Simulation

Gold nanorods (εm = -4.47643 + 2.53177i), Zinc Oxide (ZnO) (εd = 1.95908) and the surrounding material (H2O) (n = 1.33) were utilized and carried out electric

160

field intensity by calculating three dimensional Maxwell equations with commercial solver COMSOL Multiphysics^TM 3.5a, which is based on finite element method (FEM). All of the boundaries were set as periodic boundary conditions such that the layers were matched to a large number of gold nanoparticles in water. The finite-difference time-domain (FDTD) program, MEEP, has also been used to calculate induced heat from gold nanoparticles. In this simulation, all of the boundary conditions were set such that the layers were perfectly matched to eliminate interference of refracted waves with the incident light. The x, y, and z dimensions of the grid cells are all 1 nm. The boundary conditions are all set as perfect.

161

5.3 Results and Discussion

In this study, an array of ZnO nanorods was prepared with various amounts of plasmonic gold nanospheres as a platform to examine the plasmon-induced effects on a photoelectrode in the splitting of water. One-dimensional nanorods have been demonstrated to be efficient in photoelectrochemical (PEC) cell and photovoltaic cell applications because they can decouple the direction of light absorption and charge carrier collection.^[24-26] Nanorods have a small radius, and minority carriers that are generated therein can diffuse to their surfaces before they recombine. This effect increases charge separation efficiency, especially when the minority carrier diffusion length is comparable to the radius of the nanorod.

Additionally, the shape of a plasmonic gold nanostructure has been demonstrated to be easily controlled using numerous methods and the also exhibit strongly shape-dependent optical properties, meaning that present investigation can considerably operate in other systems. An array of ZnO nanorods was synthesized as previous chapter. Figure 6-1 characterizes the nanostructures of an ZnO@Au electrode with plasmonic gold nanoparticles that had been deposited for various duration. As the duration of deposition increases to 24 h, the morphology of ZnO nanorods does not significantly change. Notably, the surface of ZnO that had been deposited with Au nanoparticles for 24 h was slightly rough owing to the formation of a thin layer of Au nanoparticles upon it. Since the particles size were very small (less than 5 nm), further structural characterization was conducted using a transmission electron microscope.

162

Figure 5-1. SEM images of ZnO@Au photoelectrodes with nanoparticles deposited for various periods: (a) bare ZnO, (b) 1 h deposition, (c) 3 h deposition, (d) 6 h deposition, (e) 12 h deposition and (f) 24 h deposition.

163

Localized surface plasmon resonances, which are caused by the interaction of incident light with gold nanoparticles, are well known to be effective in for biological sensors, surface-enhanced spectroscopy, and optical devices ^[27,28], because wavelength of plasmon resonance depends strongly on the structure and composition of the material, and the local dielectric environment. Gold nanostructures serve as a platform in this investigation, and are used to reveal plasmon-induced effects on a semiconductor, because their resonant wavelength is in visible region rather than in the ultraviolet, and so overlap with the wavelengths absorbed by of a wide bandgap semiconductor (ZnO). Optical spectra of ZnO@Au photoelectrodes on which had been deposited with Au nanoparticles for various times exhibited a clear increase in absorbance corresponding to the gold nanospheres in addition to strong ultraviolet absorption of ZnO (Figure 5-2a).

As deposition time increased, the absorbance that originated in the surface plasmon resonance of gold nanoparticles (at approximately 530 nm) considerably increased, suggesting that the gold nanoparticle loading could be effectively controlled by varying the deposition duration and conditions, enabling the conditions of the photoelectrochemical reaction to be optimized. Compared to the absorption spectrum of Au nanoparticles in aqueous solution (Figure 5-2b), the deposition Au nanoparticles on the ZnO surfaces were accompanied by a red shift in a surface plasmon peak continuously toward red region. The red shift from 522 to 529 nm is attributable to the progressive depositing of Au particles upon the ZnO. This peak shift may have been caused by the dielectric variance of surrounding environment and inter-particle distance upon gold nanoparticles.^[29]

Table 5-1 shows the elemental analysis of ZnO@Au photoelectrodes from inductively coupled plasma atomic emission spectrometer, which reveals that

164

chemical bath deposition can modify the gold nanoparticles upon the surface of ZnO and significantly control the loading amount of plasmonic materials. The absorption spectra reveal that gold nanoparticles were successfully attached to the surface of the ZnO nanorods array during chemical bath deposition, but not by physical methods, since the deposition of a monolayer can clearly reveal localized surface plasmon effects without causing any undesired effects. Notably, the absorption spectrum obtained at a deposition time of 24 hours included a peak at approximately 650 nm in addition to the major plasmon resonance peak. This phenomenon is attributable to the formation of clusters aggregation that was composed of gold nanoparticles.^[29] nanoparticles deposited over various periods and (b) pure Au nanoparticles suspension.

165

Electrochemical measurements were made systematically to evaluate the photoelectrochemical properties of ZnO nanowires that were loaded with Au nanoparticles. Figure 5-3a shows a set of linear sweep voltammagrams of pristine ZnO nanorods and ZnO nanorods that were loaded with Au nanoparticles that had been deposited for various durations under 100 mW/cm² of illumination. A dark scan from -0.5 to +1.1 V revealed a small current of around 10^-6 A/cm². The photocurrent increased dramatically with deposition time to 12 h. ZnO nanorods with loading of 12 h yielded a pronounced photocurrent under illumination that started when -0.25 V was applied, and increased to 1.3 mA/cm² when +1.0 V was applied. The photocurrent density in ZnO nanorods that were loaded with Au nanoparticles (12 h) was about double that of pristine ZnO nanowires with a similar thickness (~0.7 mA/cm²) at 1.0 V, suggesting that decoration by Au nanoparticles promoted the harvesting of solar light. As the deposition time was further increased, an important effect occurred. Most interestingly, the photocurrent declined greatly as the deposition time increased to 24 h. This result is particularly interesting. This phenomenon indicates that several effects originated from either the localized surface plasmon or the collection of sunlight, which involved in the photoelectrochemical reaction on the ZnO@Au photoelectrode. Although the more loading amount of Au nanoparticles can lead to an increase in hot electrons injection and contribute a detectable photocurrent to entire device (as shown in Figure 5-3b), the ZnO@Au sample with 24 h of deposition exhibited a great decrease in its photocurrent. This may be attributed to the blocking effect of gold nanoparticle upon the ZnO nanorods, since the loading amount of 24 h sample was almost twice as much as that of 12 h depositon sample, indicated that Au nanoparticles upon the ZnO would block two times of irradiation

166

from solar simulator and suppress photoexcitation process on ZnO that was a UV active semiconductor. Therefore, metallic gold may also play as a trap centers for photoelectrons and collect some photogenerating electron from ZnO nanorods, this phenomenon results in a negative effect in photoresponse. As a result, the sample with 12 h of deposition (~10 % loading) was expected to be an optimum condition for both hot electron injection and light absorption of semiconductor support.

Figure 5-3. (a) Linear-sweep voltammograms of ZnO@Au photoelectrodes with nanoparticles deposited for various periods, and a dark scan, performed in a 0.5 M aqueous Na2SO4 with a pH of 6.8 under an AM 1.5 solar simulator. (b) Linear-sweep voltammograms of ZnO@Au photoelectrodes with nanoparticles deposited for various periods, obtained under illumination by visible light (> 420 nm).

To explore hot electron generation and plasmon damping, an ZnO@Au photoelectrode was irradiated with visible light (>420 nm) to eliminate the effects of the photoexcitation of ZnO. ZnO nanorods with a large bandgap of above 3 eV are a UV-photoactive material, such that the measured photoresponse under

167

illumination by visible light (> 420 nm) is attributable largely to plasmonic Au nanoparticles. Figure 5-3b presents a set of linear sweep voltammagrams of pristine ZnO nanorods and ZnO nanorods that were loaded with Au nanoparticles that had been deposited for various durations under illumination by visible light (>420 nm) with an intensity of 100 mW/cm². Pristine ZnO nanorods generate no obvious photocurrent because ZnO is inactive in the visible region. When few Au nanoparticles decorate the array, the photocurrent depends substantially on the number of Au nanoparticles, reaching a maximum of 0.3 mA/cm². Further increasing the number of Au nanoparticles cannot significantly increase its photocurrent, suggesting that optimal decoration had been achieved and that further increasing the number of Au nanoparticles would not increase the measured photoresponse of the PEC by contributing more photoelectrons. This result is consistent with the I-V measurements under full-spectrum solar illumination and reveals that excess Au nanoparticles could not significantly contribute to the photoresponse, but had a blocking effect that reduced the absorption of light by the ZnO nanorods. The photocurrent density of an ZnO@Au photoelectrode (12 h) increased with applied voltage, starting at -0.25 V versus Ag/AgCl, and reaching 0.3 mA/cm² at 1.0 V. It was similar to that of the ZnO@Au photoelectrode under full-spectrum solar simulator (Figure 5-3a). We suggest that the hot electrons that are generated by plasmon damping under visible illumination can be injected into the ZnO over the Schottky barrier, yielding a detectable photocurrent under visible illumination. Notably, plasmonic enhancement is attributable to the injection of hot electrons from plasmonic materials into the conduction band of the semiconductor (ZnO) rather than to the trapping of photogenerating electrons from the conduction band of the semiconductor and

168

their transfer and subsequent reaction with hydrogen ions. The transfer of photogenerated electrons from the conduction band of the semiconductor to the metal is energetically unfavorable because the photoelectrons must overcome the Schottky barrier even though this barrier is small.

Because of shape/size-dependent nature for gold nanostructure, Figure 5-4a displays the TEM image of as-prepared Au nanoparticles, which are small, with a mean diameter of 4.7 nm with a standard deviation of 0.7 nm. The TEM image (Figure 5-4b) of ZnO nanorods that are decorated with an ensemble of Au nanoparticles (12 h) reveals that these particles are uniformly distributed upon the surface and uniform in diameter. The distribution of Au nanoparticles on the surface of the ZnO nanorod may be optimal because more Au nanoparticles would have had a negative effect, blocking ZnO nanorods from absorbing solar illumination, suppressing the photogeneration process of ZnO under ultraviolet irradiation. A high-resolution TEM image of the edge of a nanorod (Figure 5-4c) provides more compelling evidence that Au nanoparticles are attached to its surface. An abrupt transition is observed between the lattice planes of the ZnO nanorod and the (111) lattice planes of the Au nanoparticles. The lattice spacing between the (111) planes, 0.235 nm, is also consistent with that of the gold bulk crystal (JCPDS no. 89-3697). Most interestingly, the selected electron diffraction pattern (Figure 5-4d) is characteristic of the two component crystalline nature. The set of spot pattern can be indexed to the [2110] zone axis of the ZnO wurtzite structure zone axis, which is a single crystalline structure (indicated by the white circle). The set of rings reveals a typical face-centered-cubic polycrystalline structure that corresponds to bulk gold and is probably associated with the large amount of Au nanoparticles on the surface of the ZnO nanowires. These results

169

indicate that Au nanoparticles were successfully attached to the surfaces of ZnO nanowires. The following section will systematically address the effects of localized surface plasmon upon the ZnO nanorods and their relationship to the water splitting reaction.

Figure 5-4. (a) TEM micrographs and distribution of sizes of as-prepared gold nanoparticles (12 h). (b) TEM micrographs and (c) high-resolution image of ZnO@Au photoelectrodes. (d) Corresponding electron diffraction pattern of ZnO@Au photoelectrode.

170