Nanoscale. Realizing omnidirectional light harvesting by employing hierarchical architecture for dye sensitized solar cells PAPER.

Transcription

1 PAPER View Article Online View Journal View Issue Cite this:, 2016, 8, 5478 Received 11th November 2015, Accepted 26th January 2016 DOI: /c5nr07948a Introduction Due to the effects of global warming and the energy crisis, energy has become more important in recent years. Solar energy is the most renewable and secure of existing energy sources. Many photovoltaic devices have already been developed over the past decades. However, their widespread use is limited by two major challenges, namely conversion efficiency and cost. 1 3 The single-crystalline silicon solar cell was invented more than 50 years ago and currently makes up 94% a Department of Electronic Engineering, Chang Gung University, 259 Wen-Hwa 1st Road, Kwei-Shan, Taoyuan 333, Taiwan. sykuo@mail.cgu.edu.tw; Fax: ; Tel: , ext b Department of Photonics Engineering, Yuan-Ze University, 135 Yuan-Tung Road, Chung-Li, 32003, Taiwan. filai@saturn.yzu.edu.tw; Fax: ; Tel: , ext c Advanced Optoelectronic Technology Center, National Cheng Kung University, No. 1, University Road, Tainan 701, Taiwan d Instrument Technology Research Center, National Applied Research Laboratories, 20 R&D Road V1, Hsinchu Science Park, Hsinchu 300, Taiwan e Institute of Electro-Optical Engineering, National Chiao-Tung University, 1001 University Road, Hsinchu, 300, Taiwan f Chang Gung Memorial Hospital, No. 5, Fuxing Street, Kwei-Shan, Taoyuan 333, Taiwan Realizing omnidirectional light harvesting by employing hierarchical architecture for dye sensitized solar cells Ming-Yang Hsieh, a Fang-I Lai,* b,c Wei-Chun Chen, d Min-Chi Hsieh, a Hsiang-Yi Hu, b Peichen Yu, e Hao-Chung Kuo e and Shou-Yi Kuo* a,f To improve the omnidirectional light-harvesting in dye-sensitized solar cells (DSSCs), here we present a dandelion-like structure composed of ZnO hemispherical shells and nanorods. Uniformly distributed hemispherical shells effectively suppress the reflection over the broadband region at incident angles up to 60, greatly improving the optical absorption of the DSSCs. In addition, modulating the length of the ZnO nanorods controls the omnidirectional characteristics of DSSCs. This phenomenon is attributed to the degree of periodicity of the ZnO dandelion-like structures. Cells with shorter rods exhibit a high degree of periodicity, thus the conversion efficiencies of the cells show specific angle-independent features. On the other hand, the cells with longer lengths reveal angle-dependent photovoltaic performance. Along with the simulation, the cells with dandelion-like ZnO structures can couple incident photons efficiently to achieve excellent broadband and omnidirectional light-harvesting performances experimentally, and the DSSCs enhanced the conversion efficiency by 48% at large incident angles. All these findings not only provide further insight into the light-trapping mechanism in these complex three-dimensional nanostructures but also offer efficient omnidirectional and broadband nanostructured photovoltaics for advanced applications. of the market. 4 However, due to the high material costs, thin film solar cells have been developed to solve the problem of production costs. Thin film solar cells, such as gallium arsenide (GaAs), cadmium telluride (CdTe), and copper indium gallium selenide (CIGS) receive much attention because of their high absorption coefficients and high conversion efficiencies. 5,6 However, these materials have certain limitations, such as the scarcity of indium and gallium, and the environmental issues associated with Se. To aim at further lowering the production costs, dye-sensitized solar cells (DSSCs) have recently emerged as a promising approach for efficient solar energy conversion Moreover, DSSCs are promising devices for environmentally compatible and largescale solar energy conversion as an alternative to conventional solid-state semiconductor solar cells. In the past few years, the development of a photoelectrode composed of various types of materials to boost the power conversion efficiency is one of the main issues in DSSCs. Titanium dioxide (TiO 2 ) is most commonly used as a photoelectrode in DSSCs, and to date, the best power conversion efficiency in DSSCs is based on TiO 2 anode materials However, the power conversion efficiency is limited because of recombination between electrons and either the oxidized dye molecules or electron-accepting species in the electrolyte during the 5478,2016,8, This journal is The Royal Society of Chemistry 2016

2 charge transport process In order to enhance power conversion efficiency, DSSCs technology based on zinc oxide (ZnO) has been studied extensively. ZnO is a wide direct band gap (3.37 ev) semiconductor with a high exciton binding energy (60 mev), excellent chemical and thermal stability and an electron affinity similar to that of TiO Moreover, due to the electron mobility of single-crystalline ZnO being higher by 2 3 orders of magnitude than that of TiO 2, it exhibits excellent electron transport collection and fast charge transfer, with reduced recombination loss when used in DSSCs. 21,22 Much research has been reported on the use of ZnO materials for application in DSSCs. In the past several years, one dimensional (1D) ZnO nanostructures, such as nanowires (or nanorods), nanotubes and nanotips have been extensively used in DSSCs, with the expected aim of significantly improving the light harvesting and charge collection in the photoelectrode. 23,24 Hsu et al. fabricated DSSCs of hydrothermally deposited ZnO nanorods and obtained a conversion efficiency of 0.22%. 25 Baxter and Aydil reported ZnO nanowire-based DSSCs and obtained a conversion efficiency of 0.5%. 26 Cheng et al. also fabricated ZnO nanorod-based DSSCs with a conversion efficiency of 0.6%. 27 Yang et al. fabricated ZnO nanowirebased DSSCs with conversion efficiencies of up to 1.5%, where the wire length and fill factor were about 17 m and 0.37 respectively. 28 However, the ideal photoelectrode structure should offer a large surface area for a long optical path, interfacial charge collection and low light reflectance loss for excellent charge transportation. 29,30 Therefore, due to the three dimensional (3D) organization of these networks, this proved a significant increase in the density of the optical pathway. Moreover, 3D structures have been reported to present interesting light trapping characteristics Indeed, a significant increase in the optical pathway in 3D ZnO nanostructures is expected to enhance the light trapping. However, to our knowledge, there are few reports that have discussed the characteristics of DSSCs with 3D ZnO nanostructures. Additionally, the angle-resolved spectral transmittance and simulation of 3D ZnO nanostructures as a photoanode has not yet been studied for DSSCs. Besides, because of the movement of the sun, solar light is incident at all angles of incidence as diffused light scattered by the earth s atmosphere. Indeed, we have made a detailed and systematic study of omnidirectional and broadband photovoltaic properties of DSSCs with 3D ZnO nanostructures. In this work, we demonstrate a scalable and inexpensive bottom-up approach for self-assembling PS nanospheres on a transparent conductive fluorine-doped tin oxide (FTO) glass substrate, and growing ZnO nanorods on the PS nanospheres using a hydrothermal method. This is one of the simplest and fastest methods of fabricating a two dimensional periodic structure using self-assembled PS nanospheres. Growing ZnO nanorods using the hydrothermal method is suitable for producing nanostructures on large-area patterns. The 3D hierarchical dandelion-like ZnO structure in this study is a combination of PS nanospheres and ZnO nanorods, as shown in Fig. 1(e). This dandelion-like structure forms a photoanode Fig. 1 Schematic process for fabricating the dandelion nanostructure on an FTO glass substrate and SEM images of the dandelion nanostructures. Scale bar: (b) 1.5 µm, (d and f) 1 µm. that enhances light absorption in the broadband and omnidirectional spectral range. For a deep understanding of the effects of the hierarchical dandelion-like ZnO structure on its optical characteristics, a finite difference time domain (FDTD) simulation was also performed. Experimental The fabrication process for the dandelion-like ZnO structure is schematically illustrated in Fig. 1. First, a commercially available polystyrene (PS) microsphere suspension (diameter 1 µm, 10 wt% aqueous dispersion) was used for dip coating. FTO transparent conductive substrates were cleaned thoroughly by consecutive sonication in acetone, ethanol, and isopropanol for 20 minutes each. Afterward, a monolayer of self-assembled PS microspheres was spun on the surface, and a closely packed PS microsphere monolayer was obtained as shown in Fig. 1(a and b). In order to fabricate the ZnO nanorods which were grown on the surface of the microspheres on the FTO transparent conductive substrates, we used a threestep method corresponding to the fabrication of seed layers and the growth of the nanorods. First, the PS microspherecovered samples were subjected to the radio frequency (RF) magnetron sputtering and sol gel method for synthesizing a ZnO thin film covering the microsphere and FTO substrate surface, which was used as a seed layer to grow the ZnO nanorods. Subsequently, the PS microspheres were burned off by calcination, leaving spherical voids in the ZnO structures as shown in Fig. 1(c and d). Finally, the ZnO nanorods were grown by the hydrothermal method. The synthesis process has been described in detail in our previous report. 34 The aqueous solution was prepared for growing the ZnO nanorods by This journal is The Royal Society of Chemistry 2016,2016,8,

3 mixing zinc nitrate hexahydrate and hexamethylenetetramine. After the two aqueous solutions were mixed, the sample was immersed in the solution at 90 C for 3 21 h. Subsequently, we used deionized water to clean the cell and it was heated to 60 C in air for 1 h, when the process of growing the ZnO nanorods was completed. D-719 dye, cis bis(isothiocyanato) bis(2,2 -bipyridyl-4,4 - dicarboxylato) ruthenium(ii) bis-tetrabutylammonium (Everlight Chemical Industrial Corp., Taipei, Taiwan), was dissolved in acetonitrile for the preparation of the 0.5 mm dye solution. Solar cells were prepared by immersing the dandelion-like ZnO structure on the FTO substrate in the dye solution for 1 hour at room temperature. A sandwich-type configuration was employed to measure the presentation of the DSSCs. The area of the active electrode was typically 1 cm 2. The morphologies of the fabricated ZnO nanostructures on the FTO substrate were observed using a field emission scanning electron microscope (FESEM, Hitachi S-4700I). The structural properties were analyzed using transmission electron microscopy (TEM). To characterize the performance of the ZnO nanostructures on the FTO substrate, the transmission spectra and angle-dependent transmittance of the samples were measured using a UV-vis-NIR spectrophotometer (Lambda 900 UV/Visible/NIR) for wavelengths ranging from 350 to 800 nm and incident-angles from 0 to 60 degrees. The performances of the DSSCs were characterized under simulated AM1.5G illumination with a power of 1000 W m 2. The external quantum efficiency (EQE) results were acquired from a system which employed a 300 W Xenon lamp (Newport 66984) light source and a monochromator (Newport 74112). The beam spot size at the sample was measured as roughly 1 mm 3 mm and was rectangular. During the measurements, the temperature was controlled at 25 ± 1 C. Results and discussion Fig. 2 shows the top-view and cross-sectional SEM images of the PS spheres and dandelion structures. Monolayers of 1 µm diameter spheres were formed by drying the suspension, and Fig. 2(b) shows the layout of the PS spheres with a closelypacked hexagonal pattern in a large area. After we fabricated approximately 100 nm-thick ZnO seed layers on the PS spheres and removed the PS spheres, the PS spheres still retained the well-ordered structure with hollow voids, as shown in Fig. 1(d). After fabrication of the ZnO hemispherical shells, the ZnO nanorods were grown on the ZnO shells using the hydrothermal method. In Fig. 2, the top-view (Fig. 2(c, e, f and h) and the cross-sectional (Fig. 2(i l)) SEM images of the ZnO nanorods which were grown on the ZnO shells in 3 h, 9 h, 15 h Fig. 2 (a) Image of monolayer PS microsphere arrays assembled on the FTO glass. (b) SEM image of monolayer PS microsphere arrays. Top-view and cross-sectional SEM images of the ZnO dandelion structures with different times of growth on FTO glass: (c and i) 3 hours, (e and j) 9 hours, (f and k) 15 hours, and (g and l) 21 hours. Scale bar: (b) 5 µm, (c, e, f, g, and i l) 2 µm, and (d and h) 1 µm. 5480,2016,8, This journal is The Royal Society of Chemistry 2016

4 and 21 h, are shown, respectively. In Fig. 2(d) it can be observed that groups of freestanding ZnO nanorods are branching out from the surface of each ZnO hemispherical shell, exhibiting a dandelion-like shape. After higher reaction times, the ZnO nanorods were vertically aligned and randomly distributed on the surface as shown in Fig. 2(h). The ZnO nanorods had diameters of approximately 50 nm and lengths ranging from 1 µm to 3 µm with the different growth times. The length of the ZnO nanorods continuously increased up to 15 h, after which it decreased. This is due to the zinc species in the solution being largely consumed, which would prohibit nanorod growth at higher reaction times. Additionally, because of the increase in the local concentration of OH ions relative to Zn 2+ ions, the solution becomes more basic, which would promote the dissolution of the ZnO nanorods. 35 It should be noted that the basicity of the solution increases with the increase in growth time. Therefore, in the 21 h reaction time, the ZnO nanorods were selectively dissolved from their top to form a random distribution on the surface. TEM images of ZnO nanorods which were grown on ZnO hemispherical shells, along with the corresponding fast- Fourier-transform patterns, are shown in Fig. 3. The low magnification image in Fig. 3(a) shows the presence of ZnO nanorods of different sizes. The tilted lengthy nanorod at the right of the image shows a diameter of 35 nm and a length of 830 nm. The presence of some short nanorods in the image is probably due to fragmentation which occurred during the TEM sample preparation. Also, the bottom two layers on the hemispherical shell with (i) sol gel ZnO and (ii) sputtered ZnO, can be observed, respectively. Fig. 3(b) and (c) show highresolution transmission electron microscopy (HRTEM) images of ZnO (0002) lattices and Fourier-filtered images. Also, Fourier-filtered images of the framed regions in the lattice images (Fig. 3(b) and (c)) were obtained by selecting the ±(0002) ZnO spots in the FFT pattern. The result indicated that the ZnO nanorod is nearly a single crystal on a ZnO hemisphere. In addition, the ZnO lattice fringes are consistent with hexagonal wurtzite ABAB stacking in all the samples, corresponding to the ABABAB stacking sequence parallel to the (0001) plane, which confirms the wurtzite structure with the c lattice parameter approximately 0.26 nm from the value for the ZnO bulk crystal. 36 Stacking faults and/or dislocations were observed at nearly the top of the ZnO rods as shown in Fig. 3 (b). Moreover, it can be clearly observed that the ZnO spots in the selected area electron diffraction (SAED) pattern exhibit sharp streaks along the direction of the c-axis, suggesting that there exists a high density of planar defects at nearly the top of the ZnO nanorod. The stacking faults can also be formed by condensation of intrinsic point defects, vacancies or interstitials. The stacking faults at the initial stage of growth, ZnO nanorods with a hexagonal prism morphology, are formed by the fastest growth rate along the <001> direction. Stacking Fig. 3 (a) Cross-sectional TEM image of ZnO nanorods grown on a ZnO hemispherical shell for 9 h. The lattice image and the electron diffraction patterns of a nanorod s top and bottom are shown in (b) and (c), respectively. Fourier-filtered images of the framed regions in (b) and (c) were obtained by selecting the ±(0002) ZnO spots. This journal is The Royal Society of Chemistry 2016,2016,8,

5 faults in the nanorods naturally introduce steps on the side walls. The stacking faults are bounded by partial dislocations, being a part of faulted dislocation loops or ends at the internal interfaces. Under the precursor vapor supersaturation environment, these stacking faults act as sites of preferential attachment for the deposited atoms, leading to the observed lateral growth. 37 In order to more clearly investigate the crystals of the ZnO dandelions, the ZnO nanorods and ZnO dandelion structures with different lengths on the FTO glass substrate are also investigated. Fig. 4(a) shows XRD patterns of the ZnO nanorods and dandelion structures on the FTO glass substrate. The strong diffraction peak at 34.4 on the ZnO nanorods and dandelion structures corresponds to the (002) plane along the c-axis. The three peaks at 31.8, 36.3 and 47.7 in the ZnO dandelion samples represent the (100), (101) and(102) ZnO crystal planes, respectively. This is assigned to the pure wurtzite ZnO phase as referred to in the JCPDS file Therefore, the XRD analysis shows that perpendicularly oriented ZnO crystals form on the FTO glass substrate, as well as other crystal planes. Additionally, the XRD analysis corresponds to the SEM images. To investigate the charge and transfer process of dandelionlike ZnO photoelectrodes, electrochemical impedance spectroscopy (EIS) measurements were conducted under open-circuit conditions under simulated solar light illumination. The Nyquist plots of the obtained EIS data are shown in Fig. 4(b). The high frequency semicircular portion in the Nyquist plots corresponds to the charge transfer in the platinum counter electrode, and the large semicircle in the low frequency range represents the charge transfer at the ZnO/dye/electrolyte interface. The measured impedance spectroscopic data were fitted using an equivalent circuit and Zsimpwin software. 38 The R CT2 value for the photo-electrode containing the 15 h dandelionlike structure (55.74 Ω) is lower than that of the photo-electrode containing the 18 h nanorod structure (89.93 Ω), whereas the R CT1 value is almost the same. The R CT2 values for the3h,9hand21hstructuresare76.49ω, Ω and Ω, respectively. One possible cause for low-load transport resistance might be that the dandelion-like structures and the longer branch structures promote the charge separation and transportation arising from the increased contact surface area. Therefore, because of the higher dye loading amount, the excited electrons were further enhanced, causing higher electron concentrations and larger driving forces for injecting electrons into the outer circuit, thus resulting in smaller values of R CT2. Fig. 4(c) displays the photovoltaic I V characteristics at a normal incident angle. DSSCs with different lengths of Fig. 4 (a) X-ray diffraction (XRD) patterns of the dandelion-like ZnO structures with different times of growth and ZnO nanorods ; (b) electrochemical impedance spectra (EIS); (c) I V characteristics and (d) external quantum efficiency spectra of the DSSCs based on the ZnO nanorod structure and dandelion structures. 5482,2016,8, This journal is The Royal Society of Chemistry 2016

6 Table 1 Photovoltaic parameters of different ZnO nanostructures ARcoated on CIGS solar cells Samples V oc (V) J sc (ma cm 2 ) FF η (%) Dandelion-3H Dandelion-9H Dandelion-15H Dandelion-21H Nanorod-18H dandelion-like structure were measured under standard test conditions with simulated air mass 1.5 global illumination. This study also compares this cell with a standard one with a two-dimensional (2D) ZnO nanorod structure. The characteristics of the DSSCs are summarized in Table 1. In Fig. 4(c), there is a similarity in conversion efficiency (η) between the 18 h ZnO nanorod and the 3 h ZnO dandelion-like structure. However, the open-circuit voltages (V oc ) decreased from 0.7 V to 0.63 V and the short-circuit density ( J sc ) increased from 2.81 ma cm 2 to 3.31 ma cm 2 when the 2D nanorod was transformed into a 3D dandelion-like structure. According to the previous literature, the V oc decrement in ZnO photoelectrodes can be explained in three ways: (1) insufficient electrolyte pore-filling in ZnO structures could not support the efficient hole scavenging at the ZnO/dye interfaces, raising the locus of recombination; (2) lower dye loading fosters less charge injection from the dye sensitizer to the conduction band of ZnO, resulting in a downward-shift in the ZnO quasi- Fermi level, thus reducing the potential difference between ZnO and the redox species. 39,40 However, in this study, explanations (1) and (2) are inadequate to explain it, due to the dandelion-like ZnO structure having higher J sc values than the nanorod structure. Therefore, the V oc decrement in the dandelion-like ZnO structure can be justified by reason (3): the inefficient ZnO photoelectrode coverage on the surface of the FTO substrate, resulting in high recombination losses at the ZnO/ FTO substrate interface. This could be observed as the crosssectional SEM images in Fig. 1(d) and (e), the hollow voids lead to inefficient coverage of the ZnO photoelectrode on the surface of the FTO substrate. It is known that the photon absorption ability and carrier collection efficiency are crucial to photovoltaic devices. As the length of the ZnO nanorods increased initially, the low current density ( J sc ) and the conversion efficiency increased due to the increase in the effective surface area. However, the conversion efficiency of the DSSCs decreased while the length of the nanorods continued to increase. The decrease in conversion efficiency might be attributed to reduced transmittance and an enhanced two-dimensional growth mechanism with longer nanorods. Accordingly, the inter-rod distance became smaller and they became more compact with the larger diameter. This phenomenon could make the dye only adsorb onto the surface, resulting in a small amount of dye reaching the nanorods. Moreover, the distance that the electrons have to travel is increased, thus lowering the carrier collection efficiency. Moreover, the J sc increment in dandelion-like ZnO structures compared with 2D nanorods can be considered a combination of two effects. First, the enhanced photon absorption associated with the significantly enlarged surface area results in higher dye loading. Second, randomly branched nanorods and the ZnO hemispherical shell promote enhanced light trapping and scattering effects in the broad wavelength region without sacrificing efficient electron transport. Therefore, the cell with the 15 hour dandelion-like structure has the best conversion efficiency (η) of 1.53% with an approximate V oc of 0.63 V, a fill factor (FF) of 0.36, and a J sc of 6.7 ma cm 2. Moreover, the quantum efficiency (EQE) measurement shown in Fig. 4(d) also supports the enhancement in conversion efficiency created by the dandelion-like ZnO structure. Compared to the 2D nanorod structure, the dandelion-like ZnO structure has higher EQE values over the whole spectral range from 400 to 700 nm, due to the higher dye loading amount and more efficient light harvesting. This observation confirms that the improvement in J sc of the DSSCs with 3D dandelion-like structures compared with a 2D nanorod structure is because of the great light trapping and scattering of the dandelion-like structures. For a deeper understanding of the light scattering effect by the hierarchical dandelion-like ZnO structures, the FDTD simulation was performed. Fig. 5(b i) shows the calculated electric fields of light passing through the (b) FTO/glass, (c) ZnO nanorod (1 µm length, 50 nm interval)/fto/glass, (d f) ZnO hemispherical shell (300 nm, 600 nm and 1 µm diameter)/fto/glass, and (g i) ZnO nanorods (400 nm, 1 µm, 2 µm height)/zno hemispherical shell (1 µm diameter)/fto/ glass. The thickness of the FTO films and the ZnO hemispherical shell were fixed at 500 nm and 100 nm, respectively. The diameter of the ZnO nanorod was fixed at 50 nm. In this simulation, the amplitude of the y-polarized electric field was simulated for the incident plane wave with a normalized Gaussian beam profile propagating in the z direction at a wavelength of 500 nm. The refractive indexes of glass, FTO, and ZnO at a wavelength of 500 nm were 1.54, 2.04, and 1.87, respectively. For the bare FTO glass substrate in Fig. 5(b), the incident light propagated steadily and unchangeably from its propagation direction. After the ZnO nanorod pattern was deposited on the FTO glass substrate, the incident light exhibited just a few changes in the distribution of the electric field intensity but was still steady in its propagation direction. Fig. 5(b) confirms that the 2-D ZnO-nanorod structure does not possess a light scattering effect, and cannot enhance light trapping, as schematically illustrated in the top of Fig. 5(a). When the hemispherical shells with 300 nm diameter were employed on FTO/glass, as shown in Fig. 5(d), the incident light propagated smoothly with its direction unchanged. However, when the diameter of the hemispherical shells increased, strong interference patterns occurred at near and far distances from the surface, which can be observed in Fig. 5(e and f). The sizes of the hemispherical shells belong to the sub-wavelength range (i.e. they are smaller than the wavelength of the light), so the Mie scattering effect dominated the results of light This journal is The Royal Society of Chemistry 2016,2016,8,

7 Fig. 5 (a) Schematic illustrations of light reflecting and scattering within the ZnO structures. Calculated electric fields of light passing through the (b) FTO/glass, (c) ZnO nanorod (1 µm length, 50 nm interval)/fto/glass, (d f) ZnO hemispherical shell (300 nm, 600 nm and 1 µm diameter)/fto/ glass, and (g i) ZnO nanorods (400 nm, 1 µm, 2 µm height)/zno hemispherical shell (1 µm diameter)/fto/glass. The thickness of the FTO films and the ZnO hemispherical shell were fixed at 500 nm and 100 nm, respectively. The diameter of the ZnO nanorods was fixed at 50 nm. propagation through the scattering center the hemispherical shells. The Mie scattering effect expresses that when the size of the scattering center is comparable with the incident wavelength, the intensity of forward scattering is greater than that of the backward scattering, and this phenomenon can be seen in Fig. 5(d) and (e). 41,42 When the size of the hemispherical shell is slightly larger than the incident wavelength, the diffraction effect is increased, as shown in Fig. 5(f). The light propagation range was enlarged after transferring through the hemispherical shell due to the diffraction effect. This will assist in light diffusing to a wider area. Afterward, the ZnO nanorods grown on the 1 µm ZnO hemispherical shell monolayer (i.e. the dandelion-like ZnO structures) further enhance the light diffraction and diffusion phenomenon, as shown in Fig. 5(g) (i). The sizes of the nanorods are about λ/10 and the scattering result was governed by Rayleigh scattering. This causes the intensity of scattered light to be almost the same in every direction. 42 Therefore, the enhanced forward light by the hemispherical shells monolayer will be scattered in the side direction by the nanorods as shown in Fig. 5(g). But, when the length of the nanorods continuously increased, the strong diffraction patterns in the forward direction were transformed smoothly as depicted in Fig. 5(h and i). This might result from the longer nanorods destroying the periodic property of the ZnO hemispherical shells, and suppressing the diffraction of light. Fig. 5(b), (c), (f) and (g) reveal that the dandelion-like ZnO structures can improve light scattering and enhance the J sc. Although the longer nanorods might slightly reduce the diffraction effect, they could enhance dye loading. Therefore, choosing a proper length for the dandelion-like ZnO structure according to the FDTD simulation results can enhance the conversion efficiency because of the better scattering effect as schematically illustrated in Fig. 5(a). Because of the movement of the sun, the incidence angledependent transmittance was further investigated using a UVvis-NIR spectrophotometer (Lambda 900 UV/Visible/NIR), which was modified by inserting a rotatable sample holder as shown in Fig. 6(a). Fig. 6(b f) show the measured angle-dependent transmittance of the FTO glass substrate with different ZnO structures for wavelengths ranging from 350 to 800 nm and incident-angles from 0 to 60 degrees. Red and blue in the color bar represent the normalized values of transmittance from high to low. In Fig. 6(b f), at wavelengths of nm and incident angles of 0 degrees, the average transmittance value of the 2D nanorods is higher than the 3D dandelion-like structures. The effective refractive index can be calculated 5484,2016,8, This journal is The Royal Society of Chemistry 2016

8 Fig. 6 (a) UV-vis-NIR spectrophotometer schematic. The measured angular transmittance spectra for FTO glass with (b) 3 hour dandelions, (c) 9 hour dandelions, (d) 15 hour dandelions, (e) 21 hour dandelions, (f) 18 hour nanorods, and (g) the average transmittance of all samples. using a volume weighted average of the refractive indices of the different constituents. Therefore, it can be explained that the nanorods have a linear and continuous graded effective refractive index profile from air to the FTO glass substrate. In contrast, there are hollow voids in the ZnO hemispherical shell, which has a refractive index exhibiting an abrupt change at the interface of the ZnO nanorods/fto glass substrate. The discontinuous refractive indexes lead to the strong light scattering effect. Therefore, when the incident angle is increased, for dandelion-like structures, the transmittance is much less dependent on the incident angle over a wide wavelength of nm than it is for the nanorods. Fig. 6(g) shows the normalized values of transmittance of the dandelion-like structures and nanorods. The transmittance of the dandelions remains almost the same at wavelengths of nm up to an incident angle of θ i = 40, but the nanorods decrease in transmittance from the incident angle of θ i = 15. Thus, the ZnO dandelion-like structure is expected to serve as a photoelectrode with efficient broadband wide-angle transmittance and light scattering effect on the DSSCs. The angle-dependent conversion characteristics were further investigated using a homemade rotatable photo I V measurement apparatus, and the incident angles were between 0 and 60, as shown in Fig. 7(a). The inset of Fig. 7(b) shows the angular conversion efficiency for the cells with nanorods and different length dandelion structures. The efficiency enhancement factor (EF EFF ) of the ZnO dandelionlike structures can be obtained using the following equation: EF Eff ¼ ΔEff Eff Eff with dandelion ¼ ð Þ Eff ð with nanorod 18H Þ Effðwith nanorodþ When the cells are measured under the normal incident angle, there is a similarity in conversion efficiency between the 18H ZnO nanorod and the 3H ZnO dandelion-like structure. As the light incident angle increased to 60, the efficiency enhancement factor increased to 48%. Moreover, in Fig. 6(b), the light incident angle increased to 60, and the efficiency decreases of the 18H nanorod and 3H dandelion structures were 39.31% and 10.13%, respectively. These results imply that cells with shorter rods exhibit a high degree of periodicity, thus the conversion efficiencies of the cells show specific angle-independent features. On the other hand, the cells with longer lengths reveal angle-dependent photovoltaic performance. Therefore, the superior light trapping and light scattering over a This journal is The Royal Society of Chemistry 2016,2016,8,

9 funded by the Ministry of Science and Technology (MOST E CC3, MOST M ) and Chang Gung Memorial Hospital (BMRP 956). Fig. 7 (a) Schematic of homemade rotatable photo I V measurement apparatus. (b) The angular photocurrent characterization of the DSSCs based on ZnO nanorod structure and dandelion structure. wide range of incident angles can further improve the solar energy conversion efficiency for the entire day, and integrating 1D ZnO nanorods with a ZnO hemispherical shell has the potential to provide enhanced broadband and omni-directional light harvesting for application in other optoelectronic devices. Conclusions In summary, DSSCs with dandelion-like ZnO structures were fabricated and characterized in detail. The photoelectrode using a dandelion-like ZnO structure involves the combination of self-assembled PS and the hydrothermal method, which shows the superior light-harvesting characteristics for DSSCs by enhancing forward scattering. The angular I V characterization confirms an enhanced photocurrent conversion for an incident angle up to 60. Additionally, as the light incident angle increases to 60, the efficiency enhancement factor increases to 48%. Theoretical and experimental investigations find that the dandelion-like ZnO structures are effective in harvesting light over a wide range of wavelengths and incident angles, showing their broadband and omnidirectional characteristics. The presented concept and the manufacturing technique for the light-harvesting dandelion-like ZnO structures represent a viable, promising path toward high-efficiency DSSCs and should benefit many other types of electro-optical devices. Acknowledgements This research was supported by the Green Technology Research Center of Chang Gung University, and also a grant References 1 D. M. Bagnall and M. Boreland, Energ. Photovoltaic technologies, Energy Policy, 2008, 36(36), M. A. Green, Recent developments in photovoltaics, Sol. Energy, 2004, 76, M. Oliver and T. Jackson, The market for solar photovoltaics, Energy Policy, 1999, 27, D. M. Chapin, C. S. Fuller and G. L. Pearson, A New Silicon p-n Junction Photocell for Converting Solar Radiation into Electrical Power, J. Appl. Phys., 1954, 25, K. Jimbo, R. Kimura, T. Kamimura, S. Yamada, W. S. Maw, H. Araki, K. Oishi and H. Katagiri, Cu 2 ZnSnS 4 -type thin film solar cells using abundant materials, Thin Solid Films, 2007, D. B. Mitzi, O. Gunawan, T. K. Todorov, K. Wang and S. Guha, The path towards a high-performance solutionprocessed kesterite solar cell, Sol. Energy Mater. Sol. Cells, 2011, 95, B. O Regan and M. Grätzel, A low-cost, high-efficiency solar cell based on dyesensitizedcolloidal TiO2 films, Nature, 1991, 353, S. Mathew, et al., Dye-sensitized solar cells with 13% efficiency achieved through the molecular engineering of porphyrin sensitizers, Nat. Chem., 2014, 6, J.-H. Yum, et al., A cobalt complex redox shuttle for dyesensitized solar cellswith high open-circuit potentials, Nat. Commun., 2012, 3, X. Y. Lai, J. E. Halpert and D. Wang, Recent Advances in micro-/nano-structuredhollow spheres for energy applications: from simple to complex systems, Energy Environ. Sci., 2012, 5, A. Wolcott, W. A. Smith, T. R. Kuykendall, Y. P. Zhao and J. Z. Zhang, Photoelectrochemical water splitting using dense and aligned TiO2 nanorod arrays, Small, 2009, 5, X. Feng, K. Shankar, O. K. Varghese, M. Paulose, T. J. Latempa and C. A. Grimes, Vertically aligned single crystal TiO2 nanowire arrays grown directly on transparent conducting oxide coated glass: synthesis details and applications, Nano Lett., 2008, 8, P. Hartmann, D. K. Lee, B. M. Smarsly and J. Janek, Mesoporous TiO2: Comparison of Classical Sol Gel and Nanoparticle Based Photoelectrodes for the Water Splitting Reaction, ACS Nano, 2010, 4, G. M. Wang, H. Y. Wang, Y. C. Ling, Y. C. Tang, X. Y. Yang, R. C. Fitzmorris, C. C. Wang, J. Z. Zhang and Y. Li, Hydrogen-Treated TiO2 Nanowire Arrays for Photoelectrochemical Water Splitting, Nano Lett., 2011, 11, J. Nissfolk, K. Fredin, A. Hagfeldt and G. Boschloo, Recombination and transport processes in dye-sensitized 5486,2016,8, This journal is The Royal Society of Chemistry 2016

10 solar cells investigated under working conditions, Phys. Chem. B, 2006, 110, M. Gratzel, Solar Energy Conversion by Dye-Sensitized Photovoltaic Cells, Inorg. Chem., 2005, 44, M. Gratzel, Conversion of sunlight to electric power by nanocrystalline dye-sensitized solar cells, J. Photochem. Photobiol., A, 2004, 164, J. B. Baxter and E. S. Aydil, Nanowire-based dye-sensitized solar cells, Appl. Phys. Lett., 2005, 86, Y. F. Hsu, Y. Y. Xi, A. B. Djursic and W. K. Chan, ZnO nanorods for solar cells: hydrothermal growth versus vapor deposition, Appl. Phys. Lett., 2008, 92, B. Pradhan, S. K. Batabyal and A. J. Pal, Vertically aligned ZnO nanowire arrays in Rose Bengal-based dye-sensitized solar cells, Sol. Energy Mater. Sol. Cells, 2007, 91, M. Law, L. E. Greene, J. C. Johnson, R. Saykally and P. Yang, Nanowire dye-sensitized solar cells, Nat. Mater., 2005, 4, C. Magne, F. Dufour, F. Labat, G. Lancel, O. Durupthy, S. Cassaignon and T. Pauporté, Effects of TiO 2 nanoparticle polymorphism on dye-sensitized solar cell photovoltaic properties, Photochem. Photobiol., A, 2012, 232, U. Pal, J. G. Serrano, P. Santiago, G. Xiong, K. B. Ucer and R. T. Williams, Synthesis and optical properties of ZnO nanostructures with different morphologies, Opt. Mater., 2006, 29, S. Kar, A. Dev and S. Chaudhuri, Simple solvothermal route to synthesize ZnO nanosheets, nanonails, and well-aligned nanorod arrays, J. Phys. Chem. B, 2006, 110, Y. F. Hsu, Y. Y. Xi, A. Djurisic and W. K. Chen, ZnO Nanorods for Solar Cells: Hydrothermal Growth Versus Vapor Deposition, Appl. Phys. Lett., 2008, 92b, J. B. Baxter and E. S. Aydil, Nanowire-based dye-sensitized solar cells, Appl. Phys. Lett., 2005, 86, A. J. Cheng, Y. Tzeng, Y. Zhou, M. Park, T. Wu, C. Shannon, D. Wang and W. Lee, Thermal chemical vapor deposition growth of zinc oxide nanostructures for dye-sensitized solar cell fabrication, Appl. Phys. Lett., 2008, 92, M. Law, L. E. Greene, J. C. Johnson, R. Saykally and P. Yang, Nanowire dye-sensitized solar cells, Nat. Mater., 2005, 4, L. S. Li, Y. H. Yu, F. Meng, Y. Z. Tan, R. J. Hamers and S. Jin, Facile solution synthesis of alpha-fef 3 center dot 3H 2 O nanowires and their conversion to a-fe 2 O 3 nanowires for photoelectrochemical application, Nano Lett., 2012, 12, C. Liu, N. Dagupta and P. D. Yang, Semiconductor nanowires for artificial photosynthesis, Chem. Mater., 2014, 1, J. Elias, C. Lévy-clément, M. Bechelany, J. Michler, G.-Y. Wang, Z. Wang and L. Philippe, Hollow urchin-like ZnO thin films by electrochemical deposition, Adv. Mater., 2010, 22, D.-M. Tang, G. Liu, F. Li, J. Tan, C. Liu, G. Q. Lu and H. M. Cheng, Synthesis and Photoelectrochemical Property of Urchin-like Zn/ZnO Core/Shell Structures, J. Phys. Chem. C, 2009, 113, C. W. Cheng, H. F. Zhang, W. N. Ren, W. J. Dong and Y. Sun, Three dimensional urchin-like ordered hollow TiO 2 /ZnO nanorods structure as efficient photoelectrochemical anode, Nano Energy, 2013, 2, S. Y. Kuo, M. Y. Hsieh, H. V. Han, F. I. Lai, Y. L. Tsai, J. F. Yang, T. Y. Chuang and H. C. Kuo, Dandelion-shaped nanostructures for enhancing omnidirectional photovoltaic performance,, 2013, 5, Y. Chaea, J. T. Parkb and J. K. Kohb, All-solid, flexible solar textiles based on dye-sensitized solar cells with ZnO nanorod arrays on stainless steel wires, Mater. Sci. Eng., B, 2013, 178, R. Triboulet, Growth of ZnO bulk crystals: A review, Prog. Cryst. Growth Charact. Mater., 2014, 60, V. Khranovskyy, M. O. Eriksson, G. Z. Radnoczi, A. Khalid, H. Zhang, P. O. Holtz, L. Hultman and R. Yakimova, Photoluminescence study of basal plane stacking faults in ZnO nanowires, Physica B, 2014, 439, Y. B. Xie, L. M. Zhou and J. Lu, Photoelectrochemical behavior of titania nanotube array grown on nanocrystalline titanium, Mater. Sci., 2009, 44, F. Sauvage, F. D. Fonzo, A. L. Bassi, C. S. Casari, V. Russo, G. Divitini, C. Ducati, C. E. Bottani, P. Comte and M. Graetzel, Hierarchical TiO2 photoanode for dye-sensitized solar cells, Nano Lett., 2010, 10, P. Sudhagar, R. S. Kumar, J. H. Jung, W. Cho, R. Sathyamoorthy, J. Won and Y. S. Kang, Facile synthesis of highly branched jacks-like ZnO nanorods and their applications in dye-sensitized solar cells, Mater. Res. Bull., 2011, 46, M. Kerker, The Scattering of Light and Other Electromagnetic Radiation, Academic Press, J. R. Meyer-Arendt, Introduction to Classical and Modern Optics, Prentice Hall, New Jersey, This journal is The Royal Society of Chemistry 2016,2016,8,