Thinking over the kinds of constraints in each technique, it was sometimes not convincing to prove the existence of OVs with just one technique. In the early research, various electron microscopes and radiation instruments, such as UV-vis and XRD, were used to characterize OVs in perovskite. In recent years, more advanced techniques that included HRTEM and EELs were used to characterize and gain insight into OVs in perovskite catalyst.

X-ray photoelectron spectroscopy (XPS). It was one of the most important characterization techniques for analyzing OVs, which was characterized by the difference in the atomic number ratio of metal ions to lattice oxygen. XPS and others were employed together to analyze the structures of perovskites and concentrations of OVs. TiO₂ was obtained with distinctive concentrations of OVs by synthesizing a series of TiO₂ samples with different Cu-doping concentrations by Liang et al. [91]. Although pure TiO₂ could not absorb the visible light, UV-vis diffuse reflectance spectroscopy could easily detect the presence of OVs in TiO₂ samples. As the copper-doping concentration climbed, the intensity of absorption in the visible area increased, which led to the remarkable rising of the concentration of OVs (Figure 3a). More information about OVs in Cu-doped TiO₂ samples was displayed in the O 1s XPS (Figure 3b–e). By curve fitting, O 1s area precisely showed three peaks, which corresponded to lattice oxygen, surface hydroxyl oxygen, and adsorbed oxygen. They were located at 529.5, 530.5, and 531.5 eV.

Magray et al. [92] used a solid–state method and a laser deposition method to prepare La₂CoMnO₆ bulk (LCMO-B) and La₂CoMnO₆ films (LCMO-F), respectively. For the purpose of detecting OVs in the above two perovskites, rapid responding XPS measurements were performed. Figure 3f,g illustrates the Mn 2p XPS spectrum of LCMO-B and LCMO-F, respectively. Firstly, the background was used for the best fit, and then the normalized spectrum was used to figure out the charge states of Mn ions. Mn⁴⁺, Mn³⁺, and Mn²⁺ peaks were fitted to the Mn 2p 3/2 peak. Manganese in the sample was present in the Mn²⁺ state because of the satellite peak at 647.9 eV. This confirmed the presence of OVs in these samples. However, the peak intensity of Mn³⁺ and Mn²⁺ in LCMO-F was higher than that in LCMO-B, indicating that LCMO-F had more OVs. XPS detection technology could reflect the presence of OVs from the side, rather than directly detect the concentration.

Raman spectroscopy was widely applied in studying various defects, such as OVs caused by ion doping and so on, because the characterization method was high-efficiency and sensitive to the structure and bond order of perovskites [93]. For example, the chemical coordination structures of Pt/Nb₂O₅-AR₁₀ (PNO-A) and Pt/Nb₂O₅-air (PNO-a) were studied by Raman spectroscopy. The peaks after several times magnification are shown Figure 4a,b. It could be clearly seen that in PNO-A, the peak concentrated at 685 cm⁻¹, while the peak of PNO-a shifted to 676 cm⁻¹, which could generate more OVs under Ar reaction.

The influence of OVs on NO selective reduction by CeO₂ was also investigated. Zhang et al. [94] found that the OV of CeO₂ roasted in the Ar atmosphere was higher than that of the roasting in ambient air. Figure 4c illustrates the Raman spectra of the two catalysts. Usually, the strain and the defect in the lattice caused the Raman band position and shape of the F2g mode at 465 cm⁻¹. The F2g oscillation frequency of the CeO₂-AR catalyst was changed to 455 cm⁻¹. The sample was calcined at 600 °C and then cooled to room temperature. Therefore, no residual strain was found in the lattice of CeO₂. Therefore, the inferred Raman band position shift might indicate that there were OVs in CeO₂-AR.

XAFS was developed on the basis of synchrotron radiation. The excited photoelectrons were dispersed by the surrounding atoms as X-rays passed through a sample, which resulted in an energy oscillation. The local electronic and geometric formation of the detected specimen could be obtained by detecting these oscillatory indicators [95].

The location of OVs of Sr₂CuO_3+δ was developed in a high-temperature superconductor by polarized extended X-ray absorption fine structure (EXAFS) by Wang et al. [96] The characterization analysis was that OVs were distributed in the material plane and apical position, about 10% and 8%, respectively. By rotating Sr₂CuO_3+δ around the b axis, polarized EXAFS at different incident angles was collected (Figure 4d), and the DFT was used in the calculation.

Xie et al. [97] used EXAFS to collect the information of SrTiO₃ (STO) doped with NiP. They concluded that the ratio of OVs in NiP/STO-60 was the highest among all samples; then, the order of oxygen octahedron structure was affected. The STO doped with NiP could gently adjust the relative concentrations of OVs, lattice oxygen, and adsorbed oxygen and significantly improve the catalytic hydrogen evolution activity due to the oxygen balance.

EPR was a kind of magnetic resonance technology that could be used to detect the unpaired electrons contained in atoms or molecules of matter qualitatively and quantitatively. Then, the structural properties of the surrounding environment could be explored. Jing et al. [98], respectively, synthesized LaFeO₃ (LFO) and 3D micron globular LaCo_0.5Fe_0.5O₃ (LCFO) by the hydrothermal method. Both of these perovskites were used to activate PMS to degrade BPA, and the electron spin resonance (ESR) spectra of the two were compared to detect OVs (Figure 5a). LCFO exhibited stronger signal intensities due to electron capture sites on OVs. The peak intensity was proportional to the concentration of OVs, which also indicated that there were more OVs in the LCFO material. OVs in LCFO could be increased by ion co-doping in a LFO structure, thereby enhancing the activation ability of PMS [99].

The OVs in the various prepared MnO₂-x catalysts were characterized by EPR [100]. Electron delocalization on OVs might result in more symmetric signal peaks. As can be seen from Figure 5b,c, each catalyst had a symmetric EPR signal peak, indicating that there were certain OVs in each material. The changing trend of the signal peak intensity of different catalysts was as follows: MnO₂-1.8 (I_peak = 795) > MnO₂-2.0 (I_peak = 758) > MnO₂-1.65 (I_peak = 578) > MnO₂-1.5 (I_peak = 333) > MnO₂-1.8R (I_peak = 226). The content of OVs was directly reflected by the signal peak strength. When x = 1.8 in the prepared MnO₂-x sample, the signal strength of the catalyst was the strongest, suggesting that the MnO₂-1.8 catalyst had the highest OVs’ concentration. Conversely, when x = 1.8R, the signal strength of the catalyst was the lowest, which indicated that it possessed the least amount of OVs [101].

Yang et al. [102] utilized EPR to probe into the hydrothermal α-Fe₂O₃ film and explored the kinetic and thermodynamic properties of its photoanodes in the presence or absence of OVs, understanding the effect of OVs on the water oxidation of the film. The consequence was that the film had an obvious EPR signal. Along with the time of hydrothermal treatment, the signal intensity of the material strengthened, indicating the increase in OVs.

The Kelvin probe force microscope (KPFM) was one of the most widely used instruments, which was susceptive to detect the concentrations of local OVs. The electronic characterization of semiconductor surfaces also could be performed by it. The preparation of BiFeO₃ (BFO) and hydrogenated BFO (HB-180-2-8) on the ITO surface was measured simultaneously with KPFM, respectively, and the morphology and CPD images are illustrated in Figure 6a,b. The variation of CPD cross-section variation values (ΔVCPD) was ~40 mV and ~140 mV, respectively, which were the original BFO and the hydrogenated BFO (HB-180-2-8); both them were clearly marked in CPD images with the marks of one, two, and three. The surface potential of unhydrogenated BFO nanocrystals was less than negative than that of hydrogenated BFO nanocrystals. The large number of OVs on the particle surface was due to the surface potential with a more negative shift of hydrogenated BFO, and these OVs could be used as active sites for electron capture [103]. The BFO particle electronic structure surface was actually altered by the use of a hydrogenation method in the image of KPFM. In an effort to better study the distribution and kinetics of OVs in NdBaCo₂O_5.5, the oxygen ions and their occupied positions were analyzed by means of neutron powder diffraction (NPD) and annular dark-field scanning transmission electron microscopy (ADF-STEM). A schematic diagram of the crystal structure of NdBaCo₂O_5.5 (NBCO_5.5) in TEM image is shown in Figure 6c. The image could be used to distinguish OVs and positions of oxygen atoms by the way of improving spatial resolution. Figure 6d displays the ordered arrangement of OVs’ channels in the ND-O layer due to alternating the CoO₅ square cone and CoO₆ octahedron along the B-axis. ADF-STEM image simulation was carried out on the measured grid parameters and the optimized DFT model in order to demonstrate the experimental results. The image showed that OVs’ channels were empty, and the atomic columns of Nd, Ba and Co were murky, which was consistent with the experimental image primely [104].

It was confirmed that the addition of Mn₂ ions into a ZrO₂ lattice could lead to a large amount of OVs, which was probably one of the causes of the reduction in lattice parameters.

Temperature programmed desorption (TPD) was a technique used to measure the surface properties of an active center. Temperature programmed reduction (TPR) could measure the reducibility of a substance. La_0.9K₀.₁CoO₃/ZrO₂ (LKC/Z), LaFeO₃/ZrO₂ (LF/Z), and pure LaFeO₃ (Pure-LF) were prepared by Guo et al. [106] The TPR profiles of all samples are shown in Figure 7a. The reduction in Fe⁴⁺ to Fe³⁺ occurred in two places: one was a small H₂ consumption peak of LF/Z material at around 370 °C, and the other was a relatively obvious H₂ consumption peak, which appeared in the sample of x = 0.2. Another significant H₂ consumption peak appeared in the x = 0.1 sample due to the reduction in Co³⁺ to Co²⁺ at low temperature. The reduction in the above metal ions led to the formation of OVs on the surface of perovskite. The O₂-TPD spectra (Figure 7b) of CuO and La₂CuO_4−δ (LCO) showed LCO had more surface chemisorbed oxygen because several of the main low-peak strengths of CuO were less than LCO at 800 °C [107].

The gravimetric method was the way to determine the component content of the measured substance by weighing the mass of the substance. For the purpose of quantitatively studying the OVs’ concentration, the OVs-rich α-Fe₂O₃ was obtained by the gravimetric method. Three specimens were prepared on the basis that the molar ratios of α-Fe₂O₃ and tartaric acid were, respectively, 1/5, 1/7, and 1/10, expressed as S1, S2, and S3. The OVs-rich sample, in which mv was 1000 mg, was transformed into an OVs-free sample (m_o) after oxygen annealing at 400 °C for 2 h.
where Δm_x was the evaporation of α-Fe₂O₃ at 400 °C, and the weight of absorbed water was calibrated by repeated blank tests.

In the absence of OVs, the mean weight loss of α-Fe₂O₃ (Δm_x) was −3.4. For OVs-rich α-Fe₂O₃ (Δm), S2 and S3 were, respectively, +1.4 mg and −2.0 mg. As a result, S2 and S3 had an OVs mass (Δm − Δm_x) of +4.8 mg and +1.4 mg. Then, in accordance with Equation (2), S3 and S2 were calculated to be 1.6% and 0.4%, respectively. The results for S1 and S4 were not clear, mainly due to the fact that they were too low to weigh (<0.1%), which exceeded the capacity of the device [90]. Other methods for the quantitative detection of OVs according to different concentrations were also studied [36].