Abstract
Granular ZnO nanostructures of single-crystalline wurtzite hexagonal phases were synthesized by a facile and low-cost chemical method in aqueous condition. The average size of ZnO nanograin increased with reflux time, and it significantly affected the open circuit potential while the short circuit current density was not changed much. The overall energy conversion efficiency was 1.82% with the smaller grain size of ca. 250 nm when it was used as photoelectrode of DSSCs. The positive shifting of the Fermi energy () and low density of surface states (DOS) were consistent with the reduction of the recombination of excited electron with electrolyte for smaller grains.
1. Introduction
Zinc oxide (ZnO) is unique in exhibiting both semiconducting and piezoelectric properties [1]. Bulk ZnO is a direct and a wide band gap semiconductor ( at 25°C) with high excitation binding energy (~0.06 eV), large saturation velocity (3.2 107 cm s−1), and high optical gain (300 cm−1) [2]. The high excitation binding energy leads to excitonic transitions (~25°C), high radiative recombination efficiency for spontaneous emission, and a low threshold voltage for laser emission. The quantum confinement effect of photo-generated electron-hole pairs and the tunability of optical and electronic properties of ZnO nanoparticles (~10 nm) [3, 4] allowed them to be used in various nanodevices such as information storage, sensing, and surface acoustic wave devices [5–7].
ZnO is potentially useful in dye-sensitized solar cells (DSSCs) because it has a similar band gap to TiO2 and shows lower electron-hole () recombination probability due to its filled valence band ()() and s-p hybridization properties of conduction band () [2]. The less population of deep traps in ZnO nanostructures compared with TiO2 nanoparticles induced the longer electron lifetime [2]. In spite of these superior physical properties of ZnO, the photoconversion efficiency of TiO2-based DSSC is generally much higher than ZnO. This was mainly attributed to the acidity of the binding groups of dyes (e.g., N719, N3, and black dyes, etc.), which reduced the chemical stability of ZnO by dissolving it to form a precipitation of dye-Zn2+ complexes to lead electron injection and dye regeneration inefficiently [8]. In general, it is known that nanostructures, such as nanorods, nanofibers, and nanowires, induce less recombination to improve the charge collection efficiency [9–11]. This work reports the synthesis of granular ZnO nanostructures by simple hydrothermal reactions of zinc nitrate and KOH in the presence of tetra butyl ammonium hexafluorophosphate (TBAPF6). The effects of the size of ZnO nanograin on the photovoltaic performances of DSSCs were studied in conjunction with their morphology as well as the optical and the electronic properties.
2. Experimental
Zinc nitrate [Zn(NO3)2·6H2O] (0.01 mol) and KOH (0.02 mol) were dissolved in 20 mL of distilled water to form a white flocculate, which was then diluted to 100 mL. After the addition of 5 10−3 mol TBAPF6 while stirring, it was refluxed for 30, 60, and 120 min to obtain white precipitates. The samples, being collected by centrifuge and dried for 30 hours at room temperature after washing with water and ethanol, were sintered in an electric muffle furnace at 500°C for 30 min in ambient condition before further characterization. They were denoted as Z1, Z2, and Z3, respectively, depending on the reflux time.
Scanning electron microscopy (SEM, Hitachi S-3000N, Japan) was used to investigate the ZnO nanograin’s surface morphologies. X-ray diffractometry (Philips, X’pert, Netherland) was conducted using Cu radiation ) over a 2 scan range of 20°–80°. Optical absorption and photoluminescence emission spectra of the ZnO dispersed in ethanol were obtained using a UV-Vis spectrophotometer (Scinco, S-3100, Republic of Korea) and spectrofluorometer (Horiba, Flourolog-3, USA). The amount of dye adsorptions on photoelectrodes was measured from the absorption spectra of desorbed-dye solutions as reported in the literature [12].
ZnO paste was prepared by mixing ZnO powder, which was pretreated by acetyl-acetone to minimize aggregation with ethanol, terpineol, and 10% ethyl cellulose as reported elsewhere [10, 11]. The ZnO films on FTO (Pilkington, were prepared by doctor blading and were sintered in an electric muffle furnace at 500°C under ambient condition. They were immersed in ethanolic solution of 0.4 mM cis-diisothiocyanato-bis (2,2′-bipyridyl-4,4′-dicarboxylato) ruthenium(II) bis-tetrabutylammonium (N719) for 12 h. Counter electrodes were prepared by spin coating of 5 mM chloroplatinic acid hexahydrate (H2PtCl66H2O) in ethanol on FTO and by sintering it in an electric muffle furnace at 380°C for 20 min. The dye-loaded photoelectrodes (active area ca. 0.2 cm2) and platinized counter electrodes were sandwiched with 50 m thick surlyn film as a spacer and sealing agent at 110°C for 10 min. The electrolyte solution composed of 0.6 M 1,2-dimethyl-3-propylimidazolium iodide (DMPII), 0.1 M LiI, 0.1 M I2, and 0.5 M 4-tert-butylpyridine (tBP) in 3-methoxypropionitrile (MPN) was injected into the cell through the drilled holes on the counter electrode. They were sealed with a transparent scotch tape for temporal sealing.
An AM1.5 solar simulator with a 200 W Xenon lamp (Polaronix K201, McScience, Republic of Korea) was used to illuminate the cells, whose current density-voltage (J-V) properties were measured using a photovoltaic power meter (Polaronix K101 LAB20, McScience, Republic of Korea). The incident light intensity was adjusted to 100 mWcm−2 (1 sun) by a standard mono-Si solar cell (PVM 396, PV Measurement Inc., USA) certified by the US National Renewable Energy Laboratory. Cyclic voltammetric (CV) experiment was performed with a CHI430A electrochemical workstation (CH instruments, Inc., USA). ZnO nanograins on FTO, a platinum wire, and a Ag/Ag+ electrode were used as working, counter, and reference electrodes, respectively. Electrochemical impedance spectra (EIS) were obtained under open circuit and dark conditions at frequencies of 105–0.1 Hz with a 5 mV ac amplitude (IM6ex, Zahner-Elektrik GmbH & Co. KG, Germany). The measured spectra were fitted to an equivalent circuit appropriate for DSSCs using Zview software (version 3.1, Scribner Associates Inc., USA).
3. Results and Discussion
The SEM images showed that the average size of ZnO nanograins increased slowly with reflux time by reducing the population of smaller size grains, while the aspect ratio (ca. 1.75) was almost unchanged. The lengths of the grains ranged from ca. 50 to 750 nm, while the upper limit of the grain length was ca. 350, 500, and 750 nm for Z1, Z2, and Z3, respectively (Figure 1(a)). The XRD patterns of all the synthesized ZnO nanograins were consistent with the single-crystalline wurtzite (hexagonal-phase) with no impurities such as Zn and Zn(OH)2. All the diffraction peaks were well indexed to the hexagonal phase of ZnO reported in JCPDS card (no. 36-1451) (Figure 1(b)(A)).
(a)
(b)
All the absorption maxima (), observed at ca. 362, 376, and 378 nm for Z1, Z2, and Z3, respectively, were blue-shifted relative to the bandgap energy () of bulk ZnO [2] with the much broader absorption edges for Z1 (Inset, Figure 1(b)(B)). The lowest value of the of Z1 indicated the larger and the positive shift of quasi-Fermi level () compared to those of Z2 and Z3 [13]. Photoluminescence emission spectra, recorded at room temperature under excitation at 350 nm for Z1 and 360 nm for Z2 and Z3, clearly showed the strong UV emissions peaked at ca. 368 nm (Z1) and ca. 385 nm (both Z2 and Z3) (Figure 1(b)(B)) which could be attributed to the direct and instant radiative recombination of excitons (RR0) (Scheme 1). The very broad and weak emission observed at ca. 565 nm for Z1 was attributed to the other radiative recombination (RR1) through the defect levels possibly caused by vacancies in the interstitial site of zinc and oxygen [14] as described in Scheme 1. Holes form the valence band edge () of ZnO recombined with electron by tunneling through these defect levels. The significantly low emission intensities in the visible range indicated little density of defect levels for Z2 and Z3 compared with Z1 [15–18]. It indicated that the Z1 has the high probability of recombination of excitons under UV illumination. Nevertheless, the photoelectrode constructed with Z1 showed the better performance in DSSCs as shown in the photocurrent density-voltage (J-V) characteristics (Figure 2(a) and Table 1). It was partly because there was no exciton generation in Z1 nanograins during the operation of DSSC and the source of electron was from dyes, which were mostly excited by visible light under AM 1.5 condition. Therefore, the existence of these intrinsic defect levels of Z1 nanograins did not contribute much to reduce the cell performance because the recombination of these electros was more significantly involved with the highest occupied molecular orbital (HOMO) of dye, which is energetically much more closer to the conduction band edge () of ZnO than these intrinsic defect levels of Z1 nanograins (Scheme 1).
 |
(a)
(b)
(c)
(d)
The overall energy conversion efficiency with Z1 was ca. 1.82% and it was up to 11% and 15.2% higher than that of Z2 and Z3, respectively. The higher efficiency for Z1 was mainly attributed to the higher open circuit potential () compared to Z2 and Z3. Meanwhile, the short circuit current density () was not affected significantly even though the dye loading for Z1 was higher up to ca. 26.8 and 41.42% than those of Z2 and Z3, respectively (Table 1). It could be attributed to the scattering effect with bigger grains [19]. The variation of in DSSC was known to be dependent significantly on the distribution and density of surface states in the mesoporous structure of photoelectrode, which was investigated by monitoring the nonfaradic capacitive current flow at these electrodes. The capacitive charging current appeared in the forward scans in CVs (Figure 2(b)) could be interpreted as the filling of these surface trap states as previously studied in similar systems based on TiO2 nanoparticles [20–22] and the current essentially reached zero during reverse scans to sufficiently positive potentials because of the regeneration of these surface states by discharging. The higher values of capacitive currents for Z2 and Z3 compared to those of the Z1 indicated the higher density of such trap states as described in Scheme 1. It was found, from the density of states (DOS) calculation [22], that they distributed exponentially with the onset potential being dependent on the size of ZnO grain (Figure 2(c), Table 1). The downshift of the onset of the state distribution was consistent with the tendency of the negative shift of the pseudo-Fermi level of the photoelectrode (Scheme 1) [23], which was responsible for the higher recombination rate to induce the decrease of . The electron transfer kinetics, measured by electrochemical impedance spectroscopy (EIS) at the ZnO|electrolyte interface under dark condition (Figure 2(d)), was also consistent with this trend by showing the lowest value of the back electron transfer rate to () for Z1 as summarized in Table 1.
4. Conclusions
Crystalline ZnO nanograins being synthesized by a facile hydrothermal method in aqueous system were used as photoanodes in DSSCs. It was found that some physicochemical properties of nanograins such as the density and the distribution of intrinsic defects and surface states were major factors to control many optoelectronic properties such as quasi-Fermi level and radiative recombination probabilities. They were highly dependent on the size of grains and this also affected the performance of DSSCs when used for photoelectrode. Even though the maximum conversion efficiency obtained was 1.82% with the grain size of ca. 250 nm (Z1), it showed an improved performance compared with similar previously reported system based on ZnO nanostructures [24–27]. It was expected that the use of ZnO nanograin can offer a promising strategy to enhance the performance of DSSCs and further optimization by controlling the size and aspect ratio to enhance the cell performance is still under investigation.
Conflict of Interests
The authors declare that they have no conflict of interests.
Acknowledgments
This work was supported by the National Research Foundation of Korea (NRF) grant funded by the government of the Republic of Korea (MEST) (NRF-20100026916) and by the Converging Research Centre Program through the Ministry of Education, Science and Technology (2012K001287) and by Seoul R&BD Program (WR090671).