Fluoro Nitrenoid Complexes FN=MF2 (M=Co, Rh, Ir): Electronic Structure Dichotomy and Formation of Nitrido Fluorides N≡MF3

Abstract The fluoronitrenoid metal complexes FNCoF2 and FNRhF2 as well as the first ternary RhVI and IrVI complexes NIrF3 and NRhF3 are described. They were obtained by the reaction of excited Group‐9 metal atoms with NF3 and their IR spectra, isolated in solid rare gases (neon and argon), were recorded. Aided by the observed 14/15N isotope shifts and quantum‐chemical predictions, all four stretching fundamentals of the novel complexes were safely assigned. The F−N stretching frequencies of the fluoronitrenoid complexes FNCoF2 (1056.8 cm−1) and FNRhF2 (872.6 cm−1) are very different and their N−M bonds vary greatly. In FNCoF2, the FN ligand is singly bonded to Co and bears considerable iminyl/nitrene radical character, while the N−Rh bond in FNRhF2 is a strong double bond with comparatively strong σ‐ and π‐bonds. The anticipated rearrangement of FNCoF2 to the nitrido CoVI complex is predicted to be endothermic and was not observed.


Introduction
Fluoronitrenoid metal complexes are underexplored compounds as only two examples have been reported so far, FNReF 5 and FNCuF 2 . [2] This is most likely due to the fact that they bear areactive FN function and are not readily available. In general, late transition-metal-nitrogenmultiple bonds have attracted particular interest after having been found to enable the conversion of ubiquitous C À Ha nd C À Cb onds into valuable CÀNbonds as either catalysts or intermediates. [3,[4][5][6] Detailed knowledge of the geometry and electronic structure of such compounds are vital to elucidate the nature and mechanism of these reactions. [4] While nitrido complexes usually feature aM Ntriple bond, [4,7] theimido ligand NR 2À exhibits aM = Ndouble bond in complexes with abent M = N À Rlinkage (Scheme 1). [4] However,ithas been mentioned that the energy required to change the M=NÀRangle from bent to linearity is often small, and indeed, transition metal imido complexes bearing sterically encumbered ligands to protect the reactive metal-nitrogen multiple bond often feature short M À Nbond lengths and nearly linear angles about the M À N À Rl inkage (Scheme 1). [4,5] In this bonding Scheme the anionic NR 2À imido ligand is expected to bear anucleophilic character in its reactions.The polarities of both the s and p bonds are important factors that govern the reactivity of imido complexes. [4,5] As one moves from early to late transition metals,the binding energy of the metal d-electrons increases,and an imido ligand becomes less nucleophilic.T his applies in particular to late transition metals in higher oxidation states.Infact, the imido ligands in late 3d transition metal complexes are often so electrophilic that these complexes can be better described as metal-nitrene complexes (Scheme 1). Formally,a ni mido complex differs from anitrene complex in the formal charge of the ligand and thus in the oxidation state of the metal (Scheme 1). An imido species suggests ad ianionic imido ligand (NR 2À ), whereas aneutral nitrene ligand is usually the result of apredominantly covalent nitrogen-metal bond. In the case of neutral, covalently bound ligands,n op olarization of bonding electrons towards the ligand and no or only little charge transfer from the metal to the ligand is generally to be expected. Additionally,n ot long-ago af ew examples were reported where an iminyl radical ( 2 NRC)i sc oordinated to at ransition metal. [8] Metal-imido cores containing iminyl radical ligands are proposed as reactive intermediates in an umber of metal Scheme 1. Simplified scheme [1] for the interactions of ametal center and an anionic imido (formal NR 2À ,t op) and an eutral nitrene (bottom) ligand, respectively, in linear and bent nitrenoid complexes. The charge of the NR ligand in the ionic (NR 2À )and neutral (NR) approximation and the corresponding formal metal oxidation states are indicated below for uncharged donor ligands L. catalyzed aziridination and amination reactions. [5] These and further possible interpretations of the electronic structures of terminal metal nitrenoid complexes (e.g.,i mido,M (RN 2À ); iminyl, M( 2 RNC); nitrene,M (RN);a nd triplet nitrene ( 3 RNCC) [9] demonstrate the diversity of the metal-nitrogen bond in MÀNÀRcomplexes.
We became interested in fluoronitrenoid complexes of the group 9m etal difluorides,F ÀN=MF 2 ,M= Co,R h, Ir. These simple nitrene complexes should allow arigorous experimental and quantum-chemical comparison of the electronic properties of the Co IV complex with those of its heavier congeners.F luoronitrenoid-metal complexes show more complex nitrenoid-metal binding modes and new reactivities. Thef luoronitrene ligand shares some similarities with the oxygen molecule,s ince the nitrogen 2p electrons involved in metal-nitrogen bonding are accommodated in ad egenerate pair of p*(F À N) orbitals.H ence any metal-to-ligand charge transfer in af luoronitrene complex will increase the occupancy of these p*(FÀN) orbital, rendering the NÀFstretching frequency ah ighly sensitive probe for the polarity and the strength of the N=Mb ond:aweak nitrogen-metal bond in am etal-nitrene complex result in as trong fluorine-nitrogen bond and vice versa. Utilizing the F À Nf unctionality and relying upon ah igh metal-fluorine bond energy we have targeted the synthesis of high-valent nitridometal trifluorides NMF 3 ,s tarting from the fluoronitrene complexes by an oxidative FÀNt oM ÀFf luorine migration, by which the formal metal oxidation state will be increased by two units.As far as we know,m olecular nitridometal trifluorides,N MF 3 , are known only for the early transition metals of group IV (M = Ti,Zr, Hf [10] )and VI (Cr,Mo, W [11] ). Notably,the formal metal oxidation state VI in NMF 3 (M = Co,R h, Ir) is rare, with IrO 3 ,I r(h 2 -O 2 )O 2 ,I rF 6 ,R h(h 2 -O 2 )O 2 and RhF 6 as the only examples. [12] Te rminal nitrido complexes of very high formal oxidation states have been predicted very recently, [13] however, such high-valent group 9m etals are still unknown. They would be of particular interest for cobalt, since the highest oxidation state reported for any molecular complex of cobalt is V, for example,the well-known [Co(1-norbornyl) 4 ] + or in the tricoordinated cationic bis(nitrene) cobalt complex [(IMes)Co(NDipp) 2 ] + . [6,14] Thelatter low-coordinated cationic complex is supported by the strongly electron-donating and sterically demanding N-heterocyclic carbene ligand IMes,and has been obtained by oxidation of the corresponding neutral bis(nitrene) Co IV complex. Interestingly,t heoretical calculations indicated that the frontier molecular orbitals of these bis(nitrene) complexes have near-equal contributions from both the cobalt center and the nitrene ligand orbitals, indicating that the spectroscopic oxidation states for these cobalt centers are likely to be lower than IV and V, respectively.A part from these bis(nitrene) complexes,t he majority of the known cobalt nitrenoid complexes have low spin Co III centers which are supported, for example,bybulky ancillary tripodal or bidentate ligands to achieve kinetic stabilization. [15] To the contrary,terminal nitrido complexes of cobalt still remain elusive. [16]

Results and Discussion
To obtain the group 9m etal difluorides,F À N = MF 2 (M = Co,R h, Ir), we have studied the gas-phase reaction of the laser-ablated free metal atoms with NF 3 seeded in a1 :1000 excess of neon or argon. Ther eaction products were deposited on ag old-plated copper mirror cooled to 5a nd 12 Kand IR-spectroscopically investigated (for experimental details see the Supporting Information). According to preliminary calculations at the DFT-B3LYP and BP86 levels of theory the direct insertion of the excited metal atoms into the FÀNb ond of NF 3 to F 2 NÀMF,a nd the subsequent fluorine migration from nitrogen to the metal center to yield the desired FN=MF 2 is highly exothermic for all three metals ( Figure 1, Table S1). However, the expected rearrangement of the fluoronitrene to ah igh-valent nitrido trifluoride N MF 3 is found to be endothermic for the cobalt complex, rendering FN=CoF 2 the most stable CoF 3 Ni somer.T ot he contrary,t his rearrangement is slightly exothermic for the rhodium nitrene complex (DH 0 = À12 kJ mol À1 ,C CSD(T)), and becomes strongly exothermic for the iridium congener (DH 0 = À98 kJ mol À1 ,C CSD(T), Table S1), which rendered the detection of the iridium nitrene complex difficult if not impossible.
These predictions were fully supported by the analysis of the experimental IR spectra of the deposits in solid neon shown for cobalt ( Figure 2), rhodium ( Figure 3) and iridium ( Figure 4). Complementary argon spectra for the experiments using rhodium and iridium have also been recorded and shown in the supporting information, Figures S1-S3. These spectra are dominated by strong bands of the NF 3 precursor ( Figure S4, Table S2 and Ref. [17]) and its plasma radiation induced decomposition products NF and NF 2 . [18] However, the assignment of IR bands associated with the targeted nitrene and nitrido complexes is facilitated by acharacteristic 14/15 Ni sotope shift exhibited by all modes in which the nitrogen atom is significantly involved. These isotope shifts are indicated in the experimental spectra shown in the Figures 2-4. They were obtained in experiments using 15 NF 3 , which was synthesized from 15 N 2 and F 2 mixtures in an electric  Table S1 for more details.

Angewandte Chemie
Research Articles discharge. [19] While ad etailed report about the spectral assignment is given in the Supporting Information, it should be mentioned here,t hat bands due to binary metal fluorides MF n also appeared in these spectra, however these were safely assigned in nitrogen-free experiments,i nw hich NF 3 was replaced by elemental fluorine.I nt hese experiments none of the bands assigned to an itrogen-containing species appeared. Furthermore,bycomparing spectra of experiments using different group 9m etals,t he desired metal dependent bands were identified. Al ist of all observed IR bands associated with the target compound is shown in Table 1 together with their approximate assignment and supporting predictions from quantum-chemical calculations.
In the experiment using laser-ablated Co atoms and NF 3 four IR bands were obtained that displayed acharacteristic 14/ 15 Ni sotope shift (labeled A-D in Figure 2) and their assignment to the targeted fluoronitrene complex FNCoF 2 is well supported by prediction on the CASPT2/cc-pVTZ-DK level of theory (Table 1). Thev alues obtained using single reference correlation methods did either not converge (CCSD(T)), or did not yield qualitatively consistent results (B3LYP and BP86). Bands associated to the desired nitrido complex NCoF 3 were not detected, which is consistent with the significant higher energy of this isomer.O nt he other side, in the spectra obtained from laser-ablated Ir atoms and NF 3 our search for bands due to FNIrF 2 was unsuccessful and only the nitrido complex NIrF 3 was formed. Again, this reflects the lower stability of the former species,w hich exothermically rearranged to the lowest energy isomer.Here,too,four bands were assigned to NIrF 3 (marked with A' '-D' ' in Figure 4), of which only two revealed a 14/15 Ni sotope shift. Quantumchemical calculations (Table 1) performed at the DFT (BP86, B3LYP) and CCSD(T) levels of theory fully support these assignments.T he IrF 3 stretching modes are split into three components due to afirst order Jahn-Teller distortion for the anticipated 5d 3 configuration, which reduced the full C 3v point group symmetry to C s symmetry (for structures see Figures 5,  S5, and Table S3). All three IrÀFs tretching modes were observed, but only the band associated with the IrÀF' bond, which resides in the mirror plane together with the NÀIr bond, show asmall 14/15 Ni sotope shift.
Thea ssignment of the IR spectra obtained after codepositing evaporated rhodium and diluted NF 3 (Figure 3) was more puzzling. In these experiments both the anticipated compounds are finally detected in the solid matrices and for each all four stretching bands were successfully assigned ( Table 1). As described above for the corresponding Iridium compound also for NRhF 3 a 14/15 Nisotope shift was observed for the N À Rh and the Rh À F' stretching modes (Figure 3, A' ' and D' ',r espectively). Our CCSD(T) calculations for this species yield two imaginary frequencies which likely are caused by ac lose-lying excited electronic state which interferes with the calculation of displaced steps during the   numerical hessian calculation where the symmetry is lowered to C 1 .However,the structure obtained at the B3LYP level of theory is close to the one obtained at the CCSD(T) level (Figures 5a nd S5) and the good agreement of the B3LYP results for NIrF 3 with the experimental frequencies suggest ag ood performance also for the NRhF 3 species.T he 14/15 N isotope shift observed for the nitrene complex FNRhF 2 is well distributed between the F À Nand the N À Rh stretching bands (Figure 3, bands A and B,r espectively) indicating as trong vibrational coupling between these two modes.Analyzing the NÀMa nd FÀNs tretching frequencies of FNCoF 2 and FNRhF 2 we found surprisingly large differences in the bonding of the fluoronitrene ligand. In general, the two singly occupied anti bonding p*(FÀN)-orbitals of the FN ligand form a s and a p bond to these metal centers,a sd epicted in the qualitative molecular orbital (MO) interaction diagram shown in Figure 6. TheF À Nm ode of FNCoF 2 (1056.8 cm À1 ) appeared red-shifted by 62.6 cm À1 from the absorption of free,n eutral FN (1119.4 cm À1 ), [11] indicating the presence of an almost neutral nitrene ligand, while the corresponding mode of FNRhF 2 (872.6 cm À1 )ismuch stronger red-shifted by 246.8 cm À1 .
On the other side,t he low NÀCo stretching frequency of FNCoF 2 (586.1 cm À1 )i sm ost likely associated with aN À Co single bond, while the N À Rh frequencyo fF NRhF 2 Table 1: Comparison of infrared band positions (cm À1 )a nd isotopic shifts (cm À1 ,inparenthesis) observed in solid neon and argon with calculated values [intensities in km mol À1 in brackets] and their assignment in terms of an approximate descriptiono fthe vibrational modes.    Table 2, and the calculated NÀMb ond lengths (M = Co: 177 pm, Rh:174 pm, Figure 5). To shed light on these striking bonding differences non-dynamical electron-correlation effects were taken into account. Ther esults of CASSCF calculations revealed that the leading configuration (s 2 p 2 d 1 p* 0 s* 0 )associated with the qualitative MO Scheme shown in Figure 6c ontributes only 48 %t ot he 2 A'' ground state of FNCoF 2 ,f ollowed by states with significant weights which contain single and double p!p*e xcitations (Table S4). For FNRhF 2 am uch smaller extend of non-dynamic correlation was determined, since the most dominant configuration as depicted in Figure 6c ontributes to 84 %t oi ts 2 A' electronic ground state.A saconsequence of these correlation effects significant higher s*a nd p*p opulations (0.37 and 0.61, Figure S6) were found for FNCoF 2 compared to FNRhF 2 (s*: 0.11, p*: 0.18). Thee ffective bond orders (EBO) [20] derived from the natural orbitals obtained at the CASSCF level are 1.1 for the Co À Na nd 1.7 for the Rh = Nb ond. We also note ac onsiderable amount of minority spin population at the N atom (À0.46) in FNCoF 2 (Table 2a nd Figure S7) antiferromagnetically coupled to the majority spin at the Co center (1.46). Thelatter spin density can mainly be attributed to the singly occupied nonbonding d-MO of a'' symmetry ( Figures 6  and S6). Taking these effects into account, the FN unit in FNCoF 2 has at least aconsiderable fraction of iminyl/nitrene radical character,w hich explains the shortened single bond.
Thed isparities in the metal-nitrogen bonds of these nitrene complexes can likely be attributed to the peculiarity of bonding of the strongly correlated first-row transition metalligand bonds.E specially the close internuclear distance required for an optimum orbital overlap for p bonding is likely hindered due to Pauli repulsion of the Co 3s,3p coreshell and the nitrogen ligand orbitals. [21] Forthe nitrido complexes NRhF 3 and NIrF 3 the computed bond length (162 pm (IrN), 159 (RhN);f or comparison: triple-bond additive covalent radii:160 pm (IrN) and 160 pm (RhN), [22] Figure 5) and the experimental stretching frequencies (Table 1) indicate strong N Mt riple bonds.A na nalysis of the CASSCF(9,8) natural molecular orbitals ( Figure S8) reveals EBOs of 2.7 and 2.8 for NRh and NIr, respectively. Consistent with the assignment of oxidation state + VI for both metal centers,ad 3 configuration and aJ ahn-Teller distorted 2 A' electronic ground state was determined for both species.The N À Ir stretching frequency in neon of 1150.4 cm À1 is moderately higher than that observed for diatomic IrN embedded in solid neon (1111.1 cm À1 ). [23] However,incase of rhodium asignificant blue-shift of the NÀRh stretching mode (1112.6 cm À1 )o f2 21.2 cm À1 occurred compared to diatomic RhN embedded in argon (891.4 cm À1 ). [24] Theincreased force constant of 920 Nm À1 in NRhF 3 from 580 Nm À1 in RhN indicates as ignificant strengthening of the nitrogen-metal bond induced by the fluorine ligands.T his fluorine effect is less pronounced for the already strong triple bond in IrN, where force constants increase from 950 to 1020 Nm À1 for IrN and NIrF 3 ,respectively.

Conclusion
In summary,w ed escribed the exothermic formation of the fluoronitrenoid complexes FNCoF 2 and FNRhF 2 and of the nitrido complexes NRhF 3 and NIrF 3 by the reaction of the free group 9m etal atoms with NF 3 .T he IR spectra of these compounds isolated in solid rare gases (neon and argon) were recorded and, aided by the observed 14/15 Ni sotope shifts and quantum-chemical predictions,a ll four stretching fundamentals of these complexes were safely assigned. Neither nitrido nor other ternary complexes of Rh VI and Ir VI have yet been reported. Thea nticipated rearrangement of FNCoF 2 to the nitrido Co VI complex was not observed, because this reaction is endotherm. Thecovalently bound FN ligand in these highvalent metal complexes is almost neutrally charged, and the formal picture of an FN 2À ligand bound to aMF 2 fragment is ar ather coarse approximation for these fluoronitrenoid complexes.However, the bonding of the FN ligand in FNCoF 2 and FNRhF 2 was found to be strikingly different. In FNCoF 2 the FN ligand is singly bonded to Co and bears considerable iminyl/nitrene radical character,while the N=Rh double bond in FNRhF 2 shows comparatively strong s-a nd p-bonds.T he stretched N-Co bond and the poor overlap especially between the ligand p-and the metal 3d-orbitals can likely be attributed to arepulsion between the ligand orbitals and the outermost core 3s,3p shell of cobalt.