Understanding the light-induced oxygen vacancy in the photochemical conversion

According to the previous reports, the light-induced V_O is mostly found on different kinds of metal-oxide-based semiconductors. Depending on the circumstance of its generation, it could be divided into three main approaches. Firstly, to avoid reoxidation, the vacuum system should be a natural choice for producing light-induced V_O. At the early stage, light-induced V_O is observed on the surface of TiO₂ in the high vacuum chamber of scanning tunneling microscope (STM) for investigating the super hydrophilicity of TiO₂ under UV light (see figures 2(a) and (b)) [5]. Although it is difficult to apply vacuum in the photo-chemical process, some researchers still used vacuum to pre-treated metal-oxide-based semiconductors to generate light-induced V_O on the surface for subsequent photo-chemical catalysis. For example, Zu et al fabricated the defective Bi₂O₂CO₃ nanosheets, in which the light-induced V_O needed to be regenerated under UV light irradiation in the vacuum in every circle [12].

Zoom In Zoom Out Reset image size

Secondly, inert atmosphere without reducing agent could be employed to achieve the similar effect as using vacuum. Inspired by heat-induced V_O, Zhang et al extensively studied the light-induced V_O in the inert atmosphere such as Ar gas [10, 13, 17–19, 22–24, 28] and have published their research outputs for CO₂ reduction and H₂O splitting by using pure TiO₂ and various decorated TiO₂. Since formation energy could be reduced and absorbance of solar light could be expanded from UV to visible light, doping and loading co-catalysts (e.g. Au, Pd, Cu) on the TiO₂ have been proven to be the efficient methods for improving generation of light-induced V_O on the material surface used in their work. As shown in figure 2(c), Xu et al introduced a LSPR effect by loading Pd NPs on the TiO₂ [18]. Since light absorption was enhanced in vis-NIR spectrum, more available charge carriers were generated to induce more V_Os on TiO₂. Besides TiO₂, various metal-oxide-based semiconductors were also found to be good substrates for generating light-induced V_O. For instance, amorphous zinc germanate (α-Zn–Ge–O) semiconductor with weak lattice constraint was used by Wang et al to promote the formation of light-induced V_O [8]; they also conducted generation and consumption of V_O simultaneously to reduce CO₂ to C in the CO₂ atmosphere (see figure 2(d)).

Thirdly, the light-induced V_O could be produced in the solution under some additional conditions such as biased voltages and reducing agents. These additional conditions can create reductive circumstance to promote O exsolution on the surface of metal-oxide-based semiconductors. Normally, the V_O could only be maintained when additional conditions were applied. Recently, Sun et al fabricated light-induced V_O on the photoanodes (e.g. BiVO₄ and WO₃) in a neutral solution by applying a reduced potential under AM 1.5 sunlight (see figure 2(e)). They demonstrated that light-induced V_O could remarkably enhance the photoelectrochemical performance [37]. Furthermore, during some photocatalytic processes including photocatalytic degradation, dehalogenation and water treatment, it is easy for metal-oxide-based semiconductors (e.g. In₂O₃, BiOCl crystals) to produce light-induced V_O with organic reactants such as rhodamine B under UV light irradiation [36, 40, 41, 54]. Additionally, to promote fabrication of light-induced V_O, the Raman reporter such as mercaptobenzoic acid is usually utilized for enhancing signal on the surface in the photo-induced enhanced Raman spectroscopy [51, 53].

Generally, the generation of light-induced V_O proceeds with the following steps [8, 17, 27]: in reaction (1) as seen below, the light-induced EHPs are produced as a result of illumination on the M_x O_y surface. In reaction (2), the photo-generated excitons are reduced on the M_x O_y surface, causing the conversion of ${{\text{M}}^{\left( {\frac{{2y}}{x} + } \right)}}$ sites to ${{\text{M}}^{\left( {\frac{{2y - 2}}{x} + } \right)}}$ sites. In reactions (3) and (4), the excited holes are trapped in the lattice O ions to produce free oxygen radicals (O^·) and generate V_O on the nonstoichiometric surface. In reaction (5), the two adjacent O^· ions combine to form an oxygen molecule in the vacuum and inert atmosphere. If reducing agents or protons are added, the O^· ions will combine with reducing agents or protons. reaction (6) shows the overall reaction of light-induced O exsolution [from reactions (3) to (5)]. It should be noted that the A(B) means item A is adsorbed on the item B in the bracket as in the following reactions.

${{\text{M}}_x}{{\text{O}}_{y,{\text{semiconductor}}}} + h\nu \to {h^ + } + {e^ - }$	Light-induced EHPs	(1)
$x{{\text{M}}^{\left( {\frac{{2y}}{x} + } \right)}}\left( {{{\text{M}}_x}{{\text{O}}_y}} \right) + 2{e^ - } \to x{{\text{M}}^{\left( {\frac{{2y - 2}}{x} + } \right)}}\left( {{{\text{M}}_x}{{\text{O}}_{y - 1}}} \right)$	Metal ion reduction	(2)
${{\text{O}}^{2 - }}({{\text{M}}_x}{{\text{O}}_y}) + {h^ + } \to {{\text{O}}^{ - }}({{\text{M}}_x}{{\text{O}}_{y - 1}})$	O exsolution	(3)
${\text{O}}^ - ({{\text{M}}_x}{{\text{O}}_y}) + {h^ + } \to {{\text{O}}^.} + {{\text{V}}_{\text{O}}}({{\text{M}}_x}{{\text{O}}_{y - 1}})$	O exsolution	(4)
${{\text{O}}^{\text{.}}} + {{\text{O}}^{\text{.}}} \to {{\text{O}}_2}$	O exsolution	(5)
$2{{\text{O}}^{2 - }}({{\text{M}}_x}{{\text{O}}_y}) + 4{h^ + } \to 2{{\text{V}}_{\text{O}}}({{\text{M}}_x}{{\text{O}}_{y - 1}}) + {{\text{O}}_2}$	Total reaction of O exsolution	(6)

Obviously, the key points are the excitation and separation of EHPs, decreasing formation energy of V_O and maintaining V_O under the required conditions, which leads to the requirements for both photoabsorbers and circumstances during the generation of light-induced V_O. In general, all of these three conditions, vacuum, inert atmosphere and solution, are available to create V_O. The light-induced V_O generated in vacuum and inert atmosphere is usually consumed in the cycle system since it could not be well maintained when vacuum and inert atmosphere are replaced by reactants. These circumstances are created to avoid re-oxidation which normally reduces the number of V_O. Because additional conditions are usually applied to keep reduction, the generated V_O could be easily created and maintained in solution and utilized in the continuous system. Researchers should create suitable circumstances in different applications to avoid re-oxidation or keep reduction. While most of works could use UV light to generate V_O [27, 55–57], the photoabsorbers need to be further modified for using visible light. Expanding the absorbance of light could provide more available EHPs to drive V_O generation. Meanwhile, recombination of EHPs has been found to be the most energy-consumption step in the light conversion [4, 56, 58, 59]. To depress the recombination of EHPs, co-catalysts that could create electron or hole traps and prolong lifetime of carriers will play an important role [4, 60, 61]. But the co-catalysts could also be a recombination center [17, 62], which should be further studied and need delicate adjustment in the fabrication of photoabsorbers. Based on density functional theory (DFT) calculations, Xu et al reported that both interstitial doping and replaced doping are demonstrated to be effective to decrease the formation energy of V_O (see figure 2(f)), which could greatly increase the number of V_O [19]. However, the easier the generation of V_O is, the harder the V_O is maintained and used. Thus, the balance between the difficulty of V_O formation and utilization needs to be fine-tuned in the experiments. This issue is important but is often neglected by researchers. It should also be noted that the valence of metal could play an important role in the formation of V_O. On one hand, because the capacity of losing and obtaining electrons could be much different between the metal ions with single oxidized valence and the metal ions with multiple oxidized valences (e.g. Zn and Ti ions), it leads to different abilities of oxygen evolution in different metal oxides. On the other hand, the valence of metal could also be important even in the same metal oxide. Generating V_O is always harder in the metal oxide with low-valence metal ions than that with high-valence metal ions. For example, the formation of V_O on the CeO₂ could be much easier than on the Ce₂O₃. Besides, Qi et al also found that the enlarged surface area allows more exposed oxygen atoms on the surface, which is favorable for the generation of V_O under the irradiation of light [46]. Constructing efficient nanostructure via light engineering on the photoabsorber to achieve multiple reflection and absorption should be a promising way to increase light-induced V_O formation.

Currently, the productions of useful fuels and chemicals such as CO₂ reduction [18, 19, 23], H₂O splitting [13, 22, 24], nitrogen fixation [63–65], reverse water gas conversion [33, 35, 46, 47] and dry reforming of methane (DRM) [25, 26, 47, 56] are the most studied applications of the light-induced V_O for the photochemical conversion. This light-induced V_O formation is a unique process whereby the exsolved atoms are initial part of the support crystal lattice of semiconductors, this phenomenon forms the base for the two functionally useful ways used in the photochemical conversion. On one hand, the remaining V_O on the non-stoichiometric semiconductor could be utilized as a reactant as well as consumables under a specified condition and regenerated under irradiation. As a result, a completed cycle of V_O consumption and regeneration could be established. Taking CO₂ reduction and H₂O splitting as the examples, they could be used to produce CO and H₂ by capturing the oxygen in the CO₂ and H₂O. Then, the dissociation of CO₂ and H₂O could be completed by forming a photochemical looping via combining reaction 7 or 8 (see figure 1). Meanwhile, the metal ion ${{\text{M}}^{\left( {\frac{{2y - 2}}{x} + } \right)}}$ will be oxidized to ${{\text{M}}^{\left( {\frac{{2y}}{x} + } \right)}}$ in reaction (9).

${{\text{V}}_{\text{O}}}({{\text{M}}_x}{{\text{O}}_{y - 1}}) + {{\text{H}}_2}{\text{O}} + 2{e^ - } \to {{\text{O}}^{2 - }}({{\text{M}}_x}{{\text{O}}_y}) + {{\text{H}}_2}$	VO consumption	(7)
${{\text{V}}_{\text{O}}}({{\text{M}}_x}{{\text{O}}_{y - 1}}) + {\text{C}}{{\text{O}}_2} + 2{e^ - } \to {{\text{O}}^{2 - }}({{\text{M}}_x}{{\text{O}}_y}) + {\text{CO}}$	VO consumption	(8)
$x{{\text{M}}^{\left( {^{\frac{{2y - 2}}{x} + }} \right)}}({{\text{M}}_x}{{\text{O}}_{y - 1}}) + 2{h^ + } \to x{{\text{M}}^{\left( {^{\frac{{2y}}{x} + }} \right)}}({{\text{M}}_x}{{\text{O}}_y})$	Metal ion oxidation	(9)

Zhang et al and Yoon et al have replaced the heat-induced V_O with light-induced V_O to lower the reaction temperature and improve solar-to-fuel efficiency for solar CO₂ reduction and H₂O splitting by photo-thermochemical cycles [10, 13, 17–19, 22–28]. The mechanism of photo-thermochemical cycle over TiO₂, which achieved CO₂ reduction by generating and consuming light-induced V_O at a temperature of less than 500 °C, is shown in figure 3(a). As a comparison, the heat-induced V_O needs a temperature of higher than 1000 °C [56]. In the CO₂ photoreduction work by Zu et al (as shown in figure 3(b)), V_Os are generated under fast UV light irradiation on the Bi₂O₂CO₃ nanosheets in near vacuum and reproduced after every 24 h cycling test [12]. Due to light-induced V_O, the formation energy of COOH* intermediate can be decreased from 1.64 to 1.13 eV, which was the rate-limiting step in the photochemical CO₂ conversion. Finally, a high CO evolution rate of 275 mmol g⁻¹ h⁻¹ using Bi₂O₂CO₃ with V_O, which was 120 times higher than that without V_O, was achieved for the visible-light-driven CO₂ photoreduction during 110 cycling tests. The light-induced V_O could also be utilized in photoelectrochemical reactions. As shown in figure 3(c), Sun et al conducted a photocharge/discharge strategy to initiate WO₃ photoanode by generating light-induced V_O [37]. There was no significant decay in more than 25 cycles with 50 h durability with the photocharged WO₃ surrounded by a 8–10 nm overlayer and V_Os. It was attributed to the prolonged charge carrier lifetime caused by light-induced V_O.

Zoom In Zoom Out Reset image size

On the other hand, the light-induced V_O could act as the activated sites for photochemical conversion on the surface [6, 30, 31, 33, 35, 38, 40–42, 44–46, 48, 66, 67]. Different from the cyclic utilization, all the steps are completed under the same condition. Taking H₂ production for example, the mechanism could be shown as follows [6, 17, 56, 67]:

According to the electronegativity of ions, H⁺ and OH⁻ could be generated and adsorbed on the reduced metal ions (${{\text{M}}^{\left( {\frac{{2y}}{x} - 1} \right)}}$) and V_O in the water ionization, respectively (reaction 10). The V_O and metal ions could be the activated sites for further reactions of O²⁻ and H₂ (reactions 11 and 12). The O²⁻ could react with holes and V_O could be unoccupied in the O₂ production (reaction 13). Different from the application in the photochemical looping, the light-induced V_O will not be consumed, but act as an activation center in the photocatalysis. Zhang et al found that light-induced V_O could be sustainable on the BiOBr nanoflower [30]. And reaction rates were found to be a greatly increased which was attributed to the V_O-induced improvement for transfer efficiency of photoinduced carriers and the light absorption capacity in photocatalytic water splitting. In addition, Qi et al fabricated a large quantity of dual-functional light-induced V_O on the 2D In₂O₃ nanostructures, where local heat was more efficiently produced to achieve a high photothermal conversion and provide active sites for the photothermal CO₂ reduction reaction simultaneously (see figures 3(d) and (e)) [46]. A great CO production rate of 103.21 mmol g_cat ⁻¹ h⁻¹ could be obtained. The work indicates that the light (acted as an ideal external stimulus) could efficiently produce a large number of V_Os on the 2D nanostructures. Besides, some reactions which do not need light could be assisted by light-induced V_O under light. Pan et al integrated photocatalysis and thermocatalysis for DRM on MgO/Pt/Zn-CeO₂ [47]. As shown in figures 3(f) and (g), they maintained the light illumination to generate light-induced V_O on CeO₂ by photoinduced electrons, which could avoid deactivation to stabilize the thermochemical DRM process. It has also been found that CO disproportionation as the major side-reaction could be restrained with the light-induced V_O, therefore the rate of carbon deposition was decreased during the process of DRM. Interestingly, Lee et al generated light-induced V_O in 2D BaSnO₃ epitaxial films to reversibly control photocurrent responsivity and persistent photoconductivity (see figure 3(h)). It indicates that light-induced V_O could be applied to develop even 2D optoelectronic devices by optically controllable manipulation of surface defect states, which should be another interesting application for light-induced V_O [9]. Additionally, the stability of light-induced V_O needs to be considered in two cases according to its application. The first one is the application of V_O consumption for reduction reaction, which requires that oxygen vacancies could be steadily decreased and easily participate in the reduction reaction. In the second case, the environment where the oxygen vacancies perform catalytic reactions mostly exists in the presence of oxygen, so how to maintain V_O is challenging. Especially, light-induced V_Os are mostly formed on the surface, which are less stable than those in the bulk. Some modification methods have been employed, such as nitrogen doping and nanostructure design to stabilize valance state of metal ions to keep V_Os stable [68, 69]. Besides, starting from the reaction circumstances, utilizing some reaction conditions to maintain a reducing environment, such as reducing gas, sacrificial agent and biased voltages, can also maintain oxygen vacancies. Both ways of utilizing light-induced V_O have been demonstrated to be efficient in improving the photochemical conversion. The production and utilization of light-induced V_O are simple, precise, nondestructive and easy to be expanded. It is foreseeable that more and more photochemical conversions including solar-driven water splitting, CO₂ reduction, synthesis gas conversion, seawater desalinization and organic pollutant disposal could benefit from the activation and catalysis of light-induced V_O.

${{\text{H}}_2}{\text{O}} \to {{\text{H}}^ + }({{\text{M}}^{\left( {\frac{{2y - 2}}{x} + } \right)}}) + {\text{O}}{{\text{H}}^ - }({{\text{V}}_{\text{O}}})$	H2O dissociation	(10)
${\text{2O}}{{\text{H}}^ - }{\text{(}}{{\text{V}}_{\text{O}}}{\text{)}} \to {{\text{H}}_{\text{2}}}{\text{O}} + {{\text{O}}^{{\text{2}} - }}{\text{(}}{{\text{V}}_{\text{O}}}{\text{) }}$	OH− catalysis on the VO	(11)
$2{{\text{H}}^ + }({{\text{M}}^{\left( {^{\frac{{2y - 2}}{x} + }} \right)}}) + 2{e^ - } \to {{\text{H}}_2}$	H2 generation	(12)
$2{{\text{O}}^{2 - }}({{\text{V}}_{\text{O}}}) + 4{h^ + } \to {{\text{O}}_2}$	O2 generation	(13)

Revealing the relationship between the light-induced V_O formation and activities is very important for designing high-performance defective catalysts. Therefore, measurements and evaluations of light-induced V_O need to be carefully handled by using various characterizations. In that perspective, the measurements could be generally divided into electron microscopy techniques and spectroscopy techniques.

The atomic structure of imaging materials could be shown by electron microscopy technology directly. Due to the low resolution, some normal microscopy techniques, including scanning electron microscopy and transmission electron microscopy, are not able to accurately measure V_O [33, 41, 43, 54]. As mentioned before, STM could directly observe and demonstrate the V_O formation (see figures 1(a) and (b)) [5]. And scanning transmission electron microscopy (STEM) which owns a resolution of atomic level is a very useful tool for observing the V_O on the surface of materials [9, 15, 40, 54, 63, 65, 70]. Light-induced V_O was well confirmed by Qi et al in the magnified image obtained by an annular bright-field scanning transmission electron microscopy (ABF-STEM), which could show the atom arrangements of the 2D In₂O_{3 − x} layers (see figure 4(a)) [46]. Furthermore, as shown in figure 4(b), Lee et al found that it was better to use the low-angle annular dark-field (LAADF) signal of STEM, which was more sensitive to the strain field from point defects (i.e. V_O), than the high-angle annular dark field signal to measure light-induced V_O [9]. It shows that LAADF-STEM is a good direct way to observe light-induced V_O optically. Besides, Dagdeviren et al employed a time-resolved atomic force microscopy with fast-detection electronics and high-frequency cantilevers to demonstrate the effect of the light-induced surface V_O on charge carrier dynamics (see figure 4(c)) [21]. They measured the resonance frequency shift (Δf₀) of the oscillating cantilever to assess the time-dependent variation in the tip-sample interaction force, which could further obtain the effective activation energy related to the migration of holes, E_a*, and the migration barrier for a single hole motion, E_a. The E_a in dark and that under light could be measured and detected to uncover the change of hole migration dynamics by this method. It has been proven to be effective to study the deep mechanism in the formation of light-induced V_O.

Zoom In Zoom Out Reset image size

In regards to the x-ray spectroscopy techniques, resolution of x-ray diffraction is usually too low to detect light-induced V_O, but x-ray photoelectron spectroscopy (XPS) [12, 13, 22–24, 30–34, 36–42, 71–73] and x-ray absorption spectroscopy (XAS) [1, 2, 24, 27, 33, 46, 50, 54, 60, 62, 63, 65, 71, 74] are effective. XAS, including x-ray absorption near-edge structure spectroscopy and extended x-ray absorption fine structure, could assess oxidation state in bulk and average coordination environment of elements in materials. As shown in figure 4(d), Qi et al studied the In K-edge of In₂O_{3 − x} nanosheets and standard In₂O₃ and found the existence of lower average oxidation states of In species in In₂O_{3 − x} nanosheets [46]. According to results in the R space (see figure 4(e)), the light-induced V_O led to the decrease of the In–O distance, which caused a split between 1 and 2 Å of In₂O_{3 − x} nanosheets. XAS could show ample information of the defects in bulk, however, the light-induced V_O is normally generated on the surface only. Thus, use of XPS gains more popularity than XAS in the study of light-induced V_O, since the former only detects the surface defects. Especially, time-resolved in-situ XPS could be more powerful to clarify the formation and function of light-induced V_O. As shown in figure 4(f), the process of light-induced V_O formation could be easily observed in Lee et al's study [9]. To reveal the interaction between light-induced V_O and metal ions, Zu et al employed a synchrotron-radiation quasi in-situ XPS to study the Bi ions near the V_O in figures 4(g) and (h) [12]. Their XPS results indicated that the Bi³⁺ sites could easily receive the photoexcited electrons with the presence of light-induced V_O and then adsorb CO₂ molecules, i.e. not only CO₂ was efficiently activated, the partially reduced Bi sites were returned into the original Bi³⁺ sites as well.

As an effective method for investigating unpaired electrons in materials, electron paramagnetic resonance (EPR) is also often used for qualitative and quantitative characterizations of light-induced V_O [13, 24, 30, 32, 34, 35, 37–39, 41, 42, 71, 75]. As shown in figure 4(i), Wang et al detected an EPR signal of V_O at g = 2.003 after α-Zn–Ge–O was illuminated by a 300 W Xe lamp in vacuum, which demonstrated the process of light-induced V_O formation [8]. They also conducted a rough quantitative EPR analysis by reference to the spin numbers of lone electrons of natural coal. Their results indicated that the V_O concentration of α-Zn–Ge–O continuously increased from 0 to 0.21 μmol g⁻¹ as the irradiation time was increased from 0 to 48 h. However, the EPR could not distinguish the surface V_O and bulk V_O. Thus, it was still hard to know the number of light-induced V_Os on the surface, which were the real effective sites. Besides, Raman spectroscopy, photoluminescence and UV-visible diffuse reflectance spectroscopy (UV–vis DRS) could also detect the V_O indirectly [4, 9, 11, 22–24, 33, 36–38, 41, 43, 44, 46, 48–51, 54, 56]. Nonetheless, the accuracy of results highly depends on the V_O concentration, materials property and conditions of measurements, making these methods not very universal. For example, the changes in the results of UV–vis DRS when producing light-induced V_O on the TiO₂ in Ar gas were much smaller than those on the α-Zn–Ge–O in vacuum (figures 4(j) and (k)) [8, 13]. In addition, the production in the V_O formation (such as O₂) could also be analyzed by using gas chromatography and traced by mass spectrum with isotopic gas [13, 24, 27, 35, 60, 63, 67]. The materials, methods and corresponding measurements of the representative studies of photo-induced V_O are summarized in table 1, which shows that various materials could be utilized to generate light-induced V_O in different kinds of circumstances. While most measurements are similar, in-situ methods and advanced electron microscope could be rare and powerful in the study of light-induced V_O. Conclusively, although the measurements and evaluations of light-induced V_O are either very complex and expensive or lack of accuracy, the development of in-situ techniques combined with electron microscopy and spectroscopy techniques is expected to bring more valuable information and knowledge about light-induced V_O.

The light-induced V_O can be implemented under mild reaction conditions with low cost energy source (sunlight). Since the reaction conditions of photochemical conversion are very similar to that for light-induced V_O formation, it is convenient to combine materials fabrication (via in situ light-induced V_O formation) with the subsequent reactions of photochemical conversion. Thus, this method demonstrates a great potential of its much broader applications for the photochemical conversion. However, achieving an effective light-induced V_O formation could face three challenges, i.e. suitable semiconductor support, measurement of V_O, mechanism of the effect of light-induced V_O on the photochemical conversion.

Firstly, in order to enable light-induced V_O formation, the semiconductor support should meet two conditions: (a) the bandgap of semiconductor should be suitable to generate enough EHPs, (b) the formation energy of V_O should be low enough to be driven by EHPs. In the previous studies, many approaches have been taken to adjusting the bandgap and V_O formation energy with some promising results, including doping with transition metal ions and loading noble metals [13, 18, 22–24, 51, 56, 76, 77]. However, more efficient semiconductor supports should be developed and fabricated to produce more light-induced V_O. Besides, many studies have shown that the formation of V_O could also be the recombination of EHPs [9, 21, 30, 39, 56]. Although it is important and should be well adjusted in the fabrication, no detailed information is available and no discussion and research have been conducted yet.

Secondly, it is difficult to evaluate the performance of light-induced V_O without the accurate information of the numbers of V_O. However, on the one hand, due to the formation of light-induced V_O occurs on the surface of metal oxide and its concentration is quite low in terms of the whole material, making it difficult to detect and measure it accurately. On the other hand, since the light-induced V_O is usually not stable in the atmosphere and tends to be transient during the photochemical conversion, critical conditions should be applied to be able to detect and evaluate it [39, 54, 56]. Therefore, in-situ techniques may need to be combined with above measuring techniques to make a detailed and real-time evaluation of light-induced V_O during photochemical conversion. In-situ tests combined with STM, STEM, EPR and XPS could be employed to observe the light-induced V_O formation directly. However, the very strict requirements of these tests are hardly to be practical, and the complete quantitative V_O measurements also remain challenging.

Thirdly, although the mechanism of the effects of light-induced V_O has been extensively investigated and many possible reasons have been proposed, there is lack of a clear picture in regards to the contributions of each effect, e.g. the synergetic effect and the effect of interaction between V_O and reduced metal ions in the process of photochemical conversion. Additionally, in addition to the number of exsolved sites, the qualities of V_O and metal ions are also important since some low-quality V_O may not be the activated sites and would not participate in the reaction. It is worth noting that some recent calculation work has noticed this point. The DFT calculations of Li et al indicate that the balance between the formation and reactivity of light-induced V_O is the key factor for the pathway preference for water splitting, which should be taken into consideration for catalyst design and screening [77, 78]. Currently, there is still no reported experimental work focusing on this. The three issues mentioned above could greatly affect the final efficiency of reactions which would dictate the future development and potential applications of the light-induced V_O for the photochemical conversion.