Operando Tracking the Interactions between CoOx and CeO2 during Oxygen Evolution Reaction

CeO2 greatly enhances the electrocatalytic oxygen evolution reaction (OER) activity of CoOx, though the enhancement mechanism beyond this synergy is yet to be understood. Here, operando hard X‐ray absorption spectroscopy (hXAS) is applied to monitor the Co K edge and Ce L3 edge in CoOx/CeO2 to shed light on the evolution of the Co and Ce oxidation states during OER. In addition, ex situ soft XAS (sXAS) characterizations provide information on the irreversible surface‐specific transformations of the Co L3 edge as well as of the O K edge. Combining the operando and ex situ spectroscopic characterizations with comprehensive electrochemical analyses, it is confirmed that CeO2 is not the active center for the OER. However, coupling CeO2 with CoOx introduces significant modifications in the Co and O species at the CoOx surface and alters the flat band potential (Efb), leading to more favorable Co oxidation state transformations during OER and possibly modifying the preferential reaction pathway. This work establishes the connections between electronic structures, Co oxidation state and the OER reaction mechanism for CoOx/CeO2 composites electrodes.


Introduction
Water electrolysis powered by renewable electricity is one of the most promising means to achieve sustainable, carbon-neutral long-term energy storage in the form of hydrogen in a multienergy carrier energy system. [1]Recently, alkaline water splitting has attracted increasing interest, with rapid progression in the development on anion exchange membrane, electrocatalysts, and devices. [2]The alkaline environment enables the application of transition-metal-based oxides as cost-effective OER catalysts. [3]and optimize the binding energy of OER intermediates.Qiu et al. [7b] suggested the built-in electron field in the Co 3 O 4 /CeO 2 pn heterojunction can reduce the Co III to Co II in the octahedral site and produce oxygen vacancies.In comparison, Liu et al. [7c] suggested there was also a high concentration of oxygen vacancies at the Co 3 O 4 /CeO 2 interface; however, rather than Co III being reduced, Ce IV was reduced to Ce III to facilitate the electron transfer.Moreover, Li et al. [7d] proposed the Ce donated more electron to the lattice oxygen to reinforce the Co─O covalency.Overall, even though different interaction mechanisms have been proposed, little attention has been paid to the interactions between Co and Ce during the OER due to the lack of operando characterizations.3c,8] Systematically linking structural properties, electronic features and electrochemical behavior is helpful to reach a more comprehensive understanding of the nature of Co─Ce interactions and subsequent enhancement of the OER activity.
Herein, using the advantages of operando hXAS characterizations at the Co K edge and Ce L 3 edge, we are able to monitor the evolution of both CoO x and CeO 2 in a CoO x /CeO 2 nanocomposite during OER.We confirm that the Ce oxidation state does not change with increasing potential in the OER region, thus it is not the active center.Conversely, the Co oxidation state in CoO x /CeO 2 responds clearly to the applied potential but behaves differently compared to CoO x without CeO 2 .With the help of ex situ sXAS and in situ impedance spectroscopic characterizations, we confirm the interactions between CeO 2 and CoO x could trigger a modification of both structural and electronic material charac-teristics.Finally, investigation of the pH-dependence of the OER activity shows that CoO x and CoO x /CeO 2 have different reaction orders, indicating that proton and electron transfer during the rate-determining step (RDS) may be also altered due to interactions between CoO x and CeO 2 .

Results and Discussion
3a] The structural properties of the as-prepared catalysts were verified by X-ray diffraction (XRD), as shown in Figure 1a.The CoO x sample is composed of CoO and Co 3 O 4 .When 10% (molar ratio) of Ce is introduced during synthesis, there is no obvious shift in the XRD peaks; rather, CeO 2 is present as a separate phase, thus the sample is denoted as CoO x /CeO 2 .Other control samples with different Ce/(Ce+Co) percentages have also been synthesized following similar protocol for comparison (Figure S1, Supporting Information).
Furthermore, the ex situ hXAS was applied to understand the Co oxidation state and the electronic structure of the as-prepared catalysts.The X-ray absorption near edge spectroscopy (XANES) spectra at the Co K edge for both CoO x and CoO x /CeO 2 are almost overlapping, suggesting no significant difference in the average Co oxidation state (Figure 1b).The energy of the absorption edge (E edge ) was determined by using either an integration method [6b,9] (inset in Figure 1b) or the energy at half (0.5 in arbitrary units) of the normalized edge jump [10] (Figure S2a, Supporting Information), obtaining almost identical trend among the samples.Furthermore, the E edge values for both CoO x and CoO x /CeO 2 sit between the reference CoO and Co 3 O 4 samples, with the Co oxidation of around 2.26+ from linear regression (Figure S2b, Supporting Information), as would be expected in a composite of CoO and Co 3 O 4 .Therefore, the Fourier transform (FT) of the Co K edge extended X-ray absorption fine structure (EXAFS) shows the combining features of CoO and Co 3 O 4 structures for both single material CoO x and CoO x /CeO 2 composite electrode (Figure S3a, Supporting Information).The Ce oxidation state in the CoO x /CeO 2 was identified from the Ce L 3 edge spectrum, using Ce(NO 3 ) 3 and CeO 2 as references (Figure 1c).The XANES spectrum, and the corresponding FT-EXAFS spectrum (Figure S3b, Supporting Information) of Ce L 3 edge in CoO x /CeO 2 are similar to those of the CeO 2 reference, proving that the Ce in the composite exists as the fluorite CeO 2 structure.
Since the OER is a surface process, the surface oxidation state of the catalyst will strongly relate to its catalytic performance.The sXAS combined with surface-sensitive total electron yield detection was used to study the Co L edge, O K edge, and Ce M edge of CoO x and CoO x /CeO 2 (Figure 1d-f).The full spectra for the Co L edge are shown in Figure S4 (Supporting Information), while the Co L 3 edge is highlighted in Figure 1d to demonstrate the Co oxidation state at the surface.The peaks, from low to high absorption energy, are attributed to octahedral Co II (≈722 eV, in region A), tetrahedral/octahedral Co II (≈779 eV, in region B) and Co III (≈780 eV, in region C), respectively. [11]The arrows in Figure 1d shows the trend of peak shift in different regions with increasing Co oxidization.Clearly, the peak intensity for both CoO x and CoO x /CeO 2 is greatly reduced in region A compared to that of reference CoO.Generally, the as-prepared CoO x from flame spray synthesis show a Co 3 O 4 -like surface.They overlap with reference Co 3 O 4 in region B, however, the peaks have lower intensity in region C.These results are consistent with the trend in average Co oxidation state observed in the Co K edge from hXAS.The peak of CoO x /CeO 2 in region C is slightly higher than that of CoO x , indicating a higher fraction of Co III species are present on its surface.Furthermore, O K edge was probed to reveal the orbital hybridization between Co 3d and O 2p (Figure 1e).The pre-edge resonance at ≈530.7 eV (in region D) is attributed to the Co III ─O hybridized state, while the one at ≈532 eV (in region E) is suggested to be the protonated bridging oxygen that adsorbs at defective sites (e.g., corner, edge and metal vacancy sites). [11,12]To find out the contribution of CeO 2 to the O K edge, a CeO 2 reference sample has also measured (Figure 1f).The peak at ≈533 eV in the reference CeO 2 , due to the hybridization between O 2p and Ce 5d-e g orbitals, [13] does not overlap with peak assigned to the protonated bridging oxygen in CoO x /CeO 2 , which suggests this interested peak is not directly contributed from the CeO 2 in the composite.Furthermore, the surface-sensitive sXAS results revealed no obvious shift in the Ce M edge of CoO x /CeO 2 compared to the reference CeO 2 , further proving that Ce is majorly present in the 4+ (Figure 1f).In general, the sXAS results indicate the interactions between CoO x and CeO 2 endow the CoO x /CeO 2 composite with a defective surface, potentially leading to improved OER performance.
To evaluate the OER activity, cyclic voltammetry (CV) was conducted in 0.1 M KOH to investigate the Co redox properties and the OER current in CoO x and CoO x /CeO 2 (Figure 2a).Typically, there are two pairs of redox peaks in both samples, which are at-tributed to Co II/III and Co III/IV , at ≈1.21 V and 1.48 V versus RHE, respectively. [14]The CoO x /CeO 2 shows more prominent Co III/IV redox peaks and a reduced overpotential for OER.To exclude impacts from the pseudocapacitive current attributed from Co redox processes, a steady-state chronoamperometric (CA) technique was applied to obtain the polarization curve for OER (Figure S5, Supporting Information).The CoO x /CeO 2 still shows earlier onset and thus has improved OER activity.Compared to CoO x , the mass activity of CoO x /CeO 2 at 1.55 V versus RHE is remarkably enhanced (≈12.6 time higher), and at a current density of 10 A g −1 the overpotential is 40 mV lower (Figure 2b).The control samples synthesized with different percentages of Ce have also been investigated (Figure S6, Supporting Information).The CoO x /CeO 2 with 10% Ce, the focus of this work, shows the best OER performance; nevertheless, also the composite samples with Ce in the 2-30% range outperform CoO x in the OER activity.Therefore, adding an optimized amount of CeO 2 to CoO x brings simultaneously two major advantages: it allows the amount of Co, whose price is constantly increasing, to be reduced and the OER activity to be increased.Furthermore, the control samples of physical mixing of CeO 2 with CoO x (or Co 3 O 4 ) do not show improved OER activity (Figure S7, Supporting Information), suggesting that the establishment of interactions at the interface of CoO x and CeO 2 particles is crucial to induce the synergy between the two phases in the CoO x /CeO 2 composite.A possible O-bridged interaction between Co and Ce at the interface may yield an improved absorption strength of OER intermediates, leading to improved OER activity. [15]e further verify the contributions of surface area and conductivity to the improved OER performance of CoO x /CeO 2 .The surface areas of the catalysts were estimated by two different methods (Figure 2c; Figure S8, Supporting Information).The Brunauer-Emmett-Teller (BET) surface area of CoO x /CeO 2 (≈28.4 m 2 g −1 ) is only slightly higher than that of CoO x (≈26.0 m 2 g −1 ).Nevertheless, CoO x /C eO 2 shows higher specific activity, whether normalized by BET surface area or double layer capacitance (C dl ) (Figure S9, Supporting Information).The double layer capacitance (C dl ), extracted from the CVs at different scan rates (Figure S8c, Supporting Information), is smaller for the CoO x /CeO 2 (≈0.79 mF g −1 ) compared to CoO x (≈ 1.43 mF g −1 ), due to the contribution from CeO 2 .Four-wire impedance spectroscopy [16] was conducted to show that the ex situ electron conductivity of CoO x and CoO x /CeO 2 is on the same order of magnitude (10 −6 S cm −1 ) (Figure 2d), proving the adding CeO 2 does not hamper electron transfer among the particles.Besides, according to electrochemical impedance spectroscopy (EIS) measurements at 1.60 V versus RHE, CoO x /CeO 2 shows better electrode kinetics with a smaller electron transfer resistance of ≈87.9 ohm at 1.60 V versus RHE compared to ≈270.5 ohm for CoO x (Figure 2d; Figure S10, Supporting Information), in agreement with OER activity measurements.In all, combining CeO 2 with CoO x improves the intrinsic OER activity.
To understand how the synergy between CoO x and CeO 2 enhances the OER activity, ex situ sXAS characterizations in total electron yield (TEY) mode have been used to monitor the evolution of the Co L 3 edge and O K edge before and after the OER.The Co L 3 edge of dropcasted CoO x and CoO x /CeO 2 (i.e., before the OER) and after the OER were compared to that of the assynthesized powder (Figure 3a,b).The major changes happening in region C correlate to the formation of octahedral Co III .To better present the differences between CoO x and CoO x /CeO 2 , the peak intensity ratio between region C and B (denoted as I C /I B ) was calculated (insets in Figure 3a,b).CoO x /CeO 2 powder has a slightly higher I C /I B compared to CoO x powder (≈2.05 versus 1.93), suggesting there is an initially higher Co oxidation state on the surface.The dropcasted samples, in which the catalyst makes contact with water during ink preparation, see an I C /I B increase of ≈0.12 and 0.05 for CoO x and CoO x /CeO 2 , respectively.It is reported that a transition from rocksalt CoO into Co 3 O 4 will happen under open circuit potential. [17]The changes in Co L 3 edge indicates there is chemical reconstruction during electrode preparation, and it is more significant for the CoO x .The I C /I B is further increased after the OER for both CoO x and CoO x /CeO 2 , suggesting a higher fraction of Co III is generated.Moreover, the discrepancy in surface oxidation between CoO x and CoO x /CeO 2 is more obvious in the O K edge spectra (Figure 3c,d).Similarly, the peak intensity ratio between regions E and D (denoted as I E /I D ) has been calculated, with I E /I D = ≈0.34 for the Co 3 O 4 reference (Figure S11, Supporting Information).Clearly, the I E /I D of CoO x is similar to that of the Co 3 O 4 reference under all conditions (i.e., as-synthesized powder, before the OER, after the OER) (inset in Figure 3c).Conversely, the I E /I D of CoO x /CeO 2 is much larger than 0.34 under all three conditions, suggesting a higher fraction of the protonated bridging oxygen is present at the defective surface sites.Notably, the I E /I D of CoO x /CeO 2 reached ≈ 1.09 after contacting with water and before the OER and was decreased down at ≈0.55 after the OER (inset in Figure 3d).Therefore, it is evident that the interactions between CoO x and CeO 2 in CoO x /CeO 2 modulate the electronic structure and surface Co oxidation state of the composite during the electrocatalytic process.Usually, the OCP of a material is related to its electronic structure. [18]As the OCP of CoO x /CeO 2 is 110 mV higher than that of CoO x , this confirms that a modulation in electronic structure is induced by CeO 2 (Figure S12, Supporting Information).
Operando hXAS characterizations of the Co K edge and Ce L 3 edge were performed to track the interactions between CoO x and CeO 2 during the OER (Figure 4).The hXAS spectra acquisition has a time-resolution of two spectra per second, thus enabling the capture of energy shifts in the E edge during CV measurements with a slow scan rate of 2 mV s −1 .To better present the results, the 20 spectra collected every 10 s are averaged to have the resolution of 20 mV at the applied potential.The applied potential window for each operando measurement has been optimized by limiting the maximum current density to avoid gas bubbles formation impeding the quality of the XAS data obtained.Due to the higher activity of the CoO x /CeO 2 composite, the upper applied potential used was reduced from 1.63 to 1.6 V (Figure S13, Supporting Information).The XANES spectra of the Co K edge for both CoO x and CoO x /CeO 2 shift to higher energy with increasing applied potential, due to the oxidation of Co (Figure 4a,b).6b,9] The energy shift (ΔE edge ), representing the change in Co oxidation state, is obtained by comparing the E edge at a specific applied potential to that seen at a potential of 1.0 V versus RHE during the first anodic scan.Taking the data for the first anodic scan as an example, the ΔE edge of CoO x increased continuously with the applied potential from 1.0 to 1.63 V versus RHE, whereas the ΔE edge of CoO x /CeO 2 showed a bi-phasic tendency, growing slowly before the Co III/IV redox peaks (≈1.32 V vs RHE in Figure 2a) but increasing steeply after that (Figure S14, Supporting Information).The ΔE edge of CoO x /CeO 2 before the Co III/IV redox peaks is only ≈10 meV, in contrast to ≈30 meV for CoO x .The increase of Co oxidation state (i.e., ΔE edge ) before the OER can be correlated to surface deprotonation. [14,19] total of 10 CV cycles were measured by scanning the potential forth and back (Figure 4d), to verify the reversibility of the Co oxidation state in CoO x (Figure 4e).The ΔE edge of CoO x overlaps in the first anodic and cathodic scan, and there are no obvious changes even after 10 cycles (Figure 4e); the ΔE edge of the Co foil used as a reference show no obvious periodic changes (Figure S15, Supporting Information), excluding the influence from the beam and other setups.In comparison, the ΔE edge of CoO x /CeO 2 shows an increase of ≈27 meV (indicated as the green dashed arrow in Figure 4f) as the potential is decreased from 1.6 to 1.0 V versus RHE at the end of the tenth cycle.10b,20] Indeed, the reversibility of Co oxidation is slightly modified by the formation of the CoO x /CeO 2 composite.
The periodic change of Co K edge showing a slight change E edge indicates that Co is the active center for OER (Figure 4e,f).The reproducible and reversible small change in the Co K edge is additionally demonstrated through multivariate analysis (Figure S16, Supporting Information).In comparison, the evolution of the Ce oxidation state was monitored via the Ce L 3 edge spectra, with no apparent changes in the E edge for Ce L 3 edge with the applied potential over the whole 10 CV cycles (Figure 4g).A similar trend has also been detected for the single phase CeO 2 reference material, which is inactive toward OER (Figure S6, Supporting Information), clearly showing that Ce is not redox-active during OER (Figure S17, Supporting Information).Thus far, we can safely conclude that CeO 2 does not directly participate in the OER as an active site.
After operando tracking of the Co oxidation state during CV, we have further conducted steady-state choropotentiometry (CP) measurements (Figures S18 and S19, Supporting Information).The ΔE edge (i.e., the change in Co oxidation state) has been plotted as the function of both the applied potential and the logarithm of current density (Figure 5a,b).The ΔE edge of CoO x /CeO 2 is ≈30 meV at ≈1.60 V versus RHE, that is, in the OER-region (Figure 5a), while for CoO x is ≈60 meV consistently with the operando hXAS results observed during CV measurements (Figure 4d,e).The ΔE edge for CoO x /CeO 2 is also smaller than that of CoO x at the same current value (Figure 5b).The variance in ΔE edge thus Co oxidation state could alter both the interfacial capacitance and total charge on the electrode surface.To reveal the change in the interfacial capacitance, in situ impedance spectroscopy at different applied potentials (Figure S20, Supporting Information) was conducted to derive Mott-Schottky plots (Figure 5c).The negative slope indicates that CoO x is p-type in nature, [8c,21] in contrast to the CeO 2 reference (Figure S21, Supporting Information).The slope for CoO x /CeO 2 is also negative, but is less steep than that of CoO x , suggesting a higher electron hole concentration is present at the surface since this is inversely proportional to the gradient. [22]Besides, the flat band potential (E fb ) was extracted from the intercept of the Mott-Schottky plots.
The different E fb of 1.32 V versus RHE for CoO x /CeO 2 compared to that of CoO x (1.15 V versus RHE) verifies the electronic modulation that arises from the addition of CeO 2 to CoO x .More interestingly, the E fb can have an impact on the evolution of the Co oxidation state during OER.Indeed, theoretically electron holes can be generated with Co being oxidized and can accumulate in the space-charge region when the applied potential is higher than E fb (inset in Figure 5c). [23]This can explain why the ΔE edge (see Figure 5a) increases after ≈1.15 V and 1.32 V versus RHE in CoO x and CoO x /CeO 2 , respectively (i.e., after the E fb ).
Furthermore, a pulse voltammetry protocol [24] (see Figures S22 and S23, Supporting Information, for more details) was used to reveal the total charge related to space charge region and electric double layer at the catalyst/electrolyte interface (Figure 5d).There is a linear relationship between the total charge and log (J), indicating that the generated current is controlled by the accumulated charge in the catalyst. [24]The total charge in CoO x at the same log (J) value is always higher than that of CoO x /CeO 2 , likely correlated to the higher ΔE edge (i.e., Co oxidation) during OER for CoO x compared to CoO x /CeO 2 as observed by operando hXAS during CP (Figure 5b).Moreover, Figure 5b also shows that the ΔE edge for CoO x has stronger dependence on the log (J) in the OER region; by comparison, the ΔE edge of CoO x /CeO 2 is almost unchanged.These results suggest two catalysts may follow different reaction pathways during the OER.S18, Supporting Information).The ΔE edge herein is calculated by comparing E edge at different condition to that seen at the initial step with a constant current of 0.001 mA.c) Mott-Schottky plots to extract the E fb ; inset shows the space charge region at the surface of the catalysts when the applied potential is higher than E fb .d) Total charge at the catalyst/electrolyte surface as a function of log (J) that measured in the rotating disk electrode; inset shows the total charge accumulated at the space charge region and electric double layer.The error bars in this figure are from the averages on three data points.
To shed light on the reaction mechanism, a pH-dependence study was conducted.Tafel plots were made to compare the OER activity of CoO x and CoO x /CeO 2 in the electrolytes of different OH − concentration, and their reaction order at 1.60 V versus RHE was determined to be ≈0.46 and 0.26, respectively (Figure 6a-c).A higher reaction order implies a stronger dependence of activity on pH, suggesting the proton (or OH − ) and electron transfer steps are decoupled during the rate-determining step (RDS). [25]In contrast, pH-independent OER activity involves a concerted proton electron transfer (CPET) during the RDS.The reaction order of CoO x /CeO 2 is lower than that of CoO x , suggesting the RDS is more likely to be dominated by a CPET process.
In an alkaline environment, the pH-dependent activity is majorly controlled by the OH − and/or electron transfer in the RDS.10b,26] Therefore, we propose two reaction pathways: one has a decoupled proton transfer as the RDS, while the other has a CPET as RDS.CoO x shows no obvious peak for the protonated bridging oxygen in the O K edge spectra (Figure 3c); the ΔE edge of the Co K edge shows stronger dependence on log (J) and is higher than that in CoO x /CeO 2 (Figure 5b).Finally, CoO x presents a reaction order of ≈0.46.Consequently, we propose that CoO x would largely follow the non-concerted reaction pathway (Figure 6d).The OER cycle may start with a decoupled electron transfer in Co III ─O─Co IV to form the Co IV dimer, and then proceed with deprotonation via OH − attack during the RDS.By comparison, the CoO x /CeO 2 shows an obvious peak for the protonated bridging oxygen in the O K edge spectra (Figure 3d); the ΔE edge of the Co K edge is both less dependence on log (J) and lower than that of CoO x (Figure 5b).In addition, CoO x /CeO 2 shows a small reaction order (≈0.26) suggesting that a CPET process likely dominates the RDS in CoO x /CeO 2 (Figure 6e).The Co III ─OH─Co IV , with protonated bridging oxygen, may proceed the O─O formation, via a one electron transfer process coupled with deprotonation by OH − attack.In this mechanism, there is no need to form the Co IV dimer, which explains the lower ΔE edge and its reduced dependence on log (J).Note that different reaction pathways could occur concurrently (and this is why we talk about a dominating mechanism) and there could be other possible reaction pathways, for instance involving a single Co as the active center (Figure S24, Supporting Information), or the direct O─O coupling in a Co IV dimer without the participation of bridging oxygen (Figure S25, Supporting Information).We anticipate those pathways are not dominant in the two catalysts, since they cannot fully explain the reaction order and the trend of Co oxidation state changes during OER.In all, the two preferred reaction pathways are based on the outlined structural features and electronic transformations, which demonstrate the different electrochemical properties of CoO x and CoO x /CeO 2 .Evidently, though CeO 2 does not directly participate in the OER reaction mechanism, its electronic modulation effects can alter the CoO x surface structure and Co redox properties to change the reaction mechanism and improve the OER activity of CoO x .

Conclusion
In summary, the OER enhancement mechanisms of CoO x /CeO 2 nanocomposite compared to CoO x have been systematically revealed by combining ex situ/operando spectroscopic characterizations with comprehensive electrochemical analyses.Introducing CeO 2 to CoO x leads to a higher fraction of surface Co III and protonated bridging oxygen, as evidenced by ex situ sXAS characterizations.With the help of operando hXAS characterizations, we showed that the evolution of the Co oxidation state was controlled by the E fb .The E fb is modulated by the introduction of CeO 2 , thus, the CoO x /CeO 2 shows different Co oxidation behavior with applied potential.The energy shift in the Co K edge of CoO x /CeO 2 is less significant and less dependent on log (J) compared that of CoO x .Besides, the CoO x /CeO 2 has a smaller reaction order for OH − than that of CoO x , as revealed by the pH-dependence study.Based on the spectroscopic characterizations and electrochemical analyses, we propose the RDS for the CoO x /CeO 2 is dominated by a CPET process, with involvement of protonated bridging oxygen, in comparison to a non-concerted reaction pathway for CoO x .Finally, we find that CeO 2 is not directly involved in the OER cycle, though it modifies the structural and electronic properties of CoO x , changes the reaction pathway and improves the OER activity.

Figure 1 .
Figure 1.Structural characterizations on CoO x and CoO x /CeO 2 .a) XRD patterns; inset shows the CoO (002) peak.b) XANES spectra at the Co K edge; inset shows the E edge of different samples.c) XANES spectra at the Ce L 3 edge.The sXAS spectra at the d) Co L 3 edge, e) O K edge, and f) Ce M 4,5 edge.

Figure 2 .
Figure 2. Electrochemical characterizations on CoO x and CoO x /CeO 2 .a) CVs collected in 0.1 M of KOH, with a scan rate of 50 mV s −1 .The inset highlights the Co II/III and Co III/IV redox pairs.b) Tafel plots obtained by CA measurement; the error bars are from the averages on three individual measurements.c) BET surface area calculated from N 2 absorption isotherm curves and C dl ; the error bars are from fitting.d) Ex situ conductivity extracted from four-wire impedance spectroscopy and electron transfer resistance obtained by EIS measurement at 1.60 V versus RHE; the error bars are from the averages on three individual measurements and the fitting, respectively.

Figure 3 .
Figure 3. Ex situ sXAS characterizations (in TEY mode) on CoO x and CoO x /CeO 2 to compare the Co L 3 edge and O K edge of powder to those before OER (prepared in the catalyst ink and dropcased on glassy carbon) and after OER.The Co L 3 edge spectra of a) CoO x and b) CoO x /CeO 2 ; the insets display the corresponding peak intensity ratio between region C and B (I C /I B ) at the Co L 3 edge, with the dash line at 2.27 from Co 3 O 4 reference for comparison.The O K edge of c) CoO x and d) CoO x /CeO 2 ; the insets display the corresponding peak intensity ratio between region E and D (I E /I D ) in the O K edge, with the dash line at 0.34 from Co 3 O 4 reference for comparison.

Figure 4 .
Figure 4. Operando hXAS characterizations on CoO x and CoO x /CeO 2 to monitor the evolution of Co K edge and Ce L 3 edge during CV measurement with a scan rate of 2 mV s −1 .The XANES spectra at the a) Co K edge of CoO x , b) Co K edge of CoO x /CeO 2 and c) Ce L 3 edge of CoO x /CeO 2 during the first anodic scan.Insets in panel (a-c) highlights the shift in absorption edge.d) The applied potential and the OER current of CoO x electrode as a function of time during the 10 CV cycles of the operando hXAS characterization.The corresponding energy shift (ΔE edge ) at the Co K edge spectra for e) CoO x and f) CoO x /CeO 2 during 10 CV cycles.g) The ΔE edge at the Ce L 3 edge spectra for the CoO x /CeO 2 .The ΔE edge in panel (e-g) was obtained by comparing the energy of absorption edge (E edge ) at the specific potential to that at 1.0 V versus RHE during first anodic scan.

Figure 5 .
Figure 5. Correlations among Co oxidation state, electronic structures and surface charge in CoO x and CoO x /CeO 2 .The ΔE edge being plotted against the a) applied potential and b) log (J) that obtained during the steady-state CP measurement in a flow cell (FigureS18, Supporting Information).The ΔE edge herein is calculated by comparing E edge at different condition to that seen at the initial step with a constant current of 0.001 mA.c) Mott-Schottky plots to extract the E fb ; inset shows the space charge region at the surface of the catalysts when the applied potential is higher than E fb .d) Total charge at the catalyst/electrolyte surface as a function of log (J) that measured in the rotating disk electrode; inset shows the total charge accumulated at the space charge region and electric double layer.The error bars in this figure are from the averages on three data points.

Figure 6 .
Figure 6.Reaction order and proposed reaction pathways.The pH-dependent Tafel plots of a) CoO x and b) CoO x /CeO 2 .c) The log (J) at 1.6 V versus RHE was plotted against the log [c(OH) − ] to extract the reaction order.The error bars are from the averages on three individual measurements.The proposed reaction pathways d) with chemical proton transfer as the RDS, or e) with concerted proton electron transfer during RDS.The oxygen that participated in the O─O bond formation is marked in red.