Direct observation of reaction intermediates for a well defined heterogeneous alkene metathesis catalyst

Abstract

Grafting of [W(≡NAr)(=CHtBu)(2,5-Me₂NC₄H₂)₂] on a silica partially dehydroxylated at 700°C (SiO_{2- (700)}) generates the corresponding monosiloxy complex [(≡SiO)W(≡NAr)(=CHtBu)(2,5-Me₂NC₄H₂)] as the major species (≈90%) along with [(≡SiO)W(≡NAr)(CH₂tBu)(2,5-Me₂NC₄H₂)₂], according to mass balance analysis, IR, and NMR studies. This heterogeneous catalyst displays good activity and stability in the metathesis of propene. Very importantly, solid state NMR spectroscopy allows observation of the propagating alkylidene as well as stable metallacyclobutane intermediates. These species have the same reactivity as the initial surface complex [(≡SiO)W(≡NAr)(=CHtBu)(2,5-Me₂NC₄H₂)], which shows that they are the key intermediates of alkene metathesis.

Alkene metathesis, a key reaction in industry, is used for the synthesis of both basic and fine chemicals as well as polymers (1–3). Because of an increasing world demand for propene, one of the most important processes involving metathesis is now the synthesis of propene through cross-metathesis of ethene and 2-butene. This process relies on heterogeneous catalysis because of its many technical and economical advantages. However, the development of more efficient systems in terms of activity, selectivity, and stability is still required because of the need to optimize the use of resources within the context of a more sustainable development. Although several breakthroughs in the preparation and characterization of single-site heterogeneous catalysts have already appeared (4–7), including those in the field of alkene metathesis catalysts (8–15), understanding deactivation phenomena is still a challenge today. In this respect, the observation and determination of the fate of reaction intermediates would help tackle this problem and thereby lead to improved catalyst performances. In the specific case of alkene metathesis, it has been proposed that carbenes and metallacyclobutanes are key reaction intermediates (16). Although they have been unambiguously prepared and observed for homogeneous catalysts (17–29), observing them as heterogeneous catalysts has proved to be a formidable challenge (30–32). Note that because the surface can play such a crucial role, one should never assume that the intermediates are identical in heterogeneous and homogeneous systems (31, 32). Currently, whereas well defined highly active silica supported metallocarbene precursors having neopentylidene and neophylidene ligands can be prepared and characterized at a molecular level, observation of propagating alkylidene or metallacyclobutanes have remained elusive even for these well defined systems (33).

Here, we describe the preparation of a well defined silica supported W alkylidene species [(≡SiO)W(≡NAr)(=CHtBu)(2,5-Me₂NC₄H₂)] (compound 1), report its activity in alkene metathesis, observe the methylidene and metallacyclobutane species, and show they are key reaction intermediates in alkene metathesis.

Results and Discussion

Grafting [W(≡NAr)(=CHtBu)(2,5-Me₂NC₄H₂)₂] (compound 2) (34) on a silica that has been partially dehydroxylated at 700°C (SiO_{2- (700)}), [2/SiO_{2- (700)}] gives the corresponding monosiloxy complex [(≡SiO)W(≡NAr)(=CHtBu)(2,5-Me₂NC₄H₂)] (compound 1) as the major species (≈90%) along with [(≡SiO)W(≡NAr)(CH₂tBu)(2,5-Me₂NC₄H₂)₂] (compound 3), according to mass balance analysis, IR, and NMR studies as previously observed in the case of the Mo homolog (Scheme 1) (14). Monitoring the grafting step by IR spectroscopy shows that most surface silanols are consumed and that aromatic and alkyl ligands appear as evidenced from the appearance of signals in the 3,100–2,700 (ν_C-H) and 1,600–1,350 (δ_C-H) cm⁻¹ regions [supporting information (SI) Fig. S1]. In addition, 0.9 eq of 2,5-dimethylpyrrole is present after grafting, and elemental analyses of the resulting solid (W = 3.73 ± 0.05% _wt, C = 5.70 ± 0.1% _wt, and N = 0.71 ± 0.1%_wt) corresponds to 0.21 mmol/g W, 24 ± 2 carbons, and 2.5 ± 0.5 nitrogen molecules per grafted W, which also is consistent with consumption of most of the surface silanols (0.21 mmol/g W for 0.26 mmol/g SiOH in SiO_{2- (700)}) and formation of 1 as the major surface species, for which 23 C/Mo and 2 N/Mo are expected. Furthermore, the ¹H magic angle spinning (MAS) solid state NMR spectrum, and more convincingly, the constant time (CT) (35, 36) ¹H MAS NMR spectrum (Fig. S2), display the following resonances: 9.0 ([W] = CHtBu), 7.0 (Csp²-H), 6.7 (Csp²-H), 5.5 (pyrrolyl-Csp²-H), 3.2 (CHMe₂), and 0.9 ([W] = CHCMe₃ + pyrrolyl-Csp³-H + CHMe₂) ppm as expected for 1, with the signal at 1.9 ppm being associated with the methylene protons in the neopentyl ligand, [W]-CH₂tBu, of 3 (see Scheme 1). Conversely, the ¹³C cross-polorization (CP) MAS spectrum (Fig. S3 and Table S1) displays the expected signals for 1, with the exception of the carbenic carbon [W]=CHtBu, which is not detected, as well as an extra signal at 69 ppm, which can be associated with the methylene carbon [W]CH₂tBu of the neopentyl ligand in 3. These assignments are confirmed by performing the 2D ¹H-¹³C heteronuclear correlation spectroscopy (HETCOR) solid state NMR on the ¹³C labeled species [(≡SiO)W(≡NAr)(=*CHtBu)(2,5-Me₂NC₄H₂)] (compound 1*), 33% ¹³C labeled on the α-carbon to W, which clearly shows a correlation associated with the neopentyl carbon [W]-*CH₂tBu in 3 at 70 ppm and its proton at 1.9 ppm, but which also, and more importantly, displays the neopentylidene carbon signal [W]=*CHtBu at 260 ppm and its correlation with the corresponding proton [W]=CHtBu at 9.0 ppm, associated with 1 (Fig. S4). All of these data are consistent with the formation of 1 as the major surface species (90%) via cleavage of the WN bond by the surface silanol, and the presence of a minor amount of 3, formed through the competing reaction of addition of a surface silanol with the W=C double bond of 2.

Scheme. 1.

Formation of [(≡SiO)W(≡NAr)(=CHtBu)(2,5-Me₂NC₄H₂)] (compound 1, 90%) and [(≡SiO)W(=NAr)(CH₂tBu)(2,5-Me₂NC₄H₂)₂] (compound 3, 10%) by grafting of [W(≡NAr)(=CHtBu)(2,5-Me₂NC₄H₂)₂] (compound 2) with the isolated silanols of SiO_{2- (700)}.

Contacting [2/SiO_{2- (700)}] with propene in a flow reactor (≈82 ml/min⁻¹; ≈420 mol of propene per mol of W per minute) selectively yields ethene and 2-butenes with an initial activity of 0.41 s⁻¹ (Fig. S5). Deactivation is slow, and the activity after 1,500 min is still ≈0.1 s⁻¹, so that 25,000 turnovers can be achieved within this time. Although not especially active, 2/SiO_{2- (700)} displays good stability compared with Mo- and Re-based catalysts, which typically deactivate rapidly (loss of 75% of the activity within 100–400 min). Stability of this catalyst led us to investigate the possibility of observing the NMR spectroscopic signatures of alkene metathesis reaction intermediates, i.e., the propagating carbenes and metallacyclobutane intermediates (10, 37). Therefore, ¹³C dilabeled ethene was contacted at −196°C with [2/SiO_{2- (700)}]. Upon warming the reaction mixture to room temperature, the yellow powder turned red. After 4 h at 25°C, the gas phase contained ≈0.4 eq of 95% ¹³C monolabeled tBuCH=*CH₂. The corresponding solid was then analyzed by solid state NMR. From ¹H, ¹³C CP and ¹H-¹³C HETCOR MAS experiments (Fig. 1), it is clear that intense new peaks have appeared that can be readily assigned as follows (Scheme 2): Correlation of the ¹H and ¹³C signals at (¹H: 9.0/¹³C: 234 ppm) is     fully consistent with a methylidene ligand [W]=CH₂ in [(≡SiO)W(≡NAr)(=CH₂)(2,5-Me₂NC₄H₂)] (compound 4) (38) resulting from the cross metathesis of the neopentylidene and the ¹³C di-labeled ethene.

Fig. 1.

2D ¹H-¹³C HETCOR solid state NMR spectrum of [2/SiO_{2- (700)}] contacted with ¹³C dilabeled ethene. This spectrum was acquired under eDUMBO-1_12.5 (44) homonuclear decoupling (F₁) and SPINAL-64 (45) heteronuclear decoupling (F₂) at ν₁^H = 100 kHz. The contact time for CP and the recycle delay were 500 μs and 2 s, respectively. A total of 48 t₁ points with 512 scans each were collected. A scaling factor (λ_exp) of 0.54 was used in F₁ for eDUMBO-1_12.5 homonuclear decoupling. Asterisks indicate spinning side bands of the alkylidene resonance [W] = CH₂. The ¹³C CP MAS spectrum (512 scans) recorded with a CP step of 500 μs and a recycle delay of 2 s is indicated above the 2D plot. To the left of the 2D plot is shown the ¹H single pulse spectrum (eight scans). Dashed and dotted correlations correspond to the methylidene [W]=CH₂ and metallacyclobutane [W](CH₂CH₂CH₂) moieties in 4 and 5, respectively.

Scheme. 2.

Formation and NMR spectroscopic signatures of methylidene (compound 4) and tungstacyclobutane (compound 5) species by of [(≡SiO)W(≡NAr)(= CHtBu)(2,5-Me₂NC₄H₂)] (1) with ¹³C dilabeled ethene.

Resonances appearing at (¹H: −0.9/¹³C: −7 ppm) and (¹H: 4.0/¹³C:     100 ppm) correspond to β- and α-carbons of the unsubstituted metallacyclobutane [(≡SiO)W(≡NAr)(CH₂CH₂CH₂)(2,5-Me₂NC₄H₂)] with a trigonal bipyramid geometry (5-TBP). Proton and carbon resonances at 2.0 and 46 ppm, respectively, can be attributed to the protons and carbons of a metallacyclobutane with a square based pyramid geometry (5-SBP).

Observation of stable metallacyclobutanes in two types of geometries is fully consistent with previous experimental (in solution) (18–25) and theoretical studies (39–43) for d⁰ metal complexes. It is also noteworthy that in solution, the SBP/TBP isomer ratio and the reactivity in metathesis are correlated, for a series of isoelectronic metallacyclobutane intermediates of the general formula [(RO)₂W(≡NAr)(CH₂CH₂CH₂)], to the electronic properties of the RO ligand as follows: (i) for RO = (CF₃)₂(CH₃)CO (the most electron-withdrawing ligand), only the TBP isomer is observed, and the complex is very reactive; (ii) for RO = (CF₃)(CH₃)₂CO, a mixture of TBP (a few %) and SBP isomers is observed, and they display intermediate reactivity; and (iii) for RO = (CH₃)₃CO, only the SBP isomer is observed, with very low reactivity because of the high stability of the metallacyclobutane (21, 22). In the case of the active silica supported system reported here, for which the two RO ligands have been replaced by one large siloxy surface ligand and one 2,5-dimethylpyrrolyl ligand, both isomers are observed in roughly equal amounts, and they show good activity (see above). Modulating the ligands in the future by probing the ratio of metallacyclobutane isomers and comparing reactivity opens up a way to access structure-reactivity relationships for heterogeneous alkene metathesis catalysts directly from reaction intermediates.

Moreover, contacting this sample with propene in a flow reactor under the same reaction conditions as those described for [2/SiO_{2- (700)}] display activities (0.51 s⁻¹) similar to that observed for the original catalyst (0.41 s⁻¹ for [2/SiO_{2- (700)}]) (Fig. S6), albeit with a slightly lower stability. The fact that the methylidene and metallacyclobutane species display similar activity with the well defined silica supported catalyst precursor 1 bearing a neopentylidene ligand clearly suggests that these species are indeed the key reaction intermediates in the alkene metathesis pathway.

Conclusions

We have shown that (i) a well defined silica supported W imido alkylidene surface complex [(≡SiO)W(≡NAr)(=CHtBu)(2,5-Me₂NC₄H₂)] (compound 1), prepared and characterized at a molecular level, displays a good activity, selectivity, and stability in the metathesis of propene; (ii) contacting 1 with ¹³C labeled ethene yields stable methylidene and metallacyclobutane intermediates, resulting from cross metathesis and [2 + 2]-cycloaddition (16, 41, 42); and (iii) these species are indeed the key intermediates of the catalytic cycle of alkene metathesis. Overall, these results serve as further evidence that well defined silica-supported systems do behave in a manner similar to homogeneous systems. Direct observation by NMR of these intermediates and their associated spectroscopic signatures is the first step toward a better molecular understanding of heterogeneous olefin metathesis catalysts, and further closes the gap between homogeneous and heterogeneous catalysis. Finally, the design of more efficient catalytic systems may be possible as more information of catalytic intermediates and their stability becomes available (2, 3).

Methods

General Procedures.

All experiments were carried out under dry and oxygen-free Ar using either standard Schlenk or glove box techniques for the organometallic synthesis. For the synthesis and treatment of surface species, reactions were carried out using high vacuum lines (1.34 Pa) and glove box techniques. [W(≡NAr)(=CHtBu)(2,5-Me₂NC₄H₂)₂] (Ar = 2,6-iPr₂C₆H₃) (compound 2) was prepared as described previously (34). [(1-¹³C 33%) W(≡NAr)(=*CHtBu)(2,5-Me₂NC₄H₂)₂] (compound 2*) was synthesized as described previously (9) using [(1-¹³C 33%) tBu¹³CH₂MgCl] as an alkylating agent. [(1-¹³C 33%) tBu¹³CH₂MgCl] was prepared by mixing (1-¹³C 99%) tBu¹³CH₂Cl and natural abundance tBuCH₂Cl followed by the Grignard reaction. Silica (200 m²/g; Aerosil Degussa) was compacted with distilled water, calcined at 500°C under air for 2 h and treated under vacuum (1.34 Pa) at 500°C for 12 h and then at 700°C for 4 h (support referred to as SiO_{2- (700)}). Benzene and C₆D₆ (SDS) were distilled from sodium benzophenone ketyl. Propene (99.95%; Scott) and ¹³C di-labeled ethene (99% ¹³C; CIL) were purified over R3–11 BASF catalyst/MS 4 Å before use. Cp₂Fe (98%; Aldrich) was used as received. Elemental analyses were performed at the Mikroanalythisches Labor Pascher. Infrared spectra were recorded on a Nicolet 550-FT by using an infrared cell equipped with CaF₂ windows, allowing in situ studies. Typically, 16 scans were accumulated for each spectrum (resolution, 2 cm⁻¹). Products were identified by gas chromatography (GC)/MS. Gas phase analyses were performed on a Hewlett Packard 5890 series II GC apparatus equipped with a flame ionization detector and a KCl/Al₂O₂ column (50 m × 0.32 mm).

NMR Spectroscopy.

Liquid state NMR spectroscopy.

Liquid state NMR spectra were recorded in C₆D₆ using a Bruker DRX 500 spectrometer and referenced to the residual protonated solvent peaks (δ_H = 7.15 ppm). The amount of 2,5-dimethylpyrrole released during grafting was monitored by quantitative ¹H NMR with spin presaturation using Cp₂Fe as an internal standard.

Solid state NMR spectroscopy.

All solid state NMR spectra were recorded under MAS on a Bruker Avance or Avance II 500 MHz spectrometer with a conventional double-resonance, 4-mm CP MAS probe. The MAS frequency was set to 12.5 kHz for all of the experiments reported here. The proton radiofrequency (RF) field strength was set to 83 kHz for pulses and 100 kHz during decoupling. For the CP step, the RF field was set to 70 kHz, and for carbon, a ramped RF field was applied on protons and matched to obtain optimal signals. All other experimental details are given in the figure legends. ¹H delayed acquisition spectra including ¹H constant-time spectra were recorded without homonuclear decoupling schemes (36). SPINAL64-¹H (45) heteronuclear decoupling was used in all ¹³C experiments. An exponential line broadening of 80 Hz was applied before Fourier transform of all ¹³C experiments. eDUMBO-1_12.5 (44) homonuclear dipolar decoupling was used for the HETCOR experiment presented in Fig. 1. The experimental scaling factor (λ_exp) of 0.54 was obtained by comparing the proton spectra of l-alanine recorded under fast MAS conditions (30 kHz) and under eDUMBO-1_12.5 decoupling as described previously (44). The samples were filled in a 4-mm zirconia rotor in a glove box and closed with tightly fitting caps. Proton and carbon chemical shifts are reported in ppm downfield from liquid SiMe₄ (0 ppm).

Grafting of [W(≡NAr)(=CHtBu)(2,5-Me₂NC₄H₂)₂] (compound 2) Monitored by in Situ IR Spectroscopy.

Silica (55 mg) was pressed into a 18-mm self-supporting disk and put into a sealed glass high vacuum reactor equipped with CaF₂ windows. After calcination at 500°C under air for 2 h, the silica disk was treated under vacuum (1.34 Pa) at 500°C for 12 h and then at 700°C for 4 h. The silica support thus obtained (SiO_{2- (700)}, ≈15 μmol SiOH) was then immersed into a benzene (10 ml) solution of [W(≡NAr)(=CHtBu)(2,5-Me₂NC₄H₂)₂] (2) (28 mg, 45 μmol, 3.1 eq) at 25°C for 12 h. After washing the pellet three times with benzene (10 ml) and drying under vacuum (1.34 Pa) at 25°C for 2 h, an IR spectrum of the yellow disk was recorded: 3,066, 3,031, 2,966, 2,936, 2,906, 2,873, 1,613, 1,589, 1,578, 1,473, 1,463, 1,433, 1,387, 1,363 and 1,351 cm⁻¹ (Fig. S1). All of the wash solutions were collected and analyzed by ¹H NMR spectroscopy (in C₆D₆), which showed that 2,5-dimethylpyrrole was formed during grafting.

Grafting of [W(≡NAr)(=CHtBu)(2,5-Me₂NC₄H₂)₂] (2) by Impregnation.

A mixture of [W(≡NAr)(=CHtBu)(2,5-Me₂NC₄H₂)₂] (2) (61 mg, 0.10 mmol) and SiO_{2- (700)} (421 mg, 0.11 mmol SiOH) in benzene (12 ml) was stirred at 25°C for 3 h. After filtration, the volatiles were distilled off, and benzene (≈8 ml) was added. The yellow–orange solid was then washed, filtered and the volatiles distilled off. This was repeated three times, and then the resulting yellow powder was dried under vacuum (1.34 Pa) at 25°C for 2 h to yield [2/SiO_{2- (700)}]. All of the filtrate solutions were collected and analyzed by quantitative ¹H NMR spectroscopy (in C₆D₆) using Cp₂Fe (13.3 mg, 71.5 μmol, 1 H) as an internal standard, and 27.4 μmol of 2,5-dimethylpyrrole was formed during grafting (0.9 eq of 2,5-dimethylpyrrole/W_surf). Elemental analysis: 3.73%_wt W, 5.70%_wt C, 0.71%_wt N, 24 ± 2 C/Mo, 2.5 ± 0.5 N/Mo.

Grafting of [W(≡NAr)(=*CHtBu)(2,5-Me₂NC₄H₂)₂] (compound 2*) by Impregnation.

This reaction was carried out as described above by using [W(≡NAr)(=*CHtBu)(2,5-Me₂NC₄H₂)₂] (compound 2*) (33 mg, 0.05 mmol), SiO_{2- (700)} (232 mg, 0.06 mmol SiOH) and benzene (7 ml), thus yielding a yellow powder, and 44 μmol of 2,5-dimethylpyrrole was formed during grafting (0.9 eq of 2,5-dimethylpyrrole/W_surf).

Reaction of ¹³C Dilabeled Ethene with [2/SiO_{2- (700)}].

To [2/SiO_{2- (700)}] (52 mg, 10.6 μmol W) was added ¹³C dilabeled ethene (65 Torr, 140 μmol, 13 eq), while maintaining the sample at liquid nitrogen temperature. The sample was then brought to room temperature, and its color changed from yellow to red. After 4 h, analysis by gas chromatography indicated the formation of 3.9 μmol of 3,3-dimethylbutene (0.37 eq) with the following isotopomeric composition: 3,3-dimethylbutene (5 ± 5%), 3,3-dimethylbutene-¹³C₁ (95 ± 5%), 3,3-dimethylbutene-¹³C₂ (<1%) and 3,3-dimethylbutene-¹³C₃ (< 1%). The red powder was then dried under vacuum (1.34 Pa) at 25°C for 30 min to yield an orange solid as a mixture of [(≡SiO)W(≡NAr)(=*CH₂)(2,5-Me₂NC₄H₂)] (compound 4) and [(≡SiO)W(≡NAr)(*CH₂*CH₂*CH₂)(2,5-Me₂NC₄H₂)₂] (compound 5).

Propene Metathesis in a Flow Reactor.

The solid catalyst (40 mg, 8.1 μmol) was loaded in a flow reactor in the glove box, the isolated reaction chamber was then connected to the propene line, the propene pressure was set to 1 bar, and the tubes were flushed with propene for 2 h. The flow rate was set to 82 ml/min (6.6 mol of propene per mol of Mo per second), the temperature was set to 30°C and the opening of the valve corresponds to the beginning of the catalysis. The reaction was monitored by GC using an autosampler.