On the mechanism of imine elimination from Fischer tungsten carbene complexes

(Aminoferrocenyl)(ferrocenyl)carbene(pentacarbonyl)tungsten(0) (CO)5W=C(NHFc)Fc (W(CO)5(E-2)) is synthesized by nucleophilic substitution of the ethoxy group of (CO)5W=C(OEt)Fc (M(CO)5(1Et)) by ferrocenyl amide Fc-NH– (Fc = ferrocenyl). W(CO)5(E-2) thermally and photochemically eliminates bulky E-1,2-diferrocenylimine (E-3) via a formal 1,2-H shift from the N to the carbene C atom. Kinetic and mechanistic studies to the formation of imine E-3 are performed by NMR, IR and UV–vis spectroscopy and liquid injection field desorption ionization (LIFDI) mass spectrometry as well as by trapping experiments for low-coordinate tungsten complexes with triphenylphosphane. W(CO)5(E-2) decays thermally in a first-order rate-law with a Gibbs free energy of activation of ΔG‡298K = 112 kJ mol−1. Three proposed mechanistic pathways are taken into account and supported by detailed (time-dependent) densitiy functional theory [(TD)-DFT] calculations. The preferred pathway is initiated by an irreversible CO dissociation, followed by an oxidative addition/pseudorotation/reductive elimination pathway with short-lived, elusive seven-coordinate hydrido tungsten(II) intermediates cis(N,H)-W(CO)4(H)(Z-15) and cis(C,H)-W(CO)4(H)(Z-15).

In principle, the formation of imines from NH carbene complexes can occur by three conceivable fundamental pathways. The first pathway starts with the dissociation of the carbene followed by a 1,2-H shift at the free carbene (elimination-migration). The second one operates via a hydrogen atom shift at the coordinated carbene followed by dissociation of the resulting imine (migration-elimination). A third conceivable pathway could start with CO loss, followed by H atom migration. To the best of our knowledge, the mechanism of the imine formation from NH carbene complexes is not yet established.
In the absence of a base, the bulky diferrocenylcarbene complex Cr(CO) 5 (E-2) is stable even in refluxing toluene and hence, a simple migration-elimination or elimination-migration reaction is not anticipated in this case. We report here the heavier tungsten analogue W(CO) 5 (E-2) which is thermally reactive and smoothly forms the imine E-3 without the need of prior deprotonation. This apparently simpler reaction allows the investigation of the mechanism of imine formation from NH carbene complexes.
Herein, the synthesis and characterization of W(CO) 5 (E-2) followed by detailed mechanistic studies regarding the formation of imine E-3 are presented including mass spectrometric, NMR, IR and UV-vis spectroscopic kinetic studies in combination with (TD)-DFT methods.

Results and Discussion
Synthesis of W(CO) 5

Characterization of W(CO) 5 (E-2)
The composition and purity of W(CO) 5 (E-2) is ascertained by mass spectrometry, showing the expected molecular ion peak at m/z = 721 with appropriate isotopic pattern, and elemental analysis (Experimental section and Supporting Information File 1). At increasing temperature in the FD mass spectrometer, peaks at m/z = 397 appear which can be assigned to a molecular ion of the composition C 21 H 19 NFe 2 . A tiny peak cluster at m/z = 693, assignable to the loss of CO from W(CO) 5 (E-2), and peaks at higher m/z ratios, assignable to tungsten clusters, are also observed when traces of oxygen/water were present. Using 1 H and 13   found in a similar region as for other (pentacarbonyl)tungsten complexes W(CO) 5 (E-14 R ) with the α-ferrocenyl NH carbene ligand :C(NHR)Fc E-14 R (R = Me, Et, n-Pr [23], n-Bu [25], n-Pent [21]). Due to additional ring-current effects and nonclassical NH···Fe hydrogen bonding [54][55][56][57][58][59] of the NH-Fc moiety, the resonance for the amine proton NH 6 (δ = 10.50 ppm in CD 2 Cl 2 ) is shifted to lower field as compared to that of alkylamine substituted NH carbene complexes M(CO) 5 (E-14 R ) (δ = 9.00-9.11 ppm in CDCl 3 ) [21,23]. The NH···Fe interaction is also supported by the low-energy NH stretching vibration of W(CO) 5 (E-2) at 3240 cm −1 in CD 2 Cl 2 , which matches to that of Cr(CO) 5 (E-2) (3233 cm -1 ) [27] (Experimental section and Supporting Information File 1). A weak absorption band at 3439 cm −1 is tentatively assigned to some W(CO) 5 (Z-2) isomer lacking the NH···Fe interaction. In the solid state (KBr) the NH stretching vibration appears at 3335 cm −1 (Experimental section and Supporting Information File 1). The C-N-H bending vibration is observed as a single sharp relatively strong band at 1508 cm −1 . These IR data reveal that the main isomer in solution as well in the solid state is the E isomer in accordance with the IR data of W(CO) 5 (E/Z-8) [46]. The carbonyl region of IR spectra of W(CO) 5 (E-2) are in accordance with those of Cr(CO) 5 (E-2) [27] and related amino(ferrocenyl)carbene(pentacarbonyl)tungsten complexes W(CO) 5 (E-14 R ) (R = Me, Et, n-Pr [23], n-Bu [25], n-Pent [21]). The UV-vis spectrum of W(CO) 5 (E-2) (Supporting Information File 1) is similar to that of Cr(CO) 5 (E-2) [27] and to those of carbene(pentacarbonyl)metal complexes (Cr, W) [60,61].
Thermolysis of W(CO) 5 (E-2) in refluxing toluene gives imine E-3 [43] after ca. 24 h in almost quantitative yield, as monitored by 1 H NMR spectroscopy. Accordingly, W(CO) 5 (E-2) is a suitable candidate to investigate the imine formation from NH carbene complexes in a simple one-component system under relatively mild conditions and, importantly, in the absence of a base.
DFT studies on the formation of imine E-3 from W(CO) 5

(E-2)
Three conceivable reaction pathways for the formation of imine E-3 have been considered. For each pathway, density functional theory (DFT) calculations on the B3LYP/LANL2DZ (IEF-PCM toluene) level of theory have been performed to localize minimum structures and energies of the intermediates which are connected by transition states. The Gibbs free energies are reported at 298 K.
The overall Gibbs free energy of activation amounts to only ΔG ‡ total = 183 kJ mol −1 with the RDI W(CO) 4
All overall Gibbs free energies of activation for the discussed pathways 1a/1b and 3a/3b in the coordination sphere of the metal center are higher than for the carbene → imine isomerization in the metal-free systems E-2 → E-3 and Z-2 → Z-3 (pathways 2a/2b). This suggests that W(CO) 5 or W(CO) 4 coordination to E-2 or Z-2 kinetically stabilizes the carbene ligand. All pathways 1a/1b, 2a/2b and 3a/3b have large overall Gibbs free energies of activation with ΔG ‡ total > 200 kJ mol −1 . The alternative pathway 3c via CO dissociation, oxidative addition, pseudorotation and reductive elimination features the lowest overall Gibbs free energy of activation of ΔG ‡ total = 183 kJ mol −1 . Although activation barriers for CO and carbene ligand dissociation steps could not be determined by DFT calculations, the formation of tetracarbonyl complexes is very probable, while the carbene dissociation is less likely. The experimentally determined barrier for CO dissociation from tungsten hexacarbonyl amounts to 193 kJ mol −1 [74]. According to the calculations, the W(CO) 4 (Z-2) isomer is accessible from the thermodynamically preferred W(CO) 4 (E-2) isomer. The following oxidative addition pathway 3c from W(CO) 4  tively ( Figure 2). These hydrido intermediates act as hydrogen atom shuttle from the nitrogen to the carbon atom in NH carbene tetracarbonyl tungsten complexes. Oxidative additions of XY bonds to low-coordinate W(CO) n fragments is a common reactivity pattern for tungsten carbonyl complexes [75][76][77][78][79][80] and appears to be operative in the present case as well.
Similarly, photochemical activation (400 nm LEDs) in toluene produces imine E-3 in 31% yield after 120 h from W(CO) 5 (E-2) already at room temperature, while only 1% E-3 is formed in the dark at room temperature. This observation additionally supports the hypothesis that the key initial step is the dissociation of CO from W(CO) 5 (E-2) to give W(CO) 4 (E-2) (Scheme 4).
Attempts to observe the tetracarbonyl intermediates in the absence of PPh 3 by LIFDI mass spectrometry were unsuccessful. The mass spectra recorded at several time intervals during the heating procedure (reflux in toluene under strictly inert conditions) display the peak of the starting material at m/z = 721 and the peak of the imine product E-3 at m/z = 397. The former peak decreases while the latter one increases during the heating process (Supporting Information File 1, Figure S26). No other intermediates appear in the FD mass spectra. Tentatively, the concomitantly formed tungsten species aggregate under these conditions and form the observed dark precipitate. This further supports the hypothesis that no intermediates accumulate during the reaction and that the rate-determining step is the CO ligand dissociation.
In full accordance with the above observations, UV-vis spectra recorded during the thermal treatment of W(CO) 5 Figure S28). The final UV-vis spectrum after 6 h closely resembles that of the calculated TD-DFT spectrum of imine E-3 (Supporting Information File 1, Figure S29). All spectroscopic and analytical data suggest that the imine formation is faster than the CO dissociation.

Conclusion
The thermally induced formation of E-1,2-diferrocenylimine E-3 from the NH carbene pentacarbonyl tungsten complex W(CO) 5 (E-2) was investigated by density functional theory methods and mechanistic experimental studies (NMR, IR, UV-vis spectroscopy, FD mass spectrometry, kinetic studies, trapping of intermediates). All available data support the initial dissociation of a CO ligand to give the tetracarbonyl complex W(CO) 4

Supporting Information
Supporting Information File 1 Experimental spectra and DFT derived data.