Nature of cobalt species during the in situ sulfurization of Co(Ni)Mo/Al2O3 hydrodesulfurization catalysts
aInorganic Chemistry and Catalysis, Debye Institute for Nanomaterials Science, Utrecht University, Universiteitsweg 99, 3584 CG Utrecht, The Netherlands, bStanford Synchrotron Radiation Lightsource, SLAC National Laboratory, 2575 Sandhill Road, Menlo Park, CA 94025, USA, and cID26, European Synchrotron Radiation Facility (ESRF), 71 Avenue des Martyrs, 38000 Grenoble, France
*Correspondence e-mail: email@example.com
The evolution in local structure and electronic properties of cobalt was investigated during in situ sulfurization. Using a combination of 1s X-ray absorption (XAS) and 1s3p resonant inelastic X-ray scattering (RIXS), the valence, coordination and symmetry of cobalt ions were tracked in two cobalt-promoted molybdenum oxide precursors of the hydrodesulfurization catalyst system, namely Co–Mo/Al2O3 and Co–Ni–Mo/Al2O3. Extended X-ray absorption fine structure shows that the Co—O bonds were replaced with Co—S bonds as a function of reaction temperature. The cobalt K pre-edge intensity shows that the symmetry of cobalt was modified from Co3+ Oh and Co2+ Oh to a Co2+ ion where the inversion symmetry is broken, in agreement with a square-pyramidal site. The 1s3p RIXS data revealed the presence of an intermediate cobalt oxy-sulfide species. This species was not detected from XAS and was determined from the increased information obtained from the 1s3p RIXS data. The cobalt XAS and RIXS data show that nickel has a significant influence on the formation of the cobalt oxy-sulfide intermediate species prior to achieving the fully sulfided state at T > 400°C.
The development of novel hydrodesulfurization (HDS) catalysts has been one of the interesting research subjects in the field of heterogeneous catalysis because of the new environmental regulations to reduce sulfur emissions of transportation fuels (Eijsbouts et al., 2008). Cobalt- and nickel-promoted molybdenum sulfide catalysts supported on porous oxides, such as alumina, have been used since the early years of the previous century. The active phase of this catalyst is assumed to consist of MoS2 slabs with cobalt and/or nickel decorating the edges. There are many factors that can affect the activity of a catalytic solid, e.g. the metal loadings, type of support and especially the synthesis conditions and treatment procedures employed (Ratnasamy et al., 1980; Keely et al., 1984; Topsøe, Clausen, Topsøe & Pedersen et al., 1986; Topsøe & Clausen, 1986; Prins et al., 1989; Breysse et al., 1991; Chianelli et al., 1994; Startsev, 1995). Numerous studies have attempted to explain the catalytic HDS process and describe the role of the involved promoters Co and Ni on the Al2O3-supported MoSx phase. The Co–MoSx structure originally proposed by Topsøe and co-workers and Grunwaldt et al. (2004) was also confirmed by Bouwens et al. (1990, 1991, 1994) and Crajé et al. (1991a,b). The catalytic performance of this active phase is determined by the structure of the oxide precursor species, which is strictly dependent on the preparation methods (Topsøe et al., 1984; Kibsgaard et al., 2010). Therefore, as a starting point it is crucial to understand the processes involved in the synthesis of the catalyst and especially the sulfurization/activation step. In the petrochemical industry, the oxide precursors are activated by the addition of sulfur-containing compounds, e.g. methyl sulfides, to the feedstock. However, usually a mixture of 10% H2S/H2 is used to sulfurize the (supported) oxide precursors. Among the many synthesis steps involved, the sulfurization process is very likely the step that determines the final structure and performance of the active Co–Ni–Mo/Al2O3 catalyst material. Following the analysis of Topsøe and Clausen, we start from the assumption that cobalt can occur in three forms in alumina-supported catalysts: (a) as cobalt ions on the edges of the MoS2 crystallites, (b) as Co9S8 particles and (c) as divalent cobalt ions at the surface or in the bulk of the alumina (Topsøe et al., 1984). Several characterization studies focused on the changes in (cobalt-doped) molybdenum compounds during the sulfidation process (Scheffer et al., 1984; Arnoldy et al., 1985; Nicosia & Prins, 2005; Rochet et al., 2016). Recently Šarić et al. performed detailed density functional theory calculations of the Co–MoSx system (Šarić et al., 2017, 2018). They calculated the structures of the edges, basal plane and corners of the MoS2 slabs and found that the HDS activity is related to the small energy differences between structures with or without a sulfur atom for the corners and the S edges, but not the Mo edges.
In this article, we study the valence, coordination and spin-state changes of cobalt in single (cobalt-) and double (cobalt-nickel-) promoted HDS catalysts, using a combination of 1s X-ray absorption (XAS) and 1s3p resonant inelastic X-ray scattering (RIXS). The 1s core levels of the 3d transition metal ions can be probed with hard X-rays (4–10 keV). In contrast to the main K-edge, which arises from the dipole allowed transition of 1s to 4p, the pre-edge involves mainly 1s3d quadrupole transitions. The K pre-edge energy position and intensity are sensitive to the metal oxidation state, the site symmetry, and the crystal-field splitting. However, because of their relatively low intensity and large core hole lifetime broadening, their analysis is often limited to the energy position and intensity of the pre-edge. In a 1s3p RIXS experiment, the resonant 1s3p X-ray emission is measured by tuning the energy of the incident energy to a 1s X-ray absorption edge. This effectively combines X-ray absorption and X-ray emission into a single experiment (Glatzel et al., 2009; Kotani et al., 2001). The RIXS process can be viewed as an inelastic scattering of the incident photon at a resonance energy of the metal ion and is theoretically described by the Kramers–Heisenberg formula (Sakurai, 1967; Rubensson, 2000). The spectral broadening of 1s XAS is determined by the lifetime of the 1s intermediate state (L1s), whereas the spectral broadening of the 1s3p resonant XES depends on the final state 3p lifetime (L3p) (Glatzel et al., 2005; Hill et al., 1998). Because L1s is larger than L3p, this results in the observation of spectral features on the X-ray emission energy scale with sharper line width than in a conventional X-ray absorption spectrum.
Hard X-rays are able to penetrate solids and probe the bulk phase properties of the samples (van Bokhoven et al., 2004; Rabe et al., 2010; Szlachetko et al., 2013). This allows the determination of the active sites within the bulk of the sample under in situ conditions. In this work, RIXS is used to study the composition of the Co–Mo/Al2O3 and Co–Ni–Mo/Al2O3 HDS catalysts as model systems to reflect the abilities of the RIXS technique to study other heterogeneous catalysts under in situ conditions. The purpose of the present work is to unravel the nature of the cobalt species in the precursor oxide of γ-Al2O3-supported Co–Mo and Co–Ni–Mo catalysts and to follow the changes in the local structure and electronic properties, during the in situ sulfurization with RIXS and XAS.
The Co–Mo/Al2O3 and Co–Ni–Mo/Al2O3 catalyst materials were both synthesized by incipient wetness impregnation of the γ-Al2O3 support (BASF Al-4184, surface area of 157 m2 g−1 and pore volume of 0.75 ml g−1) with a solution of metal salts. Aqueous solutions of ammonium heptamolybdate hexahydrate [AHM, (NH4)6Mo7O24·6H2O, Sigma Aldrich, ≥99.0%], cobalt nitrate hexahydrate [Co(NO3)2·6H2O, Acros Organics, 99+%] for Co–Mo/Al2O3 and additionally nickel nitrate hexahydrate [Ni(NO3)2·6H2O, Sigma-Aldrich, 99.999%] for Co–Ni–Mo/Al2O3 were co-impregnated to obtain weight loadings of 14 wt% for molybdenum and 5 wt% for both cobalt and nickel. Subsequently, both samples were dried at 60°C for 16 h then at 120°C for 1 h. To convert the impregnated metal precursors to their respective metal oxides, the samples were calcined at 450°C for 16 h. All drying and calcination steps were performed under a (75% N2)/(25% O2) flow (1 ml min−1). We note that the atomic Co/Mo ratio of 0.58 is higher than used in the industrial HDS catalysts. We used the increased amount of cobalt and nickel to increase the signal-to-noise of the X-ray spectroscopy experiments.
During the in situ sulfurization experiments both oxide precursor samples were simultaneously sulfurized by loading the samples on the same sample holder. Pellets of Co–Mo/Al2O3 and Co–Ni–Mo/Al2O3 samples were made from the respective powders and loaded on the same sample holder as is shown in Fig. 1. Subsequently, the sample holder was placed inside the heating stage of the reactor. The sample was heated to 450°C with a ramp of 10°C min−1, while flowing with a 10% H2S/H2 mixture (1 ml min−1). The samples were kept at 450°C for 5 h.
The reactor was assembled and sealed and the incident window was covered with Kapton foil to decrease the photon flux and prevent radiation damage and leakage of air into the reactor. The reactor includes connections for both gas inlet and gas outlet. The inlet line is connected with a mass-flow controller that regulates the type and amount of gas that passes through the reactor. On the outlet side, there are bubbler airlocks that can be used to trap the toxic gas mixture. First, the oxide precursors were measured under ambient conditions, followed by in situ sulfurization under dynamic flow of 10% H2S/H2 gas mixture and by slowly heating the samples to 450°C.
The spectra were recorded at beamline 6-2 of the Stanford Synchrotron Radiation Lightsource (SSRL) (Sokaras et al. 2013). The beamline is equipped with two double-crystal monochromators, Si(111) and Si(311). A collimating and a focusing Rh-coated mirror are positioned before and after the monochromator, respectively. The incident energy was selected using the Si(311) crystal during the measurements at the Co K-edge. After the monochromator, the beam is delivered to the Rh-coated focusing mirror and focused both horizontally and vertically to ∼140 µm × 400 µm (V × H) at the sample position. XAS spectra were measured simultaneously in total fluorescence yield (TFY) mode using a photodiode. The sample, crystal analyzer and photon detector were arranged in a vertical Rowland geometry. The Co Kβ (1s3p) RIXS spectra were recorded at a scattering angle of 90° in the horizontal plane using seven analyzer crystals (Bergmann & Cramer, 1998). The total energy resolution (ΔE/E) of the beamline and the spectrometer is ∼0.2 eV. The intensity was normalized to the incident flux. The RIXS data are shown as a contour map in a plane of incident and transferred photon energies, where the vertical axis represents the energy difference between the incident and emitted energies (energy-loss). The variations in colour on the plot correspond to the different scattering intensities.
The RIXS 2D maps were recorded in the excitation range 7704–7715 eV with steps of 0.2 eV, and 55 to 82 eV for the RIXS energy loss, with steps of 0.25 eV. The TFY XAS was also recorded with a photodiode. Radiation-damage studies were performed at ambient conditions by measuring four XAS spectra to check for any shifts in the edge jump energy, and the pre-edge intensity and its energy shift. Additionally, RIXS spectra were measured to reconfirm the results obtained with XAS. For all spectra, there was no change in the measured spectra for a 3 h experiment, confirming the absence of radiation damage.
3.1. Catalytic activity testing
Before presenting the results of the in situ sulfurization, we briefly discuss the catalytic activity of the Co–Mo–S/Al2O3 and Co–Ni–Mo–S/Al2O3 catalysts towards the thiophene desulfurization. The HDS reaction of thiophene was performed using thiophene saturated H2 flow in a single-pass tubular down-flow fixed-bed reactor at the reaction conditions of 320°C. Fig. 2 shows a schematic summary of the performed catalytic studies. Prior to the reaction, the catalyst was reduced and sulfided for 6 h at 450°C at a heating rate of 5°C min−1 under a gas mixture containing 10 mol% H2S in H2 and a flow rate of 1 ml min−1. The initial activity test (the first pulse) was performed under 0.1 µl of thiophene saturated in 10 ml min−1 H2, at a temperature of 320°C. The sample was then stabilized using 1 µl thiophene saturated H2 with a flow rate of 10 ml min−1 at 320°C. Following a stabilization step, the catalytic activity of the samples was measured under similar conditions. The gaseous products were analyzed online on a mass spectrometer through a sampling and a flame ionization detector. Fig. 3 shows the normalized conversions of thiophene by Co–Ni–Mo–S/Al2O3 and Co–Mo–S/Al2O3 during three subsequent HDS steps. Since variation in molybdenum-loading influences the HDS activity, the conversion is normalized for the molybdenum loading of the catalysts. An initial conversion of 27% was observed for the double promoted catalyst and 60% for the single promoted catalyst material. The thiophene conversion data were normalized to the Mo loading.
Because of the higher ratio of promoters to molybdenum in Co–Ni–Mo–S/Al2O3, an increase in the formation of bulk cobalt- and nickel sulfide phases is expected, resulting in a reduced catalytic activity. The lower conversions observed for Co–Ni–Mo–S/Al2O3 can be explained by a decrease in the number of promoter atoms on the edges of MoS2 slabs. In addition, as shown in Fig. 3, an increase in the conversion rate was observed for the subsequent thiophene pulses. This increase is caused by the correlation between the dispersion, degree of sulfidation of active phase and the HDS activity. Note that, because of the thiophene treatment steps, the degree of sulfidation and dispersion of the active phase is increased.
3.2. XANES and EXAFS analysis
The X-ray absorption spectra were normalized according to the procedure described in the supporting information. Fig. 4 shows the normalized cobalt K-edge X-ray absorption near-edge (XANES) region as a function of temperature. The cobalt K-edge spectrum of freshly calcined Co–Mo/Al2O3 is characterized by an intense main edge (1s → 4p) and a weak pre-edge feature with a main peak at 7709 eV followed by a weak shoulder at 7711.4 eV. This finding is similar to the previously reported Co3O4 reference (al Samarai et al., 2016). Exposing both samples to the gas mixture at room temperature results in a significant shift of the main edge to lower energies implying a partial reduction of cobalt species accompanied by a decrease in the intensity of the pre-edge feature. In addition, the pre-edge intensity grows dramatically and the spectral shape is modified for both catalysts.
Upon heating the catalysts to higher reaction temperatures, the main edge shifts to lower energies, indicating a further reduction. In Fig. 5 the normalized intensity of the pre-edge of Co–Mo/Al2O3 and Co–Ni–Mo/Al2O3 shows an increase for T < 200°C, while for T > 200°C the pre-edge intensity is not further increased. The details of the fits including all numbers are given in Tables S2 and S3 of the supporting information. The increase in pre-edge intensity is assigned to a reduction in symmetry of Co3+ Oh and Co2+ Oh in the cobalt oxide species to a Co2+ ion in a square-pyramidal cobalt sulfide species for the fully sulfided catalyst sample (Borges et al., 2012; Orita et al., 2004; Paul et al., 2008; Ma & Schobert, 2000). In this case, the crystal field splits the 3d orbitals partly belonging to the same irreducible representation of the point group as metal 4p orbitals, leading to partial mixing and hybridization (Griffith, 1964; Figgis, 1966). The increase in the pre-edge intensity is ascribed to the increase in the degree of the 3d4p mixing/hybridization character of the orbitals. We note that the pre-edge intensity is equivalent to that found in other studies (Nicosia & Prins, 2005; Rochet et al., 2016) and lower than the pre-edge in Co9S8, which contains 50% tetrahedral and 50% octahedral sites. The observed pre-edge intensity indicates that if tetrahedral sites exist they relate to a maximum of 25% of the cobalt sites, the other 75% being octahedral.
In the performed composition analysis (listed in Tables S2 and S3), the Co–Mo/Al2O3 sample shows a higher rate of sulfidation compared with the Co–Ni–Mo/Al2O3 sample (Fig. 8). The Co–Mo/Al2O3 sample shows the complete sulfidation at 190°C, whereas the Co–Ni–Mo/Al2O3 sample is only fully sulfided at 400°C. Furthermore, these results are confirmed by the increase in the integrated pre-edge intensities and the analysis of the extended X-ray absorption fine structure (EXAFS) region during the sulfurization process, which was sensitive to mainly the presence of either Co—O or Co—S bonds. The EXAFS analysis illustrates the considerable impact that the temperature has on the sulfurization process and on the formation of Co—S bonds under the flow of the 10% H2S/H2 gas mixture (Fig. S6). At room temperature, approximately 45% and 50% of the total bonds are converted to Co—S bonds for Co–Mo/Al2O3 and Co–Ni–Mo/Al2O3 samples, respectively. This increase is attributed to gradual ligand substitution and conversion of cobalt ions at Oh sites in the case of the oxide precursor to the (C4v) square-pyramidal cobalt in symmetry for the fully sulfided sample.
A series of cobalt 1s3p RIXS planes were recorded during the in situ sulfurization of Co–Mo/Al2O3 and Co–Ni–Mo/Al2O3. Fig. 6 shows the temperature-dependent 1s3p RIXS spectra of Co–Mo/Al2O3. The equivalent 1s3p RIXS spectra of Co–Ni–Mo/Al2O3 are shown in Section S3. The figure shows several changes during the in situ sulfurization reaction, including (a) the gradual conversion of two distinct pre-edge features to a single emission feature and (b) a shift of the main edge to lower absorption energies.
For every RIXS plane the emission slice at the pre-edge peak position of 7709 eV was measured as shown in Fig. 7. The equivalent graphs of the Co–Ni–Mo/Al2O3 catalyst are shown in Figs. S4 and S5. The oxide precursor has distinctive resonance peaks at 7650.7 eV (peak 1) and 7647.45 eV (peak 2) characteristic of divalent cobalt species (al Samarai et al., 2016). Subsequently, in the following steps, as the sample was exposed to the gas mixture, the RIXS spectra were affected and the resonant emission intensity ratios of the two peaks were reversed with peak 1 at 7648.15 eV and peak 2 at 7650.65 eV. This change can be ascribed to partial sulfidation of cobalt oxide species leading to modification in the symmetry of Co2+ ions.
The RIXS spectra were fit according to a procedure that involved a linear combination of the RIXS data on the freshly calcined sample (assumed to be in its oxide form) and the sample at 400°C for each temperature in addition to a small offset value to account for the counts on the tails of other emission processes. Initially, to fit the RIXS data, the combination of only two components was used to analyse the spectra. However, this proved to be impossible as additional peaks were observed in the experimental spectra during the in situ reaction. Therefore, additional empirical peaks were added to the model and collectively denoted as `intermediate species'. However, we imposed an important constraint to these empirical peaks: the broadening, shape energy position and relative intensities were forced to be the same in all spectral series (for all temperatures) by holistically fitting all spectra. We performed 100 fits in each stage of the in situ reaction using the methodology implemented in Blueprint XAS to explore uncertainty of all fit parameters (Delgado-Jaime et al., 2010; Delgado-Jaime & Kennepohl, 2010). We then selected the best fits on the basis of the lowest sum squared error (SSE).
3.4. Combined analysis of XANES, EXAFS and RIXS
The obtained phase concentrations are shown in Fig. 8. In the case of the Co–Mo/Al2O3 sample the RIXS data analysis revealed the presence of a single intermediate species (in blue) for the lower temperatures (T < 150°C). Because of its spectral shape, we can assume a species consisting mostly of Co—S bonds. However, the fitting of the Co–Ni–Mo/Al2O3 RIXS data was only possible by the inclusion of two intermediate species (blue and purple), which remain present during the in situ reaction for T < 400°C. The shape of one of the species has a predominantly oxide character [blue in (a)], while the second has a rather sulfide character [purple in (b)]. This assumption is based upon their comparison with the spectra of the pure oxide and the pure sulfide with regard to their individual shape as well as the position and intensity of the spectral features. In other words, in Co–Mo/Al2O3 the intermediate is most likely to be the Co—S5O and/or Co—S4O2 oxy-sulfide, which can be distinguished from Co—S6 in some intermediate stages. In Co–Ni–Mo/Al2O3 we see both the Co—S5O and/or Co—S4O2 oxy-sulfide (blue) and also the Co—S1O5 and Co—S2O4 oxy-sulfides (purple).
The Co–Ni–Mo/Al2O3 sample is more sensitive toward sulfidation at room temperature as 60% (versus 30% for Co-Mo/Al2O3) of the oxide species is already converted to either the sulfide or intermediate species under the dynamic flow of the gas mixture at 25°C. The details of the analysis and all numbers are given in Tables S4 and S5. We note that at 400°C oxidic cobalt is no longer detected, implying the absence of divalent cobalt ions in the alumina. In addition, a fraction of the sulfide phase could exist as Co9S8 particles.
During the gradual sulfurization of cobalt in the Co–Mo/Al2O3 and Co–Ni–Mo/Al2O3 samples, an increase in the pre-edge intensity was observed. The increase is caused by the conversion of cobalt species to a species without inversion symmetry, for example in square-pyramidal symmetry. Furthermore, because of the sulfurization process the cobalt was reduced and the Co K main edge was also shifted to lower absorption energies. The EXAFS data analysis showed the gradual ligand exchange during the process of sulfurization. The relative concentration percentages of Co—O and Co—S bonds were determined by fitting the EXAFS data as the linear combination of components at the two extremes. The analysis showed a higher tendency towards sulfurization of cobalt in the case of Co–Mo/Al2O3. In addition, the gradual change in the composition was obtained for the subsequent steps of the in situ reaction.
The Co 1s3p RIXS study has shown a gradual transition of the cobalt oxide in the active phase to the fully sulfided species. One of the findings is that the Co–Mo/Al2O3 catalyst has a higher tendency for the sulfurization reaction and achieves a >90% conversion to the sulfide phase at 190°C, compared with 400°C for the Co–Ni–Mo/Al2O3 catalyst. By analyzing the fits obtained from the RIXS data, the existence of a single intermediate species for T ≤ 150°C with a sulfide-like character was detected for the Co–Mo/Al2O3 catalyst. In the case of the Co–Ni–Mo/Al2O3 catalyst two intermediates were identified that eventually converted to the fully sulfided phase at 400°C. The RIXS data confirmed the previously obtained results by XANES and EXAFS analysis of the two samples. A summary of the proposed cobalt species involved during the in situ sulfurization of the Co–Mo/Al2O3 and Co–Ni–Mo/Al2O3 HDS catalysts is shown in Fig. 9. The two phases found in the fully sulfided phase are not in agreement with the recent studies of Šarić et al. (2017, 2018), who calculate the HDS activity as a difference between tetrahedral cobalt sites. As discussed in Section 3.2, the observed pre-edge intensity indicates that if tetrahedral sites exist they can relate to a maximum of 25% of the cobalt sites, the other 75% being octahedral. The results are in agreement with the alternative of the cobalt having square-pyramidal symmetry.
We conclude that from a combination of cobalt 1s XANES, EXAFS and 1s3p RIXS we derived several boundary conditions regarding the mechanisms of sulfurization in Co–Ni–Mo/Al2O3 catalysts. EXAFS shows that the Co—O bonds are replaced with Co—S bonds as a function of reaction temperature, and the pre-edge intensity shows that the symmetry of cobalt is modified from Co3+ Oh and Co2+ Oh to a Co2+ ion where the inversion symmetry is broken, in agreement with a square-pyramidal site. The cobalt 1s3p RIXS reveals the presence of an intermediate cobalt oxy-sulfide species.
The catalytic data shows a large influence of nickel in the Co–Ni–Mo/Al2O3 catalyst in comparison with Co–Mo/Al2O3. The cobalt XAS and RIXS data show that nickel has a significant influence on the formation of the cobalt oxy-sulfide intermediate species prior to achieving the fully sulfided state.
The supporting information is available as a PDF file containing information on (1) the catalytic activity testing, (2) the XAS data analysis method, (3) the numerical XANES fit results and (4) the 1s3p RIXS images of the CoNiMo catalyst.
‡Present address: Max Planck Institute for Chemical Energy Conversion, Stiftstrasse 34, D-45470 Mulheim, Germany.
The experiments were performed under the approvals of the Stanford Synchrotron Radiation Light source (Proposal No. 4107).
This research is performed with financial support from STW-NWO (Netherlands Organization for Scientific Research with grant number NANO-11507).
Arnoldy, P., van den Heijkant, J. A. M., de Bok, G. D. & Moulijn, J. A., (1985). J. Catal. 92, 35–55. Google Scholar
Bokhoven, J. A. van, Ressler, T., de Groot, F. M. F., Knop-Gericke, A. (2004). In In-situ Spectroscopy of Catalysts, edited by B. M. Weckhuysen. Stevenson Ranch: American Scientific Publishers. Google Scholar
Bergmann, U. & Cramer, S. P. (1998). Proc. SPIE, 3448, 198–209. CrossRef CAS Google Scholar
Borges, I. Jr & Silva, A. M. (2012). J. Braz. Chem. Soc. 23, 1789–1799. CAS Google Scholar
Bouwens, S. M. A. M., Prins, R., De Beer, V. H. J. & Koningsberger, D. C. (1990). J. Phys. Chem. 94, 3711–3718. CrossRef CAS Web of Science Google Scholar
Bouwens, S. M. A. M., Van Veen, J. A. R., Koningsberger, D. C., De Beer, V. H. J. & Prins, R. (1991). J. Phys. Chem. 95, 123–134. CrossRef CAS Google Scholar
Bouwens, S. M. A. M., van Zon, F. B. M., van Dijk, M. P., van der Kraan, A. M., de Beer, V. H. J., van Veen, J. A. R. & Koningsberger, D. C. (1994). J. Catal. 146, 375–395. CrossRef CAS Google Scholar
Breysse, M., Portefaix, J. L. & Vrinat, M. (1991). Catal. Today, 10, 489–505. CrossRef CAS Google Scholar
Chianelli, R. R., Daage, M. & Ledoux, M. (1994). J. Adv. Catal. 40, 177–232. CrossRef CAS Google Scholar
Crajé, M. W. J., de Beer, V. H. J. & van der Kraan, A. M. (1991a). Appl. Catal. 70, L7–L13. Google Scholar
Crajé, M. W. J., De Beer, V. H. J. & van Der Kraan, A. M. (1991b). Bull. Soc. Chim. Belg. 100, 953–960. Google Scholar
Delgado-Jaime, M. U. & Kennepohl, P. (2010). J. Synchrotron Rad. 17, 119–128. Web of Science CrossRef CAS IUCr Journals Google Scholar
Delgado-Jaime, M. U., Mewis, C. P. & Kennepohl, P. (2010). J. Synchrotron Rad. 17, 132–137. Web of Science CrossRef CAS IUCr Journals Google Scholar
Eijsbouts, S., Battiston, A. A. & Vanleerdam, G. (2008). Catal. Today, 130, 361–373. CrossRef CAS Google Scholar
Figgis, B. N. (1966). Introduction to Ligand Fields. New York: Interscience Publishers. Google Scholar
Glatzel, P. & Bergmann, U. (2005). Coord. Chem. Rev. 249, 65–95. Web of Science CrossRef CAS Google Scholar
Glatzel, P., Sikora, M., Smolentsev, G. & Fernández-García, M. (2009). Catal. Today, 145, 294–299. Web of Science CrossRef CAS Google Scholar
Griffith, J. S. (1964). The Theory of Transition-Metal Ions. Cambridge University Press. Google Scholar
Grunwaldt, J. D., Caravati, M., Hannemann, S. & Baiker, A. (2004). Phys. Chem. Chem. Phys. 6, 3037–3047. Web of Science CrossRef CAS Google Scholar
Hill, J. P., Kao, C. C., Caliebe, W., Matsubara, M., Kotani, A., Peng, J. L. & Greene, R. L. (1998). Phys. Rev. Lett. 80, 4967–4970. CrossRef CAS Google Scholar
Keely, W. M., Jerus, P., Dienes, E. K. & Hausberger, A. L. (1984). Catal. Rev. 26, 485–502. CrossRef CAS Google Scholar
Kibsgaard, J., Tuxen, A., Knudsen, K. G., Brorson, M., Topsøe, H., Laegsgaard, E., Lauritsen, J. V. & Besenbacher, F. J. J. (2010). J. Catal. 272, 195–203. CrossRef CAS Google Scholar
Kotani, A. & Shin, S. (2001). Rev. Mod. Phys. 73, 203–246. Web of Science CrossRef CAS Google Scholar
Ma, X. L. & Schobert, H. H. (2000). J. Mol. Catal. A Chem. 160, 409–427. CrossRef CAS Google Scholar
Nicosia, D. & Prins, R. (2005). J. Catal. 231, 259–268. Web of Science CrossRef CAS Google Scholar
Orita, H., Uchida, K. & Itoh, N. (2004). Appl. Catal. Gen. 258, 115–120. CrossRef CAS Google Scholar
Paul, J. F., Cristol, S. & Payen, E. (2008). Catal. Today, 130, 139–148. CrossRef CAS Google Scholar
Prins, R., De Beer, V. H. J. & Somorjai, G. A. (1989). Catal. Rev. 31, 1–41. CrossRef CAS Google Scholar
Rabe, S., Nachtegaal, M., Ulrich, T. & Vogel, F. (2010). Angew. Chem. Int. Ed. 49, 6434–6437. CrossRef CAS Google Scholar
Ratnasamy, P. & Sivasanker, S. (1980). Catal. Rev. 22, 401–429. CrossRef CAS Google Scholar
Rochet, A., Baubet, B., Moizan, V., Pichon, C. & Briois, V. (2016). C. R. Chim. 19, 1337–1351. CrossRef CAS Google Scholar
Rubensson, J. E. (2000). J. Electron Spectrosc. Relat. Phenom. 110–111, 135–151. CrossRef CAS Google Scholar
Sakurai, J. J. (1967). Advanced Quantum Mechanics. Reading: Addison-Wesley. Google Scholar
Samarai, M. al, Delgado-Jaime, M. U., Ishii, H., Hiraoka, N., Tsuei, K. D., Rueff, J. P., Lassale-Kaiser, B., Weckhuysen, B. M., de Groot, F. M. F. (2016). J. Phys. Chem. C, 120, 24063–24069. Google Scholar
Šarić, M., Rossmeisl, J. & Moses, P. G. (2017). Phys. Chem. Chem. Phys. 19, 2017–2024. PubMed Google Scholar
Šarić, M., Rossmeisl, J. & Moses, P. G. (2018). J. Catal. 358, 131–140. Google Scholar
Scheffer, B., De Jonge, J. C., Arnoldy, P. & Moulijn, J. A. (1984). Bull. Soc. Chim. Belg. 93, 751–762. CrossRef CAS Google Scholar
Sokaras, D., Weng, T. C., Nordlund, D., Alonso-Mori, R., Velikov, P., Wenger, D., Garachtchenko, A., George, M., Borzenets, V., Johnson, B., Rabedeau, T. & Bergmann, U. (2013). Rev. Sci. Instrum. 84, 053102. Web of Science CrossRef PubMed Google Scholar
Startsev, A. N. (1995). Catal. Rev. 37, 353–423. CrossRef CAS Google Scholar
Szlachetko, J., Sá, J., Safonova, O. V., Smolentsev, G., Szlachetko, M., van Bokhoven, J. A. & Nachtegaal, M. (2013). J. Electron Spectrosc. Relat. Phenom. 188, 161–165. Web of Science CrossRef CAS Google Scholar
Topsøe, H. & Clausen, B. S. (1986). Appl. Catal. 25, 273–293. Google Scholar
Topsøe, H. & Clausen, B. S. (1984). Catal. Rev. 26, 395–420. Google Scholar
Topsøe, H., Clausen, B. S., Topsøe, N. Y. & Pedersen, E. (1986). Ind. Eng. Chem. Fund. 25, 25–36. Google Scholar
This is an open-access article distributed under the terms of the Creative Commons Attribution (CC-BY) Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original authors and source are cited.