Cationic molybdenum oxo alkylidenes stabilized by N-heterocyclic carbenes: from molecular systems to efficient supported metathesis catalysts

Cationic d0 group 6 olefin metathesis catalysts have been recently shown to display in most instances superior activity in comparison to their neutral congeners. Furthermore, their catalytic performance is greatly improved upon immobilization on silica. In this context, we have developed the new family of molecular cationic molybdenum oxo alkylidene complexes stabilized by N-heterocyclic carbenes of the general formula [Mo(O)(CHCMe3)(IMes)(OR)[X−]] (IMes = 1,3-dimesitylimidazol-2-ylidene; R = 1,3-dimesityl-C6H3, C6F5; X− = B(3,5-(CF3)2C6H3)4−, B(ArF)4, tetrakis(perfluoro-t-butoxy)aluminate (PFTA)). Immobilization of [Mo(O)(CHCMe3)(IMes)(O-1,3-dimesityl-C6H3)+B(ArF)4−] on silica via surface organometallic chemistry yields an active alkene metathesis catalyst that shows the highest productivity towards terminal olefins amongst all existing molybdenum oxo alkylidene catalysts.


Introduction
Since the discovery of olen metathesis more than half a century ago, 1 the development of novel catalyst families has been at the core of extensive research efforts, with the goal to increase catalytic performance, from functional group tolerance to increased activities, selectivities and stabilities. 2 Apart from Ru-based "Grubbs-type" catalysts, 3 d 0 early-transition-metal metathesis catalysts mostly based on group 6 metals (Mo/W), i.e. Schrock-type catalysts of the general formula M(E)(CHR)(X)(Y) with E ¼ oxo or imido and X, Y ¼ anionic ligands of various types (alkyls, alkoxys, amidos), have emerged as a central class of olen metathesis catalysts. 4 Besides, the use of surface organometallic chemistry (SOMC) 5 has enabled to greatly improve their activity and stability through generation of the corresponding well-dened active sites dispersed at the surface of oxide supports like silica. 6 In parallel, DFT calculations have revealed that the right balance of the s-donation ability of the X, Y and E ligands is key to these improved catalytic performances. 7 The concurrent presence of both strong and weaker s-donating anionic X and Y ligands for instance avoids overstabilization of the metallacyclobutane intermediates while favouring [2 + 2] cycloaddition and cycloreversion processes as well as suppressing deactivation pathways. N-Heterocyclic carbenes (NHCs) have gained increasing importance as ancillary ligands in transition metal complexes since the 1990s. 8 In this regard, the introduction of NHCs in overall tetracoordinate, metathesis-active, cationic Mo and W species has illustrated how catalytic performance can be further improved. 9 Charge delocalization between the NHC and the cationic metal centre renders cationic complexes rather "so" according to the HSAB principle, which explains the high activity and functional group tolerance in cationic group 6 alkylidene complexes. 10 The combination of both conceptsintroduction of the strong s-donating NHC ligands and weak siloxy ligands provided by the silica surfacehas thus enabled the generation of some of the most active and stable olen metathesis catalysts to date. 9b,d,9g For the cationic tungstenbased catalysts, the oxo-bearing complexes outperformed their imido analogues by 1-2 orders of magnitude in activity; 9g this can be attributed to the combination of the smaller E ligand size, allowing for easier access of the incoming olen to the metal centre, and its strong s-donating ability, which simultaneously enhances catalyst stability. Previously published immobilized neutral Mo oxo alkylidene complexes proved to possess high activity for internal olens (Fig. 1a). 2j,11 Consequently, we wanted to investigate whether the activity of molybdenum oxo complexes can be boosted for terminal olens similar to what was shown for analogous tungsten oxo complexes. 9g

Synthesis of molecular catalysts
In this context, we developed and herein report the synthesis and characterization of the long-awaited family of molecular and well-dened silica-supported cationic Mo oxo species stabilized by NHC ligands (Fig. 1b) and investigated their activity towards internal and terminal olens. The synthesis of cationic molybdenum oxo alkylidene NHC complexes was rst attempted starting from the 4-methoxybenzylidene precursor Mo-1 (Scheme 1a). 12 Addition of 1 equiv. of 1,3-bis(2,4,6trimethylphenyl)imidazol-2-ylidene (IMes) to Mo-1 resulted in deprotonation of the alkylidene ligand. Through careful choice of a less basic chlorinated NHC, 13 this side-reaction could be vastly circumvented and the desired product was isolated in 34% yield. Single crystals of Mo-2 for X-ray diffraction were grown from a mixture of dichloroethane/n-heptane at À40 C (Fig. 2). Mo-2 crystallizes in the triclinic space group P 1 with a ¼ 1184.11(6) pm, b ¼ 1247.18(7) pm, c ¼ 1537.13(8) pm, a ¼ 80.252(2) , b ¼ 81.208(2) , g ¼ 66.112(3) , Z ¼ 2. In the solid state, the complex adopts a distorted square pyramidal geometry (s 5 ¼ 0.34) 14 with a syn alkylidene ( 1 J CH ¼ 134 Hz) in the apical position. The Mo-oxo bond (166.4 pm) is slightly shorter than in ve-coordinate W oxo alkylidene NHC complexes (169.0-176.0 pm) whereas the Mo-NHC bond distance of 226.3 pm is well in the expected range of comparable tungsten complexes.
Exchange of the uorinated alkoxides was achieved by protonation using N,N-dimethylanilinium chloride to furnish complex Mo-3. Due to their electron count of 16 (counting the free electron pair at the oxo ligand, too) and their binding site occupied by an NHC, both, Mo-2 and Mo-3 were not expected to exhibit high activity in olen metathesis. Therefore, we strived for the synthesis of highly active cationic species. 9k  Abstracting one of the X-type ligands using either N,Ndimethylanilinium B(Ar F ) 4 in the reaction with Mo-2 in the presence of pivalonitrile (pivCN) or Ag(pivCN) 3 B(Ar F ) 4 in the reaction with Mo-3 resulting in complexes of the formula [MoO(CH-4-(OMe)C 6 H 4 )(X)(IMesCl 2 )(pivCN)][B(Ar F ) 4 ] (X ¼ OC(CF 3 ) 3 , Cl) proved to be successful when conducted in CDCl 3 and monitored by 1 H NMR. However, the cationic species readily decomposed upon workup, which we attribute to the low steric demand of both, the 4-methoxybenzylidene moiety and the oxo ligand. And indeed, so far, all isolated cationic group 6 alkylidene complexes either bear a neophylidene or neopentylidene ligand 9c in line with the "almost magic properties" of the neopentylidene ligand contributing to the stability of group 6 complexes. 15 We consequently shied our attention to a lately published procedure leading to the Mo oxo neopentylidene complex MoO(CHCMe 3 )Cl(OHMT)(3-Brpy) (OHMT ¼ 2,6-dimesitylphenoxide; 3-Brpy ¼ 3-bromopyridine). 16 3-Bromopyridine as a relatively weak donor could then be readily replaced by strongly sdonating IMes (Scheme 1b). In this case, no deprotonation of the alkylidene ligand occurred. The resulting ve-coordinate NHC complex was not isolated on account of its poor crystallisation propensity. Nevertheless, the addition of NaB(Ar F ) 4 and lithium tetrakis(peruoro-t-butoxy)aluminate (LiPFTA), respectively, to the crude reaction mixture led to formation of the desired cationic complexes that could be puried by recrystallization and which were isolated in good yields of 72 and 80%, respectively, over two steps. The stable cationic species served as efficient and general precursors for immobilization but can also be further transformed by exchange of the X-ligand. Thus, upon addition of an excess of HOC 6 F 5 to Mo-7, quantitative protonation of the HMTO ligand was facilitated to form Mo-8 in a high isolated yield of 84%. The coordination of THF in complex Mo-8 that lacks the sterically demanding alkoxide is imperative for its purication by recrystallization. Single crystals of Mo-8 suitable for X-ray diffraction were grown from 1,2dichloroethane at À40 C (Fig. 3). As for Mo-2, the triclinic space group P 1 was found in Mo-8 with the unit cell dimensions being a ¼ 1296.54(8) pm, b ¼ 1388.76(9) pm, c ¼ 1930.09(12) pm, a ¼ 79.898(2) , b ¼ 78.994(2) , g ¼ 86.057(3) and two molecules in the unit cell. The distorted square pyramidal structure (s 5 ¼ 0.29) bears a syn-alkylidene ligand in the apical position and all other ligands in the equatorial plane. All bond lengths and angles are similar to those observed for ve-coordinate cationic W oxo alkylidene NHC complexes. 17 On a nal note regarding the synthesis of cationic Mo oxo alkylidene complexes, the synthesis of further species bearing uorinated alkoxides starting from aryloxide complexes Mo-6 and Mo-7 have been attempted using HOC(CF 3 ) 3 . Unfortunately, the reactions proved unsuccessful, and no conversion was observed. However, the aryloxide ligand in Mo-6 and Mo-7 was successfully protonated by the action of HCl in Et 2 O in the presence of pivCN to yield a mixture of complexes of the type [MoO(CHCMe 3 )Cl(IMes)(pivCN)][X] (X ¼ B(Ar F ) 4 or LiPFTA) and the protonated OHMT ligand as an inseparable oil. All attempts to isolate the complexes by crystallization or to further convert them in situ using MOC(CF 3 ) 3 (M ¼ Li, Ag) failed.

Immobilization of complexes
The reaction of Mo-7 with HOC 6 F 5 illustrates that protic functional groups can lead to the very selective exchange of the Xligand without notable side-reactions. Following the development of the molecular compounds, we also sought to exploit this reactivity for graing the complex on silica partially dehydroxylated at 700 C (SiO 2-700 ). For comparability with previously published complexes, 9b,d,9g the B(Ar F ) 4 -containing complex Mo-6 was selected for this purpose and yielded the corresponding supported species Mo-6@SiO 2 (Scheme 2).
Quantication by 19 F solution NMR spectroscopy of the remaining B(Ar F ) 4 of the complex indicates that ca. 29% of the surface silanols (0.32 mmol g À1 ) reacted with Mo-6, in line with a molybdenum loading of 0.79 wt% as determined by elemental analysis. IR spectroscopy conrms that a signicant amount of isolated silanols (ñ¼ 3747 cm À1 ) are consumed upon graing by protonolysis, although the large size of Mo-6 likely prevents full coverage.
Besides of the appearance of CH stretching bands between n1 3200 and 2800 cm À1 , a broad absorption band appeared at around ñ¼ 3700 cm À1 , typical of the non-covalent interactions of unreacted surface silanols with the aromatic ligands of the graed complex. 2h,9b,9d,11 Further characterization of Mo-6@SiO 2 by 1 H magic angle spinning (MAS) NMR showed the alkylidene proton resonance at 13.6 ppm, slightly upeld with respect to the molecular precursor signals in solution (13.9 ppm). The 13 C cross-polarization (CP) MAS NMR spectrum shows the presence of the different methyl moieties and aromatics, whereas the alkylidene carbon, unfortunately, was not observed (for further details see ESI Fig. S1 and S2 †).

Catalytic testing of molecular and immobilized complexes
Evaluation of the catalytic activity for Mo-6 was carried out for both molecular and immobilized complexes in the homo metathesis of the two benchmark substrates cis-4-nonene and 1nonene at 30 C (Table 1). While the homogeneous complex Mo-6 had to be tested in o-dichlorobenzene (DCB) due to solubility requirements, the immobilized analogue Mo-6@SiO 2 was tested in both DCB and toluene, prototypical metathesis solvents used in many studies. In all cases, the metathesis activity towards a-olens was higher than for internal ones. This behaviour was most pronounced for the molecular complex Mo-6, where no activity was observed for internal olens and only 86% conversion aer 8 hours was reached with 1-nonene. The lack of reactivity is likely associated with the presence of the large phenoxy (OHMT) ligand, whose steric encumbrance probably hinders the required distortion of the metal complex to a trigonal prism necessary to bind the incoming olen. 2j In fact, switching OHMT for the smaller surface siloxy ligand leads to an increase in activity towards both olens. 2g,h Notably, the initial turnover frequency recorded for the self-metathesis of cis-4-nonene was modest in comparison to other molybdenum oxo alkylidenes (TOF 3 min ¼ 26 min À1 vs. >500 min À1 ). 2j,11 This observation is in line with the reported activity difference for the cationic tungsten analogues 9b and established trends that show increased activity for internal olens in Schrock-type metathesis catalysts with decreasing s-donor strength of the X ligands (X ¼ NHC, pyrrolide, alkoxide). 9b, 11,18 Considering also the activity for terminal olens, the same correlation can only be observed in few cases. 11 The decisive factor for a high catalytic activity might be rather the right balance of strong and weak s-donor ligands as outlined above. 7 However, the activity of Mo-6@SiO 2 towards terminal olens exceeds those previously reported for this family of supported complexes with TOF 3 min ¼ 254 min À1 at 0.02 mol% catalyst loading compared to only 75 min À1 and 170 min À1 of their neutral congeners at their optimal loadings (Fig. 4). 2j,11 Indeed, this supports our working hypothesis that the catalytic activity of cationic Mo oxo alkylidenes is enhanced for terminal olens similarly as shown previously for the corresponding tungsten oxo complexes. 9g Furthermore, the catalyst remained active (94% aer 8 h), when increasing the substrate to catalyst ratio to 10 000 : 1.
The excellent performance of Mo-6@SiO 2 was additionally demonstrated by recycling tests, where maximum conversion was obtained within 8 h three times, albeit accompanied by a constant loss of activity (see ESI †).  In previous studies it was shown that the homometathesis of 1-nonene using neutral immobilized Mo alkylidene catalysts can lead to low selectivities through isomerization of the position of the double bond. This can result in a selectivity of the desired hexadec-8-ene of only 80-85%, especially in closed vessels and at prolonged reaction times. 2m,5b This type of isomerization can be attributed to Mo(IV) species resulting from methylidene complex decomposition that form p-complexes with olens in the reaction mixture. Subsequently, the pcomplexes undergo oxidative addition and reductive elimination processes and, thereby, lead to isomerization. 19 Monitoring the selectivity for the formation of the homocoupling product in the metathesis of 1-nonene using cationic Mo oxo alkylidene complexes, in contrast, demonstrated that neither the homogeneous nor the immobilized catalyst are prone to form structural isomers. Even aer long reaction times of 8 h, high selectivities of ca. 98% were obtained.

Conclusions
In conclusion, catalysts belonging to the new family of cationic Mo oxo alkylidenes NHC complexes have been developed. The immobilization on silica yielded a catalyst with preferential activity towards a-olens over internal olens. This particular reactivity is attributed to the nature of the NHC ligand, whose steric demand and strong s-donor ability have an important inuence on the formation and reversion of the TBP metallacyclobutane. These results further contribute to the overall understanding on how structural and electronic properties of metal alkylidenes affect the catalytic performance and further highlight the role of the NHC ligand in providing highly active cationic d 0 -based olen metathesis catalysts.

General considerations
All experiments were carried out under an inert nitrogen or argon atmosphere using Schlenk techniques or an MBraun or GS glovebox equipped with a purier unit. Water and oxygen levels were kept below 0.1 ppm. Diethyl ether, methylene chloride, pentane, toluene and tetrahydrofuran were puried using double MBraun SPS alumina columns. Benzene and benzene-d 6 were distilled from Na/benzophenone. 1,2-Dichlorobenzene was distilled from CaH 2 . All solvents were degassed by three consecutive freeze-pump-thaw cycles.
Elemental analyses were performed at Mikroanalytisches Labor Pascher, Germany, and at the Institute of Inorganic Chemistry, University of Stuttgart, Germany. All infrared (IR) spectra were recorded using a Bruker FT-IR Alpha spectrometer placed inside a glovebox, equipped with the OPUS soware. The IR spectrometer had a total spectral range of 275-7500 cm À1 with a resolution <2 cm À1 and consisted of a RockSolid interferometer, a DTGS (triglycine sulfate) detector and a SiC globar source. Solid samples were investigated in a magnetic pellet holder. A typical experiment consisted of 32 consecutive transmission measurements in the region from 4000 to 400 cm À1 . Solution NMR spectra were recorded on a Bruker Avance III 400.
Chemical shis are reported in ppm relative to the solvent signal (CDCl 3 : 7.26 ppm, C 6 D 6 7.16 ppm, CD 2 Cl 2 5.13 ppm). 20 Data are reported as follows: chemical shi, multiplicity (s ¼ singlet, d ¼ doublet, t ¼ triplet, q ¼ quartet, quint ¼ quintet, sept ¼ septet, br ¼ broad, m ¼ multiplet), coupling constants (Hz) and integration. The 1 H and 13 C solid-state NMR spectra were acquired on a Bruker AVANCE III spectrometer operating at 400 MHz 1 H frequency (9.4 T) and equipped with a 3.2 mm proton-heteronucleus magic angle spinning ( 1 H,X MAS) probe (Bruker). Samples were packed into a 3.2 mm MAS NMR zirconia rotor closed with a VESPEL drive cap in an argon-lled glovebox, transferred to the NMR spectrometer in a tightly sealed vial under argon, rapidly inserted into the NMR spectrometer and spun under dry nitrogen. The MAS frequency was set to 16 kHz for all experiments. For the 13 C CP MAS measurements, the Hartman-Hahn cross-polarization was performed by making use of a linear ramp from 70% to 100% with the contact time set to 3 ms. For decoupling, SPINAL64 was applied with 100 kHz irradiation on 1 H. The acquired solid-state NMR spectra are referenced externally with respect to the downeld signal of adamantane (38.5 ppm). All MAS NMR spectra were acquired at ambient temperature.
Liquid catalytic test aliquots were analysed using a GC/FID (Agilent Technologies 7890 A) equipped with a split-splitless injector heated to 250 C, injection volume 0.5 mL using hydrogen carrier gas. Chromatographic separations for 1-nonene and cis-4-nonene catalytic tests were performed using an HP-5 (Agilent Technologies) column (30 m, 0.32 mm, 0.25 mm stationary phase). Crystal data have been deposited with the Cambridge Crystallographic Data Centre (CCDC): Mo-2 CCDC 2151036, Mo-8 CCDC 2151037.
Silica. Silica (Aerosil Degussa, 200 m 2 g À1 ) was compacted with distilled water, sieved, calcined at 500 C under air for 12 h and treated under vacuum (10 À5 mbar) at 500 C for 8 h and then at 700 C for 14 h (referred to as SiO 2-700 ).

Author contributions
JVM and JDJS contributed equally.

Conflicts of interest
There are no conicts to declare.