Impact of Local Structure in Supported CaO Catalysts for Soft-Oxidant-Assisted Methane Coupling Assessed through Ca K-Edge X-ray Absorption Spectroscopy

Soft-oxidant-assisted methane coupling has emerged as a promising pathway to upgrade methane from natural gas sources to high-value commodity chemicals, such as ethylene, at selectivities higher than those associated with oxidative (O2) methane coupling (OCM). To date, few studies have reported investigations into the electronic structure and the microscopic physical structure of catalytic active sites present in the binary metal oxide catalyst systems that are known to be effective for this reaction. Correlating the catalyst activity to specific active site structures and electronic properties is an essential aspect of catalyst design. Here, we used X-ray absorption spectroscopy at the Ca K-edge to ascertain the most probable local environment of Ca in the ZnO-supported Ca oxide catalysts. These catalysts are shown here to be active for N2O-assisted methane coupling (N2O-OCM) and have previously been reported to be active for CO2-assisted methane coupling (CO2-OCM). X-ray absorption near edge structure features at multiple Ca loadings are interpreted through simulated spectra derived from ab initio full multiple scattering calculations. These simulations included consideration of CaO structures organized in multiple spatial arrangements—linear, planar, and cubic—with separate analyses of Ca atoms in the surfaces and bulk of the three-dimensional structures. The morphology of the oxide clusters was found to influence the various regions of the X-ray absorption spectrum differently. Experiment and theory show that for low-Ca-loading catalysts (≤1 mol %), which contain sites particularly active for methane coupling, Ca primarily exists in an oxidized state that is consistent with the coordination environment of Ca ions in one- and two-dimensional clusters. In addition to their unique nanoscale structures, the spectra also indicate that these clusters have varying degrees of undercoordinated surface Ca atoms that could further influence their catalytic activities. The local Ca structure was correlated to methane coupling activity from N2O-OCM and previously reported CO2-OCM reactor studies. This study provides a unique perspective on the relationship between the catalyst physical and electronic structure and active sites for soft-oxidant-assisted methane coupling, which can be used to inform future catalyst development.


Table of Contents
Table S1.Table S3.Best-fit of the EXAFS parameters of 2 mol% Ca/ZnO.Notation: N, coordination number; , amplitude correction term; ΔE 0 , energy correction factor; R, scattering path length; σ 2 , disorder  2 0 term.Values without error bounds were held constant.A k-range of 3.4-9.1 Å -1 and R-range of 1.0-3.5 Å were used.111) plane to create optimal epitaxial contact with the support.The optimized cluster has an average number of Ca-Ca near-neighbors of 4.6 ± 1.2, similar to that observed in the EXAFS, making it relevant for comparison to the experimental results.The final optimized structure expands laterally with respect to the initial one to release virtually all the epitaxial contraction, resulting in an average bond distance of 3.40 Å, nearly identical to the bulk CaO distance of 3.39 Å at the same level of theory.This results in a bond strain of 1.00 ± 0.01.Moreover, the Ca-Ca shell distribution is narrow, with a static mean square relative displacement (MSRD) of 3.6×10 -3 Å 2 .The Ca-Zn shell has an average bond distance of 3.43 Å, thus lying in the range of the Ca-Ca one.The average Ca-Zn number of near-neighbors is 3.0 ± 0.0.However, the Ca-Zn path static disorder is very large, with an MSRD of 22.4×10 -3 Å 2 .This probably results in the Ca-Zn signal being completely overwhelmed by the Ca-Ca one, thus making the support invisible in the EXAFS.Table S7.Best-fit of the EXAFS parameters.Notation: CN, coordination number; , amplitude  2 0 correction term; ΔE 0 , energy correction factor; R, scattering path length; σ 2 , disorder term.Values without error bounds were held constant.A k-range of 3.4-9.1 Å -1 and R-range of 1.0-3.5 Å were used.

Figure S1 .
Figure S1.Ex situ Ca K-edge XANES spectra characterizing Ca/ZnO catalysts of varying composition collected at room temperature in He.Dashed line to visualize pre-edge feature.

Figure S2 .
Figure S2.The magnitude (blue) and imaginary (red) portion of the k 2 -weighted Fourier transforms of the EXAFS spectra (top left) and q-space function (top right) used to calculate from 2 mol% Ca/ZnO, with fits are shown in dotted lines; k-space (k 2 -weighting) data and fits  2 0 in black and red, respectively (bottom).

Figure S3 .
Figure S3.Magnitude of the Fourier transform of the first two phase-uncorrected single-scattering paths of CaO: Ca-O at bond distance of 2.42 Å and Ca-Ca at 3.42 Å, simulated with FEFF6.The comparison of the simulated paths to 2 mol% Ca/ZnO confidently describes the first 2 peaks of the data.

Figure S4 .
Figure S4.The magnitude and imaginary portion of the k 2 -weighted Fourier transforms of the EXAFS spectra (top left) are shown using solid blue and red lines, respectively and q-space data (top right) shown in red, where fits are shown with dotted lines; k-space (k 2 -weighting) data and fits in black and red, respectively (bottom).

Figure S5 .
Figure S5.The magnitude and imaginary portion of the k 2 -weighted Fourier transforms of the EXAFS spectra (top left) are shown using solid blue and red lines, respectively, and q-space data (top right) of the 0.6 mol% Ca/ZnO sample shown in red, where fits using a model of an isolated calcium atom on ZnO are shown with dotted lines; k-space (k 2 -weighting) data and fits in black and red, respectively (bottom).

Figure S6 .
Figure S6.The magnitude and imaginary portion of the k 2 -weighted Fourier transforms of the EXAFS spectra (top left) are shown using solid blue and red lines, respectively, and q-space data (top right) of the 0.6 mol% Ca/ZnO sample are shown in red, where fits using a model of calcium oxide cluster on ZnO are shown with dotted lines; k-space (k 2 -weighting) data and fits in black and red, respectively (bottom).

Figure S7 .
Figure S7.Simulated k-space oscillations of second nearest neighbor Ca and Zn atom scattering paths compared to experimental data.

Figure S8 .
Figure S8.Continuous Cauchy wavelet transform of k 2 -weighted (left) CaO, (middle) 1% Ca/ZnO, and (right) 0.6% Ca/ZnO EXAFS spectra reveals that the R-space peak centered around 3 Å does not contain any detectable influence from a heavier scattering element.

Figure S9 .
Figure S9.Top-view (top)  and side-view (bottom) of a representative, DFT-relaxed, ZnOsupported CaO cluster cleaved along the O-exposed (111) plane to create optimal epitaxial contact with the support.The optimized cluster has an average number of Ca-Ca near-neighbors of 4.6 ± 1.2, similar to that observed in the EXAFS, making it relevant for comparison to the experimental results.The final optimized structure expands laterally with respect to the initial one to release virtually all the epitaxial contraction, resulting in an average bond distance of 3.40 Å, nearly identical to the bulk CaO distance of 3.39 Å at the same level of theory.This results in a bond strain of 1.00 ± 0.01.Moreover, the Ca-Ca shell distribution is narrow, with a static mean square relative displacement (MSRD) of 3.6×10 -3 Å 2 .The Ca-Zn shell has an average bond distance of 3.43 Å, thus lying in the range of the Ca-Ca one.The average Ca-Zn number of near-neighbors is 3.0 ± 0.0.However, the Ca-Zn path static disorder is very large, with an MSRD of 22.4×10 -3 Å 2 .This probably results in the Ca-Zn signal being completely overwhelmed by the Ca-Ca one, thus making the support invisible in the EXAFS.

Figure S10 .
Figure S10.The magnitude and imaginary portion of the k 2 -weighted Fourier transforms of the EXAFS spectra (top left) are shown using solid blue and red lines, respectively, and q-space data (top right) of the 0.6 mol% Ca/ZnO sample shown in red, where fits using a model of pure calcium hydroxide are shown with dotted lines; k-space (k 2 -weighting) data and fits in black and red, respectively (bottom).

Figure S11 .
Figure S11.The magnitude and imaginary portion of the k 2 -weighted Fourier transforms of the EXAFS spectra (top left) are shown using solid blue and red lines, respectively, and q-space data (top right) of the 0.6 mol% Ca/ZnO sample shown in red, where fits using a model of both calcium hydroxide and calcium oxide are shown with dotted lines; k-space (k 2 -weighting) data and fits in black and red, respectively (bottom).

Figure S12 .
Figure S12.Comparison of experimentally measured Ca K-edge XANES spectrum of CaO reference to simulated XANES spectrum of a Ca atom from within the bulk of a large CaO structure.

Figure S13 .Figure S14 .Figure S15 .
Figure S13.Comparison of experimental spectra of 0.6 mol% Ca/ZnO and bulk CaO reference to simulated spectra of a surface Ca atom of a CaO slab and a bulk Ca atom.

Figure S16 .
Figure S16.Decomposition of the dipole and dipole+quadrupole contributions to the theoretical spectra of (a-b) bulk CaO, (c-d) surface Ca atom, and (e-f) oxygen-terminated surface Ca atom into their s→p and s→d components (a, c, e) without and (b, d, f) with a distortion applied to the central Ca atom of +0.1 Å along the z-axis.

Figure S18 .
Figure S18.X-ray diffractograms of catalysts after N 2 O-OCM.Diffractograms were collected using a Bruker D8 Advanced Diffractometer with Cu Kα radiation.Powder catalysts were sieved through a 304 stainless steel wire cloth disc with a mesh size of 200 to ensure a random distribution of exposed facets, then adhered to silica sample holders with Dow Corning high vacuum grease.

Table S1 .
Table of Abbreviations

Table S4 .
BestΔE 0 , energy correction factor; R, scattering path length; σ 2 , disorder term.Values without error bounds were held constant.A k-range of 3.4-9.1 Å -1 and R-range of 1.0-3.5 Å were used.Both paths were generated from a cif file replacing a single Zn atom within ZnO with Ca.

Table S8 .
Total and regional XANES matching Frechet-distance-like errors between theory and experiment.The Frechet figure of merit (FOM) is defined as: Given a reference function and a  1 target represented by the curve segments with coordinates and , we define the  2 { 1 , 1 } { 1 , 1 } vector of distances from point in to any point in .With these distances we  12 = { 12 }

Table S9 .
Summary of catalytic performance over the range of Ca loadings.