Introduction

Conversion of solar energy to chemical or electric energy by semiconductor nanomaterials has attracted considerable interest1,2,3. Conventional dye or quantum dot sensitized solar cells are fabricated using mesoporous films which have large surface area4. However, a major drawback of the mesoporous nanoparticle photoanode is the inefficient electron transport due to defects such as grain boundaries and surface traps5. For instance, the electron diffusion coefficient of TiO2 nanoparticles is more than 2 orders of magnitude lower than that of TiO2 single crystals6. On the other hand, high surface area and efficient electron diffusion are both essential to improve the efficiency of solar energy conversion devices.

Over the past decades, zero-dimensional (0D), one-dimensional (1D) and two-dimensional (2D) nanostructures have attracted much attention due to their unique structure-dependent properties7,8,9. Semiconductors with various tailored morphologies, such as nanoparticles (NPs), nanorods (NRs) and nanosheets (NSs) have been extensively studied for photoelectric applications. Furthermore, heterogeneous nanostructures composed of two or three compounds, usually exhibit superior properties compared with that of a single component10,11. Therefore, optimal-assembling of multi-component semiconductor nanostructures is an attractive approach for designing next generation solar cells. Up to date, most of the work focused on utilizing one or two kinds of nanostructures to improve the photoelectric conversion performance, such as nanorod/nanoparticle, nanowire/nanosheet et al12,13,14. To simultaneously increase the transport efficiency of carriers as well as the surface area, one can assemble hierarchical 0D, 1D and 2D nanostructures on the electrode.

Here we developed a hierarchical heterostructure with multiple components (2D TiO2 NSs, 1D ZnO NRs and 0D CdS NPs) on FTO for efficient solar energy conversion. TiO2 and ZnO were chosen in our design due to their low-cost, outstanding chemical stability and environmentally friendly nature. Both of them are technologically important and have been widely used in the fields of photocatalysis, solar cells and self-cleaning devices. The integration of ZnO with TiO2 will facilitate the photo-generated charge separation and give rise to longer carrier lifetime15. More importantly, 2D single-crystal TiO2 NSs with reactive (001) facet exhibit superior photocatalytic activity and could also improve the photoelectric conversion performance16,17,18,19,20. Meanwhile, 1D ZnO NRs can significantly improve the electron transport by providing a direct conduction pathway and reducing the number of interparticle hopings21,22,23,24. However, the relatively large band gap of 3.2 eV limits further application of TiO2 and ZnO in the visible light region. CdS, with a narrow direct band-gap (2.4 eV) and flat band potential at 0.66 V (pH 7)25, can be used as a photosensitizer for the wide band-gap semiconductors TiO2 and ZnO. In this work, we synthesize a hierarchical heterostructure, that is, 2D TiO2 (001) NSs coated with well-aligned 1D ZnO nanorod arrays, on which 0D CdS NPs are anchored. This is referred to as A-TiO2/ZnO/CdS, which combines the advantages of TiO2, ZnO and CdS. In this optimized heterostructure, 0D CdS NPs gather the visible light, while 1D ZnO NRs arrays which has a high surface area facilitate sufficient CdS deposition and simultaneously harvest UV-light to generate photoelectrons. The 2D TiO2 NSs work as the efficient charge collection and transport path, meanwhile, provide more sites for ZnO deposition. As a reference, a rutile-TiO2 NRs/ZnO-NRs/CdS (1D/1D/0D), named as R-TiO2/ZnO/CdS, is also fabricated and the influence of different morphologies of the heterogeneous architectures on photoelectric performance is systematically studied.

Results

Morphology and key processes of A-TiO2/ZnO/CdS

The fabrication process of CdS sensitized A-TiO2/ZnO heterostructure on FTO glass substrate is shown in Figure 1. The 2D single-crystal TiO2 NSs were grown on FTO by a one-step hydrothermal method26. The ZnO NRs arrays were deposited on the surface of TiO2 via a seed-mediated method27, while the seed layer was pre-deposited by Pulsed Laser Deposition (PLD). After that, CdS was deposited on A-TiO2/ZnO by successive ion-layer adsorption and reaction (SILAR) technique28, which is based on the successive surface adsorption of ions and thus proceeds via a layer-by-layer buildup of the film29.

Figure 1
figure 1

Schematic illustration of the fabrication process of CdS sensitized A-TiO2/ZnO heterostructure on FTO glass substrate.

Full size image

As shown in Figure 2a, the single-crystalline TiO2 (001) NSs, with side length of ~2 μm and thickness of ~100 nm, grow vertically or obliquely on the FTO substrate (Figure S1). The XRD pattern (Figure 2d) shows that the products are pure anatase phase of TiO2 with lattice constant a = 3.7852 Å (JCPDS No. 21-1272). Compared with the powder diffraction pattern, the (004) diffraction peak is significantly enhanced, indicating the exposing of large percentage of {001} facet. After the deposition of ZnO seed layer by PLD, well-aligned ZnO NRs-arrays with average diameter of 60 nm and length of 800 nm were vertically grown on the TiO2 {001} and {101} facets, respectively (Figure S2). Notably, the low-magnification SEM images (Figure S2a) show that the A-TiO2/ZnO heterostructure retain the original morphology of TiO2 NSs. Such kind of well-aligned ZnO arrays will facilitate efficient photogenerated-electron transfer from ZnO to TiO2 surface. The XRD pattern (Figure 2d) shows that all the diffraction peaks of ZnO nanorods can be indexed to a pure wurtzite hexagonal phase of ZnO.

Figure 2
figure 2

SEM images and XRD patterns of the nanostructures obtained during the growth of hierarchical TiO2/ZnO heterostructure on FTO.

(a) Growth of anatase TiO2 (001) tetragonal nanosheet arrays on FTO (A-TiO2), (b, c) Deposition of ZnO nanorod arrays on TiO2 (A-TiO2/ZnO). (d) XRD patterns of FTO, A-TiO2 and A-TiO2/ZnO.

Full size image

Figure 3a shows a typical SEM image of ZnO NRs arrays after the deposition of CdS via SILAR process. It can be found that the surface of ZnO NRs is homogeneously covered by a thin layer of CdS. The coating of CdS NPs on ZnO NRs as shown in Figure 3a is further characterized with energy dispersive X-ray spectroscopy (EDS). The EDS spectrum in Figure 3b reveals that the nanocomposite consists of zinc, cadmium and sulphur. The titanium can also be detected as the substrate of ZnO NRs arrays.

Figure 3
figure 3

Morphological characterization of A-TiO2/ZnO/CdS.

(a) SEM images and (b) EDS spectra of CdS sensitized A-TiO2/ZnO heterostructure (A-TiO2/ZnO/CdS). (c) TEM and (d) HRTEM images of a single CdS/ZnO nanorod.

Full size image

The ZnO/CdS core-shell structure is further demonstrated with transmission electron microscopy (TEM) images in Figure 3c, which show evidence of morphological changes of the ZnO NRs upon CdS deposition, such as significant surface roughening and diameter increase. The single-crystalline core (Figure 3d) shows a fringe spacing of 0.26 nm, which matches well with the d-spacing of the (0002) plane of hexagonal ZnO, confirming that the ZnO nanorods are preferentially oriented in the c-axis direction. A lattice fringe spacing of 0.21 nm in the crystallites matches well with the interplanar spacing of the (220) plane of cubic CdS, which demonstrates that the CdS NPs deposited on the surface of ZnO NRs are well-crystallized and the interface between them is of high quality. This will facilitate the photo-generated electron transfer from CdS to ZnO. The XRD pattern of A-TiO2/ZnO/CdS (Figure S3d) does not show the diffraction peak of cubic CdS even though they are well-crystallized. One possible reason is the overlap between the (111) diffraction peak of CdS and the (110) diffraction peak of SnO2 (FTO). Another reason is the amount and particle size of CdS NPs is too small to be detected.

Synthesis and characterization of R-TiO2/ZnO/CdS

To systematically study the influence of different morphologies of the heterogeneous architectures on the photoelectric performance, a reference sample was fabricated and characterized. Figure S4 shows the fabrication process of CdS sensitized R-TiO2 NRs/ZnO NRs heterostructure (R-TiO2/ZnO/CdS) on FTO glass substrate, which is similar to that of A-TiO2/ZnO/CdS, except that rutile TiO2 NRs arrays (R-TiO2) are grown on FTO30 at the beginning instead of A-TiO2 NSs arrays.

Figure S5 shows typical SEM images of the R-TiO2 NRs arrays grown on FTO surface. The images at different magnifications reveal that the entire surface of the FTO substrate is uniformly covered with TiO2 NRs. Tetragonal NRs with square top facets are nearly perpendicular to the FTO substrate and the average diameter is 90 nm. The XRD patterns of the sample show that all the diffraction peaks match well with the tetragonal rutile phase. Absence of some diffraction peaks that are normally present in polycrystalline samples demonstrates that the nanorods are not only aligned but are also single crystalline throughout their length. After the deposition of ZnO seed layer by PLD, ZnO NRs-arrays with average diameter of ~240 nm were vertically grown on TiO2 NRs film (Figure S6). The XRD pattern (Figure S6d) clearly shows the diffraction peaks of wurtzite ZnO and rutile TiO2.

Figure S7 shows the SEM images of R-TiO2/ZnO/CdS heterostructure obtained with different SILAR deposition cycles. The CdS layer deposited on top of the R-TiO2/ZnO after 25 SILAR cycles exhibits a flat surface, which is due to the relative smaller surface area of R-TiO2/ZnO (top region for CdS deposition) compared with A-TiO2/ZnO substrate. Further increase the SILAR deposition to 50 cycles leads to cluster growth and echinus-like CdS microstructures. Thus, for R-TiO2/ZnO/CdS, the interface area between ZnO and CdS is much smaller compared with A-TiO2/ZnO/CdS heterostructure, while the latter will offer efficient charge transfer between CdS and ZnO. Furthermore, the large exposed surface area of A-TiO2/ZnO/CdS will generate efficient charge transfer between the electrolyte and the photoanodes.

Discussion

UV-vis spectra of the A-TiO2/ZnO/CdS, the R-TiO2/ZnO/CdS and the arrays of bare TiO2 on FTO substrates were characterized (Figure 4a). Compared with the R-TiO2 NRs arrays (red dotted line), the A-TiO2 (001) NSs arrays have a stronger absorption in the UV–Vis region (blue dotted line), which is due to its multiple reflection structure of the NSs-arrays. To eliminate the influence of film thickness on the light harvesting efficiency, the cross-sectional morphology was characterized (Figure S8) and the result shows that the film thicknesses of these two different structures are nearly the same. The hierarchical A-TiO2/ZnO/CdS also exhibit a significantly improved absorption in the visible region compared with the R-TiO2/ZnO/CdS, therefore, are expected to provide a more efficient light harvesting and an enhanced photoelectric performance.

Figure 4
figure 4

Photoelectric properties of hierarchical nanostructures on FTO substrates.

(a) UV–Vis diffuse reflectance spectra of the samples. (b) J–V curves of R-TiO2, A-TiO2, A-TiO2/ZnO in darkness and under illumination. (c) I−V characteristics of A-TiO2/ZnO/CdS and R-TiO2/ZnO/CdS solar cells in darkness and under illumination. (d) Schematic illustration of band energy positions and the charge transfer process of the TiO2/ZnO/CdS heterostructure.

Full size image

Figure 4b shows the J–V curves of R-TiO2, A-TiO2, A-TiO2/ZnO in dark and under illumination. The A-TiO2 NSs arrays exhibits a significantly enhanced (3.4 times in photocurrent at 0.8 V) photoelectric response compared with the R-TiO2, which is due to its highly reactive (001) facet and good crystallinity. When combined with ZnO NRs arrays, the photocurrent density at 0.8 V is about 4.9 times that of the bare A-TiO2, indicating that the TiO2/ZnO heterojunction can facilitate the photo-generated charge separation and hence improve the photoelectric performance. With Pt counter-electrodes and polysulfide electrolyte, we obtained a simple solar cell and characterized the photoelectric performance (Figure 4c). When using the A-TiO2/ZnO/CdS as the photoanode, a short-circuit photocurrent (Jsc) of 3.24 mA/cm2 is observed, which is much higher than that of the R-TiO2/ZnO/CdS photoanode (1.30 mA/cm2). The cell with A-TiO2/ZnO/CdS as the photoanode shows a fill factor (FF) and efficiency of 0.17 and 0.51%, respectively, which exhibit a significant increase when compared with that of the R-TiO2/ZnO/CdS photoanode (0.11 and 0.13%). The improved interfacial charge transfer, efficient photo-generated electron collection from the 0D to 2D nanostructures, as well as increased light absorption are the main reasons for the enhanced performance.

Moreover, band position calculations (see detail in supporting information) suggest that TiO2, ZnO and CdS have the staggered Z-scheme energy potentials that can reduce the recombination of the photogenerated carriers within the three semiconductors. Under solar light irradiation, the 0D CdS NPs are excited and the as-photoinduced electrons migrate to the 1D ZnO, which could be subsequently collected by the 2D TiO2 NSs and finally migrate to the FTO substrate. Meanwhile, the redox electrolyte (sulfide/polysulfide) scavenges the holes and thus ensures regeneration of the CdS.

To further investigate the role of each component, the photoelectric performance was characterized under different light irradiation. As shown in Figure S9, under visible light irradiation, the sample still shows photoelectric response, which is arises from the CdS NPs. Under the full spectrum light irradiation, the photocurrent density is higher than that under UV light, indicating that such a hierarchical composite could utilize both UV and visible band of solar light.

Photocatalytic H2 production from water using semiconductor photocatalysts is an attractive process for solving energy and environmental problems. As shown in Figure 5a, the hierarchical heterostructures exhibit efficient H2 evolution. Under visible light irradiation, the H2 evolution rate of A-TiO2/ZnO/CdS is 13.3 μmol h−1cm−2, which is 2.6 times that of R-TiO2/ZnO/CdS. The higher activity under UV-visible light irradiation is due to the excitation of all the three components as well as enhanced charge separation. Figure 5c displays the stability of A-TiO2/ZnO/CdS for H2 evolution under UV-visible light irradiation. There is no obvious activity decrease after each run, indicating the good stability of our sample. Furthermore, the action spectra were obtained using monochromatic light at intensity of 4.3 mW cm−2. The quantum efficiency (QE) at each centered wavelength of the monochromatic light was calculated from the ratio of twice the number of H2 molecules to the number of incident photons using the following equation: QE = (2 × the number of H2 molecules/the number of incident photons) × 100%. Figure 5b and 5d show the action spectra of H2 evolution and quantum efficiencies. For both samples, the H2 evolution rate and quantum efficiencies decreased obviously at around 520 nm with band-pass filters, revealing that the visible-light response for H2 evolution is due to the band-gap transition of CdS31. The QE of H2 evolution for A-TiO2/ZnO/CdS obtained using a 460-nm band-pass filter was still as high as 5.5%, which is much higher than that of R-TiO2/ZnO/CdS (2.1%).

Figure 5
figure 5

Photocatalytic H2 evolution.

(a) H2 evolution rate of A-TiO2/ZnO/CdS and R-TiO2/ZnO/CdS film under visible (λ > 420 nm) and UV-visible light irradiation. (c) H2 evolution over A-TiO2/ZnO/CdS during repeated photocatalytic experiments under UV-visible light irradiation. Action spectra of H2 evolution and quantum efficiency (QE) over (b) A-TiO2/ZnO/CdS and (d) R-TiO2/ZnO/CdS.

Full size image

In summary, we have developed a well-aligned, structurally-optimized heterostructure with multiple components (2D TiO2 NSs, 1D ZnO NRs and 0D CdS NPs) for efficient solar energy conversion. The structure−property relationship is systematically studied by comparing with a reference photoanode, R-TiO2 NRs/ZnO NRs/CdS NPs. The designed A-TiO2/ZnO/CdS heterostructure shows a significantly enhanced photoelectric performance and H2 evolution activity as a result of improved light absorption and efficient electrons transport through a highly connected 3D network. We believe that this strategic design of 3D architectures from 2D to 0D nano-components can be further extended to the synthesis of various heterostructures with promising applications in clean energy, optoelectronics, sensing and so on.

Methods

Materials

Tetrabutyl titanate (97%), ammonium hexafluorotitanate ((NH4)2TiF6), zinc nitrate hexahydrate (Zn(NO3)2·6H2O), hexamethylenetetramine (HMTA C6H12N4), cadmium nitrate tetrahydrate (Cd(NO3)2·4H2O, 98%), sodium sulfide hydrate (Na2S·9H2O) were purchased from Sigma Aldrich. All the chemicals were used as received without further purification.

Substrate Preparation

Fluorine-doped tin oxide (FTO) glasses (1 cm × 2 cm) were cleaned ultrasonically with acetone and ethanol for 10 min, respectively and then dried with nitrogen before TiO2 growth.

Growth of TiO2 Arrays on FTO Substrate

(1) Anatase TiO2 (001) tetragonal nanosheet arrays (A-TiO2) were synthesized with an improved method similar to that described by Yang26. In a typical procedure, 1 mL of tetrabutyl titanate was mixed with 60 mL hydrochloric acid (5 M) under vigorous stirring for 10 min, then 0.5 g (NH4)2TiF6 was added into the solution and stirred for another 10 min. The resulting solution was transferred in a dried Teflon autoclave. After that, two pieces FTO glasses were placed at the bottom of the Teflon autoclave with the conductive side facing up. The Teflon autoclave was kept at 180 °C for 16 h. After cooling down to room temperature, the FTO substrates were taken out and rinsed with deionized water thoroughly and then dried at 60 °C. (2) Rutile TiO2 nanorod arrays (R-TiO2) were synthesized following similar procedure except (NH4)2TiF6 was removed and the FTO substrates were placed at an angle against the wall of the Teflon autoclave with the conducting side facing down. The hydrothermal synthesis was conducted at 150°C for 6 h.

Growth of ZnO Nanorod-Arrays on TiO2 Arrays

ZnO nanorod-arrays were synthesized by a seed-mediated method27. First, ZnO seed layer was deposited on the surface of TiO2 arrays by Pulsed Laser Deposition (PLD). The distance between the ZnO target and the substrate was ~50 mm. The vacuum chamber was evacuated to a base pressure of 1 × 10−5 Pa and then retained at an oxygen partial pressure of 3 Pa. The energy at the target surface and repetition rate of the laser was 52 mJ and 5 Hz, respectively. Hydrothermal ZnO growth was carried out by suspending the above-mentioned substrates upside-down in an aqueous solution containing 25 mM zinc nitrate hexahydrate and 25 mM hexamethylenetetramine at 90 °C. The typical growth time is 4 hours. The samples were then removed from solution, rinsed with deionized water and dried at 50 °C. The ZnO nanorod-arrays modified TiO2 films were mentioned as A-TiO2/ZnO and R-TiO2/ZnO, respectively.

CdS Deposition

CdS was deposited on the A-TiO2/ZnO and R-TiO2/ZnO samples by successive ion-layer adsorption and reaction (SILAR) technique28. Typically, the TiO2/ZnO films were successively immersed in two different aqueous solutions for 20 s each, first in 50 mM Cd(NO3)2 in methanol and then in 50 mM Na2S in methanol/water (1:1/v:v). Between each immersion step, the samples were rinsed with methanol for 20 s to remove excess ions that were weakly bound to the surfaces. This immersion cycle was repeated for up to 50 cycles and, after finishing the coating, the samples were dried at 40 °C. The as-prepared samples were named as A-TiO2/ZnO/CdS and R-TiO2/ZnO/CdS, respectively.

Fabrication of solar cells

Platinum (Pt) coated silicon wafers counter-electrodes were prepared by PLD. Solar cells were prepared by assembling and bonding the TiO2/ZnO/CdS electrodes with the Pt counter-electrodes. The two electrodes were separated by a 60 μm thick polypropylene spacer and the internal space of the cell was filled with a polysulfide electrolyte (1.0 M S, 1.0 M Na2S and 0.1 M NaOH in deionized water).

Characterization

X-ray diffraction patterns were obtained using a Shimadzu XRD-6000 X-ray diffractometer with Cu Kα radiation (λ = 1.5418 Å). Field emission scanning electron microscopy images were obtained using a JSM-7600F microscope. The EDS was taken on an OXFORD INCA X-Max Energy Dispersive X-Ray Spectroscopy. High-resolution transmission electron microscopy (HRTEM) measurements were carried out on JEM-2010 operated at 200 KV. J–V measurements were performed using a Keithley 2400 SourceMeter with a custom-made LabTracer program. A 300 W Xenon Arc Lamp (Newport) with appropriate filter was used as the light source.

Photocatalytic Hydrogen Production Activity Tests

Typically, the TiO2/ZnO/CdS electrode (0.3 cm2) was immersed in 4 mL of aqueous solution containing 0.1 M Na2S and 0.1 M Na2SO3 bubbled with Ar and sealed with a rubber septum. A 300 W Xenon Arc Lamp (Asahi Spectra, LAX-C100) with appropriate filter was used as the light source to provide simulated sunlight (AM 1.5G). For visible-light hydrogen production, a 420 nm cutoff filter was used to remove UV light. The amount of H2 in the gas phase was measured using a Shimadzu GC-8A gas chromatograph equipped with an MS-5A column and a thermal conductivity detector (TCD). To obtain an action spectrum in this reaction system, photocatalytic H2 formation was carried out under irradiation of monochromated light from a Xe lamp (Asahi Spectra; 4.3 mW cm−2) with band-pass filters (light width: ±5 nm).