh) 2 Zn (1a), a stoichiometric reaction does not go to completion, but full conversion is achieved by performing the reaction with 3 equivalents of BF 3 ÁEt 2 O at 70 1C twice. During the reaction, a colour change from deep blue to red and the precipitation of a white solid (presumably ZnF 2 ) was observed to indicate formation of mono(formazanate) boron difluoride complex (PhNNC( p-tol)NNPh)BF 2 (2a, Scheme 1), which was isolated as an air-stable, crystalline material in 86% yield. A single crystal X-ray diffraction study (Fig. S1, ESI †) revealed a distorted tetrahedral boron centre, which is displaced out of the (planar) formazanate NNCNN backbone by 0.5 Å. A similar bonding mode was observed for b-diketiminate complexes of Sc and attributed to steric interactions. 10 In the absence of significant steric pressure in 2a, we ascribe the out-of-plane displacement of the B atom to packing effects; the observation of only one 19 F NMR resonance for 2a even at low temperature is consistent with a lowenergy C 2v symmetric structure through which the two 19 F environments exchange. As is the case in its parent zinc complex 1a and the related mono(formazanate) boron diacetate complexes reported by Hicks and co-workers, 7b full delocalization within the formazanate framework in 2a is indicated by the similar N-N and C-N bond lengths in the NNCNN backbone.
The redox-active nature of the formazanate ligand in 2a was established by cyclic voltammetry (CV) in THF solution with [Bu 4 N][PF 6 ] electrolyte. Upon scanning to negative potentials, the CV shows two quasi-reversible one-electron redox processes (system I/I 0 and II/II 0 , Fig. 1) that can be assigned to formazanate-based reductions. The first redox-couple of 2a occurs at more positive potential (E 1/2 (I/I 0 ) = À0.98 V vs. Fc 0/+ ) than that in 1a (À1.31 V vs. Fc 0/+ ) 9 and at similar potential as Hicks et al. reported for the analogous mono(formazanate) boron diacetate compound (À0.86 V vs. Fc 0/+ in CH 3 CN). 7b A second redox-event is observed at more negative potential (E 1/2 (II/II 0 ) = À2.06 V vs. Fc 0/+ ) but also this reduction is reversible. Based on these data, we infer that both one-and twoelectron reduction products of 2a are relatively stable and accessible. Importantly, the second reduction forms the dianion LBF 2 2À , in which two electrons have been added to a single formazanate ligand to result in a 'L 3À ' fragment coordinated to the boron centre (Scheme 1). Bard et al. recently explored the voltammetry of boron dipyrromethenes (BODIPY): in general these show oneelectron reductions at more negative potentials compared to 2a. 11 Related b-diketiminate compounds show irreversible reductions at much more negative potential (E p,red B À2.6 V vs. Fc 0/+ ). 12 The first redox-couple of 2a occurs at À0.98 V vs. Fc 0/+ , which suggests that cobaltocene is a suitable reducing agent to selectively synthesize a singly-reduced product. Upon reaction of 2a with one equivalent of Cp 2 Co in THF, an immediate colour change from red to green is observed to indicate formation of the radical species [Cp 2 Co] + [(PhNNC(p-tol)NNPh)BF 2 ] À (3a). In contrast to the reduction product of the diacetate analogue reported by Hicks, the difluoro complex 3a is stable in solution under an inert atmosphere for days and can even be boiled in THF-hexane without noticeable decomposition. The surprising stability of 3a is likely due to the strength of the B-F bonds in comparison to B-OAc. Crystals of 3a suitable for single-crystal X-ray diffraction analysis were obtained by diffusion of hexane into THF solution at room temperature in quantitative yield (Fig. 2, 9 suggest the ligand of 3a is a radical dianionic species (L 2À ) in which the unpaired electron occupies a N-N anti-bonding orbital.
UV-Vis absorption spectroscopy of neutral and anionic mono(formazanate) difluoride compounds provides additional evidence for ligand-based reduction and formation of the dianionic ligand radical (L 2À ) (Fig. S3, ESI †). The neutral compound 2a shows a single broad absorption in the visible at 521 nm. In the case of 3a, absorption is at longer wavelength (716 nm) and at 454 nm with shoulders at 674 and 431 nm, respectively. These absorption bands are in agreement with the presence of reduced formazanate ligands. 9,13 The EPR spectrum of 3a shows a broad EPR signal in frozen THF with g-value of B2 (Fig. S2, ESI †). Attempts to synthesize and characterize the two-electron reduction product have so far been unsuccessful.
In order to expand the scope of transmetallation reactions from bis(formazanate) zinc complexes and boron trifluoride, the reaction of (PhNNC(C 6 F 5 )NNMes) 2 Zn (1b) with BF 3 was attempted. The colour of the reaction mixture fades from deep to light orange upon heating 1b in the presence of 3 eq. BF 3 Et 2 O at 70 1C overnight, but precipitation of ZnF 2 was not observed. Orange crystals of the product 4b were obtained by slow diffusion of hexane into a toluene solution at À30 1C in 85% yield (Fig. 2, right). The crystal structure of 4b shows a distorted octahedral zinc centre. In the structure, there are two tridentate (PhNNC(C 6 F 5 )NNMes(BF 3 )) units coordinated to the Zn centre in a meridional fashion via two nitrogens and a fluorine atom to give a [NNF] 2 Zn compound. This unusual binding motif results from interaction of BF 3 with the terminal nitrogen of the formazanate fragment (the NMes group), which gives rise to 2 five-membered chelate rings upon coordination to the Zn centre. To the best of our knowledge, the structural characterization of this BF 3 Scheme 1 Synthesis of mono(formazanate) boron difluoride complexes and their reduction products.  binding mode has no precedent in the literature, although the 'frustrated Lewis pair' (tmp)MgCl/BF 3 has been postulated to contain a B-F fragment appended to a Mg-N(tmp) bond. 14 The 19 F-NMR of 4b shows six distinct resonances with integration ratio of 1 : 1 : 3 : 1 : 1 : 1 (Fig. S6, ESI †). Five resonances with the same integration (1F) suggest that all F substituents of the C 6 F 5 ring are inequivalent due to hindered rotation around the C-C 6 F 5 bond. The resonance integrating as 3F shows 11 B and 10 B coupling features and can be assigned to the BF 3 unit. The appearance of the BF 3 group in the 19 F NMR does not change upon cooling to À55 1C, which suggests that the barrier to rotation around the N-BF 3 bond is low. Conversely, heating an NMR sample of 4b to moderate temperature (65 1C for 24 h) does not result in changes in the spectroscopy, confirming that the octahedral [NNF] 2 Zn complex is quite stable. Upon heating the NMR tube to 130 1C overnight, full conversion to 2b is obtained. The 19 F NMR spectrum of the new species shows signals characteristic for a freely rotating C 6 F 5 group and a BF 2 unit ( 11 B NMR: À1.34 ppm, triplet with J B-F = 24 Hz). These data are consistent with formation of [PhNNC(C 6 F 5 )NNMes]BF 2 (2b). Cyclic voltammetry of 2b (Fig. 1) shows two quasi-reversible redox processes similar to 2a but shifted to more negative potential. This suggests that the electron-rich N-Mes group in 2b is more important than the electron-withdrawing C-C 6 F 5 moiety in modulating the redox-potential of the formazanate fragment.
The sequential transformation 1b -4b -2b suggests that a six-coordinate species related to 4b is likely also involved in the formation of 2a. Based on these observations, we propose the following mechanism for the transmetallation leading to compounds 4 (Scheme 2): (i) formazanate rearrangement from a 6-to a 5-membered chelate ring liberates the terminal N-atom, (ii) BF 3 binds to this terminal N-atom and brings a B-F group in proximity of the Zn centre, and (iii) the F atom binds to the Lewis acidic Zn(II) centre to from a tridentate [NNF] ligand with two 5-membered chelate rings. Repeating this sequence for the second formazanate ligand results in formation of the [NNF] 2 Zn complex 4. Elimination of ZnF 2 from this complex either occurs rapidly (in case of 1a -'4a' -2a), or requires heating to proceed so that the intermediate can be isolated (1b -4b -2b). A reason for the increased stability of 4b vs. that of putative intermediate 4a could be the favorable p-interactions between the electron-rich N-Mes and the electron-poor C-C 6 F 5 substituents, 15 which are present only when the formazanate ligands adopt a 5-membered chelate ring.
In conclusion, transmetallation of bis(formazanate)zinc complexes with BF 3 ÁEt 2 O provides a convenient entry into formazanate boron chemistry. The reaction likely occurs via initial binding of BF 3 to the formazanate ligand. This pathway is possible through the flexibility of the NNCNN backbone to adopt 5-membered chelate ring isomers, which are inaccessible to their b-diketiminate congeners. One-electron reduction of LBF 2 results in a fully characterized stable ligand-based radical, and cyclic voltammetry confirms a second reduction is possible, allowing access to three oxidation states (LBF 2 0/À1/À2 ) that are all based the redox-chemistry of a single formazanate ligand. The further development of this unique class of stable redox-active ligands towards applications in coordination chemistry and catalysis is the focus of ongoing work in our laboratory.