In-situ X-ray techniques for non-noble electrocatalysts

In-situ hard X-ray absorption spectroscopy (XAS)

X-ray absorption spectroscopy (XAS) contains X-ray absorption near edge spectra (XANES) and extended X-ray absorption fine structure (EXAFS). The former allows us to recognize the average oxidation number and the electronic configuration of the specific elements in materials while the latter provides the information of interatomic distance and the coordination number. The local atomic environment can be described for each type of atom in a chemical compound, so that the geometrical sites of the occupied atoms, e.g. tetrahedral or octahedral sites, can be identified. Besides, the material does not require an ordered structure, which means that atom in amorphous materials or molecular samples can be investigated. These natures make XAS as a powerful manner to in-situ probe the structural and chemical variation in the electrocatalysts during the catalysis.

Recently, various composite electrocatalysts have been developed and exhibited high activities for OER, HER, ORR, and CO2RR. In-situ XAS is frequently utilized to identify the active sites via analyzing the shift of near edge, indicative of the change in oxidation state, and the deviation of interatomic distance during the electrocatalysis. Interestingly, it is commonly observable that a particular element in the composite material is primarily in charge of catalytic center while the other elements play the auxiliary roles to alter the properties of the active element or to provide decent surroundings for active species [58], [59], [60], [61]. Even for the same element with different geometrical sites, e.g. two types of geometrical cobalt ions in spinel Co₃O₄: Co²⁺ ions in the tetrahedral site (Co²⁺_(Td)) and two Co³⁺ ions in the octahedral site (Co³⁺_(Oh)), they present distinctive characteristics during OER [62]. As shown in Fig. 1a(i), in-situ Co K-edge EXAFS reveals that a small-scale shrinkage of the Co–O bond could be observed in Co²⁺_(Td) containing spinels (Co₃O₄ and CoAl₂O₄) during OER, indicative of partial oxidization on the catalyst surface, while not in Co³⁺_(Oh) spinel (ZnCo₂O₄). Moreover, the Co K-edge XANES spectra (Fig. 1a(ii),(iii)) show that the white line intensities keep unvarying for ZnCo₂O₄ but increasing with the applied potential for CoAl₂O₄. It means that CoAl₂O₄ with an initially low oxidation state (Co²⁺_(Td)) could be further oxidized with the positive applied potential, facilitating the formation of CoOOH on the surface, which have been regarded as the active species for water electrolysis on spinel Co₃O₄ [63], [64]. These in-situ results correspond to the superb OER activities of Co²⁺_(Td) containing spinels (Co₃O₄ and CoAl₂O₄) [62]. It implies that the Co³⁺_(Oh) ions are relatively inactive than Co²⁺_(Td) ions in spinel Co₃O₄, and the oxidation process on oxidizable Co ions is the crucial characteristic for high-efficient OER catalyst.

Fig. 1:

In-situ X-ray absorption spectroscopy (XAS) for various electrocatalysis. (a) For oxygen evolution reaction of spinel Co₃O₄ in alkaline electrolyte: (i) in-situ Co K-edge EXAFS spectra for Co₃O₄ (blue), ZnCo₂O₄ (red), and CoAl₂O₄ (green), where the applied voltage is referenced to RHE, and the enlarged peak for Co–O interatomic distance. (ii) Normalized in-situ Co K-edge XANES spectra for ZnCo₂O₄ and CoAl₂O₄. Reprinted with permission from [62]. Copyright 2016 American Chemical Society. (b) For hydrogen evolution reaction of Ni-thiolate coordinated polymer: (i, ii) in-situ Ni K-edge XANES and EXAFS spectra of Ni-BDT before and after electrochemical activation. (iii) Schematic illustration of the synthesis of Ni-BDT and in situ electrochemical production of Ni NSs with Sad^δ−. Reprinted with permission from [68]. Copyright 2017 Elsevier Inc. (c) For oxygen reduction reaction of Zn–Co atomic pairs coordinated on N doped carbon: in-situ XANES spectra of Zn/CoN–C for Co K-edge and Zn K-edge. Reprinted with permission from [69]. Copyright 2019 John Wiley & Sons, Inc. (d) For carbon dioxide reduction reaction of single Ni atom dispersed anchored on nitrogen-doped carbon: in-situ Ni K-edge XANES spectra at various biases (applied voltage versus RHE) in 0.5 M KHCO₃ in 1 atm of Ar or CO₂ and Fourier transform magnitudes of EXAFS spectra (without phase correction) under open-circuit voltage bias in Ar (OCV-Ar) and CO₂ (OCV-CO₂), and at −0.7 V (versus RHE), in which an expanded Ni–N bond was detected [55].

Transition-metal non-oxides or metal-organic coordination motifs have exhibited excellent catalytic activities [65], [66], [67]. In-situ techniques can be exploited to identify the genuine active materials and structural/compositional variation during the catalysis due to the harsh condition of electrocatalysis, such as basic/acid electrolyte and high applied potential. Zheng’s group has developed two-dimensional Ni-coordination polymer (Ni-BDT) nanosheets showing superior HER activity in an alkaline environment [68]. They further activate Ni-BDT at −20 mA/cm² for 12 h. The completely activated Ni-BDT (Ni-BDT-A) displays a surged catalytic performance and a comparable activity with commercial Pt/C at 100 mA/cm². Ex-situ X-ray photoelectron spectroscopy shows that the catalyst majorly consists of metallic Ni with a slim layer of Ni(OH)₂ on the surface after electrochemical activation. However, this layer possibly forms via post-oxidation as exposing the activated catalyst to the atmosphere before measuring the XPS. To recognize the real catalytic structures in Ni-BDT-A, in-situ XAS is conducted to analyze the electrocatalysts under the electrochemical activation. During this activation, the amount of Ni–S bond in Ni-BDT-A significantly drops. The XANES spectra (Fig. 1b(i)) show that the coordination environment in Ni-BDT-A is divergent from that in either Ni-BDT or Ni foil. In Fig. 1b(ii) of the EXAFS spectra, Ni-BDT-A contained metallic Ni–Ni and Ni–S bonds, and their coordination numbers are fitted as 4.9 and 1.3, respectively. The metallic Ni–Ni with comparably low coordination number suggests the evolution of ultrathin metallic Ni nanosheets. The absent Ni–O bond in the in-situ results confirm that the observed Ni(OH)₂ in the XPS spectra originates from the oxidation of metallic nickel before the XPS measurement. Therefore, after activation, the catalysts transform to metallic Ni nanosheet with a trace amount of sulfide on the surface, exhibiting the superb HER activity (Fig. 1b(iii)) rather than the polymer form of Ni-BDT.

The above-mentioned metal-organic coordination motifs, as the atomically-dispersed metal-ion catalysts, have shown outstanding activities toward various catalysis. However, their fragile structures and poor chemical stabilities limit the catalytic application. In recent times, atomically-dispersed nitrogen-coordinated transition-metal anchored on carbon materials present superior activities and stabilities. Sun’s group further demonstrates that the well-defined Zn–Co atomic pairs coordinated on N doped carbon support, which enables to regulate the binding ability of reactants or intermediates to elongate the bond length of oxygen molecules and thus facilitate the cleavage of oxygen molecules, exhibits superior ORR activities in both basic and acidic electrolytes [69]. In-situ XANES spectra can identify the active sites of Zn–Co atomic pairs, shown in Fig. 1c. The nearly identical Zn K-edge spectra at different operating voltage reveal the fact of the inactive nature of Zn ions during the ORR while the Co K-edge spectra constantly shift to higher energy with the applied potential, indicative of the Co ions as the active center for ORR. Through the entire ORR process, the oscillation behavior of Co and Zn K-edge spectra merely deviate. This fact suggests that the robust framework structure of Zn–Co atomic pairs facilitates the catalytic stability of the atomically-dispersed metal-ion materials. Furthermore, the central metal ions in atomically-dispersed nitrogen-coordinated transition-metal anchored on carbon materials present distinctive catalytic characteristics, demonstrating high selectivities and activities for CO2RR [70], [71], [72]. Liu’s group develops an atomically dispersed nickel on nitrogenated graphene as an electrocatalyst with high performance for CO2RR and deciphers the specific adsorption and activation process of CO₂ on this electrocatalyst [55]. As shown in Fig. 1d of in-situ XAS results, a high-energy shift of the Ni K-edge (an increment of Ni oxidation state) is observed in the CO₂-saturated KHCO₃ solution under open-circuit voltage in comparison to that in the Ar-saturated one. It is seemingly attributed to the delocalization of the unpaired electron in 3dx²-y² orbital of nickel ion and the spontaneous charge transfer from central nickel ion to the carbon 2p orbital in CO₂ owing to the evolution of CO₂^δ− intermediates. After switching on the CO2RR in the voltage bias of −0.7 V vs. RHE, the Ni K-edge shifts toward low energy. This relatively small average oxidation state of Ni ions implies that the adsorbed CO₂^δ− intermediates continuously undergo the reductive catalytic cycles and thus retrieve the low-oxidation-state Ni during CO2RR. In-situ EXAFS spectra reveal that the main peak heightens on a small scale, resulting from the participating contribution of Ni–C bond due to the adsorption of CO₂ with the structural Ni–N bond. The right-shift of the main peak during CO2RR means the lengthening of the Ni–N bond. It is ascribed to the redistribution of the electrons in Ni 3d orbital with four N coordinations (Ni–N bonds) and the adsorbed CO₂ (Ni–C bond), and hence distorting the central Ni ions out of the graphene plane.

In-situ XAS provides abundant informative evidence regarding the detailed mechanism of catalysis, including the identification of the active elements for the catalysis, the verification of the authentic catalysts, and observation of steady reaction intermediates in single-atom catalysts. XAS via hard X-ray averages the signals from bulk and surface parts of the materials, weakening the information through the catalysis, so promoting the signals on the catalytic surface by reducing the X-ray penetration depth can boost and widen the development and versatility of the in-situ XAS for catalysis.

In-situ X-ray diffraction (XRD)

X-ray diffraction (XRD) offers precise crystallography information, including the lattice parameters and the preferred orientation. Thus, the strain, the crystal facet, the doping effect, and the phase separation can be analyzed through XRD. XRD distinguishes the atomic arrangement with long-range order, meaning the amorphous phase could not be recognized, but the peak sets in XRD can be employed to reconstruct the whole crystal structure. With the development of surface XRD (using a fixed ultra-small incident angle), the crystalline states at the surface of catalysts can be identified. As mentioned above, the CoOOH is formed on the Co₃O₄ surface during OER reaction according to the in-situ XAS results and the Pourbaix diagram [62], [63], [73]. Magnussen’s group demonstrates an in-situ surface X-ray diffraction analysis of epitaxial Co₃O₄(111) and CoOOH(001) [74]. As shown in Fig. 2a(i) of the cyclic voltammogram (CV), Co₃O₄(111) film (red line) exhibits two redox pairs A1/C1 (Co₃O₄ to CoOOH) and A2/C2 (CoOOH to CoO₂), while the CoOOH(001) film (blue line) appears nearly no obvious redox pairs. The in-situ SXRD results present the structural variations in vertical and lateral grain size (Fig. 2a(ii,iii)) and strain (Fig. 2a(iv,v)) during CV sweep. The structure of the CoOOH film (blue symbols) remains nearly constant, but the Co₃O₄ film (red symbols) displays remarkable changes during the potential cycle. The shrinking vertical grain size indicates that the surface layers of the Co₃O₄ grains might be transformed into CoOOH [63]. The fact that the surface layer resumes back to epitaxial crystalline Co₃O₄ after the CV loop reveals the reversible phase transition during the redox process with hysteresis in forward and reverse potential scan. An anisotropic (Δε_⊥>Δε_||) volume shrinkage of the unit cell is observed in positive potentials for the reason that the robust epitaxy of the Co₃O₄ islands with the substrate laterally clamps to its surface lattice. In the range of large current density of oxygen evolution, no apparent structural changes are observed for CoOOH(001) but a further decrease in Δd_⊥,|| and Δε_⊥,|| for Co₃O₄(111). Therefore, in-situ surface XRD can clarify the lattice-related information, i.g. phase identification, strain, grain size, which also provides a crucial segment to understand the catalytic mechanism.

Fig. 2:

In-situ X-ray diffraction (XRD) patterns for various electrocatalysis. (a) For oxygen evolution reaction (OER) of Co₃O₄ and CoOOH: (i) CV and (ii−v) in-situ XRD structural data. Presented are the potential-dependent changes in out-of-plane and in-plane (ii,iii) grain size Δd_⊥,|| and (iv,v) strain Δε_⊥,|| relative to the values at the lower potential limit. The right-hand panel in (ii−v) show the variations of these structural parameters as a function of OER current. In all panels, filled and open symbols refer to the positive and negative going potential sweep, respectively. In addition, results of the structure under steady-state conditions are included for comparison (diamonds) and a schematic illustration of the film morphologies is included (upper right corner). Reprinted with permission from [74]. Copyright 2019 American Chemical Society. (b) For hydrogen evolution reaction of CoP: in-situ XRD patterns at dry state, before and during HER. OCV meant open-circuit voltage. The incident X-ray Energy is 16 keV. Reprinted with permission from [75]. Copyright 2019 American Chemical Society. (c) For oxygen reduction reaction of Co(OH)₂: in-situ XRD for the cathode obtained during DBFC (direct borohydride fuel cells) discharge at a current of 30 mA at about 25 °C. Thirty milliampere-initial is the spectrum collected at the beginning of the discharging process; 30 mA-20 min is the spectrum recorded when the DBFC has been discharging for 20 min; 30 mA-stop was acquired when the DBFC discharge was stopped; and the background is the spectra for DBFC which just has a membrane, fuel and anode. Reprinted with permission from [78]. Copyright 2013 Royal Society of Chemistry. (d) For carbon dioxide reduction reaction of copper-indium catalysts: in-situ XRD patterns of the (i) Cu/IO and (ii) CuInO₂ electrodes showing the evolution of the catalysts following each electrocatalytic cycle. The reflections originally present in the carbon GDL are labeled with “c”. The patterns are normalized to the height of the graphitic peak at 26.6° 2θ, which retained a similar intensity throughout the runs. The inset shows in detail the region around the (200) reflection of In(OH)₃, evidencing the gradual generation of the hydroxide following several CO₂ reduction electrolyses [79].

The electrocatalysts frequently undergo phase transition during the electrocatalysis due to the basic/acidic environment and high applied potential. That is to say, the material is easy to be reduced under the reduction reaction, and vise versa. In the case of cobalt phosphide for HER in alkaline KOH_(aq) (Fig. 2b) [75], the result of in-situ XRD with transmission mode reveals that CoP was swiftly transformed into Co(OH)₂ with high crystallinity. As X-ray penetrates through the entire material in the transmission mode, this transformation is characterized as an entire-material-conversion rather than a thin-surface-conversion. Co(OH)₂ remains until a highly negative applied potential, and subsequently, it reforms into the amorphous state, identified as the formation of a pure Co metal by in-situ X-ray absorption spectroscopy. These results disclose that the genuine electrocatalyst for HER is the cobalt metal rather than the as-prepared cobalt phosphide. On the other hand, polypyrrole (PPy) modified carbon-supported Co(OH)₂ (a low-cost non-noble catalyst) shows a comparable ORR activity in comparison with Pt/C in direct borohydride fuel cells (DBFC) [76]. Qin et al. conduct the ex-situ experiment and detect CoOOH transformed from Co(OH)₂ [77]. After that, they monitor the process of phase conversion by in-situ XRD during DBFC discharge at a current of 30 mA, and the results are shown in Fig. 2c [78]. Compared with the background, which is the XRD pattern of DBFC with a membrane, fuel, and anode, a faint peak appears in cathode at 15.841° in the initial stage. This peak represents the (040) of CoOOH. The peak intensifies after discharging for 20 min, proving their previous ex-situ results, while it fades when stopping the DBFC discharge. This result concludes that Co(OH)₂ transforms into CoOOH during the ORR process in DBFC discharging. These above examples regarding in-situ XRD confirm that the electrocatalysts frequently undergo phase transition during the electrocatalysis. As to composite materials in electrocatalysis, Peŕez-Ramírez’s group demonstrates two copper-indium composites: CuInO₂ and In₂O₃-supported Cu nanoparticles (Cu/IO) for CO2RR to CO [79]. They find that these catalysts exhibit the initially different crystalline structures and the element distribution but feature a similar electrocatalytic performance with continuing the reaction. In-situ XRD is conducted to investigate the material evolution, shown in Fig. 2d. As the reduction potential was applied, Cu–In intermetallic compounds (IMCs) are observed in both CuInO₂ and Cu/IO. CuInO₂ is found to be hardly reduced compared with Cu/IO, while Cu/IO reduce to Cu–In metallic mixtures and pure In. The pure In fades after CO2RR, corresponding to the generation of HCOO⁻ in the initial stage of the catalysis [80]. Cu₂In IMC primarily appears in CuInO₂ while CuIn IMC dominates in Cu/IO. In spite of the distinct IMCs appearing in two electrocatalysts, they both exhibit high selectivity for CO in the long-term CO2RR. Revealed by in-situ XRD, this superb selectivity was associated with the formation of core-shell structure (the Cu-rich IMC as the core and the In(OH)₃, gradually forming within the reduction, as the shell). These results demonstrate the correlation between the nanostructure of composite materials and their catalytic performance.

In-situ X-ray photoelectron spectroscopy (XPS)

In contrast to XAS and XRD, which mainly determined the atomic arrangement and phase identification for bulk materials, XPS is a surface-sensitive spectroscopy evaluating chemical states, electronic states, and charge-transfer behaviors of the specific elements within a material. Using synchrotron sources with smaller incident energy than Al or Mg monochromatic source allows the analysis of chemical states at the outermost few layers of the material surface. Photoelectrons excited from materials require an ultra-high vacuum to drift into the detector, resulting in the obstacle to developing in-situ XPS. Recently, differential pumping systems have been established to survive the photoelectrons and to realize the in-situ XPS investigation for electrocatalysis [81]. Typically, the measured area cannot be immersed in the electrolyte during electrocatalysis since the thick liquid layer stops the photoelectrons from the electrode; instead, ultrathin liquid layer (20–30 nm), formed by the capillary action through immersing and subsequently partial retracting of the electrode, can pass through the photoelectrons to proceed in-situ XPS. Friebel’s group investigates a Ni−Fe electrocatalyst via in-situ XPS for OER, and the normalized XPS spectra in the Ni and Fe 2p regions are shown in Fig. 3a [82]. The XPS spectra present metallic Ni and oxidized Ni^2+/3+ while metallic Fe and oxidized Fe^2+/3+ for the as-prepared catalyst. At 0.3 V vs. Ag/AgCl, the metallic Ni⁰ and Fe⁰ vanish, and the Fe ions are oxidized to completely Fe³⁺, suggesting that Ni–Fe electrocatalyst is fully oxidized at a positive potential. An increase to 0.55 V vs. Ag/AgCl does not bring about any remarkable differences in the chemical states of Ni and Fe ions. The thin layer of electrolyte limits the solution diffusion, causes the small current for OER current density, and restricts the metal ions oxidizing during OER. In addition to identifying the chemical states, in-situ XPS can analyze the variation of valence band during the reaction, which is related to electron conduction mechanisms. Nenning et al. demonstrate that in-situ XPS study of the perovskite SrTi_0.7Fe_0.3O_3-δ (STF) thin film under the electrochemical polarization as anodes in H₂/H₂O (Fig. 3b) [83]. Fe 2p spectra present the mixed Fe^2+/3+ ions near the surface at +0 mV (open-circuit voltage) due to the appearance of the satellite peak. Moreover, the valence band edge of STF is slightly below the Fermi level, suggesting the localized electronic defects and the semiconductor-like electronic structure. Anodic bias rapidly oxidizes Fe²⁺ ions to Fe³⁺ states as the disappearance of Fe²⁺ satellite peak and the observation of Fe³⁺ one while cathodic bias augments the quantity of Fe²⁺ ions and induces the emergence of metallic Fe⁰, followed by a sharp decline of the overall intensity. Metallic Fe⁰ also extends the orbitals towards Fermi level in the valence band spectra and facilitates the hydrogen evolution for water splitting. Thus, in-situ XPS provides sufficient information on chemical states and electronic states during the electrocatalysis, making the behavior of charge transfer and the variation of electronic configurations for the electrocatalysis more understandable and comprehensible.

Fig. 3:

Ambient-pressure X-ray photoelectron spectroscopy (AP-XPS) for various electrocatalysis. (a) For oxygen evolution reaction of Ni–Fe electrocatalyst: normalized AP-XPS of (i) Ni 2p and (ii) Fe 2p transitions, recorded with 4020 eV excitation energy. Bottom to top in parts i and ii: as-prepared catalyst at 9 Torr H₂O pressure, measurement in 0.1 M KOH at 0.3 V after electrochemical conditioning (6 potential cycles from 0 to 0.65 V), measurement at 0.55 V after additional electrochemical conditioning (6 potential cycles from 0 to 0.85 V). Reprinted with permission from [82]. Copyright 2016 American Chemical Society. (b) For bifunctional reaction (OER and HER) of SrTi0.7Fe0.3O3-δ (STF): Fe 2p XPS and valence band polarized by different overpotentials in H2/H2O atmosphere. Each spectrum is plotted with the same scale. ×, ‡, and + means the position of Fe⁰ species, Fe³⁺, and Fe²⁺ satellite features [83]. (c) For oxygen reduction reaction of iron-nitrogen-carbon (Fe–N–C): N 1s spectra for Fe–N–C and N–C electrocatalysts: (i and iv) in UHV; (ii and v) in O₂/H₂O at 60°C. Overlay of spectra acquired in UHV and oxygen/water for (iii) Fe–N–C and (vi) N–C electrocatalysts. Reprinted with permission from [84]. Copyright 2017 American Chemical Society. (d) For carbon dioxide reduction reaction of atomically dispersed nickel on nitrogenated graphene (A-Ni-NG): (i) change in the XPS O 1s intensity induced by CO₂ adsorption. (ii) Valence band spectra before (black line) and after (red line) CO2 gas exposure, and after desorption of CO₂ by thermal treatment at 500°C for 20 min in vacuum (dark blue line). The blue line shows the variation in the valence band spectra induced by CO₂ adsorption [55].

Regarding the identification of the active species for the electrocatalysis, in-situ XPS can also offer the auspicious indications. Take the iron-nitrogen-carbon electrocatalysts as an example [84]. Figure 3c displays N 1s spectra collected from an iron-containing MNC#1 and metal-free NC samples in UHV and O₂/H₂O. The spectra are fitted by 6 peaks: [85] including the peak at 398 eV (imine or cyano nitrogens), 398.8 eV (pyridinic nitrogen and disordered iron-coordinated nitrogen, e.g. Fe–N₂ and Fe–N₃) [86], 399.9 eV (iron-coordinated nitrogen, Fe–N₄, in MNC#1 and amine nitrogens in NC), 400.7 eV (pyrrolic and hydrogenated H–N nitrogens), 401.8 and 402.7 eV (graphitic nitrogens) [87]. Figure 3c(ii,v) show N 1s spectra collected in O₂/H₂O for iron-containing and iron-free samples, while Fig. 3c(iii,vi) superimpose the spectra in UHV and O₂/H₂O and highlight the spectra with significant changes in their relative abundance. For MNC#1, primary changes are observed at Fe-coordinated nitrogens, whose peak intensifies as much as 56% in O₂/H₂O, and at hydrogenated pyridinic as well as pyrrolic nitrogen peaks (peak intensity also increased). As oxygen is adsorbed on the Fe that is coordinated to nitrogens, π backdonation from iron to the adsorbed oxygen reduces the electron density on the nitrogens, leading to a shift of N 1s binding energy to higher values. On the contrary, the changes in NC are hugely divergent: hydrogenated pyridinic and pyrrolic nitrogen peaks diminish remarkably while amine and graphitic peaks soar. For both cases, the oxygen adsorption on hydrogenated pyridinic or pyrrolic nitrogens shift N 1s binding energy to lower values. In-situ XPS results unveil that oxygen is adsorbed to various N atoms in M-N-C electrocatalysts and also adsorbed to iron centers that are coordinated to nitrogens, facilitating the reaction kinetics and improving the catalytic activity. Another example to identify the charge transfer process of the active sites is A-Ni-NG with high activity and selectivity for CO2RR [55]. In Fig. 3d, CO₂ adsorbed on A-Ni-NG is recognized as the chemically adsorbed CO₂^δ− species (531.1 eV) and physically adsorbed CO₂ (533.5 eV) [88], revealed by the deconvolution of differential XPS of O 1s. Charge transfer from A-Ni-NG to CO₂ molecules owing to a rise in the work function of A-Ni-NG leads to the adsorption of CO₂^δ− on the A-Ni-NG surface. This change in the valence band considerably lessens the density of states in Ni 3d orbital and descends the valence band edge, suggesting the charge transfer from the 3d_x²_-y² orbital of central nickel to the C 2π_u orbital. These results identify the catalytic characterizations, including chemical states, electronic states, as well as charge-transfer behaviors of the specific elements, confirm the structure of adsorbed species, and would be conducive to the establishment of the catalytic mechanism.

In-situ soft X-ray absorption spectroscopy

In order to replace the noble electrocatalysts, numerous 3d transition-metal electrocatalysts have been developed and exhibit outstanding catalytic activities and selectivities. The reacting orbitals (3d orbitals for first-row transition metal cations and C, N, O orbitals for anions/adsorbed species) can be analyzed by in-situ soft X-ray absorption spectroscopy. The whole measurement is necessary to proceed in an ultrahigh vacuum environment due to the low energy of the incident soft X-ray. Recently, various in-situ cells, which compose of thin Si₃N₄ window(s), three electrodes, and an electrolyte pumping system, have been developed and thus copious valuable information during electrocatalysis can be analyzed [89], [90], [91]. Lange’s group reveals the local bonding states and symmetry characteristics of O atoms in electrodeposited Ni–Fe(O_xH_y) electrocatalysts for OER by in-situ O K-edge XAS, shown in Fig. 4a [92]. The spectra are classified in three regions: (1) the energy region at 525 eV–534 eV, which is the electronic transition from O(1s) to O(2p)/M(3d); [93], [94] (2) at 534 eV–540 eV, representing the near edge feature of water molecule; and (3) above 540 eV, showing the extended X-ray absorption fine structure of O K-edge. Four peaks at 529 eV, 529.9 eV, 531.2 eV and 532.5 eV are identified as the transitions from O(1s) to O(2p) hybridized with Ni(3d)t_2g, Fe(3d)t_2g, O(π*) of O₂ gas, and Fe(3d)e_g, respectively [95]. The peak at 529 eV (O(1s)→O(2p)/Ni(3d)t_2g) waxes at 1.48 V and wanes reversibly at reductive potentials, coinciding with the Ni²⁺/Ni³⁺ redox. The rise of this peak is also related to an augment in hybridization between O(2p) and Ni(3d), causing injection/extraction electrons from O to Ni sites (Ni³⁺-O²⁻→Ni^(3-δ)+-O^(2-δ)−) and improving the OER activity. On the contrary, the peaks at 529.9 eV (O(1s)→O(2p)/Fe(3d)t_2g) and 532.4 eV (O(1s)→O(2p)/Fe(3d)e_g) show minuscule changes with the applied potentials, suggesting that iron sites undergo a tiny variation in comparison to that at nickel sites. In the above case, in-situ O K-edge unveils that the orbital hybridization of O(2p) with metal ions and the charge transfer from O(2p) to metal ions during OER, which well-elaborates the evolution of reacting orbitals during electrocatalysis.

Fig. 4:

In-situ soft X-ray absorption spectroscopy for various electrocatalysis. (a) For oxygen evolution reaction of electrodeposited Ni–Fe electrocatalyst: (i) in-situ O K-edge spectra at various applied potentials. Potential values are given vs. RHE., where “rev” means the reverse direction from higher to lower potentials. (ii) O K-edge prefeature region at 1.18 V and (iii) at 1.78 V vs. RHE. Blue dots represent the experimental points, colored dashed lines the fit components of the multi-peak fitting and solid color lines the fit data model. The peak at 529 eV is the O1s-O(2p)/Ni(t2g) transition and appears at OER potential due to the increase in oxidation state of Ni. The measurements were carried out in 0.1 M KOH. (iv) Schematic illustration showing a partial donation of the higher electron density on the oxygen site (O^δ−) to the Ni metal site (O→Ni charge transfer) [92]. (b) For bifunctional reaction (OER and ORR) of the electrodeposited manganese Oxide: the Mn L_3,2-edges recorded in inverse partial fluorescence mode using the three-electrode XAS cell with an Au|CrO_x|Si₃N₄ window while flowing O₂-saturated 0.1 M KOH. (i) Waterfall plot of the spectra during cycling from 1.65 V via 0.50 to 1.65 V vs. RHE. (ii) Trends of the Mn valence during cycling [97]. (c) For hydrogen evolution reaction of atomically dispersed cobalt supported on phosphorized carbon nitride (Cu₁/PCN): Co L-edge XANES spectra of the ex situ catalyst and under the open-circuit condition. TEY, total electron yield model. The shaded region highlights the increase in peak intensity [98]. (d) For carbon dioxide reduction reaction of electro-redeposited (ERD) copper: in-situ Cu L₃-edge sXAS spectra of ERD Cu at different applied potentials (solid lines) and sXAS spectra of reference standards Cu metal (red dotted), Cu₂O (green dotted) and CuO (purpled dotted) [99].

Soft L-edge XAS for 3d transition metal is ideally proper to analyze the valence state and coordination symmetry since the 3d orbitals are verified directly. Owing to the smaller core hole broadening of the L-edge spectrum [96], the energy resolution of the L-edge spectrum is much better than that of K-edge one, leading that valence state and coordination in mixed-valence metal ions can be finer identified. Furthermore, the attenuation lengths of soft X-ray are considerably shorter than those of hard X-ray, so the information near the catalytic surface can be acquired by soft X-ray techniques. These advantages of soft L-edge XAS allow us to in-situ identify the valence changes of catalysts with mixed valence. Valence variations at OER- and ORR-relevant potentials of an electrodeposited manganese oxide film are demonstrated by Yang’s group [97]. In Fig. 4b, remarkable changes in Mn L_3,2-edge XAS spectra are observed at various applied potentials. At applied potentials above 1.50 V vs. RHE (steps 10a, 17a, and 18a), the spectra closely correlate with that of the δ-K_xMn^3.5+O_2-y·zH₂O reference, while at those below 0.80 V vs. RHE (steps 13a, 14a, and 15a), the spectra match well with that of the Mn^2.7+₃O₄ reference containing tetrahedral Mn²⁺. Their in-situ results suggest that along with Mn³⁺, the manganese oxide is composed of tetrahedral Mn²⁺ during ORR while Mn⁴⁺ during OER. Also, hysteresis in the Mn valence during cycling the potentials is observed (Fig. 4b(ii)), indicating that oxidative kinetics of the manganese oxide is more sluggish than the reductive kinetics. Thus, the dynamic change of Mn valance near the surface with applied potential offers insightful information regarding the active metal sites that catalyze ORR and OER.

The valence change and charge transfer due to adsorbed intermediates can also be investigated by in-situ soft XAS. Take the atomically dispersed cobalt immobilized in the framework of phosphorized carbon nitride (Co₁/PCN) for HER as an example [98]. As shown in Fig. 4c, the ex-situ catalyst is the Co₁/PCN coated on conductive silicon, while the open circuit condition means that the catalyst is wetted in KOH aqueous solution (HO-Co₁/PCN). The charge density difference of HO–Co₁/PCN towards Co₁/PCN discloses that the Co atom primarily contributes its electrons to surrounding N atoms and the OH⁻ group through orbital hybridization. The valences of 2.05 and 2.16 in central Co atom are acquired for Co₁/PCN and open-circuit HO–Co₁/PCN, respectively. This result verifies the more valence of the Co atom at the initial period of catalytic HER, suggesting that the π backdonation on HO–Co₁/PCN would promote the sluggish water dissociation step and subsequently improve the overall catalytic activity. In terms of CO2RR, which requires much higher applied potential than HER, the variation of valence in electrocatalysts should be more considerable and even cause the phase transitions. Sargent’s group develops an electro-redeposited (ERD) copper, namely the dissolution and redeposition of copper from a sol-gel, to improve the catalytic activity and selectivity of copper significantly [99]. The valence is probed by in-situ Cu L₃-edge with various applied potential (Fig. 4d). The spectra display a single peak at 931 eV, corresponding to the Cu(II) spectra [100], at applied potentials higher than 0.28 V vs. RHE while a notable peak at 933 eV, presenting the Cu(I) species, appears at an applied potential lower than 0.28 V vs. RHE. As the negative potential is applied, the metallic copper signals become dominated. Eventually, at the applied potential at −1.87V vs. RHE, the spectrum perfectly matches with that of standard metallic copper, suggesting a thorough transition from Cu(I) to Cu(0). The analysis of the acquired spectra through a linear combination of Cu(0) metal, Cu(I)₂O, and Cu(II)O standards exhibits that Cu(I) remains even at the negative potential of −1.47 V versus RHE, in which proceeds major CO2RR, and enhances the ethylene production by stabilizing ethylene intermediates. These cases highlight the crucial roles of valence identification and charge transfer information in establishing a proper catalytic mechanism.

In-situ high-energy-resolution fluorescence-detected X-ray absorption spectroscopy (HERFD-XAS)

In-situ probing and analyzing the information regarding the evolution of d orbitals in central metal sites during the reaction helps understand the overall catalytic mechanism. In-situ soft X-ray absorption spectroscopy exerts its advantages as above mentioned. Sometimes, the conditions that materials are required to be deposited onto specific silicon nitride window and the signal is acquired from the substrate side rather than the side of catalytic surface significantly hinder the extensive applications in diverse catalytic systems. In-situ high-energy resolution fluorescence detected X-ray Absorption Spectroscopy (HERFD-XAS, 1s→3d) allows us to investigate 3d orbitals in a typical catalytic environment. Hung et al. demonstrates in-situ HERFD-XAS to distinguish the active site in composite iron-doped cobalt oxide for OER [101]. In Fig. 5a,b, the HERFD spectra reveal two distinct regions: (1) low energy region, referring to a metal local quadrupole transition; (2) high energy region, presenting an oxygen-mediated metal-metal interaction. The high-energy shift of spectra in Co-Dom spinel (iron-doped cobalt oxide with cobalt-dominated lattice frame) suggests that the iron ions significantly influence the oxidation behavior and/or coordinated environment of cobalt ions, and stabilize the cobalt ions in higher valence, boosting the formation of intermediates and improving the OER performance. The in-situ cobalt K-edge HERFD-XAS (Fig. 5c,e) show an additional shoulder related to the oxygen-mediated metal−metal interaction after onset potential of OER, indicating an intense interplay between the electrolyte and cobalt ions during the catalytic process. As shown in Fig. 5d,f of difference spectra, the peak referred to the oxygen-mediated metal-metal interaction in Pristine spinel shifts to low energy in the period of OER while the peak in Co-Dom spinel keeps unchanged. These results imply the distinct catalytic behaviors of cobalt ions presented between these catalysts. Moreover, a broad peak is observed in the difference spectra of Co-Dom spinel, validating a more significant orbital interaction between the adsorbed O species and Co ions in comparison with that of Pristine spinel and subsequently bettering intrinsic catalytic performance of Co-Dom spinel. Regarding iron ions, the HERFD-XAS shows the imperceptible differences with the applied potential (not shown here), which means that iron ions do not involve in the catalytic cycles. Therefore, according to in-situ HERFD-XAS for iron-doped cobalt oxide, cobalt ions serve as the active sites for OER, and iron ions promote the catalytic interaction between cobalt ions and electrolyte. To our best of knowledge, in-situ HERFD-XAS are rarely employed to explore the catalytic characteristics for various catalysis, so adopting this technique to probe diverse reactions can be beneficial to figure out the catalytic mechanism in respect of the orbital interaction/hybridization.

Fig. 5:

In-situ high-energy-resolution fluorescence detected X-ray absorption spectroscopy (HERFD-XAS) for oxygen evolution reaction of pristine and iron-doped spinel Co₃O₄: (a) Schematic illustration of the cobalt 3d orbital interaction of catalytic surface with oxygen 2p orbital of electrolyte. (b) HERFD-XAS of Co K-edge for pristine spinel and Co-Dom spinel. Operando HERFD-XAS and difference spectrum of Co K-edge for (c, d) pristine spinel and (e, f) Co-Dom spinel. Reprinted with permission from [101] Copyright 2018 American Chemical Society.