Tracking the phase changes in micelle-based NiGa nanocatalysts for methanol synthesis under activation and working conditions

Tracking heterogeneous structural motifs and the redox behaviour of copper–zinc nanocatalysts for the electrocatalytic CO2 reduction using operando time resolved spectroscopy and machine learning

“…As in our previous works, 35 we validated the accuracy of the designed NN using experimental spectra for pure well-defined reference materials (oxides, metals, and alloys). In Fig.…”

“…This method is built upon ideas from our previous works, where an artificial neural network method was used to “invert” EXAFS spectra for metals and oxides, and directly extract the radial distribution function of neighboring atoms of different types. 19,29,34–36 We rely on the observation that for a mixture of different species, the total EXAFS spectrum χ ( k ) can be expressed as a sum over partial contributions of different species where w s is the concentration of the species, and χ s ( k ) is the partial spectrum for the pure species, which is linked to its structure. For example, in the single scattering approximation, χ s ( k ) can be directly linked to the partial radial distribution function (RDF) g ( R ):Here, the summation is carried out over distinct bonds (more precisely – photoelectron scattering paths).…”

“…As in our previous works, 35 we validated the accuracy of the designed NN using experimental spectra for pure well-defined reference materials (oxides, metals and alloys). In Figure S15ab, we demonstrate that RDFs extracted by NN match well those independently obtained from the same EXAFS data by reverse Monte Carlo simulations [38][39][40] (which can be used for the analysis of pure bulk materials, but not for mixtures).…”

“…This method is built upon ideas from our previous works, where an artificial neural network method was used to "invert" EXAFS spectra for metals and oxides, and directly extract the radial distribution function of neighboring atoms of different types. 19,29,[34][35][36] We rely on the observation that for a mixture of different species, the total EXAFS spectrum 𝜒(𝑘) can be expressed as a sum over partial contributions of different species 𝜒(𝑘) = ∑ 𝑤 𝑠 𝜒 𝑠 (𝑘)…”

“…𝜑 𝑠𝑝 and 𝐴 𝑠𝑝 are the scattering phase and amplitude terms that can be calculated ab-initio for s-th species and p-th path with, e.g., the FEFF code 37 , R is the halflength of the path, therefore related to the interatomic distance between neighbouring atoms, and 𝑆 0 2 is the amplitude reduction factor originating from many-electronic effects. In our previous works, 19,29,35,36 we have shown that 𝑔 𝑠𝑝 (𝑅) functions can be extracted from EXAFS spectra using machine learning. In this process, we first construct large sets of theoretical relevant structure models.…”