Control of zeolite microenvironment for propene synthesis from methanol

Optimising the balance between propene selectivity, propene/ethene ratio and catalytic stability and unravelling the explicit mechanism on formation of the first carbon–carbon bond are challenging goals of great importance in state-of-the-art methanol-to-olefin (MTO) research. We report a strategy to finely control the nature of active sites within the pores of commercial MFI-zeolites by incorporating tantalum(V) and aluminium(III) centres into the framework. The resultant TaAlS-1 zeolite exhibits simultaneously remarkable propene selectivity (51%), propene/ethene ratio (8.3) and catalytic stability (>50 h) at full methanol conversion. In situ synchrotron X-ray powder diffraction, X-ray absorption spectroscopy and inelastic neutron scattering coupled with DFT calculations reveal that the first carbon–carbon bond is formed between an activated methanol molecule and a trimethyloxonium intermediate. The unprecedented cooperativity between tantalum(V) and Brønsted acid sites creates an optimal microenvironment for efficient conversion of methanol and thus greatly promotes the application of zeolites in the sustainable manufacturing of light olefins.


Catalyst characterisation
Powder X-ray diffraction (PXRD) patterns were recorded on a Philips X'pert X-ray diffractometer (40 kV and 30 mA) using Cu Kα1 radiation (λ = 1.5406 Å). N2 adsorption was carried out at 77 K on a Micromeritics 3Flex instrument after activating the samples for 10 h under dynamic vacuum at 623 K. The crystal morphology and size were measured by scanning electron microscopy (SEM) on a Quanta FEG 650 microscope. The ratios of Ta/Al/Si in the sample were quantified by EDX using multiple regions over a sample on a Bruker XTrace instrument. Thermogravimetric analysis was carried out with a SDTQ600 TA instrument. Samples were heated from room temperature to 800 °C at a rate of 10 °C min -1 under an air flow at 100 mL min −1 . Attenuated total internal reflection Flourier transformed infrared (ATR-FTIR) spectra were collected with Nicolet iS5 spectrometer. The acidity was measured by temperature-programmed desorption of ammonia (NH3-TPD) with a Quantachrome Autosorb-1 equipped with a thermal conductivity detector (TCD). Typically, 100 mg of sample was pre-treated in a helium stream (30 mL min −1 ) at 600 °C for 2 h. The adsorption of NH3 was carried out at 50 °C for 1 h. The catalyst was flushed with helium at 100 °C for 2 h to remove physisorbed NH3 from the catalyst surface. The TPD profile was recorded at a heating rate of 10 °C min -1 from 100 to 600 °C. Used catalysts were calcined at 600 °C for 6 h in air flow prior to the NH3-TPD experiments. The Brønsted and Lewis acid sites of the samples were investigated by FT-IR of adsorbed pyridine in an in situ cell with CaF2 windows. Wafers with a weight of 25 mg and radius of 6.5 mm were degassed for 1 h under vacuum at 600 °C. Then pyridine was admitted, and after equilibration, the samples were outgassed for 0.5 h at increasing temperatures (150, 200, 250, 350 and 450 °C). The spectra were recorded on a Nicolet iS50 FT-IR spectrometer. Solid-state 29 Si NMR spectra were recorded with a frequency of 79. 46 MHz, a spinning rate of 10.0 kHz, and a recycling delay of 10 s. Tetramethylsilane was used as the reference for the chemical shift.
For the measurement of electron paramagnetic resonance (EPR) spectroscopy, the sample was placed in a 2.8 mm o.d. quartz tube and connected to a vacuum line. Prior to irradiation, zeolite materials were activated by pumping at 10 -5 Torr for 12 h at 150 ˚C and then flame sealed under vacuum. The samples were exposed to γ-irradiation from a 60 Co source at 77 K to a total dose of 4.1 MRad at a dose rate of 0.48 MRad h -1 . CW EPR measurements were carried out at the X-band frequency (9.4 GHz) using a Bruker EMX spectrometer equipped with an Oxford Instruments temperature control system at 77 K. EPR spectra were detected with modulation amplitudes of 0.2 and 1 mT, and microwave powers varied in the range ~0.  mW. For the data presented here, 7 to 70 mW was chosen to provide optimum signal intensity. Pulsed EPR measurements of samples after irradiation were performed at the X-band frequency (ca. 9.7 GHz) on a Bruker Elexsys E580 spectrometer. The microwave frequency was measured with a built-in digital counter and the magnetic field was calibrated using the microwave pulse the sequence (π/2 − τ − π − τecho) with π/2 and π pulse length of 16 and 32 ns. The interpulse delay τ was 150 ns. HYSCORE 1 spectra were measured at X-band using a pulse sequence (π/2 -τπ/2 -T1 -π -T2 -π/2 -τecho) with π/2 = 16 ns and π = 26 ns, and fixed τ = 136 ns or τ = 200 ns; The initial values of T1 and T2 were 100 ns. The (128×128) HYSCORE data array was recorded with the time increment of 16 ns, and then, data were Fourier transformed (FT) to produce 2D spectra. Theoretical modelling of all EPR data was performed using EasySpin toolbox (Version 5.2.28) for Matlab.
The experiments of TP-MS were carried out in a stainless-steel continuous-flow reactor (6.35 mm i.d.). 0.33 g of TMO@TaAlS-1(0.013/0.027/1) or TaAlS-1(0.013/0.027/1) was loaded in the reactor, followed by injection of 0.3 mL of CD3OD or CH3OH into the reactor. The reactor temperature was increased from room temperature to 400 °C at 10 °C min -1 under a flow of helium at 100 mL min -1 . The efflux was monitored by online mass spectroscopy.

DFT calculations and modelling of the INS spectra
Simulation of INS spectra of bare TaAlS-1 zeolite and TMO species. Periodic density functional theory (periodic-DFT) calculations were carried out using the plane wave pseudopotential method as implemented in the CASTEP code 3,4 . Exchange and correlation were approximated using the Perdew-Burke-Ernzerhof (PBE) functional 5 . Ultra-soft pseudopotentials were employed to account for the effects of core electrons.
The Tkatchenko-Scheffler dispersion correction 6 was used for van der Waals interactions. The Energy cutoff for the plane-wave basis set was 380 eV. Phonon frequencies were obtained by diagonalisation of dynamical matrices computed using the finite displacement method. The atomic displacements in each mode that are part of the CASTEP output, enable visualization of the modes to aid assignments and are also all that is required to generate the INS spectrum using the program OCLIMAX 7,8 . DFT calculations of the INS spectrum for single trimethyloxonium tetrafluoroborate molecule was carried out, which were used to identify the modes of vibrational features in the experimental INS spectrum. In addition, DFT calculations of the INS spectra for bare TaAlS-1 were performed to investigate the distribution of Ta/Al/H sites. Due to the large number of configurations involved in these simulations, the CP2K code 9 based on the mixed Gaussian and plane-wave scheme 10 and the Quickstep module 11 were used. The calculation used the molecularly optimised Double-Zeta-Valence plus Polarization (DZVP) basis set 12 , Goedecker-Teter-Hutter pseudopotentials 13 , and the PBE exchange correlation functional 5 . The plane-wave energy cutoff was 400 Ry. The DFT-D3 level correction for dispersion interactions, as implemented by Grimme et al 14 , was applied.
Simulation of the INS spectrum of solid methanol. Periodic-DFT calculations were carried out using CASTEP (version 7.03) with the PBE functional. Norm conserving pseudopotentials with a plane wave cutoff of 1000 eV were employed, with a Monkhorst-Pack grid of 10×10×5 (75 k-points). After geometry optimisation, the residual forces were converged to better than |0.004| eV Å -1 . Phonon frequencies were obtained by diagonalization of dynamical matrices computed using density-functional perturbation theory 15 .
In addition to the calculation of transition energies and intensities at zero wavevector, phonon dispersion was also calculated along high symmetry directions throughout the Brillouin zone. For this purpose, dynamical matrices were computed on a regular grid of wavevectors throughout the Brillouin zone and Fourier interpolation was used to extend the computed grid to the desired fine set of points along the high-symmetry paths 16 . The INS spectrum was calculated from the atomic displacements in each mode using the program aCLIMAX 8 .
Simulation of the INS spectra of dimethyl ether and propene. Optimised geometry and vibrational frequencies of the lowest energy conformer of each molecule were calculated at the B3LYP/6-311++G(d,p) level of theory with the help of Gaussian suite of programs. INS spectra were obtained from calculated eigenvectors and eigenenergies using the aCLIMAX programme 8 , and were compared with the experimental data.

Supplementary Note 1 EPR spectroscopy
The existence of Ta-O • -Si defects after γ-irradiation was previously proposed in a Ta-modified MCM-41 zeolite 17 but the hyperfine structure associated with 181 Ta has not been resolved for that material. In the present work, γ-irradiated TaS-1 and TaAlS-1 samples show more detailed features. The spectra were simulated taking into account the hyperfine structure due to interaction of the electron with 181 Ta nuclei (I = 7/2; 99.98% natural abundance) and the considerable quadrupole interaction, which is expected to be observed in the continuous wave (CW) EPR spectrum due to the large quadrupole moment of Ta (  Table 3). The average g-value in our work is in good agreement with literature data for M-O • -Si defects 17,18 . For HZSM-5, EPR spectroscopy reveals interaction of an electron hole with Al(III) within the zeolite structure and the g and A parameters obtained are similar to those reported by Wichterlová et al. 18 . For consistency, EPR spectra of samples TaS-1, TaAlS-1 and HZSM-5 were simulated by superposition of the signals for M-O • -Si (M=Ta or Al) and Si-O  -Si defects, and the resulted simulation parameters are presented in Supplementary Table 3. EPR-active defect sites induced by γ-irradiation in HZSM-5, TaAlS-1 and TaS-1 were further analysed by hyperfine sub-level correlation (HYSCORE) spectoscopy 19  Additionally, we observe off diagonal cross peaks due to HF coupling with 29 Si (cyan colour in simulations in Supplementary Fig. 5) which are unresolved in the CW spectrum of the HZSM-5 sample. In the (+,+) quadrant, we observe peaks correspoding centered at the 29 Si larmor frequency. TaAlS-1 clearly demonstrates the presence of the same characreristic off diagonal cross peaks for 27 Al with a slight deviation in HFI values from those previously observed in HZSM-5 (A-tensor = [18.5 18.5 21.0] MHz). The latter confirms the presence of Al-O  -Si sites in the TaAlS-1. The appearance of the 27 Al larmor frequecy diagonal peak in the (+,+) quadrant supports this analysis. Off diagonal cross peaks due to HF interaction with 29 Si observed in the (-,+) quadrant deviates from those observed for HZSM-5 and therefore could refer to Ta-O  -Si sites, which is expected to display a stronger signals in Pulsed EPR due to longer relaxation times for such defects.
Supplementary Note 2 Interaction between methanol and Ta/Al/H sites Synchrotron XPD has been successfully applied to study the interaction between guest molecules and porous materials, such as the binding of hydrocarbons in porous metal-organic frameworks 20 , binding of γvalerolactone in Zn/ZSM-5 21 and the binding of pyridine in HZSM-5 22 . In this study, over 3200 hkl reflections were used for the structural refinement, which allowed for extensive structural variables to be refined in a satisfactory manner. It is almost impossible to distinguish between Al and Si (Z=13 and 14, respectively) sites using X-rays diffraction. Although Ta (Z=73) is heavier, the low occupancy of Ta precludes the determination of its precise location in the framework. Recently, the distribution of active sites in H-ZSM-5 22 and Zn/ZSM-5 21 has been successfully determined by examining the intermolecular distances and angles between guest molecules and framework sites. For example, protonic acid sites of H-ZSM-5 have been located by examining the atomic distances and angles between pyridine and framework atoms 22 .
Methanol can be adsorbed on Brønsted acid sites upon interaction between AlO(H)Si and the OH group of methanol 23 , while it can also be adsorbed on Ta(V) sites via electron donation to the d-band of a Ta(V) site from the OH group of methanol (Fig. 4b). Thus, the distribution of active sites in TaAlS-1 has been revealed by a detailed examination of the MeOH-TaAlS-1 binding distances, particularly that between OMeOH and Ozeolites and between OMeOH and T sites (see Supplementary Tables 12,13). Importantly, the distances between OMeOHII and O6zeolites, and that between OMeOHIV and O18zeolite are shortest in both MeOH-adsorbed TaAlS-1 and HZSM-5, suggesting that protonic acid sites are located at O6-T2 and O18-T9 centres. Similarly, the distances between OMeOHIII and T5, and between OMeOHIII and T21 are shortest in MeOH-adsorbed TaAlS-1 and TaS-1, respectively, indicating that Ta(V) sites are likely located at T5 and T21 positions   (Supplementary Tables 12,13).

Supplementary Note 3 Distribution of Ta/Al/H sites
In order to further study the distribution of Ta/Al/H sites, 29 Si NMR, inelastic neutron scattering (INS) and DFT calculations were employed. The 29 Si NMR spectra of TaAlS-1(0.013/0.027/1) and HZSM-5(0.027/1) show notable differences in the range of -108 to -100 ppm ( Supplementary Fig. 18a). The spectra were deconvoluted by fitting to a sum of Guassian and Lorentzian line shapes. The assignments and peak areas are listed in Supplementary Table 14. The (Al+Ta)/Si ratios calculated from the 29 Si NMR data are consistent with that obtained from EDX. The peak at -98 ppm is absent in the spectrum of HZSM-5(0.027/1), indicating the absence of Si(OSi)2(OAl)2 species (Supplementary Fig. 18b), consistent with previous reports 24,25 . The intensity of the peak at -108 ppm [assigned to Si(OSi)3(OAl/Ta)] in the spectrum of TaAlS-1(0.013/0.027/1) is higher than that in HZSM-5(0.027/1), indicating the successful incorporating of Ta(V) sites into the framework but not adjacent to Al sites (i.e., no "T pairs" sites), which is consistent with the SXPD result ( Fig   3g). The population of the different types of Si (Q n ) of Ta/Al/H sites displayed in Fig. 3g were calculated by zeoTsites software 26 (Supplementary Table 15). The Si(OSi)3(OAl/Ta) is calculated to account for 13% (Supplementary Table 15), which matches the experimental observation (14% , Supplementary Table 14).
Recently, the distribution of protons in LTA zeolite has been studied using periodic DFT calculations and INS 27,28 . The distribution of Ta/Al/H sites in TaAlS-1(0.013/0.027/1) has been further investigated by combining INS and DFT calculations. Based on 29 Si NMR results, only "isolated single" T sites are considered. From the Rietveld refinement of SXPD date (Fig. 3g), one Ta and two Al sites are positioned at T5, T2 and T9, respectively, for the calculation of the corresponding INS spectra (Supplementary Table 16 Fig. 19). As there is an uncertainty on the position of H, each calculated INS spectra of Ta/Al/H_1-16 is given a same weight to assemble the combined spectrum (combination 1 in Supplementary Fig. 20). The spectrum of combination 2 is produced by combining spectra Ta/Al/H_17-19 with a same weight. All peaks in combination 1 are consistent with those in the experimental data, particularly the peaks assigned to -OH in-plane bending and  Table 17). The intensities of peaks below 497 cm -1 (assigned to relaxation of zeolite framework and -OH out-of-plane bending in combination 1) are lower than that of the experimental data because only one unit cell and a simplified model were used and external SiOH groups are not considered in the simulation. Combination 2 is produced with other Ta/Al/H sites ( Supplementary Fig.20), and a number of peaks (53, 148, 267 and 304 cm -1 ) are absent compared with the experimental data. These results does not mean the TaAlS-1 contains various compounds (with the same Ta/Al/H sites used here for modelling), but demonstrating TaAlS-1 (0.013/0.027/1) has an optimal site distribution ( Fig. 3g) that promotes the adsorption of methanol in TaAlS-1 via a "TMO-type" mechanism.

Supplementary Note 4 Inelastic neutron scattering
Direct visualisation of the interaction between adsorbed methanol and the active sites is crucial to understand the molecular details of adsorption, activation and conversion into propene. INS is a powerful neutron spectroscopy technique to investigate the dynamics (particularly for the deformational and conformational modes) of methanol. It has several advantages: • INS spectroscopy is sensitive to the vibrations of hydrogen atoms, and hydrogen is ten times more visible than other elements due to its high neutron scattering cross-section.
• The technique is not subject to any optical selection rules. All vibrations are active and, in principle, measurable.
• INS observations are not restricted to the centre of the Brillouin zone (gamma point) as is the case for optical techniques.
• INS spectra can be readily and accurately modelled: the intensities are proportional to the concentration of elements in the sample and their cross-sections, and the measured INS intensities relate straightforwardly to the associated displacements of the scattering atom. Treatment of background correction is also straightforward.
• Neutrons penetrate deeply into materials and pass readily through the walls of metal containers making neutrons ideal to measure bulk properties of this material (in this case for 11 g catalyst).
• INS data can be collected at low temperature (< 15 K in this case), where the thermal motion of the catalyst, the adsorbed methanol and the reacted intermediate molecules can be significantly reduced.