Enhancing the Light Harvesting Capacity of the Photoanode Films in Dye-Sensitized Solar Cells

Dye-sensitized solar cell (DSC) has been receiving continuous academic and industrial at‐ tention as a potential low-cost, clean, and renewable energy source, since its inception in 1985 (Desilvestro et al., 1988; Regan et al., 1991; Gao et al., 2008; Hagfeldt et al., 2010; Yella et al., 2011). The DSC is the only photovoltaic device that uses molecules to absorb photons and converts them to electric charges without the need of intermolecular transport of elec‐ tronic excitation. It is also the only photovoltaic device that separates two functions of light harvesting and charge-carrier transport, mimicking the photosynthesis found in green leaves. The primary photon-to-electron conversion process in DSC occurs at the oxide/dye/ electrolyte interface, functioning at a molecular and nano scale. There exist many complicat‐ ed optical, electrical, and chemical processes during the light-to-electric conversion process in DSC, including the light absorption of dye molecules (producing electrons and leaving the dyes in their oxidized states), the electron injection from dyes to the conduction band of semiconductor (e.g., TiO2), the percolation of electrons through the mesoporous semicon‐ ductor network toward the bottom electrode, the recombination of photo-excited electrons in the porous electrode with either oxidized dyes or acceptors in the electrolyte, and the re‐ generation of dye molecules by iodides in the electrolyte etc. For an efficient DSC device, while above processes should be kept in a complicated and delicate balance, the high lightharvesting, the rapid electron transport, and the minimum electron-hole recombination are essential. In special, the light-harvesting efficiency of the photoanode is the most important and indispensable factor for the high-efficiency DSC, which is mainly related to the molar extinction coefficient of the sensitizer, the dye-loading capacity of the porous electrode, and the optical path of the incident light in the electrode.


Introduction
Dye-sensitized solar cell (DSC) has been receiving continuous academic and industrial attention as a potential low-cost, clean, and renewable energy source, since its inception in 1985 (Desilvestro et al., 1988;Regan et al., 1991;Gao et al., 2008;Hagfeldt et al., 2010;Yella et al., 2011). The DSC is the only photovoltaic device that uses molecules to absorb photons and converts them to electric charges without the need of intermolecular transport of electronic excitation. It is also the only photovoltaic device that separates two functions of light harvesting and charge-carrier transport, mimicking the photosynthesis found in green leaves. The primary photon-to-electron conversion process in DSC occurs at the oxide/dye/ electrolyte interface, functioning at a molecular and nano scale. There exist many complicated optical, electrical, and chemical processes during the light-to-electric conversion process in DSC, including the light absorption of dye molecules (producing electrons and leaving the dyes in their oxidized states), the electron injection from dyes to the conduction band of semiconductor (e.g., TiO 2 ), the percolation of electrons through the mesoporous semiconductor network toward the bottom electrode, the recombination of photo-excited electrons in the porous electrode with either oxidized dyes or acceptors in the electrolyte, and the regeneration of dye molecules by iodides in the electrolyte etc. For an efficient DSC device, while above processes should be kept in a complicated and delicate balance, the high lightharvesting, the rapid electron transport, and the minimum electron-hole recombination are essential. In special, the light-harvesting efficiency of the photoanode is the most important and indispensable factor for the high-efficiency DSC, which is mainly related to the molar extinction coefficient of the sensitizer, the dye-loading capacity of the porous electrode, and the optical path of the incident light in the electrode.
At present, the record efficiency of DSC has reached 12.03% (Yella et al., 2011) after the first breakthrough of M. Grätzel in 1991 and through the continuous efforts of the scientists for more than 20 years. With an aim to further improve the conversion efficiency of DSC to 15% or higher, recently, intensive efforts have been directed to enhance the functions of the major components in DSC, focusing on mainly three aspects, i.e., the ordered nanoporous electrode, the novel organic or quantum-dot sensitizer, and the new redox electrolyte system (Hagfeldt et al., 2010;Tetreault et al., 2012). During this journey toward high-efficiency DSC, the light-harvesting capacity of the photoanode is expected to play a critical and indispensable role by spanning the mesoporous electrodes utilizing varied nano/meso materials and the sensitizers with high molar extinction coefficient or quantum dots, and by connecting the optical functions of the photoanode with its electrical response together.
There are mainly two components concerning the light-harvesting property of DSC, the dye and the mesoporous electrode which supports the sensitizer. While the development of the novel sensitizers with higher molar extinction coefficient and better response to near-infrared wavelength has been long studied since the first breakthrough of DSC in 1991 (Nazeeruddin et al., 2001;Grtäzel et al., 2009;Yum et al., 2011), the significance of the light scattering and/or reflection of the mesoporous electrode on the light-harvesting properties is only recognized in recent years. Accompanying the rapid advance of the ordered or hybrid photoanode materials, novel semiconductor nanostructures or new concepts are widely used to construct new type photoanodes, including the plenary optical waveguide nanowire enhancing the multiple internal reflections (Wei et al., 2010), the plasmonic core-shell metalinsulator nanoparticles enhancing the near-infrared absorption (Brown et al., 2011), the dual-function mesoporous TiO 2 structures increasing both the dye-loading and the optical scattering effects (Koo et al, 2008;Huang et al., 2010), and so on. In view of the wide coverage of the optical and photoelectrical nanostructures involved in these studies and their rapid progresses, it is great valuable to make a comprehensive and in-depth review on the current status of the light-harvesting capacities of the photoanode in DSC.
The chapter will start with a brief description of the basic concept of the light-harvesting efficiency (LHE), and then give a review on five typical branches representing the significant advances in this area, including LHE (λ) = 1 − 10 -Г σ(λ) (1) where Г is the number of moles of sensitizer per square centimeter of projected surface area of the film, and σ is the absorption cross section in units of cm 2 /mol obtained from the decadic extinction coefficient (units of M -1 cm -l ) by multiplication with 1000 cm 3 /L (Nazeeruddin et al., 1993). The LHE is directly determined by the surface concentration of the dyes in the film, and the molar extinction coefficient of dye. The LHE, together with the quantum yield of charge injection (φ inj ) and the efficiency of collecting the injected charge at the back contact (η c ), determines the incident monochromatic photon-to-current conversion efficiency (IPCE), defined as the number of electrons generated by light in the external circuit divided by the number of incident photons (Equation 2).
Previous studies proved that φ inj is a wavelength independent parameter, and almost all of photons absorbed by the sensitizer are quantitatively converted to the conduction band electrons (Nazeeruddin et al., 1993). Meanwhile, the time-resolved laser photolysis experiments showed that the injected electrons can percolate without significant loss through the network of interconnected particles present in the nc-TiO 2 film, and the back reaction is relatively slow, indicating that the majority of the injected charges is able to reach the back contact, and the influence of η c on IPCE is minor . So the wavelengthdepedent LHE has predominant effects on the IPCE and the PCE of DSC.
In essence, the LHE of DSC is the electrical response of the photovoltaic device to the solar spectrum projected on earth. It is predominantly related to there factors, the wavelength dependent light-absorbing capacity of the dye molecules, the quantity of the dye molecules adsorbed on the porous electrode, and the optical path of the incident light traveled in the electrode (determining the collision numbers of the light with the dye molecules). As a wavelength-dependent parameter, the LHE is usually evaluated by the IPCE spectrum. Figure 2 gives the IPCE spectrum of a typical DSC with nc-TiO 2 electrode, N719 dye and I 2 /I 3   electrolyte system, and its comparison with the standard solar irradiation spectrum (ASTM  G173). To characterize the response of DSC to the solar irradiation, the IPCE spectrum can be divided into four wavelength zones, i.e., the strong absorption zone between 500-550 nm, two moderate absorption zones between 400-500 nm and 550-700 nm, and the weak absorption zone above 700 nm. While the broadband or near-infrared absorption sensitizers are devised to harvest the sun light beyond 700 nm, the modulation of the optical path of the incident light by introducing scattering or reflection is mostly used to improve the light harvesting in two moderate absorption zones. From the viewpoint of the photoanode film (i.e., the porous electrode), there are several measures to improve the LHE, including 1. increasing the internal surface area of the electrode, which has the potential to improve the dye-loading capacity of the photoanode over a specific film thickness and area; 2. increasing the optical path of the incident light in the electrode, by introducing scattering centers in the bulk film, by introducing the scattering layer on the top of nanocrystalline electrode, and by constructing hierarchical structures possessing strong light scattering effects; and 3. enhancing the absorption of dye molecules by introducing plasmonic metal-semiconductor structures. In current days, the nanocrystallite electrode is confronting the maximum photocurrent theoretically achievable with the Ru-based sensitizers (e.g. N719 dye), and the quest for the novel nanostructured photoanode has becoming prosperous Tetreault et al., 2012). Among numerous innovative nanostructured photoanodes, a considerable amount of studies are directed to improve the LHE of the photoanode, in view of the moderate surface area, the moderate electron transporting property of nc-TiO 2 electrode, and the intrinsically low-surface-area nature of the traditional submicron scattering layer. In the following sections, we will introduce five typical research branches in this area, to provide readers with a comprehensive and in-depth overview on the development of the LHE studies in DSC.

Figure 2.
The light-harvesting capacity of N719 sensitized nc-TiO 2 photoanode in DSC and its comparison with AM 1.5 Global solar spectrum.

Mesoporous photoanodes with high surface area
It is essential for the photoanode in DSC to possess a high internal surface area, to adsorb sensitizer as much as possible. Typically, the hydrothermal nc-TiO 2 with the particle size of 15-30 nm is used to fabricate the porous electrode, which usually exhibits the BET surface area of 50-100 m 2 /g. While these nanocrystallite photoanodes perform fairly well and occupy the highland of DSC with the best efficiency, their moderate dye-loading capacity motivate researchers to seek other materials with higher surface area. In fact, there are several materials which do possess much higher surface area than nc-TiO 2 electrode, such as the mesoporous materials, aerogel, and metal-organic framework (MOF), whose surface area usually range from several hundred to three thousand m 2 /g. It is greatly intriguing to construct photoanodes using these high-surface-area materials, which is expected to remarkably improve the dye-loading capacity of the photoanode and the PCE of DSC in case a good charge transport can be maintained in these mesoporous structures. Up to date, mesoporous TiO 2 in the form of powder, sphere, bead, or film, and aerogel have been used to synthesize the photoanode, which usually exhibited the surface area of 100-1100 m 2 /g. Here we present some typical mesoporous photoanodes prepared by the high-surface-area TiO 2 mesostructures and TiO 2 aerogel.
Mesoporous powders possessing a large volume of mesopores and irregular micron/ submicron secondary particles are the mostly-used materials in constructing the mesoporous photoanodes. The template methods are the common strategy to prepare these mesoporous powders. For example, T. K. Yun et al. prepared two different mesoporous TiO 2 powders, using a soft template of tri-block copolymer and a hard template of mesoporous ZnO/ Zn(OH) 2 -composite, respectively (Yun et al., 2011). Both mesoporous TiO 2 possessed the same high surface area (BET ~210 m 2 /g) but with different pore sizes of 6.8 (for soft template) and 3.0 nm (for hard template). Different photovoltaic performances were observed. While the photoanode using the mesoporous TiO 2 having larger pores showed the PCE of 6.71%, higher than P25 TiO 2 nanopowders based photoanode (η= 5.62%), only half the performance (3.05%) was observed for the mesoporous TiO 2 having small pores. The work indicated that, a suitable selection of the pore sizes was important for mesoporous materials used in DSC, and a too small pore size would inhibit the diffusion of dye molecules through the pores, thus greatly reducing the uptake of the dye molecules.
Beside the template method, the external fields such as the microwave irradiation can also be very useful in synthesizing mesoporous TiO 2 powders. C. H. Huang et al. reported the rapid synthesis of mesoporous TiO 2 via a microwave-assisted hydrothermal route using water-soluble titanium citrate complexes as the precursor, and obtained TiO 2 powders showed the surface area of 217-323 m 2 /g, and the pore diameter of 5.8-6.9 nm (Huang et al., 2011). The photoanode based on this mesoporous TiO 2 exhibited the highest conversion efficiency of 7.1% (active area: 0.25 cm 2 ). Also, the cell exhibited good long-term stability, and the PCE of a 1 cm 2 device decreased merely from 4.8% to 4.3% after being stored for more than 490 h.
For most photoanodes using the high-surface-area mesoporous materials, it is required to prepare the mesoporous powders first, and the final electrode is obtained via the doctorblade method using the mesoporous paste prepared from the mesoporous powders. This is a relatively complicated process. In contrast, W. Chen et al. developed a facile one-step preparation of crack-free thick ordered mesoporous TiO 2 films, via the combination of ''doctor blade'' technique and the evaporation induced self-assembly method (Chen et al., 2007). By employing the as-synthesized mesoporous film (7 μm in thickness) as the photoanode, a PCE of 6.53% was obtained at 30 mW/cm 2 light intensity, illustrating the potential of this simple strategy in constructing powerful DSC devices.
While the mesoporous materials of pure TiO 2 give a promising potential to improve the LHE, the introduction of other oxides (e.g., Al 2 O 3 , SiO 2 ) in the mesoporous structure of TiO 2 opens an alternative route to enhance the LHE of DSC. As an example, J. Y. Kim et al. prepared highly ordered mesoporous Al 2 O 3 /TiO 2 via a sol-gel reaction and evaporation-induced self-assembly route using Pluronic P123 as the structure-modifying reagent (Kim et al., 2011). Obtained mesoporous powders possessed an average pore size of 6.33-6.58 nm, and BET surface area of 181-212 m 2 /g, and the content of Al 2 O 3 had significant effects on the BET surface area. The thin Al 2 O 3 layer was located mostly on the surface of mesopore (as seen in X-ray photoelectron spectroscopy (XPS) spectra in Figure 4c), blocking the charge recombination during the photon-to-electron conversion process. The photoanode based on the mesoporous Al 2 O 3 /TiO 2 (1 mol % Al 2 O 3 ) exhibited ~10% efficiency improvement com-pared to the pure mesoporous TiO 2 electrode (η increasing from 5.88% to 6.50%). Though the authors didn't compare their results with traditional nanocrystalline counterpart, the much higher surface area of the mesoporous electrode than P25 TiO 2 may possibly lead to higher dye-loading quantity, and corresponding higher light-harvesting capacity. Also, the introduction of Al 2 O 3 barrier layer in mesoporous TiO 2 illustrated a viable and versatile route to further improve the functions of the mesoporous photoanodes. Though above mesoporous powders with irregular secondary particles are promising in improving the dye-loading capacity of the photoanode, it is usually difficult for them to obtain high quality electrodes with good transparency, due to the poor workability of the paste produced from the irregular micron/submicron particles. Therefore, it is intriguing to prepare mesoporous TiO 2 with regular secondary particles, e.g., TiO 2 mesoporous spheres. D. Chen et al. did a good job on this area, and synthesized uniform, crystalline, and mesoporous TiO 2 beads in size of ~800 nm through a combination of sol-gel and solvothermal processes, which exhibited the surface area of 108.0 m 2 /g and tunable pore sizes (pore diameter: 14.0 2 2.6 nm) (Chen et al., 2009). Due to the submicrometer-sized particle diameter and the high specific surface area, the mesoporous TiO 2 beads enhanced the LHE without sacrificing the accessible surface for dye loading, thereby increasing the PCE compared to P25 nanoparticles. The photoanode based on these mesoporous TiO 2 beads exhibited an overall light conversion efficiency of 7.20% (V oc = 0.777 V, J sc = 12.79 mA/cm 2 , FF = 0.72), significantly higher than that derived from standard Degussa P25 TiO 2 electrode with similar thickness (5.66%).
In a further study, the group selected Ru(II)-based dye C101 and C106 to match with the mesoporous TiO 2 beads, and realized an overall PCE of greater than 10% (η=10.7%) using a single screen-printed TiO 2 layer cell construction (without an additional scattering layer). Hence, a delicate modulation of the micron/submicron size/ shape of mesoporous materials, and a careful selection of the dye to match the dye's photon-absorption characteristics with  . Al-so, the authors investigated systematically the effects of ZnO thickness on SiO 2 aerogel template on the photovoltaic performance, and found a nearly linear relationship between the key photovoltaic index and the thickness of ZnO layer (Figure 6c,d), with the optimal ZnO thickness of 10.5 nm. These results indicated that, on one hand, the aerogel film in combination of ALD or other physical or chemical deposition processes was a powerful candidate to prepare mesoporous photoanode with controllable microstructures. On the other, the use of the dangerous supercritical drying process in the synthesis of aerogel, together with the expensive and complicated ALD process involved in the preparation of ZnO/TiO 2 layer, limited the popularization of this method greatly. There were very few studies on the aerogel photoanodes in recent five years. To crack above problems, X. D. Gao et al. developed a low-cost sol-gel and ambient-drying route to fabricate SiO 2 -TiO 2 hybrid aerogels, which possessed a very high surface area of 500-1170 m 2 /g, and incorporated them into nc-TiO 2 electrode, obtaining the aerogel hybrid photoanodes . These hybrid photoanodes yielded significantly higher photocurrent densities than nc-TiO 2 photoanodes due to the increased dye-loading capacity and the enhanced visible scattering effect, showing the highest efficiency of 9.41% at optimal condition, 16% higher than the TiCl 4 -treated TiO 2 photoanode modified with a conventional scattering layer. This work indicated that, the hybrid photoanodes integrating both the high-surface-area mesoporous materials and nc-TiO 2 represented a powerful and viable route toward high-efficiency DSCs.
In brief, current studies on the mesoporous photoanodes built up by the high-surface-area mesoporous materials have unambiguously proved that, the application of high-surfacearea powders/beads/aerogel in the photoanode can effectively increase the dye-loaing capacity and the light scattering of the electrode, and correspondingly improve the LHE of DSC in most cases. The preparation of mesoporous materials with regular and uniform size/ shape in micron/submicron scale, and their integration with nc-TiO 2 or other nanostructures, seem feasible to achieve high conversion efficiency. A remaining problem in these high-surface-area mesoporous materials is the inferior electron-transporting property, intrinsically originating from their lower particle coordination number than the common nc-TiO 2 porous electrode. A possible solution may be the development of hybrid electrodes by incorporating metal, carbon nanotube, or graphene etc into the mesoporous structures. A mesoporous material possessing both high surface area and good conductivity is by all means necessary for highly efficient DSCs.

Hierarchical photoanodes
Hierarchical nanostructures, those with different morphology at different length scale and usurally characterized by the high surface area, are intresting candidates for efficienct photoanodes in DSC. This special class of nanostrcutures is spurred at the low surface area and consequently the low PCE of nanowire based photoandes, and have receivd wide-spread and intensive studies in recent years. Hierarchical ZnO and TiO 2 in the form of grass, forest, popborn ball, and tube etc. have been developed, and a wide range of chemical and physical methods have been used to prepare the hierarchical structures, including the hydrothermal synthesis, sol-gel, chemical bath deposition, pulsed laser deposition, photolithography, and so on. Due to the huge literature in this field, here we merely present some representive results in recent three years.
ZnO nanowire array may be the piorneer in the nanostructured photoanode since the outstanding work of M. Law et al. (Law et al., 2005). However, the PCE of nanowire based photoanode was low (mostly 1%-3% for a long time) mainly due to its low surface area. So the development of hierarchical ZnO and their application in DSC are wide-spread recently. As an example, Xu et al. reported a two-step synthesis process to produce hierarchical ZnO nanoarchitectures (Figure 8a,b), in which ZnO nanosheet arrays were first prepared by the pyrolysis of the precursor Zn 5 (OH) 8 Cl 2 electrodeposited on conductive glass substrates, and then dense ZnO nanowires were grown on the surfaces of the primary ZnO nanosheets by the chemical bath deposition method . Due to the better dye loading through the increased internal surface area and the higher light scattering behavior through extend-ing the optical path length within the photoanode, the DSC based on the hierarchical ZnO nanowire-nanosheet architectures showed a PCE of 4.8%, nearly twice as high as that of DSC constructed using a photoanode of bare ZnO nanosheet array.
Apart from these hierarchical structures with relatively simple structures, several groups developed ZnO or TiO 2 forest (Figure 8c,d), more complicated hierarchical structures. F. Sauvage et al. prepared the hierarchical assemblies of nanocrystalline particles of anatase TiO 2 (Sauvage et al., 2010), by a pulsed laser deposition method, via the fine modulation of the plasma expansion dynamics by means of a reactive atmosphere during the ablation process.
In combination with the high molar extinction coefficient heteroleptic C101 dye, the authors achieved 3.1% power conversion efficiency for 2 um thick films and 5.0% for films 7 um thick, demonstrating the great potential of the physical vapor deposition method in growing hierarchical nanostructured photoanodes with high conversion efficiency. 63% v.s. 0.45%). The enhanced surface area for higher dye loading and light harvesting, and the reduced charge recombination by providing direct conduction pathways along the crystalline ZnO "nanotree" multi generation branches, were responsible for the efficiency improvement. . Exhibiting the BET surface area of 119-331 m 2 /g, DSC devices based on these multishelled NTA exhibited the highest efficiency of 6.2%, higher than that based on the singlelayer TiO 2 nanotube array (3.35%) at similar experimental parameters (Figure 9f).
Apart from the hierarchical nanowire/nanotube array, there are also a few studies which modulate the pores in the photoanode to realize the hierarchical structures. C. Y. Cho et al. reported an interesting method to generate hierarchical electrodes consisting of meso-and macroscale pores (Cho et al., 2011). Mesoscale colloidal particles and lithographically patterned macropores were used as the dual templates, with the colloidal particles assembled within the macropores. An infiltration of TiO 2 into the template and subsequent removal of the template produced hierarchical TiO 2 electrodes for DSC (Figure 10a). While the holographic lithography defined the macroporous networks, the colloidal crystal templates provided three-dimensionally organized mesoscale pores with a uniform size distribution (Figure 10 b-e). Owing to the strong scattering and the suppression of charge recombination in the hierarchical TiO 2 electrodes, the photovoltaic performance of the cell was comparable with nc-TiO 2 electrodes, showing a maximum efficiency of 5.0% with 50 nm pores and 6 μm thickness. In general, the hierarchical photoanodes, though possessing obviously higher internal surface area than the nanowire/nanotube/nanoplate counterparts and higher conductivity than nc-TiO 2 electrode, haven't show a predominant advantage over other materials up to now. The PCE of most photoanodes based on the hierarchical structures is still much lower than the state-of-the-art performance of nc-TiO 2 electrode. A possible vent for the hierarchical photoanodes may be the further increase of its internal surface area while maintaining its high electron-transporting nature. Also the combination of the hierarchical structures with the mesoporous material is a meaningful and feasible strategy.

Scattering layers on nc-TiO 2 electrode
The photoanode of DSC is traditionally composed of nc-TiO 2 film, which is basically transparent for visible light (Rayleigh scattering, mainly), resulting in a considerable part of light shone on the DSC transmitting through the TiO 2 layer without interacting with the sensitizer. To overcome this problem, scattering layers with different structures, including scattering centers and upper scattering layers, have been employed. The theoretical studies since 1998, and subsequent substantial experimental studies on scattering layers based on the submicron particles of TiO 2 , ZrO 2 , ZnO, nanowire, hierarchical structures, mesoporous spheres, photonic crystals, and upper-conversion materials, comprise a vivid picture about the research activities in this area. By briefing some important advances, we describe here some novel scattering layers developed in recent years.  (Figure 12). They proposed that the diffuse scattering layers acting as the back reflector provided the largest achievable light harvesting efficiencies, which determined an optimum overall performance, for electron diffusion lengths longer than the electrode thickness. When the electron diffusion length was shorter than the electrode thickness, the embedding of the diffuse scattering particles in the nanocrystalline paste yielded a better output even when the light harvesting was not optimal. These results can provide helpful guidance for us to design and prepare the scattering structure of the photoanode.  (Koo et al., 2008). The walls of the obtained hollow spheres were composed of nanocrystalline anatase TiO 2 particles with mesoporous structure (Figure 14 a-c), which endowed them 5 times higher dye-loading capacity (N719 dye: up to 5.0×10 -5 mol/g) than that of the normally used 400 nm-diameter scattering particles with flat surfaces. The NeHS particles scattered incoming light effectively as confirmed by the reflectance spectroscopy (Figure 14 d,e). When the NeHS TiO 2 particles were used as the secondary scattering layer in DSC, a substantial improvement in the efficiency has been achieved, with the highest conversion efficiency of 10.34%. Huang F et al. used the submicrometer-sized mesoporous TiO 2 beads as the dual function scattering layer for the nanocrystalline electrode, and observed similar results, demonstrating that the dual-function scattering layer represented a powerful alternative for the traditional large-particle scattering layer for high-efficiency DSCs (Huang et al., 2010).  The cell based on this hierarchical anatase TiO 2 exhibited a conversion efficiency of 5.53%, superior than commercial TiO 2 (P25). The much higher optical reflectance of the hierarchical structures than the nanorod array and nanoparticle films ( Figure 15) proved that this morphology was beneficial to the light-scattering capacity of the photoanode.
Yang et al. reported an innovative bilayer TiO 2 nanofiber photoanode, which combined both smaller (60 nm) and larger (100 nm) diameter TiO 2 nanofibers fabricated by electrospinning . The smaller diameter nanofiber (SNF) layer with a high surface-to-volume ratio was used to adsorb sufficiently dye molecules and directly transport electrons released from excited dyes. The bigger-diameter nanofi ber (BNF) layer worked as light scattering, adsorbed sufficient dye molecules for energy harvest, and provided higher pore volume in BNF to facilitate electrolyte diffusion for regenerating sensitized dye molecules in the photoanode. Therefore, this bilayer composite nanostructure photoanode offered excellent dye-loading, lightharvesting, and electron-transport properties. The PCE of DSC was improved from 7.14% for the single-layer to 8.40% for the bilayer of TiO 2 nanofiber photoanode, representing an increase of 17%.  As the conclusion of this section, we introduce a special scattering layer using up-converting materials. G. B. Shan et al. utilized the hexagonal nanoplatelets of β-NaYF 4 :Er 3+ /Yb 3+ , a typical up-conversion phosphor owing to Er 3+ doping with Yb 3+ codopant assisting in the energy transfer process, as the second layer on nc-TiO 2 electrode ( Figure 17) (Shan et al., 2011). Approximately 10% enhancements of photocurrent and overall DSC efficiency were achieved by the addition of the external layer, which exhibits two functions of light reflecting and near-infrared (NIR) light harvesting. Though the overall PCE of the cell and the improvement were moderate, the work opened a new opportunity to use some advanced functional materials to enhance the light-harvesting functions of DSC.
In summary, a wide variety of nanostructures including submicron particles, nanowires, nanofibers, mesoporous beads, spheres, nanopalletes etc have been used as the top scattering layer of DSC, which are proved to be very effective in improving the LHE and PCE of DSC, especially for the dual-function materials possessing both high surface area and large-particle size. At the same time, the application of multiple function materials such as the upconverting nanopalletes in the scattering layer sheds a new light for further development.

Plasmonic DSCs
Localized surface plasmon resonance (LSPR) behavior is an intriguing characteristic of metal nanoparticles (NPs), which is generated by the resonance between electric fields of electromagnetic waves and free electrons in metal NPs. The use of plasmonic effects has been proposed as a promising pathway to increase the light absorption in active layers of solar cells, and has been demonstrated on several thin-film solar-cell materials such as amorphous silicon, gallium arsenide, polymers, and DSC. Previous studies on the plasmonic solar cells with thin active layers have proved the beneficial functions of the LSPR on the photocurrent density through such following channels:

scattering of light by metallic NPs into dielectric-like waveguide modes of the solar cell;
3. coupling to propagating surface plasmon polariton (SPP) modes (Ding et al., 2011).
This has established a solid foundation for the rapid development of plasmonic DSCs in recent years.
The mostly-used metal NPs in the plasmonic DSCs are Ag and Au. While earlier studies have found the corrosion of metal particle in the electrolyte or the undesired introduction of electron-hole recombination by the plasmonic particles, it has become a common practice to protect the Ag or Au NPs by a thin TiO 2 or SiO 2 layer. Coupled with the standard thick nc-TiO 2 electrode, these plasmonic photoanodes performed fairly well compared with the control samples. . They deposited Ag NPs on TiO 2 framework via the photo-reduction of Ag + from dissolved AgNO 3 , and protected them by a thin TiO 2 layer by refluxing Ag-TiO 2 electrode in isopropyl alcohol and subsequent annealing at 370 o C. The electrode incorporating the protected silver NPs (Ag-TiO 2 ) showed enhanced extinction of a subsequently adsorbed dye (the ruthenium-containing molecule, N719), realizing an overall conversion efficiency of 8.9% and 25% improvement over the performance of otherwise identical solar cells incorporating silver NPs lacking protection ( Figure 18). Roughly half the improvement could be traced to the increased dye loading by the photoanodes following silver incorporation, with the remaining improvement coming from the localized surface plasmon resonance (LSPR) enhancement of the effective absorption of N719 dye molecules.
J. Qi et al. coated TiO 2 thin layer (~2 nm) on Ag NPs and incorporated this Ag@TiO 2 nanostructures into the TiO 2 photoanode (Qi et al, 2011), which can easily transferred the carriers to surrounding TiO 2 NPs in contact with the shell, while providing a good protection for the metal NPs. By utilizing Ag@TiO 2 core-shell nanostructures, the optical absorption of dye molecules in solution and in thin film was enhanced by the strong localized electric field generated by LSPs (Figire 19 a,b). By incorporating Ag@TiO 2 NPs, the PCE of DSCs with very thin photoanodes (1.5 μm) was increased from 3.1%to 4.4%, and a small amount of Ag@TiO 2 NPs (0.1 wt%) improved efficiency from 7.8% to 9.0% while decreasing the photoanode thickness by 25% for improved electron collection (Figure 19c,d).   (Brown et al., 2011). They successfully overcame the detrimental effects of the incorporating "bare" metal NPs into the bulk DSCs, by coating the Au NPs with a thin shell of silica, which could sustain the sintering process of the TiO 2 with negligible influence to the optical properties, resist the corrosion from the iodide/triiodie electrolyte, and enhance the photocurrent generation and solar cell performance as a result of light harvesting by the metal NPs. With a significant increase in the shortcircuit photocurren, the best solid state plasmonic cell sensitized with Z709 exhibited a PCE of 4.0% (Figure 20). The readily tunable optical properties of metallic NPs, through the use of structures such as nanobars, nanostars, or other complex shapes, provided extensive scope for further work to fully optimize the technology. NPs, they improved the cell efficiency from 9.3% for an N719 sensitized device to 10.2% upon incorporation of 0.7% Au@SiO 2 and to 9.8% upon loading of 0.7% Au@TiO 2 NPs. The plasmonic effect was observed in Au@SiO 2 incorporated DSC, which produced higher photocurrent ( Figure 21). However, in Au@TiO 2 incorporated DSC, Au NPs underwent charge equilibration with TiO 2 NPs and shifted the apparent Fermi level of the composite to more negative potentials, which resulted in a higher photovoltage. These observations opened up new opportunities to introduce both these paradigms and to synergetically enhance the photocurrent and photovoltage of DSC.
In brief, the plasmonic photoanodes are still a newly-emerged research area in DSC, spanning a wide range of materials (metal, semiconductor) and optical and electrochemical phenomena (localized surface plasmon resonance, scattering, electron-hole recombination). It is currently the main task for scientists to seek a suitable plasmonic structures (metal-oxide core-shell structures) and effective strategies to incorporate them into the nanocrystalline or nanostructured electrodes. While at present the nanoparticles of Au and Ag have been attempted to produce plasmonic DSC, further efforts may be extended to other metals (Ti, Zn, etc.) or morphologies with complex structures of metals (cubic, hexagonal, star-form etc.). Also, in-depth explorations on the mechanism of the plasmonic resonance embedded in the nanoporous electrode, and their detail influences on the photo-to-electric conversion process, are very useful to promote the development of this intriguing area.

Photonic crystals based photoanodes and more
Photonic crystals (PCs) are materials that exhibit periodicities in their refractive index on the order of the wavelength of light, and thus provide many interesting possibilities for "photon management" (John, 1987). PC layers can be coupled to the charge-generating layer of DSC, and increase the LHE via such following mechanisms, Compared with the traditional scattering techniques using geometrical optic-based elements (incoherent scattering layers, or high-reflectivity metallic mirrors), the application of PCs can obtain a more significant improvement on the LHE via the delicate control of the reflected, diffracted, or refracted light passing through the cell, apart from the apparent advantage in obtaining the transparent and colorful DSC rather than the opaque ones. The major limitation for the incorporation of PCs (especially the self-assembled 3D PCs) into DSC is the incompatibility between the fabrication routes for the photonic structures and the photoanode. So in the current stage, the preparation technology of PCs and their coupling to nc-TiO 2 layer represent the emphasis of many studies. In this section, we give several examples of PC layer in DSC with distinctive features in their structure. nm. The photocurrent of DSC based on the 420 nm-QIO layer was higher than the commercial scattering layer-based DSSC, achieving an efficiency of 5.7% ( Figure 22).  S. Colodrero, et al. couped a porous and highly reflecting 1D photonic crystal into a nc-TiO 2 electrode, by the deposition of alternate layers of NPs of different compositions by spincoating (Colodrero et al., 2009). While the porous mesostructure allowed the electrolyte to flow through it and soak the electrode without interfering with the charge transport through the cell, the PC layer with a thickness of just half a micrometer efficiently localized the incident light within the nc-dyed TiO 2 electrode in a targeted wavelength range ( Figure 24). Consequently, the average PCEs were improved to between 15 and 30% of the reference value attained for standard electrodes, while the open-circuit voltage and the transparency of the cell remained intact, contrary to what happened when the scattering layers were employed to improve the light harvesting.
Apart from the intensive studies on the PCs based DSC, other optical elements such as planar waveguide were also explored to enhance the LHE of DSC. Typically, Z. L.
Wang's group demonstrated a novel three-dimensional DSC by integrating planar optical waveguide and nanowires (NWs) (Wei et al., 2010). The ZnO NWs were grown normally to the quartz slide, and the 3D cell was constructed by alternatively stacking a slide and a planar electrode. While the slide served as a planar waveguide for light propagation, the 3D structure effectively increased the light absorbing surface area due to internal multiple reflections without increasing the electron path length to the collecting electrode. Obtained 3D DSCs exhibited a significant improvement in energy conversion efficiency by a factor of 5.8 compared to the planar illumination case. In summary, the optical elements such as the photonic crystals and the planar waveguide can provide exciting perspective to deliberately modulate the optical path of the incident lights and to significantly enhance the LHE of DSC. However, the ultimate PCE of corresponding DSCs is still far lower than that of nc-TiO 2 counterpart. How to tackle with the preparation problems related to the PC layers, and to avoid the possible adverse effects of the PC layers on the dye-loading and charge-transporting properties of the original electrode, are the kernel issues which should be solved in near future.

Conclusion
In conclusion, the light-harvesting efficiency (LHE) of the photoanode film has determinative effects on the power conversion efficiency of DSC. The deliberate modulations of the internal surface area of the nanoporous electrode and the optical path of the incident light are currently the main pathway to enhance the LHE of DSC. A wide range of novel materials or techniques have been utilized to improve the LHE of the electrode, including the high-surface area mesoporous nanostructures or aerogel, scattering-enhanced hierarchical nanostructures, up-conversion materials, plasmonic core-shell structures, and photonic crystals etc. However, while most of reported work realized obvious enhancement on one or more specific capacities of DSC, such as the dye-loading properties, optical scattering, or improved harvesting of near-infrared light, very few study can demonstrate high device performance comparable with the state-of-the-art nc-TiO 2 cell. The intrinsically different particle size, microstructures, preparation strategy of these novel materials from the traditional nc-TiO 2 electrode will inevitably result in significant changes in the microstructure or the optical/ electrical properties of the photoanode, which may greatly impair the final performance of the device. How to balance the advantageous and disadvantageous factors involved in these new-type photoanodes, and realize the solid improvement of the overall performance of DSC, are the emphasis of the scientists in near future. After all, the photoanodes based these novel materials or structures are still in an infant stage, containing infinite possibilities to improve or even revolutionize the basic principle and performance of the traditional DSC. We believe, via the intensive and extensive efforts of the scientist all over the world and through the collaborative studies among different areas spanning from material, chemistry to physics, the new-type photoanodes will certainly usher a brilliant future.