Oxygen Vacancy-Mediated Interfacial Charge Transfer of Au/ZnO Schottky Heterojunctions for Enhanced UV Photodegradation

We intend to report an interesting phenomenon related to the di ﬀ erent interfacial transfer processes between ellipsoidal-like ZnO (E-ZnO) and rod-like ZnO (R-ZnO) nanoheterojunctions witness by the nanosecond time-resolved transient photoluminescence (NTRT-PL) spectra. Fristly, E-ZnO and R-ZnO nanoarchitectures were fabricated via facilitating the electrochemical route; and then, they decorated it with dispersed Au nanoparticles (NPs) by the methods of ion-sputtering deposition, constituting Au/E-ZnO and Au/R-ZnO Schottky-heterojunction nanocomplex, which is characterized by SEM, XRD, Raman analysis, and UV-vis absorption spectra. Steady-state photoluminescence and NTRT-PL spectra of as-fabricated Au/E-ZnO and Au/R-ZnO nanocomposites were probed for interfacial charge transfer process under 266 nm femtosecond (fs) light irradiation. Simultaneously, a distinct diversi ﬁ cation for the NTRT-PL spectra is observed, closely associating with oxygen vacancies (V o ), which is con ﬁ rmed by X-ray photoelectron spectroscopy (XPS) and electron spin resonance (ESR) spectra. Furthermore, Au NPs act as an “ annular bridge ” and “ transit depot ” for interfacial charge transfer through local surface plasmon resonance (LSPR) e ﬀ ect and Schottky barrier, respectively, which is identi ﬁ ed by NTRT-PL and time-resolved PL (TRPL) decay spectrum. Moreover, this mechanism is responsible for the enhanced photoelectrochemical (PEC) performances of methyl orange (MO) photodegradation under UV light irradiation.


Introduction
Zinc oxide (ZnO), which is a representative II-VI semiconductor with a wide direct bandgap of 3.3 eV and a large excitation binding energy of 60 meV, has attracted considerable attention because of its unique optical and electrical properties. Recently, ZnO has been used as an alternative to TiO 2 for photoelectrochemical (PEC) performance due to its nontoxicity, earth-abundant, and high electron mobility efficiency [1]. Meanwhile, ZnO has a wide application field, including energy conversion and storage, optoelectronics, and environmental remediation [2][3][4]. At present, various kinds of synthetic methods were adopted for preparing the ZnO nanostructure, such as sol-gel, chemical vapor deposition, ultrasonic-assisted method, magnetron sputtering, molecular beam epitaxy, and electrochemical deposition [5][6][7][8][9][10]. Among the preparation methods above, the electro-chemical deposition route provides a better alternative for its facile and feasible operation process. It is well known that the pristine ZnO semiconductor is intrinsically not a good candidate for PEC implementation, which is limited by light absorption and higher electron-hole (e --h + ) pair recombination rate. And these issues are reported to be settled by coupling wide bandgap semiconductors (WBGs) with plasmonic metal nanocrystals (e.g., Au, Ag, and Pt), i.e., M/S Schottky contact, as they could improve the concentration of PEC active electrons and facilitate the photogenerated e --h + pair separation, which originated from local surface plasmon resonance (LSPR) effect and built-in electric field of Schottky nanoheterojunction, respectively. Researchers have confirmed [11] that pure Au NPs possess greatly photocatalytic activity irradiated by UV light, associated with interband transition from 5d to 6sp-band. We are inclined to believed that the PEC performance of Au/ZnO Schottky-heterojunction nanohybrids would effectively promote mechanisms related with the synergistic effect of energic electron generation and accelerated charge transfer.
The photocatalytic properties of ZnO can be strongly influenced by introducing native point defects, including oxygen vacancies (V o ), zinc vacancies (V Zn ), oxygen interstitials (O i ), and zinc interstitials (Zn i ), especially for V o . Recently, researchers [12] have reported the V o intrinsic defects could induced visible-light absorption and enhanced PEC activity. Wang et al. [13] concluded that the high concentration of V o defects present in ZnO not only as a trapping levels, but also creates impurity levels near the valence band (VB) of ZnO and causes bandgap narrowing. Furthermore, Xie et al. [14] have confirmed experimentally that the presence of V o can strongly promote interfacial carrier transfer of a ZnO-related nanoheterostructure. Meanwhile, the photocatalytic activities associated with inherent defects for ZnO are significantly influenced by its different morphologies, including sphere-like, rod-like, flower-like, rice-like, hexagonal disk-like, and other nanostructures [15][16][17][18][19]. Thus, an efficient PEC process for ZnO can be mediated either by the presence of defect states related to morphology, or intentionally created defects within the bandgap through doping with transition metal ions [20].
Photoinduced interfacial charge transfer is a prevalent phenomenon, which is crucial for semiconductor devices of optoelectronic conversion and photocatalysis. Of permanent interest questions are photogenerated e --h + pair separation and recombination process and to what extent native defect states can participate in charge transfer processes at the interface of the heterojunction. Fortunately, transient photoluminescence (PL) spectroscope is a nondestructive and sensitive optical characterization to investigate the fundamental physical properties of ZnO, including inherent energy band and active defects [21]. As investigated in our previous literature, the corresponding PL spectra obtained from ZnO shows UV and broad visible emission, which are usually assumed to originate from the near-band-edge (NBE) direct recombination and defect-related indirect recombination emission, respectively [22].
In this work, we prepared a series of specimens of ellipsoidal-like and rod-like ZnO nanoarchitectures decorated with loading controllable Au nanoparticles (NPs) (denoted as Au/E-ZnO and Au/R-ZnO, respectively) and investigated their PEC performance-adopted methyl orange (MO) photodegradation under UV light irradiated. The Au/ZnO Schottky heterojunction nanocomposites are fabricated via a facile electrochemical deposition and ionsputtering strategy. The loading contents of the Au NPs can be easily controlled by varying the sputtering time. Meanwhile, compared with the conventional defect-doping methods, we provide an effective way to create V o in ZnO via thermal annealing, which is more favorable on preventing from introducing any impurity and preserving the native crystal structures of ZnO. Furthermore, nanosecond timeresolved transient photoluminescence (NTRT-PL) spectra present direct evidence that the interfacial transfer mechanism of the photoinduced electrons in the Au/ZnO nanoheterojunction associated with intrinsic defect states. The PL emission of ZnO is seen to be drastically quenched in the presence of Au NPs, showing their strong influence on the charge carrier dynamics. To the best of our knowledge, there are only few reports for this interesting phenomenon, which is better understood by introducing competing mechanisms between the photogenerated carrier defect trapping and PEC redox process in the interfaces of the Au/ZnO Schottky heterostructure.

Experimental Section
High-purity reagents were used without further purification for the undoped ZnO sample synthesis. Ellipsoidal-like and rod-like ZnO nanoarchitectures were prepared with an electrochemical deposition approach reported in the previous work [23]. The typical synthetic procedure was given as follows: firstly, the fluorine-doped tin oxide (FTO) conductive glass substrates (1:2 × 1:5 cm), which with a sheet resistance of 12 Ω, were separately cleaned by ethanol and acetone, and then rinsed with distilled water for 5 min in an ultrasonic cleaner (KQ5200DE). And then, a facile electrodeposition route was applied to synthesize ZnO nanoellipsoid (E-ZnO) and ZnO nanorod (R-ZnO) thin films, respectively. The experiments were conducted in a conventional potentiostatical three-electrode system (CHI600E, CH Instruments Co. Ltd.) at 80°C (±2°C) for 10 min, in which the FTO substrate was used as the working electrode, a Pt plate was applied as the counter electrode and an Ag/AgCl (KCl saturated) was employed as the reference electrode. The electrolyte contained 0.1 M zinc nitrate and 0.1 M hexamethylenetetramine (HMT), and the pH value of it was approximately 5.5. Lastly, the electrochemical deposition was performed for the E-ZnO and R-ZnO samples at a constant potential of -1.5 V and -2.0 V versus the Ag/AgCl reference electrode, respectively. For better crystallization and morphology, the as-prepared E-ZnO and R-ZnO thin films on the FTO substrates were annealed at 450°C and 350°C for 30 min in a tubular furnace, respectively. Well-controlled sample preparations are critical to form nice nanostructures of well-discreted Au NPs on the E-ZnO and R-ZnO film substrates.
Au NPs were deposited on the samples of E-ZnO and R-ZnO thin films by using an ion-sputtering system, and the isolated Au NPs on the ZnO films were prepared by thermal annealing in a tubular furnace with different sputtering times (20 s, 40 s, and 80 s), respectively. The highest ramp speed of the temperature in the furnace was 10°C/min. When the preset temperature (400°C) was reached, the samples were annealed for 30 min.
The morphologies and microstructures of the Au/ZnO nanocomposites were evaluated by scanning electron microscopy (SEM, Hitachi S4200) operating at accelerating voltage of 15.0 kV. A UV-visible (UV-vis) diffuse reflectance spectroscopy (UV-1800, Shimadzu) was applied to study the optical absorption properties of the as-prepared Au/E-ZnO and Au/R-ZnO heterojunctions in air at room temperature. The composition and phase pure of Au NP-decorated ZnO films were characterized by X-ray diffraction (XRD, Shimadzu XRD-600) using Cu Kα radiation (λ = 0:15418 nm). The native defect states of the samples for pristine-ZnO and Au/ZnO were examined by X-ray photoelectron spectroscopy (XPS, ESCALAB 250, Thermo Fisher Scientific Ltd.), and the X-ray source was the Al anode emitting the Kα (1486.6 eV) radiation of 150 W. Crystal structure and chemical bonding states of as-formed Au/ZnO Schottky heterojunctions were investigated by using the micro-Raman spectroscopy system (Horiba JY-HR800) equipped with a confocal microscope with an Ar + laser operating at 532 nm at a excitation power level of 6 mW at room temperature. Electron spin resonance (ESR) measurements were performed on a JEOL-TE300 spectrometer operating at an X-band frequency of 9.4 GHz.
As described in our previous report [24], a Ti : sapphire femtosecond (fs) laser system (Spectra-Physics) was used to evaluated the steady state and the NTRT-PL spectra for Au/E-ZnO and Au/R-ZnO thin films, respectively. Briefly, the excitation laser beams with wavelength centered at 266 nm, is acquired accordingly by the third harmonic generation (THG) of 800 nm laser beam through a BBO (beta-BaB 2 O 4 , barium metaborate) crystal, with pulse duration of 130 fs, repetition rate of 1 kHz, and spot size of 1 mm. The PL emission is collected by a spectrometer (Bruker Optics 250 IS/SM) coupled with an intensified charge-coupled device (CCD) detector (Istar 740, Andor). Steady-state and transient-state experiments were carried out by adjusting the delay time of the CCD shutter. Laser pulse gives an external trigger signal to open 100 ms and 0.5 ns temporal gate of the intensified CCD through a synchronization/delay generator (SDG) and digital delay/pulse generator (DG 535); then, the steady-state and transient PL spectra at different decay times can be recorded by the CCD detector, respectively. Figure 1 gives a schematic diagram of the experimental setup. Simultaneously, the NTRT-PL decay spectra measurement was performed in FLUOROMAX-4 spectrofluorometer in room temperature.
The methyl orange (MO) photocatalytic degradation procedure is as follows: MO of 5 mg was dissolved in 500 mL of deionized water to yield a concentration of 10 mg/L, and the mixture was stirred in the dark for 30 min to reach an adsorption-desorption equilibrium. The supernatant was transferred to a quartz cuvette to measure its absorption spectrum, which was exposed to UV irradiation for different durations at a constant temperature of 300 K. The concentration of MO solution was checked every 30 min using a UV-vis spectrophotometer by monitoring the intensity variation for the characteristic absorption peak at 465 nm.

Results and Discussion
The surface morphology feature of the pristine-ZnO and Au/ZnO with different sputtering times (20 s, 40 s, and 80 s) was characterized by SEM as shown in Figure 2.     Figure 3(a) and denoted by the symbol "■,"which corresponds to that of the hexagonal wurtzite ZnO; some diffraction signals are identical to Figure 3(a) and marked with the symbol "◆," which is classified as the FTO substrates; other diffraction peaks lie in 38.14°and labeled as the "▼," which were attributed to the presence of Au NPs. Furthermore, the average crystallite sizes of all the deposited Au NPs (D Au ) on the ZnO matrix can be estimated from the (111) peaks using Scherrer's formula [25]: D = Kλ/β cos θ, where K is a constant (0:89 < K < 1), λ is the wavelength of the incident Xray radiation, β is the full-width at half-maximum (FWHM) of the diffraction peak, and θ is the Bragg angle. The D Au for the samples of Au/E-ZnO 20 s, Au/E-ZnO 40 s, Au/R-ZnO 20 s, and Au/R-ZnO 40 s were calculated to be 15 nm, 43 nm, 18 nm, and 45 nm, respectively. It was explicitly indicated that the D Au increased with increasing Au NP sputtering times from 20 s to 40 s, which was in good agreement with the SEM observation.
Raman spectroscopy is an efficient avenue to assess the crystal quality and structural defect states on the surface of an as-fabricated sample.  In the present investigation, the Raman spectra of Au NP unloaded and loaded R-ZnO and E-ZnO samples all showed the remarkable peak at 437 cm -1 is the typical optical phonon E 2 high mode, corresponding to the band characteristic of the wurtzite phase [27], which is consistent with the XRD analysis. Meanwhile, the peaks at 580 cm -1 (E 1 LO mode) were found to be evident in all Raman spectra, which are associated with the formation of native defects like V o and Zn i in ZnO [28]. A new vibration located at 560 cm -1 (B 1 low mode) appeared for the samples of Au/E-ZnO and Au/R-ZnO with different Au NP loading times, which was absent for the pristine E-ZnO and bare R-ZnO. This peak can be assigned to the newly created defects associated with Au NPs attached [29], and its intensity increased with the increase of Au NPs sputtering time. Although, XRD measurement does not probe any strain deformation in the ZnO crystal lattice, but the incorporation of metal NPs can introduce surface defect states, which originated from the lattice mismatch between metal NPs and ZnO. Further, the E 1 LO and E 2 high peaks are associated with the lattice oxygen and hence the lowering of its intensity with the increase Au NP amount, which indicates the evolution of V o -related defects in the lattice. Raman spectra data confirmed the preparation of single-phase hexagonal wurtzite ZnO structure and formation of defect states in ZnO.
To evaluate the effect of intrinsic defects on the energy gap of the pristine-ZnO and Au/ZnO Schottky heterojunctions prepared under different conditions, the light absorption properties of pristine E-ZnO, bare R-ZnO, Au/E-ZnO, and Au/R-ZnO samples were investigated by UV-vis diffuse reflectance spectroscopy. As can be seen from Figure   corresponding to the bandgap energy (E g ) of 3.14 eV, 3.11 eV, 3.08 eV, 3.06 eV, 3.03 eV, 2.98 eV, 2.93 eV, and 2.89 eV, respectively. The absorption edges believed to ascribe to the NBE excitonic absorption; with the annealing temperature increasing, it is distinctly observed that the band edge red-shifted and broadened as the morphology of ZnO nanoparticles changed from nanorod to nanoellipsoid, which is in accordance with the previous literature [30]. Furthermore, the decrease in optical bandgap values with increasing loading contents of Au NPs. Chen et al. [31] concluded that Au NPs could induce excess defects within the bandgap of ZnO under thermal annealing, thus reducing the valid energy gap. Simultaneously, theoretically and experimentally, we have confirmed that the concentration of V o defects could impactfully modulate the E g of metal oxide with WBGs attributed to doped noble metal NPs. Light absorption characterization indicated that the presence of Au NPs can effectively expand the light absorption range of Au/ZnO nanocomplex, especially for Au/E-ZnO, and improve its PEC performance. We are inclined to believe that the change trend of optical band gaps for prepared Au/ZnO samples originated from the concentration variety of V o defects, which was in accordance with the Raman spectroscopy investigation. Besides, the enhanced visible light absorption band from 500 nm to 600 nm is attributed to the LSPR absorption of Au NPs [32]. The increase absorption intensity is ascribed to size increasing of Au NPs as confirmed in the above SEM observations.
In order to further pinpoint the defect characteristic of Au/ZnO hybrid heterojunctions, especially for V o , the ESR measurements were carried out and the spectra of Au/R-ZnO and Au/E-ZnO with different loading times (0 s, 20 s, 40 s, and 80 s) are shown in Figure 6, respectively. It was explicitly observed that all the ESR signals of Au NP undoped and doped ZnO samples exhibited significant peaks with a g factor of 2.03 (close to the free electron value g = 2:002), which is a characteristic of unpaired electrons trapped on surface V o sites [33]. The intensity of the ESR signal for pristine E-ZnO was higher than that of bare R-ZnO, demonstrating that the concentration of V o defects was increasing with the increased aspect ratio, which was in good agreement with the results of SEM and UV-vis diffuse reflectance spectra and be useful to the visible-light photocatalytic process. For the Au/ZnO samples with different Au NP loading amount, the intensity of the ESR signal peaks were increasing with increased Au NP depositing times from 0 s to 40s, revealing more V o defects could have been induced by doping of noble metal NPs. Moreover, this indicates that there are more superoxide radical anions (·O 2 -) generated from surface adsorption oxygen molecules (O 2 ) that captured the free electron [34]. For the sample of Au/ZnO with loading time 80 s, less concentration of V o defect-associated·O 2 was observed, owing to its lower specific surface area.
For further evaluation of the chemical states and elemental composition of the as-prepared Au/ZnO Schottky nanoheterojunctions, XPS analysis was performed on Au/R-ZnO 20 s and pristine R-ZnO nanocomplex. The survey spectrum result of Au/R-ZnO 20 s in Figure 7(a) demonstrated only peaks that were attributed to Zn, O, C, and Au, indicating the sample comprised of Zn, O, and Au elements. In addition, the C 1 s peak (284.0 eV) was treated as a reference, which was probably derived from chamber contamination prior to the XPS examined [35]. The detective results were in well concordance with SEM and XRD as elucidated above. High-resolution XPS spectra of Zn, Au, and O species were demonstrated in Figures 7(b)-7(d), respectively. The Zn 2+ valence state in the prepared samples was confirmed by the XPS spectra of Zn 2p, which were shown in Figure 7(b). It was observed that the Zn 2p spectrum consists of two binding energy (BE) peaks 1021.9 eV and 1044.9 eV corresponding to the 2p 3/2 and 2p 1/2 for pristine R-ZnO, respectively.   International Journal of Photoenergy Meanwhile, the core line of Zn 2p for Au/R-ZnO 20s, including BE of 2p 3/2 and 2p 1/2 were located at 1022.2 eV and 1045.2 eV, respectively. Compared with the corresponding value of the pure R-ZnO, the shift of BE peaks could be ascribed to the reduction of the surface band bending [14]. Furthermore, the 23 eV difference between the two Zn 2p core levels originated from the spin-orbit coupling effect [36]. The XPS spectrum of Au 4f for Au/R-ZnO 20 s was presented in Figure 7(c). Two significant BE peaks located at 83.6 eV and 88.8 eV were indexed to the electronic states of Au 4f 7/2 and Au 4f 5/2 , respectively. Moreover, examination of XPS results indicated that Au was present in the Au 0 state, and further validated that as-formed Au NPs were successfully decorated on R-ZnO nanosurfaces. The locations of Au 4f two peaks were shifted to higher BE associated with that of bulk Au, which consisted of 83.1 eV and 87.2 eV corresponding to Au 4f 7/2 and Au 4f 5/2 , respectively. This shift phenomenon can be ascribed to electron transport from ZnO to the Au, which could cause the accumulation of negative charge density on the surface of Au NPs [37]. The fitting results of O 1 s XPS spectrum for Au/R-ZnO 20s are shown in Figure 7    International Journal of Photoenergy curve), which correspond to the lattice oxygen, oxygen deficient, and loosely bound oxygen, respectively. It is believed that the surface area of O b is connected to the variations in the concentration of V o [14].
Therefore, we can obtain the concentrations of V o for Au/E-ZnO with different Au loading amount associated with the surface area ratios (SAR) of O b /ðO a + O c Þ and detailed descriptions as shown in Table 1. It is observed that the area ratio value of Au/E-ZnO 40 s sample is explicitly greater in the Au/E-ZnO samples; the SAR values were increasing with increased Au NP loading times from 0 s to 40 s, as Table 1 shows, while the area ratio of Au/R-ZnO 20s is 49.5%, less than that of Au/E-ZnO 20 s (53.8%), owing to the E-ZnO possessing more specific surface area. It illustrates that the concentration of V o in Au/E-ZnO samples were increased when decorated with more Au NP deposition; this result is consistent with the former conclusion of Raman and ESR analyses.
The PL emission technique has been widely used to reveal the electronic construction of intrinsic surface defects and probe for understanding the surface radiative recombination processes in which photoexcited e --h + pairs take part, i.e., primary processes in photocatalysis. Figures 9(a)-9(d) demonstrated the steady-state PL spectra of the pristine ZnO and Au NP-decorated ZnO with different depositing times (20 s, 40 s, and 80 s) in ambient condition at room tempera-ture with excitation wavelength of 266 nm, respectively. In the detected spectral region, the integration time of steadystate PL was 100 ms. As illustrated in Figures 9(a) and 9(b), steady-state PL emission spectra for Au/E-ZnO nanoheterojunctions with different loading times 0 s, 20 s, 40 s, and 80s, respectively are shown. Emission behaviour of pristine E-ZnO was plotted, and the main profile of the PL spectrum can be deconvoluted into four categories, i.e., UV emission band and different visible emission bands, as shown in Figure 9(a). The sharp emission located at 390 nm (UV emitted) be pertained to NBE photogenerated electron direct transition recombination of ZnO between the conduction band (CB) and the VB [40], which is in accordance with ZnO bandgap energy (ca. 3.23 eV). Meanwhile, the visible PL emission band was decomposed into three Gaussian   [42], ,and photoexcited electron radiative transitions from CB to doubly ionized oxygen vacancies (V o ++ ) defects [43], ,respectively. The steady-state PL spectra of Au/E-ZnO with different Au NP loading amounts were plotted in Figure 9(b), which explicitly reveals that a charge transfer was enhanced at the Au/E-ZnO interface when E-ZnO was decorated with Au NPs. Apparently, a decrease in the UV PL-emitted intensity with increasing Au NP depositing time, demonstrating that existence of Au NPs could effectively suppress the direct radiative recombination and facilitated photoformed free electron transport at the Au/ZnO interface. Furthermore, visible PL emission intensity (especially for 2.52 eV) increased with increasing Au NP loading, which strongly suggests the V o + defect center was the primary shallow donor energy level in E-ZnO. Similarly, steady-state PL emission spectra were carried out for Au/R-ZnO nanoheterojunctions with different deposition times (0 s, 20 s, 40 s, and 80 s), as presented in Figures 9(c) and 9(d), respectively. Figure 9(c) illustrated the main fea-tures of steady PL for bare R-ZnO, which can be deconvoluted into five categories. Two typical PL emission bands were observed: emission at 390 nm (direct radiative transition from CB to VB) and emission in the visible region (indirect radiative transition from CB to defect level). Besides the fitting peaks of 390 nm, 493 nm, 606 nm, and 650 nm, the emission peak located at 670 nm was inclined to electron radiative transitions from CB to V o ++ defects [42,43]. The patterns of steady-state PL spectra for Au/R-ZnO Schottky heterojunctions with different Au NP loading times were shown in Figure 9(d), disclosing the key role for Au NPs inhibiting the photogenerated e --h + pair direct recombination and promoted the charge carriers transfer between the interface of Au NPs and R-ZnO. Moreover, the steady-PL intensities located at 1.85 eV of Au/R-ZnO hybrids distinctly increased with increasing Au NP depositing, indicating that the V o ++ defects were the dominant deep donor energy levels in R-ZnO. Summarizing the above steady-PL analysis, the number of V o + and V o ++ defects was significant increased in Au/E-ZnO and Au/R-ZnO heterojunction nanocomplex with increasing Au NP loading amount, respectively, since the PL emission intensity mainly results from the interband radiative recombination of the photoexcited free charge 9 International Journal of Photoenergy carriers. Hence, we were inclined to believe that a large number of defect active points can be induced at the interface between Au NPs and ZnO [44], improving PEC performance was expected.
The essence of the PEC process is photoredox reaction. Elucidation of photoinduced charge transfer dynamic mechanism is vital to understand the subsequent photoredox reactions and also for evaluating the interband electron transition at the interfacial energy band structures of Au/ZnO hybrid nanoheterojunctions; time-resolved transient PL spectroscopy characterization technique was evaluated. As we have seen, Figures 10(a)-10(d) presented the NTRT-PL spectra for pristine E-ZnO and Au/E-ZnO Schottky nanocomplex with different Au NP depositing times (20 s, 40 s, and 80 s) in ambient condition at room temperature under 266 nm fs light irradiation, respectively. In the tested spectral region, the time interval of NTRT-PL was 1.5 ns. Moreover, it is worth mentioning that Yanome et al. [45] had investigated the time scale for electron trapped recombination associated with defect states and holes in VB is a nanosecond order, which determines the photocatalytic activity of the heterojunction nanocomposites. Accordingly, we have reasonably inclined to believe that the NTRT-PL spectrum was in good coincidence with the time scale for free exciton interfacial charge transfer dynamic process.
Meanwhile, Figures 11(a)-11(b) presented schematic illustrations of photoexcited free exciton segregation and reunion transient process for pristine E-ZnO and Au/E-ZnO Schottky heterojunctions at room temperature and atmospheric pressure under exciting light of 266 nm irradiation, respectively. The main photoreaction processes for pristine E-ZnO, as follows: Normally, atmospheric molecular oxygen (O 2 ) is adsorbed on the surface of E-ZnO. Initially triggered by 266 nm (4.7 eV) irradiation (t = 0 ns), plenty of electrons were promoted from the VB across the bandgap to the CB since the energy of excited light is greater than the bandgap value of E-ZnO (equation (1) (5)). After 1.5 ns (t = 1:5 ns), the PL peaks located at 390 nm appeared, which could be ascribed to direct radiative recombination associated with NBE (equation (6)).
The e -CB increased with increasing spectral recording time (t = 3 ns); apparently, the intensities of transient PL peaks located at 390 nm, 493 nm, 603 nm, 650 nm, and 670 nm were enhanced compared with that before 1.5 ns. With the evolution of spectral recording time (t = 4:5 ns), a part percentage of e -CB can be trapped by electron acceptors such as adsorbed O 2 on the surface of E-ZnO to produce superoxide radical anions (O 2 -) (equation (7)), and the holes of VB (h + VB ) could be entrapped by the surface hydroxyl ions (OH -) to yield hydroxyl radicals (·OH) (equation (8)), which possessed powerful oxidation capability that can degrade the organic molecules. Subsequently, the intensity of transient PL peak located at 390 nm quenching at the end of recording time (t = 6 ns), which is caused by the immediately decreased concentration of e -CB . By contrast, the interfacial charge transfer mechanism for the Au/E-ZnO Schottky nanoheterojunctions as depicted in Figures 10(b)-10(d) and 11(b). In the initial of spectral recording time (t = 0-1.5 ns), the peak positions of transient PL spectra were identical with those of pristine E-ZnO, which originated from NBE and defect radiative recombination transition. The Schottky barrier height of Au/E-ZnO was determined by the difference between the work function of Au and the electron affinity for ZnO. Because the work function of Au is -5.3 eV [46] (vs. vacuum level) and the electron affinity for E-ZnO is approximately -4.5 eV [47] (vs. vacuum level), so the Schottky barrier should be 0.8 eV [48]. With the lapse of time (t = 3 ns), the intensity of NBE emission band, located at 390 nm, disappeared or had drastic reduction;, mainly resulting from the photogenerated e -CB transfer to Au, driven by a built-in electric field force at the interface between Au NPs and the E-ZnO caused by Schottky barrier (process ①). At the interface of Au/E-ZnO nanocomplex, the LSPR effect of Au NPs was stimulated by the defect PL emissions of E-ZnO between 500 nm and 600 nm, which coincided with the UV-vis light absorption characterization. Typically, the free excitons in Au NPs cannot transfer to the CB of ZnO, in virtue of the obstacle from the Schottky barrier. As the LSPR effect was triggered, plenty of excitons situated at E F would be excited as "hot electrons" [49] (process ②). Expectedly, under the recording time evolution (t = 4:5-6 ns), substantial energetic hot electrons will easily transfer from the Au NPs injected to the CB of E-ZnO (process ③). This interfacial charge transfer process increased the number of e -CB and resulted in an enhanced radiative intensity for NBE and defects PL emission, which were characterized by NTRT-PL spectra. Furthermore, it could be definitely observed that the transient PL intensities of defect emissions for samples of Au/E-ZnO 0 s, 80 s, 20 s, and 40 s, palpably increased sequentially at recording time of 6 ns, indicating the number of defects associated with photoredox active points induced by Au NPs enhanced in a sequence. During the process of charge transfer, the O 2 adsorbed on the surface of Au NPs, which isattached on E-ZnO, were able to extract e -CB to produce·O 2 -. Analogously, to further obtain insightful information for depicting the separation and recombination of photogenerated e --h + pairs, from which the efficacy of transportation and entrapping of charge carriers can be obtained.
At the beginning of the photoexcitation through 266 nm fs light irradiation (t = 0 ns), it is well known that the photogenerated electrons were directly stimulated from the VB to the CB since the energy of incident photons was larger than the threshold energy of R-ZnO (equation (1)). Subsequently, the transient PL peaks located at 493 nm, 606 nm, 650 nm, and 670 nm appeared, which are ascribed to radiative recombination associated with V o + , O i , and V o ++ intrinsic point defects, respectively (equations (2), (4), (8), and (16)). As recording time goes after 1.5 ns (t = 1:5 ns), the NBE emission transient PL peak located at 390 nm emerged, attributing to e -CB and h + VB in ZnO direct radiative recombination (equation (5)). While concentration of e -CB increased with increasing the recording time (t = 3 ns), the NBE emission PL intensity was remarkably enhanced, meanwhile the transient PL intensities of defects were significantly improved. Under the recording time evolution (t = 4:5 ns), all the transient PL intensities evidently depressed, compared with that at previous recording time.
We have a reason to believe the results arised from e -CB extracted by adsorbed O 2 on the surface of R-ZnO to produ-ce·O 2 -(equation (6)), accompanied by h + VB combined with OHto generate ·OH (equation (7)). The sustained consumption of e -CB with the evolution of spectral recording time (t = 6 ns), giving rise to the quenched transient PL intensity of 390 nm and depressed defects PL emission. Figures 12(b)-12(d) and 13(b) demonstrated transient PL spectra and credible mechanism explanation for Au/R-ZnO nanoheterojunctions, respectively. In the present case, considering the difference of electron affinity and work function for R-ZnO and Au NPs (-4.2 eV and -5.3 eV vs. vacuum level, respectively) was 1.1 eV [50], which formed the Schottky barrier is slightly higher than that of E-ZnO. Accordingly, the initial electrons transfer between the R-ZnO and Au NPs would cause a band bending. Irradiation of the Au/R-ZnO nanocomposites by 266 nm exciting light at initial time (t = 0-1.5 ns) would excite the free electrons in VB to the CB, which is higher than the CB edge of R-ZnO. The highdensity e -CB can be transferred to the defect levels of ZnO if they are situated below the Fermi level (E F ) of Au, causing an enhancement of defect emission. Remarkably, Gong et al. [51] emphasized that the dominant V o defect states in R-ZnO are V o ++ defects, which ascribed to V o + defects would be formed to higher total energy than V o ++ defects. Thus, the defect PL emission heavily focused on 650 nm and 670 nm, which originated from e -CB radiative trapped by V o ++ defects. The concentration of e -CB increased with increasing recording time (t = 3 ns); a great many of e -CB direct recombination with the h VB + to radiate higher 390 nm emission light, besides, transient PL intensity of defect emission turned more stronger. As we known, O 2 can be much efficiently adsorbed on the surface of Au NPs, which acted as the electron acceptor via a built-in electric field force at the interface of Au/R-ZnO Schottky nanohybrids. Au NPs served as "transit depot" of electrons in the whole transient charge transfer process. Subsequently, at the last stage of the transient PL recording time (t = 4:5-6 ns), owing to the substantial interfacial tranfer of e CB from R-ZnO to Au NPs associated with powerful Schottky barrier, the PL intensity of NBE and V o ++ defect emission decreased gradually until it vanished with the evolution of recording time. Evidently, LSPR effect cannot be excited by Au/R-ZnO Schottkyheterojunction nanocomplex, possibly originating from its defect PL irradiation around 650 nm-670 nm, which deviated from the excitation range for Au NP LSPR effect. It is worth mentioning that the difference of interfacial charge transfer from ZnO to Au NPs for Au/E-ZnO and Au/R-ZnO, mainly attributing to the specific surface area of E-ZnO is larger than that of R-ZnO, which favored more adsorption of O 2 and witness by previous SEM characterization.
Time-resolved PL (TRPL) decay spectrum is a powerful and nondestructive detection technique, which is typically used to probe the e -CB lifetime of the specific emission band. To further obtain insightful information for elucidating the charge transfer (i.e., charge injection) efficiency and evaluating the carrier lifetime at the interface between Au NPs and ZnO Schottky-heterojunctions, TRPL decay profiles were performed for the samples of Au/R-ZnO 20 s and Au/E-ZnO 0 s, 20 s, 40 s, and 80 s with the excitation wavelength of 350 nm at room temperature for NBE emission band, respectively, as illustrated as Figure 14. Currently, the TRPL decay curves were well fitted to the biexponential pattern described by the equation: IðtÞ = A 1 exp ð−t/τ 1 Þ + A 2 exp ð−t/τ 2 Þ, and the e -CB average lifetime (τ avg ) of the samples were calculated from the relation: τ avg = ðA 1 τ 1 2 + A 2 τ 2 2 Þ/ðA 1 τ 1 + A 2 τ 2 Þ, where τ 1 is the fast decay component related to the free excitons and τ 2 represents the slow decay component associated with the bound excitons. A 1 , A 2 are the relative amplitude of fast decay and slow decay, respectively. Simultaneously, the efficiency of hot carrier injection (η in ), which originated from the LSPR effect excited by Au NPs, can be obtained from the defined relation: η in = ½τ avg ðAu/E−ZnOÞ − τ avg ðE−ZnOÞ /τ avg ðE−ZnOÞ , where τ avg ðAu/E−ZnOÞ represents the average lifetime for Au/E-ZnO nanohybrid and τ avg ðE−ZnOÞ is the average lifetime for pristine E-ZnO. Table 2 presents the values of A 1 , A 2 , and τ avg of e -CB , and η in for Au/R-ZnO 20 s and Au/E-ZnO Schottkyheterojunctions with different Au NP amount (0 s, 20 s, 40 s, and 80 s), respectively.
As illustrated in Table 2, the τ avg for Au/R-ZnO 20s, pristine E-ZnO, and Au/E-ZnO with different Au NP depositing times for 20 s, 40 s, and 80 s are 4.04 ns, 5.15 ns, 6.46 ns, 7.31 ns, and 5.68 ns, respectively. It can be distinctly discovered that the τ avg of e -CB for Au/R-ZnO 20s is minimum, and the τ avg and η in increased with the Au NP depositing time increasing from 20 s to 40 s and then decreased with a continuous distribution of Au NP deposition time 80 s was applied. Logically, the τ avg of the photogenerated carrier is heavily depended on the concentration of e -CB . The NBE emission band located at 390 nm resulted from the direct radiative recombination of ZnO between CB and VB. Therefore, the raise of τ avg for Au/E-ZnO samples compared with 13 International Journal of Photoenergy pristine E-ZnO, resulting from the e -CB injection excited by Au NPs; yet, the depression of τ avg for Au/R-ZnO 20 s sample associated with the reduction of e -CB concentration, which were attributed to intenser built-in electric field force induced by Schottky barrier and higher charge transfer efficiency [52]. The parameter η in with a direct correlation to the amount of hot electron injection, which is sensitive with the discrete particle size of Au NPs. The increased particle size with the increasing Au NP deposition time from 20 s to 40 s coincides with promoting intension of hot electron cyclic injection. As the Au NP depositing time was 80s, the surface of E-ZnO is covered with successive films constructed by Au NPs, which presumaed strong decoupling the resonance effect between E-ZnO and Au NPs.
In the Au/E-ZnO Schottky nanocomposites, the Au NPs can be regarded as a circular bridge of carrier transport, which transfer the V o + defects PL emission energy to the NBE emission, and resulted in the prolonged τ avg of photoexcited charge carriers. Meanwhile, the longer τ avg is associated with more effective separation of e --h + pairs, which demonstrates more photogenerated charges would participate in PEC reactions. By contrast, the Au NPs really act as transit depot of unidirectional e -CB transfer in the Au/R-ZnO Schottky nanocomplex, which relied on vigorous internal electronic driving force in the charge depletion region and acquired effective transfer capability for interfacial photoinduced carriers. There is an obvious contradiction between impressive injection of energetic electrons and rapid interfacial electroic transport for more effectively boost the PEC activity.
In order to validate the hypothesis, we carried out the experiments of MO photodegradation for the samples of Au/R-ZnO 20 s and Au/E-ZnO with different Au NP content under UV light irradiation, respectively. The excitation wavelength of UV irradiated was 254 nm with luminescence power 11 W, which is in accordance with the excited wavelength of 266 nm for the steady-and transient-state PL investigation. Detailed photodegradation experimental procedures were described above, using concentration of MO (10 mg/L) in the photocatalytic experiment. The photodegradation efficiency (η) of MO could be calculated as η ð%Þ = ½ðC 0 − CÞ/ C 0 × 100%, where C 0 is the initial concentration of the MO, and C is the concentration after photoirradiation. Figure 15 presents the experimental results of five cycles under the same conditions under UV irradiation, respectively. The average η for self-degradation efficiency of MO, Au/R-ZnO 20 s, and the Au/E-ZnO Schottky nanocomposites with different Au NP depositing times (0 s, 20 s, 40 s, and 80 s) are equal to 10.6%, 51.3%, 25.2%, 64.1%, 72.8%, and 55.6%, respectively. This results distinctly indicate that Au NPs play a vital role in facilitating a photoexcited electron transfer to the interface before it is radiatively recombined at defects or band edge emission. Meanwhile, the UV photocatalytic activities increased with the Au NP-decorated E-ZnO film depositing time increased from 0 s to 40s and then decreased with deposition time 80 s, which coincided with the ultrafast dynamics of charge injection between E-ZnO and Au NPs.
In order to evaluate the reaction kinetics of the MO photodegradation catalyzed by Au/R-ZnO 20 s, pristine E-ZnO, and Au/E-ZnO with Au NP deposition times 20 s, 40 s, and 80 s, the experimental data in Figure 16 were fitted by the Langmuir-Hinshelwood equation: ln ðC 0 /CÞ = kt, where C 0 and C are the initial and residual concentration of MO in reaction mixture (mg/L), t refers to exposure time (hour), and k is the reaction rate constant. According to the approximate linear fitting curves of ln ðC 0 /CÞ versus irradiation time t, the values of k for above samples were calculated to be 0.00042 min -1 , 0.0063 min -1 , 0.002 min -1 , 0.0067 min -1 , 0.0073 min -1 , and 0.0081 min -1 , respectively. It is explicitly demonstrated that k of ZnO decorated with Au NPs were significantly higher than that of bare ZnO, benefiting from the synergetic effect of energetic charge injection and expeditious interfacial charge transfer. Furtermore, Au NPs also absorb the UV light to cause electron transition from 5d to 6spband, which leads to the improved PEC performance of Au/ZnO in comparison with pristine ZnO. In addition, the magnitude of the reaction rate is strong depending on the   14 International Journal of Photoenergy amount of defect sites, which is associated with the aspect ratio of the specimen. Besides the strong electronic interaction between Au NPs and ZnO can be induced by the LSPR effect, a higher specific surface area is another vital factor improving photocatalytic activity, which resulted in a remarkable promotion in photodegradation rate of Au/E-ZnO compared with Au/R-ZnO. The k of Au/E-ZnO heterojunctions increased from 0 s to 40s and then decreased with deposition time 80s, originating from injection efficiency of hot electrons induced by more appropriate particle size of Au NPs, which is in accordance with our expectation.

Conclusion
In summary, Au/R-ZnO and Au/E-ZnO Schottky heterojunction nanocomposites were successfully synthesised through a two-step electrochemical and ion-sputtering depo-sition method, and they exhibited fascinated interfacial charge transfer process mediated by V o defects under 266 nm irradiation, which is witnessed by NTRT-PL spectra and XPS analysis. Furthermore, the surface decoration of plasmonic Au NPs has been demonstrated as an effective approach to improve the PEC performance irradiated by UV light, ascribing to the synergistic effects of energenic electrons injection and rapid interfacial charge transfer. For Au/E-ZnO nanocomplex, the injection electron process from plasmonic Au NPs into the E-ZnO matrix is the key factor for enhancing the efficiency of the photocatalytic degradation process. Yet for Au/R-ZnO nanocomposites, the attached Au NPs at the surface of R-ZnO form close Schottky barrier contacts between the metal and semiconductor play a crucial role for improving the PEC activity during UV light irradiation. In addition, the difference roles of Au NPs in Au/ZnO Schottky nanoheterojuntions are discussed separately, associated with transient PL emission of NBE as well as defects, which acts as "annular-bridge" and "transit depot" in Au/E-ZnO and Au/R-ZnO nanohybrids, respectively. This proposals are supported by the characterisations of steadyand transient-state PL spectra, which explicitly reveal the intrinsic competition relations between the interfacial transfer of photogenerated e --h + pairs and recombination process. We believe that the design and synthesized Au/ZnO nanostructure in this work not only provides new insights into the mechanism of interfacial charge transfer assoiated with V o defects but also exploits new prospects for the development of M/S Schottky heterojunction-improved PEC devices.

Data Availability
All data, models, and code generated or used during the study appear in the submitted article.

Conflicts of Interest
There are no conflicts of interest to declare.