Formation of surface defects by thermal shock method for the improved photocatalytic activity of ZnO nanoparticles

ABSTRACT Surface-modified ZnO photocatalysts with enhanced UVA-light-driven and visible-light-driven activities were synthesized by the thermal shock method with Cu(NO3)2 at different thermal shock temperatures (300 – 600°C). The influences of thermal shock temperatures on the crystal structure, morphology, surface functional groups and surface composition of modified catalysts were investigated by XRD, TEM, Raman and XPS spectra, respectively. Their photocatalytic activity was evaluated via the degradation of methylene blue under both UVA and visible light irradiation. According to the results, by combining the thermal shock method and an agent with low thermal stability such as Cu(NO3)2, we did not modify the crystal structure, phase composition nor the morphology of ZnO nanoparticles, but successfully modified the surface of ZnO nanoparticles with the migration of zinc ions, leading to the creation of new environments of Zn2+ and O2 – ions as well as the formation of surface zinc vacancies. These evolutions were found to be able to enhance the photocatalytic performance in the UVA light region and also in the visible light region.


Introduction
Over the past decades, photocatalysis based on the performance of various transition metal oxides like ZnO has been considered as one of the most promising methods for water treatment [1][2][3][4]. However, the recombination of photogenerated electrons-holes which occurs during the photocatalytic process as well as the possibility to operate under UV light irradiation due to the large bandgap are the main reasons limiting the applications of ZnO in industrial scale. Therefore, a wide range of studies have been carried out to modify the oxide with nonmetal or transition metal ions. Recently, several works revealed that the creation of intrinsic or extrinsic defects such as oxygen vacancies, zinc vacancies, zinc interstitials and impurities can help to effectively control the photocatalytic performance of ZnO. Specifically, the presence of these defects can improve or decline the oxide activity, depending on the type and the position of defects [5][6][7][8]. According to the work of Chen et al. [9], when the oxygen vacancies or zinc vacancies are formed in the bulk structure of ZnO, they can act as the recombination centers of photogenerated charges, preventing the transport of photogenerated electrons and holes to the surface and consequently their reactions with adsorbed species. Likewise, by using a simple ballmilling technique, Aggelopoulos et al. successfully created the oxygen vacancies in ZnO lattice and observed the systematic decrease in photocatalytic activity when the milling time increased [10].
By contrast, the formation of different defects on the surface of ZnO has been recognized as a potential way to ameliorate its photocatalytic response. In fact, by using the ZnO 2 precursor, Wang et al. successfully synthesized ZnO nanoparticles with a high concentration of oxygen vacancies on the material surface [11]. These surface oxygen vacancies are supposed to create an intermediate energy level near the valence band of this oxide, which induces the bandgap narrowing and resultantly the photocatalytic performance in the visible light region [12,13]. Moreover, the surface oxygen vacancies were also proved to behave as electron traps, promoting the separation of photogenerated electrons-holes and thereby effectively improving the photocatalytic activity [14,15]. Similar to oxygen vacancies, it was reported that the surface zinc vacancies can also create impurity energy levels in the bandgap of ZnO [16]. As a result, the ZnO photocatalysts with surface zinc vacancies are capable of absorbing the visible light and resulting in the visible-light-induced activity.
Nevertheless, in order to create the surface defects, ZnO was usually prepared and treated by complicated procedures (atomic layer deposition [17], hydrothermal process [18] or vacuum deoxidation [13]) with expensive chemicals (ZnO 2 [11], NaBH 4 [15]). Recently, we developed the thermal shock technique, a simple, easy-to-implement method which is based on the thermal treatment of a material during a short time [19,20]. Different from the combustion synthesis method relying on a thermally induced redox reaction between an oxidant and a fuel [21][22][23][24], the thermal shock method is simply an abrupt temperature change of a solid material with appropriate precursors at some high temperatures during a short time, which allows us to forcefully modify the surface of this material without affecting its bulk structure. In fact, due to the short heating time (only 5 min), the modifications caused by thermal shock are likely to be limited to the surface with the formation of surface defects. Since most of the photocatalytic reactions occur on the surface of a heterogeneous semiconductor, this modification method is found to be very suitable for the performance control of photocatalysts. Moreover, in order to create a great number of surface defects, we assume that it is necessary to use the precursors which can be thermally decomposed during the thermal shock, thereby creating disturbances on the sample surface. Therefore, in this work, we suggest to modify ZnO nanoparticles by using the thermal shock technique with Cu(NO 3 ) 2 precursor. Since Cu(NO 3 ) 2 is a compound that can be decomposed and produce gas at low temperature, we believe that using this precursor during the thermal shock method is a promising way to improve the photocatalytic activity of ZnO under both UVA and visible light illumination. The influences of thermal shock with Cu(NO 3 ) 2 precursor at different temperatures on the crystal structure, surface morphology and surface composition of ZnO were also investigated and discussed in detail.

Catalyst preparation and modification
In this study, the precursors Zn(CH 3 COO) 2 .2H 2 O, Cu(NO 3 ) 2 .2.5H 2 O (>98%, ACS reagent) and H 2 C 2 O 4 .2H 2 O (>99%, ACS reagent) were purchased from Sigma Aldrich (USA). Methylene blue (MB, analytical grade) was also purchased from Sigma Aldrich (USA). All chemicals were used as received without further purification. Distilled deionized water was used in all experimental work.
In the first stage, ZnO nanoparticles were prepared by a simple precipitation-annealing method. Briefly, 0.5 mol.L -1 Zn(CH 3 COO) 2 solution was added to H 2 C 2 O 4 solution (0.5 mol.L -1 ), magnetically stirred at room temperature during 1 h to obtain a white precipitate. This precipitate was thoroughly washed with deionized water until there is no more ion detected in the filtrate (using conductimetry tests). Then, the precipitate was dried at 120°C for 1 h and subsequently heated at 500°C for 2 h to produce ZnO nanoparticles.
In the second stage, the as-prepared ZnO nanoparticles were surface-modified by thermal shock method with Cu(NO 3 ) 2 precursor at different temperatures. First of all, Cu(NO 3 ) 2 solution was poured into the beaker containing ZnO powder to reach the molar ratio of Zn: Cu fixed at 1:0.005. The mixture was regularly stirred for 20 min to well distribute Cu(NO 3 ) 2 to the surface of ZnO nanoparticles and dried at 150°C during 3 h. Next, the thermal shock for 5 min was carried out by putting the obtained powder into the Nabertherm muffle furnace (maximum 9 kW) preheated at different temperatures (300°C, 400°C, 500°C and 600°C with a heating rate of 1°C .min -1 ). After 5 min, the powder samples were immediately removed from the furnace and cooled in an ambient atmosphere to room temperature. Finally, the products were washed with distilled deionized water, dried at 150°C for 1 h in order to remove nitrate species from the ZnO surface. In the following manuscript, the ZnO samples surface-modified with Cu(NO 3 ) 2 by thermal shock were labeled as mod-ZnO-X (with X the thermal shock temperature). The ZnO sample modified only by thermal shock without Cu(NO 3 ) 2 was also prepared by the same procedure (thermal shock temperature of 300°C) and denoted as TS-ZnO.

Characterization
The crystal structure and the phase composition of unmodified ZnO and surface-modified ZnO samples at different thermal shock temperatures were investigated by the powder X-ray diffraction on a BRUKER-Binary V3 X-ray diffractometer equipped with a Cu Kα source (λ = 1.5406 Å). The accelerating voltage and the applied current were fixed to 40 kV and 40 mA, respectively. The scanning rate was held constant at 0.019°s -1 . In order to characterize the surface morphology of catalysts, the TEM (transmission electron microscopy) images were captured on a JEOL JEM-1400 system at the accelerating voltage of 100 kV.
The Raman spectroscopy was used to investigate the vibrations of chemical bonds on the surface of catalysts. The Raman spectra were taken by a HORIBA Jobin-Yvon spectrophotometer consisting of diode pumped solid-state laser with 532 nm wavelength.
Besides, the surface atomic composition and chemical environment of unmodified ZnO and surface-modified ZnO samples were also studied by X-ray photoelectron spectroscopy (XPS) on a Kratos Axis Ultra DLD spectrometer (Kratos Analytical Ltd, UK) using the Al-Kα source (1486.6 eV) and the pressure of 10 −7 mbar. All binding energies of elements in different chemical environments on the surface of catalysts were referenced to the C1s line at 285 eV, corresponding to the surface contamination carbon. The Rietveld refinement was carried out by using the Fullprof 2009 structure refinement software [25].

Catalytic tests
The photocatalytic performance of unmodified ZnO and surface-modified ZnO samples was examined via the MB degradation under UVA light and visible light illumination. In a typical photocatalytic test, 250 mL of MB solution (10 −5 mol.L −1 ) was mixed with 0.125 g catalyst powder in a glass beaker under continuous stirring. The initial pH of suspensions was determined at 7. An Osram 9-W UV light lamp and an Osram 9-W visible light lamp were used for the irradiation in UVA and visible light region, respectively. These lamps were placed about 10 cm above the suspension surface. Before the lamps were turned on, the suspension was vigorously stirred in the dark for 60 min to ensure the adsorption-desorption equilibrium of MB aqueous solution with the photocatalysts. Then, the solution containing MB and catalysts was exposed to the UVA light or visible light for a certain period of time. A water circulation system was used to maintain the reactor temperature at 30°C during the experiments. In order to monitor the progress of MB degradation, the absorbance of MB at 664 nm was measured at regular time intervals by a UV-visible Helios Omega spectrophotometer.

Crystal structures and morphology
In order to investigate the impacts of thermal shock modification on the crystal structure, phase composition and surface morphology, the unmodified ZnO, TS-ZnO and mod-ZnO catalysts were, respectively, characterized by powder XRD and TEM techniques. The XRD patterns of all samples (Figure 1) exhibit the intense and symmetrical diffraction peaks at 31.82°, 34.41°, 36.22°, 47.60°, etc., which are characteristic of ZnO zincite phase (space group P63mc, JCPDS file no. 36-1451) in wellcrystallized states. Moreover, no impurity phase was observed in these patterns. Likewise, the TEM images of all samples show the same morphology, with polyhedral particles in the size distribution from 20 to 100 nm ( Figure 2). These results demonstrate that the thermal shock process with Cu(NO 3 ) 2 at different temperatures did not modify the phase composition, crystal structure and the particle size of ZnO nanoparticles.

Surface characterization
As mentioned above, in this study, we focused on studying the surface modification of ZnO nanoparticles by thermal shock method with Cu(NO 3 ) 2 . Figure 3 shows the Raman spectra of three samples ZnO, mod-ZnO-300 and mod-ZnO-500. The unmodified ZnO sample presents two peaks with high intensity, situated at 96.28 and 437.84 cm -1 , which are featured for ZnO wurtzite phase [26,27]. The former peak is attributed to the E 2 (low) vibration mode of zinc ions while the latter peak belongs to the E 2 (high) vibration mode of oxygen ions on the oxide surface. In addition, a weak peak was observed at 330.84 cm -1 in the Raman spectrum of unmodified ZnO sample, corresponding to the second-order scattering processes of E 2 (high)-E 2 (low) multi-phonon mode [28,29]. When ZnO was modified with Cu(NO 3 ) 2 by thermal shock method at different temperatures, the Raman spectra still present these three characteristic peaks, but the intensity of the peaks associated with E 2 (low) and E 2 (high) modes dramatically decreases. Moreover, in the Raman spectra of surface-modified catalysts, the E 2 (high) peak is found to be shifted to the lower wavenumber. These changes reveal that the thermal shock with Cu(NO 3 ) 2 affected the Zn-O bonds on the catalyst surface, which can lead to disturbances of zinc and oxygen ions or the formation of intrinsic defects on the surface of ZnO nanoparticles.
Therefore, in order to elucidate the influences of thermal-shock modification on the surface composition and chemical environment of atoms on the surface, all our catalysts were analyzed by XPS, a specific method for surface characterization of solid materials. The XPS results are summarized in Table 1. Figure 4 presents the high-resolution Zn 2p spectra of unmodified ZnO, TS-ZnO, mod-ZnO-300 and mod-ZnO-500 samples. For unmodified ZnO catalyst, the Zn 2p 3/2 and 2p 1/2 spin-orbit peaks are observed at 1021.4 and 1044.5 eV, respectively, corresponding to the binding energies of Zn 2+ ions in ZnO lattice [30,31]. When ZnO was surface-modified at 300°C without Cu(NO 3 ) 2 , the position and the shape of these peaks were not altered. However, when ZnO was surface-modified at 300°C with Cu(NO 3 ) 2 , the Zn 2p peaks became asymmetrical, indicating the evolution in composition and  chemical environment of zinc ions on the surface. For this sample, the Zn 2p peaks can be deconvoluted into 3 components: the main components (Zn I ) situated at 1021.3 (Zn 2p 3/2 ) and 1044.4 eV (Zn 2p 1/2 ), characteristic of Zn 2+ ions in ZnO lattice [30,31] and two other spin-orbit components with lower intensity at 1019.-2-1042.3 eV (Zn II ) and 1017.6-1040.4 (Zn III ). Since their binding energies are lower than that of Zn 2+ in ZnO lattice, the new minor components can be attributed to the zinc ions less oxidized or zinc ions in oxygendeficient regions. However, when the thermal shock temperature was up to 500°C, the intensity of these components decreased, indicating the declined content of Zn 2+ ions in the new chemical environments. Figure 5 presents the O 1s XPS spectra of unmodified ZnO, TS-ZnO, mod-ZnO-300 and mod-ZnO-500 samples. It was observed that the O 1s core peaks of both four samples are asymmetric and comprise more than one component. For unmodified ZnO and TS-ZnO samples, the core peaks O 1s can be deconvoluted into two components: the main and intense component (O I ) at 530.2 eV assigned to O 2ions in ZnO lattice [32] whereas the weak component at 531.8 eV attributed to surface hydroxyl groups [33]. However, the XPS O 1s spectra of mod-ZnO-300 and mod-ZnO-500 samples are more complicated. In addition to O I and O II components, the O 1s peaks of mod-ZnO-300 catalyst also contain two new components at lower binding energies: 528.4 (O III ) and 526.4 eV (O IV ). This is very interesting because, according to our best understanding, there seems to be no report about oxygen environments having lower binding energy than that of O 2 -   Table 2, the unmodified ZnO and TS-ZnO samples show the molar Zn 2+ /O 2ratio of 1/1, which is consistent with the stoichiometric ratio of elements in ZnO. However, for mod-ZnO-300 and mod-ZnO-500 samples, this molar ratio greatly decreased, indicating the zinc-deficiencies on their surfaces. The zinc-deficiencies are likely to correspond to the formation of zinc vacancies on ZnO surface after the thermal-shock modification. Moreover, the content of surface hydroxyl groups of ZnO powders modified with Cu(NO 3 ) 2 also tends to decline. It should  be noted that when we changed the thermal shock temperature from 500°C to 300°C, the evolution of elemental composition and chemical environments on the catalyst surface became stronger, which was proved by the further decrease in molar Zn 2+ /O 2ratio (from 0.868 to 0.835) and the further increase in the content of new components of surface zinc and oxygen ions. This suggests that the thermal shock at 300°C affects the surface of ZnO nanoparticles more than that at 500°C. Besides, we do not detect any copper species on the surface of modified ZnO samples, indicating that the content of copper species may be below the detection limit of XPS technique. All these results demonstrate that using Cu(NO 3 ) 2 allows us to effectively modify the surface of ZnO, creating the disturbance in chemical environments of Zn 2+ and O 2ions without leaving impurities on the surface. Figure 6 and 7 compare the MB degradation over different photocatalysts under UVA light ( Figure 6) and visible light (Figure 7). It was observed that the photocatalytic MB degradation always followed the pseudo-first-order Langmuir-Hinshelwood kinetic model, which allows us to evaluate the performance of our photocatalysts via their apparent rate constants k (Table 3). Under UVA and visible light, the rate constant of MB degradation in the presence of unmodified ZnO was found to be 3.20 and 0.17 h -1 , respectively. These rate constants remained practically  unchanged for TS-ZnO sample. But when ZnO was modified with Cu(NO 3 ) 2 by thermal shock method at 300°C, the photocatalytic activity was greatly enhanced. In fact, the rate constant of MB degradation using mod-ZnO-300 sample was nearly 3 times (under UVA light) and 2 times (under visible light) higher than those using unmodified ZnO sample. However, further increase of the thermal shock temperature (from 300°C to 600°C) resulted in a gradual decrease in photocatalytic performance. Under UVA light or visible light, the mod-ZnO-600 only exhibited the rate constant of 5.32 h -1 and 0.18 h -1 , respectively. These results depict that the thermal shock temperature plays an important role for the performance of photocatalysts. Among all samples, the mod-ZnO-300 was found to be the best photocatalyst under both UVA light and visible light.

Discussion
The experimental results demonstrated that the thermal shock method with or without Cu(NO 3 ) 2 barely lead to any modification in the crystal structure, phase composition and morphology of ZnO nanoparticles. However, in the presence of Cu(NO 3 ) 2 , the thermal shock can induce various changes in the chemical composition and environments on the surface of ZnO. It should be reminded that all mod-ZnO samples depicted higher photocatalytic activity under both UVA light and visible light than ZnO and TS-ZnO samples. Therefore, the enhanced performance of mod-ZnO catalysts is most likely due to the surface modifications after the thermal shock process. In fact, when ZnO was thermal-shock-modified with Cu(NO 3 ) 2 at 300°C, the XPS study indicated many changes on the material surface, including the formation of new environments of Zn 2+ and O 2ions, the decrease in surface hydroxyl groups and the formation of surface zinc vacancies. Specifically, on the surface of mod-ZnO -300, Zn 2+ ions are not only situated in the chemical environment of ZnO, but also located in the environment less oxidized than ZnO or in the oxygen-deficient regions. Likewise, a number of O 2ions were also found in the zinc-excessive regions. These evolutions were not clearly observed in the TS-ZnO sample, proving the vital role of Cu(NO 3 ) 2 for the surface modification of ZnO nanoparticles. It should be noted that before the thermal shock process, ZnO nanoparticles were mixed in Cu(NO 3 ) 2 solution. Owing to this step, all the surfaces of ZnO nanoparticles were covered by Cu(NO 3 ) 2 . Since Cu(NO 3 ) 2 can be decomposed at low temperature (263°C [34]), in the next step, the thermal shock at 300°C possibly induced the thermal decomposition of Cu(NO 3 ) 2 , which causes numerous disturbances on the surface of ZnO, including the migration of Zn 2+ ions. Due to their small radius, a portion of Zn 2+ ions can leave their balance positions to move to the interstitial sites, consequently forming at the same time zinc vacancies and new chemical environments where Zn 2+ ions are less oxidized and O 2ions are connected to Zn 2+ ions more than the conventional environment of O 2ions in ZnO. Moreover, the interstitial Zn 2+ ions can interact with surface hydroxyl groups, convert the Zn-O-H bonds to Zn-O-Zn bonds. Hence, the surface hydroxyl groups tend to decrease on the surface of mod-ZnO samples.
Previously, many studies reported that the enhanced photocatalytic activity can be obtained by increasing the surface hydroxyl groups [35,36]. Nevertheless, in this work, the photocatalytic performance of our catalysts was effectively improved despite the decrease in surface hydroxyl groups, which indicates that the content of surface hydroxyl groups is not the main factor determining the activity. In contrast, according to several recent studies, surface zinc vacancies have been considered as a positive factor that can contribute to the enhancement of photocatalytic activity. Through DFT calculations, Pan et al. successfully proved that the formation of zinc vacancies is able to create the intermediate energy level in the bandgap, above the valence band of ZnO [16]. Owing to these intermediate levels, the electrons in the ZnO valence band require less energy to move to the conduction band, which hinders the recombination of electron-hole pairs, promotes the visible-lightabsorption, then improves the photocatalytic performance under both UVA light and visible light. This argument is further reinforced through the bandgap values of ZnO and mod-ZnO-300 samples measured by the UV-Visible diffuse reflectance spectra which were recorded at room temperature on a Perkin-Elmer Lambda 850 Spectrophotometer. Accordingly, the bandgap value of mod-ZnO-300 (2.87 eV) is found to be lower than that of unmodified ZnO (3.19 eV), which demonstrated the impact of surface zinc vacancies on the optical properties of ZnO nanoparticles. However, when the thermal shock temperature overpassed 300°C, the quantity of Zn 2+ and O 2ions in the new chemical environments as well as the content of surface zinc vacancies gradually decreased. It seems that at thermal shock temperatures above 300°C, Cu(NO 3 ) 2 decomposes too quickly, which reduces the impacts of modification on the oxide surface thus declines the photocatalytic activity. This problem needs to be further clarified in our future studies.

Conclusion
In this work, by using the thermal shock method with Cu(NO 3 ) 2 agent at different temperatures, we successfully modified the surface of ZnO nanoparticles without affecting their crystal structure, phase composition and morphology. All surface-modified ZnO samples exhibitedan enhanced photocatalytic activity under both UVA light and visible light. This improvement of activity can be explained by the impacts of thermal shock with Cu(NO 3 ) 2 , which promotes the migration of zinc ions and then the formation of zinc vacancies. The thermal shock temperature was found to be closely related to the content of surface zinc vacancies and the photocatalytic performance. Among all our samples, the mod-ZnO-300 is the best photocatalyst owing to the highest content of surface zinc vacancies. When the thermal shock temperature overpassed 300°C, the surface zinc vacancies decreased, leading to the declined photocatalytic activity.

Disclosure statement
No potential conflict of interest was reported by the authors.