Microstructure-Dependent Visible-Light Driven Photoactivity of Sputtering-Assisted Synthesis of Sulfide-Based Visible-Light Sensitizer onto ZnO Nanorods

The ZnO-CdS core-shell composite nanorods with CdS shell layer thicknesses of 5 and 20 nm were synthesized by combining the hydrothermal growth of ZnO nanorods with the sputtering thin-film deposition of CdS crystallites. The microstructures and optical properties of the ZnO-CdS nanorods were associated with the CdS shell layer thickness. A thicker CdS shell layer resulted in a rougher surface morphology, more crystal defects, and a broader optical absorbance edge in the ZnO-CdS rods. The ZnO-CdS (20 nm) nanorods thus engaged in more photoactivity in this study. When they were further subjected to a postannealing procedure in ambient Ar/H2, this resulted in the layer-like CdS shell layers being converted into the serrated CdS shell layers. By contrast, the ZnO-CdS nanorods conducted with the postannealing procedure exhibited superior photoactivity and photoelectrochemical performance; the substantial changes in the microstructures and optical properties of the composite nanorods following postannealing in this study might account for the observed results.


Introduction
Advancements have been substantial in the design and understanding of the physical properties of low-dimensional semiconductor heterostructures [1,2]. Low-dimensional semiconductor heterostructures, because of their special band alignment at the interfaces, have potential applications in photoactivated and gas-sensing devices [3,4]. Among various heterostructures consisting of two types of semiconductors, the type-II heterojunction provides an effective spatial separation of charge carriers upon photoexcitation by tuning the energy levels of the valence and conduction bands of the constituent semiconductors [5]. A wide-bandgap semiconductor is coupled with a narrow bandgap semiconductor to form a heterostructure, which extends the photoresponse into the visible-light region, after which enhanced light harvesting is expected [3]. Researchers have investigated numerous semiconductor heterostructure systems consisting of oxide and sulfide semiconductor-for examples, TiO 2 -CuS [6], ZnO/In 2 S 3 [7], ZnO-CuS [8], ZnO-Ag 2 S [9], and ZnO-CdS [10,11]. ZnO, because of its high electron mobility and various fabrication methods, has garnered considerable attention for the fabrication and investigation of ZnO-based heterostructures for their enhanced light-driven properties [7,12]. The drawback of ZnO is its wide-bandgap value (approximately 3.3 eV), which induces the activation of ZnO only through UV irradiation. A visible-light sensitizer is therefore crucial for integration into ZnO to broaden the use of solar light when the device functions of ZnO-based materials necessitate operation under photoexcitation.
Among various narrow-bandgap metal sulfide semiconductors, the CdS is a suitable visible-light sensitizer for ZnO, because CdS has a direct bandgap value in the visible-light region, and the crystallographic features of CdS are similar to those of ZnO; these advantages engender a close energy-band interaction between the two semiconductors when they are formed in a heterostructure [13,14]. An effective interband photoexcited charge transfer occurs between CdS and ZnO [14]. Several synthesis methods have been reported for fabricating CdS with various crystal features, including radio-frequency (RF) sputtering [15], successive ionic layer adsorption and reaction [16], the hydrothermal method [13], chemical bath deposition [17], thermal evaporation [18], chemical vapor deposition [19], and spray pyrolysis [20]. However, most ZnO-CdS heterostructure systems are synthesized by decorating CdS crystallites onto ZnO templates through aqueous chemical growth methods [10,11,14]. By comparison, RF sputtering is more advantageous for preparing CdS with various crystallite thicknesses and crystalline qualities. Incorporating CdS into the ZnO template is also advantageous for forming ZnO-CdS heterostructures through RF sputtering. The synthesis methods have a considerable influence on the microstructures of the semiconductors. Understanding the correlation between the crystal feature and photoactivated properties of the low-dimensional ZnO-CdS heterostructures by sputtering deposited CdS crystallites onto ZnO templates through a physical method is of scientific importance and relevant reports are limited in number. The sputtering-assisted synthesized ZnO-CdS heterostructures are favorable to investigate effects of structural modification of CdS crystallites on photoexcited properties of the heterostructures; this information is important for designing ZnO-CdS nanorods through a vacuum process with a satisfactory photoactivity. In the present study, CdS crystallites with various thicknesses were RF-sputtered onto ZnO rods to form ZnO-CdS core-shell heterostructure rods; moreover, a simple postannealing procedure was adopted to modulate the crystal properties of CdS shell layer. The microstructures and photoactivated properties of the sputtering technique facilitated the formation of ZnO-CdS heterostructure nanorods, which were investigated in this study.

Materials and Methods
In this study, ZnO-CdS core-shell nanorods with two different CdS shell layer thicknesses were fabricated by sputtering CdS thin films onto the surfaces of the hydrothermally derived ZnO nanorod templates (hereafter referred to as ZnO-CdS (5 nm) and ZnO-CdS (20 nm)). The CdS thin films were fabricated using RF magnetron sputtering in pure Ar ambient at 250 • C. The gas pressure during thin-film deposition was fixed at 2.67 Pa, and the sputtering power was fixed at 40 W. Detailed experiments on the synthesis of vertically aligned ZnO nanorods have been described elsewhere [21]. Some of the ZnO-CdS (20 nm) nanorods are subjected to a postannealing procedure in ambient Ar/H 2 (3% H 2 balanced) for 20 min at 400 • C to modulate crystal feature of the CdS crystallites.
Crystal structures of the as-synthesized core-shell rods were investigated by X-ray diffraction (XRD) using Cu Kα radiation.
The surface morphologies of the various samples were characterized by scanning electron microscopy (SEM), and high-resolution transmission electron microscopy (HRTEM) was used to investigate the detailed microstructures of the core-shell rods. Room-temperature-dependent photoluminescence (PL) spectra were obtained using the 325 nm line of a He-Cd laser. The analysis of absorbance spectra of the core-shell rods were conducted by using UV-Vis spectrophotometer. Photocatalytic activity of various samples were performed by comparing the degradation of aqueous solution of methylene blue (MB; 10 −6 M) with various core-shell nanorods as catalysts under visible light (λ > 420 nm) irradiation excited from the 100 W Xe arc lamp equipped with an ultraviolet light filter. The solution volume of MB is 30 mL and nanorods are grown on the pure glass substrates with a fixed coverage area of 2.0 cm × 2.0 cm for photocatalytic test use. The photoelectrochemical (PEC) properties were measured in a convenient three electrodes electrochemical system. A mixed aqueous solution composed of Na 2 S (0.25 M) and Na 2 SO 3 (0.35 M) were used as electrolyte. Work electrodes were made of various core-shell nanorods on the conductive fluorine-doped tin oxide glasses. Ag/AgCl (1 M KCl) electrode was used as a reference electrode and a platinum wire was used as a counter electrode. The PEC measurements were conducted under visible light irradiation in this study.

Results and Discussion
The surface morphologies of the various ZnO-CdS composite nanorods are shown in Figure 1a-c. Figure 1a displays the smooth surface of the ZnO-CdS (5 nm) nanorods. Moreover, the hexagonal morphology of the ZnO nanorods was maintained when the ultrathin 5-nm-thick CdS layer was coated onto the surfaces of the ZnO rods. By contrast, the ZnO nanorods coated with 20-nm-thick CdS shell layers exhibited increased surface roughness (Figure 1b); the rod surfaces changed to exhibit ruggedness. When the ZnO-CdS (20 nm) nanorods were further postannealed in ambient Ar/H 2 , the rod surfaces became relatively uneven. The surfaces of the composite rods were composed of many irregularly shaped, tiny island-like crystallites, according to an observation of the SEM image ( Figure 1c). reference electrode and a platinum wire was used as a counter electrode. The PEC measurements were conducted under visible light irradiation in this study.

Results and Discussion
The surface morphologies of the various ZnO-CdS composite nanorods are shown in Figure 1a-c. Figure 1a displays the smooth surface of the ZnO-CdS (5 nm) nanorods. Moreover, the hexagonal morphology of the ZnO nanorods was maintained when the ultrathin 5-nm-thick CdS layer was coated onto the surfaces of the ZnO rods. By contrast, the ZnO nanorods coated with 20-nm-thick CdS shell layers exhibited increased surface roughness (Figure 1b); the rod surfaces changed to exhibit ruggedness. When the ZnO-CdS (20 nm) nanorods were further postannealed in ambient Ar/H2, the rod surfaces became relatively uneven. The surfaces of the composite rods were composed of many irregularly shaped, tiny island-like crystallites, according to an observation of the SEM image ( Figure 1c).   Figure 2b; this Bragg reflection was ascribed to the hexagonal CdS (103). When the CdS shell layer thickened, the crystallographic orientation of the CdS crystallites became more randomly oriented. The degree of preferred (101)-crystallographic orientation was evaluated by calculating the peak intensity ratio I101/(I101 + I103) and the background intensity was deducted before the calculation of peak intensity ratio. The peak intensity ratio was reduced to 78% for the thicker CdS shell layer. Furthermore, as displayed in Figure 2c, when the ZnO-CdS (20 nm) nanorods were further postannealed in ambient Ar/H2, the Bragg reflections of the CdS (101) and (103) became more sharp and intense; however, this annealing treatment further engendered an increased randomness of the crystallographic orientation of the 20 nm-thick CdS shell layer and the degree of preferred (101)-crystallographic orientation was decreased to 72%. The change of average crystallite size of thicker CdS shell layers with and without the postannealing procedure was evaluated from the XRD Bragg reflection of (101) according to Scherrer equation [22]. The average crystallite size of the CdS shell layer was increased approximately from 12.1 nm to 16.9 nm after the postannealing procedure. The XRD results revealed that the crystalline quality of the sputtering deposited CdS crystallites was markedly improved after the postannealing procedure. Figure 3a displays a low-magnification TEM image of the morphology of a single ZnO-CdS (5 nm) nanorod. The TEM image shows that the CdS crystallites were homogeneously coated onto the surface of the ZnO core to form a flat and continuous layer-like shell structure. The surface of the CdS shell layer was smooth. Figure 3b,c display high-resolution (HR) TEM images obtained from the edges of the ZnO-CdS nanorod. The clear and highly ordered lattice fringes with an interval of  The XRD results revealed that the ZnO crystals exhibited a high c-axis orientation, and the as-deposited CdS crystallites exhibited a strong (101)-crystallographic orientation. By contrast, a weak Bragg reflection centered at approximately 48.8 • was observed for the thicker CdS shell layer in Figure 2b; this Bragg reflection was ascribed to the hexagonal CdS (103). When the CdS shell layer thickened, the crystallographic orientation of the CdS crystallites became more randomly oriented. The degree of preferred (101)-crystallographic orientation was evaluated by calculating the peak intensity ratio I 101 /(I 101 + I 103 ) and the background intensity was deducted before the calculation of peak intensity ratio. The peak intensity ratio was reduced to 78% for the thicker CdS shell layer. Furthermore, as displayed in Figure 2c, when the ZnO-CdS (20 nm) nanorods were further postannealed in ambient Ar/H 2 , the Bragg reflections of the CdS (101) and (103) became more sharp and intense; however, this annealing treatment further engendered an increased randomness of the crystallographic orientation of the 20 nm-thick CdS shell layer and the degree of preferred (101)-crystallographic orientation was decreased to 72%. The change of average crystallite size of thicker CdS shell layers with and without the postannealing procedure was evaluated from the XRD Bragg reflection of (101) according to Scherrer equation [22]. The average crystallite size of the CdS shell layer was increased approximately from 12.1 to 16.9 nm after the postannealing procedure. The XRD results revealed that the crystalline quality of the sputtering deposited CdS crystallites was markedly improved after the postannealing procedure. Figure 3a displays a low-magnification TEM image of the morphology of a single ZnO-CdS (5 nm) nanorod. The TEM image shows that the CdS crystallites were homogeneously coated onto the surface of the ZnO core to form a flat and continuous layer-like shell structure. The surface of the CdS shell layer was smooth. Figure 3b,c display high-resolution (HR) TEM images obtained from the edges of the ZnO-CdS nanorod. The clear and highly ordered lattice fringes with an interval of approximately 0.33 nm in the CdS shell layer corresponded to the (101) lattice plane. An ultrathin CdS shell layer with a high crystalline quality was formed on the surface of the ZnO core. Figure 3d displays a low-magnification TEM image of the morphology of a single ZnO-CdS (20 nm) nanorod. The thickness distribution of the CdS shell layer over the ZnO core was relatively inhomogeneous compared with that in the ZnO-CdS nanorod with a 5 nm-thick CdS shell layer. Figure 3e,f display the HRTEM images obtained from the outer regions of the ZnO-CdS nanorod. The ordered lattice fringes in the HRTEM images show that the excellent crystalline CdS shell layer covered the ZnO core. Certain portions of the CdS layer still had flat surfaces (Figure 3e). However, partial regions of the CdS shell layer exhibited irregular surfaces ( Figure 3f); boundaries between the CdS crystallites were visible. The energy-dispersive X-ray spectroscopy (EDS) spectra displayed in Figure 3g show that Zn, Cd, O, and S are the main constituent elements of the selected composite rod.          [20,24]. A similar visible-light emission band at approximately 506 nm was also reported for a TiO2-CdS composite rod system [25]. Comparatively, the intensity of NBE emission band of ZnO core decreased and visible-light emission band intensity increased when the CdS shell thickness was increased. This is attributable to the factor that an increased size of surlfide surface defect density is highly associated with the thickness of CdS shell layer and a larger amount of CdS crystallites as a shell layer promoted the photoexcited charge separation efficiency of the ZnO core. The subsequently postannealing procedure for the ZnO-CdS nanorods in ambient Ar/H2 engendered the intensities of the ZnO core-dependent NBE band and visible-light emission band from the CdS shell quenched. The PL analysis result revealed that the charge separation efficiency of the ZnO core and the crystalline quality of the CdS shell were substantially improved when the ZnO-CdS core-shell nanorods were subjected to the postannealing procedure. Figure 5b displays the optical absorbance spectra of the ZnO-CdS nanorods with various CdS shell layers. By comparison, a clear redshift was observed for the ZnO nanorods coated with the CdS shell layers. The room-temperature PL spectra of ZnO-CdS nanorods with various CdS shell layers are displayed in Figure 5a. A UV emission band centered at approximately 378 nm was ascribed to the near-band edge (NBE) emission of the ZnO nanorods [1,23]. Moreover, a clear visible-light emission band centered at approximately 495 nm was observed for the ZnO-CdS nanorods. For the ZnO-CdS composite nanorod system, this visible-light emission band is referred to as the deep-level or trap-state emission band, and is associated with structural defects arising from the sulfur-related point defects of the CdS shell layer [20,24]. A similar visible-light emission band at approximately 506 nm was also reported for a TiO 2 -CdS composite rod system [25]. Comparatively, the intensity of NBE emission band of ZnO core decreased and visible-light emission band intensity increased when the CdS shell thickness was increased. This is attributable to the factor that an increased size of surlfide surface defect density is highly associated with the thickness of CdS shell layer and a larger amount of CdS crystallites as a shell layer promoted the photoexcited charge separation efficiency of the ZnO core. The subsequently postannealing procedure for the ZnO-CdS nanorods in ambient Ar/H 2 engendered the intensities of the ZnO core-dependent NBE band and visible-light emission band from the CdS shell quenched. The PL analysis result revealed that the charge separation efficiency of the ZnO core and the crystalline quality of the CdS shell were substantially improved when the ZnO-CdS core-shell nanorods were subjected to the postannealing procedure. Figure 5b displays the optical absorbance spectra of the ZnO-CdS nanorods with various CdS shell layers. By comparison, a clear redshift was observed for the ZnO nanorods coated with the CdS shell layers. This is because CdS crystallites have an optical bandgap value in the visible-light region (approximately 2.4 eV), which improved light harvesting to the solar lights of the wide-bandgap ZnO [26]. Furthermore, the thicker 20-nm-thick CdS shell layer engendered the ZnO-CdS nanorods with a larger redshift and a broader optical absorbance edge in comparison with the ZnO-CdS nanorods containing a 5-nm-thick shell layer. Thicker CdS crystallites have a rougher surface; this broadened the optical absorbance edge of ZnO-CdS with a thicker CdS shell layer [4]. The broadness of optical absorbance edge of the ZnO-CdS nanorods was further widened when the layer-like CdS shell was converted into the serrated CdS shell. This might be attributed to the serrated CdS shell layer is more favorable for light harvesting because of multiple light reflections and scattering [4]. This is because CdS crystallites have an optical bandgap value in the visible-light region (approximately 2.4 eV), which improved light harvesting to the solar lights of the wide-bandgap ZnO [26]. Furthermore, the thicker 20-nm-thick CdS shell layer engendered the ZnO-CdS nanorods with a larger redshift and a broader optical absorbance edge in comparison with the ZnO-CdS nanorods containing a 5-nm-thick shell layer. Thicker CdS crystallites have a rougher surface; this broadened the optical absorbance edge of ZnO-CdS with a thicker CdS shell layer [4]. The broadness of optical absorbance edge of the ZnO-CdS nanorods was further widened when the layer-like CdS shell was converted into the serrated CdS shell. This might be attributed to the serrated CdS shell layer is more favorable for light harvesting because of multiple light reflections and scattering [4].   The photocatalytic performance of the ZnO and ZnO-CdS nanorods with various CdS shell layers under visible-light irradiation was further investigated. The time-course photodegradation curves for an aqueous MB solution with various rods samples are displayed in Figure 6a-c. The visible and intense peaks of the absorbance spectra at approximately 663 nm were due to monomeric MB. The ratio of the remaining MB concentration (C) after visible-light irradiation to the initial MB concentration without visible-light irradiation (C o ) i.e., C/C o was used to determine the photodegradation degree of the MB solution containing various rod samples. Therefore, the C/C o value could be evaluated from the absorbance spectra intensity ratio at 663 nm before and after the MB solution was subjected to visible-light irradiation with various nanorod samples. Figure 6d shows the absorbance spectra of the MB solution containing various rod samples at the irradiation duration of 180 min in the wavelength range from UV to visible region. The decrease in the absorbance peak intensity in the UV region is associated with the mineralization of the MB dye during the photodegradation process. To evaluate the influence of MB amount adsorbed on photocatalytic activity, dark adsorption tests were performed at different durations (90 and 180 min). Figure 6e      and 0.58 mA/cm 2 , respectively, under visible-light illumination. The reduction-oxidation reactions in the electrolyte during the PEC measurements involved the electron-hole pairs that presented in the CdS shell layers upon exposure to visible-light irradiation (Figure 7c). The photoexcited holes remained in the valence band of the CdS; however, the CdS shell layers inject the photoexcited electrons from its conduction band into that of the ZnO nanorods. The injected electrons flowed through the ZnO nanorods, and reached the F-doped tin oxide layer [29]. The band alignment between the ZnO-CdS in this study was similar to that of the TiO 2 -CdS and TiO 2 -CdSe heterostructure systems, and these heterostructures have exhibited improved charge injection efficiency and decreased charge recombination loss compared with the constituent single counterpart under light irradiation [30]. For the ZnO-CdS nanorod system in this study, the composite rods with a thicker CdS shell layer exhibited superior PEC performance. This is associated with the composite rods with a thicker CdS shell layer being larger in visible-light harvesting; moreover, as the shell layer, larger CdS crystallites might provide a higher number of photoexcited carriers under light irradiation [31]. These factors enabled the ZnO-CdS nanorods with a thicker CdS shell layer to exhibit a higher photocurrent density compared with those with a thinner CdS shell layer. A similar shell layer crystallite-size effect on the photocurrent density was reported in oxide composite rods with two substantially different bandgap values [4]. By contrast, the ZnO-CdS nanorods conducted with the postannealing procedure exhibited a superior PEC performance compared with the ZnO-CdS nanorods without the postannealing procedure under the same effective CdS shell layer thickness. Based on the aforementioned structural analyses, this is attributable to the higher efficiency in visible-light harvesting in ZnO-CdS nanorods associated with the substantial microstructure and crystalline quality differences among the nanorod samples.
(20 nm), and ZnO-CdS nanorods conducted with the postannealing procedure at 0.3 V were approximately 0.23, 0.38, and 0.58 mA/cm 2 , respectively, under visible-light illumination. The reduction-oxidation reactions in the electrolyte during the PEC measurements involved the electron-hole pairs that presented in the CdS shell layers upon exposure to visible-light irradiation (Figure 7c). The photoexcited holes remained in the valence band of the CdS; however, the CdS shell layers inject the photoexcited electrons from its conduction band into that of the ZnO nanorods. The injected electrons flowed through the ZnO nanorods, and reached the F-doped tin oxide layer [29]. The band alignment between the ZnO-CdS in this study was similar to that of the TiO2-CdS and TiO2-CdSe heterostructure systems, and these heterostructures have exhibited improved charge injection efficiency and decreased charge recombination loss compared with the constituent single counterpart under light irradiation [30]. For the ZnO-CdS nanorod system in this study, the composite rods with a thicker CdS shell layer exhibited superior PEC performance. This is associated with the composite rods with a thicker CdS shell layer being larger in visible-light harvesting; moreover, as the shell layer, larger CdS crystallites might provide a higher number of photoexcited carriers under light irradiation [31]. These factors enabled the ZnO-CdS nanorods with a thicker CdS shell layer to exhibit a higher photocurrent density compared with those with a thinner CdS shell layer. A similar shell layer crystallite-size effect on the photocurrent density was reported in oxide composite rods with two substantially different bandgap values [4]. By contrast, the ZnO-CdS nanorods conducted with the postannealing procedure exhibited a superior PEC performance compared with the ZnO-CdS nanorods without the postannealing procedure under the same effective CdS shell layer thickness. Based on the aforementioned structural analyses, this is attributable to the higher efficiency in visible-light harvesting in ZnO-CdS nanorods associated with the substantial microstructure and crystalline quality differences among the nanorod samples.

Conclusions
For this study, the ZnO-CdS core-shell composite nanorods were synthesized by sputtering CdS crystallites onto hydrothermally derived ZnO rods. The ZnO-CdS composite nanorods with two CdS shell thicknesses of approximately 5 and 20 nm were fabricated. Microstructural analyses revealed that the CdS crystallites covered the surfaces of the ZnO nanorods homogeneously and formed a layer-like structure. The ZnO-CdS (20 nm) composite nanorods exhibited a rougher surface and more surface crystal defects compared with the ZnO-CdS (5 nm) nanorods. These differences engendered the ZnO-CdS (20 nm) nanorods with higher photoactivity and PEC performance. Further postannealing the ZnO-CdS (20 nm) nanorods in ambient Ar/H 2 engendered the morphology change of the layer-like CdS shell to a serrated CdS shell layer. In this study, the photoactivity and PEC performance of the ZnO-based nanorods was further enhanced through a postannealing procedure, because of improvements in visible-light harvesting and crystalline quality of the CdS shell layer in the ZnO-CdS nanorods.