Cobaltacycles May Collapse Upon Oxidation V

Home , Organomercury

On the Fate of Cobaltacycles in Cp*Co-mediated C-H bond Functionalization Catalysis: Cobaltacycles May Collapse upon Oxidation via Co(IV) species Fule Wu, Christophe Deraedt, Yann Cornaton, Laurent Ruhlmann, Lydia Karmazin, Corinne Bailly, Nathalie Kyritsakas, Nolwenn Le Breton, Sylvie Choua, Jean-Pierre Djukic

To cite this version:

Fule Wu, Christophe Deraedt, Yann Cornaton, Laurent Ruhlmann, Lydia Karmazin, et al.. On the Fate of Cobaltacycles in Cp*Co-mediated C-H bond Functionalization Catalysis: Cobaltacycles May Collapse upon Oxidation via Co(IV) species. 2021. hal-03292255

HAL Id: hal-03292255 https://hal.archives-ouvertes.fr/hal-03292255 Preprint submitted on 20 Jul 2021

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. On the Fate of Cobaltacycles in Cp*Co-mediated C-H bond Func- tionalization Catalysis: Cobaltacycles May Collapse upon Oxidation via Co(IV) species. Fule Wu, a Christophe Deraedt, a Yann Cornaton,a Laurent Ruhlmann, b Lydia Karmazin, c Corinne Bailly,c Nathalie Kyritsakas,c Nolwenn Le Breton,d Sylvie Chouad and Jean-Pierre Djukica,* a Laboratoire de Chimie et Systémique Organométalliques, Institut de Chimie de Strasbourg (UMR 7177) CNRS/Université de Strasbourg, 4 rue Blaise Pascal, F-67000 Strasbourg, France b Laboratoire d’Electrochimie et Chimie Physique du Corps Solide, Institut de Chimie de Strasbourg (UMR 7177) CNRS/Université de Strasbourg, 4 rue Blaise Pascal, F-67000 Strasbourg, France c Service de Radiocristallographie Fédération de Chimie Le Bel – FR2010 BP 296R8, 1 rue Blaise Pascal, F-67008 Stras- bourg Cedex, France d Laboratoire Propriétés Optiques et Magnétiques des Architectures Moléculaires Institut de Chimie de Strasbourg (UMR 7177) CNRS/Université de Strasbourg, 4 rue Blaise Pascal, F-67000 Strasbourg, France

ABSTRACT: Recent reports have identified Cp*Co-based complexes as powerful catalysts for aromatic C-H bond activation under oxidative conditions. However, little is known about the speciation of Cp*Co species during catalysis. We now show that key intermediates, Cp*Co(III) metallacycles derived from 2-phenylpyridine (phpy-H), react swiftly in solution with one-electron oxidants to irreversibly collapse by a cyclocondensation of the organic ligands to afford cationic alkaloids in yields of >70 %. Low temperature EPR analysis of a mixture of cobaltacycle with the tritylium cation reveals the signatures of trityl and Co(IV)-centred radicals. Electrochemical analyses show that the oxidation of these cobaltacycles is irreversible and gives rise to several products in various amounts, among which the most salient ones are a cationic alkaloid resulting from the cyclocondensation of the phpy and + Cp* ligands, and the dimeric cation {[Cp*Co]2(-I)3} . DFT investigations of relevant noncovalent interactions using QTAIM- based NCI plots and Intrinsic Bond Strength Index suggest a ligand-dependent predisposition by “NCI-coding” for the Co(IV)- templated cyclocondensation, the computed reaction network energy profile for which supports the key roles of a short lived Co(IV) metallacycle and of a range of triplet state organocobalt intermediates.

INTRODUCTION iodo-Ir(III) complexes are left untouched. In the majority of 5 1 reports dealing with group 9 metallacycles bearing an hetero- The -pentamethylcyclopentadienyl ligand - or Cp* - in spite - of an abundant literature2 is often considered as a "robust", chelating carbanionic ligand {N,C} with Cp* bound metals, sterically and electronically stabilizing1,3 ligand in its use in the latter ligand never interferes explicitly with ligands bound catalysts.4 If in specific cases the endo migration of a hydrido to the metal by other means but steric repulsion and synergetic ligand to Cp* was reported in the case of a bis-pyridine che- “through bonds” electronic effects. late of Rh(III),5 literature is rather silent on occurrences of PhCO3H - or explicit reactions of the Cp* ligand with {N,C} heteroche- NiCp Ar-Cl 6 N N lating ligands in metallacycles. This issue has been over- N N (1) o-xylene, reflux looked in most reports dealing with the use of Cp*Co-based Ustinyuk et al. 1969 & 1970 catalysts in aromatic C-H bond functionalization employing redox co-catalysts where cobaltacycles are central intermedi- Cp* oxidation ? 4,7 N N 2 ates. One can find a rare report on the metal-templated "H" 1 (2) Co 3 collapse of a metallacycle of a CpNi (II) complex reported by room temp. 5 Wu et al. 2020 I 4 Ustinyuk et al.8 where an azo-benzene-based nickelacycle H , I3 undergoes interligand condensation and metal decoordination 1a rac-[2a]+ under oxidative conditions, affording a new organic extended In a recent report6 we disclosed without further reporting on its - aromatic product (1). It is probably because of the scarcity of origin the crystal structure of the I3 salt of an unprecedented similar reports that Cp* bound metallacycles of group 9 metals cationic Cp*-2-phenylpyridine condensation product, i.e [2a]+ at the +I and +III formal oxidation states are frequently con- (2), serendipitously discovered whilst attempting to grow 9 7c,10 sidered as structurally viable candidate catalysts. A report crystals of 1a. It was obvious that [2a]+ resulted from the 11 by Mayer et al. outlined that the two-electron oxidation of collapse of cobaltacycle 1a by the cyclocondensation of the Cp*Ir metallacycles by PhI(OAc)2 only affects the Cp* ligand Cp* and 2-phenylenepyridine (abbr. phpy) ligands with the in the case of a methyl-Ir(III) complex, whilst other chloro and capture of an exogenous H atom (2). We therefore undertook the investigation of a series of Co(III) metallacycles derived identified readily from their specific signature: 2-(2’- from 2-phenylpyridine (abbr phpy-H), because such a demetal- iodophenyl)pyridine, tris--iodo dimeric Cp*cobalt (III) cation + + lative cyclocondensation reaction, if relevant in its extent, {[Cp*Co]2(-I)3} and compound [2a] were formed in 40%, 4 could limit further developments of Cp*Co(III) catalysts at a 66% and 33% yield according to 1H NMR analysis (3). Com- larger scale where recycling of metal catalysts might be cru- pound {[Cp*Co]2(-I)3}[I3] produces one signal for the Cp* cial for environmental reasons.12 1 ligand at  1.80 ppm in CDCl3. For [2a][I3] the H NMR spec- " " trum depicts the expected magnetic inequivalence of the five

N Cp* -1e N Cp* N methyl groups of the cyclopentenyl fragment that show up at Co Co S 0.96 (d), 1.31 (s), 1.56 (s), 1.75 (s) and 2.33(s) ppm. Proton SCF3 (IV) SCF3 CF3 H5 (2) appears at  2.49 ppm as a broad multiplet integrating for 1 proton. Additional electrospray-mass spectroscopy Perez-Temprano et al. 2021 (abbr. ES-MS) analysis of the raw reaction mixture confirmed Scheme 1. The oxidation of the singlet state Co(III)- the presence of the latter three compounds alongside a strong thiolato complex leading to C-S bond formation. signal of a [Cp*CoI]+ moiety at m/z 321, stemming from the + dimer {[Cp*Co]2(-I)3} showing up with lower intensity at Interestingly, it has been recently proposed that transient m/z 769. Worthy to note, compound {[Cp*Co]2(-I)3}[I3] was 13 Co(IV) species may play a central role in oxidative C-H successfully isolated and subsequently recrystallized to afford 14 bond thiolation. Indeed, in an effort to clarify the mecha- stable dark green crystals suitable for X-ray diffraction analy- 14 nism of the catalysis reported by Glorius et al., Perez- sis. The structure of {[Cp*Co] (-I) }[I ] shows three sym- 13 2 3 3 Temprano et al. showed in a report that appeared while the metrically bridging iodo ligands between the two Co(III) cen- present manuscript was being finalized that the treatment of a tres that are distant from each other by 3.177(5) Å (Figure 1). trifluoromethylthiolato complex of a 2-phenylpyridine Cp*- This first test reaction clearly revealed that at least two path- bound cobaltacycle with excess Ag(I) salts may lead to the ways are involved in the observed oxidative decomposition of oxidation of the Co(III) centre into a Co(IV) complex purport- 1a: 1) the reductive elimination of 2-(2’-iodophenyl)pyridine edly responsible for the formation of the expected ortho- and 2) Cp*-phpy cyclocondensation, both triggered by the substituted product of C-S coupling (Scheme 1) in various formal oxidation of the Co(III) centre to either the open shell yields (from ca. 99% to 0%) depending on the amount and Co(IV) or the diamagnetic Co(V) state. The major question nature of the oxidant. In this recent report13 the fate of the remaining to be answered being that of the origin of the “H ” remaining amounts of cobaltacycle was not addressed by the 5 atom in [2a]+ (2). authors though.

In the present report we disclose our conclusions on the ob- N I 15 2 eq I2 1a [2a][I ] and and *CpCo CoCp* (3) served ligand cyclocondensation in cobaltacycles of 2- 3 I I DCM, room temp. I , [I3] phenylpyridine derivatives and its direct cause, i.e the irre- overnight versible oxidation of Co(III) centre into the Co(IV) state, 33% (NMR) 40% (NMR) 66% (NMR) which sheds a new light on the possible limitations of Cp*Co(III) based aromatic C-H functionalization catalysts when placed under oxidative conditions. The recently introduced quantum theory of atom in molecule (QTAIM16)-based ginter interaction score17and IBSI index18 are used here to reveal the effect of the oxidation of Co(III) cobaltacycles on the attractive noncovalent Cp*-phpy interactions and their possible driving role in this unexpected cyclocondensative collapse of cobaltacycles.

RESULTS AND DISCUSSION

Synthetic outlook. In our original report that briefly de- 6 scribed compound [2a][I3] (2) it was stated that it crystallized from a solution that originally contained 1a. The presence of - I3 as counter anion suggested that the initial solution of 1a might have been contaminated by adventitious amounts of Figure 1. ORTEP-type drawings of the structures of 1a, 1d, - either I2 or some I3 salts introduced or formed in the course of 1e, {[Cp*Co] (-I) }[I ], and 1h, at 50% probability with 6 2 3 3 the synthesis of [Cp*CoI2]2, i.e the precursor of 1a. In the partial atom numbering. Molecules of solvent were omitted present study, suspecting an oxidoreductive process responsi- for clarity. 1d,e were synthesized by the lithium acetamidate- + ble for the formation of the condensation product [2a] , 1a assisted cyclocobaltation method.6 was hence reacted with two equiv. of I in dichloromethane 2 Given the intricate mixture produced by the reaction of 1a (abbr. DCM) for 16 h at 20°C. The 1H NMR analysis of the with I , as an alternative to the latter, we decided to probe the reaction mixture revealed the full conversion of 1a into a 2 oxidizing property of the one-electron oxidant mixture of various products among which a few could be 19 + [Ph3C][BArF24]. The tritylium [Ph3C] , which undergoes a 2 swift decomposition20 in the gas phase, is in solution a well- Noteworthy, adding 1 equiv. of (2,2,6,6-tetramethylpiperidin- known stable multi-facetted Lewis acidic cation.21 It may act 1-yl)oxy (TEMPO) to the reaction mixture resulted in a sig- as a hydride22 and/or a halide23 abstraction agent but also as a nificant inhibition (Table 1, entries 2 and 3) of the reaction 24 + 25 one-electron oxidant (Ph3C / Ph3C couple ) in the ground perceptible from an NMR yield decrease to 35% after 17 h, as well as in the excited state26 and may be involved also in suggesting that open shell species and radicals essential to the the hydride abstraction process as reported by Mandon, Astruc formation of [2a]+ are partly trapped by the rather bulky ni- et al.27 Quite interestingly, to the best of our knowledge there troxyl radical. Tritylium-triggered condensation reactions has been no mention of any major drawbacks related to the were repeated with other Cp*-bound cobaltacycles, i.e 1b-e, oxidative properties of the tritylium as “activator”, i.e as a which were all reactive towards [Ph3C][BArF24], although halido ligand “remover”, in transition metal-mediated homo- showing distinct outcomes depending on the presence (1b, geneous catalysis. For instance hydrosilylation catalysis oth- 1c)6 or absence (1d, 1e) of methyl hydrogens substractable by 28 + erwise ineffective with Cp*-bound chloridoiridacycles can [Ph3C] (Table 1). The data listed in Table 1 were all acquired be triggered if a [Ph3C][BArF24] salt is added to the reaction under standard conditions consisting of a reaction time of 17 h mixture seemingly acting as a chloride scavenger.29 at room temperature. Table 1. Reaction of Co(III) cobaltacycles with 1b, 1c untraceable products [Ph3C][BArF24] in various halogenated solvents and conditions R' (atmosphere: dry argon or ambient air). Cp* N [Ph3C][BArF24] Co R= H, [2a] Entry Complex R, Solvent Product(s) (yield) DCM or DCE N R= Br, [2d] I room temp. R’ (atm.) R= CF3, [2e] 1a, 1d, 1e R H 1 1a H,H DCM (Ar) [2a][BArF24] a b R= H, R'=H, 1a , [BArF24] (29% , 72% ) R= NMe2, R'= tBu, 1b R R= Me, R'= H, 1c and "CoI" and {[Cp*Co]2(-I)3}[BArF24] 2 1a H,H DCM (Ar) + [2a][BArF24] R= Br, R'= H, 1d c b R= CF3, R'= H, 1e TEMPO (35% ) CDCl3  H2O/air F3C 3 1a H,H DCM (Ar) + [2a][BArF24] TEMPOd (17%b) [BArF24] = B N R= H, [3a] R= Br, [3d] 4 1a H,H DCM and [2a][BArF24] F3C R= CF , [3e] 4 OH 3 b CDCl3 (am- (44% ), , [BArF24] bient air) [3a][BArF24] R b Scheme 2. The Co-templated oxidative cyclocondensation (44% ) of 1a, 1d and 1e. 5 1a H,H DCE (Ar) [2a][BArF24] (40%b) The reaction between 1a and [Ph C][BArF ] (Scheme 2) 3 24 6 1a H,H d .PhCl (Ar) [2a][BArF ] carried out at 20°C in DCM for 16 h produced [2a][BArF ] 5 24 24 (51%b) with an NMR yield of 72% (Table 1, entry 1). In fact 1H NMR spectroscopy monitoring showed that the reaction 7 1d Br, DCM (Ar) [2d][BArF24] b reached 100% conversion after only 30 min. Also noteworthy, H (57% ) the same experiment carried out with the chloridoiridacycle 8 1d Br, DCM (ambi- [2d][BArF24] analogue left the substrate unchanged. The same reaction H ent air) (37%a, 84%b) carried out using 2 equiv. of AgPF6 as oxidant instead of 9 1d Br, DCM (Ar) [3d][BArF24] [Ph3C][BArF24] was completed with 100% conversion of 1a a H and CDCl3 (21% ) into a mixture of [2a][PF6] (70% NMR yield, not isolated) and (air) other aromatic products. Despite the fact that Ag+ was also 10 1e CF3, DCM (Ar) [2e][BArF24] efficient in this reaction, the significant content in side organic b products and the overall higher heterogeneity of the reaction H (63% ) mixture led us to privilege [Ph3C][BArF24] as a standard oxi- 11 1e CF3, DCM (ambi- [2e][BArF24] a b dant for further physical investigations. H ent air) (35% , 88% )

Compound [2a][BArF24] was isolated pure by repeated recrys- 12 1e CF3, DCM (Ar) [3e][BArF24] a tallization with a major loss of material due to the difficulty H and CDCl3 (17% ) that arose in the separation of this salt from other polar Co- (air) containing residues (isolated yield: 29 %). The structure of the 13 1g H, H DCM (Ar) [2a][BArF24] b cation of [2a][BArF24] obtained by X-ray diffraction analysis (21% ) is in all aspects identical (Figure 2) to that already published6 a isolated yield. b 1H NMR yield in the presence of 1,3,5- for [2a][I3]. Alongside this organic salt, amounts of Ph3C-I c + trimethoxybenzene as an internal standard. 1 equiv and {[Cp*Co] (-I) } were also identified by ES-MS analy- d 2 3 TEMPO/Co, aliquot taken at 17h of reaction time. 4 equiv sis. More interesting was the formation of small amounts of TEMPO/Co, aliquot taken after 17 h of reaction. N-protio,2-phenylpyridinium indicating that even with

[Ph3C][BArF24] the hydrodechelation of the phpy ligand jointly with the release of {[Cp*Co]2(-I)3}[BArF24] is a com- Only the substrates bearing –Br and –CF3 substituents (Figure peting process. 1), i.e 1d and 1e (Table 1, entry 7 and 10), afforded cyclic 3 pyridiniums, i.e [2d]+ and [2e]+, the 1H NMR signature of Compounds [3a,d,e]+ were present alongside variable amounts which could be recognized readily in the spectrum of the raw of [2a]+, [2d]+ and [2e]+ (Table 1). The structures of [3d]+ and mixture from the specific pattern of methyl groups showing up [3e]+ were inferred from X-ray diffraction analysis (Figure 2) in the  0.9-2.5 ppm range. The reaction with compounds 1b and indicate that they result from the hydroxylation of the 31 and 1c yielded untraceable mixtures of soluble and insoluble same C5-H position. The interference of Co species released residues with no distinguishable main product. The structures by the cyclocondensation process with the unreacted tritylium + of [2d,e] were both established by X-ray diffraction analysis, cation capable to interact with O2 and/or water to form possi- which revealed features akin to those of [2a]+ (Figure 2). ble tritylperoxo species32 can be suspected in this hydroxylation. Particularly informative is the reaction of the chlorido analogue of 1a, i.e 1g, with [Ph3C][BArF24], which afforded Cp* [2a][BArF ] in 21 % yield (NMR yield), at room temperature N N 24 [Ph3C][BArF24] H Co (4) after about 17 h of reaction in DCM (scheme 3, Table 1 entry DCM or DCE 13). I room temp. , [BArF24] 1f The case of the Cp-containing complex 1h (Figure 1) is par- 6 ticularly striking (Scheme 4). The reaction of 1h with Cobaltacycle 1f reacted with [Ph3C][BArF24] and fully con- verted to afford the product of hydrodemetallation, that is [Ph3C][BArF24] carried out in DCM at room temperature namely the N-protonated benzo[h]quinoline,30 the structure of yielded a mixture of mostly untraceable products among + which was ascertained by X-ray diffraction analysis (cf. Sup- which condensation product [2h] (Figure 3) was isolated in + porting Material). ca. 12% yield and crystallized. Compound [2h] differs from the previous condensation products by the cationic spiro motif Quite intriguing is the outcome of the reactions of 1a, 1d and resulting formally from the condensation of the phpy ligand to 1e with [Ph C][BArF ] when adventitious moisture and air 3 24 the same carbon atom at the Cp ligand combined with a proto- contamination made its way (Table 1, entries 4, 8, 9, 11, 12) to tropy. The incorporation of two 2-phenylenepyridine units in a new solution in CDCl prepared by dissolving the raw reac- 3 [2h]+ obviously entails two distinct processes of intra and tion residue obtained upon DCM evaporation after an over- intermolecular nature that, for obvious structural reasons, night reaction: compounds [3d][BArF ], and [3e][BArF ] 24 24 cannot occur with the Cp* complex 1a. were readily isolated as they showed much lower solubility in

CDCl3 than their respective precursors (scheme 2). In con- Cp N N N N [CpCoI2]2 [Ph3C][BArF24] H trast, [3a][BArF24], which showed solubility in chloroform Co DCM or DCE MeC(O)NHLi H similar to that of [2a][BArF24] could not be isolated pure. DCM 50°C I room temp. H 40% , [BArF24] [nBu N]Cl 12% (isolated yield) 4 1h [2h] MeCN, [Ph3C][BArF24] room temp. Cp* N [D2].DCM Scheme 4. Synthesis of 1h and its conversion to overnight room temp. 1a [2h][BArF ] upon treatment with [Ph C][BArF ]. Co [2a][BArF24] 24 3 24 79% Cl 21% (NMR) 1g Scheme 3. Synthesis of 1g and its reaction with [Ph3C][BArF24].

Figure 3. ORTEP-type drawing of the structure of [2h]+ at - 50% probability with partial atom numbering. The [BArF24] anion and solvent were omitted for clarity.

Thermochemistry of the reaction of 1a with [Ph3C][BArF24]. The thermochemistry of the reaction of 1a with [Ph3C][BArF24] was addressed by isotherm titration calorimetry at +25.00°C in 1,2-dichloroethane (DCE), placing the cobalt complex in the 1 mL cell and the tritylium salt in the Figure 2. ORTEP-type drawings of the structures of [2a]+, servo-controlled injection syringe. Quite interestingly the heat [2d]+, [2e]+, [3d]+, [3e]+, at 50% probability with partial atom flow decreased steadily down to a marked inflexion point at a numbering. The [BArF24] anion and lattice solvent were omit- [Ph3C][BArF24] /1a ratio n of ~1:1 (Figure 4). The enthalpy of ted for clarity. reaction was determined from the raw values of heat flow corrected by subtraction of the residual heat peaks at n>1.5:1 4 as H= 13.7 ± 0.5 kcal/mol (from three independent titra- no deuterium incorporation tions), which is the total exothermicity of all the chemical N processes that take place under those conditions up to the heat ~70% inflexion point before athermicity, that is namely: the oxida- H ,[BArF ] tion of 1a and the cyclocondensation reaction, dechelation by [2a]+ 24 [Ph3C][BArF24]. protonation, possible radical captures and recombination in CD2Cl2 N argon atmosphere N [d15.Ph3C][BArF24] which it is expected that the solvent or any H-atom donor D2O CD2Cl2 CD2Cl2 H 1a species might be a player. air argon atmosphere H ,[BArF ] or [2a]+ 24 + ,[BArF24] argon CF3SO3H [2a] A control reaction carried out under identical conditions was atmosphere 80% under air CH2Cl2 1 10% under argon argon atmosphere 64% checked by H NMR spectroscopy and showed that a yield of no deuterium incorporation no deuterium incorporation 40% in [2a]+ could be achieved (Table 1, entry 5), suggesting N that about 60% of the amount of tritylium cation might be H diverted to an oxidative chemical pathway either disconnected quantitative + from or instrumental to the formation of [2a] . Scheme 5. The use of CD2Cl2 and/or [d15.Ph3C][BArF24] 2 does not result in the incorporation of H at the C5 posi-

-33 tion in [2a][BArF24]. Treatment with triflic acid results in hydrodemetallation. Treatment with D2O promotes -34 the formation of [2a][BArF24] under an atmosphere of 2 -35 air without incorporation of H.

-36 A second experiment, ran in CD2Cl2, using -37 33 [d15.Ph3C][BArF24] afforded [2a][BArF24] in 64% yield with no incorporation of deuterium at the C5 position. The latter RawHeat Rate (µcal s) / -38 result suggests that the most probable source of H atom is -39 either the BArF24 anion, 1a or any of the products of its degra- dation such as 2-phenylpyridine, {[Cp*Co]2(-I)3}[BArF24]. -40 - To eventually rule out [BArF24] anion as a potential source of 33 -41 0 10000 20000 30000 40000 50000 60000 70000 80000 H atom, the reaction of 1a with [Ph3C][PF6] was undertaken, Time (seconds) which led to the formation of [2a]+ in yields similar to those Figure 4. Thermographic trace of the isotherm calorimetric titration of a obtained with the tritylium BArF24 salt. Whether the H atom solution of 1a (0.94 mM) by sequential addition of 2 L of a solution of is captured directly from 1a by an oxidized form of 1a or from [Ph3C][BArF24] (19.6 mM) in DCE at 25°C in an argon-filled glovebox. an organic source such as Ph3CH, was not elucidated experi- Individual heat flow peaks correspond to the response of the cell contain- mentally. Triphenylmethane could indeed arise from an hy- ing 1a after the addition of portions of a solution of [Ph3C][BArF24]. The + th dride abstraction from the Cp* ligand of 1a by [Ph3C] if one inflexion point considered at the 29 peak (at ca. 55000 s) corresponds to refers to the work of Tilley et al.34 that was originally carried a n=1:1 stoichiometry. out with [Ph3C][Ph4B]. However, the presence of Ph3CH was not overwhelming in all analysed reaction media. Further- Attempts to Trace the H-atom source down. more, like shown below in the theoretical investigation, + Ph CH is a very unlikely source of H radical for thermochemi- To trace the origin of the H atom incorporated in [2a,d,e] two 3 2 cal reasons. sets of experiments were carried out using H-labeled molecules, expecting that a major isotopic effect could allow the It is well established that the trityl radical recombines to give a identification of the main source of H atom in the reaction of polyaromatic species, i.e the so-called the Gomberg-Jacobson- 32 1a with [Ph3C][BArF24] (Scheme 5). The first set consisted of Lankamp-Nauta-MacLean asymmetric quinoid-like structure two parallel reactions in anhydrous deuterated and non- (cf. Scheme 6) that is more favorable than Ph3C-CPh3, which deuterated solvents, namely CD Cl and CH Cl , under identi- displays a weak but reactive C-C bond. However, the absence 2 2 2 2 + cal conditions. 1H NMR analysis of the raw reaction mixtures of deuterium incorporation into [2a] in the experiment carried + 33 indicated that the yields in [2a] were the same in both cases out with [d15.Ph3C][BArF24] seemingly rules out this latter with no incorporation of deuterium at the key position C5 scenario too. (Scheme 5). Worthy of note, the treatment of 1a with 1 equiv. of CF3SO3H in CH2Cl2 (Scheme 5) at room temperature swiftly led to the hydro-de-cobaltation of the phpy ligand with the formation of N-protio,2-phenylpyridinium, suggesting that any formation of + + H in the reaction of 1a with [Ph3C] should result in a significant hydrodemetallation of the phpy ligand, which is not the case. Treatment with 1 equiv. of a weaker acid, e.g. pure AcOH, under identical conditions resulted in the hydrodemetallation of 1a and the formation of 2-phenylpyridine, in lower yields though: 14% after 5 h, 34% after 24 h. Also noteworthy is the effect on the reaction of 1a with [Ph3C][BArF24] in the presence of 5 eq of H2O or D2O that 5 was probed under atmospheres of either argon or air (Scheme anhydrous 1,2-dichloroethane using [N(nBu)4][PF6] (0.1 mol 1 -1 5): disregarding the atmosphere, with D2O H NMR analysis L ) as conducting electrolyte (Figure 6, Figures S13-S21 in of the reaction mixture revealed no incorporation of deuterium the Supporting Material and Table 2). The cyclic voltammo- + - at the C5 position of [2a][BArF24]. The formation of the latter gram of {[Cp*Co]2(-I)3 }[I3 ] (Figure 6 black curve) shows in the presence of water occurs with a lower yield when the three successive oxidation processes at -0.02 V, 0.36 V and reaction was run under argon (~10%) as compared to the one 1.25 V vs. Ag/AgCl. The last oxidation at 1.25 V is irreversi- ran under air (~80 %)(cf. Supporting Material). This result ble and may correspond to the oxidation of the cobalt centre, suggests that the combination of water contamination with while the two first oxidations are similar to the redox behav- oxidizing conditions (1 electron oxidant + air) favours the iour of tetraethylammonium iodide (abbr. TEAI) (Supporting + + formation of [2a] . Noteworthy, [Cp*Co]2(-I)3 , i.e the main Material, Figure S15). It corresponds to the redox properties - side product of the collapse of 1a, fully decomposes in the of the triodide I3 counter-anion. In fact, at the potential ap- - - reaction carried out in the presence of water under air (cf. plied at the start of the sweep I3 is first reduced to I , and Supporting Material). redox behaviour similar to that of TEAI is observed. Indeed, cyclic voltammogram of 0.4 mmol.L-1 solution of TEAI re- The body of evidence gathered here points at 1a or some de- -1 corded in 0.1 mol L [N(nBu)4][PF6]/DCE at a platinum elec- graded from of it as a probable source of the H atom that in- -1 corporates into [2a]+. A contribution of the solvent cannot be trode and scan rate of 0.1 V s (Figure 6, blue curve and Fig- completely excluded neither at this stage as a major kinetic ure S10) exhibited also two anodic oxidative peaks at Epa1 (a’) barrier to deuterium incorporation such as a transfer of H by = -0.08 V vs. Ag/AgCl and at Epa2(c’)= 0.40 V vs. Ag/AgCl 35 quantum tunneling (not addressed here) may well lead to the coupled with an cathodic peaks at Epc1(b’) = -0.28 V vs. consumption of the residual protio-dichloromethane in the Ag/AgCl and Epc2(d’) = 0.35 V. The presence of two coupled experiments ran in this deuterated solvent. peaks is well documented in the literature. It indicates the presence of two chemical processes following the two steps of 39 g = 5.2 g = 2.02 the quasi-reversible electron transfer (E1C1E2C2 mechanism). According to the literature,40 the peak a’ (Figure 6a) is attributed to the E1C1 mechanism: first the oxidation of two iodides

to I2 (electrochemical step E1, two electron oxidation wave) - - followed by the chemical reaction C1 with I giving triodide I3 (chemical step C ). The peak b’ (Figure 6a) is attributed to the

1 - - reduction of I3 to regenerate I . The second anodic peak (c’, - Figure 6a) yielded from I3 is a one electron oxidation wave - (electrochemical step E2) attributed to the oxidation of I3 obtained after the E1C1 process to form I2, while peak d’ (Fig- - ure 6a) is attributed to the reduction of I2 to I3 . 0 100 200 300 400 B (mT) Figure 5 X-band EPR spectrum recorded at 4.3 K after 5 min of reaction between 40 mM solutions of 1a and [Ph3C][BArF24] in DCM . The spectrum was recorded at a frequency of 9.345 GHz.

EPR and electrochemical investigations. The mixture produced by the reaction of a 40 mM solution of 1a and [Ph3C][BArF24] in DCM was analysed by EPR spectroscopy at low temperature in the frozen state. The EPR spectrum (Figure 5) recorded at 4 K after 5 minutes of reaction showed the superposition of the signatures of two species.

One intense peak is centered at g2.003 (Hpp=1.25 mT) and supports the presence of an organic radical36 in frozen solution. The second signal of lower intensity displays a distorted axial pattern with broad resonances and effective g values at g 5.2 and g 2.02 similar to that already reported by Wang et al.37 These main peaks point to high-spin Co(IV) species37-38 (S= 3/2 or S= 5/2) as a result of the oxidation of 1a and in agreement with their disappearance at higher temperature in frozen solution, i.e from 30 K.38 After 1 hour of reaction of the molten mixture, both signals disappeared when the sample was frozen back to 4 K and analysed. The electrochemical behaviour of the compounds involved in the cyclocondensation of 1a was studied by cyclic voltammetry under argon with solutions in the moderately volatile 6

redox potential reduction wave at -0.51 V. The potential is a little different, which can be explained by the nature of the - - counter-anion (PF6 vs. I3 ) for {[Cp*Co]2(-I3}[I3].

Table 2. Electrochemical data for 1a, 1d, 1e, 1h, 1g, + - [[Cp*Co]2(-I)3 ][I3 ], 2a, and TEAI.

Compounds E1/2 1a 1.11irr 0.52irr 1d 1.19irr 0.59irr -1.60 1e 1.23irr 0.61irr 1hb 1.11irr 0.49irr 1g 0.80irr -1.21irr - 1.76irr irr a {[Cp*Co]2(- 1.25 0.36 -0.02 -0.51 - irr I)3}[I3] (168) (293) (120) 1.76 irr [2a][ BArF24] -1.39 TEAI 0.40 -0.08a (116) (400) Fc 0.16 (72) Potentials (in V) vs. Ag/AgCl were obtained from cyclic voltammetry -1 -1 in DCE with 0.1 mol L [N(nBu)4][PF6]. Scan rate = 100 mV s . Work- ing electrode: Pt, d= 3 mm in diameter. The given half-wave potentials irr are equal to E1/2 = (Epa+ Epc)/2. Under bracket: △Ep = |Epa-Epc|. : Irre- a - - versible peak potential. Slow process. Italic (blue color): couple I3 /I or - b I2/I3 . see (9). + - Figure 6 a) Cyclic voltammograms of {[Cp*Co]2(-I)3 }[I3 ], -1 The last reduction at -1.37 V (peak c) may be due to the irre- TEAI, 1a and [2a][BArF24] (0.4 mmol L ) on Pt electrode in a -1 versible reduction of the pyridinium group of [2a]+. Indeed, DCE solution in the presence of [N(nBu)4][PF6] (0.1 mol L ). v = -1 100 mV s . b) Cyclic voltammograms of 1a running first anodic the cyclic voltammogram of [2a][BArF24] shows similar irre- or cathodic sweep. **: Ill-defined peak which may be due to the versible reduction at -1.39 V (Figure 6a, green curve). This reduction of 1a. irreversible reduction is attributed to the formation of a reactive pyridyl radical which may undergo homocoupling through For 1a, two irreversible oxidation peaks were detected at 0.52 C-C bond formation.41 and 1.11 V vs. Ag/AgCl (Figure 6a, red curve). The first oxidation corresponds to the oxidation of Co(III) to Co(IV).13 In the case of the cathodic scanning before the anodic scan- Note that the intensity of the peak current of the second irre- ning, we did not detect the additional signals obtained only versible oxidation at 1.11 V is a little less than half of the after the irreversible oxidation of 1a (Figure 6b). However, intensity of the first oxidation. It may corresponds to the oxi- one small reoxidation peak at around -0.5 V was detected and dation of the intermediate of the reaction giving [2a]+ (cf. is probably arising from the species generated during the re- Supporting Material) or may be related to the oxidation of the duction of the solvent when reaching the cathodic limit or released {[Cp*Co(III)] (-I) }+ (cf. Supporting Material). from the reduction of 1a (ill-defined peak (**) detected in 2 3 Figure 6b). The irreversibility of the oxidation of 1a is fully in agreement Finally, compounds 1d, 1e and the new “CpCo” complex 1h with the reactivity of 1a with [Ph3C][BArF24] leading to the + give a redox behaviour similar to that of 1a (Table 2 and Sup- formation of [2a] , “CoI” and {[Cp*Co] (-I) }. During the 2 3 porting Material) showing that the potential of oxidation of the reverse sweep after the oxidation 1a, three new irreversible Co(III) is little affected by the nature of substituent R (R = H, peaks are detected at -0.25 V (peak a), -0.57 V (peak b) and - Br or CF ) or by substitution of Cp* by Cp in cobaltacycles, 1.37 V (peak c). The three signals should correspond to the 3 like reported for other metallacycles.42 detection of the compounds formed during the irreversible oxidation. However, in the case of 1g the redox behaviour is dissimilar with 1a: replacing the iodo by a chlorido ligand leads to only The first irreversible cathodic peak at -0.25 V has a redox one irreversible oxidation and two irreversible reduction at potential similar to that of TEAI and may correspond to the - - - 0.80 V, -1.21 V and -1.76 V respectively. reduction of I3 to I indicating probably the release of I during the oxidation of 1a. The second signal at -0.57 V (peak b) can be attributed to the reduction of the {[Cp*Co]2(-I)3} species which gives similar 7

Recent reports6,49 have indeed stressed the assisting effect of local attractive NCI interactions in the reactant complexes of base assisted aromatic C-H bond activation by Pd(II), Ni(II), Co(III) and Ir(III) centres. NCI plots are particularly informative of the interactions in reactant complexes that may prefig- ure new bonds appearing once the transition state is passed.50 NCI plots43 were therefore scrutinized for peculiar features related to the uncommon cyclocondensation reported here. The NCI plots43 depicted in Figure 7 display the isosurfaces of attractive domains (red coloured isosurfaces) and repulsive or non-bonding ones (blue coloured isosurfaces) in various singlet and doublet state geometries of complexes relevant to this report, including the one reported by Perez-Temprano.13 The circular attractive domain isosurface located underneath the Cp*ligand in all the cases depicted in Figure 7 is frequently observed in 5-cyclopentadienyl and 6-arene- metal complexes.51 One uncommon feature however is when this ring-like isosurface is no more circular like in Ir-1a but displays attractive outgrowths, like in 1a: these outgrowths are symptomatic of a peculiar attractive relationship between one carbon atom at the cyclopentadienyl moiety and an atom of a group of atoms at a different vicinal ligand.

This is the case for the attractive domains between carbon C4 of the Cp* and carbon C2’ of the phpy ligand that bind during the cyclocondensation and which are present in singlet state 1a but not in singlet state Ir-1a. This difference in attractive NCI is even more pronounced when the Co(III) centre is oxidized to Co(IV) to give [1a’]+ (Figure 7) . 43 Figure 7. NCI plots from ADFview2019 for optimized ge- The doublet state Co(IV) complex [1a’]+ not only exhibits a ometries of a) singlet ground state 1a, b) its singlet ground + larger attractive domain between C4 of the Cp* and C2’ of the state Ir analogue Ir-1a, c) doublet ground state [1a’] , d) dou- phpy ligand, but an attractive domain between C of the Cp* blet ground state [Ir-1a’]+ e) doublet ground state 3 + + and N of phpy also appears. None of these attractions appear [Cp*Co(SCF3)(phpy)] and f) doublet ground state [1g’] for the Ir(IV) analogue. It is also worthy to note the slight (SCM-ADF201944 ZORA45-PBE46-D4(EEQ)47 / all electron 48 rapprochement between Cp* and phpy induced by oxidation in TZP / COSMO (CH2Cl2) level). Color code for atoms: dark both cases. blue N, gray C, white H, yellow S, light green F, orange Ir; NCI plots of the doublet states of Co(IV) purple I; indigo Co. NCIs are materialized by reduced density + 13 + gradient isosurfaces (cut-off value s= 0.5 a.u.,=0.02 a.u.) [Cp*Co(SCF3)(phpy)] and [1g’] , i.e the oxidized form of coloured according to the sign of the signed density i.e. in red 1g, (Figure 7e,f respectively) all show a significant Cp*-phpy for attractive NCI domains and in blue for repulsive (or non- C-C attractive domain while the N-C3 attractive domain de- velops weakly only for [1g’]+ and is absent in Co(IV) bonding) NCI domains. Specific NCI attraction domains + [Cp*Co(SCF3)(phpy)] . Worthy to note, in the case of between the Cp* and the phpy ligand are circled in green (cf. + [Cp*Co(SCF3)(phpy)] the phpy-related C2’-S interaction is Supporting Material for larger drawings). + strongly attractive whereas in [1a] the C2’-I interaction bears a

repulsive/non-bonding character inwards C2’. + DFT evaluation of the mechanism of formation of [2a] . To get a more quantitative insight onto the difference in inter- NCI plots and IBSI suggest a predisposition to cyclocon- action strength between the Co and Ir complexes 1a and Ir-1a, densation. A preliminary comparative study between the at the +III and +IV formal oxidation states and for reactive cobalt complex 1a and the iodoiridium analogue Ir-1a [Cp*Co(SCF3)(phpy)] and 1g at the +III and +IV formal oxi- was carried out to address the propensity of 1a to undergo the dation state, the integrated independent Gradient Model condensation of Cp* with phpy under oxidative conditions (IGM)-based δginter descriptor17 (also called Δginter interaction while its Ir analogue supposedly does not undergo cyclocon- score) and the recently developed Intrinsic Bond Strength densation. Note that according to Mayer et al.11 Ir-1a is re- Index18 (IBSI) were computed to quantify the interaction be- portedly insensitive to PhI(OAc)2. The computed difference in tween phpy and each of the two other ligands. the Gibbs free energy of oxidation for the Co(III) and Ir(III) complexes is negligible and cannot explain the difference of behaviour (ΔΔG=0.1 kcal/mol). Our attention turned to the possible existence of attractive noncovalent interactions that could predispose 1a to its collapse by cyclocondensation.

Table 3. Interatomic “Intrinsic Bond Strength Indexes” (IBSI)18 and the integrated difference between the user-defined fragment-based density gradients with and without interference IGM17ginter computed for 1a and [1a’]+, Ir-1a [Ir-1a]+, 1g, + + [1g’] , [Cp*Co(SCF3)(phpy)] and [Cp*Co(SCF3)(phpy)] .

Me Me Me

Me C3 C4 Me M X N C2'

+ + + + 1a Ir-1a [1a’] [Ir-1a’] 1g [1g’] [Cp*Co(SCF3)(phpy)] [Cp*Co(SCF3)(phpy)] Atom-pair IBSI (au) interactions

C4-C2’ 0.047 0.038 0.052 0.020 0.044 0.053 0.038 0.048

C3-N 0.026 0.012 0.048 0.009 0.026 0.027 0.021 0.029

C4-Co/Ir 0.253 0.274 0.232 0.248 0.244 0.203 0.242 0.192

C3-Co/Ir 0.259 0.301 0.213 0.247 0.250 0.223 0.235 0.204

C2’-Co/Ir 0.479 0.488 0.469 0.498 0.497 0.460 0.485 0.459 N-Co/Ir 0.385 0.402 0.376 0.410 0.385 0.378 0.392 0.391 a a a a b b c c C2’-X 0.026 0.024 0.024 0.020 0.030 0.027 0.046 0.032 interacting ginter (au) fragments Cp*-phpy 0.865 0.630 0.935 0.633 0.857 0.920 0.859 0.917 phpy-X 0.292a 0.343a 0.220a 0.216a 0.288b 0.270b 0.458d 0.513d

Noncovalent interactions (NCIs) are evidenced18 by 0 < IBSI < 0.15, transition-metal coordination bond by 0.15 < IBSI < 0.60, and covalent bonding by a b c d 0.15 < IBSI < 4.0 values. X= I. X= Cl. X= S. X= SCF3

the +IV formal oxidation state. Oxidation increases the inter- The IGM-δginter descriptor can be seen as a measure of the action between Cp* and phpy in the Co case, while the in- electron sharing caused by the electron density interference crease is almost null in the Ir one. Conversely, the interaction between two user-defined fragments. It assesses the mutual between phpy and the iodo ligand decreases with the oxidation density penetration by the deviation of the true electron den- on both cases and, while it is stronger in the Co case than in sity gradient from a non-interacting reference, i.e the inde- the Ir one at the +III formal oxidation state, it is the opposite at pendent gradient model bearing the same electron density as the +IV formal oxidation state. 37,38 the real system. Integrating this descriptor gives rise to the IBSI evidences a stronger interaction between C3 of the Cp* inter Δg interaction score that measures the interaction between moiety and C2’ of the phpy ligand, as well as between C4 of the the two user-defined fragments. IBSI is another new efficient Cp* and N of the phpy, in 1a than in its Ir analogue, while the quantitative index allowing to intra-molecularly probe the interaction between the four aforementioned atoms and the “strength” of a given covalent bond or weaker interaction also metal center is stronger in the Ir analogue. The same observa- based on the so-called Independent Gradient Model,18 which tion stands when the complex is oxidized. Moreover, in the was validated by tests on 677 bonds in 235 molecular systems. Co case oxidation increases the strength of the interaction Like many QTAIM16-based analytical methods the Δginter score between the Cp* and the phpy, while decreasing the strength and IBSI are exempt of issues related to basis set superposition of the interaction between the ligands and the metal center. error. Both Δginter score and IBSI have previously been suc- This analysis suggests that the condensation of the phpy ligand cessfully applied in mechanistic studies for the rationalization with the Cp* moiety could be somewhat encoded in the favor- of the discrimination between competing reaction pathways52 able IBSI and NCI signatures observed in 1a and its oxidized and are particularly useful in assessing reactional predisposi- Co(IV) form [1a’]+. tions in so-called reactant complexes. The general qualitative behaviour of the Δginter interaction Like shown below, oxidized cobaltacycles can be considered score and of the IBSI remains the same when replacing the as typical reactant complexes at a mechanistic crossroad that iodo ligand by a chloro one in the Co complex (Table 3). The are capable of either evolving by reductive elimination by the main difference lies in the Δginter between phpy and the halido coupling of ligand X with the phpy, which corresponds to ligand which is lower in the chloro analogue than in the iodo observation reported by Perez-Temprano et al.13 for the at the +III formal oxidation state, while it is higher at the +IV thiolated complex [Cp*Co(SCF3)(phpy)] or by cyclocondensa- formal oxidation state. tion of phpy with the Cp* ligand (this report). 13-14 When comparing 1a to [Cp*Co(SCF3)(phpy)], the general In the cases addressed here (cf. Table 3), Δginter evidences a quantitative behaviour of the Δginter interaction score between stronger interaction between the Cp* and phpy ligands in the Cp* and phpy, and of the IBSI remain the same (Table 3). iodo-Co case than in the iodo-Ir one, in the +III as well as in However, the Δginter interaction score between phpy and the X 9 ligand is higher when X= SCF3 than when X= I : while it decreases in the latter case, it increases in the former, consis- tently with the favoured C-S bond formation by reductive elimination.13 More important is the fact the the ginter for the Cp*-phpy interaction in almost identical in 1a/[1a’]+ and + [Cp*Co(SCF3)(phpy)]/[Cp*Co(SCF3)(phpy)] whereas in the latter case no metallacycle collapse has been reported.13

The DFT exploration of the formation of [2a]+. Several scenarios were considered for the mechanism of formation of [2a]+. In this study, only intramolecular mechanisms of condensation were considered, keeping in mind that from the experimental investigations mainly two questions remain unanswered, that are: a) where does the hydrogen H5 originate from? b) what is the fate of the Co after the release of [2a]+? We considered that the formulation of a plausible reaction energy profile should be based on the following crucial three steps starting from 1a: i) attack of the carbanionic carbon of the phpy ligand, on carbon C4 at Cp* ii) attack of the nitrogen of the phpy on carbon C3 at Cp* and iii) capture of ≠ hydrogen H5 (likely as a H•). The oxidation of 1a. Like suggested by EPR spectroscopy, cyclic voltammetry and experimental evidence of the ready oxidation of 1a, the first step of the condensation reaction conspicuously involves the oxidation of singlet-state 1a into Co(IV) complex [1a’]+ which was considered therein in its doublet-state (by about 20 kcal/mol more stable than the quar- tet 3/2 spin state) in a way similar to the treatment operated by 13 Pérez-Temprano et al. for [Cp*Co(SCF3)(phpy)]. The com- Scheme 7. Proposed network of reaction pathways for the puted Gibbs free energies of the single electron oxidation of + condensation of Cp* on phpy. pathway1 (light pink): 1a → 1a by Ph3C and by I2 amount to + 9 kcal/mol and + 19 + + + + + [1a'] → [IntC’] → [IntCH] → [pre-2a] → [pre-2a-sol] kcal/mol respectively. The Gibbs enthalpy value for Ph3C is → [2a]+, pathway 2 (light blue): 1a → [1a’]+ → [IntC’]+ → a rather low endergonic value for an equilibrated oxidation + + + + + [pre-2a’] → [pre-2a] → [pre-2a-sol] → [2a] , pathway 3 given that the irreversible consumption of [1a’] draws this (light green): 1a → [1a’]+ → [1a-H]+ → [IntCH] → [pre- oxidation step to completion by virtue of the mass action law. 2a]+ → [pre-2a-sol]+ → [2a]+. An additional single electron oxidation of [1a’]+ into singlet or + triplet-state [1a”] was eventually discarded, for its Gibbs free The intrinsic mechanism of cyclocondensation and the “H” energy exceeded + 40 kcal/mol. atom capture question. Owing to the uncertainty existing + Note that the spin density of [1a'] is essentially located at the around the origin of H5 all calculations were carried out con- Co center. sidering the intervention of a fictitious H• as a starting model to compute reaction “intrinsic” energy profiles. This hypothe- Ph Ph Ph sis has the advantage of providing a lower bound reference Ph Ph Ph Ph value of the H• capture energy to which actual sources of H Ph H Ph - H atom can be confronted. For instance, considering intuitively Ph Ph Ph that Ph3CH could be the source of H• in either oxidative Ph Gomberg-Jacobson-Lankamp-Nauta-MacLean Ph Ph mechanisms entailed a prohibitive surge in the overall reaction quinoid structure Ph Ph Ph Ph Gibbs enthalpy of ΔΔG= + 71 kcal/mol, which discards Ph3CH as a reasonable H atom source (Figure 8b). Scheme 6. The so-called Gomberg-Jacobson-Lankamp- Nauta-MacLean asymmetric quinoid-like structure as a source of H.

onic hydrido-cobalt(V) species corresponds formally to the capture of a H+ by 1a, which has already been ruled out ex- perimentally for it leads exclusively to the demetallation of the phpy ligand. Moreover, all attempted optimisations of the hydrido-Co(V) species in the singlet state, that is the intermediate arising from the capture of a H• by [1a']+ (or of a H+ by 1a) evolved into “agostic”6,53 C-H…Co complexes (Figure 9) where the hydro- + gen is bound either to the C2’ of the phpy ([ag-phpy] ) or to the + + C5 of the Cp* ([ag-Cp*-H] ), [ag-phpy] being more stable than [ag-Cp*-H]+ by 20 kcal/mol (ΔΔG).

Figure 8. Proposed computed Gibbs free energy profiles of the formation of [2a]+. The three possible pathways are materialized by solid coloured lines (green, blue, pink). a) Exer- gonic intrinsic Gibbs energy profile considering a fictitious H. b) Endergonic intrinsic Gibbs energy profile considering Ph3CH as the source of H. c) Endergonic intrinsic Gibbs energy profile considering Ph3C-C6H5C(Ph2) as the source of H. The so-called Gomberg-Jacobson-Lankamp-Nauta-MacLean Figure 9. Singlet and triplet state C-H…Co agostic transients‘ geometries computed at the (COSMO)-ZORA-PBE-D4(EEQ)/all electron TZP level asymmetric quinoid-like structure Ph3C-C6H5=C(Ph2) arising from the reaction of two units of Ph C (G = + 9 kcal/mol) with selected interatomic distances. Color code for atoms: gray C, blue N, 3 indigo Co, white H, purple I. was also considered as a potential source of H as shown in Scheme 6. This hypothesis proved to be unfavourable as it would result in a surge in the overall reaction Gibbs enthalpy Triplet spin states were also considered for the different inter- of G= + 46 kcal/mol compared to the reference reaction mediates arising after the capture of H as spin transitions Gibbs enthalpy with H (Figure 8c). The same observation between triplet and singlet states cannot be fully7a ruled out in could be made when considering the Cp* ligand from external chemical reactions implying Co(III).54 1a or [1a’]+ to be the source of H• (ΔΔG= +82 kcal/mol and +68 kcal/mol, respectively). Tentative optimization of the hydrido-intermediate Co(V) in + the triplet state evolved unavoidably either into a non-agostic The doublet-state Co(IV) species [1a’] was therefore consid- 4 + + ( -Cp*H)Co(III)complex [1a-H] that might undergo the ered as pivotal in the mechanism of formation of [2a] . condensation of Cp* with the phpy ligand or by the insertion Three pathways appeared to be realistic with respect to the of H in the Co-I bond leading to the release of HI, the former crucial H atom capture step (Scheme 7): 1) the capture of a + being the most favourable by 6 kcal/mol (ΔΔG). Both prod- formal H• by [1a'] occurs before the condensation reaction ucts in the triplet state are less stable than other products (pathway 1), 2) the capture of a formal H• after the migration formed along a singlet state path. of phpy’s C to the Cp* ligand and before the migration of the N centre (pathway 2) and 3) the condensation of the Cp* with Pathway 1, cyclocondensation. Considering the attack of the Cp* on the phpy ligand from the (η4-Cp*H)Co complex in the the phpy ligand occurs prior to the capture of a formal H• + (pathway 3). triplet state [1a-H] , three possible paths can be followed: i) attack on C3 of Cp* by the nitrogen of the phpy ligand first, Pathway 1, the early capture of H• and spin state. The then on C4 of the Cp* by the carbanionic C2' of phpy, ii) attack formation of a hydrido-cobalt(V) species by capture of a H• on C4 of Cp* by C2' of phpy, then on C3 of Cp* by N of phpy, was considered to account for the exclusive endo positioning iii) concerted attack on C3 and C4 of Cp* by N and C2' of phpy of H5. The oxidation of singlet state 1a into doublet state respectively. [1a’]+ followed by the capture of H• to yield a putative cati- 11

Out of these possible paths, the attack on C4 of Cp* by C2' [Co(Cl)(I)(CH2Cl)] complex, becomes exergonic (ΔG= - 8 (phpy) displays the lowest activation barrier (ΔG= + 7 kcal/mol). kcal/mol, while ΔG=+ 48 kcal/mol for the concerted attack In summary, all three proposed pathways start with the oxida- and ΔG= + 42 kcal/mol for the attack of N). Each of the three tion of 1a into doublet [1a’]+ and subsequently diverge: paths from the singlet state agostic [ag-Cp*-H]+ complex -pathway 1: H is first captured by the Co(IV) radical yielding display an activation barrier higher than + 30 kcal/mol. There- the triplet-state [1a-H]+, then C and N of the phpy sequen- fore, all of these considered transition states are higher in 2' tially attack the Cp*, energy than the one corresponding to the coupling of C2’ (phpy) with C4 in the triplet state. -pathway 2: C2' of the phpy first attacks C4 at the Cp*, then H is captured by the Co(IV) radical yielding the triplet state Following this triplet-state path, the formation of the interme- + + intermediate [IntCH] before N-C bond formation, diate where the C-C bond has been formed, i.e [IntCH] , is exergonic (ΔG= - 14 kcal/mol). This intermediate is more -pathway 3: the C-C and N-C couplings occur stepwise in the stable in the triplet state than in the singlet by a G of +15 Co(IV) species before H is captured. kcal/mol. In the triplet state the formation of the final com- Discrimination between these three pathways, for which Fig- plex [pre-2a]+, i.e when the final condensation product [2a]+ is ure 8a materializes the intrinsic energy profile considering a still bound to the “CoI” moiety, from [IntCH]+ is then ender- fictitious H source, depends on the nature of the latter, which gonic (ΔG= + 6 kcal/mol). Computations suggest this reaction remains still to be identified. Figure 8b-c shows how two 2 as barrier-less, as the shift of the Co from a N position to a  different H sources, namely Ph3CH and Ph3C-C6H5C(Ph2), position on the aryl fragment combined with the search for may affect the overall energy reaction profile changing the planarity of the phpy moiety leading to a proximity between energy precedence of pathways 1, 2 and 3 in the overall reac- C3 of the former Cp* and the N, favours the formation of the tion network by increasing or lowering activation barriers at C-N bond. Intermediate [pre-2a]+ is more likely to stay in the different stages. triplet state, as its singlet equivalent lies 7 kcal/mol above. Pathway 2 and 3, the early C-C coupling. Also considered CONCLUSION was the attack on the Cp* by the phpy ligand in [1a’]+ before In this study the sensitivity of Co(III) complexes to oxidation the capture of the hydrogen atom occurs later in the process. has been established by experimental methods and its relation- Again three options were considered: the prior attack on C at 4 ship to the formation of [2a,d,e]+ demonstrated. The in- Cp* by C2’ at the phpy ligand shows the lowest activation - barrier (ΔG= + 19 kcal/mol, while ΔG= + 48 kcal/mol for the tramolecular cyclocondensation of the {N,C} heterochelat- concerted attack and ΔG= + 42 for the prior attack by N). In ing ligand with the Cp* moiety is a rather fast process induced these conditions, the formation of the C-C bond in intermedi- by the formation of a Co(IV) transient. DFT calculations ate [IntC']+ from [1a']+ is endergonic with a ΔG of + 12 suggest a pathway where C-C and C-N bond forming steps kcal/mol. may occur before the capture of a H atom, the origin of which has not yet been settled. From this investigation it appears that Pathway 2, the capture of H. The capture of H by inter- + consideration for spin states at Co is important. Also impor- mediate [IntC'] was considered by attempting the optimiza- tant is the scrutiny of NCI plot patterns, which suggests that tion of a hydrido-Co(III) species, which eventually ended up the studied cobaltacycles particularly in their Co(IV) state are in the migration of H to C5, thus yielding the triplet-state in- + somewhat prepared to cyclocondensate owing to marked at- termediate [IntCH] which undergoes the N-C coupling as tractive NCI domains that seemingly predispose the interac- mentioned for pathway 1. tion of the {N,C}- ligand with Cp*. The relevance of this Pathway 3, the N-C coupling followed by the H capture. unprecedented process of collapse of the Co metallacycle From [IntC']+, the formation of the C-N bond is also ender- particularly raises the fundamental question of the viability of gonic (ΔG= + 20 kcal/mol) and barrierless. Tentative optimi- such organometallic intermediates in C-H bond functionaliza- zation of a hydrido-Co(III) species accounting for the capture tion catalysis where “redox” co-catalysts are commonly used 14,55 of a H eventually ended up in the migration of H to C5, yield- on a trial-and-error basis. If Pérez-Temprano et al. evi- ing [pre-2a]+. denced the role of Co(IV) species in C-H bond sulfurization13 The fate of "CoI". The question of the fate of the “CoI” they did not mention the herein described irreversible collapse moiety when [2a]+ is released has been addressed heuristi- of the metallacycle. We speculate that in the context of com- cally. Calculation reveals that the release of “naked” CoI as plex catalytic mixtures containing reagents and co-catalysts such from triplet-state [pre-2a]+ is highly unfavourable what- endowed with oxidoreductive mediation roles, where similar 7b,7c,56 ever the spin multiplicity of CoI may eventually be (ΔG= + 81 metallacycles are claimed to be key intermediates, the kcal/mol for the singlet, ΔG= + 48.8 kcal/mol for the triplet, occurrence of such an unwanted cyclocondensation pathway and ΔG= + 69 kcal/mol for the quintet). It was therefore in- depends on the nature of ligand X (Scheme 8), like suggested inter ferred that the “CoI” moiety might undergo the oxidative- by the IBSI and g scores. addition of a solvent molecule. Insertion of the Co(I) of CoI Particularly crucial in such catalytic reactions are the kinetics in the C-Cl bond of a solvent molecule after the release of of the reductive elimination step that releases the pivotal + [2a] was found highly stabilizing (ΔG= - 48 kcal/mol). The “Cp*Co” moiety in the medium alongside the main organic + addition of a solvent molecule before the release of [2a] was product: a disfavoured reductive-elimination may result in low also considered. The insertion of the Co in the C-Cl bond is conversions due to a more favourable "autolysis" of the cata- endergonic (ΔG= + 4 kcal/mol) in the triplet state (lowest spin lyst by the herein described cyclocondensation (Scheme 8). state). From there, the release of [2a]+, generating a 12

Electrochemical investigation suggests that the dechelation of CF3-phpyH, 98%), 2-(p-tolyl)pyridine (98%). The cobalt(III) + 5 the Co centre to give the dimeric {[Cp*Co]2(-I)3} competes complexes [Co(η -C5Me5)I2]2 and LiNHAc were prepared as with the cyclocondensation process, which might be a crucial reported in the literature.6,59 All NMR yields were calculated information in understanding all the underlying issues of the using 1,3,5-trimethoxybenzene as internal 1H NMR standard. actual mechanism of catalytic C-H bond functionalization and All solvents were distilled over sodium or CaH2 under argon in improving the performance of catalysts. before use. 1,2-Dichloroethane used in cyclic voltammetry experiments was distilled over CaH2 under argon. Deuterated solvents were dried over sodium or CaH2, filtered over acti- vated neutral alumina, and stored under argon before use. 1H (300, 400, 500, and 600 MHz) and 13C (75 and 126 MHz) NMR spectra were measured on Bruker DPX 300 and 400, Avance I 500, and Avance III 600 spectrometers. Chemical shifts (expressed in parts per million) were referenced against solvent peaks. Mass spectra were run on a MicroTOF Bruker Daltonics spectrometer, using a TOF-ESI coupling analysis system. Elemental analyses were achieved with Thermo Sci- entific FLASH 2000 CHNS/O analyzers.

X-ray diffraction analysis. The crystals were placed in oil, and a single crystal was selected, mounted on a glass fibre and placed in a low-temperature N2 stream. X-ray diffraction data collection was carried out on a Bruker PHOTON III DUO CPAD diffractometer equipped with an Oxford Cryosystem liquid N2 device, using Mo-Kα radiation (λ = 0.71073 Å). The crystal-detector distance was 37 mm. The cell parameters were determined (APEX360 software) from reflections taken from 1 set of 180 frames at 1s exposure. The structures were solved Scheme 8. Which pathway prevails in the catalysis ? using the program SHELXT-2014.61 The refinement and all The collapse of the metallacycle or the reductive elimi- further calculations were carried out using SHELXL-2014 or 62 nation of the targeted product ? How does the nature of SHELXL-2018. For compound [3e][BArF24], the hydrogen ligand „X“ determine the outcome ? atom of the OH group was located from Fourier difference. The other H-atoms were included in calculated positions and In the context of the catalysis, the cyclocondensative collapse treated as riding atoms using SHELXL default parameters. The non-H atoms were refined anisotropically, using weighted of cobaltacycles represents an irreversible destructive deacti- 2 vation pathway, which further confirms that establishing the full-matrix least-squares on F . A semi-empirical absorption whereabouts of the Cp(*)Co(III)-based catalysts after catalysis correction was applied using SADABS in APEX3. is an imperious need. The combination of QTAIM-based IBSI, IGM and NCI-plot might prove particularly informative Cyclic Voltammetry. All glassware was dried for 24 hours at in this endeavour like shown above. From the view point of 90°C prior to use. Voltammetric data were recorded using an metal-mediated synthesis, the oxidation-induced cycloconden- Autolab PGSTAT30 potentiostat (Eco Chemie, Holland) sation of Cp*Co(III) metallacycles reported herein is a new driven by the GPSE software with a standard three-electrode example of oxidation-triggered Co(IV)-centred "metal- system flushed with a gentle stream of argon. The electrolyte 57 templated" synthesis of valuable three-dimensional alkaloids consisted of a solution of 0.1 mol L-1 of anhydrous tetrabu- of the Berberine58 class, the hydroxylation reaction of which tylammonium hexafluorophosphate ([NBu4][PF6]) in dry dis- deserves further investigations. tilled 1,2-C2H4Cl2. The reference electrode was the Ag/AgCl electrode, which was electrically connected to the solution by

a junction bridge filled with 5 mL of anhydrous CH3CN + 0.1 M [NBu4][PF6]. The working electrode was a platinum disk EXPERIMENTAL PART electrode (d = 3 mm in diameter). All potentials are given vs. General information. All experiments were conducted under Ag/AgCl used as external reference and are uncorrected from a dry argon atmosphere using standard Schlenk lines and dry ohmic drop. glovebox techniques. The following compounds were purchased from Sigma Aldrich: 2-phenylpyridine (abbr. phpyH, 98%), tritylium chloride (97%). The following compounds EPR spectroscopy. EPR spectra were recorded on a continu- were purchased from Alfa Aesar: 1,2,3,4,5- ous‐ wave X‐ band EMX‐ plus equipped with TE102 rectan- pentamethylcyclopentadiene (abbr. Cp*, 94%), and sodium gular cavity (ER 4102ST, Bruker) and an ESR900 continuous tetrakis[3,5-bis(trifluoromethyl)phenyl]borate (abbr. Na- flow cryostat controlled with an Oxford ITC503S. EPR spectra were recorded with the following parameters: microwave BArF24, 97%). The following compounds were purchased from TCL chemicals: 2-(4-bromophenyl)pyridine (abbr. Br- power of 1 mW, a modulation frequency and amplitude of phpyH, 98%), 2-[4-(trifluoromethyl)phenyl]pyridine (abbr. respectively 100 kHz and 0.4 mT, a conversion time of 300 ms 13 and a time constant of 20.48 ms. A 2nd order polynomial simulating dichloromethane (ε= 8.9, r= 2.94 Å) as a solvent on baseline was subtracted to the spectra recorded between 0 and the previously optimized geometry using the ORCA program 450 mT. The simulation was done using Easyspin63 toolbox system72 version 4.1.1. working under MatLab (MathWorks). Preparation of the Tritylium tetrakis[3,5- ITC experiments. Isothermal titration calorimetry (ITC) bis(trifluoromethyl)phenyl]borate ([Ph3C][BArF24]). experiments were carried out on a Waters-SAS nano-ITC [Ph3C][BArF24] was prepared according to a literature proce- device (TA Instruments®) equipped with two stainless steel dure with minor modification.73 Under argon atmosphere, a Hastelloy cells of 1 mL volume each) and housed in an argon- mixture NaBArF24 (886.2 mg, 1 mmol) and tritylium chloride filled glovebox. Auto equilibration of the ITC was performed (278.1 mg, 1 mmol) was stirred overnight in ~20 mL of before every experiment to reach an acceptable baseline. The CH2Cl2 at room temperature in a sealed Schlenk vessel. The solutions of reactants were prepared in the same glovebox by reaction mixture was then filtrated through Celite, the filtrate dissolving a mass of substrate placed in a volumetric flask in was evaporated under reduced pressure. The resulting solid pure, freshly distilled and degassed 1,2-dichloroethane. The was first washed with n-pentane and then purified by recrys- ITC experiments were performed by sequential injection at tallization with CH2Cl2/n-pentane to afford bright yellow 25.00⁰ C with a stirring rate of 300 rpm. The solution of microcrystals (973 mg, 88% yield). 1H NMR (300 MHz, tritylium salt was placed in the injection syringe and the solu- CDCl3) δ 8.15 (ddt, J = 8.8, 7.3, 1.3 Hz, 3H), 7.80 – 7.73 (m, tion of compound 1a in the measure cell. Three thermograms 6H), 7.68 (p, J = 2.1 Hz, 8H), 7.59 – 7.54 (m, 6H), 7.48 (s, were acquired by sequential injection of 2-3 L with a delay 4H). between two injections not exceeding 900 s, and the raw enthalpy of the reaction was inferred by summing up the heat Preparation of the d -trityl chloride . d -Tritylium chloride flow released from which was deduced the residual heat flow 15 . 15 was prepared according to a literature procedure with minor after the thermal inflexion point. 74 modification . Under argon atmosphere, a mixture of d6- benzene (5.0 mL, 0.05 mmol) and carbon tetrachloride (1.0 Computational details mL, 0.01 mmol) was dissolved in 20 mL of 1,1,2,2- DFT-D computations. Geometry optimization of the reac- tetrachloroethane in a sealed Schlenk vessel, and cooled down tants, intermediates, transition states, and products were per- to 0 °C. Aluminium trichloride (1.6 g, 0.01 mmol) was then formed using the SCM-ADF2019.0144 package, at the Density added slowly during stirring. After 15 min of stirring, the Functional Theory (DFT) level. The Perdew-Burke-Ernzerhof solution was refluxed for 90 min and then cooled down to (PBE)46 functional augmented with Grimme’s dispersion room temperature. The solution was treated with 1 M HCl, and corrections with the electronegativity equilibrium model the organic phase was extracted with DCM and dried over (DFT-D4(EEQ))47 was used in all geometry optimizations. Mg2SO4. After removal of the solvent under vacuum, the All geometry computations were carried out using scalar rela- crude solid was purified by column chromatography (hex- tivistic effects with ad hoc all-electron (abbr. ae) single polari- anes/ethyl acetate = 9:1) to afford d15-tritylium chloride (1.2 g, zation functions triple-ζ Slater-type basis set (TZP)64. Solva- 37%). tion was treated by the Conductor-like Screening Model (COSMO)65 procedure assuming CH Cl (ε= 8.9, r= 2.94 Å) as 2 2 Synthesis of CpCo(CO)I2. CpCo(CO)2 (1 mL, 7.5 mmol) a solvent. Geometry optimization by energy gradient minimi- and 20 mL of distilled Et2O were introduced in a Schlenk tube zation were carried out in all cases with integration grid accu- under argon atmosphere. I (2 g, 7.9 mmol) dissolved in dis- 66 2 racy “Normal”. All transition states were found using the tilled Et O was added to the Schlenk tube with a cannula. The 67 2 Nudged Elastic Band (NEB) procedure and submitted to the reaction was stirred overnight at room temperature. The sol- 68 Intrinsic Reaction Coordinate (IRC) procedure to verify the vent was removed under reduced pressure and the residual connection to their reactive complexes and products. Vibra- solid washed 4 times with distilled pentane. 1H NMR analysis tional modes were analytically computed to verify that the in CDCl3 revealed the disappearance of the signal of the start- optimized geometries were related to energy minima or to ing compound at 5.06 ppm that was replaced by a new Cp transition states. For the monoatomic systems that were con- singlet at 5.68 ppm. Analytical data were in accord with those - - sidered (H and I ), as the enthalpy and entropy corrections to reported in the literature.75 the energy are usually computed from the vibrational energies, these contributions were neglected and the Gibbs free energy was approximated as the bond energy. Synthesis of [CpCoI2]2. The synthesis of the [CpCoI2]2 was carried out by applying the methodology reported for IBSI computations. Intrinsic Bond Strength Index (IBSI) of 6 [Cp*CoI ] . In a round bottom flask, 1 g of [CpCo(CO)I ] the key atom pairs were computed using the IGMPLOT pro- 2 2 2 were introduced with 50 mL of distilled heptane. The solution gram17 using gaussian-type wavefunctions obtained from was stirred and refluxed during 2 days. After reaction, the single-point calculations with the PBE functional46 augmented solid was filtrated and washed 3 times with distilled pentane. with Grimme’s correction to the dispersion with a Becke- [CpCoI ] was insoluble in most solvents, but slightly soluble Johnson damping function69(PBE-D3(BJ)), scalar relativistic 2 in coordinating solvents such as CH CN and THF giving rise effects with the balanced Karlsruhe 2nd generation default 3 70 to quite air and light sensitive solutions. All analytical data triple-ζ valence plus polarization (def2-TZVP) basis set and 75b,76 were in accord with data published in literature. Conductor-like Polarizable Continuum Model (CPCM)71

General Procedure for the Synthesis of Iodocobaltacycles. mixture of 1a (95 mg, 0.2 mmol) and dry tetrabutylammonium 1a~e were prepared according to our published procedure.6 A chloride (100 mg, 0.35 mmol) was stirred overnight in mixture of [Cp*CoI2]2 (1 equiv), LiNHAc (6 equiv.), and 2- ∼10−15 mL of acetonitrile at room temperature in a sealed arylpyridine ligand (2 equiv.) were stirred in ∼10−15 mL of Schlenk vessel. After reaction, washed and filtrated with H2O, CH2Cl2 at 50 °C for 48 h in a sealed Schlenk vessel. The the resulting residue was purified by recrystallization with reaction mixture was filtrated over Celite and the resulting CH2Cl2/n-pentane and/or washing with n-pentane, to afford the filtrate was stripped of solvent. The resulting residue was then pure compound the analytical data of which matched pub- 77 1 purified by column chromatography (DCM/n-pentane= 4/1) lished data (60 mg, 79 %). H NMR (400 MHz, CDCl3): δ and recrystallization with CH2Cl2/n-pentane and/or washing 9.24 (dt, J = 5.6, 1.2 Hz, 1H), 8.28 (dd, J = 7.7, 1.1 Hz, 1H), with n-pentane. 7.70 – 7.63 (m, 2H), 7.53 (dd, J = 7.6, 1.5 Hz, 1H), 7.32 (td, J = 7.4, 1.5 Hz, 1H), 7.15 – 7.07 (m, 2H), 1.29 (s, 15H). Iodo[2-(4-methylphenyl)pyridine-κ2- C,N](pentamethylcyclopentadienyl)cobalt(III) (1c). Synthesis of iodo[2-phenylpyridine-κ2- [Cp*CoI2]2 (448 mg, 0.5 mmol), LiNHAc (195 mg, 3 mmol), C,N](cyclopentadienyl)cobalt(III) (1h). The synthesis of 1h and 2-(4-methylphenyl)pyridine (169 mg, 1 mmol) in 20 mL was performed similarly to the one of 1a published recently.6 of CH2Cl2 afforded 1c in 60 % yield (145 mg). Anal. Calcd [CpCoI2]2 (40 mg, 0.053 mmol), 2-phenylpridine (15.1 µL, for C22H25CoIN: C 54.01, H 5.15, N 2.86. Found: C 53.64, H 0.106 mmol) and LiNHCOCH3 (20 mg, 0.307 mmol) were 1 5.20, N 3.02. H NMR (500 MHz, CDCl3): δ 9.16 (dt, J = 5.8, introduced in a Schlenk tube and 5 mL of distilled CH2Cl2 1.3 Hz, 1H), 8.07 (dd, J = 1.6, 0.8 Hz, 1H), 7.62 – 7.53 (m, were added. The reaction was stirred two days at 50°C. The 2H), 7.44 (d, J = 7.7 Hz, 1H), 7.00 (td, J = 5.9, 2.7 Hz, 1H), reaction was filtrated and the resulting filtrate was stripped of 6.84 (ddd, J = 7.8, 1.6, 0.7 Hz, 1H), 2.46 (d, J = 0.7 Hz, 3H), solvent under reduced pressure. The resulting solid residue 13 1.48 (s, 15H). C NMR (126 MHz, CDCl3): δ 181.16, 167.16, was washed 3 times with distilled pentane and further purified 154.92, 143.21, 143.00, 139.08, 136.54, 124.07, 122.96, by chromatography. A gradient of eluent starting from 100% 120.76, 118.10, 93.36, 22.13, 10.39. HRMS (ESI, m/z) calcd. of pentane to 100% of CH2Cl2 to elute the desired purple com- + for C22H25CoN : 362.1319, found: 362.1318. pound. The eluate containing the latter was stripped of solvent under reduced pressure to afford 1h in 40% yield (17 mg). 2 Anal. calcd. for C16H13CoIN·1/2 H2O: C 46.40, H 3.41, N Iodo[2-(4-bromophenyl)pyridine-κ - 1 3.38. Found: C 46.26, H 3.24, N 3.29. H NMR (500 MHz, C,N](pentamethylcyclopentadienyl)cobalt(III) (1d). CDCl ): δ 9.69 – 9.63 (dd, 1H), 8.48 (dd, J = 7.7, 1.1 Hz, 1H), [Cp*CoI ] (448 mg, 0.5 mmol), LiNHAc (195 mg, 3 mmol), 3 2 2 7.70 – 7.60 (m, 3H), 7.31 (td, J = 7.4, 1.5 Hz, 1H), 7.14 – 7.02 and 2-(4-bromophenyl)pyridine (235 mg, 1 mmol) in 20 mL of 13 (2dd, 2H), 5.15 (s, 5H). C NMR (126 MHz, CDCl ): δ CH Cl afforded a yield of 71% (393 mg). Anal. calcd for 3 2 2 169.83, 167.67, 157.17, 146.86, 144.25, 137.09, 129.44, C21H22BrCoIN: C,45.52, H 4.00, N 2.53. Found: C 45.33, H 1 124.25, 123.28, 121.62, 119.03, 86.99. HRMS (ESI, m/z) 4.05, N 2.62. H NMR (500 MHz, CDCl ): δ 9.16 (ddd, J = + 3 calcd. for C H CoN : 278.0380. Found: 278.0374. 5.7, 1.5, 0.8 Hz, 1H), 8.34 (d, J = 1.9 Hz, 1H), 7.63 (ddd, J = 16 13 8.0, 7.1, 1.5 Hz, 1H), 7.59 (ddd, J = 8.2, 1.8, 0.8 Hz, 1H), 7.40 (d, J = 8.2 Hz, 1H), 7.17 (dd, J = 8.2, 1.9 Hz, 1H), 7.08 (ddd, J General Procedure for the oxidative cyclocondensation of = 7.3, 5.7, 1.7 Hz, 1H), 1.50 (s, 15H). 13C NMR (126 MHz, 1a,d,e. A mixture of iodocobaltacycles (1 equiv) and CDCl3): δ 184.10, 166.10, 154.92, 144.21, 143.88, 136.78, [Ph3C][BArF24] (1 equiv) was stirred overnight in ∼10−15 125.80, 124.07, 123.94, 121.52, 118.54, 93.58, 10.26. HRMS mL of CH2Cl2 at room temperature in a sealed Schlenk vessel. + (ESI, m/z) calcd. for C21H22BrCoN : 426.0268. Found: The reaction mixture was filtrated through Celite and the 426.0262. resulting filtrate was stripped of solvent under reduced pressure to afford a solid residue that was purified by column chromatography. Recrystallization of the eluted compound Iodo[2-(4-(trifluoromethyl)phenyl)pyridine-κ2- from a CH Cl /n-pentane mixture and further trituration of the C,N](pentamethylcyclopentadienyl)cobalt(III) (1e). 2 2 residue with n-pentane afforded the product as a white pow- [Cp*CoI ] (448 mg, 0.5 mmol), LiNHAc (195 mg, 3 mmol), 2 2 der. and 2-(4-(trifluoromethyl)phenyl)pyridine (224 mg, 1 mmol) in 10 mL of CH2Cl2 afforded 1e in 62% yield (335 mg). Anal. calcd for C22H22CoF3IN: C,48.64; H,4.08; N,2.58. Found: C, [2a][BArF24]. 1a (48 mg, 0.1 mmol) and [Ph3C][BArF24] 1 48.55; H, 4.04; N, 2.64. H NMR (500 MHz, CDCl3): δ 9.25 – (111 mg, 0.1 mmol) in 10 mL of CH2Cl2, isolated yield: 29%

9.20 (m, 1H), 8.50 (dd, J = 1.8, 0.9 Hz, 1H), 7.72 – 7.66 (m, (33 mg). Anal. calcd for C53H36BF24N: C 55.18, H 3.15, N 2H), 7.63 (dd, J = 8.1, 0.9 Hz, 1H), 7.29 – 7.26 (m, 1H), 7.14 1.21. Found: C 55.21, H 3.34, N 1.21. 1H NMR (500 MHz, 13 (ddd, J = 6.6, 5.7, 2.2 Hz, 1H), 1.50 (s, 15H). C NMR (126 CDCl3): δ 8.30 (dd, J = 6.5, 1.5 Hz, 1H), 8.06 (dd, J = 8.4, 1.6 MHz, CDCl3): δ 181.93, 165.71, 155.20, 148.65, 138.28, Hz, 1H), 7.97 (ddd, J = 8.5, 7.3, 1.4 Hz, 1H), 7.87 (dd, J = 8.0, 137.04, 130.04, 122.49, 122.31, 119.76, 119.35, 93.86, 10.36. 1.3 Hz, 1H), 7.73 (td, J = 7.7, 1.3 Hz, 1H), 7.70 (dt, J = 5.2, 19 F NMR (282 MHz, CDCl3): δ -62.1 ppm. HRMS (ESI, m/z) 2.3 Hz, 8H), 7.59 – 7.51 (m, 2H), 7.50 (s, 4H), 7.45 (ddd, J = + calcd. for C22H22CoF3N : 416.1036, found:416.1035. 7.8, 6.6, 1.6 Hz, 1H), 2.42 – 2.36 (m, 1H), 2.09 (dq, J = 2.7, 1.3 Hz, 3H), 1.70 (p, J = 1.3 Hz, 3H), 1.40 (s, 3H), 1.29 (s, 13 2 3H), 0.96 (d, J = 7.2 Hz, 3H). C NMR (126 MHz, CDCl3): δ Synthesis of chloro(2-phenylpyridine- - 162.43, 162.04, 161.64, 161.25, 150.29, 148.75, 144.57, C,N)(pentamethylcyclopentadienyl)cobalt(III) (1g). A 15

141.09, 138.75, 135.37, 130.92, 129.49, 128.76, 127.87, [3e][BArF24]. 2a (15 mg, 0.03 mmol) and [Ph3C][BArF24] 127.17, 125.94, 125.70, 123.54, 123.41, 122.54, 121.37, (30 mg, 0.03 mmol) in 10 mL CH2Cl2, isolated yield: 17% (6

117.65, 85.48, 53.28, 49.20, 19.84, 16.59, 13.52, 13.00, 9.71. mg). Anal. calcd for C54H35BF27NO: C 52.41, H 2.85, N 1.13. + 1 HRMS (ESI, m/z) calcd. for C21H24N : 290.1903. Found: Found: C 52.03, H 2.80, N 1.13. H NMR (500 MHz, 290.1897. CD2Cl2): δ 8.53 (dd, J = 6.5, 1.4 Hz, 1H), 8.33 (ddd, J = 8.7, 7.4, 1.4 Hz, 1H), 8.26 (dd, J = 8.4, 1.6 Hz, 1H), 8.18 (d, J = 8.2 Hz, 1H), 7.91 (d, J = 8.2 Hz, 2H), 7.79 (ddd, J = 7.9, 6.5, [2d][BArF ]. 1d (27 mg, 0.05 mmol) and [Ph C][BArF ] 24 3 24 1.6 Hz, 1H), 7.72 (p, J = 2.3 Hz, 8H), 7.55 (s, 4H), 2.21 (q, J = (55 mg, 0.1 mmol) in 10 mL of CH Cl , NMR yield: 57%; 2 2 1.4 Hz, 3H), 1.81 – 1.75 (m, 3H), 1.51 (s, 3H), 1.44 (s, 3H), under air, NMR yield : 84%, isolated yield : 37% (22 mg). 13 1.32 (s, 3H), 1.11 (s, 1H). C NMR (126 MHz, CD Cl ): δ Anal. calcd for C H BBrF N·0.15 CH Cl : C 51.26, H 2.86, 2 2 53 35 24 2 2 162.69, 162.30, 161.90, 161.50, 149.61, 147.87, 145.42, N 1.12. Found: C 51.24, H 2.95, N 1.26. 1H NMR (500 MHz, 141.17, 137.81, 135.14, 134.69, 129.74, 129.59, 129.36, CDCl ): δ 8.29 (dd, J = 6.6, 1.4 Hz, 1H), 7.98 (dd, J = 8.5, 1.7 3 129.34, 129.32, 129.29, 129.11, 129.09, 129.07, 129.04, Hz, 1H), 7.90 (ddd, J = 8.4, 7.3, 1.4 Hz, 1H), 7.71 (s, 3H), 128.18, 127.71, 127.20, 126.58, 126.55, 126.01, 125.82, 7.69 (dd, J = 5.1, 2.7 Hz, 8H), 7.50 (s, 4H), 7.45 – 7.38 (m, 125.79, 124.68, 123.85, 121.68, 117.88, 117.85, 117.81, 1H), 2.42 – 2.37 (m, 1H), 2.08 (dd, J = 2.8, 1.4 Hz, 3H), 1.72 19 85.96, 82.60, 55.52, 21.99, 21.23, 20.79, 13.99, 11.34. F – 1.68 (m, 3H), 1.40 (s, 3H), 1.30 (s, 3H), 0.98 (d, J = 7.2 Hz, NMR (282 MHz, CDCl ):  -62.8 ppm. HRMS (ESI, m/z) 3H). 13C NMR (126 MHz, CDCl ): δ 162.43, 162.03, 161.64, 3 3 calcd. for C H F NO+: 374.1726. Found: 374.1700. 161.24, 150.27, 147.88, 144.83, 141.27, 140.61, 134.88, 22 23 3 132.96, 132.11, 131.01, 130.92, 129.24, 128.99, 128.40,

127.85, 126.28, 125.69, 123.52, 123.19, 121.45, 121.35, [2h][BArF24]. 1h (15 mg, 0.03 mmol) and [Ph3C][BArF24] 117.67, 85.61, 53.15, 49.14, 19.89, 16.65, 13.48, 13.02, 9.81. (30 mg, 0.03 mmol) in 10 mL CH2Cl2, isolated yield: 17% (6 + HRMS (ESI, m/z) calcd. for C21H23BrN : 368.1008. Found: mg). Anal. calcd for C59H33BF24N2·1.1 CH2Cl2: C 54.27, H 368.10. 2.67, N 2.11. Found: C 54.15, H 2.80, N 2.04. 1H NMR (300 MHz, CDCl3): δ 8.52 (d, J = 5.4 Hz, 1H), 8.25 (d, J = 6.2 Hz, 1H), 8.01 – 7.89 (m, 2H), 7.89 – 7.75 (m, 3H), 7.69 (d, J = 4.2 [2e][BArF ]. 1e (16 mg, 0.03 mmol) and [Ph C][BArF ] 24 3 24 Hz, 8H), 7.67 – 7.60 (m, 2H), 7.58 – 7.50 (m, 2H), 7.48 (d, J = (33 mg, 0.03 mmol) in 10 mL of CH Cl , NMR yield: 63%; 2 2 1.3 Hz, 4H), 7.47 – 7.38 (m, 3H), 7.26 – 7.17 (m, 1H), 6.83 under air, NMR yield: 88%, isolated yield: 35% (13 mg). (dd, J = 5.4, 2.0 Hz, 1H), 5.75 (dd, J = 5.4, 2.7 Hz, 1H), 5.30 Anal. calcd for C54H35BF27N: C 53.09, H 2.89, N 1.15. Found: 1 (s, 1H), 5.08 (d, J = 7.5 Hz, 1H), 3.06 (dd, J = 15.9, 8.0 Hz, C 52.86, H 3.06, N,1.30. H NMR (500 MHz, CDCl ): δ 8.36 3 1H), 2.80 (dd, J = 15.9, 6.7 Hz, 1H). 13C NMR (126 MHz, (dd, J = 6.6, 1.4 Hz, 1H), 8.06 – 8.01 (m, 1H), 7.98 (d, J = 8.3 CDCl ): δ 162.41, 162.02, 161.62, 161.23, 149.12, 148.74, Hz, 1H), 7.93 (td, J = 7.9, 1.4 Hz, 1H), 7.83 (dd, J = 8.2, 1.8 3 146.05, 145.93, 140.15, 139.67, 137.33, 135.62, 134.87, Hz, 1H), 7.80 (s, 1H), 7.72 – 7.67 (m, 8H), 7.51 – 7.48 (m, 131.30, 130.89, 130.38, 130.01, 129.70, 129.26, 128.99, 4H), 7.47 (d, J = 7.7 Hz, 1H), 2.40 (s, 1H), 2.11 (dq, J = 2.6, 128.17, 127.83, 126.34, 125.66, 125.26, 124.45, 123.80, 1.3 Hz, 3H), 1.73 – 1.70 (m, 3H), 1.41 (s, 3H), 1.35 (s, 3H), 13 123.49, 123.29, 122.53, 120.25, 117.68, 91.44, 85.17, 83.21, 0.99 (d, J = 7.2 Hz, 3H). C NMR (126 MHz, CDCl ): δ 3 82.47, 53.59, 49.67, 48.10. HRMS (ESI, m/z) calcd. for 162.43, 162.04, 161.65, 161.24, 150.54, 147.08, 145.25, C H N +: 373.1699. Found: 373.1700. 141.70, 139.85, 134.86, 131.00, 129.28, 129.01, 127.81, 27 21 2 127.17, 126.39, 125.67, 123.86, 123.51, 117.71, 85.98, 53.26, 49.14, 20.00, 16.68, 13.49, 13.06, 9.81. 19F NMR (282 MHz, ASSOCIATED CONTENT CDCl3):  -62.4 ppm. HRMS (ESI, m/z) calcd. for Supporting Information. Deposition numbers for {[Cp*Co]2(- C H F N+: 358.1777. Found: 358.18. 22 23 3 I)3}[I3] (CCDC 2077967), 1a (CCDC 2077960), 1d (CCDC 2077958), 1e (CCDC 2077959), 1h (CCDC 2077956), [2a][BArF24] (CCDC 2077961), [2d][BArF24] (CCDC 2077964), [3d][BArF24]. 3a (16 mg, 0.03 mmol) and [Ph3C][BArF24] [2e][BArF24] (CCDC 2077963), [3d][BArF24] (CCDC 2077965), (30 mg, 0.03 mmol) in 10 mL CH2Cl2, isolated yield: 21% (8 [3e][BArF24] (CCDC 2077968), [N-protiobenzo[h]quinolinium] mg). Anal. calcd for C53H35BBrF24NO: C 50.99, H 2.83, N 1 [BArF24] (CCDC 2077966) and [2h][BArF24] (CCDC 2077962) 1.12. Found: C 50.69, H 2.76, N 1.18. H NMR (500 MHz, contain the supplementary crystallographic data for this paper. CDCl3): δ 8.46 (dd, J = 6.6, 1.4 Hz, 1H), 8.28 (ddd, J = 8.7, These data are provided free of charge by the joint Cambridge 7.4, 1.4 Hz, 1H), 8.18 (dd, J = 8.3, 1.6 Hz, 1H), 7.91 (d, J = Crystallographic Data Centre and Fachinformationszentrum 8.5 Hz, 1H), 7.85 – 7.77 (m, 2H), 7.73 (d, J = 6.4 Hz, 1H), Karlsruhe Access Structures service 7.73 – 7.70 (m, 8H), 7.56 (s, 4H), 2.21 (q, J = 1.3 Hz, 3H), www.ccdc.cam.ac.uk/structures. Full experimental procedures 1.80 (q, J = 1.3 Hz, 3H), 1.51 (s, 3H), 1.40 (s, 3H), 1.32 (s, and details, voltammograms, EPR, Mass and NMR spectra, ener- 13 3H), 1.06 (s, 1H). C NMR (126 MHz, CDCl3): δ 162.90, gies and Cartesian coordinates, high resolution NCI figures. This 162.50, 162.11, 161.71, 149.63, 148.79, 145.21, 141.00, material is available free of charge via the Internet at 138.82, 135.36, 134.91, 133.33, 132.47, 130.34, 129.54, http://pubs.acs.org. 129.30, 128.55, 128.39, 126.57, 126.23, 125.74, 124.24, 124.06, 118.03, 86.18, 82.57, 55.64, 22.27, 21.38, 21.01, AUTHOR INFORMATION + 14.22, 11.54. HRMS (ESI, m/z) calcd. for C21H23BrNO : Corresponding Author 384.0958, found: 384.1006. * J.-P. Djukic, Email: [email protected]

Author Contributions S.; Rovis, T., Correction to “Correlating Reactivity and Selectivity to Cyclopentadienyl Ligand Properties in Rh(III)-Catalyzed C–H Activation Reactions: The manuscript was written through contributions of all authors. An Experimental and Computational Study” J. Am. Chem. Soc. 2020, 142, 7709- All authors have given approval to the final version of the manu- 7709. (e) Piou, T.; Romanov-Michailidis, F.; Romanova-Michaelides, M.; Jackson, K. E.; Semakul, N.; Taggart, T. D.; Newell, B. S.; Rithner, C. D.; Paton, R. S.; Rovis, script. T., Correlating Reactivity and Selectivity to Cyclopentadienyl Ligand Properties in Rh(III)-Catalyzed C–H Activation Reactions: An Experimental and Computational Study J. Am. Chem. Soc. 2017, 139, 1296-1310. (f) Piou, T.; Rovis, T., Electronic and Steric Tuning of a Prototypical Piano Stool Complex: Rh(III) Catalysis for C–H The authors declare no competing financial interests. Functionalization Acc. Chem. Res. 2018, 51, 170-180. (g) Yamada, T.; Shibata, Y.; Tanaka, K., Functionalized Cyclopentadienyl Ligands and Their Substituent Effects on a Rhodium(III)-Catalyzed Oxidative [4+2] Annulation of Indole- and Pyrrole-1- Carboxamides with Alkynes Asian J. Org. Chem. 2018, 7, 1396-1402. (h) Campos, Funding Sources J.; Hintermair, U.; Brewster, T. P.; Takase, M. K.; Crabtree, R. H., Catalyst Activation by Loss of Cyclopentadienyl Ligands in Hydrogen Transfer Catalysis The Centre National de la Recherche Scientifique with Cp*IrIII Complexes ACS Catalysis 2014, 4, 973-985. (i) Jutzi, P.; Reumann, The University of Strasbourg G., Cp* Chemistry of main-group elements J. Chem. Soc., Dalton Trans. 2000, 2237-2244. (3) Rottink, M. K.; Angelici, R. J., Ligand and metal effects on the enthalpies of GENCI-IDRIS Grant 2020-A0080811408 protonation of Cp'M(PR3)(PR'3)X complexes (M = ruthenium or osmium) J. Am. Chem. Soc. 1993, 115, 7267-7274. HPC Center of the University of Strasbourg Grant g2020a78c. (4) Park, J.; Chang, S., Comparison of the Reactivities and Selectivities of Group 9 [Cp*MIII] Catalysts in C−H Functionalization Reactions Chem. – Asian J. 2018, 13, 1089-1102. Notes (5) (a) Pitman, C. L.; Finster, O. N. L.; Miller, A. J. M., Cyclopentadiene- ≠ The first discarded mechanisms consisted in the spontaneous mediated hydride transfer from rhodium complexes Chem. Commun. 2016, 52, 9105-9108. (b) Quintana, L. M. A.; Johnson, S. I.; Corona, S. L.; Villatoro, W.; evolution of singlet state 1a wherein the Cp* first binds the phpy Goddard, W. A.; Takase, M. K.; VanderVelde, D. G.; Winkler, J. R.; Gray, H. B.; ligand before catching a hydrogen atom in a final step. This Blakemore, J. D., Proton–hydride tautomerism in hydrogen evolution catalysis Proc. mechanism could indeed proceed in three different modes (cf. Nat. Acad. Sci. 2016, 113, 6409-6414. (c) Peng, Y.; Ramos-Garcés, M. V.; Lionetti, Scheme 2 and Figure S123 of the Supporting Material): a) the D.; Blakemore, J. D., Structural and Electrochemical Consequences of [Cp*] Ligand Protonation Inorg. Chem. 2017, 56, 10824-10831. Cp* and the phpy condense concertedly, ii) the Cp* ligand is (6) Wu, F.; Deraedt, C.; Cornaton, Y.; Contreras-Garcia, J.; Boucher, M.; sequentially first attacked by the N center and then the Karmazin, L.; Bailly, C.; Djukic, J.-P., Making Base-Assisted C–H Bond Activation carbanionic C atom of the phpy ligand, or iii) the reverse, i.e the by Cp*Co(III) Effective: A Noncovalent Interaction-Inclusive Theoretical Insight and Experimental Validation Organometallics 2020, 39, 2609-2629. Cp* is first attacked by the carbanionic carbon and secondly by (7) (a) Ma, P.; Chen, H., Ligand-Dependent Multi-State Reactivity in Cobalt(III)- the N center of the phpy ligand. For each of these three scenarios Catalyzed C–H Activations ACS Catalysis 2019, 9, 1962-1972. (b) Ghorai, J.; applied to 1a, the first activation barrier exceeds + 35 kcal/mol Anbarasan, P., Developments in the Cp*Co(III) Catalyzed C-H Bond (ΔG= + 63 kcal/mol for the concerted binding, ΔG= + 62 Functionalizaitons Asian J. Org. Chem. 2019, 8, 430 -455. (c) Gandeepan, P.; Müller, T.; Zell, D.; Cera, G.; Warratz, S.; Ackermann, L., 3d Transition Metals for kcal/mol for the binding on N and ΔG= + 37 kcal/mol for the C–H Activation Chem. Rev. 2019, 119, 2192-2452. + binding on C). The abstraction of iodide from 1a by Ph3C was (8) Ustynyuk, Y. A.; Barinov, I. V., Nickelocene Interaction with o-halogenated also considered as the first step of the mechanism, but was even- azobenzenes. 4-Phenyl-4H-cyclopenta[c]cinnoline, a new pseudo azulene. J. Organomet. Chem. 1970, 23, 551-557. tually discarded due to its unfavorable thermochemistry (ΔG= + (9) Hull, J. F.; Balcells, D.; Blakemore, J. D.; Incarvito, C. D.; Eisenstein, O.; 22 kcal/mol). Brudvig, G. W.; Crabtree, R. H., Highly Active and Robust Cp* Iridium Complexes for Catalytic Water Oxidation J. Am. Chem. Soc. 2009, 131, 8730-8731. (10) (a) Frederic, P.; Joanna, W.-D.; Frank, G., Cp*Rh-Catalyzed C-H Activations. Versatile Dehydrogenative Cross-Couplings of Csp2 C—H Positions ACKNOWLEDGMENT with Olefins, Alkynes, and Arenes Aldrichim. Acta 2012, 45, 31-45. (b) Kuhl, N.; The CNRS and the University of Strasbourg are acknowledged for Hopkinson, M. N.; Wencel-Delord, J.; Glorius, F., Beyond Directing Groups: Transition-Metal-Catalyzed C-H Activation of Simple Arenes Angew. Chem., Int. their financial support. F.W. thanks the China Scholarship Coun- Ed. 2012, 51, 10236-10254. (c) Kuhl, N.; Schröder, N.; Glorius, F., Formal SN- cil for financial support. Y.C. would like to acknowledge the Type Reactions in Rhodium(III)-Catalyzed C-H Bond Activation Adv. Synth. Catal. GENCI-IDRIS for providing access to national HPC resources 2014, 356, 1443-1460. (11) Park-Gehrke, L. S.; Freudenthal, J.; Kaminsky, W.; DiPasquale, A. G.; and to the High Performance Computing Center of the University Mayer, J. M., Synthesis and oxidation of Cp*IrIII compounds: functionalization of a of Strasbourg for providing access to computing resources. We Cp* methyl group Dalton Trans. 2009, 1972-1983. gratefully thank Prof. Dr Eric Hénon (University of Reims) for (12) (a) [Online Early Access]. Published Online: 2020. edifying discussions on IGM and IBSI. https://echa.europa.eu/substance-information/-/substanceinfo/100.028.325. (b) Payne, L. R., The Hazards of Cobalt Occupational Medicine 1977, 27, 20-25. (c) . Leyssens, L.; Vinck, B.; Van Der Straeten, C.; Wuyts, F.; Maes, L., Cobalt toxicity in humans—A review of the potential sources and systemic health effects ABBREVIATIONS Toxicology 2017, 387, 43-56. (d) Mahey, S.; Kumar, R.; Sharma, M.; Kumar, V.; Bhardwaj, R., A critical review on toxicity of cobalt and its bioremediation strategies DCM, dichloromethane; DCE, 1,2-dichloroethane; ITC, isother- SN Applied Sciences 2020, 2, 1279. (e) Egorova, K. S.; Ananikov, V. P., Toxicity of mal titration calorimetry; DFT, density functional theory; Metal Compounds: Knowledge and Myths Organometallics 2017, 36, 4071-4090. (13) López-Resano, S.; Martínez de Salinas, S.; Garcés-Pineda, F. A.; Moneo- QTAIM, quantum theory of atoms in molecule; Fc, ferrocene. Corcuera, A.; Galán-Mascarós, J. R.; Maseras, F.; Pérez-Temprano, M. H., Redefining the Mechanistic Scenario of Carbon−Sulfur Nucleophilic Coupling via REFERENCES High-Valent Cp*CoIV Species Angew. Chem. Int. Ed. 2021, 60, 11217-11221. (14) Gensch, T.; Klauck, F. J. R.; Glorius, F., Cobalt-Catalyzed C−H Thiolation through Dehydrogenative Cross-Coupling Angew. Chem. Int. Ed. 2016, 55, 11287- (1) Jutzi, P., The Versatility of the Pentamethylcyclopentadienyl Ligand in Main- 11291. Group Chemistry Comments Inorg. Chem. 1987, 6, 123-144. (15) (a) Sanjose-Orduna, J.; Gallego, D.; Garcia-Roca, A.; Martin, E.; Benet- (2) (a) Deeming, A. J. In Comprehensive Organometallic Chemistry; Wilkinson, Buchholz, J.; Perez-Temprano, M. H., Capturing Elusive Cobaltacycle Intermediates: G., Stone, F. G. A., Abel, E. W., Eds.; Pergamon: Oxford, 1982, p 377-512. (b) A Real-Time Snapshot of the Cp*CoIII-Catalyzed Oxidative Alkyne Annulation Espinet, P.; Albéniz, A. C. In Comprehensive Organometallic Chemistry III; Mingos, Angew. Chem., Int. Ed. 2017, 56, 12137-12141. (b) Sanjose-Orduna, J.; Sarria Toro, D. M. P., Crabtree, R. H., Eds.; Elsevier: Oxford, 2007, p 315-403. (c) Thiel, I.; J. M.; Perez-Temprano, M. H., HFIP-Assisted C-H Functionalization by Cp*CoIII: Hapke, M., The broad diversity of CpCo(I) complexes Rev. Inorg. Chem. 2014, 34, Access to Key Reactive Cobaltacycles and Implication in Catalysis Angew. Chem., 217-245. (d) Piou, T.; Romanov-Michailidis, F.; Romanova-Michaelides, M.; Int. Ed. 2018, 57, 11369-11373. Jackson, K. E.; Semakul, N.; Taggart, T. D.; Newell, B. S.; Rithner, C. D.; Paton, R. 17

(16) Bader, R. F. W. Atoms in Molecules: A Quantum Theory; Clarendon: Development and Mechanistic Investigation ChemCatChem 2017, 9, 2009-2017. Oxford, 1990. (d) Corre, Y.; Rysak, V.; Nagyhazi, M.; Kalocsai, D.; Trivelli, X.; Djukic, J.-P.; (17) (a) Lefebvre, C.; Rubez, G.; Khartabil, H.; Boisson, J.-C.; Contreras-García, Agbossou-Niedercorn, F.; Michon, C., One-Pot Controlled Reduction of Conjugated J.; Hénon, E., Accurately extracting the signature of intermolecular interactions Amides by Sequential Double Hydrosilylation Catalyzed by an Iridium(III) present in the NCI plot of the reduced density gradient versus electron density Phys. Metallacycle Eur. J. Org. Chem. 2020, 2020, 6212-6220. Chem. Chem. Phys. 2017, 19, 17928-17936. (b) Lefebvre, C.; Khartabil, H.; (30) Robertson, K. N.; Bakshi, P. K.; Lantos, S. D.; Cameron, T. S.; Knop, O., - Boisson, J.-C.; Contreras-García, J.; Piquemal, J.-P.; Hénon, E., The Independent Crystal chemistry of tetraradial species. Part 9. The versatile BPh4 anion, or how Gradient Model: A New Approach for Probing Strong and Weak Interactions in organoammonium H(N) atoms compete for hydrogen bonding Can. J. Chem. 1998, Molecules from Wave Function Calculations ChemPhysChem 2018, 19, 724-735. 76, 583-611. (18) Klein, J.; Khartabil, H.; Boisson, J.-C.; Contreras-Garcia, J.; Piquemal, J.-P.; (31) Nesterov, D. S.; Nesterova, O. V.; Kopylovich, M. N.; Pombeiro, A. J. L., Hénon, E., New Way for Probing Bond Strength J. Phys. Chem. A 2020, 124, 1850- Pronounced retention of stereoconfiguration upon sp3 CH bonds hydroxylation of 1860. dimethylcyclohexanes and decahydronaphthalenes with m-CPBA oxidant and a Co- (19) Connelly, N. G.; Geiger, W. E., Chemical Redox Agents for Organometallic phthalocyanine catalyst Mol. Catal. 2018, 459, 8-15. Chemistry Chem. Rev. 1996, 96, 877−910. (32) McBride, J. M., The Hexaphenylethane Riddle Tetrahedron 1974, 30, (20) Denekamp, C.; Yaniv, M., Benzene Loss from Trityl Cations-A Mechanistic 2009-2022. Study J. Am. Soc. Mass Spectrom. 2006, 17, 730-736. (33) (a) O’Brien, Z. J.; Karlen, S. D.; Khan, S.; Garcia-Garibay, M. A., Solid- (21) (a) Schade, C.; Mayr, H., Tritylium ions as initiators and co-initiators in State Molecular Rotors with Perdeuterated Stators: Mechanistic Insights from cationic polymerizations Makromol. Chem., Rapid Commun. 1988, 9, 477-482. (b) Biphenylene Rotational Dynamics in Ordered and Disordered Crystal Forms J. Org. Leleu, J. M.; Bernardo, V.; Polton, A.; Tardi, M.; Sigwalt, P., About the various Chem. 2010, 75, 2482-2491. (b) Hinz, A.; Labbow, R.; Reiss, F.; Schulz, A.; types of transfer reactions in the cationic polymerization of isobutyl vinyl ether Sievert, K.; Villinger, A., Synthesis and structure of tritylium salts Struct. Chem. Polym. Int. 1995, 37, 219-230. (c) Ofial, A. R., Benzhydrylium and tritylium ions: 2015, 26, 1641-1650. complementary probes for examining ambident nucleophiles Pure Appl. Chem. (34) Straus, D. A.; Zhang, C.; Tilley, T. D., Trityl tetraphenylborate as a reagent 2015, 87, 341-351. (d) Huang, Y.; Qiu, C.; Li, Z.; Feng, W.; Gan, H.; Liu, J.; Guo, in organometallic chemistry J. Organomet. Chem. 1989, 369 C13-C17. K., Tritylium Cation as Low Loading Lewis Acidic Organocatalyst in Povarov (35) (a) Le Roy, R. J.; Sprague, E. D.; Williams, F., Quantum mechanical Reactions ACS Sustainable Chem. Eng. 2016, 4, 47-52. (e) Courant, T.; Lombard, tunneling in hydrogen atom abstraction for solid acetonitrile at 77-87.deg.K J. Phys. M.; Boyarskaya, D. V.; Neuville, L.; Masson, G., Tritylium assisted iodine catalysis Chem. 1972, 76, 546-551. (b) Cuccato, D.; Dossi, M.; Polino, D.; Cavallotti, C.; for the synthesis of unsymmetrical triarylmethanes Org. Biomol. Chem. 2020, 18, Moscatelli, D., Is Quantum Tunneling Relevant in Free-Radical Polymerization? 6502-6508. Macromol. Reac. Engin. 2012, 6, 496-506. (22) Horn, M.; Schappele, L. H.; Lang-Wittkowski, G.; Mayr, H.; Ofial, A. R., (36) (a) Boelsing, F.; Spanuth, E., Triphenylmethyl radicals by inverse Stevens Towards a Comprehensive Hydride Donor Ability Scale Chem. - Eur. J. 2013, 19, rearrangement Z. Naturforsch., B: Anorg. Chem., Org. Chem. 1976, 31B, 531-532. 249-263. (b) Fukuzumi, S.; Kitano, T.; Ishikawa, M., 10-Methylacridine dimer acting as a (23) Hayatifar, A.; Borrego, A.; Bosek, D.; Czarnecki, M.; Derocher, G.; unique two-electron donor in the one-electron reduction of triphenylmethyl cation J. Kuplicki, A.; Lytle, E.; Padilla, J.; Paroly, C.; Tubay, G.; Vyletel, J.; Weinert, C. S., Am. Chem. Soc. 1990, 112, 5631-5632. (c) Janzen, E. G.; Buchheit, M., Detection of Transition metal-free hydrodefluorination of acid fluorides and organofluorines by the triphenylmethyl radical by electron spin resonance in the thermal decomposition Ph3GeH promoted by catalytic [Ph3C][B(C6F5)4] Chem. Commun. 2019, 55, 10852- of sodium triphenylacetate J. Phys. Chem. 1972, 76, 937-939. (d) Kirste, B.; Harrer, 10855. W.; Kurreck, H., ENDOR studies of novel di- and triphenylmethyl radicals generated (24) (a) Bly, R. S.; Bly, R. K.; Hossain, M. M.; Silverman, G. S.; Wallace, E., from galvinols J. Am. Chem. Soc. 1985, 107, 20-28. (e) Ponomarchuk, M. P.; Reactions of alkyldicarbonyl-η5-cyclopentadienyliron compounds with tritylium Kasukhin, L. F.; Fileleeva, L. I., Triphenylmethane derivatives in single-electron tetrafluoroborate: a search for hydrogen-atom abstraction following one-electron- transfer reactions Zh. Obshch. Khim. 1972, 42, 2712-2715. (f) Schreiner, K.; transfer Tetrahedron 1986, 42, 1093-1108. (b) Diversi, P.; Iacoponi, S.; Ingrosso, Berndt, A.; Baer, F., Hyperconjugation in twisted π-radicals. IV. Spin density G.; Laschi, F.; Lucherini, A.; Pinzino, C.; Uccello-Barretta, G.; Zanello, P., Electron distribution and stereochemistry of triphenylmethyl radical in solution. Transfer Catalysis in the Activation of C-H Bonds by Iridium Complexes Reinvestigation of carbon-13 coupling constants Mol. Phys. 1973, 26, 929-939. (g) Organometallics 1995, 14, 3275-3287. (c) Casper, L. A.; Wursthorn, L.; Geppert, Trapp, C.; Kulkarni, S. V., Isotope effects in the electron spin resonance spectra of M.; Roser, P.; Linseis, M.; Drescher, M.; Winter, R. F., 4-Ferrocenylphenyl- deuterated triphenylmethyl radicals J. Phys. Chem. 1984, 88, 2703-2708. (h) Van + Substituted Tritylium Dyes with Open and Interlinked C Ar2 Entities: Redox der Hart, W. J., ESR spectra of triarylmethyl radicals Mol. Phys. 1970, 19, 75-84. Behavior, Electrochromism, and a Quantitative Study of the Dimerization of Their (37) Wang, B.; Lee, Y.-M.; Tcho, W.-Y.; Tussupbayev, S.; Kim, S.-T.; Kim, Y.; Neutral Radicals Organometallics 2020, 39, 3275-3289. (d) Uglova, E. V.; Seo, M. S.; Cho, K.-B.; Dede, Y.; Keegan, B. C.; Ogura, T.; Kim, S. H.; Ohta, T.; Makhaev, V. D.; Reutov, O. A., Mechanism of reduction of trityl perchlorate by Baik, M.-H.; Ray, K.; Shearer, J.; Nam, W., Synthesis and reactivity of a symmetric organomercury compounds Zh. Org. Khim. 1970, 6, 1337-1338. (e) mononuclear non-haem cobalt(IV)-oxo complex Nat. Commun. 2017, 8, 14839. Plesch, P. H.; Stasko, A.; Robson, D., Cathodic reduction of the trityl cation in (38) Zanders, D.; Bačić, G.; Leckie, D.; Odegbesan, O.; Rawson, J.; Masuda, J. methylene dichloride J. Chem. Soc. B 1971, 1634-1637. (f) El Murr, N.; Laviron, D.; Devi, A.; Barry, S. T., A Rare Low-Spin CoIV Bis(β-silyldiamide) with High E., Syntheses using electrochemically generated cobaltocene or cobaltocene anion Thermal Stability: Steric Enforcement of a Doublet Configuration Angew. Chem. Tetrahedron Lett. 1975, 875-878. (g) Baidak, A.; Naumov, S.; Brede, O., Kinetic Int. Ed. 2020, 59, 14138-14142. and Energetic Analysis of the Free Electron Transfer J. Phys. Chem. A 2008, 112, (39) Dane, L. M.; Janssen, L. J. J.; Hoogland, J. G., The iodine/iodide redox 10200-10209. couple at a platinum electrode Electrochim. Acta. 1968, 13, 507-518. (25) Bank, S.; Ehrlich, C. L.; Zubieta, J. A., Substituent effect on the (40) (a) Popov, A. I.; Geske, D. H., Studies on the Chemistry of Halogen and of electrochemical oxidation of trityl anions J. Org. Chem. 1979, 44, 1454-1458. Polyhalides. XVI. Voltammetry of Bromine and Interhalogen Species in Acetonitrile (26) (a) Azarani, A.; Berinstain, A. B.; Johnston, L. J.; Kazanis, S., Electron J. Am. Chem. Soc. 1958, 80, 5346-5349. (b) Baucke, F. G. K.; Bertram, R.; Cruse, transfer reactions between excited diarylmethyl and triarylmethyl carbocations and K., The iodide-iodine system in acetonitrile: Evaluation of standard thermodynamic aromatic donors J. Photochem. Photobiol., A 1991, 57, 175-189. (b) Chenneberg, data on the association I− + I2→ I3− from potentiometric measurements at 25 and L.; Leveque, C.; Corce, V.; Baralle, A.; Goddard, J.-P.; Ollivier, C.; Fensterbank, L., 50°C J. Electroanal. Chem. 1971, 32, 247-256. (c) Boschloo, G.; Hagfeldt, A., Single-Electron-Transfer Oxidation of Trifluoroborates and Silicates with Organic Characteristics of the Iodide/Triiodide Redox Mediator in Dye-Sensitized Solar Cells Reagents: A Comparative Study Synlett 2016, 27, 731-735. Acc. Chem. Res. 2009, 42, 1819-1826. (d) El-Hallag, I. S., Electrochemical (27) Mandon, D.; Toupet, L.; Astruc, D., Electron-transfer pathway in the oxidation of iodide at a glassy carbon electrode in methylene chloride at various hydride abstraction by tritylium from exo-substituted cyclohexadiene iron(0) temperatures J. Chil. Chem. Soc. 2010, 55, 67-73. complexes: a key step in the synthesis of heterobifunctional cyclohexadiene from (41) Huo, Z.; Yang, S.; Zang, D.; Farha, R.; Goldmann, M.; Xu, H.; Antoine, B.; benzene J. Am. Chem. Soc. 1986, 108, 1320-1322. Izzet, G.; Proust, A.; Ruhlmann, L., Photocurrent Generation from Visible Light (28) (a) Hamdaoui, M.; Ney, M.; Sarda, V.; Karmazin, L.; Bailly, C.; Sieffert, N.; Irradiation of Covalent Polyoxometalate-Porphyrin Copolymer Electrochim. Acta Dohm, S.; Hansen, A.; Grimme, S.; Djukic, J.-P., Evidence of a Donor-Acceptor (Ir- 2021, 368, 137635. H)→SiR3 Interaction in a Trapped Ir(III) Silane Catalytic Intermediate (42) (a) Zhang, M.-X.; Zuo, G.; Yang, X.; Liu, S. H., Synthesis and redox Organometallics 2016, 35, 2207-2223. (b) Hamdaoui, M.; Desrousseaux, C.; properties of cyclometallated iridium (III) complexes modified with arylamino Habbita, H.; Djukic, J.-P., Iridacycles as Catalysts for the Autotandem Conversion of groups J. Organomet. Chem. 2020, 930, 121580. (b) Lu, G.-Z.; Han, H.-B.; Li, Y.; Nitriles into Amines by Hydrosilylation: Experimental Investigation and Scope Zheng, Y.-X., Photoluminescence and electroluminescence of four platinum Organometallics 2017, 36, 4864-4882. complexes with trifluoromethyl-substituted 2-phenylpyridine and (29) (a) Corre, Y.; Werle, C.; Brelot-Karmazin, L.; Djukic, J.-P.; Agbossou- tetraphenylimidodiphosphinate ligands Dyes and Pigments 2017, 143, 33-41. Niedercorn, F.; Michon, C., Regioselective hydrosilylation of terminal alkynes using (43) (a) Contreras-Garcia, J.; Johnson, E. R.; Keinan, S.; Chaudret, R.; Piquemal, pentamethylcyclopentadienyl iridium(III) metallacycle catalysts J. Mol. Catal. A: J.-P.; Beratan, D. N.; Yang, W., NCIPLOT: A Program for Plotting Noncovalent Chem. 2016, 423, 256-263. (b) Corre, Y.; Rysak, V.; Trivelli, X.; Agbossou- Interaction Regions J. Chem. Theo. Comput. 2011, 7, 625-632. (b) Johnson, E.; Niedercorn, F.; Michon, C., A Versatile Iridium(III) Metallacycle Catalyst for the Keinan, S.; Mori-Sanchez; Contreras-Garcia, J.; Cohen, A. J.; Yang, W., Revealing Effective Hydrosilylation of Carbonyl and Carboxylic Acid Derivatives Eur. J. Org. Noncovalent Interactions J. Am. Chem. Soc. 2010, 132, 6498-6506. Chem. 2017, 2017, 4820-4826. (c) Corre, Y.; Trivelli, X.; Capet, F.; Djukic, J.-P.; (44) te Velde, G.; Bickelhaupt, F. M.; Baerends, E. J.; Fonseca Guerra, C.; van Agbossou-Niedercorn, F.; Michon, C., Efficient and Selective Hydrosilylation of Gisbergen, S. J. A.; Snijders, J. G.; Ziegler, T., Chemistry with ADF J. Comput. Secondary and Tertiary Amides Catalyzed by an Iridium(III) Metallacycle: Chem. 2001, 22, 931-967. 18

(45) Lenthe, E. v.; Ehlers, A.; Baerends, E.-J., Geometry optimizations in the (68) (a) Deng, L.; Ziegler, T., Reaction path following in mass-weighted internal zero order regular approximation for relativistic effects J. Chem. Phys. 1999, 110, corrdinates Int. J. Quantum Chem. 1990, 94, 5523-5527. (b) Deng, L.; Ziegler, T.; 8943-8953. Fan, L., A combined density functional and intrinsic reaction coordinate study on the (46) Perdew, J. P.; Burke, K.; Ernzerhof, M., Generalized Gradient ground state energy surface of H2CO J. Chem. Phys. 1993, 99, 3823. Approximation Made Simple Phys. Rev. Lett. 1996, 77, 3865-3868. (69) (a) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H., A consistent and accurate (47) Caldeweyher, E.; Ehlert, S.; Hansen, A.; Neugebauer, H.; Spicher, S.; ab initio parametrization of density functional dispersion correction (DFT-D) for the Bannwarth, C.; Grimme, S., A generally applicable atomic-charge dependent London 94 elements H-Pu J. Chem. Phys. 2010, 132, 154104. (b) Grimme, S.; Ehrlich, S.; dispersion correction J. Chem. Phys. 2019, 150, 154122. Goerigk, L., Effect of the damping function in dispersion corrected density (48) Pye, C. C.; Ziegler, T., An implementation of the conductor-like screening functional theory J. Comput. Chem. 2011, 32, 1456-1465. model of solvation within the Amsterdam density functional package Theor. Chem. (70) Weigend, F.; Ahlrichs, R., Balanced basis sets of split valence, triple zeta Acc. 1999, 101, 396-408. valence and quadruple zeta valence quality for H to Rn: Design and assessment of (49) Cornaton, Y.; Djukic, J.-P., A noncovalent interaction insight onto the accuracy Phys. Chem. Chem. Phys. 2005, 7, 3297-3305. concerted metallation deprotonation mechanism Phys. Chem. Chem. Phys. 2019, 21, (71) Barone, V.; Cossi, M., Quantum calculation of molecular energies and 20486-20498. energy gradients in solution by a conductor solvent model J. Phys. Chem. 1998, 102, (50) Yepes, D.; Munarriz, J.; Daniel, l. A.; Contreras-Garcia, J.; Jaque, P., Real- 1995-2001. Space Approach to the Reaction Force: Understanding the Origin of (72) Neese, F., The ORCA program system. WIREs Comput. Mol. Sci. 2012, 2, Synchronicity/Nonsynchronicity in Multibond Chemical Reactions J. Phys. Chem. A 73-78. 2020, 124, 1959-1972. (73) Bahr, S. R.; Boudjouk, P., Trityl Tetrakis(3,5- (51) (a) Werlé, C.; Karmazin, L.; Bailly, C.; Ricard, L.; Djukic, J.-P., Bis(Trifluoromethyl)Phenyl)Borate - a New Hydride Abstraction Reagent. J. Org. Stabilization of an Electron-Unsaturated Pd(I)–Pd(I) Unit by Double Hemichelation Chem. 1992, 57, 5545-5547. Organometallics 2015, 34, 3055-3064. (b) Werlé, C.; Bailly, C.; Karmazin-Brelot, (74) O'Brien, Z. J.; Karlen, S. D.; Khan, S.; Garcia-Garibay, M. A., Solid-State L.; Le Goff, X.-F.; Ricard, L.; Djukic, J.-P., Hemichelation, a Way To Stabilize Molecular Rotors with Perdeuterated Stators: Mechanistic Insights from Biphenylene Electron-Unsaturated Complexes: The Case of T-Shaped Pd and Pt Metallacycles J. Rotational Dynamics in Ordered and Disordered Crystal Forms J. Org. Chem. 2010, Am. Chem. Soc. 2013, 135, 17839-17852. 75, 2482-2491. (52) Khartabil, H.; Doudet, L.; Allart-Simon, I.; Ponce-Vargas, M.; Gérard, S.; (75) (a) Moran, M., Reactions of (η-C5H5)Co(CO)2 and (η-EtMe4C5)Co(CO)2 Hénon, E., Mechanistic insights into Smiles rearrangement. Focus on – stacking with iodine and cyanogen halides, XCN (X = Br or I) Z. Naturforsch., B: Anorg. interactions along the radical cascade Org. Biomol. Chem. 2020, 18, 6840-6848. Chem., Org. Chem. 1981, 36, 431-433. (b) Roe, D. M.; Maitlis, P. M., (53) Brookhart, M.; Green, M. L. H.; Parkin, G., Agostic interactions in Disproportionation of dihalogenocyclopentadienylcobalt complexes to transition metal compounds Proc. Nat. Acad. Sci. USA 2007, 104, 6908-6914. cobaltocenium salts J. Chem. Soc. A: Inorg. Phys. Theor. 1971, 3173-3175. (54) Iwakura, I.; Ikeno, T.; Yamada, T., A DFT Study on Hetero-Diels–Alder (76) (a) Knizhnikov, V. A.; Shirokii, V. L.; Sutormin, A. B.; Novikova, M. G.; Reactions Catalyzed by Cobalt Complexes: Lewis Acidity Enhancement as a Maier, N. A.; Ol'dekop, Y. A., Reactions of cyclopentadienyl(cyclopentadiene)cobalt Consequence of Spin Transition Caused by Lewis Base Coordination Angew. Chem. with iodine Metalloorg. Khim. 1989, 2, 1160-1162. (b) Shirokii, V. L.; Knizhnikov, Int. Ed. 2005, 44, 2524-2527. V. A.; Mosin, A. V.; Novikova, M. G.; Maier, N. A.; Ol'dekop, Y. A., (55) Liu, X.-G.; Li, Q.; Wang, H., (Pentamethylcyclopentadienyl)cobalt(III)- Electrochemical synthesis of (η5-cyclopentadienyl)cobalt-η4-tetraphenylbutadienes Catalyzed Direct Trifluoromethylthiolation of Arenes via C−H Activation Adv. and their interaction with iodine Metalloorg. Khim. 1991, 4, 1059-1063. (c) Pauson, Synth. Catal. 2017, 359, 1942-1946. P. L.; John Wiley & Sons, Ltd.: 2001, p 1-2. (56) Yoshino, T.; Matsunaga, S., Cp*CoIII-Catalyzed C–H Functionalization and (77) Ikemoto, H.; Yoshino, T.; Sakata, K.; Matsunaga, S.; Kanai, M., Asymmetric Reactions Using External Chiral Sources Synlett 2019, 30, 1384-1400. Pyrroloindolone synthesis via a Cp*Co(III)-catalyzed redox-neutral directed C-H (57) (a) Thompson, M. C.; Busch, D. H., Reactions of Coordinated Ligands. VI. alkenylation/annulation sequence J Am Chem Soc 2014, 136, 5424-5431. Metal Ion Control in the Synthesis of Planar Nickel(II) Complexes of α-Diketo-bis- mercaptoimines J. Am. Chem. Soc. 1964, 86, 213-217. (b) Costes, J.-P.; Chiboub Fellah, F. Z.; Dahan, F.; Duhayon, C., Role of the kinetic template effect in the syntheses of non symmetric Schiff base complexes Polyhedron 2013, 52, 1065- 1072. (c) Dötz, K. H.; Stendel, J., Fischer Carbene Complexes in Organic Synthesis: Metal-Assisted and Metal-Templated Reactions Chem. Rev. 2009, 109, 3227-3274. (d) Dötz, K. H.; Wenzel, B.; Jahr, H. C. In Templates in Chemistry I: -/-; Schalley, C. A., Vögtle, F., Dötz, K. H., Eds.; Springer Berlin Heidelberg: Berlin, Heidelberg, 2004, p 63-103. (58) Sucunza, D.; Cuadro, A. M.; Alvarez-Builla, J.; Vaquero, J. J., Recent Advances in the Synthesis of Azonia Aromatic Heterocycles J. Org. Chem. 2016, 81, 10126-10135. (59) Sun, B.; Yoshino, T.; Matsunaga, S.; Kanai, M., Air-Stable Carbonyl(pentamethylcyclopentadienyl)cobalt Diiodide Complex as a Precursor for Cationic (Pentamethylcyclopentadienyl)cobalt(III) Catalysis: Application for Directed C-2 Selective C-H Amidation of Indoles Adv. Synth. Catal. 2014, 356, 1491-1495. (60) M86-EXX229V1 APEX3 User Manual; Bruker AXS Inc.: Madison, USA, 2016. (61) Sheldrick, G. M., SHELXT - Integrated space-group and crystal-structure determination Acta Cryst. 2015, A71, 3-8. (62) Sheldrick, G. M., Crystal structure refinement with SHELXL Acta Cryst. 2015, C71, 3-8. (63) Stoll, S.; Schweiger, A., EasySpin, a comprehensive software package for spectral simulation and analysis in EPR J. Magn. Res. 2006, 178, 42-55. (64) (a) vanLenthe, E.; Baerends, E. J.; Snijers, J. G., Relativistic regular two- component Hamiltonians J. Chem. Phys. 1993, 99, 4597-4610. (b) vanLenthe, E.; Baerends, E. J.; Snijders, J. G., Relativistic total energy using regular approximations J. Chem. Phys. 1994, 101, 9783-9792. (c) vanLenthe, E.; Ehlers, A.; Baerends, E. J., Geometry optimizations in the zero order regular approximation for relativistic effects J. Chem. Phys. 1999, 110, 8943-8953. (65) (a) Klamt, A.; Schüürmann, G., COSMO: a new approach to dielectric screening in solvent with explicit expressions for the screening energy and its gradient. J. Chem. Soc., Perkin Trans.2 1993, 799-805. (b) Klamt, A., Conductor- like Screening Model for real solvents: a new approach to the quantitative calculation of solvation phenomena. J. Phys. Chem. 1995, 99, 2224-2235. (c) Klamt, A.; Jonas, V., Treatment of the outlying charge in continuum solvation models. J. Chem. Phys. 1996, 105, 9972-9981. (66) Franchini, M.; Philipsen, P. H. T.; Visscher, L., The Becke fuzzy cells integration scheme in the Amsterdam Density Functional program suite. J. Comput. Chem. 2013, 34, 1819-1827. (67) Henkelmann, G.; Uberuaga, B. P.; Jonsson, H., A climbing image nudged elastic band method for finding saddle points and minimum energy paths J. Chem. Phys. 2000, 113, 9901-9904. 19

The one electron oxidation of Cp*Co(III) metallacycles of 2-phenylpyridine derivatives by iodine, a silver (I) salt or a tritylium cation leads to the irreversible cyclocondensation of the organic ligands to afford cationic alkaloids. Under optimal conditions in either CH2Cl2, ClCH2CH2Cl or PhCl, the reaction produces the cationic aromatic alkaloids in up to 88% yield; it proceeds via a reactive Co(IV) intermediate and the capture of a H atom.