Imaging spatiotemporal evolution of molecules and active sites in zeolite catalyst during methanol-to-olefins reaction

Direct visualization of spatiotemporal evolution of molecules and active sites during chemical transformation in individual catalyst crystal will accelerate the intuitive understanding of heterogeneous catalysis. So far, widespread imaging techniques can only provide limited information either with large probe molecules or in model catalyst of large size, which are beyond the interests of industrial catalysis. Herein, we demonstrate a feasible deep data approach via synergy of multiscale reaction-diffusion simulation and super-resolution structured illumination microscopy to illustrate the dynamical evolution of spatiotemporal distributions of gas molecules, carbonaceous species and acid sites in SAPO-34 zeolite crystals of several micrometers that are typically used in industrial methanol-to-olefins process. The profound insights into the inadequate utilization of activated acid sites and rapid deactivation are unveiled. The notable elucidation of molecular reaction-diffusion process at the scale of single catalyst crystal via this approach opens an interesting method for mechanism study in materials synthesis and catalysis.

Energy dispersive X-Ray spectroscopy (EDX) was analyzed by a FE-SEM Hitachi SU8020 equipped with a Horiba X-max silicon drift X-ray detector. X-rays fluorescence (XRF) was measured by a Philips Magix-601 spectrometer. NH3 temperature-programmed desorption (NH3-TPD) was carried out with an Autochem 2920 equipment (Micromeritics). In the analysis, all SAPO-34 zeolite samples were activated at 873 K for 1 h under He atmosphere, and cooled down and saturated with NH3 at 373 K for 0.5 h. Then, the samples were purged with He for 0. 5   The boundary condition of catalyst crystal is where a is the crystal size of zeolite and function y represents the adsorption at the external surface from gas phase. and qcoke max the amount of retained carbonaceous species after deactivation (wt %). In our kinetic model, for simplicity, we defined a virtual HCP species that is a lump of active carbonaceous species covering the acid sites. In this way, we assume that 1 mol HCP species would cover 1 mol acid site, and use a virtual molecular weight of acid site of 140 g·mol −1 , i.e. the average molecular weight of HCP species, in the simulations.
The virtual molecular weight, despite being easily implemented in the model, is overestimated. But in the simulations, the acid sites do not appear in the mass balance of hydrocarbon conversions, and the virtual molecular weight of acid sites will not affect the simulation results of the quantity of acid sites for an individual SAPO-34 zeolite crystal. In our simulations, we obtained that the quantity of acid sites for an individual SAPO-34 zeolite crystal is about 1.00 ± 0.06 mmol·gzeo. −1 .
The effect of carbonaceous species on the intracrystalline diffusivity of gas molecules can be related to the formation of HCP species and coke precursors where Ai and Bi (kgzeo.·kg -1 ) are, respectively, the dimensionless quantity of HCP species an coke precursor deposited in SAPO-34 zeolites. Here qHCPs and qcoke are the mass loading of HCP species and coke precursor inside catalyst (kg·kg -1 ).
It is well-known the dual-cycle mechanism can better describe the MTO reaction 3 .
In the Supplementary Table 3 constants against the results of density functional theory (DFT) calculations 8,9,10 . DFT calculations show that the propylene is more likely to be formed than ethylene during olefins-cycle 8,9 , and we indeed found that the reaction kinetic constant of propylene formation is larger than that of ethylene formation in the olefins-cycle. We obtained that the kinetic constant of ethylene formation is larger than that of propylene formation in aromatics-cycle, which are well consistent with the DFT calculations 10 that show that the overall free energy of ethylene formation is lower than that of propylene formation.     13,14 , which were considered the confinement by CHA topology. Bn + , Nn + , PHn + and PYRn + stand for benzenic, naphthalenic, phenanthrenic and pyrenic carbocation with n methyl substituents, respectively.

Supplementary Note 5: Super resolution structured illumination microscopy
A unique feature of SIM fluorescence microscopy is its ability to detect the spatial location of retained carbonaceous species 17 . There are four channels with corresponding wavelengths in SIM, i.e. 405 (detection at 435-485 nm), 488 (detection at 500-545 nm), 561 (detection at 570-640 nm) and 640 nm (detection at 663-738 nm).
These wavelengths can cover the characteristic area of excitation and emission wavelengths of Bn + , Nn + , PHn + and PYRn + as shown in Supplementary The direct comparison between UV-vis spectra and SIM images is quite challenge.
The absorbed and emissive response of a given carbonaceous species might be different even if it is excited at the same wavelength. Figure 2a shows the oscillator strength that represents the absorbance of carbonaceous species to excited wavelength, which manifests an overall decrease from Bn + , Nn + , PHn + to PYRn + . On the meantime, in Figure   2b,