Oxygen Vacancy-Mediated ZnO Nanoparticle Photocatalyst for Degradation of Methylene Blue

: ZnO nanoparticles (NPs) are synthesized by deoxidizing ZnO powder in a vacuum drying process. This process reduces the size of the NPs and increases the concentration of oxygen vacancies on their surfaces. ZnO NPs with sufﬁcient oxygen vacancies are highly effective for the photodecomposition of methylene blue (MB) dye in water under ultraviolet irradiation. The MB degradation efﬁciency exceeds 99 percent after 50 min of light irradiation, and the catalytic property of the NPs remains stable over several complete degradation cycles. It is revealed that the concentration of oxygen vacancies on the surface, and the photocatalytic activity, are both higher for smaller NPs. Oxygen vacancies reduce the recombination rate of photo-generated charge carriers by capturing the electrons and hence, improve the efﬁciency of redox reactions. In addition, a smaller particle size leads to a larger speciﬁc surface area and a higher photonic efﬁciency for the ZnO NPs. molecules are gradually adsorbed on the surface of a catalyst, before the photocatalytic reactions occur, and ultimately reach saturation. According to Equation (1), the percentages of MB absorbed on the surfaces of Sample A and Sample B during its adsorption–desorption process (30 min), were 2.3% and 4.2%, respectively. This clearly shows that a large number of oxygen vacancies on the surface of ZnO NPs contribute to the adsorption of dye molecules onto the photocatalytic surface. Figure 7a shows the decomposition of MB (4 mg/L), using different ZnO nano-Figure


Introduction
The relentless escalation in environmental pollution presents a critical and detrimental global problem. In particular, the release of large quantities of toxic dye in wastewater is responsible for the evident and severe destruction of our ecosystem [1]. Semiconductor nanomaterials promise to play a key role in wastewater purification, a prospect that has attracted major attention from the government, industries, and the public. In general, photocatalysis by semiconductors results from the interaction between photo-formed electron and hole pairs on semiconductor surfaces irradiated by UV light [2].
Electrons ( all play important roles in photocatalytic reactions [1]. Compared with the traditional wastewater treatment process, semiconductor photocatalysis is a promising way to treat toxic, nocuous, and non-biodegradable contamination in wastewater, at room

Synthesis of ZnO Photocatalyst
ZnO powder (analytical grade) was purchased from Tianjin Binhai Cody Chemical Reagent Co., Ltd. Tianjin, China. The preparation of the vacuum-treated ZnO photocatalyst proceeded as follows: 1.5 g of pristine ZnO powder (denoted as Sample A) was loaded into a quartz tube with an opening at one end, and then placed in a vacuum drying oven. It was treated under low vacuum (1 mTorr), provided by a dry pump, for 5 h, at 250 • C (heating rate, 10 • C/min). After the white ZnO powder was vacuum-treated, its color changed to pale yellow. The treated ZnO NPs were denoted as Sample B.

Characterization
X-ray diffraction (XRD) spectra were recorded on a powder diffractometer (Rigaku DX-2000, Rigaku Co., Tokyo, Japan), with Cu Kα as the X-ray source (λ = 1.5418 Å); the 2-theta range was 20 • to 80 • , with a step width of 0.01 and count time of 4 s/step. The morphology of samples was characterized by a scanning electron microscope (SEM, JEOL-5900LV, Peabody, MA, USA). The Brunauer-Emmett-Teller (BET) surface area was measured on a Micromeritics ASAP 2020 (Norcross, GA, USA) adsorption apparatus, using N 2 adsorption, at −196 • C. The chemical states of the ZnO samples were characterized by X-ray photoelectron spectroscopy (XPS, PHI 5000 VersaProbe, ULVAC-PHI, Chigasaki, Japan). The electron spin state and the structure of the surface of the ZnO NPs were observed by an electron paramagnetic resonance (EPR) spectrometer (Bruker EMX-10/12, Karlsruhe, Germany), operating in the X-band frequency (f ≈ 9.774 GHz). Photoluminescence (PL) measurements were conducted in ambient conditions, with a RF-5310pc fluorescence spectrophotometer (Shimadzu, Kyoto, Japan), with an excitation wavelength of 325 nm.

Photocatalytic Tests
The photocatalytic efficiency of the ZnO samples was evaluated by monitoring the change in optical absorption of a methylene blue (MB) solution, at~662 nm, during its photocatalytic decomposition process, under UV-light irradiation. Methylene blue (C 16 H 18 N 3 SCl, Color Index No. 52015) exists as a solid, odorless, dark green powder that yields a blue solution when dissolved in water, at room temperature. It is a common contaminant in wastewater released from industrial (e.g., textile) dyeing processes, which is toxic to human health. All the photocatalysis experiments were carried out in a customized photocatalytic reactor, with a 250 W GYZ (HWL-125, Osram Lighting Co., Ltd., Berlin, Germany) as the UV light source (λ ≤ 365 nm), as shown in our previous work [5]. In a typical experiment, 50 mg each of Sample A and Sample B were dispersed, respectively, into 100 mL of MB aqueous solution (4 mg/L), in 250 mL glass beakers. Before exposing the solutions to UV-light, they were ultrasonicated for 30 s, magnetically stirred for 20 min, and then kept in darkness for 0.5 h, to establish an adsorption-desorption equilibrium.
The system was then exposed to UV-light. At 10 min intervals, 5 mL aliquots were taken out and centrifuged (5000 rpm, 5 min) to remove the catalyst from the suspension. The absorbance of MB was monitored by a UV-vis spectrophotometer. We used the following equation to calculate the degradation rate: where A 0 is the absorbance of dyes before illumination and A t is the absorbance of dyes at time, t. For multi-cycle performance tests, once the photocatalytic reaction of a testing cycle is complete, the subsequent cycle is then started, after an aliquot of 5 mL of MB mother liquor (80 mg/L) and deionized water are added to the glass beaker, to bring the concentration of the solution to approximately 4 mg/L. The length of each cycle was 100 min.

XRD and SEM Analysis
The XRD spectra of different ZnO NPs is shown in Figure 1. Three major peaks, observed at 2θ ≈ 31.8 • , 34.5 • , and 36.3 • , can be assigned to the (100), (002) and (101) lattice planes, respectively, which is in a good agreement with the standard JCPDS file No. 36-1451 for ZnO [34]. This reveals that Sample A and Sample B both belong to the hexagonal wurtzite structure with the space group P63mc [35]. No change occurred in the structure of ZnO during the vacuum treatment. The average grain sizes of the two samples, according to the Scherrer's equation, were both about 60 nm.
As shown in SEM images (Figure 2), ZnO samples are composed of particles with a hexagonal rod shape and relatively uniform particle size distribution. The crystal structure and morphology of the ZnO NPs are generally unchanged after the vacuum treatment. This is consistent with the results obtained by XRD ( Figure 1). However, the average particle size (~150 nm) of Sample B is slightly smaller than those (~200 nm) of Sample A, owning to a decrease in particle aggregation-the smaller the particle size, the larger the specific surface area. The BET measurement shows that the specific surface areas of Sample A and Sample B are 4.29 and 6.34 m 2 /g, respectively.
Appl. Sci. 2018, 8, x FOR PEER REVIEW 4 of 11 smaller than those (~200 nm) of Sample A, owning to a decrease in particle aggregation-the smaller the particle size, the larger the specific surface area. The BET measurement shows that the specific surface areas of Sample A and Sample B are 4.29 and 6.34 m 2 /g, respectively.

XPS Analysis
To determine the chemical state of the ZnO nanoparticle surfaces, XPS measurements were performed. From Figure 3, one can find that the energy differences between the Zn 2p3/2 and Zn 2p1/2 peaks of the two samples are 23.09 and 23.11 eV, respectively. These values are within the range of the XPS standard reference value of ZnO [36]. The Zn 2p XPS peak is sharp, and its corresponding binding energy indicates the presence of Zn 2+ lattice ions [37]. Figure 4 shows that O 1s core level spectra are quite asymmetric, with a broad shoulder by the side of the higher binding energy, indicative of oxygen-rich states (-OH, H2O, O2, etc.) on the surface [38]. The O 1s spectra of ZnO NPs can be described as the superposition of two Gaussian peaks. The main peak, at about 530.1 ± 0.1 eV, is attributed to oxygen in the ZnO crystal lattice (OL) [39,40], while the second peak, at a higher binding energy (OH) of 532.0 ± 0.1 eV, is related to the presence of hydroxyl groups (-OH) [37]. smaller than those (~200 nm) of Sample A, owning to a decrease in particle aggregation-the smaller the particle size, the larger the specific surface area. The BET measurement shows that the specific surface areas of Sample A and Sample B are 4.29 and 6.34 m 2 /g, respectively.

XPS Analysis
To determine the chemical state of the ZnO nanoparticle surfaces, XPS measurements were performed. From Figure 3, one can find that the energy differences between the Zn 2p3/2 and Zn 2p1/2 peaks of the two samples are 23.09 and 23.11 eV, respectively. These values are within the range of the XPS standard reference value of ZnO [36]. The Zn 2p XPS peak is sharp, and its corresponding binding energy indicates the presence of Zn 2+ lattice ions [37]. Figure 4 shows that O 1s core level spectra are quite asymmetric, with a broad shoulder by the side of the higher binding energy, indicative of oxygen-rich states (-OH, H2O, O2, etc.) on the surface [38]. The O 1s spectra of ZnO NPs can be described as the superposition of two Gaussian peaks. The main peak, at about 530.1 ± 0.1 eV, is attributed to oxygen in the ZnO crystal lattice (OL) [39,40], while the second peak, at a higher binding energy (OH) of 532.0 ± 0.1 eV, is related to the presence of hydroxyl groups (-OH) [37].

XPS Analysis
To determine the chemical state of the ZnO nanoparticle surfaces, XPS measurements were performed. From Figure 3, one can find that the energy differences between the Zn 2p 3/2 and Zn 2p 1/2 peaks of the two samples are 23.09 and 23.11 eV, respectively. These values are within the range of the XPS standard reference value of ZnO [36]. The Zn 2p XPS peak is sharp, and its corresponding binding energy indicates the presence of Zn 2+ lattice ions [37]. Figure 4 shows that O 1s core level spectra are quite asymmetric, with a broad shoulder by the side of the higher binding energy, indicative of oxygen-rich states (-OH, H 2 O, O 2 , etc.) on the surface [38]. The O 1s spectra of ZnO NPs can be described as the superposition of two Gaussian peaks. The main peak, at about 530.1 ± 0.1 eV, is attributed to oxygen in the ZnO crystal lattice (O L ) [39,40], while the  The XPS data of Zn 2p and O 1s of the vacuum-treated and pristine ZnO NPs are shown in Table 2. As the atomic ratio of Zn to OL at the surface of the samples is much higher than 1.0, this suggests that there are many oxygen vacancies at the surface of the ZnO NPs. An oxygen vacancy acts as an active site, which can easily combine and form stable structures with other active species [37]. This accounts for the high fraction of hydroxyl groups adsorbed on the ZnO surface. In addition, the atomic ratios of Zn to OL and OL to OH on the surface of Sample B are both higher than those of Sample A. Therefore, the concentration of oxygen vacancies in ZnO increases when pristine ZnO is treated. One possible reason for this is that a large number of hydroxyl groups are desorbed from the ZnO surface and then pumped out of the vacuum chamber.   The XPS data of Zn 2p and O 1s of the vacuum-treated and pristine ZnO NPs are shown in Table 2. As the atomic ratio of Zn to OL at the surface of the samples is much higher than 1.0, this suggests that there are many oxygen vacancies at the surface of the ZnO NPs. An oxygen vacancy acts as an active site, which can easily combine and form stable structures with other active species [37]. This accounts for the high fraction of hydroxyl groups adsorbed on the ZnO surface. In addition, the atomic ratios of Zn to OL and OL to OH on the surface of Sample B are both higher than those of Sample A. Therefore, the concentration of oxygen vacancies in ZnO increases when pristine ZnO is treated. One possible reason for this is that a large number of hydroxyl groups are desorbed from the ZnO surface and then pumped out of the vacuum chamber.  The XPS data of Zn 2p and O 1s of the vacuum-treated and pristine ZnO NPs are shown in Table 2. As the atomic ratio of Zn to O L at the surface of the samples is much higher than 1.0, this suggests that there are many oxygen vacancies at the surface of the ZnO NPs. An oxygen vacancy acts as an active site, which can easily combine and form stable structures with other active species [37]. This accounts for the high fraction of hydroxyl groups adsorbed on the ZnO surface. In addition, the atomic ratios of Zn to O L and O L to O H on the surface of Sample B are both higher than those of Sample A. Therefore, the concentration of oxygen vacancies in ZnO increases when pristine ZnO is treated. One possible reason for this is that a large number of hydroxyl groups are desorbed from the ZnO surface and then pumped out of the vacuum chamber.

EPR Analysis
To further confirm the XPS results, EPR measurements were performed for two samples. Figure 5 shows the EPR spectra of different ZnO NPs, recorded at room temperature. One can see that the-EPR signal (g = 1.9568) intensity is the same in both samples.

EPR Analysis
To further confirm the XPS results, EPR measurements were performed for two samples. Figure 5 shows the EPR spectra of different ZnO NPs, recorded at room temperature. One can see that the-EPR signal (g = 1.9568) intensity is the same in both samples. However, a significant EPR signal with a g factor of 2.0035 was detected in Sample B. This is commonly attributed to an unpaired electron trapped in an oxygen vacancy site [41][42][43]. Hence, a stronger g-2.0035 EPR signal corresponds to the higher concentration of oxygen vacancies in ZnO. The surface of Sample B has more oxygen vacancies than the surface of Sample A. Figure 6 shows the room temperature PL spectrum of different ZnO NPs, where the pristine and vacuum-treated ZnO both display strong UV and weak visible emission peaks, at 384 and 470 nm, respectively. The UV emission is attributed to the direct recombination of free excitons [5]. The blue-green (470 nm) emission implies that electrons trapped by the oxygen vacancies recombine with photo-generated holes [5,29]. This further confirms the existence of oxygen defects on the surface of the ZnO photocatalyst. Compared with Sample A, Sample B, with a higher concentration of surface oxygen vacancies, shows stronger visible, and lower UV, PL signals, which indicates an increase in the separation rate of photo-generated electron-hole pairs by oxygen vacancies.

Comparison of Catalytic Activity
In general, dye molecules are gradually adsorbed on the surface of a catalyst, before the photocatalytic reactions occur, and ultimately reach saturation. According to Equation (1), the percentages of MB absorbed on the surfaces of Sample A and Sample B during its adsorption-desorption process (30 min), were 2.3% and 4.2%, respectively. This clearly shows that a large number of oxygen vacancies on the surface of ZnO NPs contribute to the adsorption of dye molecules onto the photocatalytic surface. Figure 7a shows the decomposition of MB (4 mg/L), using different ZnO nano- However, a significant EPR signal with a g factor of 2.0035 was detected in Sample B. This is commonly attributed to an unpaired electron trapped in an oxygen vacancy site [41][42][43]. Hence, a stronger g-2.0035 EPR signal corresponds to the higher concentration of oxygen vacancies in ZnO. The surface of Sample B has more oxygen vacancies than the surface of Sample A. Figure 6 shows the room temperature PL spectrum of different ZnO NPs, where the pristine and vacuum-treated ZnO both display strong UV and weak visible emission peaks, at 384 and 470 nm, respectively. The UV emission is attributed to the direct recombination of free excitons [5]. The blue-green (470 nm) emission implies that electrons trapped by the oxygen vacancies recombine with photo-generated holes [5,29]. This further confirms the existence of oxygen defects on the surface of the ZnO photocatalyst. Compared with Sample A, Sample B, with a higher concentration of surface oxygen vacancies, shows stronger visible, and lower UV, PL signals, which indicates an increase in the separation rate of photo-generated electron-hole pairs by oxygen vacancies.

Comparison of Catalytic Activity
In general, dye molecules are gradually adsorbed on the surface of a catalyst, before the photocatalytic reactions occur, and ultimately reach saturation. According to Equation (1), the percentages of MB absorbed on the surfaces of Sample A and Sample B during its adsorption-desorption process (30 min), were 2.3% and 4.2%, respectively. This clearly shows that a large number of oxygen vacancies on the surface of ZnO NPs contribute to the adsorption of dye molecules onto the photocatalytic surface. Figure 7a shows the decomposition of MB (4 mg/L), using different ZnO nano-catalysts exposed to UV light. Remarkably, MB is unstable under UV irradiation, and over 13% of the dye was photo-decomposed, even in the absence of a catalyst. catalysts exposed to UV light. Remarkably, MB is unstable under UV irradiation, and over 13% of the dye was photo-decomposed, even in the absence of a catalyst. Therefore, the decomposition of MB is related to photolysis of itself and photocatalytic oxidation of the ZnO catalyst, in which absorption of the dye molecules and photocatalysis both occur on the surface of the catalyst. Without considering adsorption, photodegradation was complete (over 99% of MB was degraded) after 50 min for Sample B, whereas Sample A, under the same testing conditions, took nearly 80 min. Photodegradation of various dyes fits a pseudo first-order kinetic model [44]: where C0 is the concentration of dyes before illumination and Ct is the concentration of dyes at time, t, and kapp is the apparent rate constant. Figure 7b shows that the photodegradation of MB, in both Sample A and Sample B, conforms to the pseudo first-order kinetics approximation. The rate constants are calculated to be 0.091 ± 0.005 min −1 (R = 0.989) and 0.110 ± 0.005 min −1 (R = 0.996), respectively. The above results indicate that photocatalytic efficiency in photodecomposing MB is higher for vacuum-treated ZnO NPs than for pristine ZnO NPs. The improvement in photocatalytic efficiency is due to the decrease in particle size, caused by the reduction in particle aggregation and the increase in surface oxygen vacancies.  Therefore, the decomposition of MB is related to photolysis of itself and photocatalytic oxidation of the ZnO catalyst, in which absorption of the dye molecules and photocatalysis both occur on the surface of the catalyst. Without considering adsorption, photodegradation was complete (over 99% of MB was degraded) after 50 min for Sample B, whereas Sample A, under the same testing conditions, took nearly 80 min. Photodegradation of various dyes fits a pseudo first-order kinetic model [44]: where C 0 is the concentration of dyes before illumination and C t is the concentration of dyes at time, t, and k app is the apparent rate constant. Figure 7b shows that the photodegradation of MB, in both Sample A and Sample B, conforms to the pseudo first-order kinetics approximation. The rate constants are calculated to be 0.091 ± 0.005 min −1 (R = 0.989) and 0.110 ± 0.005 min −1 (R = 0.996), respectively. The above results indicate that photocatalytic efficiency in photodecomposing MB is higher for vacuum-treated ZnO NPs than for pristine ZnO NPs. The improvement in photocatalytic efficiency is due to the decrease in particle size, caused by the reduction in particle aggregation and the increase in surface oxygen vacancies.
Appl. Sci. 2018, 8, x FOR PEER REVIEW 7 of 11 catalysts exposed to UV light. Remarkably, MB is unstable under UV irradiation, and over 13% of the dye was photo-decomposed, even in the absence of a catalyst. Therefore, the decomposition of MB is related to photolysis of itself and photocatalytic oxidation of the ZnO catalyst, in which absorption of the dye molecules and photocatalysis both occur on the surface of the catalyst. Without considering adsorption, photodegradation was complete (over 99% of MB was degraded) after 50 min for Sample B, whereas Sample A, under the same testing conditions, took nearly 80 min. Photodegradation of various dyes fits a pseudo first-order kinetic model [44]: where C0 is the concentration of dyes before illumination and Ct is the concentration of dyes at time, t, and kapp is the apparent rate constant. Figure 7b shows that the photodegradation of MB, in both Sample A and Sample B, conforms to the pseudo first-order kinetics approximation. The rate constants are calculated to be 0.091 ± 0.005 min −1 (R = 0.989) and 0.110 ± 0.005 min −1 (R = 0.996), respectively. The above results indicate that photocatalytic efficiency in photodecomposing MB is higher for vacuum-treated ZnO NPs than for pristine ZnO NPs. The improvement in photocatalytic efficiency is due to the decrease in particle size, caused by the reduction in particle aggregation and the increase in surface oxygen vacancies.

Catalytic Stability of Vacuum-Treated ZnO NPs
Both the stability and recyclability of photocatalysts are extremely important for practical applications. Figure 8 shows the results of cycling tests of the UV-light driven photocatalytic activity of Sample B in degrading MB. Sample B shows good photocatalytic activity under UV-light irradiation, and no appreciable loss of activity after eight cycles. The remarkable photocatalytic stability of the vacuum-treated ZnO NPs indicates its feasibility in practical applications.

Catalytic Stability of Vacuum-Treated ZnO NPs
Both the stability and recyclability of photocatalysts are extremely important for practical applications. Figure 8 shows the results of cycling tests of the UV-light driven photocatalytic activity of Sample B in degrading MB. Sample B shows good photocatalytic activity under UV-light irradiation, and no appreciable loss of activity after eight cycles. The remarkable photocatalytic stability of the vacuum-treated ZnO NPs indicates its feasibility in practical applications.

Photocatalytic Mechanism
The mechanism for the photocatalytic degradation of MB with a ZnO photocatalyst is shown in Figure 9. The effectiveness of ZnO NPs in MB decomposition can be explained on the basis of the •OH radicals and holes present on the surface of ZnO NPs, which break down MB dye in aqueous solutions. When a semiconductor particle absorbs a photon with an energy (hν) matching or exceeding its bandgap energy (Eg), an electron (ecb − ) is excited from the valence band (VB) to the conduction band (CB), leaving a hole (hvb + ) behind. The photo-generated electrons and holes can recombine and

Photocatalytic Mechanism
The mechanism for the photocatalytic degradation of MB with a ZnO photocatalyst is shown in Figure 9. The effectiveness of ZnO NPs in MB decomposition can be explained on the basis of the •OH radicals and holes present on the surface of ZnO NPs, which break down MB dye in aqueous solutions.

Catalytic Stability of Vacuum-Treated ZnO NPs
Both the stability and recyclability of photocatalysts are extremely important for practical applications. Figure 8 shows the results of cycling tests of the UV-light driven photocatalytic activity of Sample B in degrading MB. Sample B shows good photocatalytic activity under UV-light irradiation, and no appreciable loss of activity after eight cycles. The remarkable photocatalytic stability of the vacuum-treated ZnO NPs indicates its feasibility in practical applications.

Photocatalytic Mechanism
The mechanism for the photocatalytic degradation of MB with a ZnO photocatalyst is shown in Figure 9. The effectiveness of ZnO NPs in MB decomposition can be explained on the basis of the •OH radicals and holes present on the surface of ZnO NPs, which break down MB dye in aqueous solutions. When a semiconductor particle absorbs a photon with an energy (hν) matching or exceeding its bandgap energy (Eg), an electron (ecb − ) is excited from the valence band (VB) to the conduction band (CB), leaving a hole (hvb + ) behind. The photo-generated electrons and holes can recombine and When a semiconductor particle absorbs a photon with an energy (hν) matching or exceeding its bandgap energy (E g ), an electron (e cb − ) is excited from the valence band (V B ) to the conduction band (C B ), leaving a hole (h vb + ) behind. The photo-generated electrons and holes can recombine and dissipate the input energy as heat, get trapped in metastable surface states, or react with electron donors and acceptors adsorbed on the semiconductor surface or within the surrounding double electrical layer of charged particles [14]. However, surface defect states, related to oxygen vacancies, are available to trap electrons or holes. Thus, the probability of recombination is reduced and subsequent redox reactions may occur. As mentioned above, oxygen vacancies play an important role in the degradation of dye under UV-light irradiation. They create an impurity level between the C B and V B of the vacuum-treated ZnO, which acts as a deep donor defect level, near the top of V B [45]. The V o defect energy levels used in Figure 9 are obtained by calculating the emission energy (2.64 eV) corresponding to the peak around 470 nm (PL spectrum shown in Figure 6). The calculated bandgap energy of 3.23 eV corresponds to the band-edge emission peak at around 384 nm, as shown in Figure 6. Indeed, an "empty" oxygen site (V o 2+ ) can trap either one (V o + ) or two (V o 0 ) electrons [41], thereby effectively reducing the recombination rate of the e cb − -h vb + pairs. More e cb − and h vb + migrate to the surface of the vacuum-treated ZnO and promote the photocatalytic degradation reactions. Moreover, particle size and photocatalytic activity are related. In particular, semiconductor NPs exhibit quantum size effects as the particle size decreases. This size reduction may improve the photocatalytic efficiency by increasing the effective bandgap [46]. In addition, reducing the particle size leads to a larger specific surface area. The latter is defined as the total surface area used in the redox reaction, with a corresponding increase in the number of active sites exposed. A decrease in particle size should also lead to higher photonic efficiency, due to a higher interfacial charge carrier transfer rate [47].
For wastewater treatment, dispersibility of the catalyst is another performance factor. As more catalyst particles are dispersed in the liquid, the total reaction surface area increases and this leads to a higher light harvesting efficiency. This will promote the adsorption of molecules on the surface of the catalytic particles. Consequently, the overall photocatalytic efficiency of the pristine ZnO NPs increases after vacuum treatment.

Conclusions
In summary, pale yellow ZnO nanoparticles (NPs) have been synthesized, using a vacuum drying process, to deoxidize ZnO powder. In comparison with pristine ZnO NPs, this vacuum treatment technique results in smaller ZnO NP sizes and an increase in oxygen vacancies. Moreover, these vacuum-treated ZnO NPs preserve their pristine crystal structures. In photocatalytic experiments under UV-light irradiation, the vacuum-treated ZnO NPs exhibit better photocatalytic activity (over 99% of MB is degraded after 50 min UV irradiation time, with the effective reaction rate k app ≈ 0.10977 min −1 ) and stability (excellent photocatalytic activity still exists after eight cycles) in the process of photodecomposing MB. The SEM, BET XPS, EPR and PL characterizations suggest that the observed higher photocatalytic activity of the vacuum-treated ZnO NPs stems from the smaller particle sizes and the abundance of oxygen vacancies on their surfaces.