Predicting Catalytic Activity from 13CCH Alkylidene Chemical Shift in Cationic Tungsten Oxo Alkylidene N‐Heterocyclic Carbene Complexes

A series of cationic tungsten oxo alkylidene N‐heterocyclic carbene (NHC) complexes was synthesized and structurally characterized by single crystal X‐ray diffraction. The 13C NMR chemical shifts of the alkylidene C atoms of these complexes were correlated with the diamagnetic, paramagnetic and spin‐orbit chemical shifts calculated by DFT. A good correlation (R2=0.90) between the DFT isotropic chemical shifts and the experimental chemical shift as well as a strong correlation between the DFT isotropic chemical shifts and the TOF1min for the RCM of 1,7‐octadiene was found. Further, a comparison of the catalyst geometries allowed for assigning tetracoordinate pseudotetrahedral catalysts to the most deshielded alkylidenes and to the highest TOF1min, pentacoordinate square‐planar catalysts to the intermediate deshielded alkylidenes and intermediate TOF1min, and hexacoordinate and octahedral catalyst to the most shielded alkylidene and lowest TOF1min. Analysis of the magnetic shielding tensors allowed for ascribing variations in the chemical shifts to electronic transitions between occupied molecular orbitals corresponding to the alkylidene‐C and alkylidene‐H σ‐bonds and the empty molecular orbital corresponding to the W‐alkylidene σ*‐bond.


Introduction
Molybdenum (VI) and tungsten (VI) alkylidene N-heterocyclic carbene (NHC) complexes are highly active and productive catalysts for olefin metathesis allowing for productivities (turnover numbers, TONs) > 500,000 in solution. [1] Upon immobilization on silica, TONs > 1,200,000 have been reported for cationic tungsten oxo alkylidene NHC complexes. [2] We were interested in the question, whether highly active representatives of this class of olefin metathesis catalysts, i. e. catalysts that allow for high turnover frequencies, TOFs, could be identified or even predicted by any theoretical or analytical means, e. g. by the 13 C NMR chemical shifts of the alkylidene carbon. Indeed, Copéret et al. recently outlined the importance of the analysis of the chemical shift tensors for an in-depth understanding of the reactivity of metal olefin complexes, e. g. towards oxidative cyclization and protonolysis. [3] Here, we report on a correlation between the magnetic shielding and the 13 C NMR chemical shifts of the alkylidene carbon of cationic tungsten oxo alkylidene NHC complexes based on 1, , the 1,3-dimesitylimidazol-2-ylidene (IMes) and the 1,3-diisopropylimidazol-2-ylidene ligand (IiPr), respectively, containing different anionic ligands such as chlorides, alkoxides, phenoxides and carboxylates. We finally put forward a correlation between the magnetic shielding tensors and transition between occupied and empty molecular orbitals around the alkylidene center. These values have additionally been correlated with the activity of selected complexes in the ring-closing metathesis (RCM) of 1,7-octadiene and in the cross-metathesis (CM) of 1-octene, 1dodecene as well as allylbenzene, expressed as turnover frequency after 1 min (TOF 1min ).

Synthesis of Complexes
Complexes 2-5 were prepared exploiting the route outlined in Scheme 1. [1g] The tungsten oxo alkylidene NHC dichloride complex 2 a was prepared from W(O)(CHCMe 2 Ph)Cl 2 (PMe 2 Ph) 2 (1) via reaction with IMes. The use of CuCl to remove the remaining phosphine from 2 a resulted in decomposition. However, upon addition of two equiv. of tris(pentafluorophenyl)borane as phosphine scavenger, phosphine-free 2 b was obtained, in which the borane is coordinated to the oxo group. With only one equiv. of the borane reagent a mixture of several complexes was obtained; in this case, all attempts to purify the mixture by recrystallization failed. 2 b crystallizes in the triclinic space group P � 1 with a = 1064.85 pm, b = 1274.94 pm, c = 1723.47 pm, α = 88.65°, β = 87.78°, γ = 80.77°with two molecules in the unit cell. Relevant bond lengths and angles are summarized in Table 1. The tungsten center has a distorted square pyramidal ligand sphere (τ 5 = 0.38). [4] As might be expected, the W-oxo bond is stretched after borane coordination (WÀ O1 = 176 pm). The WÀ NHC bond length is comparable to the one in 2 a. [1g] The WÀ O1À B1 angle was found to be 166°; the stretched alkylidene ligand (WÀ C40À C41) shows an angle of 160°. Consequently, the W alkylidene bond is significantly shorter than in 2 a (184 vs. 195 pm) and the alkylidene proton is in proximity to the W center (WÀ H40 = 210 pm). In solution, the alkylidene proton is observed at δ = 3.57 ppm (CDCl 3 ). This is very unusual for alkylidene resonances but is clearly a result of borane coordination and the concomitant distortion of the alkylidene ligand. A second complex (approximately 20 %) with H α at δ = 9.70 ppm was observed in solution. This signal is assigned to the boranefree complex, which is in equilibrium with 2 b, indicating that borane coordination is reversible in solution. The reaction of 2 a with one equiv. of the silver or lithium salts of various alkoxides, carboxylates or trifluoromethanesulfonate resulted in the formation of complexes 3 and 4 a-f. The mixed monotriflatemonochloride NHC complex 3 still contained the phosphine ligand. In the solid state the metal in the carboxylate-ligated complex 4 c adopts an octahedral configuration with the NHC and the chloride in the apexes (Figure 1). Consequently, the chloride is almost trans to the NHC (C1À WÀ Cl1 = 158.32°). 4 c crystallizes in the monoclinic space group P2 1 /n with a = 1156.72 pm, b = 1258.11 pm, c = 2927.0 pm, α = 90°, β = 95.92°, γ = 90°, Z = 4. The pentafluorocarboxylate is bound in an unsymmetrical η 2 -fashion (WÀ O3 = 239 pm, WÀ O2 = 224 pm). In solution, the carboxylate ligand most likely switches between η 2 -and η 1 -binding as suggested by the very broad 1 H NMR signals of this compound.
Complex 4 d, which features the decafluoroterphenolate (DFTP) ligand, crystallizes from acetonitrile (MeCN) in the triclinic space group P � 1with a = 1088.15 pm, b = 1173.07 pm, Scheme 1. Synthesis of complexes 2 a, 2 b, 3, 4 a-4 d, 5 a-5 f.  Figure S62, S.I. 4 e is the only example of a neutral W oxo alkylidene NHC complex having a distorted trigonal bipyramidal structure. The distortion is most likely caused by the steric repulsion of the 4 mesityl groups. Cationic complexes 5 a-h were prepared from the monoalkoxide/chloride precursors 4 a-f via reaction with Ag(MeCN) 2 B-(Ar F ) 4 and NaB(Ar F ) 4 , respectively. In the presence of acetonitrile, the corresponding MeCN adducts were isolated in high yield (typically > 90 %). In most cases, solvent-free NaB(Ar F ) 4 facilitated the formation of the four-coordinate cationic complexes. Surprisingly, 4 d did not react readily with NaB(Ar F ) 4 via abstraction of the chloride ligand. A mixture of unidentified products was obtained. However, the reaction with the silver salt proceeded smoothly to the corresponding cationic complex, coordinated by MeCN, and accompanied by the precipitation of silver chloride. Finally, the cationic tungsten oxo alkylidene IMes monochloro complex 5 d was prepared via the reaction of the dichloro complex 2 a with NaB(Ar F ) 4 . In 5 d, the phosphine ligand still is coordinated, since no efficient scavengers are present and the steric hindrance around the metal is comparably low. The coordination of phosphines to cationic complexes is expected to lower the reactivity in metathesis reactions dramatically and to lead to a crowded coordination sphere. [5] Indeed, in the preparation of the compounds discussed here, the fast separation of any free phosphines or phosphine-adducts was crucial to the successful isolation of the products. It is assumed that the oxo ligand is readily removed by free phosphine to form the corresponding phosphine oxides and unstable tungsten (IV) complexes. In view of these difficulties, an alternative, more direct synthetic approach was developed (Scheme 2). For these purposes, compound 1 was reacted with two equiv. of lithium hexafluorotbutoxide, followed by the addition of one equiv. of IMes. This way, the tungsten oxo alkylidene NHC bisalkoxide complex 6 was obtained in 76 % isolated yield after precipitation in hexane and quick filtration, since it decomposes in concentrated solution in the presence of phosphines. Subsequent protonation of one of the alkoxide ligands by dimethylanilinium B(Ar F ) 4 resulted in the formation of 5 b in high yield (89 %). Additionally, reduced pressure was applied during the reaction to quickly remove the volatile alcohol that formed upon protonation. Dimethylaniline was easily removed by washing the product with pentane. Compound 6 ( Figure S65, S.I.) crystallizes in the triclinic space group P � 1 with a = 1110.33 pm, b = 1878.91 pm, c = 2131.8 pm, α = 66.06°, β = 78.43°, γ = 83.27°, Z = 4 in form of two conformers. Both metal centers have a distorted square pyramidal ligand sphere (τ 5 = 0.14 and 0.13). Both W-alkoxide bond lengths are virtually identical (200 and 201 pm). All other angles and bond distances are comparable to the other structures discussed earlier.
The structural analysis of cationic complexes 5 a-h and 12 was particularly challenging since they are less prone to form crystals suitable for single-crystal X-ray analysis compared to their imido counterparts. [1f] Complex 5 g is one of the rare examples where suitable crystals could be grown from 1,2dichloroethane in the absence any additional coordinating solvents ( Figure 2). 5 g crystallizes in the monoclinic space group P2(1)/n with a = 1871.90 pm, b = 2388.31 pm, c = 1877.39 pm, α = γ = 90°, β = 106.675°, Z = 4. Compared to its highly distorted pentacoordinate precursor complex 4 e, tetracoordinate 5 g has a less crowded coordination sphere, which results in a less distorted structure (τ 4 = 0.80). [6] Complex 5 h, on the other hand, could only be crystallized in a mixture of dichloroethane and diethyl ether, which resulted in coordination of the ether functionality to the complex.
Since the ether-coordinated complex 5 h-Et 2 O has not been used in catalytic testing or calculations, the structure is of less interest and therefore is only depicted in the S.I. ( Figure S64). The first crystals grown from 1,2-dichloroethane in an attempt to crystallize 12 were resolved as the neutral complexes W(O)(CHCMe 3 )(IMes)(OCMe(CF 3 ) 2 ) 2 ( Figure S63, S.I.), which probably is a trace impurity from the chloride exchange in an earlier step but has a much higher propensity to form crystals than the cationic species. A second crop, however, yielded crystals of the desired complex 12. It crystallizes in the triclinic group P-1 with a = 1232.86 pm, b = 1638.96 pm, c = 1775.43 pm, α = 83.535°, β = 77.397°, γ = 82.581°, Z = 2. The structure is disordered in a ratio of 94 : 6 and a second weakly populated, strongly disordered tungsten alkylidyne species was found. Its structure could not unambiguously be solved. Lastly, we also obtained single crystals of 5 b-MeCN from a mixture of 1,2-dichloroethane and diethyl ether. 5 b-MeCN crystallizes in the monoclinic space group P2 1 /n with a = 2229.73 pm, b = 1620.08 pm, c = 2411.41 pm, α = γ = 90°, β = 117.3675°, Z = 4. The strongly coordinating MeCN is not replaced by diethyl ether in the solvent mixture. A non-resolvable density-peak at W (dist.: 0.8 Å) causes a large Hirschfeld factor for the W=O bond, which is not uncommon for structures of this type.
All relevant bond lengths and angles of the cationic complexes are summarized in Table 2. In general, all NHCÀ W bond lengths are shorter than in their neural precursors, which can be attributed to a delocalization of the positive charge by the NHC. [7] In the same way, the alkoxide ligands are more tightly bound to the cationic metal centers, which is reflected in bond distances of only between 185.0 and 194.2 Å, whereas in the neutral complexes those are mostly > 200 Å. The alkylidene-W and W-oxo bond distances, on the contrary, do not significantly differ from those observed in neutral W oxo alkylidene NHC complexes.

Tungsten Oxo Alkylidene NHC Complexes Bearing Bulky Aryloxides and Small NHCs
Further, complexes featuring a small NHC in combination with terphenoxides of different steric demand have been prepared (Scheme 3). The cationic complex 10 a was prepared via two different routes. First, the terphenoxide ligand was introduced to yield the phosphine-coordinated monoaryloxide monochloride complexes 7 a-c. Then, the silver iodide adduct of IMeCl 2 or IiPr (2,3-diisopropylimidazol-2-ylidene) was reacted with complexes 7 a-c. Silver iodide was found to be an excellent phosphine scavenger and thus, transmetalation proceeded smoothly to yield the pentacoordinate neutral complexes 8 a-c in good yield (60-75 %). Complex 8 b crystallizes in the monoclinic space group P2 1 /c with a = 1884.32 pm, b =  1173.60 pm, c = 1687.54 pm, α = γ = 90°, β = 97.59°, Z = 4. The metal center is coordinated by a distorted square pyramidal ligand sphere (τ 5 = 0.18). Unlike in 4 b, the smaller NHC leads to a much less distorted structure, even though the aryloxide ligand is more sterically demanding. The tungsten-NHC distance is comparable to distances observed in complexes bearing the IMes ligand. Thus, the significantly less pronounced electrondonation does not have an influence on the structure. As in 4 b, the chloride is fairly trans to the NHC (C1À WÀ Cl1 = 156.87). All other bond lengths and angles are in the expected range ( Figure S66, S.I.). By increasing the steric bulk employing the HIPTO aryloxide, compound 8 c was successfully prepared. Its structure is significantly more distorted ( Figure  In an attempt to prepare a cationic complex from 8 b and NaB(Ar F ) 4 , no precipitation of NaCl was observed. Instead, the bimolecular complex 11 formed quantitatively, containing one equivalent of NaB(Ar F ) 4 . Its solid-state structure reveals a Na + cation that is incorporated into a square pyramidal structure with the oxo and chloride ligands in the pyramidal plane (Naoxo = 231 pm, Na-chloride = 270 pm) and one of the phenyl rings from the HMTO ligand in the apex (NaÀ C distance = 293 pm). This structure is remarkably stable and is only disrupted by the addition of coordinating solvents.
The use of [Ag(MeCN) 2 B(Ar F ) 4 ] or the addition of coordinating molecules to the reaction mixture of NaB(Ar F ) 4 and 8 b led to the formation of the cationic complexes. Compound 10 a was also synthesized from the bisaryloxide complex 9. In principle, the same approach as presented in Scheme 2 for the synthesis of 5 b was used. The intermediary bisaryloxide phosphine complex 7 d was reacted with AgI . IMeCl 2 . 9 crystallizes in the monoclinic space group P2 1 /c with a = 1228.84 pm, b = 4555.8 pm, c = 1626.55 pm, α = 90, β = 111.08°, γ = 90°, Z = 8 ( Figure S68, S.I.). Two conformers co-crystallized in the unit cells. In these two conformers, the tungsten center adopts either an almost perfect or a slightly distorted square pyramidal geometry (τ 5 = 0.05 or 0.11, respectively). The aryloxide ligand trans to the NHC has a slightly longer WÀ O bond than the one trans to the oxo ligand (201 vs. 197 pm). Interestingly, the aryloxide ligand trans to the NHC is bent (C24AÀ O3AÀ W1A = 136.1°) while the one trans to the oxo functionality is almost linear (C6AÀ O2AÀ W1A = 165.5°).

Reactivity
The reactivity of complexes 5 a-d, 5 f-g, 10 a-b, 12, 5 b-MeCN and 5 c-MeCN was tested in the RCM of 1,7-octadiene and the HM of 1-octene, 1-dodecene and allylbenzene, respectively. Figure 4 shows an overview of all complexes used in the catalytic testing and DFT calculations. Although cationic tungsten oxo alkylidene complexes proved to possess an outstandingly high functional group tolerance, e. g. against nitriles, sec-amines or thioethers, [1g] we intentionally avoided Scheme 3. Preparation of neutral and cationic tungsten oxo alkylidene NHC catalysts bearing terphenoxides and a small NHC ligand. ODPP = 2,6diphenyl-phenolate. substrates bearing possibly coordinating functional groups to eliminate any unwanted, additional influence on catalytic activity. In earlier studies, this strategy was successfully utilized for the correlation of charge distribution and reactivity of cationic molybdenum imido alkylidene NHC complexes. [7] The kinetic profiles as well as TONs, E/Z-selectivity and reaction conditions are available from the Supporting Information. In addition to that, the turnover frequencies after 1 and 3 min (TOF 1min and TOF 3min ) were determined as average of three independent measurements.
Most of the tested complexes showed moderate to high activity for all four substrates. The measured TOFs range from 190 min À 1 to 18300 min À 1 . In general, the highest TOFs were observed for the RCM of 1,7-octadiene, which is in accordance with previous kinetic studies with cationic molybdenum imido alkylidene complexes. [7] Judging from the kinetic profiles of most catalysts, the reaction with 1,7-octadiene is virtually complete after a few minutes. A linear correlation (R 2 = 0.953) between the 13 C CH alkylidene shift of all tested complexes 5 a-d, 5 f-g, 10 a-b, 12, 5 b-MeCN and 5 c-MeCN and the TOF 1min was found ( Figure S60, S.I.), i. e. a higher shielding of the alkylidene carbon resulted in a higher activity of the catalyst. The highest TOF 1min of 18300 min À 1 was measured in the reaction with 5 b, which has the lowest 13 C CH shift of 297.3 ppm.
By contrast, the lowest TOF 1min was found in the reaction of 9, which has a strongly downfield shifted 13 C CH alkylidene peak (324.3 ppm). As we showed before, the σ-donor properties of the NHC can influence the catalytic activity of high oxidation state cationic group 6 alkylidene NHC complexes. [7] And indeed, by considering only the IMes-containing complexes 5 a-c, 5 f-g, 12, 5 b-MeCN and 5 c-MeCN, an excellent correlation between TOF 1min and the 13 C NMR shift of the alkylidene for 1,7-octadiene with R 2 = 0.981 was obtained ( Figure S61, S.I.).
Notably, the reaction of 5 a with 1,7-octadiene resulted in a mixture of the RCM and acyclic diene metathesis (ADMET) products of which only the dimer, 1,7,14-tetradecatriene, could be found in GC/MS. Nonetheless, higher ADMET products cannot be ruled out due to the detection limit of GC/MS for larger molecules.
In addition to the σ-donor strength of the NHC, the coordination of L-type ligands such as nitriles, phosphines or pyridines can well be expected to have an influence on both the activity and productivity of cationic tungsten and molybdenum alkylidene complexes. [8] In some cases and especially in tungsten alkylidene complexes this can lead to complete loss of activity [5] and, indeed, 10 a and 5 d, which contain two acetonitrile molecules or a sterically demanding phosphine in combination with the O-2,6-Ph 2 C 6 H 3 ligand did not show any catalytic activity.
We, therefore, strived to rule out the influence of any additional coordinating ligand such as nitriles and compared the TOF 1min of five nitrile-free complexes with three pure hydrocarbon-based substrates (Table 3, Figure 5), i. e. with 1octene, 1,7-octadiene and 1-dodecene. Very good linear correlations with high R 2 values of 0.95, 0.97 and 0.93, respectively, were found for all three substrates. Again, catalysts with higher shielding of the alkylidene carbon showed higher activity. Obviously, in nitrile-free complexes, TOF 1min can be  reliably predicted for both reactions, RCM and HM, from the 13 C alkylidene shift.

DFT modeling
Starting from the single-crystal X-ray structural data available, DFT geometry optimizations (at the PBE0-D3/SVP level of theory using the Gaussian D.01 package), [9] and NMR analysis (using the gauge-including atomic orbital (GIAO) method at the PBE0/ ZORA/TZ2P level of theory, with inclusion of spin-orbit coupling using the ADF 2019 package), [10] [11] were performed for the nine IMes-containing complexes 5 a-c, 5 f-g, 12, 5 b-MeCN and 5 c-MeCN (for computational details see the SI). The following discussion is based on ample literature on the calculation of chemical structure, on their interpretation based on the electronic structure of the metal complexes, and on the potential impact on catalysis. [12] To benchmark the chosen protocol, we initially compared the X-ray and the DFT optimized geometries of 12, 5 b-MeCN and 5 g, by calculating the rootmean-square deviation (RMSD) between the cores of the two geometries, composed of the W atoms and the atoms up to two bonds from W. The resulting RMSDs of 0.29, 0.20 and 0.14 Å indicate that the DFT geometries well reproduce the coordination sphere around the W atom, which is the region of the complex influencing mostly the electronic structure and NMR properties at the alkylidene C atom. Comparison of the DFT and experimental 13 C NMR chemical shifts of the alkylidene carbon, also resulted in good correlation (R 2 = 0.90, see Figure 6), and validated further the following analysis. Analysis of the diamagnetic, δ dia , paramagnetic, δ para , and spin-orbit, δ SO , components of the chemical shift tensors (Table 4) [13] indicated that within the examined nine complexes, δ dia and δ SO span a range of only 3.4 and 4.9 ppm, respectively, while δ para spans a range of 18.2 ppm, almost matching the isotropic DFT chemical shift (δ DFT ) window of 20.3 ppm. This suggests that the following analysis can be focused on the paramagnetic plus spin-orbit term, δ p + SO . Analysis of the principal components of the chemical shift tensor, δ 11 , δ 22 and δ 33 , indicates that the most and the least deshielding components, δ 11 and δ 33 , span a range of 24.5 and 23.1 ppm, respectively, while the intermediate deshielding component δ 22 spans a slightly larger range of 34.6 ppm, indicating that all the components similarly affect δ DFT . This preliminary analysis is in line with similar studies [14] and, more specifically, with previous work on metal-alkylidene complexes, [15] which thus set the basis for the following analysis of the magnetic shielding tensor and of the molecular orbitals (MO) responsible for the variation in the paramagnetic term. Consistently with a previous study on metal-alkylidene complexes, [15a] this MO analysis is performed within a molecular axes frame, with the WÀ C(alkylidene) bond along the Y C axis and the WÀ C(NHC) bond in the Y C Z C plane, which results in the X C axis being nearly orthogonal to the WÀ C(alkylidene) bond and nearly laying in the mean WÀ C(H)(R) plane (see Figure 7).
Next, we checked if there was a correlation between the chemical shifts and the number of ligands or geometry around the W center. This analysis (see Table 4) indicated that the tetracoordinate pseudotetrahedral (pth) complexes are systematically more shielded than the pentacoordinate square pyramidal (sp) ones, with the hexacoordinate octahedral (oh) complex 5 a being the least shielded. Considering that the experimental chemical shifts range in a window of only 27.0 ppm (20.3 ppm by DFT), which renders a detailed analysis hard to be conclusive, we focused on the two complexes corresponding to the most and to the least shielded ones, which are 5 g and 5 a, as representative cases.
Representation of the directions corresponding to the principal components in these complexes indicate that δ 11 is approximately in the plane of the C(H)(R) alkylidene moiety and nearly perpendicular to the W=C π-bond (see Figure 7), almost  Table 4. Table 4. Calculated and experimental 13 C ethylidene NMR chemical shifts for complexes of Figure 4. δ dia , δ para and δ SO are the diamagnetic, paramagnetic and spin-orbit components of the isotropic chemical shift, δ DFT . δ 11 , δ 22 and δ 33 are the principal components of δ DFT . δ Exp is the experimental chemical shift, and Δδ = δ DFT À δ Exp is error in the calculated chemical shift. coincident with the X C axis in the molecular frame. In strict similarity to the direction of the most deshielding component, [16] δ 11 only changes by 9 ppm from 5 g to 5 a, being the least relevant for the 20 ppm variation in δ iso . On the basis of a natural chemical shielding (NCS) analysis using the NBO 6.0 program, , [16] [17] as embedded in the ADF package, and in line with previous work, [15] δ 11 is mostly impacted by electronic transitions from the occupied MO corresponding to the σ W-alk bond to the empty MO corresponding to the π* W-alk bond, or by transitions from the MO corresponding to the π W-alk bond to the MO corresponding to the σ* W-alk bond, both through a coupling via 90°rotation around the X C direction. Differently, the intermediate and less shielding components are neither collinear nor perpendicular to the W=C bond, both in 5 g and 5 a (see Figure 7), preventing a straightforward description of these variations based on MOs centered at the W=C moiety. Nevertheless, the occupied MOs that should mostly contribute to δ 22 and δ 33 should correspond again to the σ W-alk and π W-alk bonds, via coupling with the MOs corresponding to the empty σ* alk-C and σ* alk-H bonds, with the addition of transitions from the occupied σ alk-C and σ alk-H MOs to the empty π* W-alk and σ* W-alk MOs.
Visual inspection of Figure 7 shows that one among the σ 22 and σ 33 components is more aligned to the WÀ C bond, specifically δ 22 in 5 g, forming an angle of 30°with the WÀ C axis, and δ 33 in 5 a, forming an angle of 27°. Instead, the other component is more perpendicular to the WÀ C bond, namely δ 33 for 5 g, forming an angle of 60°with the WÀ C axis, and δ 22 for 5 a, forming an angle of 63°. We thus compared contributions to δ 22 for 5 g with those to δ 33 for 5 a, and similarly we compared those to δ 33 for 5 g with those to δ 22 for 5 a. This means we compared components according to similarity in their direction in the molecular frame, rather than based on their magnitude.
With this caveat in mind, the variation of the principal components reported in Table 4 is 9, 14 and 66 ppm for the components mostly aligned with the X C , Y C and Z C axes, pointing to the component perpendicular to the mean = C(H)(R) plane as the one resulting in a clearly larger deshielding in 5 a compared to 5 g. Inspection of the NCS contribution to the components aligned to the Y C and Z C axes for 5 a and 5 g (see Table 5), shows negligible variations for the σ W-alk and π W-alk terms, and reduced variations for the σ alk-C and σ alk-H terms for the component aligned to the Y C axis. Instead, clearly larger variations in the shielding of the σ alk-C and σ alk-H terms occur for the component aligned to the Z C axis. As already shown, [15] these terms are dominated by transitions from the filled σ alk-C and σ alk-H MOs to the empty filled σ* WÀ C MO ( Figure 8). Finally, non-negligible contributions are also due to the σ WÀ NHC , σ W=O and π W=O bonds, indicating participation of the NHC and oxo moieties in determining the chemical shift of the alkylidene C atom.
As a last check, we focused on the different lengths of the W-alkylidene bond in 5 g and 5 a, which are 1.902 and 1.942 Å, respectively, as the shielding is impacted by both the energy gap between the MOs involved in the transition, and their overlap (see Figure 8). [13,14g-j] The latter, of course is impacted by the distance between the bonded atoms. We thus rigidly translated the entire = C(H)(R) alkylidene group in 5 g along the WÀ C bond to a W-alkylidene bond length of 1.942 Å, as in 5 a. This deformed complex 5 g experiences a downfield shift of 12.8 ppm, reducing the difference between 5 g and 5 a from 20.3 to only 8.1 ppm.  Table 5. Natural chemical shift analysis of the σ p + SO contributions of the principal components of the shielding tensor. Contributions are assigned to X C , Y C and Z C , depending on the orientation of principal components (PC) in the molecular frame.
PC aligned to 5 g 5 a 5 a-5 g X c σ W-alk À 471 À 498 À 27 π W-alk À 147 À 104 43 σ alk-C À 91 À 92 0 σ alk-H À 28 À 55 À 27 Figure 8. a) Schematic localized occupied and vacant orbitals contributing to the deshielding variation of the δ 33 term at the alkylidene C atom; b) representation of the equation determining the magnitude of the induced field upon electronic transition from one filled MO, ψ occ , to an empty MO, ψ vac , coupled by rotation of 90°of ψ occ , promoted by the angular momentum operator l. e vac À e occ is the energy gap between the ψ occ and ψ vac orbitals.

Conclusion
A correlation between the catalytic activity of a series of cationic tungsten oxo alkylidene complexes in olefin metathesis and, both, the experimentally determined 13 C alkylidene shift and the DFT isotropic chemical shifts was found. Further analysis confirmed that the variance of the total isotropic shielding, in agreement with literature, can be mainly attributed to the paramagnetic term. Comparison between catalysts geometries, the 13  Analysis of the magnetic shielding tensors of the most and least shielded complexes allowed for ascribing variations in the chemical shifts to electronic transitions between occupied molecular orbitals corresponding to the alkylidene-C and alkylidine-H σ-bonds and the empty molecular orbital corresponding to the W-alkylidene σ*-bond. Those findings provide a better understanding of the activity of tungsten oxo alkylidene NHC complexes and will be a useful tool for a more rational approach to the design of novel cationic group 6 alkylidene NHC complexes.

Preparation
of 5 b from 6: 500 mg of W-(O)(CHCMe 2 Ph)(IMes)(OCMe(CF 3 ) 2 ) 2 were dissolved in 15 mL toluene and cooled to À 40°C. To this solution 485 mg of dimethylanilinium B(Ar F ) 4 etherate (0.98 equiv) were added as a solid. After 5 min of stirring, vacuum was applied for approx. 30 min. to remove the alcohol that formed and to reduce the amount of solvent by ca. 50 %. An orange oily residue was obtained, which was triturated with 15 mL pentane until an orange solid precipitated. The product was filtered off and washed with pentane. Yield 750 mg (89 %). Analytical data were in accordance to already published data. [1g] [W(O)(CHCMe 2 Ph)(IMes)(Cl)(PMe 2 Ph)][B(Ar F ) 4 ], 5 d: 2 a (80 mg, 0.0943 mmol) was dissolved in 5 mL of CH 2 Cl 2 and cooled to À 40°C. The solution was added to solid NaB(Ar F ) 4 (84 mg, 1 equiv) and the suspension was stirred for 30 min. A colorless precipitate formed. The mixture was again cooled to À 40°C and then filtered through celite. The filtrate was reduced in vacuo to one third of the volume and filtered again. After removing the solvent an oily foam formed that was triturated with pentane until a bright yellow solid precipitated. The pentane phase was decanted and the solid was  3 Hz,1P). Elemental analysis (%) calcd. for C 71 H 59 BClF 24 N 2 OPW: C,50.96;H,3.55;N,1.67. Found: C,51.20;H,3.35;N,1.44. [W(O)(CHCMe 2 Ph)(IMes)(DFTP)(MeCN)][B(Ar F ) 4 ], 5 e: 4 d (42 mg, 0.0378 mmol) was dissolved in 5 mL of CH 2 Cl 2 and the solution was cooled to À 40°C. Then, the solution was added to solid Ag-(MeCN) 2 B(Ar F ) 4 (40 mg, 1 equiv). The resulting suspension was stirred for 30 min. A colorless precipitate formed. The mixture was cooled to À 40°C again and filtered through celite. The filtrate was reduced in vacuo to one third of the volume and filtered again. After removing the solvent, an oily foam formed that was triturated with pentane until a bright yellow solid precipitated. The pentane phase was decanted and the solid was dried in vacuo.   4 ], 5 f: 4 c (100 mg, 0.113 mmol) was dissolved in 5 mL of CH 2 Cl 2 , the solution was cooled to À 40°C and added to solid Ag(MeCN) 2 B(Ar F ) 4 (119 mg, 1 equiv). The resulting suspension was stirred for 30 min. A colorless precipitate formed. The solution was again cooled to À 40°C for 30 min and filtered through celite. The filtrate was reduced in vacuo to one third of the volume and filtered again. After removing the solvent, an oily foam formed that was triturated with pentane. The pentane phase was decanted, and the foam was dried in vacuo.   4 ], 5 g: 4 e (55.8 mg, 55.7 μmol) was dissolved in 2 mL of CH 2 Cl 2 , the solution was cooled to À 35°C and added to solid NaB(Ar F ) 4 (49.4 mg, 1 equiv). The resulting suspension was stirred for 1 h. A colorless precipitate formed. The solution was reduced in vacuo to one third of the volume and filtered over celite. After removing the solvent, an oily foam formed that was triturated with pentane (2 mL) and a lightyellow solid formed which was dried in vacuo. Yield: 90 mg (88 %).   8, 31.2, 22.9, 21.6, 21.4, 21.1, 21.0, 18.2, 17.2, 14.4 4 ], 5 h: 4 f (93 mg, 87.0 μmol) was suspended in 2 mL of CH 2 Cl 2 , the solution was cooled to À 45°C and added to solid NaB(Ar F ) 4 (77.0 mg, 1 equiv). The resulting suspension was stirred for 2.5 h. The starting material dissolved and a colorless precipitate formed. The solution filtered through celite. After removing the solvent, an oily foam formed that was triturated with pentane and a light-yellow solid formed which was dried in vacuo. The product can be recrystallized from a mixture of CH 2 Cl 2 /diethyl ether/pentane. Yield: 147 mg (89 %).   2 was dissolved in 50 mL of toluene. The solution was cooled to À 40°C and solid lithium alkoxide (576 mg, 2.05 equiv) was added in one portion. The yellow suspension was stirred for 3 h at room temperature. The solvent was reduced to 20 mL and the suspension was filtered over celite. IMes was added as a solid in one portion to the filtrate (477 mg, 1.05 equiv). The reaction was stirred for 30 min and the solvent was reduced to 5 mL. 20 mL of hexane were added to the green/yellow oily residue to precipitate the product. After filtration, the filtrate was cooled to À 40°C to precipitate a second crop of product. Combined yield: 1.13 g (76 %) of a pale yellow solid. For high yields, the quick separation of the product from the phosphine is crucial since free phosphine causes decomposition, especially in concentrated solutions (dark green colored byproducts  ,4.24;N,2.81. Found: C,46.77;H,4.46;N,2.89. W(O)(CHCMe 2 Ph)(PMe 2 Ph)(O-2,6-Ph 2 -C 6 H 3 )(Cl), 7 a: 1 (731 mg, 1.07 mmol) was dissolved in 10 mL of toluene. The solution was cooled at À 40°C for 30 min. Solid Li-2,6-diphenylphenolate was added to the solution and the mixture was stirred overnight at room temperature. Subsequently, the turbid solution was filtered through celite and the solvent was removed in vacuo, leaving a yellow/orange oil. The oil was taken up in 5 mL toluene and filtered again. After reducing the solvent, a yellow solid started to precipitate from the solution. The solution was put in the freezer overnight at À 40°C. Yield: 690 mg (86 %) of a yellow to pale yellow solid. 1 H NMR (400 MHz, CD 2 Cl 2 ) δ 1.03 (d, 3H, PMe 2 , J PH = 9.9 Hz), 1.14 (d, 3H, PMe 2 , J PH = 10.1 Hz), 1.34 (s, 3H, CMe 2 Ph), 1.52 (s, 3H, CMe 2 Ph), 6.89 (m, 2H, CMe 2 Ph), 6.97 (m, 1H, CMe 2 Ph), 7.10 (m, 3H, Ar), 7.14 (m, 3H, Ar), 7.18 (m,2H,Ar) 3-dimethyl-4,5-dichloroimidazol-2-ylidene (122 mg, 1.4 equiv) was added to the stirring solution (more equivalents of AgI · 1,3dimethyl-4,5-dichloroimidazol-2-ylidene and gentle heating to 50°C might be necessary on larger scales to obtain full conversion). The reaction mixture was stirred for 12 h at room temperature. All solids were filtered through celite and the filtrate was reduced to dryness. The residue was taken up in a mixture of diethyl ether and pentane (5 mL, 1 : 1). The colorless insoluble solid that formed was filtered off. The filtrate was reduced to 2 mL and filtered once again. Then the solution was stored at À 35°C overnight. A pale-yellow solid material precipitated. Yield 113 mg, 67 %. 1 H NMR (400 MHz Compound 7 c (61 mg, 0.0611 mmol) was dissolved in 6 mL of benzene. Solid AgI · 1,3dimethyl-4,5-dichloroimidazol-2-ylidene (30 mg, 1.2 equiv) was added under stirring (more equivalents of AgI · 1,3-dimethyl-4,5dichloroimidazol-2-ylidene and gentle heating to 50°C might be necessary on larger scales to obtain full conversion). The reaction mixture was stirred for another 12 h at room temperature. All solids were filtered off over celite and the filtrate was reduced to dryness. The residue was taken up in a mixture of diethyl ether and pentane (5 mL, 1 : 1). The colorless insoluble solid that formed was filtered off. The filtrate was reduced to 1 mL and filtered again. Then the solution was cooled at À 35°C overnight. An off-white solid material precipitated. Yield 47 mg, 74 %. 1 H NMR (400 MHz, CD 2 Cl 2 ) δ 10.25 (s, 1 J CH = 122. 9 Hz, 1H, W=CH), 7.16 (m, 1H, Ar), 7.07 (m, 1H, Ar), 7.05 (m, 1H, Ar), 6.94 (m, 4H, Ar), 6.86 (m, 4H, Ar), 6.75 (m, 1H, Ar), 3.34 (m, 2H, CH-iPr), 2.56 (br s, 6H, MeÀ NHC), 2.83 (m, 2H, CH-iPr), 2.24 (m, 2H, CH-iPr), 1.61 (s, 3H, CMe 2 Ph), 1.34 (s, 3H, CMe 2 Ph), 1.33 (d, 3