1. Introduction

Since the first report on ultraviolet lasing from ZnO nanowires^[1], remarkable effort has been dedicated to the development of novel one-dimensional (1D) ZnO nanostructures. As one of the typical 1D ZnO nanostructures, ZnO nanowires have great potential in electronic and optoelectronic device applications, such as piezo-nanogenerators^[2], electroluminescent devices^[3], field-emission devices^[4], and solar cells^[5].

At present, ZnO nanowires can be synthesized through various techniques, including physical vapor deposition^{[6, 7]}, chemical vapor deposition^{[8, 9]}, laser ablation^[10], and the solution method^[11]. Because the solution method has the merit of low temperature, large scale, and economical synthesis, it has attracted the interest of numerous researchers worldwide. As with the previous solution methods, which derive from a seed layer, it usually requires the preparation of a layer of uniform and high-quality ZnO film. There are two main ways to prepare such a ZnO seed layer: sputtering and wet coating. The former requires expensive equipment and complex conditions^{[12, 13]}. The latter usually requires several cycles of wet coating and annealing which restricts the reliability and reproducibility of the fabrication process of ZnO nanowires^[14].

In this work, we presented a simplified approach to prepare a ZnO seed layer on FTO substrate through one-time spin-coating followed by annealing. Long and well-aligned ZnO nanowires were synthesized based on this ZnO seed layer. The effect of refreshing the growth solution on the morphology of ZnO nanowires had been investigated and the possible mechanism was also discussed. The application of ZnO nanowires with various growth cycles into dye-sensitized solar cells showed that the prepared ZnO nanowires possessed excellent photovoltaic performance.

2. Experiment

ZnO nanowires were grown on ZnO-seeded fluorine-doped tin oxide (FTO) (Nippon sheet glass, 14 $\Omega$/$\square$) substrates by hydrothermal method. As shown in Fig. 1, the fabrication process was as follows: firstly, the FTO glass was rinsed ultrasonically in acetone, isopropanol and deionized water each for 10 min, successively. Then it was blown dry with nitrogen gas. Secondly, the ZnO seed layer (SL) was prepared by spin-coating and annealing of the sol seed solution which was prepared by dissolving 4 wt% polyvinyl alcohol (PVA) and 1 wt% zinc acetate dihydrate in 5 mL DI water. Next, conventional photolithography was conducted to pattern the ZnO SL in order to define the active area. Finally, the substrate was suspended upside-down in the growth solution for growing ZnO nanowires. The growth solution contained 50 mL deionized water, 25 mM zinc nitrate hexahydrate (Zn(NO$_{3})$$_{2}$$\cdot$6H$_{2}$O), 12.5 mM hexamethylenetetramine (HMTA), 0.15 g polyethyleneimine (PEI) and 2.2 mL ammonia (NH$_{4}$$\cdot$H$_{2}$O, 28%–29%). The growth solution was refreshed every 3 h and the growth temperature was set at 95~ ℃ in the oven. For comparison, ZnO nanowires were also prepared using the growth solution without refreshing regularly. To modify the surface condition and crystallinity of ZnO nanowires, a post-thermal treatment was performed on ZnO nanowire samples at 450 ℃ for 30 min in the air.

For DSSCs fabrication, ZnO nanowire-based electrodes were immersed in a 0.3 mM ethanol solution of N3 at 40 ℃ for 30 min for dye loading. The dye-adsorbed ZnO nanowire electrodes were then rinsed with ethanol and dried under a nitrogen stream. The Pt counter electrode was prepared by sputtering a layer of 10 nm thick Pt on FTO glass. The two electrodes were clipped together and used a thermoplastic film (60 $\mu$m in thick, Solaronix, SX1170-60) as sealant to prevent the electrolyte solution (Solaronix, Iodolyte PN-50) from leaking. The active electrode area was 0.25 cm$^{2}$.

The morphology and structure of ZnO nanowires were characterized by scanning electron microscopy (SEM, Hitachi, S-800). The thickness of ZnO nanowire film was measured by a profiler (Bruker, DektakXT). The photocurrent-voltage characteristics of the cells were recorded using a computer-controlled potentiostat (e-DAQ) at room temperature. A light intensity-tunable xenon lamp (Schott, KL1500-T, 150 W) was used as the light source. The amount of dye adsorbed onto the ZnO nanorods of each cell was determined by desorbing the dye molecules from a 0.25 cm$^{2}$ nanowire array area in 5 mL KOH (1 mM) aqueous solution for 30 min and then measuring the absorbance using a UV–vis spectrophotometer (Lambda 45, PerkinElmer).

3. Results and discussion

In the typical hydrothermal experiment, ZnO nanowires are synthesized based on the following reactions^[15-17]:

$({\rm CH}_2 )_6 {\rm N}_4 +6{\rm H}_2 {\rm O}\leftrightarrow 4{\rm NH}_3 +6{\rm HCHO},$

(1)

${\rm NH}_3 +{\rm H}_2 {\rm O}\leftrightarrow {\rm NH}_3 \!\cdot\! {\rm H}_2 {\rm O}\leftrightarrow {\rm NH}_4^+ +{\rm OH}^-,$

(2)

${\rm Zn}^{2+}+2{\rm OH}^-\leftrightarrow {\rm Zn(OH)}_2 \leftrightarrow {\rm ZnO}+{\rm H}_2 {\rm O}.$

(3)

During the nanowire growth process, HMTA hydrolyzes into formaldehyde and ammonia (Eq. (1)^[18]), acting as a pH buffer by slowly decomposing to provide a gradual and controlled supply of ammonia, which can form ammonium hydroxide and support OH$^{-}$. Zn$^{2+}$ can complex with OH$^{-}$ to form several monomeric hydroxyl species^[19], including Zn(OH)$^+$ (aq), Zn(OH)$_2$ (s), Zn(OH)$_3^-$ (aq) and Zn(OH)$_4^{2-}$ (aq). Then solid ZnO nuclei are formed by the dehydration of these hydroxyl species at the surface of the ZnO seed layer. The ZnO crystal continues to grow by the condensation of the surface hydroxyl groups with the zinc-hydroxyl complexes^{[20, 21]}. Additionally, ligands such as HMTA and ammonia can kinetically control the species in solution by coordinating to Zn$^{2+}$ and keeping the concentration of the free Zn$^{2+}$ low. HMTA and ammonia can also coordinate to the ZnO crystal, hindering the growth of certain surfaces^[20].

The growth solution plays a key role in synthesizing the ZnO nanowires, so it is possible to obtain long and well-aligned ZnO nanowires through optimizing the growth conditions. Here we investigated the effect of refreshing the growth solution regularly on the morphology of ZnO nanowires. The results are shown in Figs. 2 and 3. We can see that the diameter and length of ZnO nanowires increase with the growth duration when refreshing the solution every 3 h. After growing for 15 h, they are $\sim$ 200 nm in diameter and $\sim$ 20 $\mu$m in length, respectively. The aspect ratio is about 100. As for the ZnO nanowires grown without refreshing the growth solution, both diameter and length increase slowly within the initial 12 h, which are $\sim$ 150 nm and $\sim$ 9 $\mu$m, respectively. The growth rate is obviously smaller than that of ZnO nanowires grown with refreshing the growth solution regularly. In addition, when increasing the growth duration further, both the diameter and length reduced sharply. That is because, at the initial stage of the hydrothermal reaction, the forward reaction would be favorable to forming ZnO sites (Eq. (3)). Accordingly, Zn$^{+}$ and OH$^{-}$ ions from the nutrient solution arrive at the sites of these nuclei to grow and form nanowires continuously, where Zn$^{+}$ ions are supplied from zinc nitrate salt and OH$^{-}$ ions are mainly supplied from the decomposition of the initially introduced ammonia solution (Eq. (2)). At this stage, the decomposition rate of HMTA is continuous and stable because of the high concentration of NH$_{4}$OH (Eq. (1)), which ensures the growth of ZnO nanowires^[22]. With the consumption of the zinc precursor, the reverse dissolution reaction in Eq. (3) becomes important and competitive with the forward crystallization reaction. After growing for 12 h, the zinc precursor in the unchanged growth solution is nearly depleted due to long time incubation. The excess OH$^{-}$ ions in aqueous solution will erode the formed ZnO nanowires instead of being consumed in growth, which results in the decrease of diameter and length of ZnO nanowires^[23]. So it is necessary to refresh the growth solution regularly when ultralong and well-aligned ZnO nanowires are expected.

DSSCs with N3 dye-sensitized ZnO nanowires of different growth cycles as photoanodes were assembled and their photocurrent–voltage ($I$–$V$) characteristics were systematically measured to study the photoelectrochemical performance under dark and illuminated conditions. As shown in Fig. 4, all DSSCs based on ZnO nanowires with different growth cycles (1, 3, 5) present negligible positive current or photocurrent under dark conditions. It is universally known that the dark current in DSSCs is primarily caused by the recombination of electrons at the semiconductor electrode and electrolyte interface. The results indicate that our devices possess excellent diode characteristics.

Figure 5 shows the $I$–$V$ characteristics of DSSCs with ZnO nanowires of different growth cycles under the same illumination condition. We can see that the short-circuited current ($J_{\rm SC})$, open-circuited voltage ($V_{\rm OC})$ and fill factor (FF) for the cell constructed using ZnO nanowires with 5 growth cycles represent clear improvement over the cells constructed using ZnO nanowire arrays with fewer growth cycles. As the conversion efficiency $\eta$ of the solar cell is defined as the ratio of the generated maximum electric output power (the product of the available maximum photovoltage $V_{\rm m}$ and photocurrent $I_{\rm m})$ to the total power of the incident light $P_{\rm in}$,

$\eta =\frac{V_{\rm m} I_{\rm m} }{P_{\rm in} }=\frac{V_{\rm OC} J_{\rm SC} {\rm FF}}{P_{\rm in} },$

(4)

in which the FF of the solar cell is defined as FF $=V_{\rm m} I_{\rm m}/(V_{\rm OC} J_{\rm SC}$). Due to the much improved $J_{\rm SC}$ and FF, the total power conversion efficiency of a cell based on ZnO nanowires of 5 growth cycles is 3.14 times larger than that of the cell based on ZnO nanowires of only 1 growth cycle. As presented in Table 1, the FF value is a little low, which may be ascribed to the recombination loss of electrons on the ZnO nanowire photoanode-electrolyte interface. To improve the cell performance, it is necessary to modify naked ZnO nanowires with a passivating layer in order to suppress the recombination loss of electrons.

Figure 6 shows the $I$–$V$ characteristics of the same DSSC based on ZnO nanowires with 5 growth cycles under different illumination intensities, which increased from level 1 to level 5. The results showed that our device possessed good photovoltaic response to the change of illumination intensity. With the increase of illumination intensity, both $J_{\rm SC}$ and $V_{\rm OC}$ increased obviously.

Previous reports ascribed the low current density of nanowire-based DSSCs to the small surface area of the nanowire arrays with low dye loading and light harvesting. The better performance of DSSCs based on ZnO nanowires with 5 growth cycles shown in Fig. 5 is possibly caused by a better dye loading and light harvesting, which can be verified by desorption experiment. Figure 7 shows the UV–vis absorption spectra of N3 dye detached from the ZnO nanowires with different growth cycles. Here the spectra exhibited two intense absorption bands at 366 nm and 493 nm, which are characteristics of N3 dye. Notably, the amount of dye loading increases with the increase of growth cycles of ZnO nanowires. So it is contributing to the increase of $I_{\rm SC}$ and DSSC photovoltaic efficiency, which resulted from an increased number of dye molecules chemically adsorbed onto the ZnO nanorod surface.

The amount of dye loading mainly depended on the surface area and surface conditions of the ZnO nanowires. Since different growth cycle of the ZnO nanowires did not change the surface conditions of the ZnO nanowires, so the difference in the amount of dye loading shown in Fig. 7 primarily resulted from the density, length, or size of the nanowires.

4. Conclusion

In summary, long and well-aligned ZnO nanowires have been synthesized on FTO substrate at low temperature via hydrothermal technique. The simplified method of preparing ZnO seed layer on FTO substrate is favorable to improve the reliability and reproducibility of the fabrication process of ZnO nanowires. The effect of refreshing the growth solution on the morphology of ZnO nanowires has been systematically investigated and the possible mechanism was analyzed. After growing for 5 cycles, ZnO nanowires of $\sim$ 20 $\mu$m in length have been obtained. Dye-sensitized solar cells (DSSCs) with the prepared ZnO nanowires as photoelectrodes exhibited excellent photovoltaic performance. With the increase of growth cycles of ZnO nanowires, the photocurrent of DSSCs increases obviously. The conversion efficiency of DSSC with ZnO nanowire of 5 growth cycles was improved by more than 2 times compared with that of only 1 growth cycle due to the increased dye loading on the surface of ZnO nanowires. The present results imply that the long and well-aligned ZnO nanowires are promising materials for DSSCs and other optoelectronic applications.

Acknowledgments: Han Zhitao is grateful to the Chinese Scholar Council for the grant.