Critical Step in the HCl Oxidation Reaction over Single-Crystalline CeO 2−x (111): Peroxo-Induced Site Change of Strongly Adsorbed Surface Chlorine

Graphical Abstract Abstract The catalytic oxidation of HCl by molecular oxygen (Deacon process) over ceria allows the recovery of molecular chlorine from omnipresent HCl waste produced in various industrial processes. Previous density functional theory (DFT) model-calculations 1 proposed, that the most critical reaction step in this process is the displacement of tightly bound chlorine at a vacant oxygen position on the CeO 2 (111) surface (Cl vac ) towards a less strongly bound cerium on-top (Cl top ) position. This step is highly endothermic by more than 2 eV. On the basis of a dedicated model study, namely the re-oxidation of a chlorinated single crystalline Cl vac - CeO 2−x (111)-( √ 3 × √ 3)R30° surface structure, we provide unique spectroscopic data (high resolution core level spectroscopy (HRCLS) and X-ray adsorption near edge structure (XANES)) for this oxygen-induced de-chlorination process. Combined with theoretical evidence from DFT calculations, the Cl vac → Cl top displacement reaction is predicted to be induced by a surface-adsorbed peroxo species (O 22  ), making the displacement step concerted and exothermic by 0.6 eV with an activation barrier of only 1.04 eV. The peroxo species is shown to be important for the re-oxidation of Cl vac -CeO 2−x (111) and is considered essential for understanding the function of ceria in oxidation catalysis.


Introduction
−4 A sustainable way to cope with this waste problem is to recover molecular chlorine from HCl by the heterogeneously catalyzed oxidation of HCl with molecular oxygen (Deacon process): 4 HCl + O2 → 2 Cl2 + 2 H2O rH = 114 kJ/mol Although this process is known for more than 150 years, 5,6 only in the late 1990s a stable and active catalyst was discovered by Sumitomo Chemical based on RuO2 covered on rutile TiO2. 7ome 15 years later, the HCl oxidation reaction over CeO2-based catalysts 1,8,9 has been shown to be a promising and economically viable alternative to commercial RuO2-based materials.
Based on experiments on ceria powder-catalysts, a reaction pathway for the HCl oxidation over the single crystalline CeO2(111) surface was proposed, 1,9 and the energies of the reaction intermediates were predicted using density functional theory (DFT) calculations (cf. Figure 1).The catalytically active phase was identified with the chlorinated CeO2(111)-(2  2) surface, where chlorine resides in an oxygen vacancy position and one surface lattice O site is occupied by hydrogen forming a hydroxyl group (A).The catalytic cycle 1 starts with the dissociative adsorption of HCl (acid-base reaction), where chlorine resides in a Ce on-top position (Cltop) and the H atom is transferred to the OH surface group (B).The Cltop species displaces the formed H2O molecule and fills the nascent surface oxygen vacancy (VO,S), thus becoming a second Clvac, while water desorbs at a typical reaction temperature of 700 K (C).A second HCl molecule dissociates on the surface and forms a Cltop species (D).The most critical but also innovative step in the proposed reaction pathway is the displacement of Clvac to an on-top position Cltop, a reaction step that is endothermic by 2.15 eV (E).The activation of Clvac leaves an oxygen vacancy behind that is filled by dissociative adsorption of 1/2 O2 from the gas phase, a process that is exothermic by 3.4 eV (F).Last, the two Cltop species recombine, leading to the desired Cl2 product and closing the catalytic cycle (G).The latter steps of the reaction path in Figure 1 can be considered an oxidative de-chlorination process (light orange background), while the first steps can be summarized as chlorination process of the surface (light blue background).
https://doi.org/10.26434/chemrxiv-2023-rjcz6ORCID: https://orcid.org/0000-0001-7689-7385Content not peer-reviewed by ChemRxiv.License: CC BY-NC-ND 4.0 The proposed reaction pathway in Figure 1 remains elusive in two critical points: First, the activation step Clvac → Cltop is highly endothermic and therefore would only be realistic if it can be coupled with the exothermic adsorption process of 1/2 O2 gas.The second point concerns the replenishment of a surface oxygen vacancy by 1/2 O2 gas, which is not an elementary step.Instead, a full O2 gas molecule has to be adsorbed, but where does then the other half of the O2 molecule remain on the surface?
With these two questions in mind, we designed a dedicated surface science experiment, employing the recently developed Clvac-CeO2−x(111)-(√3 × √3)R30° model system. 10The reoxidation of Clvac-CeO2−x(111)-(√3 × √3)R30° is studied in situ by high-resolution synchrotronbased spectroscopy techniques, including high resolution core level shift spectroscopy (HRCLS) and X-ray absorption near edge spectroscopy (XANES), while the underlying reoxidation steps are studied by density functional theory (DFT) calculations.Specifically, we elucidate the activation of displacing Clvac towards Cltop.This process is shown to be initiated by O2 gas adsorption into an oxygen vacancy via the formation of a surface peroxide species (peroxo: O2 2 ), while its dissociation is coupled with the Clvac → Cltop displacement.This renders the concerted Clvac activation step slightly exothermic with an activation energy of only 1.04 eV.

Experimental and theoretical details
The lab-based experiments are performed in an ultrahigh vacuum (UHV) system that is equipped with Low-Energy Electron Diffraction (LEED) using a video-LEED (Specs ErLEED 1000-A) system.Additionally, Thermal Desorption Spectroscopy (TDS)/Temperature Programmed Desorption (TPD) is performed with a high sensitivity (ion counting) mass spectrometer (Hiden Analytical HAL 301/3F). 11,12A second UHV system 13 is dedicated to labbased XPS (Omnivac Al/Mg dual anode x-ray source and a Leybold EA 200 analyzer) that runs at a photon energy of 1253.6 eV (Mg).For the dosing of atomic oxygen O, the system is equipped with a thermal cracker (Oxford Applied Research: TC50 -Universal Thermal Cracker).Basis of all experiments are single crystalline CeO2(111) films on Ru(0001) (MaTeck GmbH, disk, ⌀ = 8 mm) prepared by physical vapor deposition (PVD) of Ce foil (HMW Hauner GmbH, 99.9%) from a well-outgassed electron beam evaporator (Focus GmbH, EFM 3, U = 800 V, P = 40−50 W, Flux = 2.0 µA, deposition time tdep = 70−120 min, Mo-crucible capacity = 250 mm³, contained Ce ≈ 25 mg).During evaporation, the sample is kept in a background O2 atmosphere of p(O2) = 5•10 8 mbar at a temperature of T = 700 K. Postoxidation is achieved by annealing the sample to T = 1000 K in p(O2) = 10 6 mbar for 15 min followed by flashing to T = 1200 K and subsequent cooling to room temperature in O2.−20 The film thickness of stoichiometric CeO2(111) is determined by X-ray reflectometry (XRR, PANalytical X'Pert PRO MRD) to be 5.6 nm, corresponding to 18 OCeO trilayers. 10e stoichiometric CeO2(111)-Ru(0001) samples (prepared in the homelab at the JLU) are subsequently transferred to the preparation chamber of the permanent endstation EA01 (Surface and Material Science branch) at the FlexPES beamline 21 at MAX IV in Lund, Sweden.On-site the CeO2(111)-Ru(0001) sample is annealed for t = 15 min in p(O2) = 10 6 mbar followed by a quick annealing step to T = 1200 K in p(O2) = 10 6 mbar to remove potential carbon contamination during the sample transfer.The reduced CeO2−x(111) films are prepared by deposition of Ce metal under UHV conditions via PVD at room temperature on the preformed fully oxidized CeO2(111) layer that is followed by an annealing step of the sample to T = 900 K for 15 min and a final annealing step T = 1100 K under UHV conditions. 19,22,23We employed https://doi.org/10.26434/chemrxiv-2023-rjcz6ORCID: https://orcid.org/0000-0001-7689-7385Content not peer-reviewed by ChemRxiv.License: CC BY-NC-ND 4.0 the same electron beam evaporator as used at the JLU for depositing metallic Ce under UHV conditions for a deposition time of treox ≈ 0.12 tdep.After the beamtime XRR was employed to determine the thickness of the prepared CeO2−x(111) film on Ru(0001) to be 7.5 nm that corresponds to 24 O−Ce−O trilayers (cf. Figure S1).The fully chlorinated CeO2−x(111) film is prepared by exposing the CeO2−x(111) film to 15 L HCl (ultrapure HCl, Air Liquide, 2.8) at room temperature and subsequent flashing to T = 1000 K, revealing a (√3 × √3)R30° LEED pattern (cf. Figure S2).The annealing step leads to the removal of hydrogen from the surface via water or H2 formation.Recently, it was demonstrated 10 that adsorbed chorine on CeO2−x(111) only resides exclusively in surface oxygen vacancies (Clvac): Clvac-CeO2−x(111)-(√3 × √3)R30°.
The measurement protocol in the re-oxidation experiments of fully chlorinated Clvac-CeO2−x(111) is adjusted to the requirements of the FlexPES beamline at MAX IV in Lund, Sweden in that the O2 pressure in the analysis chamber must not raise above 10 8 mbar.This excludes an operando experiment so that we exposed O2 to the Clvac-CeO2−x(111) at a sample temperature of 700 K in the preparation chamber first and then transferred the sample back to the analysis chamber without breaking the UHV conditions, where high resolution core level shift spectroscopy (HRCLS) and X-ray absorption (XANES) data are acquired.HRCLS of Ce 4d, Cl 2p, and O 1s at two photon energies (Ce 4d: hν = 250 eV/850 eV; Cl 2p: hν = 250 eV/850 eV; O 1s: hν = 580 eV/1180 eV) allows us to probe both near-surface and bulk-like properties.Bulk-like properties are probed by photoelectrons with kinetic energy of Ekin = 650 eV, while near-surface properties are probed by photoelectron of Ekin = 50130 eV kinetic energy, close to the minimum of the universal curve of inelastic mean free path of electrons. 24The energy of the Ce 4d was chosen to be hν = 250 eV as a compromise between achieving high surface sensitivity and avoiding the cooper minimum 25 at hν = 175 eV 26 with its low cross section.HRCLS of Ce 4d is used instead of Ce 3d to have sufficient flux at higher kinetic energies of Ekin = 650 eV.The slit-width of the beam is set to 50 µm and the pass energy Epass of the detector is set to Epass = 50 eV, except for Cl 2p (hν = 250 eV, Epass = 10 eV) and O 1s (hν = 580 eV, Epass = 20 eV).The spectra are calibrated via the Fermi level EF of gold foil mounted next to the sample.From the Ce 4d spectra at hν = 250 eV the near-surface fraction of Ce 3+ /Ce 4+ and therefore the near-surface reduction degree x (Ce 3+ fraction) can be determined.XANES data of Ce M4,5 edge (Ce 3d) in total electron yield (TEY) mode provides true bulk information of the Ce 3+ /Ce 4+ fraction and therefore on the bulk reduction degree x (Ce 3+ fraction).Beam-induced damages of the studied layers are not encountered, as routinely checked during the synchrotron beamtime.Additionally, the defocused spot is moved between each measuring point over the homogeneous sample.Altogether, this powerful combination of synchrotron-based techniques allows us disentangling surface from bulk properties and is applied here to study the reoxidation and de-chlorination of Clvac-CeO2−x(111).
We perform spin-polarized DFT calculations utilizing the slab−supercell technique. 27The Vienna Ab-initio Simulation Program (VASP, version 5.3.5 and 5.4.4) 28−32 is employed for this purpose.In our calculations, the Ce (4f, 5s, 5p, 5d, 6s), O (2s, 2p), Cl (2s, 2p), and H (1s) electrons are explicitly treated as valence states using the projector augmented wave (PAW) method. 33A plane-wave cutoff energy of 400 eV is applied, while the remaining electrons are considered part of the atomic cores.For the determination of energies and forces, we adopt the DFT+U approach proposed by Dudarev et al. 34 with a Ueff value of U − J = 4.5 eV for the Ce 4f electrons.Additionally, we employ the generalized gradient approximation (GGA) as suggested by Perdew, Burke, and Ernzerhof (PBE). 35e CeO2−x(111) surfaces are modeled with an optimized lattice constant of 5.485 Å for bulk CeO2.The Cl surface phase is modeled based on (3  3) unit cell reported by Olbrich et al. 36 https://doi.org/10.26434/chemrxiv-2023-rjcz6ORCID: https://orcid.org/0000-0001-7689-7385Content not peer-reviewed by ChemRxiv.License: CC BY-NC-ND 4.0 for describing the formation of reduced CeO2−x(111) surface reconstructions with one surface (VO,s) and two subsurface (VO,ss) vacancies in the outermost O−Ce−O trilayer.All surface models used in this work have three O−Ce−O trilayers and ~21 Å of vacuum separation between consecutive slabs.A (2  2  1) k-point mesh according to the Monkhorst−Pack method is used to sample the Brillouin zone. 37For the gas-phase calculations of the HCl, O2 and Cl2 molecules, a cell of (15 × 15 × 15) Å 3 is employed, using only the Γ-point.
The average adsorption energy per Cl atom on the CeO2−x(111) surface is calculated according to the following equation: where n = 1,2,3 is the number of adsorbed Cl atoms, E[nCl/CeO2−x(111)] is the total energy of the Cl atoms adsorbed on the surface, E[CeO2−x(111)] is the total energy of the surface without the adsorbate, and E[Cl2gas] is the energy of the chlorine molecule in gas phase.
To locate transition state (TS) structures, we employ the climbing image nudged elastic band method (CI-NEB) 38 with seven images for each reaction pathway.For all the TS reported in this work, we have found only one imaginary frequency, and the full geometry optimizations starting from its back and forward nearest configurations (along the reaction path) end in a nondissociated and dissociated state, respectively.In the calculated potential energy profiles, the energy barrier, EBarrier = ETS − EIS, equals the difference between the energy of the transition state, ETS, and the initial state, EIS, whereas the effective or apparent energy barrier is given by the energy of the transition state, ETS, referenced to n/2 Cl2 gas and the clean surface.

Synchrotron-based characterization of Clvac-CeO2−x(111)-(√3 × √3)R30°
The on-site prepared Clvac-CeO2−x(111)-(√3 × √3)R30° is thoroughly characterized by HRCLS and XANES measurements.The (bulk) reduction degree x and corresponding Ce 3+ fraction (2x•100%) of CeO2−x(111) is derived from the Ce M4,5 edge XANES spectrum (TEY yield) by fitting the XANES spectrum of CeO2−x(111) as a linear combination of the reference spectra of the fully oxidized CeO2(111) film (0% Ce 3+ ) and a fully reduced Ce2O3 film (100% Ce 3+ ): (1−2x)•XANES(CeO2) + 2x•XANES(Ce2O3).The fully reduced Ce2O3 films is prepared in-situ by depositing (U = 800 V, P = 40-50 W, Flux = 2.0 µA, deposition time tdep = 30 min) metallic Ce on Ir(111) under UHV conditions and annealing the film to 1250 K.The residual water pressure in the UHV chamber is sufficient to oxidize the Ce film to Ce2O3.−43 All XANES spectra including the reference spectra are normalized by the energy dependent flux and the integral intensity in the range from 865.5 eV to 918.5 eV (Figure S3).The procedure of linear combination is summarized in Figure S3b, resulting in a bulk Ce 3+ fraction of 37% that is in between those of (3 × 3) and (√7 × √7)R19.1°.Accordingly, in LEED a faint (3 × 3) is visible (cf. Figure S2a).Upon 15 L HCl exposure and subsequent stepwise annealing a (√3 × √3)R30° overlayer of chlorine is formed (cf. Figure S2b).Recall that the very same (√3 × √3)R30° chlorine overlayer structure is observed for a variety of reduction degrees CeO2−x(111) ranging from a Ce 3+ fraction of ~40% to ~80%. 10 Ce 4d HRCL spectra of CeO2−x(111) and Clvac-CeO2−x(111) including the reference spectra of Ce 4+ and Ce 3+ are summarized in Figure S4.Again, by employing the linear combination of CeO2(111) and Ce2O3 Ce 4d reference (HRCL) spectra, the Ce 3+ fraction in the near-surface region can be determined (cf. Figure S4b).The near surface Ce 3+ fraction of 71% is significantly higher than in the bulk, fully consistent with recent studies from the Matolin´s group. 22he Cl 2p HRCL spectra after 15 L HCl exposure to CeO2−x(111) and subsequent annealing to 1000 K are presented in Figure S5.The shoulders at lower binding energy of the Cl 2p doublet at 198.6 eV/200.2eV are assigned to chlorine still forming a hydrogen bond. 44These shoulders disappear upon annealing to 1000 K, leaving only a single Cl component on the surface.Concomitant with the decline of the shoulder in Cl 2p, also the OH species 45 in the O 1s spectrum (cf. Figure S6) disappear.The O 1s spectrum of fully chlorinated Clvac-CeO2−x(111) is shifted by 0.39 eV when compared to that of the stoichiometric CeO2(111) (cf. Figure S7).

Re-oxidation and De-chlorination experiments of Clvac-CeO2−x(111)
The re-oxidation experiments of the chlorinated CeO2−x(111) is carried out stepwise by a specific exposure of O2 at 700 K followed by a subsequently characterization of the sample with XANES and HRCLS at room temperature.The temperature of 700 K is chosen to be identical with the typical reaction temperature of the ceria-based Deacon process.We start the discussion with XANES Ce M4,5 edge (Ce 3d) data in the total electron yield (TEY) mode (cf. Figure 2).This detection mode is truly bulk sensitive to Ce, so that the Ce 3+ fraction of the entire Clvac-CeO2−x(111) film can be quantified by a linear combination of the reference spectra, as discussed in section 3.1.The high quality of these fits can be judged from inspection of Figure 2, while the derived bulk Ce 3+ fraction as a function of O2 exposure is compiled in Figure 3 (circles).Clearly, the bulk Ce 3+ fraction decreases linearly with O2 exposure: first steeply from 38% to 11% upon 300 L of O2 and then slows down from 11% to 7% when increasing the O2 exposure from 300 L to 700 L (and 5% for 2700 L O2).While XANES in the TEY mode is bulk-sensitive, photoemission data of Ce 4d employing a photon energy of hν = 250 eV photoelectrons with a kinetic energy of Ekin ≈ 120 eV are highly surface sensitive (the escape depth of photoelectrons is about 1nm). 24Also the Ce 4d spectra in Figure 4 can be fitted by a linear combination of references Ce 4d spectra of Ce2O3 and CeO2.Even without fitting it is obvious (consider the two peaks W''' and X''', assigned to a Ce 4d 9 O 2p 6 Ce 4f 0 final state of Ce 4+ at EBE ≈ 122126 eV) 41,46 that the Ce 4d spectra decompose in two sets.At low O2 exposures up to 200 L, the Ce 4d spectra vary slightly with exposure from 72% to 61% Ce 3+ .Between 200 L and 300 L of O2 an abrupt change in the Ce 4d spectra occurs reducing the Ce 3+ fraction from 61% to 43%.For higher O2 exposures above 300 L the variation in Ce 4d spectra is small again (43% to 35% Ce 3+ ).In fact, this observation is fully reconciled and quantified with the fitting procedure (Figure 4) whose derived nearsurface Ce 3+ fractions are overlaid in Figure 3 (triangles).For O2 exposure below 200 L (Figure 3, blue background), the fraction of near surface Ce 3+ of CeO2−x(111) decreases more slowly than that of bulk Ce 3+ as derived from the XANES data.
Keeping in mind that XANES in TEY mode is bulk-sensitive, while Ce 4d spectra is essentially surface sensitive, the difference in Ce 3+ fraction (bulk versus surface) as a function of O2 exposure is explained by a preferential oxidation of bulk CeO2−x(111).In the O2 exposure ranging from 200 L to 300 L, the near surface Ce 3+ fraction decreases quickly, while the bulk Ce 3+ fraction (XANES) of the entire film still decreases with the same rate as for lower exposures (Figure 3, transition blue/orange).This behavior points to a preferential oxidation of the surface region for O2 exposures 200 L300 L. Above 300 L both bulk and surface region further oxidize slowly with the Ce 3+ fraction declining faster in the surface than in the bulk region (Figure 3, blue background).Even after exposure of 2700 L of O2 the Ce 3+ fraction in the near surface region is still 30%, while that of the bulk Ce 3+ is 5%.The mechanisms for the initial bulk oxidation followed by the surface oxidation will be further discussed in section 3.3.
It is known that the O 1s binding energy depends sensitively on the reduction degree x of CeO2−x. 19,47In Figure S7a we show surface-sensitive O 1s spectra (hν = 580 eV) depending on the applied O2 exposure during re-oxidation.Up to 200 L O2, a shift by 0.1 eV is observed in the O 1s binding energy, while from 200 L to 300 L, it additionally shifts by 0.1 eV, remaining then nearly constant for even higher O2 exposures.The final O 1s is shifted to lower binding energy when compared with the stoichiometric CeO2(111) surface (cf. Figure S7a).For resolving this small energy shift, high resolution at FLEXPES is mandatory.More bulk-  In Figure 5 we summarize the surface sensitive Cl 2p photoemission data (hν = 250 eV) of Cl-CeO2−x(111) exposed to various doses of O2 at 700 K. Up to 200 L of O2 no significant changes are discernible in the Cl 2p spectra (Cl 2p1/2 and Cl 2p3/2 at 200.9 eV and 199.3 eV).That is, the changes are minor as long as the Ce 3+ fraction in the surface near region does not change significantly (cf. Figure 3).For higher O2 exposures the Cl 2p duplet shifts then to lower binding energies by 0.3 eV (majority species) and by 1.0 eV (minority species: shoulder at 198.3 eV, cf. Figure S9).This abrupt change in the Cl 2p spectra at 300 L coincides with the sudden decrease in the surface Ce 3+ fraction as indicated in Figures 3.For a site change the shift in Cl 2p is likely to be too small, when compared to chlorine adsorption on RuO2(110), where a site change of Cl from on-top to bridge position leads to a Cl 2p binding energy shift of 1.5 eV. 48,49A change in the oxidation state of Clvac from 1 to 0 can clearly be excluded since Cl  is substantially stronger bound (DFT: 2.65 eV) than Cl 0 (DFT: 0.46 eV) in a vacancy position (cf.Figure S8).
Therefore, the small energy shift of 0.3 eV (main peak) and 1.0 eV for the shoulder in Cl 2p with respect to the 0L spectrum may be traced to the existence of two different chlorine Clvac species as a consequence of the re-oxidation of the surface.In Figure S9 we exemplify the deconvolution of the surface sensitive Cl 2p spectra of Clvac-CeO2−x(111) in comparison with that after exposing 2700 L of O2 at 700 K. Similar Cl 2p spectra are recorded with hν = 850 eV (Figure S10; more bulk-sensitive), evidencing that both chlorine species are located solely at the surface and the quantification of chlorine species is not affected by diffraction effects.When https://doi.org/10.26434/chemrxiv-2023-rjcz6ORCID: https://orcid.org/0000-0001-7689-7385Content not peer-reviewed by ChemRxiv.License: CC BY-NC-ND 4.0 increasing the surface temperature from 700 K to 800 K, 900 K, 1050 K, and finally to 1200 K, the Cl 2p spectra in Figure S11 indicate that both Cl species are stable up to 900 K and the minority Cl component desorbs at slightly lower temperatures than the chlorine in the main peak.This indicates a lower thermal stability of the minority species than that of the majority chlorine species.The evaluation of the total amount of Cl (cf. Figure 5b) as a function of O2 exposure at 700 K indicates a ~25% loss of surface chlorine above an exposure of 300 L O2. Oxygen induced removal of chlorine is only observed for the majority Cl 2p component, while the minority Cl species is not affected by O2 exposure at 700 K and is therefore not relevant for the present dechlorination step.This indicates a higher chemical stability of the minority species with respect to O2 exposure at 700K.
The loss of 25% Cl may be traced to a destabilization of the chlorine species due to re-oxidation.In a recent study, 10 the Cl-overlayer on CeO2−x(111) was shown to destabilize with decreasing Ce 3+ fraction of the underlying preformed CeO2−x(111) thin film of ~40% to ~80% (0.2 < x < 0.4).Therefore, we compare in Figure 6 lab-based Cl2 thermal desorption spectra of Clvac-CeO2−x(111) with ~37% Ce 3+ fraction (identically prepared as at the beamtime) before and after re-oxidation, clearly indicating that re-oxidation of the chlorinated CeO2−x(111) substantially shifts the desorption temperature of chlorine to lower temperatures, i.e., the adsorbed chlorine is significantly destabilized by the re-oxidation process.However, these https://doi.org/10.26434/chemrxiv-2023-rjcz6ORCID: https://orcid.org/0000-0001-7689-7385Content not peer-reviewed by ChemRxiv.License: CC BY-NC-ND 4.0 experiments also show that the loss of 25% of Cl cannot be explained by this destabilization effect, since the desorption onset occurs at a temperature of 800 K (cf. Figure 6) which is significantly higher than 700 K. Therefore, the O2 induced removal of Clvac needs to be a surface reaction rather than a simple desorption process.
The broad TD feature of Cl2 in TDS (Figure 6) of re-oxidized Clvac-CeO2−x(111) from 800 K to 1200 K may be traced to a varying Ce 3+ fraction near the surface.During Cl2 desorption two oxygen vacancies per Cl2 are left on the surface so that the near-surface Ce 3+ fraction increases, which leads to stronger Cl bonding leading to the broad feature in Cl2-TDS (cf. Figure 6) consistent with previous DFT calculations. 10Let us start with the O2 activation.For O2 adsorption to occur we need the presence of a surface oxygen vacancy (VO,S).O2 species adsorbing on surface Ce sites are weakly bound by less than 0.4 eV, 51,52 thus being irrelevant for the re-oxidation process studied at 700 K.The formation of a surface O-vacancy at the 3Clvac-Ce3O5(111)-(3  3) surface by desorption of molecular oxygen (1/2 O2) requires 2.18 eV (Figure S12a-b, no VO,SSS in the slab) and can therefore be safely ruled out at 700 K. Instead, the diffusion of an O vacancy from a deeper oxygen layer (VO,SSS) of Ce3O5(111) towards the surface is preferred as summarized in the energy diagram, presented in Figures 7a and S12d (one VO,SSS in the slab).The O vacancy can easily diffuse towards the surface with an activation energy of 0.80 eV.The subsurface oxygen vacancy (VO,SS) is the most preferred site.−55 The probability for an oxygen vacancy to be encountered at the surface is given by Boltzmann statistics.For instance, at 700 K, the probability of a surface oxygen vacancy (VO,S) is about 7%.Whenever an O2 molecule from https://doi.org/10.26434/chemrxiv-2023-rjcz6ORCID: https://orcid.org/0000-0001-7689-7385Content not peer-reviewed by ChemRxiv.License: CC BY-NC-ND 4.0 the gas phase strikes a surface oxygen vacancy (VO,S), O2 is strongly adsorbed by about 2 eV in the form of a peroxo-species (O2 2 ) and pins the vacancy to the surface.The O-O bond length is 144 pm that agrees well with a typical bond length of peroxide groups. 56The oxidation state of O2 is determined by counting the number of Ce 3+ sites left in the unit cell.Upon O2 adsorption this number is reduced by two per (3  3) cell, clearly evidencing 57 that the adsorbed O2 is in the oxidation state 2, i.e., forming a peroxo species O2 2 .The peroxo species cannot further dissociate to Otop and Ovac species, since this process is endothermic by about 0.48 eV (Figure 7b).The energy profile for the re-oxidation of Clvac-CeO2−x(111) (cf. Figure 8) starts from the adsorbed surface peroxo species and assumes the existence of oxygen vacancies in deeper ceria layers (VO,SSS in the slab).If one of these vacancies diffuses towards the surface and the subsurface vacancy (VO,SS) is near the peroxo species, dissociation of the peroxo species occurs.This dissociation step is exothermic by 2.412.58eV and is only slightly activated by 0.310.68eV depending on the local configuration.Two lattice oxygen atoms are formed, thereby annihilating two oxygen vacancies, one on the surface (VO,S) and the other in the https://doi.org/10.26434/chemrxiv-2023-rjcz6ORCID: https://orcid.org/0000-0001-7689-7385Content not peer-reviewed by ChemRxiv.License: CC BY-NC-ND 4.0 subsurface (VO,SS) underneath.This process reduces the concentration of oxygen vacancies in the near-surface region so that the chemical potential difference drives the diffusion from bulk vacancies to the surface (activation barrier: 0.80 eV), thereby re-oxidizing the bulk of CeO2−x(111).We consider now the displacement step of Clvac to Cltop, a process that is shown to be induced by the peroxo species at the surface and whose energy profile is depicted in Figure 9, calculated using a structure with two O2 2 species (no VO,SSS in the slab).As soon as the peroxo species is in the proximity of a Clvac species, the dissociation of the O2 2 species leads to the displacement of Clvac to Cltop which is exothermic by 0.60 eV with an activation barrier of 1.04 eV.This concerted process is shown as movie frames in the supporting information (Figure S13).If the peroxo species is not close to Clvac, the peroxo species needs to diffuse along the surface to approach Clvac.In this diffusion process 51 the top O atom of the peroxo species hops to a neighboring lattice O site forming a new peroxo species with a barrier of 1.23 eV (Figure S14, no VO,SSS in the slab).Both barriers can readily be overcome at 700 K.This structure diffusion process is reminiscent of the diffusion of protons in water (Grotthuss mechanism).
If more than two Cltop species form at the surface, they can recombine and the Cl2 is released directly into the gas phase.According to DFT calculations this association step of two Cltop is activated by 0.87 eV and is endothermic by 0.79 eV (cf. Figure 9).These DFT calculations can readily explain the experiments presented in section 3.2.The activation energy for the bulk re-oxidation is governed by the diffusion barrier of O vacancies from the bulk towards the surface (0.80 eV), while the activation barrier for the displacement step Clvac → Cltop is 1.04 eV.From this difference in activation energies, the branching ratio at 700 K between these competing reactions is inferred to be about 1:50 in favor of bulk reoxidation.This is fully consistent with the experimental observation that the Cl 2p HRCL spectra (cf. Figure 5) are practically unaltered as long as the bulk CeO2−x(111) is not nearly fully re-oxidized (cf.bulk-sensitive XANES Figure 3).
When the bulk and the near surface region of CeO2−x(111) are essentially re-oxidized after exposing 300 L of O2 at 700 K, the lifetime of the adsorbed peroxo species is long enough to induce the competing displacement reaction of Clvac → Cltop.Since the surface concentration of O vacancies (VO,S) is expected to be small after bulk re-oxidation, only very few peroxo species are formed and as a consequence the displacement step Clvac → Cltop is quite slow.The Cltop species is not long-lived on the surface since the recombination of two neighboring Cltop species to form Cl2 gas is activated by only 0.87 eV and proceeds therefore rapidly at 700 K; recall the great entropy gain during desorption process.This explains also, why Cltop cannot be detected in the Cl 2p spectrum (cf. Figure 5a), thus supporting the interpretation of the observed shift in Cl 2p spectra being due to re-oxidation of the surface near region.After exposure of 2700 L O2 the reduction in Clvac coverage saturates and amounts to 25% (cf. Figure 5b).A possible explanation is that there are not enough oxygen vacancies available at the surface into which O2 can adsorb as peroxo.This terminates the displacement reaction Clvac → Cltop and the subsequent recombination step of neighboring Cltop species.

Removal of Clvac by exposure to atomic O at 300 K and O2 at 500 K
The DFT calculations in section 3.3 motivate further experiments.The low activation energies found by DFT calculations suggest that partial removal of Clvac should even occur at 500 K.This set of experiments is presented in Figure 10a.Even at 500 K the Cl 2p signal decreases with O2 exposure, although we need approximately 10 times higher O2 doses in comparison with re-oxidation at 700 K.This is explainable since both the formation of surface vacancies and the displacement reaction are activated by 0.80 eV and 1.04 eV, respectively (cf.Figures 8  and 9).In order to stabilize the peroxo species upon O2 exposure, we need the presence of oxygen vacancies at the surface (VO,S).Since the concentration of surface oxygen vacancies after bulk re-oxidation of Clvac-CeO2−x(111) is very low, the peroxo species remains a kind of "ghost species" that is present only in spurious amounts, and it is easily consumed by re-oxidation and the displacement reaction.However, there is an alternative and efficient way to produce the peroxo species, namely by exposure to atomic oxygen O. 58 At room temperature, surface lattice O 2 sites can be transformed to O2 2 by atomic O exposure without changing the oxidation state of Ce.Of course, atomic oxygen can also adsorb on surface Ce sites, but this adsorption process is by 0.48 eV less favorable than the adsorption on surface lattice O followed by the formation of O2 2 (cf. Figure 7b).The peroxo species in turn induces the shift of Clvac species from the vacancy position to the top position Cltop where chlorine can subsequently recombine and desorb from the surface as Cl2.In this way, part of the Clvac at the surface can be removed by exposing the re-oxidized Clvac-CeO2−x(111) surface to atomic oxygen at 300 K while monitoring the Cl 2p signal.From this set of experiments summarized in Figure 10b it is clear that low exposures of ~50 L O are sufficient to remove half of the surface Clvac species, which, in turn, corroborates the preferential formation of peroxo species and its important role in the Clvac → Cltop displacement reaction.The peroxo species on the stoichiometric CeO2(111) can also be verified in the O 1s spectrum (cf. Figure S15).The O 1s component at 531.3 eV can clearly be discriminated from the OH species at 532.2 eV (cf. Figure S6) and according to a recent XPS study the low energy component was assigned to the peroxo species based on DFTderived binding energy shifts. 58

Catalytically active surface of CeO2 under typical Deacon reaction conditions
The Deacon reaction with CeO2 as catalyst typically runs under oxidizing reaction conditions (feed O2 : HCl > 1:4) as the reaction order of O2 is positive. 1−61 This reduction is enabled by HCl adsorbing in an acid-base reaction leading towards OH group formation on surface lattice O and Cl atom adsorption on surface Ce.With high enough temperature chlorine atoms are able to recombine to form Cl2 whereas neighboring OH groups are able to form water and an oxygen vacancy upon recombination.The formed oxygen vacancy can be occupied by chlorine upon dissociative HCl adsorption or by peroxo species.Note that the peroxo species can desorb at 700 K and the further dissociation of peroxo would need another subsurface O vacancy in its proximity.For kinetic reasons, chlorine adsorption is preferred over peroxo adsorption, leading to an accumulation of chlorine in surface oxygen vacancies.Ultimately, this will chlorinate the CeO2−x surface, where chlorine is strongly adsorbed in surface oxygen vacancies as Clvac.
Previously the chlorination degree of CeO2 powder was quantified to be less than 1 ML (ML: monolayer) in an experiment with Prompt Gamma-ray Neutron Activation Analysis (PGAA). 59uch closer to our model system are shape-controlled CeO2 octahedrons exposing (111) facets. 62After HCl oxidation reaction of CeO2 nano-octahedrons with an oxidizing reaction feed O2 : HCl = 2:1 (cf.stoichiometric feed: O2 : HCl = 1:4), a mean Ce 3+ fraction of 29% in the near-surface region and a Cl/Ce ratio of 15% was determined. 60The fresh CeO2 catalyst possesses a Ce 3+ fraction of only 20%.From these experiments, the chlorine coverage can be estimated to be about 0.4 ML that would account for 9% of the Ce 3+ ; recall that each Cl  needs one Ce 3+ for charge compensation.Therefore, the residual 20% of Ce 3+ needs to be compensated by 10% oxygen vacancies.From the observation of a (√3 × √3)R30° pattern in LEED, 10 the Cl coverage is either 1/3 or 2/3, in broad agreement with the coverage estimated for CeO2 nano-octahedrons. 60In conclusion, this experimental evidence from powder and nanoparticle CeO2 renders a chlorinated CeO2−x(111) surface with a Clvac coverage of 1/3 and 10% oxygen vacancies a suitable model system for studying the elementary steps in the reoxidation reaction as part of the Deacon process.
That the partially reduced Clvac-CeO2−x(111) surface is the active phase for the Deacon process is also consistent with previous studies showing that neither the stoichiometric 13 nor the deeply reduced CeO2−x(111) surfaces 10 are (very) active in the HCl oxidation reaction.The stoichiometric CeO2(111) surface leads preferentially to re-combinative desorption of HCl instead of Cl2 formation, whereas the strongly reduced CeO2−x(111) surface binds chlorine in vacancy positions too strongly, thus suppressing direct Cl2 evolution at 700 K.But also the water formation step needs a partially reduced CeO2−x(111) surface.On stoichiometric CeO2(111), water formation is not observed, 13 whereas on the reduced CeO2−x(111) surface, water formation is observed at 620 K, 10 if the Ce 3+ fraction (reduction degree x) of CeO2−x(111) is not too high.
With pure HCl exposure a reduced CeO2−x(111) surface may be completely covered by chlorine in vacancies.From an energy point of view a (1  1)-Clvac overlayer would be feasible as shown https://doi.org/10.26434/chemrxiv-2023-rjcz6ORCID: https://orcid.org/0000-0001-7689-7385Content not peer-reviewed by ChemRxiv.License: CC BY-NC-ND 4.0 by the differential heat of adsorption in Figure S16.A complete chlorination of the CeO2−x(111) surface would, however, de-activate the surface since neither O2 nor HCl can further adsorb on this surface termination.Therefore, the oxygen-induced de-chlorination is important to prevent the CeO2-x(111) catalyst from over-chlorination of the CeO2−x(111) surface close to 1 ML and thus from de-activation.

Deacon reaction over Clvac-CeO2−x(111)
The Deacon reaction can be broken down into two separate processes in which the catalyst undergoes a solid state redox cycle. 2,4,63This separation in partial reactions can also be utilized in chemical looping. 64,65First, HCl reduces the oxide catalyst to a (surface) chloride, during which water is formed as couple-product.This chlorination process is exothermic.In the second step the chlorinated surface is re-oxidized by molecular oxygen to recover the active phase of the catalyst, thereby releasing the desired Cl2 and closing the catalytic cycle.This re-oxidation process can be considered as an oxygen-driven de-chlorination step, a process that is endothermic and that is in the focus of the present study starting from Clvac-CeO2−x(111).In Figure 1, all steps in the blue background comprises the chlorination process, while the steps with orange background belong to the de-chlorination process.Amrute et al. 1 have introduced the Clvac to Cltop displacement as critical step in the dechlorination process of the ceria catalyst.They proposed that one of the oxygen atoms in the subsurface layer can diffuse toward the surface, pushing a Clvac atom towards the top site on Ce and leaving one oxygen vacancy in the subsurface region.This step has been considered rate limiting for the Deacon process and is endothermic by 2.15 eV. 1 The following surface reoxidation step has remained, however, largely elusive and was proposed to be carried out through "a complex diffusion-reaction mechanism" 1 that would release nearly 3.4 eV.Altogether, this combined process of oxidative displacement is exothermic by 1.2 eV; a lower energy of 0.53 eV was reported by Wolf et al. 66 consistent with the present DFT study.
With a combination of in-situ synchrotron-based techniques (XANES and HRCLS) the reoxidation is shown to start from the bulk of Clvac-CeO2−x(111) and then propagates towards the surface.From bulk-sensitive XANES experiment (cf. Figure 2) the re-oxidation of bulk CeO2−x(111) is shown to proceed nearly linearly with O2 exposure up to 300 L of O2 and above 300 L the re-oxidation is significantly slowed down.Quite in contrast, the surface sensitive Ce 4d XP spectra (cf. Figure 4) indicate only little change of the Ce 3+ fraction in the surface region up to 200 L of O2, while oxidation in the near surface region is then practically completed for exposures up to 300 L. Above 200L O2 also Cl 2p spectra (cf. Figure 5) alter in that the spectral features shift slightly to lower binding energies and the integral intensity starts to decrease with O2 exposure.We do not directly observe the actual activation step from Clvac to Cltop, but instead monitor with the Cl 2p spectra the oxygen-driven removal of Clvac, i.e. the partial de-chlorination of the Clvac-CeO2−x(111) surface, by the recombination of Cltop species to form Cl2 and its instant release to the gas phase at 700 K.
These experiments are fully explained by the present DFT study.−69 Whenever an additional vacancy appears in the subsurface region directly below the peroxo species, the peroxo species dissociates and replenishes two oxygen vacancies.From the calculated activation energies bulk re-oxidation (0.80 eV) precedes the Clvac → Cltop displacement step (1.04 eV).The recombination of two Cltop species to form Cl2 is activated by 0.87 eV that results in its instant release to the gas phase at 700 K.The Clvac → Cltop displacement step is demonstrated to be directly coupled with the surface re-oxidation step that is induced by a neighboring peroxo species.This makes the concerted displacement step exothermic by 0.6 eV with an activation energy of about 1 eV.https://doi.org/10.26434/chemrxiv-2023-rjcz6ORCID: https://orcid.org/0000-0001-7689-7385Content not peer-reviewed by ChemRxiv.License: CC BY-NC-ND 4.0 A de-chlorination experiment with atomic oxygen O provides indirect experimental evidence for the important role of the peroxo species in the de-chlorination process.Atomic O exposure is able to form peroxo species at the surface without the need of surface O-vacancies. 58Indeed experiments in Figure 10b indicate that room temperature exposure of ~50 L of atomic O is able to remove 40% of Clvac from the surface, a larger portion than that achieved by exposure of 2700 L of O2 at 700 K (25%).

Peroxo species is essential for oxidation catalysis over CeO2
The ability to remove/uptake oxygen from the lattice while maintaining structural integrity endows ceria with unique properties in catalysis science and technology. 42,70The mass-specific amount of oxygen that can be exchanged with the oxide catalyst is called the oxygen storage capacity (OSC). 71,72The O2 incorporation requires two oxygen vacancies and electrons supplied from the oxidation of Ce 3+ . 73Frequently, the OSC serves as a simple descriptor for the activity in oxidation catalysis of reducible oxides. 9,74−80 On the basis of previous spectroscopic experiments on CeO2 powders 81−86 and CeO2-x(111) platelets, 69 superoxo-and peroxo species have been identified and assumed to play an important role in the oxidation chemistry of ceria.In the present study we show that the adsorption energy of peroxo species at O-vacancies of CeO2−x(111) is quite high (~2 eV).Therefore, the peroxo species has a long lifetime on the surface even at high temperatures of 700 K.For the reoxidation of CeO2−x, we do not need to assume two neighboring surface O vacancy sites for direct O2 dissociation.Instead, the re-oxidation process can take place in two consecutive steps, namely the activation of O2 gas to form the peroxo species at the surface oxygen vacancy VO,S and its dissociation as soon as another subsurface O vacancy approaches the peroxo species.Therefore, the peroxo species is mandatory for the oxygen storage capacity (OSC). 71,72combination of infrared vibrational spectroscopy and DFT calculations 68 demonstrated that the activation of O2 at vacancies on single crystalline CeO2−x(100) and CeO2−x(110) takes place via peroxo and superoxo species, but not on CeO2−x(111).The main reason for this apparent discrepancy with the present study is the low adsorption temperature of 110 K in Ref. 68 For such low temperatures, oxygen vacancies of CeO2−x(111) are frozen in the subsurface region and are not accessible to O2 adsorption from the gas phase.In our case O2 adsorption takes place at higher temperatures and therefore oxygen vacancies, although with low concentration, are available on the surface of CeO2−x(111).
The peroxo group at the surface can diffuse across the surface.In fact, not the whole peroxo species diffuses, but rather the upper O atom of O2 2− diffuses to the neighboring surface lattice O site and transforms it into another peroxo species.With an activation energy of 1.2 eV, this barrier can easily be overcome at 500−700 K, making the peroxo species a mobile surface species.For supported catalysts with CeO2 being the carrier, the mobility of this peroxo species will be important for facile oxygen exchange between the active particle and the CeO2 support in the form of spill-over and back spill-over species.Superoxide (O2  ) and peroxide (peroxo: O2 2 ) are known to be important oxygen species in catalytic oxidation reactions 87 on ceria-based catalysts that were studied both experimentally and theoretically (see, e.g., Ref. 52,56,87−97 ).In this work, we demonstrate the crucial role of the peroxo species in the HCl oxidation reaction during the de-chlorination process.The peroxo species induces the displacement of Cl from the vacancy position (Clvac) towards the on-top position (Cltop) and the subsequent desorption in the form of Cl2.The concerted displacement of Cl maintains the oxidation state of −1, while the desorption step leads to the oxidation of chlorine towards zero oxidation state.Concerted displacement of reaction intermediates may be important in oxidation catalysis over ceria.

Conclusion
During the HCl oxidation reaction ceria catalysts undergo a redox cycle and their catalytic activity is governed by the ease with which the chlorinated CeO2−x surfaces can be dechlorinated by oxygen exposure, thereby releasing the desired product Cl2.Based on a previous DFT study, the displacement step of Clvac species towards Cltop was predicted to be the critical reaction step in the HCl oxidation activity of ceria. 1 Unfortunately, this displacement step is highly endothermic. 1Synchrotron-based methods, including in situ X-ray absorption spectroscopy (XANES) and in situ high-resolution core level shift spectroscopy (HRCLS), together with first principles DFT calculations are employed to elucidate this critical reaction step in the HCl oxidation reaction with a dedicated model experiment, namely the re-oxidation of a chlorinated single crystalline CeO2−x(111) model catalyst at 700 K, a typical reaction temperature for the HCl oxidation reaction.The synchrotron photon energies in HRCLS are carefully chosen to be highly surface sensitive, while Ce M4,5 edge XANES (Ce 3d) in total electron yield probes bulk properties.With this combination of surface and bulk sensitive methods, we demonstrate that the re-oxidation of the chlorinated CeO2−x(111) surface at 700 K starts from the bulk and propagates subsequently towards the surface.The re-oxidation of Clvac-CeO2−x(111) considerably weakens the adsorption energy of the Clvac species.Ultimately, part of the Clvac species are shifted to on-top positions Cltop, where they recombine to form Cl2. Both the re-oxidation of CeO2−x(111) and the Clvac → Cltop displacement step of surface chlorine are predicted by DFT calculations to be induced by peroxo species (O2 2 ).In this way the displacement step and surface re-oxidation are coupled so that the concerted displacement step becomes now exothermic by 0.6 eV with an activation barrier of about ~1 eV.With such a low activation barrier the Clvac → Cltop site change is shown to take place even at 500 K, although requiring a higher O2 exposure than for the 700 K re-oxidation.The peroxo species do not only impact the Deacon process, but are of general importance for catalytic oxidation reactions on CeO2 supported catalysts.Oxygen spill-over effects for supported particles on CeO2 are intimately correlated with the peroxo species and their facile (structure) diffusion across the surface.

Figure 1 .
Figure 1.Proposed reaction pathway for the HCl oxidation over the single crystalline CeO2(111) surface (top view), re-arranged and based on Amrute et al. 1 Only the outermost OCeO trilayer is depicted.Color code: Ce 4+ cations are white spheres, the O surface (subsurface) atoms are red (light red), Cl atoms are green and H atoms are yellow.

Figure 2 .
Figure 2. Bulk re-oxidation of Clvac-CeO2−x(111): Bulk sensitive XANES Ce M4,5 edge (Ce 3d) data (total electron yield: TEY) after exposing of Clvac-CeO2−x(111) to various doses of O2 at 700 K indicated in Langmuir (L).The reference spectra for fully oxidized CeO2 and fully reduced Ce2O3 are superimposed and serve as references for fitting by a linear combination of these reference spectra to determine the bulk Ce 3+ fraction.The optimized linear combination for each spectrum are shown as thin double line (shifted for clarity reasons) in the same color as the experimental spectra.
sensitive O 1s spectra are shown in Figure S7b.Here a more continuous shift of the O 1s binding energy is evident, consistent with the XANES experiments shown in Figure 2.

Figure 4 .
Figure 4. Surface re-oxidation of Clvac-CeO2−x(111): Surface sensitive Ce 4d spectra (hν = 250 eV) after exposing of Cl-CeO2−x(111) to various doses of O2 at 700 K indicated in Langmuir (L).The reference spectra for fully oxidized CeO2 and fully reduced Ce2O3 are superimposed and serve as references for fitting by a linear combination of these reference spectra to determine the near-surface Ce 3+ fraction.The optimized linear combination for each spectrum are overlaid as thin double line (shifted for clarity reasons) in the same color as the experimental spectra.

Figure 6 .
Figure 6.HCl thermal desorption spectra of a saturated Cl overlayer on CeO2−x(111) surfaces before and after re-oxidation (2700 L O2) at 700 K and 600 K.

Figure 7 .
Figure 7. O2-activation at 3Clvac-Ce3O5(111)-(3  3): a) Energy diffusion path of the oxygen vacancy from the third oxygen layer (VO,SSS) to the surface (VO,S) of 3Clvac-Ce3O5(111) and the O2 2− adsorption state on the oxygen vacancy (peroxo species: O2 2 ).b) Adsorption and c) Dissociation of adsorbed peroxo to Otop and Ovac species.Color code: Ce 4+ on the first/second ceria trilayer are white/yellow and Ce 3+ atoms are gray, the oxygen surface (subsurface) atoms are red (light red), and Cl atoms are in green.The blue star indicates the oxygen vacancy.

Figure 8 .
Figure 8. Re-oxidation process of 3Clvac-Ce3O5(111)-(3  3).The first step is the diffusion of an oxygen vacancy VO,SSS to VO,SS.Then O2 2− dissociates and fills in the VO,SS.Depending on the local configuration this dissociation process is activated by 0.31 eV or 0.68 eV.Color code: Ce 4+ are white and Ce 3+ atoms are gray, the oxygen surface (subsurface) atoms are red (light red), and Cl atoms are in green.The adsorbed O2 2 is depicted in violet, the blue arrow indicates the movement of the O atom that fills the VO,S upon O2 2− dissociation.

Figure 9 .
Figure 9. De-chlorination process of 3Clvac-Ce3O5(111)-(3  3).Two peroxo species at the surface displace Clvac to the Cltop.This concerted process is activated by 1.04 eV and is exothermic by 0.6 eV.Subsequently, the two Cltop species recombine to form the desired product Cl2.Color code: Ce 4+ are white and Ce 3+ atoms are gray, the oxygen surface (subsurface) atoms are red (light red), and Cl atoms are in green.The adsorbed O2 2− species are depicted in violet.O* indicates the lattice oxygen atoms that formed after O2 2− dissociation.

Figure 10 .
Figure 10.Cl 2p lab XP spectra: a) de-chlorination and re-oxidation at 500 K with 43.2kL O2. b) de-chlorination and re-oxidation at 300 K with 55L atomic O, c) and d): integrated intensity of the Cl 2p peak of a) and b) respectively.Note that the exposures in panel c) are given in kL.Detector angle θ = 60° for enhanced surface sensitivity.