Cheng2012.pdf

ARTICLE pubs.acs.org/JPCC

Photoelectrochemical Performance of Multiple Semiconductors (CdS/CdSe/ZnS) Cosensitized TiO2 Photoelectrodes Shuli Cheng, Wuyou Fu, Haibin Yang,* Lina Zhang, Jinwen Ma, Hui Zhao, Meiling Sun, and Lihua Yang State Key Laboratory of Superhard Materials, Jilin University, Changchun 130012, P. R. China

bS Supporting Information ABSTRACT: The morphology of TiO2 nanotubes with nanowires directly formed on top (designed as TiO2 NTWs) would be a promising nanostructure in fabricating photoelectrochemical solar cells for its advantages in charge separation, electronic transport, and light harvesting. In this study, a TiO2 NTWs array film was prepared by a simple anodization method. The formation of CdS, CdSe, and ZnS quantum dots (QDs) sensitized TiO2 NTWs photoelectrode was carried out by successive ionic layer adsorption. The as-prepared materials were characterized by scanning electron microscopy, high-resolution transmission electron microscopy, and X-ray diffraction. Our results indicate that the nanocrystals have effectively covered both inner and outer surfaces of TiO2 NTWs array. The interfacial structure of QDs/TiO2 was also investigated for the first time in our experiment, and the growth interface when annealed to 300 °C was verified. Under AM 1.5G illumination, we found the photoelectrodes have an optimum short-circuit photocurrent density of 4.30 mA/cm2 and corresponding energy conversation efficiency of 2.408%, which is 28 times higher than that of a bare TiO2 NTWs array. The excellent photoelectrochemical properties of our photoanodes suggest that the TiO2 NTWs array films (2.6 2.8 μm) cosensitized by CdS, CdSe, and ZnS nanoclusters have potential applications in solar cells.

1. INTRODUCTION Increasing demand for energy is forcing us to research into new resources, among which solar energy is ideal to meet the target because of its abundant, clean, and inexhaustible characteristics. The solar-to-electric energy conversion is one important way in solar applications, and various solar cells such as photoelectrochemical (PEC),1 4 photovoltaic,5 and photocatalysis cells6 have been widely used. However, the conversion efficiency is always a bottleneck in these solar cells. Regarding the conventional PEC cell which is made up of a semiconductor photoanode and a platinum cathode in an electrolyte solution,1 materials of photoanode with good performance at photoresponse, charge separation, and electrons transference are of importance for the improvement of the photoelectric conversion efficiency. TiO2 films have been demonstrated to be promising candidate as photoanodes due to their appropriate energy band position and both thermal and chemical stability in solution. Among all the TiO2 films, one-dimensional single crystalline array films, including nanotubes,7 nonorods,8 and nanowires,9 caught particular attention as they have enlarged surface area and can provide direct pathway for photogenerated electron transfer, which accordingly leads to the enhancement of light harvesting and charge separation. The morphologies of TiO2 nanotubes with nanowires directly formed on top (denoted as TiO2 NTWs) have ever been reported and were supposed to be advantageous in various applications,10,11 for example, to enhance photocatalysis and to increase energy-conversion efficiency for dye or quantum dot (QD) solar cells. However, the large band gap of TiO2 (3.2 eV) limits its absorption region to the ultraviolet which takes only r 2011 American Chemical Society

3 5% of the whole solar spectrum. To extend the activity of a photoelectrode into the visible light region, various approaches were employed including doping TiO2 with other impurities,12,13 using dye-sensitized solar cells (DSSCs),14,15 and very recently using the composite semiconductors such as QDs-sensitized solar cells (QDSCs).16 18 Many kinds of small-band-gap semiconductors that can absorb light in the visible region, e.g., CdS,8,18 PbSe,19 CdSe,20 InP,21 and CdTe,22 have been used as sensitizers. Among them, CdS and CdSe are semiconductors with direct band gap (Egap) of 2.25 and 1.70 eV, respectively, which means CdS and CdSe can trigger wider light absorption range compared to TiO2. However, to accelerate charge separation in bulk/single crystal material, the level of the conduction band (CB) edge which provides the electron injection driving force from QDs to TiO2 should be taken into account. It is known that CdS and CdSe have the high and low CB edges with respect to that of TiO2, and a higher electron injection efficiency of CdS was indeed found in comparison with CdSe.23 On the basis of the above understanding, we propose that the comodified TiO2 NTWs with CdS as underlayer and CdSe as outer layer should be a promising photoanode for its wide absorption spectrum, high electron injection efficiency, and fast electrons transference. What is more, as composite semiconductors, the interfacial structure of the two semiconductors will affect the efficient charge transfer channel at the interface. As for the QDs modified TiO2, the area of the Received: September 25, 2011 Revised: December 14, 2011 Published: December 20, 2011 2615

dx.doi.org/10.1021/jp209258r | J. Phys. Chem. C 2012, 116, 2615–2621

The Journal of Physical Chemistry C channel is named effective coverage area. We think that achieving an effective coverage of the oxide with the QDs is key to the improvement of this type of QD solar cell. Three methods are mainly used in the QD modification on TiO2: (i) in-situ growth of QDs by chemical bath deposition (CBD) and successive ionic layer adsorption and reaction (SILAR), (ii) deposition of presynthesized colloidal QDs by direct adsorption (DA), and (iii) deposition of presynthesized colloidal QDs by linker-assisted adsorption (LA). However, the adsorption of presynthesized colloidal QDs in DA and LA has a low surface coverage of ∼14%,20 which results in inferior efficiency compared with directly grown QDs. As to the CBD methods, lack of stoichiometry has been observed in some cases. The method of SILAR turns out to be a proper approach for its nucleation and in-situ growth mechanism which can bring in a high coverage of the effective TiO2 surface and control of the size distribution of the deposited QDs by different cycles. Previously, Yuh-Lang Lee and co-workers prepared the TiO2/ CdS/CdSe electrode with a maximum energy conversion efficiency of 4.22%,17 while Shen et al. fabricated a CdSe QDs-sensitized TiO2 nanotube array (the tube length is about 10 μm) photoelectrode with a photocurrent of 6.2 mA/cm2 and the efficiency of 1.80%.7 Yet, to our knowledge, there is no work dedicated to the semiconductor-sensitized solar cells based on the TiO2 NTWs array films. In this paper, we report an investigation on the CdS/CdSe/ ZnS nanoclusters cosensitized TiO2 NTWs array photoelectrodes that fabricated by SILAR. The effective coverage of the oxide with nanoclusters has been realized, and the interfacial structure of QDs/TiO2 is clearly observed for the first time. The ZnS coating can be considered to be a potential barrier at the CdSe electrolyte interface, providing that the band gap of ZnS (3.8 eV) is much larger than that of CdSe.24 A detailed synthesis process, characterization, and photoelectrochemical properties of this photoelectrode are discussed. Comparisons of the properties between pure TiO2 NTWs array films and semiconductors comodified TiO2 NTWs array films are also presented.

2. EXPERIMENTAL DETAILS 2.1. Preparation of Self-Organized TiO2 NTWs Array. All chemicals used were of the highest purity available and used as received without further purification. Deionized water (18 MΩ 3 cm) was used in all cases. Prior to anodization, the pure Ti foil (4 cm  3 cm, 0.4 mm thick, 99.9% purity) were cleaned in an ultrasonic bath with acetone, isopropanol, and ethanol successively, followed by rinsing with deionized water and drying in flow of N2. A simple two-electrode cell at the temperature of 20 °C by using a direct current (dc) stabilized voltage and current power supply (WYJ60V3A, Pingguo instrumentation Co. Ltd., China) were employed, in which a pure Ti sheet served as the anodic electrode and a graphite plate as the cathode; the distance between the two electrodes was 4 cm. The backside of the pure Ti sheet was protected by coating a layer of waterproof rubberized fabric. Electrochemical anodization was performed at 40 V for 1 h in an electrolyte mixture containing 0.3 wt % NH4F and 5.0 vol % deionized H2O dissolved in ethylene glycol. After anodization, the as-prepared TiO2 NTWs were soaked in deionized water overnight and then dried in air. Then the TiO2 NTWs were annealed at 450 °C for 2 h. 2.2. Preparation of QDs Sensitized TiO2 NTWs Array Films. The method of successive ionic layer adsorption and reaction (SILAR) with slight modification is used to assemble QDs onto

ARTICLE

the TiO2 photoelectrode. For CdS modification, the TiO2 NTWs films were successively immersed into two different solutions for 10 min each, first in 0.50 M Cd(NO3)2 3 4H2O in ethanol and then in 0.50 M Na2S in deionized water. Following each immersion, the films were respectively rinsed with pure ethanol and deionized water to remove excess precursors and dried at 150 °C for 10 min before the next dipping. This is called one SILAR cycle. For CdSe sensitization, 0.50 M Cd(NO3)2 3 4H2O in ethanol and freshly prepared Na2SeSO3 aqueous solution which was kept at a 50 °C water bath were used, with a dipping time of 10 and 60 min, respectively, still dried at 150 °C for 10 min before the next dipping. The Na2SeSO3 aqueous solution used as Se source is prepared by refluxing Se (0.30 M) in an aqueous solution of Na2SO3 (0.60 M) at 90 °C until Se was fully dissolved in Na2SO3 solution. For ZnS capping, 0.10 M Zn(NO3)2 in ethanol and 0.10 M Na2S in methanol/water (7:3 v:v) were used for a SILAR process with a dipping time of 1 min each. In this work, all samples prepared were capped with three cycles of ZnS. Then, all the samples were annealed at 300 °C for 1 h. 2.3. Characterization. A model JEOL JSM-6700F fieldemission scanning electron microscopy (FESEM) was used to characterize the morphologies of the samples. Transmission electron microscope (TEM) and high-resolution TEM (HR-TEM) images were taken by a JEM-2100F high-resolution transmission microscope operating at 200 kV with a point resolution of 0.23 nm. Scanning TEM and energy-dispersive X-ray spectroscopy analyses (STEM-EDX) were taken by a FEI TECNAI F20 transmission electron microscope with an accelerating voltage of 200 kV. The crystal structure of the as-prepared films were characterized by a Rigaku D/max-2500 X-ray diffractometer with Cu Kα radiation (λ = 1.5418 Å). Optical characterization of the films was performed using a UV-3150 double-beam spectrophotometer. 2.4. Photovoltaic Measurements. The photoelectrochemical properties were probed using the conventional three-electrode system which is made of quartz cell and linked with the electrochemical workstation (CH Instruments, model CHI601C). The as-prepared film electrodes were used as the working electrode while a platinum mesh as the counter electrode and a saturated calomel electrode (SCE) as the reference electrode. The electrolyte was a mixture of 0.25 M Na2S and 0.35 M Na2SO3 aqueous solution where the Na2S works as holes scavenger and is oxidized into S22 to prevent the photocorrosion of CdSe.25 The CHI electrochemical workstation was used to measure dark and illuminated current at a scan rate of 10 mV/s. Sunlight was simulated with a 500 W xenon lamp (Spectra Physics). The light intensity was calibrated to AM 1.5G at 100 mW/cm2 by a laser power meter (BG26M92C, Midwest Group). The active area was strictly kept within 1 cm2 by coating a layer of waterproof rubberized fabric on the excess area.

3. RESULTS AND ANALYSIS 3.1. SEM Images. The FESEM images of the bare TiO2 NTWs and the TiO2/CdS/CdSe NTWs are shown in Figure 1. Figure 1A gives the top view of the as-prepared TiO2 NTWs where the wires on top of the nanotube are so loosely arranged that some bare pore openings are exposed. Since the wires are formed by chemical dissolution (etching) of oxide at the top and there is no grain boundary between the tube and the wires, the NTWs structure would inherit the advantages of both the nanotube and nanowire in charge separation, electronic transport, and light harvesting.10,26 Figure 1B is the cross-section view of the 2616

dx.doi.org/10.1021/jp209258r |J. Phys. Chem. C 2012, 116, 2615–2621

The Journal of Physical Chemistry C

Figure 1. Typical SEM images: (A) and (B) are top view and crosssection view of bare TiO2 NTWs, respectively; the inset in (B) represents a lower magnification of such array; (C) and (D) are top view and cross-section view of TiO2NTWs/CdS(4c)/CdSe(7c) electrode, respectively; (E), (F), and (G) are cross-section views of TiO2NTWs/CdS(4c)/ CdSe(3c), TiO2NTWs/CdS(4c)/CdSe(5c), and TiO2NTWs/CdS(4c)/ CdSe(7c), respectively. The samples prepared for FESEM did not cap ZnS for the sake of a better investigation on the growth of CdS and CdSe.

as-prepared TiO2 NTWs which obviously shows that the relatively smooth TiO2 NTWs arrays are uniformly formed with tube length of 2.6 2.8 μm and diameters ∼110 nm. After assembled with CdS for 4 cycles and CdSe for 7 cycles, the ordered TiO2 NTWs array structure is retained and the TiO2/CdS/CdSe NTWs electrodes with rougher surfaces are observed as shown in Figure 1C,D, which reveals that nanocrystals have covered the entire surfaces of TiO2 NTWs. The TiO2/CdS(4c)/CdSe NTWs array that assembled with CdSe for cycles from 3 to 5 to 7 are shown in Figure 1E,F,G. We can find that the amount of CdSe adsorbed on TiO2/CdS NTWs arrays increased with the SILAR cycles. 3.2. TEM and HRTEM Observation. The detailed microscopic characterization of the TiO2/CdS/CdSe/ZnS nanostructure was performed by using TEM and high-resolution TEM (HR-TEM). The typical TEM image of a TiO2 NTWs array deposited with CdS/CdSe/ZnS for 4 cycles/5 cycles/3 cycles is shown in Figure 2A where we can see that the whole surface area (including inner surface) of TiO2 NTWs array have been covered with nanoclusters. The nanoclusters have size smaller than 20 nm, and each nanocluster consists of a number of nanoparticles with diameters smaller than 10 nm. HR-TEM images of TiO2/CdS/CdSe/ZnS heterojunction region have indicated the high crystallinity of TiO2, CdS, CdSe, and ZnS. The measured lattice spacings in Figure 2B D are consistent with the d-spacings of TiO2, CdS, CdSe, and ZnS. Explicitly speaking, lattice fringes with interplanar spacings d{101} = 0.351 nm in the upper left of Figure 2B,D and d{004} = 0.237 nm in the bottom right of Figure 2D are consistent with the antase phase of TiO 2 [JCPDS no. 86-1157].

ARTICLE

Figure 2. (A)TEM image of TiO2NTWs/CdS/CdSe/ZnS electrode; (B), (C), and (D) are HR-TEM images of a TiO2NTWs/CdS/CdSe/ ZnS electrode; (E) and (F) are enlarged HR-TEM images of TiO2/CdS interface and TiO2/CdSe interface shown in (D), respectively. STEM images and corresponding STEM-EDX elemental mapping of the sample (TiO2NTWs/CdS(4 cycles)/CdSe(5 cycles)) are shown, revealing the homogeneous distribution of Cd, S, and Se elements over the whole TiO2NTWs.

While d{111} = 0.335 nm, d{200} = 0.291 nm corresponds to the cubic phase of CdS [JCPDS no. 80-0019] and d{100} = 0.359 nm to the hexagonal phase [JCPDS no. 75-1545]; d{100} = 0.372 nm, d{002} = 0.351 nm, d{101} = 0.329 nm, and d{110} = 0.215 nm correspond to the hexagonal form of CdSe [JCPDS no. 08-0459]. The wurtzite phase of ZnS with d{100} = 0.332 nm [JCPDS no. 75-1547] can also be observed even though it exists in a small amount. STEM images and corresponding STEM-EDX elemental mapping of the sample (TiO2NTWs/CdS (4 cycles)/CdSe (5 cycles)) was also shown to reveal the distribution of Cd, S, and Se elements over the whole TiO2NTWs. These results confirm that CdS, CdSe, and ZnS nanocrystals have been successfully deposited on the surface of the TiO2 NTWs. In addition, the HR-TEM images imply that there exist three dominant interfaces: TiO2/CdS, TiO2/CdSe, and TiO2/CdS/ CdSe, among which the interfacial structures of the QDs/TiO2 are identified by HRTEM as shown in Figure 2D. We can see that the {100} plane of CdS are stacked on the {101} planes of TiO2 with a certain angle, while the {101} planes of CdSe are connected to the {004} planes of TiO2 via lattice distortion area. In other words, the CdS and CdSe have grown on the TiO2 NTWs. Figure 2E,F shows the magnified HRTEM images for the marked 2617

dx.doi.org/10.1021/jp209258r |J. Phys. Chem. C 2012, 116, 2615–2621

The Journal of Physical Chemistry C

Figure 3. XRD patterns of (a) TiO2 NTWs, (b) TiO2/CdS/ZnS electrode, and (c) TiO2/CdS/CdSe/ZnS electrode.

regions in Figure 2D. In Figure 2E, misfit dislocation and moire fringes on the upper part of the image were observed. The moire fringes are due to the stacking fault that caused by the overlapping of CdS and TiO2 lattice, which implies the interface plane is not perpendicular to the observation plane. However, in return, the misfit dislocation and moire fringes would relax the lattice mismatch and thus reduce he interfacial lattice misfit dramatically,27,28 which leads to a less-defective interface between {100}CdS and {101}TiO2. In Figure 2F, we observed lattice distortion is created between {101}CdSe and {004}TiO2. We know that the lattice spacings are 0.329 and 0.237 nm, respectively. This lattice misfit is so enormous that the interface adhesion affinity cannot compensate the corresponding strain, and rows of atoms can therefore leave their adsorption sites and create the interface lattice distortion. This mechanism was also found in the Au/TiO2 system.29 In contrast to the two interfacial structures, the decreased density of trap states that act as recombination center and lead to faster electron transport at the interface of TiO2/CdS than TiO2/ CdSe, which explains the higher electron injection efficiency in TiO2/CdS.23 However, compared with the simple adsorption interface, the growth interfaces (heterojunction) of QDs/TiO2 result in more charge transfer channel and more effective coverage, which can enhance the electronic injection. More importantly, the QD/NTW and QD/QD interfaces were grown without noticeable contamination from organic solvent residues and thus reduced interfacial carrier loss is expected. 3.3. Phase Structure. Figure 3 shows the XRD patterns for both bare TiO2 NTW arrays and CdS/CdSe/ZnS sensitized TiO2 NTW arrays. The bare annealed TiO2 NTW arrays have crystal planes of anatase and rutile phases. After the sensitization of CdS nanocrystals, no fresh evident diffraction peaks are observed for the little amount of CdS QDs and their highly dispersion on TiO2 surface. The annealed CdS/CdSe/ZnS cosensitized TiO2 NTW arrays exhibit new small peaks at 2θ of ca. 23.93°, 41.99°, and 49.86°. These peaks can be indexed to the {100} and {110} planes of the hexagonal-phase CdSe [JCPDS No. 08-0459] and {311} planes of the cubic-phase CdSe [JCPDS No. 19-0191], respectively, which are consistent with the HRTEM results above. The peaks of rutile TiO2 at 25.18° are evidently broadened which is attributed to the new peak at 25.38° derived from the {111} crystal planes of cubic CdSe. The patterns for the annealed TiO2/CdS/CdSe/ZnS electrode confirm that CdSe is mixed structure of metastable cubic phase and stable hexagonal phase when annealed to 300 °C.

ARTICLE

Figure 4. UV vis absorption spectra of (a) TiO2NTWs, (b) TiO2/ CdS(4c)/ZnS(3c), (c) TiO2/CdS(4c)/CdSe(1c)/ZnS(3c), (d) TiO2/ CdS(4c)/CdSe(3c)/ZnS(3c), (e) TiO2/CdS(4c)/CdSe(5c)/ZnS(3c), and (f) TiO2/CdS(4c)/CdSe(6c)/ZnS(3c). For its wide band gap (3.8 eV), ZnS is not thought contribute to the absorption edges.

It is noteworthy that the XRD patterns show no peaks corresponding to CdO or SeO2, which indicates the thermal stability of CdS and CdSe nanocrystals and the high purity of our sample. 3.4. UV vis Absorption Spectroscopy. The variation of UV vis spectra with SILAR cycles obtained in this work are shown in Figure 4, from which enhanced absorption of visible light by TiO2/CdS/CdSe/ZnS heterostructure can be confirmed. The maximum peak of the TiO2 NTWs film occurs at around 370 nm, and no significant absorbance for visible light can be seen because of its large energy gap (3.2 eV). However, the UV vis spectra of TiO2/CdS/ZnS exhibit broad absorption bands from 350 to 530 nm, while in the UV vis spectra of TiO2/CdS/ CdSe/ZnS, broader bands appear from 350 to 730 nm. This variation indicates that the deposition of QDs has significantly extended the photoresponse of TiO2 electrode into the visible light region. The enhanced ability to absorb visible light makes this type of QDs/TiO2 promising applications in photovoltaic devices. Furthermore, the absorption intensity increases with the number of SILAR cycles, which indicates the increase of QDs amount simultaneously, and this is consistent with the results of SEM. We also found that the absorption edge showed significant red-shifts when CdS and CdSe further modified, and this is owing to the decreased band gap of deposited semiconductors. When it comes to CdSe, the slightly red-shifts with increasing SILAR cycles correspond to the growth of the CdSe particles. As known from the quantum-confinement effect, when the particle size is reduced to the quantum size, the number of atoms in a particle will enormously decrease and a larger interval between the energy level will emerge. Therefore, the energy level will change from quasi-continuous phase to split phase, and the energy gap will become so wide that the blue-shift in the absorption occurs. It is thus thought that the growth of CdSe particles may be responsible for the red-shift of the absorption edge. It is also noteworthy that the absorption edges are ca. 530 nm for TiO2/CdS(4c)/ZnS(3c) while the corresponding band gap is 2.34 eV, which is higher than that reported for bulk CdS (2.25 eV). This difference reveals that the sizes of the CdS on the TiO2 NTWs array films in our work are still within the scale of QD (less than 6 nm).8 For comodification, the absorption edges, mainly represent that of CdSe, are ca. 726 nm (1.71 eV) for TiO2/CdS(4c)/CdSe(3c)/ZnS(3c), ca. 735 nm (1.69 eV) for TiO2/CdS(4c)/CdSe(5c)/ZnS(3c), and ca. 750 nm (1.65 eV) for TiO2/CdS(4c)/CdSe(6c)/ZnS(3c). 2618

dx.doi.org/10.1021/jp209258r |J. Phys. Chem. C 2012, 116, 2615–2621

The Journal of Physical Chemistry C

ARTICLE

Figure 5. Photocurrent voltage (J V) curves of photoelectrodes with 7 cycles of CdSe and 3 cycles of ZnS as overlayers: (a) dark current; (b) bare TiO2 NTWs; (c) TiO2/CdS(1c)/CdSe(7c)/ZnS(3c); (d) TiO2/ CdS(3c)/CdSe(7c)/ZnS(3c); (e) TiO2/CdS(4c)/CdSe(7c)/ZnS(3c); (f) TiO2/CdS(6c)/CdSe(7c) /ZnS(3c). The active area was strictly kept within 1 cm2, and the photoelectrodes were measured versus SCE under simulated sunlight with an illumination intensity of 100 mW/cm2 in 0.25 M Na2S and 0.35 M Na2SO3.

Table 1. Parameters Obtained from the Photocurrent Density Voltage (J V) Measurements of Multiple Semiconductors Cosensitized TiO2 NTWs Solar Cells for Different CdS Cycles with 7 Cycles CdSe and 3 Cycles of ZnS Coating electrode (CdS/CdSe/ZnS)

Jsc (mA/cm2)

Voc (V)

FF

η (%)

0c/0c/0c

0.36

0.91

0.268

0.086

1c/7c/3c

3.50

0.92

0.399

3c/7c/3c

4.09

0.94

0.514

4c/7c/3c 6c/7c/3c

3.91 3.39

0.95 0.94

0.567 0.545

Figure 6. Photocurrent voltage (J V) curves of TiO2/CdS(4c)/ CdSe/ZnS(3c) photoelectrodes: (a) dark current; (b) bare TiO2 NTWs; (c) TiO2/CdS(4c)/CdSe(0c)/ZnS(3c); (d) TiO2/CdS(4c)/ CdSe(2c)/ZnS(3c); (e) TiO2/CdS(4c)/CdSe(4c)/ZnS(3c); (f) TiO2/ CdS(4c)/CdSe(5c)/ZnS(3c); (g) TiO2/CdS(4c)/CdSe(7c)/ZnS(3c). The active area was strictly kept within 1 cm2, and the photoelectrodes were measured versus SCE under simulated sunlight with an illumination intensity of 100 mW/cm2 in 0.25 M Na2S and 0.35 M Na2SO3.

Table 2. Parameters Obtained from the Photocurrent Density Voltage (J V) Measurements of Multiple Semiconductors Cosensitized TiO2 NTWs Solar Cells for Different CdSe cycles with 4 Cycles Underlayer of CdS and 3 Cycles Capper of ZnS electrode (CdS/CdSe/ZnS)

Jsc (mA/cm2)

Voc (V)

FF

η (%)

1.283

0c/0c/0c

0.36

0.91

0.268

0.086

1.977

4c/0c/3c

1.31

1.07

0.574

0.802

2.095 1.728

4c/2c/3c

2.31

0.96

0.567

1.257

4c/4c/3c

3.30

1.00

0.543

1.788

4c/5c/3c

4.30

0.96

0.583

2.408

4c/7c/3c

3.91

0.95

0.567

2.095

By comparing them with the values for bulk CdSe (1.70 eV), we can know that the average CdSe particles have exceed the size of QD (1.5 4.5 nm)30 when the SILAR cycle reaches more than 5. These results are in agreement with findings from the HRTEM images. 3.5. Photovoltaic Performance of the Electrodes. In this work, we studied the effect of CdS/CdSe/ZnS cosensitization on TiO2 NTWs and found the electrode prepared by using 4 cycles of CdS plus 5 cycles of CdSe with 3 cycles ZnS as capper(TiO2/ CdS(4c)/CdSe(5c)/ZnS(3c)) exhibits the best performance. The optimal cycle found for the CdSe-sensitized TiO2 NTWs electrodes is about 7 (see Figure 1 in Supporting Information). So we kept CdSe at 7 cycles first when we fabricated TiO2/CdS/ CdSe/ZnS(3c) photoelectrodes, and the corresponing photocurrent voltage (J V) curves are shown in Figure 5. The opencircuit potential (Voc), short-circuit current density (Jsc), fill factor (FF), and the total energy conversion efficiency (η) of these cells are listed in Table 1. The observed dark current density for both the TiO2 and the TiO2/CdS/CdSe/ZnS electrodes are negligible. In Figure 5, the photocurrent of TiO2/CdS/CdSe(7c)/ZnS(3c) electrodes changes little with increasing SILAR cycles of CdS, which indicates that high incorporated amount of CdSe dominates the absorbance of visible light and induces a higher current density. The highest photocurrent is 4.09 mA/cm2

with 3 cycles of CdS and 7 cycles of CdSe; then the output is found to somewhat decreased with more cycles, which may be attributed to that excess nanoparticles goes against electron transfer and would cause blockage in NTWs. On the other hand, the FF of TiO2/CdS/CdSe(7c)/ZnS(3c) electrodes shown in Table 1 is improved with increasing CdS cycles, which arise from the increasing electron injection efficiency. With increasing amount of CdS, the effective coverage area of CdS on TiO2 increased and that of CdSe decreased, leading to a higher electron injection efficiency and larger FF in the composite structure. However, the highest FF and efficient was found to be electrode with 4 cycles of CdS, so we chose 4 cycles as the best cycle of CdS for the TiO2/CdS/CdSe/ ZnS cascade photoelectrode. Then we fabricated electrodes with 4 cycles of CdS as underlayer and 3 cycles of ZnS as capper, and their photocurrents are shown in Figure 6. The open-circuit potential (Voc), short-circuit current density (Jsc), fill factor (FF), and the total energy conversion efficiency (η) of these cells are listed in Table 2. Significant improvement can be found with increasing CdSe cycles, and we think this is due to the increasing amount of CdSe as main absorber of visible light. Moreover, because of energy level alignment that occurred between CdS and CdSe in the TiO2/CdS/CdSe/ZnS 2619

dx.doi.org/10.1021/jp209258r |J. Phys. Chem. C 2012, 116, 2615–2621

The Journal of Physical Chemistry C Scheme 1. Schematic Diagrams Illustrating (a) the Architecture, (b) Relative Fermi Level and Band-Edge Positions of TiO2, CdS, and CdSe in Bulk; Proposed Band-Edge Structures and Fermi Level Alignment for the Electrodes of TiO2/ CdS/CdSe (c) and TiO2/CdSe (d) in Equilibrium with the Redox Couples in the Electrolytea

a

The dashed lines are Fermi level of sigle semiconductor or composite sructures. For the sake of a clear understanding of the effect of CdS and CdSe modification, we do not show ZnS in this schematic.

heterojunction electrodes,23 the stepwise band-edge structure would create efficient charge transfer channel and would trigger a higher resistance to transport excited electrons back to the electrolyte, and this results in a higher performance of the cosensitized electrode. Besides, the 4 cycles of CdS cannot cover the whole surface of TiO2 as discussed in the HR-TEM observation, so the increase of TiO2/ CdSe interface will aggrandize the effective cover area and further improve the photocurrent. However, the output is found to somewhat decreased when the deposition are increased to 5 cycles. This phenomenon may be attributed to that more deposition cycles would cause conglomeration and growth of the CdSe crystal nucleus. Then the grain boundaries between excess CdSe nanoparticles can act as potential barrier for charge transfer, and the oversized CdSe particles will lose the dominance as QDs (large QD extinction coefficients and generating multiple electron hole pairs). Those hindrance factors are dominant over the aggrandizement of effective cover area and cause a lower current density. In addition, we observed that Voc increased with CdS modification and then decreased with CdSe sensitization but still kept higher than that of TiO2 in Table 2. This variation trends can be qualitatively evaluated from the shift of the Fermi level (Ef) of CdS and CdSe due to their contact. The alteration of the Ef after CdS and CdSe modification is shown in Scheme 1. Compared with the bare TiO2, a shift in Ef to more negative potential happens when TiO2 coupled with 4 cycles of CdS nanoclusters, which results in a movement of Voc from 0.91 V (vs SCE) to 1.07 V (vs SCE). When CdSe comodified, energy realignment occurs in TiO2/CdS/CdSe composite electrodes23 as shown in Scheme 1c. But the Ef of CdSe is less negative than that of CdS even after this energy level realignment,31 which results in the Ef movement of TiO2/CdS(4c)/CdSe/ZnS(3c) composite system to less negative potential and thus a lower Voc than the TiO2/CdS(4c)/ZnS(3c) composite system. Because Ef of

ARTICLE

both CdSe and CdS are all more negative than bare TiO2, so Voc of TiO2/CdS(4c)/ZnS(3c) and TiO2/CdS(4c)/CdSe/ ZnS(3c) would keep larger than that of bare TiO2. Our device configuration is depicted in Scheme 1. Comparing to that in bare TiO2 NTWs, obvious enhancement on the photoelectrochemical property has been achieved in the CdS, CdSe, and ZnS cosensitized TiO2 NTWs array with Jsc of 4.30 mA/cm2 and Voc of 1.07 V (vs SCE). We here propose several reasons for the enhancement: First, the morphology of TiO2 NTWs keeps the advantage of the tube such as efficient charge separation and transport properties as well as superior light harvesting efficiency while the wires that exposed in the electrolyte favor more nanoparticles. Second, compared with bare TiO2 NTWs electrode the TiO2/CdS/CdSe/ZnS electrodes have an intense absorption in the visible region, which greatly raised the utilization rate of the solar energy. Third, the stepwise band-edge structure would create efficient charge transfer channel and would trigger a higher resistance to transport excited electrons back to the electrolyte, resulting in a higher performance of the cosensitized electrode. Fourth, the fine heterojunction formed between CdS and TiO2 is helpful for CdS to collect excited electrons from CdSe to TiO2 in the stepwise band-edge structure. Lastly, CdSe has broader spectral response than CdS, and the existence of TiO2/CdSe interface will aggrandize the effective cover area and consequently improve the photocurrent. In Scheme 1d, the upward shifts of the band edges for CdSe which favor charge separation arises from the Fermi level realignment. We know that when a semiconductor is brought into contact with a solution containing a redox couple, the Fermi levels of the semiconductor and solution will be identical after electrostatic equilibrium and thus cause downward or upward shifts of the band edges for the semiconductor.

4. CONCLUSIONS The CdS and CdSe nanoclusters cosensitized TiO2 NTWs array films have been fabricated by SILAR deposition. The photoelectrodes have been characterized by means of different techniques, and their photoelectrochemical properties were investigated carefully. It is shown that the CdS and CdSe nanoclusters covered both inner and outer surfaces of TiO2 NTWs array efficiently. As verified by the interfacial structure of QDs/TiO2, the close contact between the sensitizer and the TiO2 NTWs favors electronic injection. The TiO2 NTWs allow growth of more nanoclusters which can be induced to generate more photoelectrons by incident light and can efficiently separate and transfer photogenerated electrons from nanoclusters to the collecting Ti substrates. The stepwise band-edge structure in the TiO2/CdS/CdSe/ZnS heterojunction electrode would create efficient charge transfer channel and would trigger a high resistance to transport excited electrons back to the electrolyte, while the existence of TiO2/ CdSe interface will aggrandize the effective cover area; consequently, the photocurrents are improved. On the basis of these advantages, we found the photoelectrode has an optimum shirtcircuit current density of 4.30 mA/cm2 and corresponding energy conversation efficiency of 2.408% under AM 1.5G illumination. The photoanodes prepared in this work show photoelectrochemical properties superior to the photoelectrodes that comprised of single QD and TiO2 nanotubes with length of tens of micrometers,24 which suggests that the relative short TiO2 NTWs array films (2.6 2.8 μm) cosensitized by CdS, CdSe, and ZnS nanoclusters have a potential application in solar cells. 2620

dx.doi.org/10.1021/jp209258r |J. Phys. Chem. C 2012, 116, 2615–2621

The Journal of Physical Chemistry C

’ ASSOCIATED CONTENT

bS

Supporting Information. Figure 1. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

ARTICLE

(28) Majimel, J.; Bacinello, D.; Durand, E.; Vallee, F.; elapierre, M. T. Langmuir 2008, 24, 4289–4294. (29) Majimel, J.; Lamirand-Majimel, M.; Moog, I.; Feral-Martin, C.; Delapierre, M. T. J. Phys. Chem. C 2009, 113, 9275–9283. (30) Yu, W. W.; Qu, L. H.; Guo, W. Z.; Peng, X. G. Chem. Mater. 2003, 15, 2854–2860. (31) Lee, Y. L.; Chi, C. F.; Liau, S. Y. Chem. Mater. 2010, 22, 922– 927.

*E-mail: [email protected]. Tel: +86-431-85168763. Fax: +86-43185168763.

’ ACKNOWLEDGMENT The authors gratefully acknowledge the financial support of Science and Technology Development Program of Jilin Province (20110417). ’ REFERENCES (1) Fujishima, A.; Honda, K. Nature 1972, 238 (5358), 37–38. (2) Heller, A. Science 1984, 223 (4641), 1141–1148. (3) Hensel, J.; Wang, G.; Li, Y.; Zhang, J. Z. Nano Lett. 2010, 10, 478–483. (4) Atwater, H. A.; Polman, A. Nature Mater. 2010, 9, 205–213. (5) Kamat, P. V. J. Phys. Chem. C 2007, 111, 2834–2860. (6) Liu, J.; Cao, G. Z.; Yang, Z. G.; Wang, D. H.; Dubois, D.; Zhou, X. D.; Graff, G. L.; Pederson, L. R.; Zhang, J. G. ChemSusChem 2008, 1, 676–697. (7) Shen, Q.; Yamada, A.; Tamura, S.; Toyoda, T. Appl. Phys. Lett. 2010, 97, 123107. (8) Chen, H.; Fu, W. Y.; Yang, H. B.; Sun, P.; Zhang, Y. Y.; Wang, L. R.; Zhao, W. Y.; Zhou, X. M.; Zhao, H.; Jing, Q.; Qi, X. F.; Li, Y. X. Electrochim. Acta 2010, 56, 919–924. (9) Shankar, K.; Basham, J. I.; Allam, N. K.; Varghese, O. K.; Mor, G. K.; Feng, X.; Paulose, M.; Seabold, J. A.; Choi, K. S.; Grimes, C. A. J. Phys. Chem. C 2009, 113, 6327–6359. (10) Lim, J. H.; Choi, J. Small 2007, 3, 1504–1507. (11) Roy, P.; Berger, S.; Schmuki, P. Angew. Chem., Int. Ed. 2011, 50, 2904–2939. (12) Hong, X. T.; Wang, Z. P.; Cai, W. M.; Lu, F.; Zhang, J.; Yang, Y. Z.; Ma, N.; Liu, Y. J. Chem. Mater. 2005, 17, 1548–1552. (13) Irie, H.; Watanabe, Y.; Hashimoto, K. J. Phys. Chem. B 2003, 107, 5483–5486. (14) Mor, G. K.; Shankar, K.; Paulose, M.; Varghese, O. K.; Grimes, C. A. Nano Lett. 2006, 6, 1789–1792. (15) Hsiao, P. T.; Tung, Y. L.; Teng, H. J. Phys. Chem. C 2010, 114, 6762–6769. (16) Yang, Z.; Chen, C. Y.; Roy, P.; Chang, H. T. Chem. Commun. 2011, 47, 9561–9571. (17) Lee, Y. L.; Lo, Y. S. Adv. Funct. Mater. 2009, 19, 604–609. (18) Wang, H.; Bai, Y. S.; Zhang, H.; Zhang, Z. H.; Li, J. H.; Guo, L. J. Phys. Chem. C 2010, 114, 16451–16455. (19) Schaller, R. D.; Klimov, V. I. Phys. Rev. Lett. 2004, 92, 186601. (20) Guijarro, N.; Villarreal, T. L.; Sero, I. M.; Bisquert, J.; Gomez, R. J. Phys. Chem. C 2009, 113, 4208–4214. (21) Blackburn, J. L.; Selmarten, D. C.; Ellingson, R. J.; Jones, M.; Micic, O.; Nozik, A. J. J. Phys. Chem. B 2005, 109, 2625–2631. (22) Gao, X. F.; Li, H. B.; Sun, W. T.; Chen, Q.; Tang, F. Q.; Peng, L. M. J. Phys. Chem. C 2009, 113, 7531–7535. (23) Chi, C. F.; Cho, H. W.; Teng, H.; Chuang, C. Y.; Chang, Y. M.; Hsu, Y. J.; Lee, Y. L. Appl. Phys. Lett. 2011, 98, 012101. (24) Lee, H. J.; Bang, J.; Park, J.; Kim, S.; Park, S. M. Chem. Mater. 2010, 22, 5636–5643. (25) Rao, N. N.; Dube, S. Int. J. Hydrogen Energy 1996, 21 (2), 95–98. (26) Feng, X. J.; Macak, J. M.; Schmuki, P. Chem. Mater. 2007, 19, 1534–1536. (27) Kehagias, Th.; Komninou, Ph.; Nouet, G.; Ruterana, P.; Karakostas, Th. Phys. Rev. B 2001, 64, 195329. 2621

dx.doi.org/10.1021/jp209258r |J. Phys. Chem. C 2012, 116, 2615–2621