Cationic Heterobimetallic Mg(Zn)/Al(Ga) Combinations for Cooperative C–F Bond Cleavage

Abstract Low‐valent (MeBDI)Al and (MeBDI)Ga and highly Lewis acidic cations in [(tBuBDI)M+⋅C6H6][(B(C6F5)4 −] (M=Mg or Zn, MeBDI=HC[C(Me)N‐DIPP]2, tBuBDI=HC[C(tBu)N‐DIPP]2, DIPP=2,6‐diisopropylphenyl) react to heterobimetallic cations [(tBuBDI)Mg–Al(MeBDI)+], [(tBuBDI)Mg–Ga(MeBDI)+] and [(tBuBDI)Zn–Ga(MeBDI)+]. These cations feature long Mg–Al (or Ga) bonds while the Zn–Ga bond is short. The [(tBuBDI)Zn–Al(MeBDI)+] cation was not formed. Combined AIM and charge calculations suggest that the metal–metal bonds to Zn are considerably more covalent, whereas those to Mg should be described as weak AlI(or GaI)→Mg2+ donor bonds. Failure to isolate the Zn–Al combination originates from cleavage of the C−F bond in the solvent fluorobenzene to give (tBuBDI)ZnPh and (MeBDI)AlF+ which is extremely Lewis acidic and was not observed, but (MeBDI)Al(F)‐(μ‐F)‐(F)Al(MeBDI)+ was verified by X‐ray diffraction. DFT calculations show that the remarkably facile C–F bond cleavage follows a dearomatization/rearomatization route.

TheK + cation in these and related systems has been shown to play an eminent role in their reactivity and selectivity, [30] for example,i nt he para-selective dialumination of benzene via transition state II. [29] Most recently,w er eported similar RMgNa complexes (III)w hich based on metal electronegativities and reactivity should be seen as RMg À Na + reagents with nucleophilic magnesyl units. [31] These electron-rich reagents have considerable potential in bond activation by oxidative addition or in creation of Mg-metal bonds.T he advantages of mixing metals is not restricted to main group metals but is also actively pursued in transition metal catalysis. [32][33][34][35] We and others recently reported as eries of cationic bdiketiminate alkaline earth (Ae) metal complexes (IV) [36,37] which, being free of additional Lewis bases,d isplay high Lewis acidity. [38][39][40] It was shown that ac ombination of the Lewis acidic cation ( Me BDI)Ca + and Roeskysl ow-valent bdiketiminate Al I complex ( Me BDI)Al [41] is able to reduce benzene (V). [42] Mixing ( Me BDI)Al with [( Me BDI)Ca(m-H)] 2 led to C À Ha ctivation of benzene (VI)t og ive ( Me BDI)Al-(H)Ph. [43] Thel atter oxidative addition of ( Me BDI)Al at ab enzene CÀHb ond is catalytic in the calcium hydride reagent. Crimmin and co-workers showed that low-valent ( Me BDI)Mg-Mg( Me BDI) complexes are able to cleave Ar-F bonds by oxidative addition, provided the Ar rest is electronpoor and carries at least four F-substituents. [44] It was attempted to raise the reactivity of the low-valent complex by using mixed-metal complexes with polarized metal-metal bonds (e.g. VII), however,these mixed-metal complexes were found to be less reactive than homometallic ( Me BDI)Mg-Mg( Me BDI). [45] Most recently,the Hill group introduced lowvalent mixed-metal Ae-Al complexes VIII (Ae = Mg, Ca) with polarized Ae d+ -Al dÀ bonds. [26] Although the Ca-Al complex showed ahigher reactivity than the Mg-Al complex, both are stable in benzene.T his contrasts strongly with the high reactivity of the ( Me BDI)Ca + /( Me BDI)Al I combination which instantaneously reduces benzene at room temperature (V). [42] Thecationic nature of this mixed-metal reagent could be the key to its high reactivity.I ncreased reactivity of cationic vs.n eutral complexes was also observed by Okuda and has been attributed to higher metal Lewis acidity imparted by the positive charge on the complex. Thus, cationic Ca hydride complexes showed increased reactivity [46] and activity in alkene hydrogenation catalysis. [47] This motivated our exploration of cationic heterobimetallic complexes with reactive metal-metal bonds.H ere we report as eries of cationic heterobimetallic complexes that have been obtained by combining (BDI)Mg + or (BDI)Zn + cations with lowvalent (BDI)Al I or (BDI)Ga I complexes.W ed emonstrate that metal choice strongly influences structure,b onding and reactivity and report facile cleavage of the non-activated CÀF bond in fluorobenzene by the Zn-Al combination. DFT calculations suggest an ovel dearomatization-rearomatization mechanism for C-F bond cleavage.

Results and Discussion
Forthe cationic fragment we chose the recently reported ( tBu BDI)Mg + or ( tBu BDI)Zn + entities stabilized by al arge tBu BDI ligand with tBu groups in the backbone. [48,49] This bulky ligand prevents contact between the cation and the B(C 6 F 5 ) 4 À anion, leaving the metal free for solvent interaction [50] or molecule activation (Scheme 1). Reaction of their benzene adducts with the low-valent species ( Me BDI)Al or ( Me BDI)Ga [51] in fluorobenzene gave crystals of the heterobimetallic complexes 1-Mg/Al, 1-Mg/Ga and 1-Zn/Ga in good to excellent yields (56-88 %). These complexes can also be obtained by in situ generation of the cationic fragments followed by introduction of the low-valent Al I or Ga I reagent (see ESI).
TheC ÀCand CÀNbond distances in the BDI ligands are not notably different from those in the precursors.
While 1-Zn/Ga is reasonably soluble in C 6 D 5 Br,c omplexes 1-Mg/Al and 1-Mg/Ga could only be analyzed by NMR spectroscopy in the more polar solvent C 6 D 5 F. 1-Mg/Ga shows two sets of broadened 1 HNMR signals which are shifted in respect to signals for the reactants.T he chemical shifts are temperature sensitive and at + 80 8 8Conly sharp signals for the educts are observed ( Figure S16), indicating an asssociationdissociation equilibrium. Complex 1-Mg/Al also shows broad 1 HNMR signals with as trong temperature dependencyb ut even at + 80 8 8Cf ull dissociation cannot be observed (Figure S8), indicating that the Mg-Al bond is stronger than the Mg-Ga bond. Complex 1-Zn/Ga in C 6 D 5 Br (or C 6 D 5 F) gave for each iPr group unique 1 HNMR signals,that is,8methine resonances and 16 methyl resonances ( Figure S17), which is in agreement with the tight embrace of BDI ligands as observed in the crystal structure (Figure 1b). In contrast to similar BDI···BDI interactions in ( Me BDI) 2 Ae complexes, [61] heating as olution of 1-Zn/Ga in C 6 D 5 Br to + 80 8 8Cd id not result in coalescence of 1 HNMR signals ( Figure S25), supporting at ightly bound complex. From these dynamic NMR studies in fluorobenzene it can be deduced that the metal-metal bond strength increases along the series:M g-Ga < Mg-Al < Zn-Ga.
All four heterobimetallic cations,i ncluding the one in hypothetical 1-Zn/Al, have been analyzed by DFT methods (wB97XD/6-311 + G**//wB97XD/6-31 + G**), Atoms-In-Molecules (AIM) and Natural-Bond-Orbital (NBO) analysis (Scheme 2). Thecalculated geometries fit quite well with the crystal structures (Scheme 2b)a part for the metal-metal bonds which are calculated systematically 0.07-0.08 too short. This discrepancy was previously also noted for Mg-Mg complexes. [67,68] Although Zn 2+ and Mg 2+ have similar ionic radii (Table 1), metal-metal bonds to Mg are systematically 0.23-0.24 longer than those to Zn. This is afirst indication that metal bonds to Mg and Zn differ in nature.T he shortest

Angewandte Chemie
Research Articles metal-metal bond of 2.391 is calculated for the cation in 1-Zn/Ga. Since the calculated Zn-Al bond of 2.463 for the cation in 1-Zn/Al is slightly longer, failure to isolate 1-Zn/Al is not related to steric problems.T he considerably shorter bonds to Zn are likely related to ah igher degree of covalency: [49] Zn, Al and Ga have similar electronegativities whereas Mg is significantly less electronegative (Table 1).
Bonding metal-metal interactions in the four heterobimetallic cations are indicated by bond-critical-points (bcps) in the AIM analysis (Scheme 2a). Contour plots of the negative Laplacian, Àr 2 1(r), show the differences in electron distribution along the metal-metal axes.T he lone-pair of electrons on Al I is much more pronounced than that on Ga I .It is strongly polarized towards Mg in 1-Mg/Al but, unlike the more symmetrically arranged electron distribution in aM g-Mg bond, [69][70][71] it should not be described as aN on-Nuclear-Attractor (NNA). Thelow electron density and small positive value for r 2 1 in the bond-critical-point (bcp)indicate aweak electrostatic closed-shell interaction best described as ( Me BDI)Al coordination to the ( tBu BDI)Mg + cation. The considerably higher degree of covalencyi n1-Zn/Al is demonstrated by ah igher electron density 1(r)i nt he bcp and as mall negative value for r 2 1(r), typically found for covalent bonds.A lso the Zn-Ga bond is characterized by higher electron density in the bcp and although r 2 1(r)isnot negative,i ti sc lose to zero.T his is corroborated by the total energy density ratio in the bcp which is defined as H(r)/1(r) and is close to zero for ionic bonds but becomes more negative for covalent bonds.T he calculated values (in a.u.) clearly demonstrate that bonds to Zn are more covalent:Mg-Al À0.170, Mg-Ga À0.136, Zn-Al À0.423 and Zn-Ga À0.402.
TheNPA charges on Mg in the Mg-Al (+ 1.42) and Mg-Ga (+ 1.52) complexes are only slightly lower than that in the free cation ( tBu BDI)Mg + (+ 1.82);Scheme 2c.Incontrast, the NPAc harges on Zn in Zn-Al (+ 0.63) and Zn-Ga (+ 0.74) complexes are considerably lower than that in the free cation ( tBu BDI)Zn + (+ 1.43). Consistently,t he Mg-Al and Mg-Ga complexes show relatively low positive charges on Al (+ 1.12) and Ga (+ 0.95). These charges are close to those in ( Me BDI)Al (+ 0.82) and ( Me BDI)Ga (+ 0.79). On the other hand, the Zn-Al and Zn-Ga complexes show much higher charges on Al (+ 1.57) and Ga (+ 1.39). These charge distributions are consistent with the view that the most electropositive metal Mg forms electrostatic bonds with electron-rich Al I and Ga I "ligands" in which the electron pair is mainly located on the p-block metal. Bonds between the more electronegative Zn and Al or Ga are more covalent in character and there is considerable charge transfer from the p-block metal to Zn. AIM analyses (vide supra) support this view.T he HOMO-LUMO presentations of all cationic heterobimetallic complexes (Figures S61-64) consistently show HOMOsm ainly located on the BDI ligand at Mg or Zn while the HOMOÀ1either has the character of alone-pair at Al or Ga (for Mg-Al and Mg-Ga) or indicates more covalent metal-metal bonding (for Zn-Al and Zn-Ga). The LUMOsa re in all cases mainly concentrated on the BDI ligand at Al or Ga while the LUMO + 1h as metal-metal pbond character.
Thei nstantaneous cleavage of the C-F bond in an unactivated substrate like C 6 H 5 Fb yamixture of [( tBu BDI)Zn + ·C 6 H 6 ][B(C 6 F 5 ) 4 À ]a nd ( Me BDI)Al at room temperature is remarkable.D inuclear Mg I complexes of type (BDI)MgMg(BDI) only cleave activated C-F bonds in polyfluorinated aromatics.F or thermodynamic as well as kinetic reasons,a tl east four F-substituents are needed and the presence of an ortho-F atom has been found highly beneficial. [45] Also ( Me BDI)Al only reacts with activated polyfluorated aromatics (at least three F-substituents are needed). [72] There are very few main group metal systems that are able to cleave the C À Fb ond in C 6 H 5 F. We recently reported C À Fb ond cleavage in C 6 H 5 Fw ith ahighly reactive dinuclear Mg complex with ab ridging C 6 H 6 2À anion but conditions were harsh (5 days,100 8 8C). [67] More recently it was shown that C 6 H 5 Fr eacts at room temperature with photoactivated ( Me BDI)MgMg( Me BDI) in ar adical process resulting in [( Me BDI)MgF] 2 and biphenyl. [73] Hill reported that the highly reactive Ca hydride complex [( Me BDI)Ca(m-H] 2 slowly reacts with C 6 H 5 Ft og ive an intractable mixture of products, [74] while acationic Ca hydride complex from the Okuda group reacted with C 6 H 5 Ft oaC af luoride complex (60 8 8C, 24 h), however, the latter is likely formed by direct nucleophilic substitution and not by oxidative addition. Crimmin and co-workers reported aPd 0 catalyzed oxidative addition of ( Me BDI)Al to Ph-F which is instanteneous at room temperature. [75] Them ildest conditions (À30 8 8C) reported for oxidative addition of Al I to Ph-F need support from Rh I . [76] Inspired by the facile CÀFb ond cleavage in C 6 H 5 F, we probed whether even more electron-rich fluoroarenes could be converted. Thec ombination of [( tBu BDI)Zn + ·(C 6 H 6 )][B-(C 6 F 5 ) 4 À ]a nd ( Me BDI)Al reacted instantaneously when dissolved in p-fluorotoluene,i ndicated by ar apid color change from light-yellow to orange.Extensive NMR investigation of the hexane-soluble fraction, using two-dimensional methods and DOSY,show formation of ( tBu BDI)Zn(p-tolyl), aproduct which was also confirmed by X-ray diffraction ( Figure S57). NMR data for the second product ( Figures S40-51) have similarities with NMR data earlier reported for ( Me BDI)Al-(Ph)F which was formed by Pd-catalyzed oxidative addition of ( Me BDI)Al to Ph-F. [75] Our comprehensive NMR study indicates formation of ( Me BDI)Al(p-tolyl)F but poor crystallization did not allow for confirmation by X-ray diffraction. Thes imultaneous formation of Zn and Al p-tolyl species is likely due to ligand scrambling that is controlled by thermodynamics and product solubilities (vide infra).
Theh erein observed facile reductive C-F bond cleavage by oxidative addition to aZ n/Al combination is due to asynergistic effect:the cationic Zn fragment ( tBu BDI)Zn + and

Angewandte Chemie
Research Articles ( Me BDI)Al alone do not react with C 6 H 5 F, but Zn-Al cooperation cleaves the C À Fbond readily.This high reactivity is likely due to its cationic nature but also to the choice of metals.O ther cationic Mg/Al, Mg/Ga or Zn/Ga complexes, but also the Ca/Al pair (V), [42] do not react with fluorobenzene.I ti sn oteworthy that while the C-F bond in the fluorobenzene solvent is rapidly cleaved by the cationic Zn/Al combination, there is no indication for C-F bond activation in the anion B(C 6 F 5 ) 4 À .This may be related to the bulky tBu BDI ligand which prevents formation of a( tBu BDI)Zn + ···B(C 6 F 5 ) 4 À contact ion-pair but allows for ( tBu BDI)Zn + ·(p-C 6 H 5 F) formation. [49] In contrast to ( tBu BDI)Mg + ,which binds C 6 H 5 Fby Mg···F interaction, [50] the ( tBu BDI)Zn + binds C 6 H 5 Fa sapcomplex that is accessible for nucleophilic attack at the Ph ring.
DFT calculations on am odel system in which DIPPsubstituents were replaced for Ph groups and B(C 6 F 5 ) 4 À was neglected give insights in the possible mechanism for synergistic C-F bond activation (Scheme 2d). Starting with the "naked" L*Zn + cation (L* = HC[C(tBu)NPh] 2 ), PhF has strong preference for p-bonding (C1 vs. C2). This strongly contrasts with ( tBu BDI)Mg + ···F-Ph bonding,w ith ac lear preference for Mg···F interaction, [50] but is in agreement with isolation of a p-complex similar to C2 which was structurally characterized by X-ray diffraction. [49] Previously reported DFT calculations show at otal charge of + 0.15 on the PhF ligand which indicates some extent of charge transfer and partially covalent Zn···Ph bonding (cf. for ( tBu BDI)Mg + ···F-Ph acharge of + 0.02 on PhF was calculated, indicating amerely electrostatic interaction). While the HOMO in L*Zn + ·(p-PhF) is located on the BDI ligand, the LUMO shows major coefficients on Zn and PhF ( Figure S60). The p-complex C2 has only limited space for interaction with LAl I (L = HC[C-(Me)NPh] 2 )a nd forms al oosely bound complex (C3)w ith multiple C À H···p interactions among the BDI ligands (L and L*) and PhF.However, the free coordination site in C1 allows for formation of aZ n-Al bond (C8). Complex C9,w ith an Al···FPh interaction, is slightly more stable.T he transition state for frontal attack is too high for af acile room temperature process (C9!TS3, + 29.5 kcal mol À1 ). Although TS3 is the typical s-bond metathesis transition state for (hetero)bimetallic C-F bond activation, [45] we found an alternative mechanism. Starting from the p-bound complex C2,inwhich PhF is activated for nucleophilic attack by complexation with L*Zn + ,w es earched the potential energy suface for at ransition state with rear-side attack by LAl I according to aS N Ar mechanism. However, all efforts culminated in the identification of TS1,at ransition state which is in line with nucleophilic 1,2-addition to an aromatic C = Cd ouble bond. Despite dearomatization of the Ph ring, indicated by an onplanar ring with localized C=Cb onds,t he barrier of + 14.2 kcal mol À1 is in line with smooth C-F bond activation, as observed experimentally.Analternative minimum for 1,4addition was found (C5). Thelatter is reminiscent of complex V,which previously has been verified experimentally. [42] Such cooperation of electron-poor and electron-rich metals could also be described as bond activation by aF rustrated-Lewis-Pair (FLP). [77] Theherein proposed cooperation between the Zn and Al metal centers has been verified by calculating the mechanism for the direct oxidative addition of LAl I to the Ph-Fb ond ( Figure S59). In agreement with previous calculations, [78,79] the high activation enthalpy of + 27.8 kcal mol À1 for this reaction is nearly double that calculated for the conversion: C3!TS1!C4.N ote that the C3!C4 conversion is endothermic (due to loss of aromaticity) and that the CÀF bond in C4 (or C5)i ss till intact. Subsequent cleavage of the strong C-F bond by elimination of LAl-F + from C4 (or C5)is essentially without barrier (TS2)a nd exothermic (C4!C6, À38.3 kcal mol À1 ). Thel atter step is strongly facilitated by rearomatization of the p-system which supplies the energy needed for C-F bond cleavage.Nature uses similar dearomatization/rearomatization protocols for challenging transformations (e.g.N ADH/NAD + ). Thes ame principle is also increasingly applied in the development of contemporary catalysts with ligand-metal cooperation (e.g.o ft ype Noyori, Shvo or Huang). [80] Theenormous Lewis acidity of LAl-F + is demonstrated by its ability to substract the Ph group from the product L*ZnPh, aprocess which in the absence of further Fsources is highly exothermic (C6!C7, À52.1 kcal mol À1 ). Complex C7 could further eliminate LAl(F)Ph to give L*Zn + (C10), the starting point of the energy profile.This shows that the oxidative addition of LAl I to the Ph-F bond is overall an exothermic process and suggests that the reaction may be catalytic in L*Zn + .A ll attempts to run this C-F bond activation in ac atalytic protocol failed. This is likely due to ligand scrambling equilibria that are controlled by thermodynamics and product solubilities.I nt he present case we observed L*ZnPh as the major product, explaining why the reaction is not catalytic in Zn.
As the herein presented energy profile does not include the influence of the weakly coordinating B(C 6 F 5 ) 4 À anion, it should be treated with care.According to our calculations,the complexes C8 and C9 should be quite stable and could be at hermodynamic sink, impeding the C-F bond activation. However,e xperimentally there is no indication for Zn-Al bond formation:asolution of [( tBu BDI)Zn + ][B(C 6 F 5 ) 4 À ]a nd ( Me BDI)Al in PhF gave at low temperature crystallization of [( tBu BDI)Zn + ·(p-C 6 H 5 F)][B(C 6 F 5 ) 4 À ]w hich is the starting point for the low energy route C3!TS1!C4.
These data suggest that the facile C-F bond activation in fluorobenzene does not proceed through apreviously formed cationic heterobimetallic complex but is the result of cooperating L*Zn + and LAl I species.F ailure to locate aS N Ar pathway is likely due to unfavorable formation of L*ZnÀF, asoft-hard combination and the unusual stability of the p-PhF complex C2 vs.aL*Zn···FPh complex (C1)w hich would be the first step in aS N Ar mechanism. Theh ere presented pathway results in LAl À F + ,ah ard-hard combination, and L*ZnPh, which is in accordance with experimental observation.

Conclusion
Combining the electron-rich, low-valent b-diketiminate complexes ( Me BDI)Al or ( Me BDI)Ga with "naked" (Lewis base-free) cations like ( tBu BDI)Mg + or ( tBu BDI)Zn + gave heterobimetallic cations with MgÀAl, MgÀGa or ZnÀGa bonds.T he Zn-Al combination could not be obtained due to fast decomposition by reaction with fluorobenzene.C rystal structures of the borate salts of [( tBu BDI)Mg-Al( Me BDI) + ], [( tBu BDI)Mg-Ga( Me BDI) + ]a nd [( tBu BDI)Zn-Ga( Me BDI) + ] reveal that B(C 6 F 5 ) 4 À is at ruly non-coordinating anion. Although Mg 2+ and Zn 2+ have comparable ionic radii, metal-metal bonds to Mg are considerably longer and weaker than those to Zn. This is supported by DFT calculations and bond analyses by AIM. Based on NPAc harge calculations, the metal-metal bonds in the Mg complexes should be considered as Al I (or Ga I )!Mg 2+ donor bonds while bonding in the Zn complexes is more covalent. Dynamic NMR studies in fluorobenzene indicate that the metal-metal bond strength increases along the series:Mg-Ga < Mg-Al < Zn-Ga. This is conform Hard-Soft-Acid-Base (HSAB) theory according to which hard metal (Mg and Al) and soft metal (Zn and Ga) combinations form the strongest bonds.
Thecationic Zn-Al complex, which according to HSAB is as oft-hard mismatch, could not be obtained. While the metal-metal bound complex is not formed at lower temperatures,atroom temperature fast cleavage of the CÀFbond in the fluorobenzene solvent was observed. Thep roducts ( tBu BDI)ZnPh and [( Me BDI)Al(F)-(m-F)-(F)Al( Me BDI) + ][B-(C 6 F 5 ) 4 À ]h ave been identified. This is ar are example of transition metal-free cleavage of an unactivated CÀFb ond. Interestingly,t he generally much more reactive polyfluorinated rings in the borate anion are left intact. Since the sterically congested ( tBu BDI)Zn + cation does not interact with B(C 6 F 5 ) 4 À but prefers formation of the solvent adduct ( tBu BDI)Zn + ·(p-PhF), this complex with a p-bound fluorobenzene ligand could be the key to C-F bond activation. The fluorobenzene ligand in the latter complex is bound to Zn with its p-system and is activated for nucleophilic attack (calculated NPAc harge on PhF: + 0.15). Indeed, DFT calculations suggest aprocess in which ( Me BDI)Al reacts with ( tBu BDI)Zn + ·(p-PhF) by rear-side 1,2-addition to an aromatic C=Cb ond which is followed by ( Me BDI)AlF + elimination. Thea ctivation energy for this reaction is,d espite loss of aromaticity,l ow: + 14.2 kcal mol À1 .T he barriers for ad irect, concerted C-F bond cleavage by either the (BDI)Al complex (+ 27.8 kcal mol À1 )orbythe heterobimetallic Al-Zn complex (+ 29.5 kcal mol À1 )a re both considerably higher.
Them ost important conclusion of this work is that the right combination of electron-rich and electron-poor metal centers can cooperate to rapidly cleave the strong, unactivated, C-F bonds in fluorobenzene and fluorotoluene at room temperature.T he mechanism for this FLP-type process is proposed to proceed through an unusual intermediate in which the phenyl ring is first dearomatized. Subsequent rearomatization delivers the energy needed for C-F bond cleavage.