We have prepared and employed TiO2/CdZnS/CdZnSe electrodes for photochemical water splitting. The TiO2/CdZnS/CdZnSe electrodes consisting of sheet-like CdZnS/CdZnSe nanostructures (8–10 m in length and 5–8 nm in width) were prepared through chemical bath deposition on TiO2 substrates. The TiO2/CdZnS/CdZnSe electrodes have light absorption over the wavelength 400–700 nm and a band gap of 1.87 eV. Upon one sun illumination of 100 mW cm−2, the TiO2/CdZnS/CdZnSe electrodes provide a significant photocurrent density of 9.7 mA cm−2 at −0.9 V versus a saturated calomel electrode (SCE). Incident photon-to-current conversion efficiency (IPCE) spectrum of the electrodes displays a maximum IPCE value of 80% at 500 nm. Moreover, the TiO2/CdZnS/CdZnSe electrodes prepared from three different batches provide a remarkable photon-to-hydrogen efficiency of 7.3 ± 0.1% (the rate of the photocatalytically produced H2 by water splitting is about 172.8 mmol·h−1·g−1), which is the most efficient quantum-dots-based photocatalysts used in solar water splitting.

1. Introduction

Developing environmentally clean energy resources from abundant solar energy has attracted considerable attention these years [1]. Hydrogen production by photochemical water splitting is a promising route because hydrogen has the highest energy density values per mass (140 MJ kg−1) and its oxidation product (H2O) is more eco-friendly [24]. Hitherto, many semiconductors with band-gap energy exceeding the oxidation potential of water (1.23 V versus normal hydrogen electrode (NHE)) at pH 1.0 have been employed for water splitting [5]. Albeit metal oxides including TiO2, ZnO, and their derivatives are the most common photocatalysts used in water splitting, yet they provide low overall photon-to-hydrogen efficiency () attributed to their wide band gaps [6, 7]. To overcome these limitations, doping other metal or inorganic ions to TiO2 and ZnO materials has been demonstrated [8]. However, this strategy is not quite successful, mainly because their band gaps are greater than 2.0 eV (620 nm) [9], whereas photocatalysts having band gaps less than 2.0 eV can absorb solar light in the visible to near-infrared region more efficiently.

Quantum dots (QDs) such as CdTe [10], CdS [1113], and CdSe [1418] have been anchored to TiO2 and ZnO electrodes to harvest visible light for more efficient water splitting. Photoelectrochemical cells (PECs) incorporating QDs-sensitized TiO2 electrodes provide several advantages: (i) ease of fabrication, (ii) generation of multiple electron/hole [19], (iii) high visible light harvesting capability in solar spectrum, and (iv) tunable band gaps due to the quantum size effect of QDs. Remarkably, CdTe-sensitized ZnO nanowire electrodes provided a superior of 1.83% relative to that of ZnO nanowires (0.66%) in a nonsacrificial electrolyte [10]. Under white light illumination of 100 mW cm−2, CdS/CdSe QDs-sensitized ZnO nanowire electrodes and TiO2 electrodes had high photocurrent densities of 12 and 14.9 mA cm−2, respectively [1418]. Although PECs featuring QDs-sensitized TiO2 and ZnO electrodes provided greater values for water splitting than corresponding bare electrodes, their values are still less than 4% [1113]. Moreover, an external potential is mandatory to separate the charges efficiently to prevent the recombination of electron-hole pair under illumination [20].

Therefore, to fabricate highly efficient PECs for water splitting, we prepared TiO2/CdZnS/CdZnSe electrodes from Cd(NO3)2, ZnSO4, Na2S, and Na2SeSO3 via chemical bath deposition (CBD) approach [21]. Under AM 1.5 illumination (100 mW cm−2) and in presence of a sacrificial electrolyte (2 M Na2S and 0.25 M Na2SO3), the three as-prepared TiO2/CdZnS/CdZnSe electrodes provided of 7.3 0.1%. To the best of our knowledge, this is the most efficient QDs-sensitized electrode used in solar water splitting.

2. Materials and Methods

2.1. Chemicals and Instrumentation

Cadmium nitrate, commercially available P-25 TiO2 powders (Degussa), ethylene glycol, methyl cellulose, nitric acid, potassium chloride, poly (vinylpyrrolidone) (PVP, 55,000), polyethylene glycol ( 5,000), selenium powder, sodium hydroxide, sodium sulfide, sodium sulfite, and zinc sulfate were purchased from Sigma-Aldrich (Milwaukee, WI, USA). Fluorine-doped tin oxide (FTO) glass was obtained from Plasma Technology (Taoyuan, Taiwan).

2.2. Preparation of TiO2 Electrodes

A repetitive dry method was used to prepare the TiO2 electrodes [21]. In brief, TiO2 paste was prepared by mixing TiO2 powder (after HNO3 treatment; 0.6 g), PVP (0.18 g), methyl cellulose (0.06 g), and ultrapure H2O (3 mL). The paste (0.05 ml) was then applied to one of the bare edges of a FTO glass; it was flattened by sliding a glass rod over the tape-covered edges. A single-layer TiO2 electrode was obtained after drying at 50°C in an oven for 1 h. The process was repeated to obtain a double-layer TiO2 electrode with a thickness of 20 m.

2.3. Preparation of TiO2// Electrodes

The as-prepared TiO2 electrodes each with an effective area of 1 cm2 were immersed into a solution (2 mL) containing Cd(NO3)2 (0.5 M) and ZnSO4 (0.75 M) for 5 min, rinsed with ultrapure H2O (1 mL), and dried with an air gun. They were then dipped for 5 min into an aqueous solution (2 mL) of 0.5 M Na2S, and then rinsed with ultrapure H2O (1 mL), and finally dried with an air gun. The process was repeated up to cycles ( is integer 1~7). These as-prepared electrodes are designated herein as TiO2/ electrodes. The TiO2/ electrodes were further dipped into a solution of Cd(NO3)2 (0.5 M) and ZnSO4 (0.75 M) for 5 min at 27°C and then immersed into aqueous Na2SeSO3 (0.08 M) at 50°C for 60 min followed by rinsing with ultrapure H2O and then dried with an air gun. The process was repeated up to cycles ( is integer 1~4). The as-prepared electrodes are designated as TiO2// electrodes.

2.4. Preparation of TiO2///ZnS Electrodes

The as-prepared TiO2// electrodes were immersed separately into ZnSO4 solution (0.5 M) for 5 min, rinsed with ultrapure H2O, and dried with an air gun. They were then dipped for 5 min into aqueous Na2S solution (0.5 M) followed by rinsing with ultrapure H2O and dried with an air gun. The as-prepared electrodes are represented as TiO2///ZnS electrodes.

2.5. Measurements

The UV-vis absorption spectra of the as-prepared electrodes were recorded using a Cary 100 UV-Vis spectrophotometer from Varian (Palo Alto, CA, USA). Scanning electron microscopy (SEM) images and energy dispersive spectra (EDS) were recorded using an S-2400 SEM system from Hitachi (Tokyo, Japan). Cyclic voltammetry (CV) tests were performed using a CHI 700D electrochemical analyzer (CH Instruments, Austin, TX, USA). CV measurements of the TiO2// electrodes (effective area: 1.0 cm 1.0 cm) were performed using a three-electrode system: a TiO2// electrode as a working electrode, a Pt counter electrode, and an SCE as a reference electrode. CV measurement was carried out in a solution containing 2.0 M Na2S and 0.25 M Na2SO3. The solution was purged by nitrogen to remove dissolved oxygen before the experiment. The irradiation source was a 450-W xenon arc lamp (Oriel, Stratford, CT, USA) equipped with an AM 1.5 filter. A commercially available silicon-based reference cell (Oriel, Stratford, CT, USA) was employed to measure the light intensity (100 mW cm−2). Incident photon-to-electron conversion efficiency (IPCE) spectra were recorded using a PEC-S20 instrument (Peccell Technologies, Kanagawa, Japan). Photocatalytic hydrogen generation experiments have been carried out in a lab-made photochemical reactor. A typical glass cell consisted of photoanode (TiO2//) and cathode (Pt foil) that were connected by a fine porous glass frit. The H2 produced in our experiments was measured by applying a gas volume headspace method.

3. Results and Discussion

3.1. Properties of Electrodes

The CBD layers of QDs play a significant role in light harvesting [22], so we investigated the role that the amounts (layers) of CdZnS and CdZnSe nanostructures played in determining the values (Scheme 1). Moreover, upon increasing / layer numbers, the light absorption increased and reached a maximum value at , and (not shown). The as-prepared TiO2// electrode absorbs light in the wavelength range of 400–700 nm (Figure 1(a)). The band gap corresponding to the absorption edge of this electrode was 1.87 eV (662 nm) [23]. This value is in the range of 1.23–2.0 eV, validating it as a suitable photocatalyst for water splitting [5, 918]. The existence of Cd, Zn, S, Se, Ti, and O components in the as-prepared electrode was further confirmed from energy dispersive spectroscopy (EDS) results (Figure 1(b)). The scanning electron microscopy (SEM) image (inset to Figure 1(b)) elucidates that the as-resulting electrode was decorated with sheet-like nanostructures, having lengths of 8–10 m and widths of 5–8 nm, respectively. The cross-sectional view of the photoelectrode is shown in Figure 2, which reveals that sheet-like nanostructures are not only found on the surface but also deeply embedded within the film as denoted by the dotted circles.


The electrodes were tested for solar water splitting in presence of sacrificial electrolyte [24, 25] consisting of Na2S and Na2SO3. The use of Na2S/Na2SO3 mixture provided an advantage of negligible photocorrosion effect on QDs. Moreover, Na2S in solution acted as a hole scavenger and it was oxidized into , which was reduced back to by Na2SO3 [26]. Photogenerated holes irreversibly oxidized the reducing agents (Na2S/Na2SO3) instead of water, providing the photocatalyst electron rich and an enhanced H2 evolution reaction. The reaction mechanism of each electrode can be described as follows.Anode: Cathode:

3.2. Photocatalytic Hydrogen Evolution

It was evident from linear sweep voltammetry (LSV) characteristic curves (Figure 3(a)), under one sun light illumination (AM 1.5, 100 mW cm−2), the TiO2// electrode produced a noteworthy limiting photocurrent density value of 9.7 mA cm−2 at −0.9 V versus SCE, which was superior to those produced by most QDs-sensitized electrodes [1018]. Such a high photocurrent density can be attributed to the cascade structure of /. Although electrodes having higher limiting photocurrent densities have been reported [1418], their relative high applied voltages (0.356 and −0.574 V versus SCE) are disadvantageous over ours. The as-prepared electrode required a very low onset potential (−0.9 V), implying that it required less energy to drive the reaction. The negative shift of the onset potential confirms the enhanced charge transfer which may possibly be due to a high recombination rate and kinetic hindrance by / sensitization [16]. We further conducted amperometric - measurements to examine the photocatalytic activity and photoresponses of the TiO2// electrode (Figure 3(b)). Notably, the potential set at −0.9 V produced stable limiting photocurrent density. Upon illumination, spiked photoresponses of 9.7 mA cm−2 were observed, which was much higher than the dark current density (~50 μA cm−2). The reproducibility of the photoresponses was also excellent (relative standard deviation 0.4% from 10 replicated measurements). The photoresponses were fast (<4 s), revealing their promising photocatalytic activity. By using (7) [27], the value of for the data obtained from Figure 3 is as follows: where, is the photocurrent density (mA cm−2), is the total power output, is the electrical power input, and is the power density of incident light (100 mW cm−2). (1.23 V/NHE) is the standard reversible potential for water splitting. The applied voltage is calculated as 0.48 V using , where is the working electrode potential at which photocurrent was measured under illumination and is the potential measured at this working electrode under open circuit potential (OCP) and same experimental conditions, respectively. As shown in Figure 3(c), a maximum value of 7.3% was achieved at an of 0.48 V. We also recorded values at various values from three different batches of TiO2// electrodes, revealing of 7.3 0.1%, which is superior to those (e.g., 3.67%) [1113] of other QDs-sensitized TiO2 electrodes [1116]. Relative to some reported electrodes such as TiO2/CdS (6.4 mmol·h−1·g−1) and CdSe/CdS (40 mmol·h−1·g−1) [17, 18], our QDs-sensitized electrodes provided a higher H2 generation rate of 172.8 mmol·h−1·g−1 (). However, in many PECs competing side reactions dominate, resulting in different products and less than ideal faradaic efficiency. If we consider an average current of 9.7 mA cm−2 flowing through the circuit, one would expect to observe H2 formation at a theoretical rate of 200.9 mmol·h−1·g−1 (at 25°C). The observed H2 rate of 172.8 mmol·h−1·g−1 accounts for 86% of the amount predicted on the basis of current flow.

To further quantify the performance of PECs incorporating TiO2// electrodes, their IPCE values were acquired in the wavelength range of 400–700 nm (Figure 3(d)), [10, 16, 27, 28]. The IPCE values were determined from [16] where, is the photocurrent density, is the incident light wavelength, and is the measured irradiance. The TiO2// electrodes exhibited a pronounced IPCE value of 80% at 500 nm. We further estimated the values from the following [28]: where is the electron charge and is the incident photon flux of the solar light. The value of calculated from the maximum IPCE value at AM 1.5 photon flux was ca. ~9.74 mA cm−2, which was quite close to the value (~9.7 mA cm−2) measured from LSV measurements. The close agreement between values obtained from IPCE spectra and curves indicates that the photocurrent was indeed generated from the electrode upon illumination [28]. Moreover, photochemical stability of the TiO2// electrodes in sacrificial electrolyte was quite appreciable as negligible changes in the curve was observed over 100 cycles (not shown). Table 1 summarizes the LSV measurements of various electrodes employed for solar water splitting under the same conditions. These results revealed that the cascade structure of / played a major role in determining the values of . Upon increasing the CdZnS layers, the value increased and reached a plateau at , when was kept 2. The absorption spectra results (Figure 4) validate that the generation of different photocurrent densities is a vital factor responsible for superior light harvesting efficiency of these electrodes. The band gaps of the as-resulting CdZnS and CdZnSe QDs were 2.40 and 1.85 eV, respectively. Thus, the insertion of CdZnS layer between TiO2 and CdZnSe of the cascade structure elevates the conduction band edge of CdZnSe, and thus provides higher driving force for the injection of excited electrons out of CdZnSe layer. However, upon increasing the values of from 5 and from 2 (the thickness of the QDs layer), the photoexcited electrons could not be effectively injected into the TiO2 conduction band, likely due to the presence of additional QDs. Nevertheless, the position of the absorption peak exhibits a red shift with increase in the number of layers, mainly due to the size quantization and confinement effects.

We further measured OCP of the TiO2/, TiO2//, and TiO2/ electrodes under dark conditions (Figure 5). The OCP values were −1.46 and −1.33 V versus SCE for TiO2/ and TiO2/ electrodes, indicating that CdZnS had a higher Fermi level than that of CdZnSe. When the CdZnS/CdZnSe junction is formed, electrons will flow from CdZnS to CdZnSe until the Fermi energy of the electrons in the CdZnS equals to that of CdZnSe. Band offset occurred at the interface of CdZnS/CdZnSe electrodes under illumination, causing a shift of Fermi level (−1.38 V versus SCE) in the TiO2// electrode. As a result, higher value (9.7 mA cm−2) was generated in the TiO2// than those (2.9 and 4.6 mA cm−2) of the TiO2/ and TiO2/ electrodes.

To highlight the features of the TiO2// electrodes, we have listed some of the most effective electrodes for solar water splitting (Table 1). The as-prepared TiO2// electrodes provide the highest value as a result of their lowest applied voltage and high-current density. However, they had a negative applied voltage versus SCE, mainly because their Fermi energy level will be changed due to the band realignment of TiO2 by CdZnS and CdZnSe sensitization. These results also reveal that the photocurrents and applied voltage are highly dependent on the structure of the photoelectrodes.

4. Conclusions

In conclusion, we have employed TiO2// electrodes for highly efficient solar water splitting. These electrodes yielded a significant photocurrent density of 9.7 mA cm−2 at −0.9 V versus SCE, mainly attributing to high light harvest efficiency (e.g., IPCE of 80% at 500 nm). To the best of our knowledge, the electrodes provided the highest value of for the solar water splitting among QDs-based photocatalysts, mainly because they provide high photocurrent density at low applied voltage. Although the TiO2// electrodes hold great promise for commercial use in water splitting, further increase in their efficiency is still required.


The authors thank the National Science Council, Taiwan, for the financial support (NSC 101-2627-M-002-007-). Z. Yang and A. P. Periasamy thank the National Taiwan University for the award of a postdoctoral fellow of the Department of Chemistry, National Taiwan University.