Coligand role in the NHC nickel catalyzed C–F bond activation: investigations on the insertion of bis(NHC) nickel into the C–F bond of hexafluorobenzene†

The reaction of [Ni(Mes2Im)2] (1) (Mes2Im = 1,3-dimesityl-imidazolin-2-ylidene) with polyfluorinated arenes as well as mechanistic investigations concerning the insertion of 1 and [Ni(iPr2Im)2] (1ipr) (iPr2Im = 1,3-diisopropyl-imidazolin-2-ylidene) into the C–F bond of C6F6 is reported. The reaction of 1 with different fluoroaromatics leads to formation of the nickel fluoroaryl fluoride complexes trans-[Ni(Mes2Im)2(F)(ArF)] (ArF = 4-CF3-C6F42, C6F53, 2,3,5,6-C6F4N 4, 2,3,5,6-C6F4H 5, 2,3,5-C6F3H26, 3,5-C6F2H37) in fair to good yields with the exception of the formation of the pentafluorophenyl complex 3 (less than 20%). Radical species and other diamagnetic side products were detected for the reaction of 1 with C6F6, in line with a radical pathway for the C–F bond activation step using 1. The difluoride complex trans-[Ni(Mes2Im)2(F)2] (9), the bis(aryl) complex trans-[Ni(Mes2Im)2(C6F5)2] (15), the structurally characterized nickel(i) complex trans-[NiI(Mes2Im)2(C6F5)] (11) and the metal radical trans-[NiI(Mes2Im)2(F)] (12) were identified. Complex 11, and related [NiI(Mes2Im)2(2,3,5,6-C6F4H)] (13) and [NiI(Mes2Im)2(2,3,5-C6F3H2)] (14), were synthesized independently by reaction of trans-[Ni(Mes2Im)2(F)(ArF)] with PhSiH3. Simple electron transfer from 1 to C6F6 was excluded, as the redox potentials of the reaction partners do not match and [Ni(Mes2Im)2]+, which was prepared independently, was not detected. DFT calculations were performed on the insertion of [Ni(iPr2Im)2] (1ipr) and [Ni(Mes2Im)2] (1) into the C–F bond of C6F6. For 1ipr, concerted and NHC-assisted pathways were identified as having the lowest kinetic barriers, whereas for 1, a radical mechanism with fluoride abstraction and an NHC-assisted pathway are both associated with almost the same kinetic barrier.


Introduction
Fluorinated organic compounds have exceptional properties that are being exploited in many applications including materials, pharmaceuticals and agrochemicals. The development of methods to introduce uorinated aromatic building blocks selectively into organic molecules is thus of fundamental interest in many areas of chemical research. 1 One strategy for such transformations is the selective activation and subsequent functionalization of C-F bonds of readily available uoroorganic compounds such as uoroaromatics. The challenge here is the selective cleavage of very stable C-F bonds. 2 We have recently established a protocol for the transformation of commercially available uoroaromatics via a selective C-F deuoroborylation process to obtain polyuorinated arylboronic esters, 3 which may be further used in late stage functionalization, for example in Suzuki-Miyaura cross-coupling reactions. 4 Deuoroborylation of polyuoroaromatics can be achieved by a thermal [Ni(Mes 2 Im) 2 ]-catalyzed (Mes 2 Im ¼ 1,3dimesityl-imidazolin-2-ylidene) transformation of poly-uoroarenes into uoroaryl boronic acid pinacol esters via C-F bond activation and transmetalation with bis(pinacolato) diboron (B 2 pin 2 ) as the boron source (see Scheme 1). 3a Various arenes with different degrees of uorination were converted into their corresponding boronate esters in this way. One particularly interesting nding of our study was that activation of the C-F bond by the nickel(0) complex is fast at ambient temperature. This step yields the oxidative addition product trans-[Ni(Mes 2 Im) 2 (F)(Ar F )] (Ar F ¼ uoroaryl), which represents the resting state in the catalytic cycle. The subsequent deuoroborylation step with B 2 pin 2 is the rate determining step and requires elevated temperatures. A boryl complex trans-[Ni(Mes 2 Im) 2 (Bpin)(Ar F )], a likely intermediate, was never observed and stoichiometric reactions of trans-[Ni(Mes 2 Im) 2 (-F)(Ar F )] with B 2 pin 2 led directly to the formation of Ar F -Bpin. This nding implied that reductive elimination is very fast and that [Ni(Mes 2 Im) n (Bpin)(Ar F )], once formed, will eliminate Ar F -Bpin immediately (Scheme 1). 5a As an alternative to the thermally-induced C-F bond activation and subsequent borylation of uoroarenes, we have recently developed a process that employs visible-light photocatalysis, which has emerged as a powerful tool in organic synthesis. 6 Our highly selective and general photocatalytic C-F borylation protocol 3b employs a rhodium biphenyl complex 7 as triplet sensitizer combined with the nickel catalyst [Ni(Mes 2 -Im) 2 ] (1) for the C-F bond activation step and the deuoroborylation process. This Rh/Ni tandem catalyst system operates with visible light (400 nm) and achieves the highly selective borylation of a wide range of polyuoroarenes with B 2 pin 2 at room temperature in excellent yields. Both procedures, the thermal and photochemical deuoroborylation, work well for partially uorinated aromatics but fail, or afford only low yields, for peruoroaromatics such as hexauorobenzene or octauorotoluene.
We report herein on the reactivity of 1 with polyuorinated arenes. We compare the results with those of earlier studies on C-F bond activation processes using nickel complexes with sterically less demanding NHCs, employing i Pr instead of Mes substituents, i.e., using [Ni( i Pr 2 Im) 2 ] (1 ipr ) as the nickel source. We demonstrate that the complex of the small NHC ligand i Pr 2 Im favors a concerted oxidative addition proceeding through an h 2 (C,C) intermediate in reactions with uoroarenes to yield trans-[Ni II (NHC) 2 (F)(Ar F )] complexes, whereas the complex of the larger Mes 2 Im ligand leads to uorine atom abstraction to yield [Ni I (NHC) 2 (F)] and a phenyl radical. For both mechanisms, competitive NHC-assisted pathways are found which account for the formation of diamagnetic products by a C-F bond activation step across the Ni-C NHC bond. These NHC-assisted pathways play an important role for complexes of both sterically demanding and less bulky NHC ligands, and should thus be of general importance and widely applicable for the reactivity of NHC-stabilized transition metal complexes.

C-F bond activation of uoroaromatics
To gain insight into the C-F bond activation process using [Ni(Mes 2 Im) 2 ] (1), we rst investigated stoichiometric reactions of peruorotoluene, peruorobenzene, peruoropyridine and the partially uorinated arenes pentauorobenzene, 1,2,3,5-tetrauorobenzene and 1,3,5-triuorobenzene with 1 (see Scheme 3). We monitored the reactions by 1 H and 19 F{ 1 H} NMR spectroscopy and observed a signicant effect of the degree of uorination on both reaction rate and yield. Reactions of 1 with hexauorobenzene and octauorotoluene proceed within seconds at room temperature, whereas the reactions with tetraand pentauorobenzene take minutes to complete. With 1,3,5-triuorobenzene, full conversion of 1 takes weeks at room temperature (see ESI, Fig. S1 †), but can be accelerated at 80 C in thf to reach completion aer 5 days.
Scheme 2 Stoichiometric C-F bond activation of C 6 F 6 using sources of [Ni( i Pr 2 Im) 2 ] 1 iPr . not seem to affect the yield of the insertion product 3. Complexes 2-7 were characterized by elemental analysis, 1 H, 19 F { 1 H} and 13 C{ 1 H} NMR spectroscopy (see ESI †). In the 19 F{ 1 H} NMR spectra of these complexes, the resonances of the nickelbound uoride ligand were observed in the typical range between À361.9 and À333.1 ppm. Within the series presented (see ESI, Table S1 †), the NMR shi of this resonance depends on the degree of uorination of the uoroaryl ligands, i.e., an increase of the degree of uorination of the aryl ligand leads to an upeld shi of the Ni-F resonance.
As the low yield of trans-[Ni(Mes 2 Im) 2 (F)(C 6 F 5 )] (3) is in sharp contrast with the results we obtained previously for the reaction of [Ni 2 ( i Pr 2 Im) 4 (m-(h 2 :h 2 )-COD)] or [Ni( i Pr 2 Im) 2 (h 2 -C 2 H 4 )] with C 6 F 6 , 9 we decided to take a closer look at the corresponding reaction using [Ni(Mes 2 Im) 2 ] (1). Performing the stoichiometric reaction of 1 with C 6 F 6 in an NMR tube in C 6 D 6 led to an immediate color change from dark-violet, the color of concentrated complex 1, to orange aer addition of C 6 F 6 at room temperature. A quantitative conversion of 1 was achieved aer 5 min as monitored by 1 H NMR spectroscopy (see ESI,Fig. S2 †). However, the spectroscopic yield determined by 19 F{ 1 H} NMR spectroscopy aer 5 min at room temperature, vs. a Ph-F containing capillary as internal standard, revealed the formation of 3 in approximately 17% yield and, in addition, the formation of small amounts of uoride-containing side products (see ESI,Fig. S3 †). Even aer 72 h at room temperature, no increase in the spectroscopic yield of 3 was observed. In further control experiments, neither the use of an excess of 1 (2.85 equiv.) nor C 6 F 6 (2.5 equiv.) increased the yield of 3 substantially. These experiments demonstrate that the low isolated yield of 3 is not a problem of the isolation process for this complex, but rather an intrinsic problem associated with its formation and the C-F bond activation step. Low temperature NMR experiments (À50 C to +20 C) revealed that a nickel uoride resonance at À358 ppm appeared for this reaction in the 19 F{ 1 H} NMR spectrum already at À50 C (see ESI, Fig. S4 †), but also that, at these temperatures, all resonances are signicantly broadened in the 1 H NMR spectrum of the reaction mixture (see ESI, Fig. S5 †). Although we previously observed some line Scheme 3 The reactions of [Ni(Mes 2 Im) 2 ] (1) with (a) octafluorotoluene, (b) hexafluorobenzene, (c) perfluoropyridine, (d) pentafluorobenzene, (e) 1,2,3,5-tetrafluorobenzene and (f) 1,3,5trifluorobenzene to give the complexes trans-[Ni(Mes 2 Im) 2 (F)(4-CF 3 -C 6 F 4 )] (2), trans-[Ni(Mes 2 Im) 2 (F)(C 6 F 5 )] (3), trans-[Ni(Mes 2 Im) 2 (-F)(2,3,5,6-C 5 F 4 N)] (4), trans-[Ni(Mes 2 Im) 2 (F)(2,3,5,6-C 6 F 4 H)] (5), trans-[Ni(Mes 2 Im) 2 (F)(2,3,5-C 6 F 3 H 2 )] (6) and trans-[Ni(Mes 2 Im) 2 (F)(3,5-C 6 F 2 H 3 )] (7), respectively. Isolated yields are given. broadening for the N-alkyl groups of the related complex trans-[Ni( i Pr 2 Im) 2 (F)(C 6 F 5 )], 9a which arose due to hindered rotation of the NHC ligand about the Ni-C axis, all resonances observed for the reaction of 1 with C 6 F 6 are involved in the broadening. This led to the assumption that radical species are involved in the process. Subsequent EPR experiments were performed at À203 C for the reaction of 1 with C 6 F 6 which conrmed the presence of metal-centered radicals in the mixture.
For EPR spectroscopic investigations, 1 and C 6 F 6 were combined in an EPR tube with thf at À78 C and the sample was frozen immediately in liquid nitrogen. The EPR tube containing the frozen reaction mixture was transferred to the cooled EPR cavity at À203 C and a spectrum was recorded. 10 The resulting EPR spectrum displays a superposition of resonances of three different products, of which I and II represent the two dominant species (Fig. 2, I: 40%, II: 50%, III: 10%).
Cyclic voltammetry results exclude a simple electron transfer from 1 to C 6 F 6 as the origin of radical generation in the reaction mixture (see ESI, Fig. S6 †), as 1 shows a reversible oxidation/ reduction associated with a redox potential of À2.03 V for the redox-couple Ni 0 /Ni I , and an irreversible oxidation at 0.14 V for the redox-couple Ni I /Ni II . Although the reduction of C 6 F 6 at À2.87 V is irreversible, we exclude simple one electron transfer because of the large separation of 0.84 V.
For further scrutiny, complex 1 was oxidized by adding ferrocenium tetrauoroborate in thf at room temperature to a suspension of 1 in thf. A few min aer addition of the ferrocenium salt the metal-centered radical [Ni(Mes 2 Im) 2 ][BF 4 ] (8) precipitated as an off-white solid (83% isolated yield, Scheme 4), which is only sparingly soluble in common organic solvents. The Ni I complex 8 was characterized by 11 B{ 1 H} and 19 F{ 1 H} NMR spectroscopy in acetonitrile (decomposition occurs aer Table 1 Crystallographic data for compounds 1, 3,4,5,6,8,9,11,13,14, Table 1; see also ESI, Table S2 and Fig. S37 †) were obtained by slow evaporation of a saturated solution of 8 in a 1 : 1 toluene/ ethanol mixture under an argon atmosphere at room temperature. The X-ray crystal structure reveals a nearly linear alignment of the NHC ligands with slightly elongated Ni-C distances compared to those of the starting material 1.
The calculated g values for the radical cation [Ni(Mes 2 Im) 2 ] + (g xx ¼ 2.01, g yy ¼ 2.65, g zz ¼ 2.98), computed under gas-phase conditions, strongly differ from the experimental data with a maximum deviation of 0.36 (8a) and 0.85 (8b; see Table 2 and ESI, Table S3 †). However, computations in the presence of the counter ion result in further structural motifs with impact on    (Table 2, entry 4, Fig. 5b) results in g tensor components closely corresponding to those of 8a (maximum deviation: 0.03), while no species matching the EPR parameters of 8b were identied in our computational exploration. However, none of the EPR signatures detected for the electrochemically-formed complex 8 appeared during the reaction of 1 with C 6 F 6 ( Fig. 2) and, in light of our CV results, it is unlikely that the [Ni(Mes 2 Im) 2 ] + cation is involved here. We then focused on identifying the byproducts of the reaction of 1 with C 6 F 6 . Stoichiometric reaction of 1 with C 6 F 6 in thf overnight at room temperature led to a very small amount of a dark-green precipitate which was removed by ltration. Aer removal of all volatiles from the ltrate, the residue was washed with a large amount of hexane to extract the C-F bond activation product. The yellow residue, which remained aer washing, was identied as the diuoride complex trans-[Ni(Mes 2 Im) 2 (F) 2 ] (9) by elemental analysis, X-ray diffraction and 1 H, 19 F{ 1 H} and 13 C { 1 H} NMR spectroscopy (see ESI †). Most signicantly, the uoride resonance, detected as a singlet at À560 ppm in the 19 F{ 1 H} NMR spectrum, is shied ca. 200 ppm to higher eld compared to those of the mono-uoride complexes 2-7 (À333 ppm to À362 ppm, vide supra). A similar high-eld shied uoride resonance was also observed for the phosphine-stabilized platinum complex [Pt(P i Pr 3 ) 2 (F) 2 ] (À455.9 ppm) compared to [Pt(PPh 3 ) 2 (F)(C 6 H 5 )] (À107.6 ppm). 13 Crystals of 9 suitable for Xray diffraction (Fig. 6, Table 1; see also ESI, Table S2 and Fig. S38 †) were obtained aer storing a saturated solution of the complex at room temperature in C 6 D 6 . Crystallographic analysis revealed a square planar coordination environment about the Ni II center with a trans-arrangement of NHC and uoride ligands.
An independent sample of complex 9 was synthesized in 38% yield by uorination of [Ni(Mes 2 Im) 2 (I) 2 ] (10) using an excess (2.5 equiv.) of silver(I) uoride in CH 2 Cl 2 at 0 C (Scheme 5). Complex 10 was synthesized by reaction of 1 with I 2 , isolated in 80% yield and characterized by elemental analysis, and 1 H and 13 C{ 1 H} NMR spectroscopy (see ESI †). Interestingly, the resonance of the carbene carbon atoms is almost unaffected by substitution of the uoride by the more electropositive iodide ligand, and was detected at 176.5 ppm (cf. [Ni(Mes 2 -Im) 2 (F) 2 ] (9): 174.6 ppm). Thus, [Ni(Mes 2 Im) 2 (F) 2 ] (9) was clearly identied as one of the side products of the reaction of 1 with C 6 F 6 . This complex is formed in low yield (17%) but in an amount similar to that of the insertion product trans-[Ni(Mes 2 Im) 2 (F)(C 6 F 5 )] (3). The amounts of complexes 9 and 3 total ca. 40% when the reaction of 1 with C 6 F 6 is performed at room temperature, and thus the majority of the products formed in this reaction is still unaccounted for.
Storing the concentrated hexane mother liquor of the extract from the isolation of 9 (vide supra) for 3 days at À30 C led to crystallization of the remaining C-F bond insertion product trans-[Ni(Mes 2 Im) 2 (F)(C 6 F 5 )] (3) and a novel nickel(I) complex trans-[Ni I (Mes 2 Im) 2 (C 6 F 5 )] (11) as yellow (3) and orange (11) crystals, respectively, which were manually separated in a glovebox (see ESI, Fig. S8 †). The paramagnetic compound 11 was characterized by elemental analysis, EPR spectroscopy and Xray diffraction. Determination of the room-temperature magnetic moment of 11 in solution (Evans method) gave a m eff value of 1.80 m B , which is consistent with the presence of one unpaired electron. The molecular structure (Fig. 7, top, Table 1; see also ESI, Table S2 and Fig. S39 †) and the EPR spectrum (Fig. 7, bottom) of 11 conrm that this complex is a threecoordinate nickel(I) radical. Simulation of the EPR spectrum of 11 gave a g tensor of g xx ¼ 2.04, g yy ¼ 2.16 and g zz ¼ 2.31, which was also observed in the EPR spectrum of the crude reaction mixture of 1 and C 6 F 6 (Fig. 2). With the experimentally obtained g tensors and the molecular structure of the radical species 11 in hand, we carried out computational studies of the electronic properties of complex 11 and a likely radical counterpart from the reaction of 1 and C 6 F 6 , [Ni I (Mes 2 Im) 2 (F)] (12) (Fig. 8). Both complexes 11 and 12 would be the result of a oneelectron oxidative addition reaction of two equiv. of 1 with one equiv. C 6 F 6 (Scheme 6).
Molecular geometries, electronic structures and EPR parameters (g tensors) were thus calculated for the metal radicals trans-[Ni I (Mes 2 Im) 2 (C 6 F 5 )] (11) and trans-[Ni I (Mes 2 Im) 2 (F)] (12) (Fig. 8) in order to connect the experimentally observed EPR spectra from the reaction mixture of 1 and C 6 F 6 (Fig. 2), the EPR spectra of the isolated compound 11, and the corresponding isotropic g tensor components with the assigned structure of 11 (Fig. 7).
According to DFT calculations, complexes 11 and 12 are C 2symmetric doublet ground state species. The spin density is located at the metal center and the unpaired electron resides in an s/d z 2-type orbital, yielding 2 A electronic ground states (Fig. 8). Calculated and experimental g tensor components are in good agreement for species 11, with a maximum difference of 0.03 in g zz . With the largest deviation being 0.08 for 12, the agreement is still reasonable (Table 3).
The complex trans-[Ni I (Mes 2 Im) 2 (C 6 F 5 )] (11) as well as related trans-[Ni I (Mes 2 Im) 2 (2,3,5,6-C 6 F 4 H)] (13) and trans-[Ni I (Mes 2 Im) 2 (2,3,5-C 6 F 3 H 2 )] (14) can be synthesized from the reaction of trans-[Ni(Mes 2 Im) 2 (F)(Ar F )] (Ar F ¼ C 6 F 5 3, 2,3,5,6-C 6 F 4 H 5, 2,3,5-C 6 F 3 H 2 6) with PhSiH 3 (Scheme 7, see also ESI Fig. S9 and S10 †). 14 The metal radicals were characterized by elemental analysis, IR and EPR spectroscopy as well as single-crystal X-ray diffraction. All compounds are stable in the solid state as well as in solution for several days. If the reactions are performed in an NMR tube and followed by 1 H and 19 F{ 1 H} NMR spectroscopy (see ESI; Fig. S9 and S10 †), the resonances for the Mes 2 Im, pentauorophenyl and uoride ligands vanish, indicating the formation of a paramagnetic species. For complexes of the type trans-[Ni(NHC) 2 (H)(Ar F )], we expect hydride resonances in the region of ca. À13 ppm in the 1 H NMR spectrum, 9b,d and a strong absorption in the IR spectrum in the region between 1600 and 2200 cm À1 15 (we expect the Ni-H stretch to be at ca. 1850 cm À1 based on DFT calculations). However, such signals were absent for 11, 13 and 14. Thus, although complexes of the type trans-[Ni I (Mes 2 Im) 2 (Ar F )] cannot easily be distinguished from the corresponding hydride complexes trans-[Ni I (Mes 2 Im) 2 (H)(Ar F )] by X-ray diffraction (see below), we are condent that 11, 13 and 14 are the metal radicals. Crystals of trans-[Ni I (Mes 2 Im) 2 (C 6 F 5 )] (11), trans-[Ni I (Mes 2 Im) 2 (2,3,5,6-C 6 F 4 H)] (13) and trans-[Ni I (Mes 2 Im) 2 (2,3,5-C 6 F 3 H 2 )] (14) suitable for X-ray diffraction (Fig. 9, Table 1; see also ESI Table S2 and Fig. S39-S41 †) were obtained by storing saturated solutions of these compounds either in pentane or hexane at À30 C. Complexes 11-13 adopt a distorted T-shaped structure, in which the NHC ligands occupy mutually trans positions. Due to the absence of the uoride ligand, 11, 13 and 14 exhibit shortened Ni-C distances to the uoroaryl ligand and reduced C1-Ni-C2 angles compared to nickel(II) complexes 3, 4, 5 and 6, which is also a further indication of the absence of a metal hydride. The data is in line with the data observed for [Ni I (P i Pr 3 ) 2 (C 6 F 5 )] reported by Johnson and co-workers previously (Table 1, see also ESI Table  S2 †). 16 EPR spectra of compounds 11, 13 and 14 were recorded in frozen thf solutions and reveal similar g tensors for the complexes, which are in good agreement with the calculated parameters (see ESI, Fig. S11-S13 and Table S4 †).

Mechanistic investigations
Experimental investigations and DFT studies reported previously 9a for the reaction of [Ni 2 ( i Pr 2 Im) 4 (m-(h 2 :h 2 )-COD)] and [Ni( i Pr 2 Im) 2 (h 2 -C 2 H 4 )], used as source of [Ni( i Pr 2 Im) 2 ] (1 ipr ), with C 6 F 6 suggested a concerted mechanism for the insertion of 1 ipr into the C-F bond, and no indications for radical reactivity were obtained. As presented above, however, paramagnetic complexes clearly emerge in the reaction of 1 and C 6 F 6 . To obtain further insight, we performed a quantum-chemical investigation (COSMO(THF)-PBE0-D/def2-TZVP, for details see ESI †) 17 on the reaction pathways of C 6 F 6 with [Ni(Mes 2 Im) 2 ] (1) and with the sterically less encumbered [Ni( i Pr 2 Im) 2 ] (1 ipr ).
C-F bond activation in the latter reaction commences with the formation of a rather stable 16-electron h 2 adduct between 1 ipr and C 6 F 6 (I1, Scheme 8; see ESI, Fig. S17 †). The DFToptimized geometry of I1 is in good agreement with the structure of the closely related complex [Ni( i Pr 2 Im) 2 (h 2 -C 10 F 8 )]. 9a Three distinct reaction pathways are then possible. First, direct oxidative addition of the C-F bond to the nickel atom proceeds through TS1 to yield the trans product 3 ipr with an effective activation barrier of D ‡ G ¼ 23 kcal mol À1 relative to I1 (see ESI, Fig. S18 †). Alternative formation of the corresponding cis-[Ni( i Pr 2 Im) 2 (F)(C 6 F 5 )] (I2) and subsequent isomerization is kinetically disfavored (D ‡ G eff ¼ 27 kcal mol À1 , see ESI Fig. S19 and S20 †), as is dissociation of an NHC ligand (DG 298 ¼ 28 kcal mol À1 , see Fig. S32 †).
Second, NHC ligand cooperativity (see ESI; Fig. S21 and S23 †) opens a kinetically competitive pathway to the transproduct 3 ipr , that is, addition of the C-F bond across the Ni-C NHC bond through TS2 to yield intermediate I3, in which coordination of the uorinated NHC-F ligand to the nickel atom involves a bridging C-N bonding interaction. In TS2, the C aryl -F bond of 1.93 A is strongly elongated compared to C 6 F 6 (C aryl -F bond: 1.32 A) and TS1 (C aryl -F bond: 1.77 A), while NHC-F bond formation is hardly visible (C/F distance: 2.40 A). From I3, uoride migration onto the nickel ion (TS3, with a low barrier of D ‡ G ¼ 15 kcal mol À1 ) leads to 3 ipr with an overall barrier of D ‡ G eff ¼ 24 kcal mol À1 . Third, homolytic C-F bond cleavage involves an effective barrier of D ‡ G eff ¼ 31 kcal mol À1 (TS4) and, hence, radical abstraction is kinetically disfavored here (see ESI, Fig. S22 †).
C-F bond activation with the sterically more congested Mes-NHC complex 1 shows marked differences. Formation of the h 2 -C 6 F 6 adduct I5 (see ESI, Fig. S24 †) is now endergonic by 12 kcal mol À1 , and consecutive oxidative C-F bond addition via TS5 (D ‡ G eff ¼ 21 kcal mol À1 , see ESI; Fig. S25 †) leads to the cisproduct I6. We attribute the endergonicity of the h 2 -C 6 F 6 adduct formation (I5, DDG ¼ 28 kcal mol À1 compared to the exergonic formation of I1) mainly to the increased steric demand of the mesityl groups. A trajectory to the trans-product is precluded by the steric demand of the mesityl substituents. NHC dissociation to yield [Ni(Mes 2 Im)(h 6 -C 6 F 6 )] and subsequent insertion into the C-F bond is associated with a large barrier (D ‡ G eff ¼ 34 kcal mol À1 , see ESI; Fig. S32 and S33 †) and is Scheme 8 Calculated pathways for the C-F bond activation of C 6 F 6 with 1 ipr (DG 298 in kcal mol À1 ).
This journal is © The Royal Society of Chemistry 2020 Chem. Sci., 2020, 11, 11009-11023 | 11017 irrelevant here. Note that an alternative adduct formation stabilized by p-stacking interactions between C 6 F 6 and one of the NHC mesityl substituents, 18 such as I7 (see ESI; Fig. S26 †), is also endergonic and less favorable than I5. Furthermore, a "concerted" NHC-assisted process as in the i Pr system does not exist. We found a multi-step sequence for the mesityl system instead (Scheme 9 and ESI; Fig. S27 †), commencing with heterolytic C-F bond cleavage in I5, which exhibits a partially reduced C 6 F 6 fragment (q NPA (C 6 F 6 ) ¼ À0.69). The uoride anion expelled from the nickel coordination sphere is loosely held within the cle formed by the mesityl substituents in I8. A similar stabilizing association of a uoride anion by the methyl Scheme 9 Calculated pathways for the heterolytic C-F bond cleavage of C 6 F 6 by 1 and further reaction steps (DG 298 in kcal mol À1 ).
Scheme 10 Calculated pathways for the homolytic C-F bond cleavage of C 6 F 6 by 1 and further radical reaction steps (DG 298 in kcal mol À1 ; energies of TS10 and 9 are given relative to 12 + C 6 F 6 ). groups of mesityl substituents has been reported by Macgregor et al. for the C-F bond activation step in hydrodeuorination reactions. 19 Formation of the trans-product 3 from here involves binding to the carbene carbon atom and subsequent F-shi onto the Ni center. The overall path involves a low effective barrier of 16 kcal mol À1 (TS6).
Fluorine radical abstraction to yield C 6 F 5 and radical complex 12 via TS9 is slightly endergonic and exhibits a barrier of 16 kcal mol À1 (Scheme 10 and ESI, Fig. S26 and S28 †). Recombination of C 6 F 5 and 12 to 3 then provides a large thermodynamic driving force. Alternative addition of C 6 F 5 to the initial complex 1 to yield radical species 11 is also a highly exergonic process (À69.5 kcal mol À1 ), as well as addition of a second equivalent of C 6 F 5 to yield 15 (À108.0 kcal mol À1 ). Endergonic formation of diuoride complex 9 from 12 and another equiv. of C 6 F 6 , can be compensated by consumption of C 6 F 5 ; however, a second uorine abstraction step is prevented by the high kinetic barrier of 37 kcal mol À1 via TS10 (see ESI; Fig. S29 †). The mechanism for the formation of 9 remains obscure to us thus far. We compute the ligand exchange reaction 3 + 3 / 9 + 15 to be exceedingly endergonic (25.7 kcal mol À1 ), and also the disproportionation reactions of radicals 11 and 12 yielding 1 + 15 (31.0 kcal mol À1 ) or 1 + 9, (14.9 kcal mol À1 ), are unlikely to contribute to the formation of 9 (see ESI, Fig. S30 †). A dinuclear complex [{Ni(Mes 2 Im) 2 } 2 (m-(h 2 :h 2 )-C 6 F 6 )], which would be an intermediate for an oneelectron oxidative addition, is too high in energy to be considered (35 kcal mol À1 , see ESI; Fig. S31 †). Hence both, the radical pathway and the NHC-assisted multistep pathway represent kinetically competitive C-F bond activation steps in the reaction with [Ni(Mes 2 Im) 2 ] (1).

Discussion
It is now well established that nickel(0) complexes with phosphine, carbene, and even some nitrogen ligands undergo C-F oxidative addition with peruoroarenes to yield complexes trans-[Ni(L) 2 (F)(C 6 F 5 )]. 1,9,20 Although the lack of clean kinetics for many of the C-F oxidative additions indicate complex mechanistic scenarios, there were strong indications that the conversion of C 6 F 6 to the aryl uoride complex follows the same type of mechanism as observed for typical C-H activation reactions of benzene. It has been demonstrated, for nickel NHC and phosphine complexes, that the rst stage of C-F oxidative addition is the h 2 -coordination of the uoroarene. 1g, 9a,21,22 The introduction of uorine substituents on the arene results in a lower lying LUMO, which renders the uorinated arene a better electron acceptor compared to H-arenes and makes the reaction of electron-poor C 6 F 6 with an electron-rich, suitable nickel precursor more exothermic. The uoroarene of [Ni(L) 2 (h 2 -C 6 F 6 )] is ene-diene distorted, and the arene uoride substituents are bent out of the plane, as observed for I1 and I5. Subsequent C-F oxidative addition is strongly exothermic for trans-[Ni( i Pr 2 Im) 2 (F)(C 6 F 5 )] (DG 298 ¼ À57 kcal mol À1 ) and trans-[Ni(Mes 2 Im) 2 (F)(C 6 F 5 )] (DG 298 ¼ À52 kcal mol À1 ). Computational studies reported previously 9a,23 of the reaction pathways have supported the idea of concerted mechanisms involving a s-complex as a three-center transition state between the C 6 F 6 carbon and uorine atoms and the transition metal atom. The transition state structures typically show limited elongation of the C-F bond and interaction of the electron-rich transition metal ion with the C-F s* orbital leads to C-F bond breaking and formation of the M-C and M-F bond. We have demonstrated now for [Ni 2 ( i Pr 2 Im) 4 (m-(h 2 :h 2 )-COD)] and the related [Ni( i Pr 2 Im) 2 ] (1 ipr ) synthon complexes that C-F bond activation of C 6 F 6 occurs via both a concerted and an NHC-assisted pathway, as both are associated with very similar kinetic barriers of D ‡ G eff ¼ 23 kcal mol À1 for the concerted and of D ‡ G eff ¼ 24 kcal mol À1 for the NHC-assisted pathway. This situation will probably change if other substrates with other leaving groups, such as partially uorinated arenes, uoropyridines or other aryl halides, are involved in the reaction with the nickel complex; however, our calculations demonstrate that both reaction paths are feasible, at least for uoroarenes.
The direction of the concerted oxidative addition in TS1 to give the trans product is rather unusual. 24 For the oxidative addition of A-B to d 10 -ML 2 the important orbital interactions of the transition state are those between the lled s(A-B) orbital and the empty d s -type orbital of the metal, leading to electron donation from A-B to the metal center, and a second interaction between the lled d p -orbital of the metal and the s*(A-B), leading to electron transfer from the metal to the ligand. Strong back-donation will lead to ssion of the A-B bond. This backdonation is strongest if A-B lies within the bent-d 10 -ML 2 plane and the s*(A-B) orbital can interact with the d x 2 -y 2 orbital (actually a d-p hybrid orbital), which is pointing at the two ligands L. 24 However, it was also shown previously that concerted oxidative addition reactions may take place through a nonplanar transition state structure even for non-polar substrates with dihedral angle between ML 2 and M(A-B) planes larger than 70 . 25 It was demonstrated that this nonplanar transition state is connected to the planar product on the singlet surface and suggested that steric rather than electronic factors are responsible for the nonplanar transition state structure. Martin et al., 25c for example, calculated at the B3LYP/LanL2DZ-level of theory a nonplanar transition state for the oxidative addition of C 6 H 5 -I to [Pd(dmpe)] (dmpe ¼ bis{dimethylphosphino}ethane), in which the P-Pd-P and C-Pd-I planes are almost perpendicular to one another. Another example was provided by Jones et al. 25d for the oxidative addition of the C-CN s-bond of organonitriles to the low-valent nickel complex [Ni(dmpe)]. The C-C-N plane of the transition state (calculated at the B3LYP/6-31G(d,p)-level of theory), which leads to C-CN bond cleavage, is rotated by 38 relative to the P-Ni-P plane.
The h 2 (C,C)-bonded complex [Ni( i Pr 2 Im) 2 (h 2 -C 6 F 6 )] (I1) is also the crucial reaction intermediate for the NHC-assisted pathway. The key step here is the addition of the C-F bond across the Ni-C NHC bond and, thus, the unoccupied NHC p porbital plays a central role for this pathway as intramolecular uoride acceptor. Fluoride transfer from the arene to the NHC leads to a h 2 -uoro-imidazolyl intermediate (I3; Scheme 8) which rearranges with a second uoride transfer step from the NHC to the nickel atom to give trans-[Ni( i Pr 2 Im) 2 (F)(C 6 F 5 )] (3 ipr ).
A phosphine-assisted process has been proposed before for the C-F bond activation of pentauoropyridine with [Ni(PR 3 ) 2 ], based on the experimental observation of an unusual selectivity for the insertion into the 2-position of C 5 NF 5 and on DFT calculations. 26 However, another study performed on the reaction of pentauoropyridine with [Ni(PEt 3 ) 2 ] suggested that pathways other than a concerted oxidative addition or a phosphine-assisted pathway account for the unusual selectivity. 27 The detailed experimental analysis of the reactivity of a [Ni(PEt 3 ) 2 ] precursor with peruoropyridine demonstrated the formation of a mononuclear adduct [Ni(PEt 3 ) 2 (h 2 -C 5 F 5 N)], of dinuclear adducts [{Ni(PEt 3 ) 2 } 2 )(m-(h 2 :h 2 )-C 5 F 5 N)], some of which exhibit C-F bond activation, and a nickel(I) radical species [Ni(PEt 3 ) 2 (2-C 5 F 4 N)]. Other heteroatom-assisted C-F bond activation processes have also been proposed for other metals mainly including boryl or silyl moieties. 28 Despite precedent in the oxidative addition of other aryl carbon-halide bonds to nickel, 29,30 there is only little experimental evidence for the involvement of radicals in C-F bond activation processes. It is known that some polyuoro pyridines react with [Ni(PR 3 ) 2 ] to yield EPR-active complexes as likely intermediates, 16,27 and some studies on C-F bond activation have shown unusual products with highly-uorinated arenes that may be indicative of radical pathways. 16,22,31 However, the clear identication of radical intermediates has not been possible so far and alternate mechanisms cannot be ruled out. Although DFT calculations were performed to examine the traditional concerted oxidative addition and phosphineassisted pathways for C-F bond activation, radical pathways involving Ni(I) intermediates were rarely considered computationally.
Thus, the reaction of 1 with different uoroarenes leads to nickel insertion into the C-F bond to give the nickel uoroaryl uoride complexes trans-[Ni(Mes 2 Im) 2 (F)(Ar F )], but EPR spectroscopy also provided evidence that at least three paramagnetic species are intermediates or products of the reaction of C 6 F 6 with 1. We provide evidence that simple electron transfer from [Ni(Mes 2 Im) 2 ] (1) to C 6 F 6 , oen considered as the rst step in radical oxidative additions at nickel, 29 is unlikely to occur. The redox potentials are not in line with intermolecular electron transfer to yield [Ni(Mes 2 Im) 2 ] + and C 6 F 6 À and the EPR resonance of [Ni(Mes 2 Im) 2 ] + , which has been established for the authentic complex [Ni(Mes 2 Im) 2 ][BF 4 ] (8), was not detected in the reaction mixture. Furthermore, many diamagnetic and radical products of the reaction of [Ni(Mes 2 Im) 2 ] (1) to C 6 F 6 were identied, namely the insertion product trans-[Ni(Mes 2 Im) 2 (-F)(C 6 F 5 )] (3), the diuoride complex trans-[Ni(Mes 2 Im) 2 (F) 2 ] (9), the bis(aryl) complex trans-[Ni II (Mes 2 Im) 2 (C 6 F 5 ) 2 ] (15), the nickel(I) complex trans-[Ni I (Mes 2 Im) 2 (C 6 F 5 )] (11), and the metalcentered radical trans-[Ni I (Mes 2 Im) 2 (F)] (12). DFT calculations performed on the reaction of [Ni(Mes 2 Im) 2 ] (1) with C 6 F 6 explain the occurrence of the radical species observed. Both an NHC-assisted and a radical process are kinetically equally favored routes for this reaction. Fluorine radical abstraction from C 6 F 6 by 1 is associated with a barrier of only 16 kcal mol À1 and subsequent radical recombination steps provide the thermodynamic driving force required.
Matsubara et al. and Louie et al. reported the clean isolation of T-shaped three-coordinate radical species [Ni I (NHC) 2 (X)] (X ¼ Cl, Br, I; NHC ¼ Mes 2 Im, Dipp 2 Im) from the reaction of [Ni(NHC) 2 ] with aryl halides. 30a,b,d We have demonstrated earlier that [Ni 2 ( i Pr 2 Im) 4 (m-(h 2 :h 2 )-COD)], a source of [Ni( i Pr 2 Im) 2 ] (1 ipr ), reacts cleanly with aryl chlorides to yield the nickel(II) complexes trans-[Ni(NHC) 2 (Cl)(Ar)]. 32 Our calculations show now that a trajectory to the trans-product by a concerted oxidative addition is precluded for [Ni(Mes 2 Im) 2 ] (1) (and most probably also for [Ni(Dipp 2 Im) 2 ]) by the steric demand of the mesityl substituents. As a consequence, other pathways such as electron transfer and radical abstraction must occur which are responsible for a limited or altered reactivity of complex [Ni(Mes 2 Im) 2 ] (1) and analogues containing even more bulky Naryl substituents compared to complexes of sterically less demanding NHCs. However, uoride abstraction occurs for the reaction of 1 and C 6 F 6 even at À78 C to yield trans-[Ni I (Mes 2 -Im) 2 (C 6 F 5 )] (11) and trans-[Ni I (Mes 2 Im) 2 (F)] (12). The latter is, in contrast to the complexes of the heavier homologues, very reactive and has deed thus far isolation. In turn, the complexes trans-[Ni I (Mes 2 Im) 2 (C 6 F 5 )] (11), [Ni I (Mes 2 Im) 2 (2,3,5,6-C 6 F 4 H)] (12) and [Ni I (Mes 2 Im) 2 (2,3,5-C 6 F 3 H 2 )] (13) seem to be much more stable than [Ni I (NHC) 2 (C 6 H 5 )] and have been synthesized and characterized. The increased stability of [Ni I (Mes 2 -Im) 2 (2,3,5,6-C 6 F 4 H)] (12) can be explained by the increased Ni-C Ar bond strength of the uoroaryl ligand with respect to C 6 H 5 . 33 Nelson and Maseras 34 reported computational investigations of the reaction of [Ni(NHC) 2 ] complexes with aryl halides Ph-X (X ¼ Cl, Br, I) and demonstrated that steric effects determine the mechanism. Small NHC ligands (NHC ¼ Me 2 Im Me ) favor concerted oxidative addition via a h 2 (C,C) p-coordinated intermediate leading to trans-[Ni II (NHC) 2 (X)(Ar)] complexes whereas larger NHC ligands (e.g. NHC ¼ Mes 2 Im) lead to halide abstraction to form [Ni I (X)(NHC) 2 ] and a phenyl radical. We conrm here, by means of experiment and theory, that [Ni(NHC) 2 ] complexes of sterically less demanding NHCs favor the reaction with uoroarenes via a concerted oxidative addition proceeding through an h 2 (C,C) intermediate, and that for the bulkier NHC Mes 2 Im, C-F bond activation is achieved more easily by uorine atom abstraction. However, for both mechanisms, we found an NHC-assisted pathway which is competitive, that accounts for the formation of diamagnetic products by a C-F bond activation step across the Ni-C NHC bond. NHCassisted pathways play an important role for complexes of both sterically demanding and less bulky NHC ligand. We believe that this dual reaction pathway concept, including NHCassisted reaction pathways, should be of general importance and widely applicable for the reactivity of NHC transition metal complexes.

Conclusions
We present herein a detailed account of the C-F bond activation of polyuoroaromatics, especially of C 6 F 6 using the nickel(0) complex [Ni(Mes 2 Im) 2 ] (1). The reaction of 1 with different uoroarenes leads to insertion of nickel into the C-F bond of the