Structure of copper sites in zeolites examined by Fourier and wavelet transform analysis of EXAFS†

Copper-exchanged zeolites are a class of redox-active materials that find application in the selective catalytic reduction of exhaust gases of diesel vehicles and, more recently, the selective oxidation of methane to methanol. However, the structure of the active copper-oxo species present in zeolites under oxidative environments is still a subject of debate. Herein, we make a comprehensive study of copper species in copper-exchanged zeolites with MOR, MFI, BEA, and FAU frameworks and for different Si/Al ratios and copper loadings using X-ray absorption spectroscopy. Only obtaining high quality EXAFS data, collected at large k-values and measured under cryogenic conditions, in combination with wavelet transform analysis enables the discrimination between the copper-oxo species having different structures. The zeolite topology strongly affects the copper speciation, ranging from monomeric copper species to copper-oxo clusters, hosted in zeolites of different topologies. In contrast, the variation of the Si/Al ratio or copper loading in mordenite does not lead to significant differences in XAS spectra, suggesting that a change, if any, in the structure of copper species in these materials is not distinguishable by EXAFS.


Computational Details
All ground-state total energy calculations in this work have been performed with the allelectron full-potential DFT code FHI-aims (S1, S2) within the periodic boundary conditions model. Electronic exchange and correlation was treated on the hybrid functional level with the PBE0 functional (S3). All geometry optimization were done with the "tier2" atomcentered basis set using "tight" settings for numerical integrations. Tkatchenko-Scheffler dispersion correction (S4) has been used to account for the van der Waals energies arising from the attraction between induced dipoles formed due to charge fluctuations in the interacting species. Mordenite geometries reported herein correspond to the locally optimized configurations. Faujasite geometries correspond to locally optimized configurations under the assumption that the copper oxide cluster inside of the pore exhibits the C 2v symmetry, consistent with the experimentally observed bond lengths distribution.
Mordenite centers correspond to the periodic model of the mordenite structure, having 8-and 12-ring channels running parallel to the c axis, which are intersected by sinusoidal 8ring channels that run parallel to the b axis. These form side pockets that connect the 12-ring channel with the 8-ring channel. There are two 12-ring and two 8-ring channels per unit cell of mordenite. Two symmetrically located aluminium atoms per 8-ring channel have been assumed, which corresponds to the Si/Al ratio of 11.
Faujasite centers correspond to the periodic model, composed of sodalite cages with diameter of 6.6 Å connected to supercages having a diameter of 12.4 Å. These two units are interconnected by hexagonal prisms whose opening is of 2.3 Å. These supercages are linked together by a 12MR ring with diameter of 7.4 Å, forming the porous accessible network.

Wavelet transform details
For the continuous wavelet transform of EXAFS data Morlet mother function was used: , ( ) = 1 4 ( 2 (2)) ( - which corresponds to the frequency ω = and bandwidth σ 2 = 1. The used    Table S1. The obtained data points to the negligible contribution of multiple scattering paths in the range of 1.0-3.3 Å. For the atoms of a particular nature, having a low to moderate Z number, there is a single main maximum on the curve describing backscattering factor dependence on k value. This is indeed the case for the isolated atoms. However, in case, multiple non-equivalent atoms of similar nature are located at similar distances, the addition of backscattered waves can lead to destructive interference, which results in the appearance of several maxima on the corresponding magnitude curve. As an example, the magnitude of the wave, obtained by a combination of two backscattering waves from copper atoms in copper oxide is shown. It is clear, that those two atoms individually have magnitudes with one maximum. However, the sum of these waves results in beating and the appearance of two maxima on the magnitude curve and, respectively, in the WT counterplots.          Table S1. Best-fit parameters optimized by EXAFS fits of the k 3 -weighted spectrum of standard CuO. The tenorite crystallographic structure was used to generate scattering paths. Degeneracies were fixed according to the crystal structure. Other parameters, including Debye-Waller factors and effective radii were released. The fit was performed in R-space in the range of 1.0-3.3 Å, employing the k-range of 3.0-16.4 Å −1 for the FT, resulting in a number of independent points N ind > 19.  Table S2. Best-fit parameters optimized by EXAFS fits of the k 3 -weighted spectrum of activated copper-exchanged MOR and MFI. The fit obtained for Cu(2.7)FAU(15) with fixed atom coordinates was used as the starting guess. Other parameters, including Debye-Waller factors and amplitudes were released. The fit was performed in R-space in the range of 1.0-3.2 Å, employing the k-range of 3.0-16.0 Å −1 for the FT, resulting in a number of independent points N ind > 16.
Parameter Scatter Cu(4.3)MOR(6) Cu(4.0)MFI (12) ΔE, eV +4.9(9) +4.6(8) The obtained fits returned negative values for the second copper atom in the second coordination sphere, together with unrealistic coordination numbers for aluminum, indicating the absence of this scattering pathway in copper-exchanged MFI and MOR, hence confirming the different nature of copper-oxo species in zeolites of different structure.