ZnO:N@CdTe Photoelectrochemical Cell

We further introduced nitrogen into the ZnO nanowire array (ZnO:N). The synthesized ZnO:N nanowire arrays were annealed in ammonia at 500 °C to 700°C for 30 min and then in nitrogen for another 30 min. CdTe QDs were deposited on the ZnO:N nanowire array by chemical-bath deposition, followed by thermal treatment to remove the linker between the QDs and the ZnO:N nanowires.

Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) were used to characterize the specific nanostructures of the ZnO:N and ZnO:N@CdTe samples. The SEM images show the compact and vertically aligned ZnO nanowires after annealing in ammonia at 500°C to 700°C, the nanostructures of ZnO nanowire array was unchanged (Figures 3-10a to 3-10c).

Figure 3-10. Scanning electron microscopy (SEM) images of top views of ZnO.

SEM images of ZnO:N at annealing temperatures of (a) 500, (b) 600 and (c) 700°C.

Figure 3-11 display the HRTEM image and the absorption spectrum of prepared CdTe QDs in the solution. Figure 3-11b shows the absorption spectrum of CdTe QDs were prepared by wet-chemical processes. The prepared CdTe QDs

98

were capped by the Mercaptopropionic acid (MPA) ligands which could help the CdTe QDs disperse in water solution and self-assembling on the ZnO nanowires surface. The inset displays photographs of CdTe QDs solution under ambient and UV light with a wavelength of 365 nm, which exhibit highly dispersity in water solution. The solution absorbs in the visible region with an onset at ~720 nm, implying that the use of the solution as a sensitizer enables a wider range of visible wavelengths in sunlight to be harvested. Figure 3-11a shows a typical HRTEM micrograph of the CdTe QDs. The size of the nanoparticles is around 4~6 nm, were well separated and roughly spherical with clear lattice fringes. The average particle size obtained from the images was 5.3 + 0.8 nm. The particles were oriented along the (311) axis in the plane of the images with a lattice spacing of 0.20 nm. This value is consistent with that of CdTe bulk crystal (JCPDS file no. 89-3053). The CdTe quantum dots were then deposited on the surface of the ZnO nanowires-array for further structural and photoelectrochemical measurement.

Figure 3-11. (a) High-resolution transmission electron microscopy (HRTEM) image of the CdTe QDs. (b) The absorption spectrum of the CdTe QDs suspensions in water.

The inset displays photographs of CdTe QDs solution under ambient and UV light (365 nm).

99

X-ray diffraction (XRD) studies were conducted to examine the structural properties of the ZnO:N nanowires (Figure 3-12). The samples prepared at different temperatures yielded similar diffraction patterns. Pattern indexing showed that all diffraction peaks are consistent with the wurtzite ZnO structure with lattice constants of a = 3.250 Å and c = 5.207 Å. The high-intensity (002) diffraction peaks in the patterns indicate that the reflections from the (002) plane of the ZnO nanowires are stronger than those from other planes because of the [00l]-oriented nanowire growth. The lattice constants and phase changes of the ZnO and ZnO:N nanowires show no significant difference, indicating that N doping has no significant effect.

Figure 3-12. X-ray diffraction (XRD) patterns of ZnO, ZnO:N (500), ZnO:N (600) and ZnO:N (700).

To investigate the N incorporation in the ZnO:N nanowires, X-ray photoelectron spectroscopy (XPS) was performed to determine the quantitative concentration and the chemical state of N. The results are shown in Figure 3-13.

A plot of the N concentration in the ZnO:N nanowires versus the ammonia

100

annealing temperature is shown in Figure 3-13a. To prevent errors in the oxygen concentration measurements of the ZnO nanowires on the FTO substrate, all XPS samples were prepared on a silicon substrate. As the annealing temperature is increased from 500°C to 700°C (Figures 3-13 b-d), the N concentration slightly increases from 1.0 wt% to 4.0 wt%. The diffusion distance of nitrogen increases with increasing annealing temperature, subsequently increasing the doped N concentration. High-resolution XPS studies were then performed to identify the chemical state of the N dopant in the ZnO:N nanowires (Figure 3-13b). The core level spectrum of the N 1s region shows an asymmetric broad peak centred at 396.3 eV. All experimental line profiles show two peaks centred at 396.3 and 399.6 eV. The peak centred at 396.3 eV is attributed to typical Zn–N bonds, thus confirming the successful N doping in the ZnO crystal structure.^[2,10] The peak centred at 399.6 eV is characteristic of the N 1s binding energy in amines.

Therefore, N is successfully doped at the O sites of ZnO during the nitridation reaction.

101

Figure 3-13. (a) X-ray photoelectron (XPS) plot of the ZnO:N nanowire nitrogen concentration versus the annealing temperature. High-resolution XPS of (b) ZnO:N (500), (c) ZnO:N (600) and (d) ZnO:N (700).

UV-Vis absorption spectroscopy was conducted to investigate the optical changes in the ZnO:N samples. The results are shown in Figure 3-14. The absorption edge of the pristine ZnO nanowires is at 380 nm. Ammonia annealing at 500°C to 700°C produced a pale yellow ZnO:N nanowire electrode and caused a red shift of the absorption edge to 550 nm. The broader absorption in the 400 nm to 550 nm range is attributed to the observable changes in the band structure of the ZnO nanowires as a result of nitridation. The first excitonic peak of the CdTe QDs at 690 nm was not exhibited by the ZnO:N nanowires, possibly improving absorption in the visible light region. The UV-Vis absorption spectroscopy of the

102

ZnO:N@CdTe samples exhibit almost same curvature from 550 nm to 800 nm are shown in the Figure 3-14b. The increase absorption in the visible light region of the ZnO:N@CdTe as compared with ZnO:N samples is contributed from the decorated CdTe QDs on the ZnO nanowires surface. Nitridation of the ZnO nanowires with CdTe QDs sensitization induces efficiently harvests solar light over a wider range of wavelengths.

Figure 3-14. UV-vis absorption spectra of (a) ZnO, ZnO:N (500), ZnO:N (600) and ZnO:N (700) nanowires. (b) ZnO:N (700), ZnO:N (500)@CdTe, ZnO:N (600)@CdTe and ZnO:N (700)@CdTe nanowires.

As the proof of concept, the photocurrent measurement for the photoactivity under light illumination, PEC studies were performed by using 0.5 M Na2SO4 (pH

= 6.8) as supporting electrolyte medium. Figure 3-15a displays a set of linear-sweep voltammograms that were recorded on pristine ZnO nanowires and ZnO:N nanowires array. ZnO:N (700) nanowires array showed a pronounced photocurrent starting at ~ –0.2 V which increased to 0.31 mA/cm² at 0.5 V under illumination.

The photocurrent density of the ZnO:N nanowire array was higher (~0.31 mA/cm²) than that of the pristine ZnO nanowires (~0.15 mA/cm²) at 0.5 V, suggesting that

103

the synthesized ZnO:N is more efficient in harvesting solar light and converting it to electricity compared with the pristine nanowires. The two most important metrics associated with photocurrent are the plateau current and the onset potential.^[29,33] The plateau current depends mainly on the photogenerated holes/electrons that reach the semiconductor/liquid junction and, instead of recombining, subsequently react with water. The overpotential must be considered in determining the onset potential. A large overpotential is caused mainly by the slow kinetics of water oxidation and results in the accumulation of hole-electrons on the surface. Subsequent surface recombination occurs until sufficiently positive potentials are applied. Modification of the electrode surface lowers the kinetic barrier to interfacial charge-transfer, thus reducing the required overpotential and shifting the curve to the left. The onset potentials of the ZnO:N nanowires prepared at different annealing temperatures are more negative than that of pristine ZnO.

These samples exhibited identical behaviour, suggesting that their surfaces are similar. Notably, the plateau current of the ZnO:N annealed at 500°C is nearly equal to that of the pristine ZnO nanowires, indicating that a low dopant concentration does not significantly improve the plateau photocurrent. The improvement in the plateau current is caused by the formation of the ZnO:N nanowires, indicating that the ZnO:N nanowires can generate more photoelectrons by harvesting a higher amount of sunlight. To improve the PEC performance, we sensitized the ZnO:N nanowires with CdTe QDs, which could extend the absorption region of the photoelectrode to allow the harvesting of more visible light so as to generate photoelectrons. The set of the linear-sweep voltammograms of the ZnO:N@CdTe photoelectrode is shown in Figure 3-15b. After sensitization, the onset potential slightly shifted to a more negative potential. The ZnO:N@CdTe construct heterojunction increased the charge-transfer efficiency, thus reducing the

104

overpotential. The plateau current of the ZnO:N@CdTe is higher (~0.46 mA/cm²) than that of the pristine ZnO (~0.15 mA/cm²) and ZnO:N nanowire array. This increase is attributed to the photoelectrons generated by the CdTe QDs. We performed amperometric I–t studies to investigate the photoresponses of ZnO:N and ZnO:N@CdTe over time. The plot of efficiency against applied potential (Figure 3-15c) revealed a maximum efficiency of ~ 0.75%, which is obtained at an applied potential of +0.5 V. Importantly, the ZnO nanowires that were sensitized using CdTe QDs solution were over two times as efficient as bare ZnO nanowires, with a typical photoconversion efficiency of 0.28%. The photoconversion efficiency of the ZnO:N(700)@CdTe is more efficiency than the ZnO:N(700) photoanode, which could be considered as the contribution of the quantum dots sensitization. To quantify the photoresponse of ZnO:N@CdTe photoanodes, incident-photon-to-current-conversion efficiency (IPCE) measurements were made to examine their photoresponse as a function of incident light wavelength (Figure 3-15d). The ZnO:N nanowires that were exhibited substantially greater IPCE than bare ZnO nanowires in both the visible and UV regions, due primarily to the increase in light absorption by the nitridation. The sample of ZnO:N nanowires that were sensitized with CdTe QDs exhibited photoactivity over a broader range of wavelengths, from 430 to 660 nm, because of the nitridation and the sensitization with an IPCE value of ∼4%. At the same incident wavelength (430-660nm), the higher IPCE of the ZnO:N(700)@CdTe composite revealed that it was more efficient than bare ZnO in separating and/or collecting photoexcited electrons in the visible region, which finding is consistent with the larger potential difference between the conduction bands of CdTe and ZnO. Compared with the absorption spectrum of ZnO:N(700)@CdTe, it is worth noting that the photoanode doesn’t have any photoactivity in the wavelength longer than the 660 nm. This

105

phenomenon may owing to the photoenergy in the longer wavelength is not enough to overcome the overpotentail to drive the water splitting reaction. The I–

t curves of the ZnO:N@CdTe samples with cut-off light cycles at 100 mW/cm² at 0.5 V are shown in Figure 3-15e. Extremely low dark currents at 10⁻⁷ mA/cm² were observed during the measurement. The photocurrent intensities of ZnO:N and ZnO:N@CdTe show a pile in the photoresponse, which is caused by the transient effect of power excitation. The photocurrent then rapidly returned to the steady state, indicating an efficient photoelectron transfer. The I–t curves of all samples show that the photocurrent does not decrease with increasing measurement time, thus confirming rapid electron transport and photoelectrode stability. To demonstrate the occurrence of the water-splitting reaction under simulated solar light illumination, the gas evolution of PEC was measured by using a two-electrode system with 0.5 V bias under a solar simulator with power density of 100 mW/cm² (Figure 3-15f). Approximately 20.8 and 6.8 mol·h⁻¹ of H2 and 10.2 and 3.3 mol·h⁻¹ of O2 were produced when the ZnO:N(700°C)@CdTe and pristine ZnO photoelectrodes were used, respectively, indicating that water splitting was more efficient on the ZnO:N@CdTe photoelectrode. The slightly decreased oxygen evolution after 2 h of measurement may have contributed to the low Faraday efficiency. These findings show that the combination of N-doped ZnO nanowires and CdTe QD sensitization can improve the efficiency of PEC water splitting.

106

Figure 3-15. Photoelectrochemical (PEC) performance measurement in 0.5 M Na2SO4 under 100 mW/cm². Linear-sweep voltammograms of (a) ZnO:N and (b) ZnO:N@CdTe. (c) photoconversion efficiency of the bare ZnO, ZnO:N and ZnO:N@CdTe (d) Measured IPCE spectra of bare ZnO, ZnO:N and ZnO:N@CdTe.

(e) Chronoamperometric measurement of ZnO:N@CdTe. (f) Gas evolution of the ZnO and ZnO:N(700)@CdTe nanowire photoelectrodes (100 mW/cm² solar simulator in Na2SO4 as electrolyte).

To investigate the electronic state of the ZnO:N nanowires, the O K-edge of

107

all ZnO:N nanowires was determined. The results are shown in Figure 3-16. The white-line peaks correspond to the O 1s to O 2p transition, which indicates that N was incorporated into the ZnO nanostructure and increased the valence state of oxygen. The increased oxygen valence state results in increased the adsorption of water on the ZnO surface for the water-splitting reaction ^[34,35], thus improving the water-splitting performance.

Although XPS has demonstrated the N incorporation in ZnO nanostructures, little structural information from the viewpoint of zinc is revealed. Extended X-ray absorption fine structure (EXAFS) is local correlations around the absorbing atom with a short-range probe of structure and yielded results concerning, specifically the nearest neighbour interatomic distances and coordination numbers.^[26,27] EXAFS can obtain more convinced evidence of the structural parameters of ZnO:N nanowire array. The 01C1 beam line of the National Synchrotron Radiation Research Center (NSRRC), Taiwan, was designed for such experiments. Structural parameters from each spectrum are obtained by EXAFS refinement (Table 3-1). The EXAFS spectra of Zn K-edge were adopted to determine the structural parameters based on a two-shell model that involves Zn-O, Zn-N, and Zn-Zn shells to characterize the short-range structure around Zn atoms. Figure 3-17 shows the Zn K-edge of ZnO and ZnO:N nanowires array. In pristine ZnO nanowires, these results suggest that the interatomic distance scattered from the first nearest neighboring O atoms and the second nearest neighboring Zn atoms are ~1.9 Å and 3.2 Å, respectively. A strong peak at ~1.97 Å was observed in the Fourier transform (FT) of the Zn K-edge EXAFS spectrum of the ZnO nanowires with phase correction, suggesting that central Zn atoms were surrounded by O atoms in first shell scattering. Another strong peak at ~ 3.2 Å was obtained with phase correction, indicating that the second shell around the Zn

108

atoms included neighboring atoms of Zn atoms. The coordination number (CN) of Zn-O (2.5) and Zn-Zn (11.8) paths were similar to theoretical values (Zn-O: 4 and Zn-Zn: 12). This is consistent with the value for the bulk wurtzite phase of ZnO.

In the case of ZnO:N nanowire array, there is another peak at ~ 1.5 Å was obtained with phase correction scattered from first shell atoms , which indicated that the first shell around the Zn atoms may affect by incorporation of N atom and include neighbouring atoms of two elements (O and N). It was worth noting that second shell scattering (~ 3.2 Å) around central Zn atom were identical in both of ZnO and ZnO:N nanowires array, this could be attributed to the existence of only Zn atoms in second scattering shell. The CN of Zn-N nanowires increase with increasing annealing temperature, subsequently increasing the doped N concentration. This demonstrated that ZnO:N nanowire array exhibited wurtzite crystal structure, which is able to provide the additional potential advantage of improved charge transport over traditional doping approach, which results in a dramatic increase in the plateau current.

Figure 3-16. X-ray absorption near-edge structure measurement of the ZnO:N nanowire O K-edge.

520 530 540 550 560

Normalize Intensity (a.u.)

Binding Energy (eV)

ZnO:N (500) ZnO:N (600) ZnO:N (700) ZnO

109

Figure 3-17. EXAFS spectra of Zn K-edge for ZnO, ZnO:N (500), ZnO:N (600) and ZnO:N (700) nanowires.

Table 3-1. Zn K-edge EXAFS structural parameters of ZnO, ZnO:N (500), ZnO:N (600) and ZnO:N (700) nanowires.

sample Path R(Å) CN σ² (Å ²) △E (eV) ZnO Zn-O 1.97(3) 2.5(2) 0.0035(5) 5.7(3)

Zn-Zn 3.23(2) 11.8(5) 0.0104(2) 1.0(5) ZnO:N (500)

Zn-O 1.97(6) 2.2(4) 0.0036(4) 6.6(7) Zn-N 1.49(5) 0.3(3) 0.0020(3) -7.5(6) Zn-Zn 3.21(5) 12.1(7) 0.0106(5) 0.4(3) ZnO:N (600)

Zn-O 1.98(5) 2.9(2) 0.0027(3) 7.9(4) Zn-N 1.53(3) 0.8(3) 0.0057(6) 0.5(8) Zn-Zn 3.21(3) 12.5(6) 0.0105(2) -1.1(7) ZnO:N (700)

Zn-O 1.99(2) 2.8(1) 0.0024(3) 11.1(2) Zn-N 1.52(4) 1.2(3) 0.0101(3) 18.4(2) Zn-Zn 3.20(2) 12.3.(9) 0.0099(4) -2.9(4)