Olefin-Surface Interactions: A Key Activity Parameter in Silica-Supported Olefin Metathesis Catalysts

Molecularly defined and classical heterogeneous Mo-based metathesis catalysts are shown to display distinct and unexpected reactivity patterns for the metathesis of long-chain α-olefins at low temperatures (<100 °C). Catalysts based on supported Mo oxo species, whether prepared via wet impregnation or surface organometallic chemistry (SOMC), exhibit strong activity dependencies on the α-olefin chain length, with slower reaction rates for longer substrate chain lengths. In contrast, molecular and supported Mo alkylidenes are highly active and do not display such dramatic dependence on the chain length. State-of-the-art two-dimensional (2D) solid-state nuclear magnetic resonance (NMR) spectroscopy analyses of postmetathesis catalysts, complemented by Fourier transform infrared (FT-IR) spectroscopy and molecular dynamics calculations, evidence that the activity decrease observed for supported Mo oxo catalysts relates to the strong adsorption of internal olefin metathesis products because of interactions with surface Si–OH groups. Overall, this study shows that in addition to the nature and the number of active sites, the metathesis rates and the overall catalytic performance depend on product desorption, even in the liquid phase with nonpolar substrates. This study further highlights the role of the support and active site composition and dynamics on activity as well as the need for considering adsorption in catalyst design.


General procedures
All experiments were carried out under dry and oxygen-free nitrogen or argon atmosphere using Schlenk techniques or an MBraun or GS glovebox equipped with a purifier unit. Syntheses were carried out using high vacuum lines (10 -5 mbar) and glovebox techniques. Toluene was purified using a double MBraun SPS alumina column. C6H6 and C6D6 were distilled from Na/benzophenone. 1,2-Dichlorobenzene and 1,1,2,2-tetrachloroethane were distilled from calcium hydride. Al2O3 was dried at 250 °C under high vacuum overnight. Molecular sieves were activated at 300 °C under high vacuum overnight. BASF Selexsorb ® CD1/8''was calcined at 550 °C in air for 12 h and heated under high vacuum at 500°С for 2 h. All solvents were degassed by three consecutive freeze-pump-thaw cycles and stored over molecular sieves. The organosilicon reductant 1-methyl-3,6-bis(trimethylsilyl)-1,4-cyclohexadiene (MBTCD) was prepared following literature procedures. [1] Elemental analyses were performed at Mikroanalytisches Labor Pascher, Germany. Solution 1 H NMR spectra were recorded in C6D6 at room temperature using Bruker DRX 250 or DRX 300 NMR spectrometers and were referenced to the 1 H signal from residual C6H6 at 7.16 ppm. All infrared (IR) spectra were recorded using a Bruker α-T spectrometer in an Ar glovebox equipped with OPUS software. A typical IR measurement consisted of acquisition of 32 scans in the region from 4000 to 400 cm -1 .
MoOx@SiO2-red was prepared following an incipient wetness impregnation approach based on literature procedures. [4] Amorphous fumed silica (Aerosil Degussa, 200 m 2 /g) with a pore volume of ca. 1.5 cm 3 /g found by N2 adsorption was calcined under flowing synthetic air (30 standard S5 cubic centimeters per minute, sccm) at 500 °C to remove adsorbed organics. Freshly calcined SiO2 (1 g) in a glass flask was impregnated with 1.5 mL/g of an aqueous solution of (NH4)6Mo7O24·(H2O)4 in milliQ water (55 mg/mL). The (NH4)6Mo7O24·(H2O)4 solution was added in 100 μL increments while stirring with a glass stir rod. The impregnated silica was allowed to dry overnight under ambient conditions, then loaded into a glass flow reactor and heated under flowing synthetic air to 80 °C for 4 h, 110 °C for 4 h, and finally calcined at 500 °C for 4 h (ramp rate 1 °C/min.). The resultant material (MoOx@SiO2) was then evacuated under high vacuum (10 -5 mbar) and stored in a glovebox under inert argon or nitrogen atmosphere prior to the catalytic reaction tests. Elemental analysis of MoOx@SiO2 yielded 3.52 wt% Mo. The IR spectrum of MoOx@SiO2 is shown in Figure S2. Immediately before catalytic reaction tests, MoOx@SiO2red was prepared by reduction of MoOx@SiO2 with 2 equiv. MBTCD per Mo in a benzene solution (0.08 mg/μL).
97% pure by GC, with the impurities being predominantly internal isomers of the terminal olefins.

S6
Note that prolonged exposure to Selexsorb or molecular sieves promotes isomerization of the terminal olefins to their internal isomers. [7] Decalin was passed 3 times through a pad of activated alumina and stored over Selexsorb. 1 M stock solutions of the terminal olefins in 1,2dichlorobenzene with 0.1 M decalin as an internal standard were prepared before the catalytic tests and used immediately.

Catalytic activity measurements
The terminal olefin metathesis reactions were conducted in a nitrogen-filled glovebox using the High-Throughput Experimentation facility at ETH Zürich (HTE@ETH). The catalysts were weighed out into 10 mL reaction vials, heated to the reaction temperature (30 or 70 °C), and shaken at 300 rpm. The reaction vials were either left open over the course of the reaction to allow for ethylene removal or closed with a Teflon plate. 30 μL aliquots of the reaction mixture were taken at the specified time points, diluted with 600 μL toluene, and analyzed by GC/FID. Conversions were calculated from substrate concentration, while product formation rates were calculated from the concentrations of the metathesis products.

Preparation of post-reaction catalysts for spectroscopic analysis
After reaction of (SiO)2Mo(=O)2-red of Mo + /SiO2-700 with 1-hexadecene or 1-nonene for the specified reaction time at 30 °C, the reaction solution was filtered off and the post-reaction catalyst was washed with benzene (3x 1 mL), dried overnight under high vacuum (<10 -5 mbar), and transferred into a glovebox where the dried material was powdered and packed into an NMR rotor.

Preparation of olefin-adsorbed SiO2-700
The adsorption of olefins on silica devoid of Mo centers was also tested by contacting dehydroxylated silica SiO2-700 with olefin solution for 3 h, after which the material was filtered S7 from the olefin solution, washed with benzene (3x 1 mL), and dried overnight under high vacuum (<10 -5 mbar). The olefin solution was either 1 M 1-hexadecene or 0.35 M 15-triacontene, the product of 1-hexadecene metathesis which was purified by vacuum distillation (est. >95% purity by GC).

Solid-state NMR analyses
The 1D and 2D solid-state 1 H and 1 H{ 13 C} NMR spectra were acquired on a 700 MHz (16.4 T) Bruker NMR spectrometer equipped with a 1.3 mm HX MAS probe head and operating at Larmor frequencies of 176.061 and 700.135 MHz for 13 C and 1 H, respectively. The rotors were transferred to the NMR spectrometer in a tightly sealed vial with a screw-cap top, rapidly inserted into the MAS probe head, and spun with dry N2 gas. MAS rates of 50 kHz were used with relaxation delays of 3.5 s, which was ca. 1.4 times the longest measured 1 H spin-lattice T1 relaxation time (measured by 1 H saturation recovery). The measurements used 1 H π/2 pulse lengths of 2 μs (125 kHz) and 13 C π/2 pulse lengths of 3 μs (83 kHz). The 2D 1 H{ 13 C} correlation spectrum was acquired using a 2D dipolar-mediated HMQC sequence with 60 rotor periods of SR4 recoupling [8] to reintroduce the 1 H-13 C dipole-dipole couplings. 64 t1 increments were used in the indirect dimension and 480 transients for a total acquisition time of 30 h. 1 H spin-spin T2 relaxation times were measured using a pseudo-2D CPMG sequence [9] with varied lengths (τ) of rotor-synchronized echo trains.
The 1D and 2D 13 C{ 1 H} and 29 Si{ 1 H} DNP-NMR spectra were acquired on a Bruker Avance III 600 MHz (14.1 T) DNP NMR spectrometer equipped with a low-temperature 3.2 mm tripleresonance MAS cryo-probe operating at Larmor frequencies of 599.900, 150.858, 119.166 MHz for 1 H, 13 C, and 29 Si, respectively. For the DNP-enhanced NMR measurements, (SiO)2Mo(=O)2red was reacted with 1-hexadecene for 4 h, washed and dried as described above, and transferred S8 into a glovebox where the dried material was ground and impregnated with a 16 mM solution of TEKPol biradical [10] in 1,1,2,2-tetrachloroethane. The material was then packed into a sapphire MAS NMR rotor with a zirconia cap, transferred to the NMR spectrometer in a tightly sealed vial with a screw-cap top and rapidly inserted into the pre-cooled (100 K) MAS NMR probe head.
DNP-NMR measurements were acquired at 10 kHz MAS under continuous microwave irradiation at 395 GHz. The microwave on/off DNP enhancements were measured by comparison of the 13 C{ 1 H} CPMAS spectra acquired with and without microwave irradiation and were found to be ca. 4 for the 13 C signals associated with surface-bound organic species. For all DNP-NMR measurements, a relaxation delay of 3 s was used. The 2D 29 Si{ 1 H} DNP-HETCOR spectra in Figures 5 and S20 were acquired with 29 Si-1 H contact times of 0.5 or 5 ms, 512 transients, and 16 or 30 t1 increments in the indirect dimension for total acquisition times of 7 or 13 h, respectively.
The 2D 13 C{ 1 H} DNP-HETCOR spectrum in Figure S19 was acquired with a 13 C-1 H contact times of 2 ms, 64 transients, and 64 t1 increments in the indirect dimension for total acquisition time of 3.5 h. All of the 2D DNP-NMR spectra were acquired with 100 kHz eDUMBO homonuclear decoupling [11] in the indirect 1 H dimension and 100 kHz SPINAL-64 heteronuclear 1 H decoupling [12] during the acquisition period.
The system was then equilibrated at 295 K in the NVT ensemble using a 1 fs for each of the 3000 steps and the temperature was controlled using a 295 K canonical sampling through velocity rescaling (CSVR) thermostat [29] (time step 100 fs). All hydrogen atoms were substituted by deuterium ones in order to allow a longer time step and to prevent artificial bond breaking due to the fast motion of H atoms. Furthermore, a weak external potential of the form 1*10 -12 *((Z-12) 2 ) 4 was added to disfavor desorption of the olefin from the surface. The four lowest energy structures for each olefin found by AIMD were again optimized with the same level of theory in CP2K to obtain optimized geometries of the olefins at the surface. From these four geometries the values in Table S7 were extracted. S10 FTIR Spectra of as-prepared (pre)catalysts

S33
Spectroscopic and elemental analyses of post-reaction catalysts      Discussion of signal assignments for Figure S38 The 1 H signal at 0.0 ppm is assigned to -TMS moieties based on its correlation in Fig. S36 to a 13 C signal at 0 ppm (purple shaded region in Figure S36). The 1 H signals at 0.7 to 2.7 ppm are assigned to aliphatic moieties (green shaded regions in Fig. 5 and S36) based on their correlation to 13 Fig. S36 likely because of their low signal intensity and spectral broadening under the lowtemperature measurement conditions. Nevertheless, the signal assignments are corroborated by the correlation of the 1 H signal at 5.6 ppm to the 13 C signal at 132 ppm in the 1 H{ 13 C} D-HMQC spectrum acquired at room temperature in Figure 4. Finally, the 1 H signals between 7.1 to 8.5 ppm are assigned to aromatic species (e.g., 1,2-dichlorobenzene, blue shaded regions) based on the weak correlation of the 1 H signal at 7.1 ppm to the 13 C signal at 128 ppm in Figure S36. We note that 1 H signals at >7 ppm could also arise in part from strongly hydrogen-bonded or hydrated -OH moieties,  Figure 5b in the main text with an expanded frequency axis showing the signals arising from surface -OTMS moieties).  [30] and fit to stretched exponential fitting functions [31] Figure S40. Transmission FTIR spectra of SiO2-700 (a) as prepared or contacted for 3 hours with (b) 1 M 1hexadecene or (c) 0.3 M 15-triacontene (distilled product of 1-hexadecene metathesis) in dichlorobenzene followed by washing 3x with benzene and drying overnight at 10 -5 mbar. Figure S41. Solid-state 1D 1 H echo MAS NMR spectrum of SiO2-700 contacted for 3 hours with (a) 1 M 1hexadecene or (b) 0.3 M 15-triacontene (distilled product of 1-hexadecene metathesis) in dichlorobenzene followed by washing 3x with benzene and drying overnight at 10 -5 mbar. Figure S42. Transmission FTIR spectra of Mo + /SiO2 (a) before and (b) after 1-hexadecene metathesis at 30 °C for 1 hour, followed by washing 3x with benzene and drying overnight at 10 -5 mbar. Signals at 2320 and 2295 cm -1 from acetonitrile are removed, indicating decoordination of acetonitrile under reaction conditions. Signals in the 1300-1600 cm -1 region are also reduced in intensity likely due to remove of the aromatic -C(Me)2Ph alkylidene moiety on catalyst initiation. Notably, the signals from SiOH species at 3700 and 3615 cm -1 are minimally perturbed after reaction, indicating that these OH sites participate little in olefin metathesis product adsorption.