Thermocatalytic CO2 Conversion over a Nickel-Loaded Ceria Nanostructured Catalyst: A NAP-XPS Study

Despite the increasing economic incentives and environmental advantages associated to their substitution, carbon-rich fossil fuels are expected to remain as the dominant worldwide source of energy through at least the next two decades and perhaps later. Therefore, both the control and reduction of CO2 emissions have become environmental issues of major concern and big challenges for the international scientific community. Among the proposed strategies to achieve these goals, conversion of CO2 by its reduction into high added value products, such as methane or syngas, has been widely agreed to be the most attractive from the environmental and economic points of view. In the present work, thermocatalytic reduction of CO2 with H2 was studied over a nanostructured ceria-supported nickel catalyst. Ceria nanocubes were employed as support, while the nickel phase was supported by means a surfactant-free controlled chemical precipitation method. The resulting nanocatalyst was characterized in terms of its physicochemical properties, with special attention paid to both surface basicity and reducibility. The nanocatalyst was studied during CO2 reduction by means of Near Ambient Pressure X-ray Photoelectron Spectroscopy (NAP-XPS). Two different catalytic behaviors were observed depending on the reaction temperature. At low temperature, with both Ce and Ni in an oxidized state, CH4 formation was observed, whereas at high temperature above 500 °C, the reverse water gas shift reaction became dominant, with CO and H2O being the main products. NAP-XPS was revealed as a powerful tool to study the behavior of this nanostructured catalyst under reaction conditions.

. Representative (S)TEM images registered for the as-prepared CeO2 NCs sample: (a) TEM image at low magnification and (b) HRTEM image.

Ni 2p Peak Fitting
To fit Ni 2p3/2 spectra for all 5Ni-CeO2 NCs samples, procedure described in reference [1] was followed. Firstly, as-prepared and lower temperature treated samples, which showed a similar peak shape, was fitted with a combination of NiO and Ni(OH)2. For these compounds, reference [1] provides an empirical fitting in its table 1. Shapes generated with a set of peaks proposed by the authors, reproduce the experimental peak shape obtained for standard samples with known composition. In this case, NiO, Ni(OH)2 and NiOOH were chosen, and binding energy differences, intensity ratios and FWHM ratios for the three phases were reproduced as detailed in ref. [1] Table 1. Figure S2 shows the set of 5 peaks for NiO (blue) and 6 peaks for Ni(OH)2 (red) used to fit experimental data. No data for NiOOH is shown here, as no successful fitting was obtained including this phase. Figure 3 in the main text of this paper shows all peaks blended and filled for each phase, to show more clearly the contribution of NiO and Ni(OH)2 to the experimental spectrum. The same applies to Figure S5.
After reducing at 500 °C, and above this temperature, nickel appears to be reduced. In this case, ref. [1] was used to extract data for metallic nickel fitting. Figure S6 shows the result, with a good fitting of Ni 2p3/2 using only Ni(0) contribution.

C 1s and O 1s Peak Fitting for NAP-XPS Data
Peak fitting for C 1s and O 1s was performed using mixed Gaussian-Lorentzian (70-30%) lineshapes, and Shirley type backgrounds.
It is known that Ce 4s signal overlaps with C 1s signals, so its contribution to the total spectra had to be taken into account before performing the fittings. First of all, using a clean sample almost free from adventitious carbon or other carbon contributions, C 1s region was adquired to obtain Ce 4s peak shape. The resulting signal corresponds to Ce 4s contribution, so a peak shape could be modeled using this Ce 4s signal. As data was obtained at two different kinetic energies, 550 eV and 190 eV, and two different samples were used, CeO2 NCs and 5Ni-CeO2 NCs, a set of four different lineshapes were obtained for Ce 4s signal, depending on the KE and the sample used. This shape can be seen in dark red in Figures 7-9.
Then, a set of peaks was used for spectral decomposition, using the parameters summarized in Table S1. FWHM for the peaks are not chosen to be the same for different species appearing in the same spectrum. This is due to the different origin of those signals. All signals coming from gas-phase species were set to the same FWHM. These peaks use to be the narrowest peaks in the spectrum. Also, their BEs are linked, so fixed displacements from CO2 are expected for other gas-phase species.
The rest of C 1s signals are fitted by linking their FWHM, and giving a narrow interval for the peaks to move, according to literature [2].
For O 1s, the nature of the signal corresponding to O 2− and adsorbed species (including OH − ) is different, so FWHM corresponding to O 2− was not linked to them. Finally, all adsorbed O 1s species (including OH − ) were set to the same FWHM, and given a limited margin to move [2]. Notes: * Signals corresponding to gas phase reaction products are given with respect to gaseous CO2 signal position; ** Significant variability in peak positions for gas phase signals are due to charging effects on the sample. Although charging effects were small enough to avoid significant peak broadening and distortion, it however produced a shift for signals coming from the samples and the adsorbed species. Correction of BE scale was done to shift sample peaks in correct positions, so gas phase signals were artificially shifted due to this correction, resulting in the observed variability in their positions, as they were not affected by sample charging; § O 2− signals should include contributions from CeO2 and NiO, that cannot be distinguished; † Signals from adventitious carbon were too weak in most of the cases, so in O 1s core levels could not be detected.