Recent Advances in In Situ/Operando Surface/Interface Characterization Techniques for the Study of Artificial Photosynthesis

Abstract

(Photo-)electrocatalytic artificial photosynthesis driven by electrical and/or solar energy that converts water (H₂O) and carbon dioxide (CO₂) into hydrogen (H₂), carbohydrates and oxygen (O₂), has proven to be a promising and effective route for producing clean alternatives to fossil fuels, as well as for storing intermittent renewable energy, and thus to solve the energy crisis and climate change issues that we are facing today. Basic (photo-)electrocatalysis consists of three main processes: (1) light absorption, (2) the separation and transport of photogenerated charge carriers, and (3) the transfer of photogenerated charge carriers at the interfaces. With further research, scientists have found that these three steps are significantly affected by surface and interface properties (e.g., defect, dangling bonds, adsorption/desorption, surface recombination, electric double layer (EDL), surface dipole). Therefore, the catalytic performance, which to a great extent is determined by the physicochemical properties of surfaces and interfaces between catalyst and reactant, can be changed dramatically under working conditions. Common approaches for investigating these phenomena include X-ray photoelectron spectroscopy (XPS), X-ray absorption spectroscopy (XAS), scanning probe microscopy (SPM), wide angle X-ray diffraction (WAXRD), auger electron spectroscopy (AES), transmission electron microscope (TEM), etc. Generally, these techniques can only be applied under ex situ conditions and cannot fully recover the changes of catalysts in real chemical reactions. How to identify and track alterations of the catalysts, and thus provide further insight into the complex mechanisms behind them, has become a major research topic in this field. The application of in situ/operando characterization techniques enables real-time monitoring and analysis of dynamic changes. Therefore, researchers can obtain physical and/or chemical information during the reaction (e.g., morphology, chemical bonding, valence state, photocurrent distribution, surface potential variation, surface reconstruction), or even by the combination of these techniques as a suite (e.g., atomic force microscopy-based infrared spectroscopy (AFM-IR), or near-ambient-pressure STM/XPS combined system (NAP STM-XPS)) to correlate the various properties simultaneously, so as to further reveal the reaction mechanisms. In this review, we briefly describe the working principles of in situ/operando surface/interface characterization technologies (i.e., SPM and X-ray spectroscopy) and discuss the recent progress in monitoring relevant surface/interface changes during water splitting and CO₂ reduction reactions (CO₂RR). We hope that this review will provide our readers with some ideas and guidance about how these in situ/operando characterization techniques can help us investigate the changes in catalyst surfaces/interfaces, and further promote the development of (photo-)electrocatalytic surface and interface engineering.

1. Introduction

Recognition of the climate crisis, environmental degradation and other negative impacts [1,2,3] caused by excessive usage of fossil fuels has led to the enacting of various policies targeting renewable energy development [4,5,6]. The entry into force of the Paris Agreement officially kicked off a new era of energy transition, and also formulated targets for all the signatories [7]. To fulfill the agreement, Net Zero (or called Carbon Neutrality elsewhere) [8,9] should be listed as the first priority to keep these targets attainable between 2050–2060. Both energy transitions and circular economy strategies [10] are key for accelerating progress towards “Net Zero” of greenhouse gas emissions and increasing resilience to climate change. Among the many energy transition technologies, artificial photosynthesis has emerged as a promising approach in pursuit of accomplishing “Net Zero” [9]. Artificial photosynthesis mimics natural photosynthesis, using either water splitting to generate hydrogen and reduce carbon emissions, or the CO₂ reduction reaction (CO₂RR) to directly produce other high-value-added chemicals [11,12,13,14] and convert atmospheric carbon. After Fujishima and Honda reported water splitting on TiO₂ photoanode under UV light (Honda–Fujishima effect) [15] in 1972 for the first time, research involving (photo-)electrocatalytic artificial photosynthesis has attracted widespread attention [16,17,18,19,20,21].

Catalytic reactions are typically carried out on the surfaces of catalysts, and thus the microscopic reaction mechanism strongly depends on the physicochemical properties of surface/interface. The activity, selectivity and stability of catalysts are determined by several processes, including: (1) chemical adsorption of reactant on the surface or interface of catalyst; (2) activation of reactant to generate the intermediate, and (3) desorption of the final product from surfaces. In order to achieve a full understanding of the surface/interface engineering [22], enthusiastic efforts have been devoted to obtaining the physicochemical properties of the interface, and the structure of the double layer, including the study of substrate–adsorbate interactions that involve the sharing of electrons and orbital overlap (covalent bonds), as well as non-covalent electrostatic interactions that are weakly connecting the substrate and solvated ions and affect the potential drop (e.g., hydrogen bonding and van der Waals forces) [23].

Research on heterogeneous catalytic systems (such as gas/solid, gas/liquid and solid/liquid interfaces) [24] still faces a number of challenges. Conventional ex situ characterization techniques can only give a steady picture of the catalysts in a non-real-time state, instead of providing a real-time snapshot. For instance, as illustrated in Ref. [25], the authors loaded cobalt borate on a hematite surface and investigated the changes of cobalt borate through in situ XAS and ex situ XPS methods. While Co⁴⁺ was detected under a real-time reaction condition by in situ XAS, the ex situ XPS did not capture the presence of Co⁴⁺. Therefore, the ex situ characterization techniques are far from satisfying the need for measurement of the unsteady state of catalysts under working conditions, and do not provide a clear and accurate picture of the underlying reaction mechanisms (i.e., the reaction kinetics could not be studied).

A full understanding of reaction mechanisms can provide deep insights into the design of efficient and stable catalytic systems. In recent years, the burgeoning development of advanced in situ/operando characterization techniques has reshaped how the field of catalysis is studied. By employing in situ/operando techniques, people can facilitate the real-time analysis of interfacial carrier behavior [26], monitor the morphological and structural changes [27,28], probe the surface active sites [29,30,31,32,33], identify reaction transition states [34,35] and track the changes in adsorbed species [36,37]. This information can help us to understand the surface corrosion of materials, improve the catalytic efficiency of a system, establish the right reaction pathway, implement effective protection strategies, develop new catalysts, etc.

In this review, the applications of in situ/operando characterization techniques are summarized, with emphasis on the utilization of scanning tunneling microscopy (STM), scanning electrochemical microscopy (SECM), kelvin probe force microscopy (KPFM), X-ray photoelectron spectroscopy (XPS) and X-ray absorption spectroscopy (XAS) for studying (photo-)electrocatalysis, including surface potential, active sites, reaction intermediates, adsorption/desorption, phase and structural changes, etc. Furthermore, the aforementioned surface characterization techniques have also been extensively employed for the study of redox biomolecules coming from natural sources, which is another important area in the renewable energy field, demonstrating the wide applicability of in situ/operando in the investigations of various energy-related disciplines. The working principles of the listed techniques have been briefly described as well. We hope that this review will shed light on the necessity of in situ/operando characterization technologies for the development of the field of catalysis.

2. Scanning Probe Microscopy

Scanning probe microscopy (SPM) is a class of microscopic techniques that use probes to scan and image the surface topography of a specimen, including atomic force microscopy (AFM), scanning tunneling microscopy (STM) and scanning near-field optical microscopy (SNOM), etc. The surface imaged by SPM is attained by fully employing the physical interaction between the sharp tip and the topmost atoms from the sample, thus providing a nanoscale approach to investigate various systems. Furthermore, local chemical and physical properties other than morphology can also be interrogated by coupling SPM with other analytical methods and external fields, such as Kelvin probe, infrared spectroscopy (IR), X-ray photoelectron spectroscopy (XPS) and electrochemistry, to name a few. In this section, we will go through some applications of in situ/operando SPM techniques for the study of water splitting and CO₂RR.

2.1. Scanning Tunneling Microscopy

In 1981, Binnig and Rohrer invented scanning tunneling microscopy (STM) with atomic resolution, which demonstrated a much improved spatial resolution compared to scanning electron microscopy (SEM) [38]. The emergence of STM offered an opportunity to observe the arrangement of the surface atoms, as well as their related physical and chemical properties in real time, which greatly promoted the development of surface/interface science [39]. The working principle of STM is based on the quantum tunneling effect. In brief, when an ultrasharp tip (e.g., a tungsten tip with 7 nm tip radius) approaches to a conductive sample within a sufficiently close distance (in nanometer range), the tunneling current will occur between the sample and the tip under an external applied bias [40]. This current exponentially increases with the decreasing distance between the tip and the sample surface. In general, the tip–sample separation is kept around 4–7 Å, which is barely above the sample surface, for the tip to sense the interactions (repulsive and attractive forces) between the two. For the sake of acquiring other physical/chemical information simultaneously with the surface scanning during a chemical reaction, multiple fields can be added and thereby the corresponding phase transformation and physical response can be obtained in situ. For instance, combining electrochemical techniques and STM together, termed EC-STM, not only allows atomic-level imaging of liquid/solid interfaces to be achieved, but also provides us with unique insight into real-time electrochemical processes. In 1988, Itaya and Tomita constructed the STM for operation in electrochemical conditions for the first time [41], pioneering the observation of solid/liquid surfaces under in situ conditions. Up to now, EC-STM has become a powerful means to provide structural information about the solid side of the electrode/electrolyte interface at the atomic-level [42].

One of the most common applications of in situ STM is to directly perceive the morphological change of materials during reaction processes. Fester and co-workers investigated the morphological changes of Co-O nanoislands deposited on the surface of Au (111) after hydrogen evolution reaction (HER) and oxygen evolution reaction (OER) [43]. After 10 CV cycles at negative potential, well-defined hexagonal nanoislands with a size of 10–30 nm (Figure 1a) were found transforming into larger thin layers with uniform flat features (Figure 1b). The surface morphology after 10 OER cycles showed three distinctive regions (Figure 1c), of which region-(i) was consistent with O-Co-O/Au (111) after HER with respect to both height and morphology, and the typical smooth part of region-(ii) and Au single-atom deep pits (region-(iii)) were also observed.

Since Fujishima and Honda first reported water splitting on TiO₂ under UV illumination in 1972 [15], TiO₂ has received extensive attention and many water splitting models based on TiO₂ have been proposed [44,45,46]. In the process of establishing numerous models, STM is indispensable because it provides a way to directly monitor the water splitting process and its products at the atomic level. Tan et al. [47] observed the photodissociation of H₂O molecules adsorbed on the five-fold coordinated Ti (Ti_5c) sites of the R-TiO₂ (110)—1 × 1 surface, as shown in Figure 1d. The two H₂O molecules adsorbed at the Ti_5c sites (marked by white circles) under UV illumination can be dissociated into two types of OH species; one appeared at the adjacent oxygen bridging site (OH_br, marked by triangles), while the other occurred at the Ti_5c sites away from where the initial H₂O molecules stay, and even desorb from the surface.

The detection of active sites is of paramount importance in (photo-)electrocatalysis, which is not only a key step towards the improvement of the efficiency of catalysts, but also plays a pivotal role in determining the reaction mechanism of materials [48,49]. Mitterreiter et al. [50] investigated the active sites of MoS₂ on the Au (111) layer under HER (the boundary between MoS₂ and Au, as well as the steps on MoS₂, are marked by open and filled triangles, respectively, in Figure 1e). From the EC-STM images near the boundary (Figure 1f,g) at different potentials and corresponding line profiles (Figure 1h,i), it was found that the sharp edge between Au (111) and MoS₂ was clearly visible at −600 mV (corresponding to “HER off”). However, when the potential was gradually changed to −900 mV (corresponding to “HER on”), additional features appeared near the boundary, and the tunneling current around the edge became noisier, while the tunneling current of Au (111) and MoS₂ surface was basically unaffected. According to Pfisterer et al. [51], the place where the additional “noise” characteristic of EC-STM occurred under reaction condition can be understood as actual active reaction sites, so the authors considered that the HER sites of the above sample were located near the edges and steps. Liang and co-workers [52] also revealed the hydrogen evolution catalytic sites in monoatomic-thick Pt and Pd on Au (111) substrate (Au/Pt and Au/Pd) by analyzing the tunneling current noise. The 3–4 Pd atoms (about 1 nm wide) near the boundary of the Au/Pd system showed the highest noise level (Figure 2a), demonstrating that the most active hydrogen evolution sites were located near the boundary of Pd atoms. However, the whole Pt overlayer of the Au/Pt system showed the highest noise level (Figure 2b), indicating that the catalytic centers were likely to be uniformly distributed over the Pt overlayer. The above research demonstrates that there is a significant difference in the location of active electrocatalytic centers for Pd/Au and Pt/Au systems. STM has also been extensively employed in the investigation of CO₂RR. Gao et al. [53] prepared CeO_x nanoislands on Au (111) to study the interaction between Au-CeO_x interface and CO₂. The in situ STM images (Figure 2c,d,g) clearly showed that the CO₂ adsorption first occurred at the interfacial boundary near CeO_x. In the following studies, CO₂ tended to attach to the previous CO₂ species and gradually diffused towards the center of the island (Figure 2e,f,h), but did not adsorb on the Au (111) surface. This result showed that the Au (111)-CeO_x interface was not only the site of initial CO₂ adsorption and activation, but also acted on the subsequent activity of CO₂. This phenomenon also occurs in the field of water splitting; for instance, the reactants are first adsorbed and activated in the boundary region and gradually expand to the center of the island. In a recent work of Fester et al. [54], cobalt oxide nanoislands deposited on Au (111) surface (Figure 2i) were exposed to water vapor at room temperature. XPS analysis revealed that the lattice oxygen of the nanoisland decreased with exposure to water, while the number of hydroxyl groups continued to increase. The surface changes of single bilayers and double bilayers exposed to water vapor over time are shown in Figure 2j and Figure 2k, respectively. Point-like patterns appeared on the surface of both islands, which were the characteristic of the adsorption of H on lattice oxygen sites to form hydroxyl groups, but the hydroxyl groups on single bilayers appeared as weak protrusions, and the hydroxyl groups on double bilayers appeared as circular depressions. Combined with the atom-resolved STM movie, it was clear that the hydroxyl first originated from the edge of the island and then diffused to the basal plane, indicating that the edge of the island was the hydroxyl diffused from the edge of the island to the center. The applications of STM in the biological field are also not negligible. STM was successfully exploited to assess the electron transfer properties of cytochrome b protein molecules immobilized on gold surfaces [55]. This was the first study where the electron transfer was measured by controlling the protein orientation under pseudo-physiological conditions.

Figure 1. STM images of Co−O bilayer nanoislands on Au (111) in STM (a), the cobalt oxide/Au (111) surface after HER (b). The Co−O/Au (111) surface after OER (c). Italic numbers indicate cobalt oxide (i), gold islands (ii) and single atom deep pits in the Au (111) surface (iii). Adapted with permission from ref. [43]. Copyright 2018 John Wiley & Sons, Inc. STM images (6.3 × 6.6 nm², imaged at 1.0 V and 10 pA, 80 K) (d) showing the dissociation of water molecules under 400 nm UV irradiation for 1 h. Adapted with permission from ref. [47]. Copyright 2012 American Chemical Society. A boundary between the Au (111) substrate and a MoS₂ flake (e). Constant current mode electrochemical STM images visualizing a MoS₂/Au (111) boundary and a step edge at the surface of the MoS₂ film under HER conditions with increasing overpotentials. A set of images characterizing the (f) MoS₂/Au (111) boundary (open triangle) and (g) a step edge at the surface of the MoS₂ film (filled triangle) at different electrode potentials as indicated in the figure. Line profiles across (h), (i) the same position for each potential from (f,g), respectively. Adapted with permission from ref. [50]. Copyright 2019 Springer Nature.

Figure 1. STM images of Co−O bilayer nanoislands on Au (111) in STM (a), the cobalt oxide/Au (111) surface after HER (b). The Co−O/Au (111) surface after OER (c). Italic numbers indicate cobalt oxide (i), gold islands (ii) and single atom deep pits in the Au (111) surface (iii). Adapted with permission from ref. [43]. Copyright 2018 John Wiley & Sons, Inc. STM images (6.3 × 6.6 nm², imaged at 1.0 V and 10 pA, 80 K) (d) showing the dissociation of water molecules under 400 nm UV irradiation for 1 h. Adapted with permission from ref. [47]. Copyright 2012 American Chemical Society. A boundary between the Au (111) substrate and a MoS₂ flake (e). Constant current mode electrochemical STM images visualizing a MoS₂/Au (111) boundary and a step edge at the surface of the MoS₂ film under HER conditions with increasing overpotentials. A set of images characterizing the (f) MoS₂/Au (111) boundary (open triangle) and (g) a step edge at the surface of the MoS₂ film (filled triangle) at different electrode potentials as indicated in the figure. Line profiles across (h), (i) the same position for each potential from (f,g), respectively. Adapted with permission from ref. [50]. Copyright 2019 Springer Nature.

Due to the strict working conditions of STM, e.g., ultrasharp tip and extremely small separation between tip and sample that prefers a mirror-like flatness for the specimen, the quantum tunneling effect requires the specimen to be very conductive and uniform. These prerequisites limit the application of STM in a larger range of photoelectrode studies, where many of them are less conductive (studies of protection layers, metal–insulator–semiconductor structure), or flatness is not appreciated (nanostructures to increase surface aspect ratio, e.g., MOF, COF, inverse-opal). Moreover, the STM tips in EC-STM measurement are prepared using complex methods to deposit an electrophoretic paint coating to minimize the faradaic current at the tip/electrolyte interface. Thus, other techniques that can provide nanoscale information while having less stringent working requirements are needed to complement the study of surface/interface in photoelectrochemistry (PEC).

2.2. Atomic Force Microscopy

Ever since its invention by Binnig and Quate in 1986 [56], atomic force microscopy (AFM) has been extensively used in many different areas, such as biology, materials science, chemistry, physics, mechanical engineering, etc. In an AFM measurement, the sample’s morphological information (height fluctuation and shape, etc.) at the nanoscale or even atomic scale is obtained by the atomic force (attraction or repulsion) between the tip and the surface atoms. Such interaction exists in a range of 0.1–100 nm, regardless of the conductivity of the specimen. Therefore, AFM can not only overcome the material and flatness limitations of STM, but also can be operated under any environment (vacuum, air or liquid). The different operation modes of AFM such as contact and non-contact can be found in Reference [57]. Nowadays, there are many AFM-derived techniques, which make it not only possible to obtain information about the local topography of the sample, but they also provide surface potential [58], local conductance [59], active site distribution [60], electronic gaps [61] and other information simultaneously. Besides inorganic photocatalyst and electrocatalyst analysis, it has been successfully demonstrated that AFM also serves as a powerful tool for the study of biomolecule structure, redox mediators, surface charge density and EDL force [62]. For instance, AFM can be applied to elucidate the transition states of redox proteins like ferredoxin nicotin adenine dinucleaotide phosphate (NADP⁺) reductase towards its redox mediators, ferredoxin and flavodoxin [63] and also with the respective NADP⁺ cofactor molecule [64]. These were the first studies where redox flavoenzymes were studied at the single molecule level.

2.2.1. Kelvin Probe Force Microscopy

In 1898, the scanning Kelvin probe (SKP) technique, which was used to obtain the difference between the electronic work functions of two different metals, was first carried out by Kelvin [65]. For the first time in 1991, Nonnemacher et al. [66] introduced Kelvin probe technology into AFM, called Kelvin probe force microscopy (KPFM), which simultaneously allowed the imaging of surface topography and the measurement of the contact potential difference between the conductive tip and the sample. In KPFM analysis, when the conductive probe is close to the substrate, the contact potential difference is generated due to the different surface work functions between the two. The resulting contact potential difference can be eliminated with a known applied bias between the probe and the sample. Since the work function of the tip is known, the sample work function can be calculated [67]. A detailed explanation of the two different modes of KPFM (amplitude-modulated KPFM and frequency-modulated KPFM) can be found in references [67,68,69]. Thus far, KPFM has been developed as a powerful means to detect the surface charge distribution and/or work function on metal/semiconductor surfaces [70,71,72], and in situ/operando experiments can even image photogenerated charge carriers on the surface and interface of catalysts [73]. Although splendid progress has been made in KPFM, it is unfortunate that most KPFM operations thus far have been carried out under vacuum or ambient conditions, and few KPFM experiments in aqueous solution have been reported. This is due to the fact that the DC or AC bias will cause changes in the dynamics of the EDL [74], the electrochemical reactions and electrokinetic effects of both polar water molecules and electrolyte ions, resulting in difficulties in acquiring accurate and reliable measurements in aqueous solution [75,76,77,78,79,80,81]. Thus, in this section, we mainly focus on in situ observations under vacuum or ambient conditions rather than operando observations.

How to promote the separation and interfacial transfer of photogenerated charge carriers is a critical issue for improving the overall solar energy conversion efficiency in photocatalysis, and one of the effective methods is to load cocatalysts on photoelectrodes. Recently, various hydrogen evolution and oxygen evolution cocatalysts have been widely studied for promoting the separation and transfer of photogenerated electrons and holes at the interface, reducing the overpotential required for interfacial redox reactions, and inhibiting the corrosion of photoelectrodes. Yang et al. [17] investigated the surface potential of different oxygen evolution cocatalysts (NiO, CoO and NiCoO₂) decorated on BVO, and found that the surface potential difference between the FTO substrate and NiCoO₂/BVO was the smallest (Figure 3e,f). This result indicated that the upward bending of the band at the NiCoO₂/BVO interface was the most severe among all the samples. This band structure could facilitate the separation of electron-hole pairs and guide the transport of charge carriers, resulting in improved charge separation efficiency, oxidation kinetics efficiency and photocurrent density of the NiCoO₂/BVO photoanode. Hu et al. [82] electrodeposited AgO_x, NiO_x and AgO_x/NiO_x on the surface of BVO (AgO_x/BVO, NiO_x/BVO and AgO_x/NiO_x/BVO), and studied the surface potential changes of the above three photoanodes under illumination (Figure 3a–d). Among these three samples, AgO_x/NiO_x/BVO had the highest surface photovoltage (SPV, which is defined as the difference between contact potential difference (CPD) under light and CPD under dark [82,83]), which indicated that photogenerated holes in BVO were rapidly transferred towards the interface and accumulated in the composite AgO_x/NiO_x electrocatalyst. The concentration of accumulated photogenerated holes was the highest among the three. Meanwhile, the maximum SPV of AgO_x/NiO_x/BVO indicated that an efficient Schottky junction had been successfully established at the electrode–electrolyte interface, as demonstrated in the subsequent quantitative study of the absorbed photon to current efficiency on AgO_x/NiO_x/BVO. By artificially creating band bending by combining the cocatalysts with the semiconductor, the separation of electron-hole pairs can be effectively promoted, as is the improvement of the photocatalytic performance. In addition, spatial separation of photogenerated charge carriers can also be achieved through diffusion (sometimes called Dember effect, this has often been ignored because the built-in electric field is the main factor that facilitates the separation of charge carriers). In a recent work by Chen’s team [84], charge separation induced by diffusion (exploiting the difference in the mobility of electrons and holes) was found after cubic Cu₂O was exposed to non-uniform illumination (asymmetric illumination), and the effect was stronger than that induced by the built-in electric field. By artificially creating asymmetric illumination, they found that the potential of the irradiated surface increased by 30 mV, while the potential of the unirradiated side decreased by 10 mV (Figure 3g,h). Subsequently, periodic illumination was used to continue the irradiation, and an opposite trend of the two facets was observed (Figure 3i). Combined with other auxiliary experiments, it was deduced that the SPV generated by charge separation caused by diffusion was 20 mV (holes move to the illuminated facet and electrons move to the shadow facet). This result quantitatively indicated that the efficiency of carrier separation caused by diffusion was even better than that of the built-in electric field.

KPFM technology can also be used to probe the location of carriers and the distribution of active sites, which is due to the change of surface potential caused by the migration of charge carriers to a certain position on the surface. Wang et al. deposited Au NPs on the surface of rutile TiO₂ (100) (Au/TiO₂, as shown in Figure 4a) and detected the position distribution of the plasmonic holes for water oxidation on Au/TiO₂ [85]. The surface potential of Au/TiO₂ in the dark state (Figure 4b) clearly showed that the CPD of the interface between Au NP and TiO₂ was significantly lower than that of TiO₂, revealing that the Au/TiO₂ interface has higher work function due to the formation of Schottky junction. In addition, a ring highlighted in purple can be found by differential spectroscopy (Figure 4d, which was obtained by Figure 4b,c), indicating that charge separation occurs at the Au/TiO₂ interface, and holes were mainly concentrated at the interface. Moreover, DFT calculations also confirmed that this structure was favorable for water oxidation. Luna et al. [86] decorated Au nanoparticles (Au NPs) on TiO₂ surface and investigated the effect of a single Au NP on photogenerated charge transfer and trapping under illumination. The CPD on top of a single Au NP in the dark was about 12 mV lower than that of the surrounding TiO₂ (Figure 4e), from which it was inferred that the work function of Au NP was higher than that of TiO_2. Further increase in the CPD difference between the single Au NP and TiO₂ under illumination (up to 25 mV, Figure 4f) showed the transfer of photogenerated electrons from TiO₂ to Au NPs. Following studies also quantitatively revealed that SPV difference increased logarithmically with the increase in light intensity in a certain range (Figure 4g). Wardhana et al. [87] deposited a Cr₂O₃ thin film on SrTiO₃ (100) and found the accumulation of photogenerated holes on the surface of Cr₂O₃ thin film by recording photoinduced surface potential variations. Next, the effect of film thickness was also investigated, and it was found that the CPD in light and dark state was basically the same when the thickness was increased to 90 nm (Figure 5a), indicating that the inelastic mean free path of the photogenerated holes from the interface to the surface of the Cr₂O₃ film was within 90 nm. This work provides an approach for qualitatively calculating the mean free path of photogenerated carriers.

Combining Kelvin probe technique and photoconductive AFM, Eichhorn et al. studied water and oxygen adsorption on the surface of polycrystalline BVO film, and their effect on the transport and transfer of photogenerated charge carriers [26]. By observing the change of surface potential through KP technology, it was found that the adsorbed oxygen acted as an electron trap state on the surface of BVO, while the adsorbed water only formed a dipole layer on the surface to reduce the surface potential. The current distribution images of BVO under oxygen-rich, dry nitrogen and humid nitrogen are shown in Figure 5b–e, respectively. The minimum current in the oxygen-rich state was because the adsorbed oxygen acted as an electron trap site and captured the photogenerated electrons for desorption, leading to the decrease in the surface conductivity. The reason for the current increase under dry nitrogen was that the oxygen on the surface of BVO was replaced by nitrogen, while the apparent increase in conductivity under wet nitrogen might be related to the replacement of the residual oxygen on the surface of BVO by water molecules, and the water dipole arrangement affected the band alignment between the surface and the probe, thereby changing the injected current. Sometimes the presence of mobile ions can also affect KPFM experiments, and another new SPM technique that can be applied in liquid environments, electrochemical force microscopy, was developed and implemented by Collins et al. [79], which can also be used to study ion dynamics and electrochemical processes (such as adsorption/desorption, electron transfer, etc.).

2.2.2. Scanning Electrochemical Microscopy

Scanning electrochemical microscopy (SECM) can be categorized as an SPM technology that utilizes an ultramicroelectrode (UME) as electrochemical scanning probe to image and characterize the local morphology together with the corresponding electrochemical properties of various materials [88,89]. In the SECM analysis, the UME scans across the substrate surface to image the morphology, while the electrochemical responses occurring in the region around the probe and substrate have been simultaneously recorded [90]. A detailed explanation of the various operating modes, such as the feedback mode, the generation/collection mode and surface interrogation mode, can be found in References [91,92,93,94]. After decades of development, SECM has become a mature and attractive technology, which has received tremendous attention in the field of artificial photosynthesis, and it has been widely used in the detection of active sites, the study of interface reaction kinetics, etc. [95,96]. Moreover, the incorporation of SECM into the AFM system to form AFM-SECM (sometimes called potential-sensing electrochemical atomic force microscopy (PS-EC-AFM) elsewhere), further improves the spatial resolution and detection limits of this technique.

Bismuth vanadate (BVO) has become one of the most promising photoanode materials due to its narrow bandgap, non-toxicity, good photostability, appropriate band edge position to the redox potential of water, etc. [97,98]. Unfortunately, the poor electronic conductivity and rapid recombination of photogenerated carriers severely diminish its photocatalytic performance [33,99]. In response to this shortcoming, various optimization strategies have emerged, among which cobalt phosphate exhibits the outstanding advantage of easy incorporation onto the BiVO₄ surface for surface modification [100]. Nellist et al. [101] prepared a thick BVO layer that completely covered the FTO substrate and supported a layer of cobalt (oxy)hydroxide phosphate (CoPi) on the surface. A bipotentiostat was in use to drive the PEC process, wherein the potential of the BVO (V_sub) was controlled by WE1, whilst the nanoelectrode probe (WE2) simultaneously scanned the electrode surface and measured the potential of CoPi (V_tip), as shown in Figure 6a. The potential of CoPi remained unchanged initially, then increased with an increase in the photocurrent when the potential of BVO reached to −0.75 V vs. ε_O2/OH⁻, indicating that CoPi collected photogenerated holes from the BVO. Another experiment by Nellist et al. corroborated these results [102]. They deposited CoPi on the surface of planar hematite semiconductor photoelectrodes, and as the potential of planar hematite (V_sub) varied from −0.8 V to −0.43 V vs. ε_O2/OH⁻, the tip potential (V_tip) was essentially unchanged (Figure 6c) because CoPi was not charged/reduced (and therefore not conductive). However, when V_sub = −0.33 V vs. ε_O2/OH⁻, the photogenerated holes were transferred to CoPi, which led to the oxidation of CoPi and thus turned it into a conductive state. They also investigated the influence of the thickness of CoPi coating and found that if the BVO layer was not dense (or defective), a thick CoPi would cause collected holes to be transferred to the FTO by forming a good electrical contact with the FTO (Figure 6b) [101]. This phenomenon reveals that the thickness of the co-catalyst is also an important factor affecting the PEC performance of the photoelectrode. Laskowski et al. deposited Ni nanoislands of different sizes on n-Si surface and found that both the Schottky barrier height in air and the photovoltage under open circuit potential were independent of the size of the Ni nanoislands. However, the photovoltage after photoelectric activation was actually size-dependent, due to the surface conversion of the deposited Ni to Ni(OH)₂/NiOOH after the photoelectrochemical cycle. The authors explained this phenomenon by the pinch-off effect, and argued that the difference in photocurrent caused by the different thicknesses of the previous catalyst coatings should be explained by the pinch-off effect in addition to resistive losses and parasitic light absorption [103].

SECM can also be used to identify electrochemical CO₂RR products in a real-time PEC process. Shaughnessy et al. [104] prepared In⁰-In₂O₃ composites by electrochemically reducing the oxide layer on the surface of In₂O₃ to In, and they reported the application of substrate generation-tip collection scanning electrochemical microscopy (SG-TC SECM) to analyze the selectivity of the In⁰-In₂O₃ electrode. A schematic of SG-TC SECM is shown in Figure 6d. The amount of charge collected on the Pt tip under a series of potentials was measured to obtain the accurate reaction potential of CO production (Figure 6e). Mayer et al. [14] prepared three Sn/SnO_x catalysts with different compositions (different relative contents of Sn, SnO and SnO₂) and morphologies (from smooth surfaces to spherical nanoparticles) by pre-electroreduction. The specific SEM images and XPS spectra are shown in Figure 7a,b. These catalysts were then subjected to cyclic voltammetry tests under specific conditions, and the resulting peak current density features of each spatially resolved CV were attributed to formate, adsorbed CO (CO_ads) and H₂ oxidation, and the obtained Pt UME tip current densities were plotted as Figure 7c–e. The figures illustrated the highest formate oxidation current density and CO oxidation current density for pre-electroreduced Sn/SnO_x at −1.25 V vs. Ag/AgCl, and lack of waves associated with H₂ oxidation, which indicated that the CO₂ to carbonaceous products (formate and CO) faradaic efficiency of electroreducted Sn/SnO_x approached 100%. Sreekanth et al. [105] studied the pH and potential dependence on CO₂RR behavior of Au substrate in 0.1 M KHCO₃ solution, and showed different cyclic voltammetric responses under low alkaline (Figure 8a) and neutral conditions (Figure 8b). The characteristics of formic acid oxidation were shown under low alkaline conditions at −0.85 V vs. Ag/AgCl (¹³C NMR spectroscopy confirmed formic acid as the main product). However, when the bicarbonate solution was saturated with CO₂, the pH changed from 8.3 to 6.8, resulting in the emergence of a sharp CO oxidation peak at 0.6 V vs. Ag/AgCl and a suppression of formic acid formation. The above results demonstrated that the products obtained by CO₂RR can be controlled based on specific pH values and applied potentials. SECM based on specific modified UMEs can also be used to detect the local pH measurements. Monteiro et al. [106] first found a CO oxidation peak (peak I) at a substrate potential of −0.5 V, and a shoulder peak (peak II) emerged when the applied bias negatively shifted to −0.6 V. As the sample potential continued to move negatively, peak II gradually increased while peak I decreased. This phenomenon suggested that there were two different mechanisms of CO oxidation within a certain potential range. Since the applied potential caused a change in pH, the authors measured the pH of the local diffusion layer by a modified UME (Figure 8c), and they proposed three mechanisms of CO oxidation controlled by pH (

Table 1. Relationship between pH, j_lim species and bulk CO.

Electrooxidation Voltammetry [106]
pH	O-Donor	jlim Species	Voltammetry
≤7	H2O	CO	peak I
11 > pH > 7	H2O, OH−	OH−	peak I + peak II
≥11	OH−	CO	peak II

) in combination with rotating disk electrode experiments. When pH ≤ 7, CO was oxidized by water, corresponding to peak I. When 11 > pH > 7, both peak I and peak II were presented, with peak II corresponding to CO oxidation by OH⁻. When pH ≥ 11, CO was only oxidized by OH⁻.

Electrochemical AFM (EC-AFM) is another type of AFM-based technique that focuses on the monitoring of morphology variation during the electrochemical reaction. Toma et al. [107] carried out an in situ EC-AFM measurement of BVO with an applied bias of 1.23 V_RHE in dark, and they monitored the corrosion process of BVO films over time (Figure 8d,e). Area 1 exhibited uniform etching of the three solid–liquid interfaces. Area 2 had only one solid/liquid interface (diagonally across the upper left corner of the region, directly exposed to the electrolyte). Area 3 did not have a solid/liquid interface but there was a solid/solid grain boundary, exposing the top surface of the central grain, so this area was more vulnerable to chemical attack, with corrosion occurring almost entirely on the top surface. These results indicated that the corrosion of BVO was inevitable. Therefore, how to tackle the degradation of photoelectrodes has always been brought to the fore in PEC. Applying a protection layer onto an otherwise susceptible photocatalyst has become one of the most effective methods to solve this problem [108]. Recently, a kind of photoelectrode called “metal-insulator-semiconductor (MIS)” structure not only provides protection to the semiconductor, but also improves the solar-to-hydrogen efficiency with careful designs of band alignment and band bending [109,110,111]. Esposito et al. created a novel Si-based MIS photoelectrode [112]. They used a double-layer collector and replaced the native SiO₂ layer with a rapid thermal oxidized SiO₂. The former can increase the photovoltage while the latter can reduce the recombination of electron-hole pairs. It is noteworthy that the conventional MIS photocathode model failed to explain such high photocurrents found on this system and observed by scanning photocurrent microscopy and SECM. Therefore, the authors proposed an electrolyte-induced inversion channel layer and hydrogen spillover effect to elucidate this peculiar phenomenon that significant photocurrent and UME tip current can be observed at almost all locations on the surface of the photoelectrodes. AFM-SECM can also be used for bioelectrocatalytic research. Electron transfer of glucose oxidase (GO_x) protein molecules can be quantified by AFM-SECM [113]. The resolution of this technique is in the order of femtoamperes (fA), which enables it to visualize the current of individual molecules through ferrocene chemical mediators to gold tips.

3. X-ray Characterization Techniques

X-ray is a kind of electromagnetic wave with short wavelength and high energy. First discovered by W. C. Röntgen in 1895 [114], X-ray now has been widely used in medical treatment, industrial flaw detection, communication engineering, material characterization, etc. Based on the interaction between X-ray and the irradiated sample, we can obtain a lot of useful information, including material thickness, elemental composition, chemical valence, material structure, etc. In addition, X-ray crystallography can be employed to decipher the tridimensional structure of redox biomolecules [115]. Thanks to the recent advancement of synchrotron radiation light sources, the generated X-ray possesses high luminous flux, high brightness, high purity, remarkable collimation, excellent monochromaticity and tunable energy, which allows us to obtain more accurate and specific information about the material. All these improvements of the synchrotron X-ray source pave the way for in situ/operando study of photoelectrocatalytic materials [116,117,118]. In this section, two X-ray characterization techniques (X-ray photoelectron spectroscopy (XPS) and X-ray absorption spectroscopy (XAS)) will be introduced, and their application in the investigation of the underlying mechanisms of PEC water splitting and CO₂RR is discussed.

3.1. X-ray Photoelectron Spectroscopy

X-ray photoelectron spectroscopy (XPS) is a surface-sensitive X-ray technology with a probing depth less than 7–10 nm. The working principle of XPS is based on the renowned photoelectric effect. The incident X-ray can excite the core level electrons as well as the valence electrons from the topmost atoms. Once these electrons receive sufficient energy, they are ejected from the surface (photoejected free electrons) and then detected by the analyzer. These photoejected electrons carry bonding and valence state information, presented as the kinetic energy they possess [119,120]. However, the conventional XPS method cannot detect information alterations of the surface and interface under actual conditions (including light, humidity, applied potential, etc.). Because of the limited inelastic mean free path of photoelectrons generated by conventional X-ray sources, an ultra-high vacuum (UHV) environment is needed to minimize the electron scattering [120,121,122]. Ever since Siegbahn et al. developed the first ambient pressure (AP) XPS with a laboratory-based X-ray source in 1973, XPS has been extensively employed for liquid and/or gas phases surface/interface chemistry studies [123]. Recently, owing to the “dip and pull” method and optimized X-ray energy, we are able to probe solid/liquid and solid/gas interfaces under realistic (photo-)electrochemical reaction conditions [124,125,126].

We can utilize APXPS to determine whether water undergoes a simple adsorption process or a complex dissociation process on the catalyst surface. Through the different chemical behaviors of water vapor on pure Pt (111) and Pt/GaN (0001) systems, Zhang et al. [37] found that the selection of appropriate semiconductor supports can drive the dissociation of water. In the H₂O/Pt (111) system, only the gas-phase water and the physically adsorbed water signals were observed; neither hydrolyzed dissociated products nor oxides were detected. However, when thin Pt films were deposited on GaN (0001), it was found that the dissociation of water was observed on Pt. Starr et al. [127] studied the chemical transformation on the surface of BVO (010) after the introduction of water vapor, and found that not only were water molecules adsorbed immediately at the surface, but also extensive hydroxylation occurred, and a large amount of reduced V (from V⁵⁺ to V⁴⁺) was also observed. Subsequently, the authors further studied the composition changes of BVO photoanode at 1.0 M KOH before and after illumination. The energy of incident photons was increased to probe the BVO thin film/0.1 M KPi interface, and the light-induced surface changes at the open circuit potential were also studied. O 1s spectra and Bi 4f spectra are shown in Figure 9a and Figure 9b, respectively. After illumination, it was found that (1) the BVO O 1s signal disappeared and the Bi 4f signal decreased, which was caused by an increase in the thickness of the aqueous electrolyte layer; (2) the ratio of the sum of adsorbed or dissolved phosphate ion peak and adsorbed hydroxyl group peak, to that of the overall liquid water peak, increased, which might originate from the migration of phosphate ions from the BVO thin film/0.1 M KPi interface to the 0.1 M KPi/vapor interface; and (3) Bi 4f spectra shifted to a higher binding energy due to the formation of BiPO₄. Once back in the dark condition, the ratio of the peak intensity in (2) recovered to that of the original unilluminated state, and neither the peak associated with BiPO₄ nor the BVO O 1s could be observed. Favaro et al. [128] conducted a similar experiment in polycrystalline BVO photoanode/0.1 M aqueous potassium phosphate (KPi) system. The intensity of H_xPO₄^y− was also significantly increased relative to that of liquid-phase water (LPW) after illumination, and decreased upon subsequent reversion to the dark state (Figure 9c). Meanwhile, the ions in the electrolyte have also been studied, and are presented in Figure 9d,e for the K 2p and P 2p core levels, respectively. It can be seen that the intensity of K⁺ and P in H_xPO₄^y− increases significantly under illumination, and the intensity variation caused by illumination was reversible after returning to dark, which coincided with the results of Starr et al.

In recent years, the HER mechanism of many materials has been extensively explored [129,130,131,132]. However, OER has severely limited the overall development of water splitting due to its slow kinetic characteristics [133,134] Therefore, the study of efficient and stable OER catalysts has become a hot topic in the field of water splitting. Furaro et al. studied the surface chemistry and structure of polycrystalline Pt under 1.0 M KOH electrolyte as a function of applied potential [135]. The increase in the area of Pt-O and OH_ads was observed in the O 1s spectra (Figure 10a). Meanwhile, the area of H₂O_chem increased only under OER, which was in line with the increase in electrode surface area. The relative positions and thicknesses of various substances as a function of the applied potential (Figure 10c) were further determined qualitatively in combination with Pt 4f spectra (Figure 10b). Interestingly, Pt^(II) (including Pt^(II)(OH)₂ and Pt^(II)O) and Pt^δ-OH_ads appeared at relatively low potentials, and their quantities increased rapidly during OER, while Pt^(IV)O₂ was also detected. Stoerzinger conducted a similar study on polycrystalline Pt [136]. The O 1s and Pt 4f spectra are shown in Figure 10d and Figure 10e, respectively. The authors found that OER under alkaline conditions led to a partially irreversible oxidation of Pt, which was manifested in the increase in the amount of Pt^δ-OH_ads and the formation of Pt^(IV)O₂. However, Pt^(II)O and Pt^(IV)O₂ disappeared during HER, while the oxygen-free species Pt-H appeared. After driving HER for ~1 h, the chemical composition of the surface was found mostly unchanged under OCP.

Despite the superior performance of Pt catalysts, its scarcity and thus high and volatile price severely limits its large-scale application, so other non-noble metal catalysts are also being developed. Recently, transition metals and their derivatives have attracted increasing attention [133,134]. Han et al. [137] observed the surface species evolution of Co OER catalysts at different potentials in 0.1 M KOH electrolyte. Combined with Co 2p spectra (Figure 11a) and O 1s spectra (Figure 11b), it was found that when the potential increased to −0.4 V vs. Ag/AgCl, the formation of Co(OH)₂ was detected, accompanied by the attenuation of the Co signal. When further increased to +0.4 V, the outmost surface of the Co(OH)₂ layer was converted to CoOOH, as confirmed by APXPS with lower incident X-ray energy. The results indicated that, after the certain portion of Co that was in touch with the electrolyte was converted to Co(OH)₂, it was consecutively oxidized to CoOOH under OER condition. Favaro et al. [138] studied the interfacial chemistry and structural changes of disordered Co(OH)₂ biphase electrocatalysts covered on the Co₃O₄ surface at different potentials, as shown in Figure 11c. It was found that at the potential of converting Co³⁺ to Co⁴⁺ and above (including OER), all Co(OH)₂ was converted to CoOOH, and a portion of Co₃O₄ also underwent a similar transformation (converted to CoO_xOH_y). It seems that in order to reduce the OER overpotential, the formation of CoOOH and CoO_xOH_y phase may need to be promoted kinetically. When the potential switches back from OER to OCP, part of the top layer CoO(OH) is restored to Co(OH)₂, but there is still residual CoO(OH) remaining underneath the surface.

Many transition-metal-based catalysts can also be applied to CO₂RR, but their surface orientations may cause differences in CO₂ activation and product. Cai et al. [139] investigated the interfacial reactions on Ni (111) and Ni (100) surfaces under pure CO₂ and found that the amount of products with different surface orientations was significantly different. The C 1s spectra and O 1s spectra of Ni (111) and Ni (100) surfaces under 0.2 Torr CO₂ are shown in Figure 11d. The Ni (100) surface was covered with adsorbed CO*, graphitic carbon and a small amount of carbonate, while the Ni (111) surface was topped with a plethora of carbonate species with a reduced amount of adsorbed CO* and graphitic carbon, which was also verified by DFT calculation. Sometimes the electrolyte also plays a key role in a catalytic system, as it may affect the performance of the photoelectrode under reaction condition. Liu et al. [140] used operando APXPS to track the variation of Cu₂O in KHCO₃ and Na₂SO₄ electrolytes. Although Cu²⁺-OH species were presented in both different electrolytes under illumination, the OH⁻ feature remained at the interface of Cu₂O-KHCO₃ in dark (Figure 11f), while no OH⁻ species could be found at the interface of Cu₂O-Na₂SO₄ (Figure 11e).

3.2. X-ray Absorption Spectroscopy

When interacting with materials, the X-ray will be scattered and absorbed at first, then followed by an attenuation of the intensity of the X-ray during the transmission through the material. The X-ray absorption spectroscopy (XAS) is the technique that studies the relationship between the intensity of transmitted X-ray and the intensity of incident X-ray. According to the energy level of the emitted photoelectrons, X-ray absorption spectroscopy (XAS) is mainly composed of two regions, namely X-ray absorption near edge spectra (XANES) and extended X-ray absorption fine structure (EXAFS), which are formed by the interaction between photoelectrons and neighboring atoms. The former is located in a spectral region of about tens of eV around the absorption edge, which is a powerful technique to observe the alterations in the valence and electronic structure (including density of unoccupied states, and chemical bond hybridization) of absorbed atoms [141,142,143] The latter lies in the photon energy range of 50–1000 eV above the absorption edge, which can unravel the local geometric environment of the absorption atom, such as coordination number, coordination atoms, bond length and structural disorder [118,144,145]. Therefore, in situ/operando XAS can identify active components and track the structural evolution of catalysts and reaction intermediates under real-time electrochemical conditions [143,146,147].

Many reversible chemisorptions and electron transfer processes in interfacial reactions lead to the variations in the valence state and structure of catalysts. These processes are difficult (or even impossible) to detect by ex situ characterization techniques. A suitable approach to address these issues is in situ/operando XAS. Xi et al. reported that a cobalt borate (CoBi)-modified hematite photoanode achieves 1.12 mA cm⁻² current density at 1.23 V_RHE. An enormous enhancement of OER kinetics was attained when compared to the current density of pristine hematite (0.88 mA cm⁻²) [25]. Only Co²⁺ and Co³⁺ species were observed in the XPS spectra of the Co 2p_3/2 scans of the fresh and used CoBi-modified hematite photoanodes, but the presence of Co⁴⁺ was revealed by in situ XAS results determining the average oxidation state. Specifically, the XAS results under different potentials and illumination showed the mean oxidation state of Co ions in the CoBi-modified hematite photoanode to be 3.38, which was in stark contrast to the XPS results the authors observed. The possible reason is that Co⁴⁺ species are unstable and exist only under in situ conditions. That work strengthens the importance of in situ spectroscopy when studying catalytic process. Merte et al. [148] described the surface adsorption and oxidation of Pt NPs loaded on glassy carbon at different potentials by XAS. Their work demonstrated the chemical states of Pt NPs at different potentials (Figure 12a), which can be classified into hydrogen chemisorption between 0 V_RHE and 0.3 V_RHE, oxygen chemisorption (including OH and/or O) from 0.3 V_RHE to 0.96 V_RHE, and platinum oxide formation from 0.96 V_RHE to 1.26 V_RHE. Particularly, when the potential was kept at 1.16 V_RHE, the oxidation of Pt was found to proceed most fiercely. Möller et al. [149] tracked the chemical state and local structure changes of unsupported CuO nanocubes (U-NCs) during electrochemical CO₂RR. XANES (Figure 12b) showed that, although the potential was kept constant, the absorption edge was shifted towards the lower energy side, and evidenced a decrease in the valence state of Cu. The local environment of Cu probed by EXAFS revealed an enhancement of the Cu-Cu peak accompanied by an attenuation of the Cu₂O-related peak, demonstrating an increase in the metallic Cu-Cu coordination number and a decrease in the Cu-O coordination number. Linear fitting (Figure 12c) further found that the concentration of Cu⁰ species obtained from XANES was significantly higher than the coordination number of Cu⁰ species extracted from EXAFS, showcasing the existence of a large number of uncoordinated Cu sites. The above studies indicate that the initial cubic Cu₂O was electrochemically reduced to the defect-rich Cu. A similar conclusion was reached by Velez et al. [150]. In their study, the bulk- and surface-sensitive versions had been achieved by total fluorescence yield mode (TFY), and by auger electron yield (AEY) and total electron yield (TEY), respectively, as shown in Figure 12d,e. Although the initial composition of ingredients was different, both surface and bulk were reduced to copper during CO₂RR (in situ ECSEM also bore out this finding). Kornienko et al. [151] developed an amorphous cobalt sulfide (CoS_x) HER catalyst and investigated the composition, local atomic structure and catalytic activity of CoS_x under practical electrochemical conditions. Operando XANES (Figure 12f) showed the disappearance of the Co k-edge white line peak under HER conditions, and the spectrum resembled that of a transition metal sulfide. These results implied the disappearance of oxides and formation of cobalt sulfides. On the other hand, the increase in the intensity of the pre-edge peak can be explained by p-d hybridization of newly added cobalt sulfide. The Fourier transform EXAFS (Figure 12g) also confirmed an increase in the amount of cobalt sulfide, as a slight increase in average bond length was observed after HER.

Pt single-atom catalysts that consist only of isolated single atoms anchored onto FeO_x nanocrystallites were first reported by Qiao and co-workers with high activity and stability for both CO oxidation and preferential oxidation of Co in H₂ [152]. Since then, single-atom catalysts have aroused enthusiasm among researchers. Single-atom catalysts have many advantages, such as ultra-high atomic utilization (nearly 100%), large specific surface area, uniform distribution, etc. [153,154], and thus can tremendously improve the PEC performance of the photoelectrode. Fang’s group [155] synthesized atomically dispersed Pt on the N-C framework, and further tracked the valence state and atomic structure variations of the Pt sites under HER by operando XAS. Interestingly, the interaction and charge transfer between the Pt sites and the N-C carrier were weakened, tending to transition to a free metallic atom state. Wei et al. [156] synthesized a single-atom cobalt catalyst loaded on phosphorized carbon nitride, which exhibited outstanding HER performance. XANES and EXAFS showed that Co sites underwent a rise in oxidation state and a rearrangement in their local atomic structure from the ex situ state to the OCP and in situ states. Subsequently, the fitting and analysis spectra confirmed that HO-Co₁-N₂ acted as the real active site for HER, and the corresponding reaction mechanism was further proposed (Figure 13). The existence of the Co₁-N₄ moiety under ex situ condition was also speculated by Li et al. [157] through EXAFS.

4. Conclusions

In situ/operando analysis is crucial for the acquisition of a full understanding of catalytic reaction processes such as water splitting and CO₂RR, in order to gain more insight into the dynamic changes of the catalyst under working conditions. Given that the property of the catalyst surface and interface is a vital factor affecting the catalytic performance, enthusiastic efforts have been devoted to researching surface and interface engineering in (photo-)electrocatalysis.

As shown in the aforementioned studies, in situ/operando characterization techniques have unique functions. The application of in situ/operando SPM can study active sites, surface potential, surface adsorption and dissociation, morphology and structure. In situ/operando X-ray spectroscopy can provide information on elemental composition, valence changes, atomic short-range structure and other information. Coupling advanced in situ/operando measurements and catalytic systems can definitively allow monitoring of interfacial carrier behavior, probing surface active sites, identifying reaction transition states and tracking changes in adsorbed species, providing valuable and accurate details of reaction dynamics in real time. The integration of fragmented information obtained with different in situ/operando characterization techniques will piece together the microscopic reaction mechanisms that occur at the surface/interface scale.

Numerous reviews have been published involving the application of in situ/operando characterization in the field of artificial photosynthesis, but most of them focus on a particular characterization method, or a particular type of catalytic reaction. In this review, we have summarized the recent advances in the in situ/operando surface/interface characterization techniques used in the study of (photo-)electrocatalytic reaction mechanisms on the basis of previous studies, covering STM, AFM (including KPFM, SECM, etc.), XPS and XAS, and we thus hope to provide readers with a more comprehensive reference and some inspiration when choosing appropriate surface/interface techniques for studying reaction mechanisms. The study of dynamic changes is important, not only as the basis of mechanism research, but also because it allows formulating design principles for catalytic systems.