Recent Progress in Dye-Sensitized Solar Cells for Improving Efficiency : TiO 2 Nanotube Arrays in Active Layer

Dye-sensitized solar cells (DSSCs) have been widely studied due to several advantages, such as low cost-to-performance ratio, low cost of fabrication, functionality at wide angles and low intensities of incident light, mechanical robustness, and low weight. This paper summarizes the recent progress in DSSC technology for improving efficiency, focusing on the active layer in the photoanode, with a part of the DSSC consisting of dyes and a TiO 2 film layer. In particular, this review highlights a huge pool of studies that report improvements in the efficiency of DSSCs using TiO 2 nanotubes, which exhibit better electron transport. Finally, this paper suggests opportunities for future research.


Introduction
For nearly two centuries, mankind has employed fossil fuels as the primary energy source, and now we are facing serious problems as a result.Excessive emission of carbon dioxide and other greenhouse gases leads to environmental risk [1].The price of fossil fuels continues to rise because of the increasing worldwide energy consumption.The need for an affordable, sustainable, and carbon-free renewable energy source is greater than ever, as the rapid growth rate of renewable energy consumption suggests [2,3].Amongst all the renewable energy sources, solar energy has been regarded as very promising due to the abundance of its resource-sunlight-and the fact that it yields no harmful byproducts.The amount of solar energy that radiates upon earth in one hour is equivalent to the annual energy need of mankind [4].Such high energy output, however, has not been achieved due to thermodynamic and technical constraints as well as few other reasons.Conventional commercial solar cells, which use crystalline and polycrystalline silicon, achieve over 20% energy conversion efficiency for residential end users [5].However, there are some problems, such as complicated and difficult fabrication processing and high cost [6].
Dye-sensitized solar cells (DSSCs) are one potential alternative to silicon solar cells.DSSCs separate the light absorption and charge transfer processes, unlike Si-based cells.The organic dyes, or sensitizer molecules, adsorbed on the surface of the metal oxide nanostructure take the role of absorbing the incoming light, and the rest of the structure transfers the generated charge.An early version of a DSSC fabricated by O'Regan and Grätzel used ruthenium-based dye and 10-m-thick porous TiO 2 nanoparticle films for the photoelectrode [7].The cell yielded 7% power conversion efficiency (PCE), which was high enough to motivate further research on cells with similar structures.DSSCs provide several advantages, such as low price-to-performance ratio, low processing cost, ability to work at wide angles and low intensities of incident light, mechanical robustness, light weight, and aesthetically appealing transparent design.DSSCs have a number of remarkable properties that allow them to be used for several niche applications; they are low in weight and able to display various colors and transparency, and they work effectively in a broad range of wavelengths [8,9].However, there are still several weaknesses that require improvement, such as relatively low efficiency, high cost of the ruthenium dyes, conducting glass, and platinum, temperature stability problems, and low scalability.Among the various potential solutions, some of the most widely investigated ones are the development of new sensitizers to cover a broader solar spectrum and the fabrication of new photoanode structures with enhanced light utilization.Studies have reported several different methods to improve the efficiency of DSSCs, such as the use of panchromatic sensitizers or 3-dimensional metal oxide nanostructures [10][11][12][13][14].This review highlights another huge pool of studies that report improvements in the efficiency of DSSCs, especially those that employ the use of TiO 2 nanotubes.Moreover, we will introduce recent papers on high PCEs and the use of TiO 2 nanotube arrays in DSSCs.These papers describe the use of the space in TiO 2 nanotube array channels and how to overcome the barrier layer that exists at the bottom of these arrays to achieve a high PCE in DSSCs.

Principle of DSSCs
A DSSC consists of a photoelectrode, counter electrode, and electrolyte.The photoelectrode is made up of a transparent conductive oxide (TCO) glass substrate, a metal oxide such as mesoporous TiO 2 nanoparticles and a sensitizer such as ruthenium-(Ru-) complex dye.The electrolyte consists of a redox couple and diffusing reducing/oxidizing agent ions, and the counter electrode is coated with Pt as shown in Figure 1(a).In general, fluorine-doped tin oxide (FTO) glass is used on DSSCs instead of indium tin oxide (ITO) glass because of the better thermal stability at high temperatures [15].During the sintering at 450-500 ∘ C, which crystallizes the mesoporous TiO 2 nanoparticles, the sheet resistance of ITO increases and affects the energy conversion efficiency of the DSSCs.However, the sheet resistance of FTO is independent of the temperature up to 500 ∘ C, which makes it favorable for applications that utilize the sintering step.TiO 2 nanoparticles are primarily used as electron acceptors on DSSCs that incorporate ruthenium dye.The counter electrode consists of FTO glass coated with platinum (Pt) or carbon to resist the oxidation and to work as a catalyst during the redox reaction in the electrolyte.
The principle of DSSCs is shown in Figure 1(b).When sensitizers absorb the light energy, and electrons are generated by the sensitizers and are transferred to the metal oxide films.During the hopping process in the metal oxide films, the generated electrons are transferred to the TCO and circulated.After the circulation, the sensitizers are reduced by an iodide redox process.In general, a Ru-complex dye is used as the sensitizer, and TiO 2 nanoparticles are used as the electron acceptors on DSSCs due to their high efficiency [20][21][22].
The energy diagram shown in Figure 1(c) demonstrates the electron transfer process of DSSCs.When the Rucomplex dyes absorb the light, the electrons change from the ground state to the excited state within a time on the order of nanoseconds (ns), as shown in the reaction (Figure 1(c)-( 1)): Although the detailed mechanism of the injection process is under debate, the electron injection from the Ru-complex dye into the conduction band of mesoporous TiO 2 nanoparticles takes about a femtosecond (fs) following the reaction shown in Figure 1(c)-( 2).The energy level of electrons in the Rucomplex dye is higher than that of the conduction band of mesoporous TiO 2 nanoparticles, which enables the transfer of electrons into the mesoporous TiO 2 nanoparticles: The excited state of the Ru-complex dye lasts around 20-60 ns, which is long enough for kinetic competition between electron injection and excited state decay in the Ru-complex dye.
The electrons injected into TiO 2 nanoparticles diffuse through mesoporous TiO 2 nanoparticles in about a millisecond (ms) in the reaction shown in Figure 1(c)-(3).However, when the injected electrons are trapped on mesoporous TiO 2 nanoparticles, they return to the oxidized Ru-complex dye by the reaction shown in Figure 1(c)-( 4) or reduce the triiodide by the reaction shown in Figure 1(c)- (5).These processes are called as the recombination: The Ru-complex dye is reduced by an iodide ion to accept the electron, as shown in the reaction in Figure 1(c)-( 6): The injected electrons on TiO 2 nanoparticles drive the external circuit and perform work, as in the reaction shown in Figure 1(c)- (7).
When the triiodide ion reduces to iodide to accept the electrons from the counter electrode by the reaction shown in Figure 1(c)-( 8), the iodide ion causes the electron in the Rucomplex dye to return to the ground state, with the reaction shown in Figure 1(c)-(6).
Figure 1(d) illustrates the performance of DSSCs, and the solar power () can be calculated by the equation: The energy conversion efficiency of DSSCs is described by the ratio of the maximum electrical power output  max to the solar energy input .The maximum electrical power output  max is calculated by the maximum current  mp and the maximum voltage  mp : (6) E versus NHE (V) Electrolyte Dye Counter electrode FTO (7) Load The solar power input is taken as the product of the irradiance of the incident light that is measured in W/m 2 : The fill factor  is a measure of the quality of the solar cell, and it can be calculated by comparing the maximum shortcircuit current  sc and maximum open-circuit voltage  oc to the maximum current  mp and maximum voltage  mp : (100 mW/cm 2 ) is the solar cell testing standard under terrestrial conditions with air mass AM1.5.The ideal power density of DSSCs is the product of  sc and  oc .The maximum power density is the product of  mp and  mp , and the difference between the two comes from the resistances of the cell, electrolyte, electrode, and other components.The energy conversion efficiency of DSSCs is related to the ideal power density and maximum power density as shown in Figure 2(a).When the fill factor is equal to 1,  sc and  oc are the same as  mp and  mp .The series resistance   includes the cell, electrolyte, electrode, and interface resistance.The increasing   means that electrons are lost, so the voltage decreases earlier as shown in Figure 2(b).The shunt resistance  sh results from the recombination of electrons that are transferred in another way or are trapped by something.The decreasing  sh means that there are few recombination sites, so the electrons are transferred to each layer as shown in Figure 2(c).

Dye
The dyes commonly used for DSSC applications share certain characteristics [23]: (i) excellent spectral absorption profile,

Power density
Voltage (V) like the panchromatic absorption; (ii) a surface-anchoring group, such as a carboxylate or phosphonate group; (iii) a LUMO matching the edge of the conduction band of the photoanode; (iv) a HOMO comparable to the electrolytes' Fermi level; and (v) the stability to endure 10 8 turnovers (20 years of exposure).Generally, dyes fall into two categories: metal-complexbased polypyridyl dyes and metal-free organic dyes.While metal-complex-mostly ruthenium-dyes show high efficiencies due to high absorbance ranges extending to the near-infrared range, the metal-free organic dyes appeal to the market with their low processing cost and limited Ru requirements.

Metal-Complex-Based Dyes.
Metal-complex dye sensitizers generally consist of a central metal ion and a subsidiary ligand, typically a bipyridine or tetrapyridine that contains anchoring groups.The central metal ion performs the light absorbance.Subsidiary ligands can be structurally modified to fine-tune the photovoltaic properties.
Ruthenium complexes, as shown in Figure 3, have demonstrated excellent photovoltaic properties, such as broad absorbance profiles and well-fitted energy levels [24].These factors, combined with electrochemical stability, make Ru complexes unrivaled for sensitizer uses [10,25,26].One representative Ru-based dye is the N3 dye, whose reported conversion efficiency is over 10%.
Another one is N749 dye, which is called the black dye due to its enhanced absorption spectrum.It shows an enhanced efficiency as high as 10.4%.Recent research on this family of dyes includes the work conducted by Robson et al., in which the NCS ligands of the black dye were replaced with anionic tridentate chelating ligands to adjust the HOMO and LUMO of the dye sensitizers to increase the photocurrent without compromising other DSSC properties [27].When a triphenylamine group was substituted for the NCS group and a phenyl-bipyridine-OMe group was used as the chelating ligand, an 8.02% PCE was measured.
There have been attempts to use metals other than Ru as the dye sensitizer.Such metals include Os [28], Re [29], Fe [30][31][32], Pt [33][34][35][36], and Cu [37,38].Os(II) complexes have better absorption spectra compared to Ru complexes.However, Alebbi et al. reported a lower energy conversion efficiency, attributing it to the slower electron transfer from the iodine electrolyte to the osmium complex [39].Pt(II) diimine-dithiolate complexes showed a high IPCE of 47% at 500 nm [33] but had a poor light absorption capacity at longer wavelengths.The tenability of photophysical properties, due to possible structural modification of the substituents on the diimine and dithiolate, was notable.Cu(I) complexes have very similar photophysical properties to Ru complexes.According to Constable et al., the - characteristic of these complexes was sensitive to the location of the conjugated pibond.
Polyphyrins are also of interest due to their good absorption and stability.With zinc-polyphyrin dye and a Co(III) tris(bipyridyl)-based redox electrolyte, Yella et al. reported a breakthrough efficiency of 12.3% and attributed this high efficiency to the suppression of electron backtransfer by the dye structure [40].are relatively low.Moreover, there are concerns about the cost effectiveness of ruthenium and the environmental impacts.Organic dyes also enable easy synthesis and design of the dye molecules.The predominant structure of metal-free organic dyes is the donor--acceptor (D--A) structure.It works as a basic template to design new dye molecules by adjusting the functional group or the whole moiety of the donor, pi-bridge, or acceptor of an existing dye.

Metal-Free Organic
Zhang et al. reported a high power conversion efficiency of 11.5-12.8%with cyclopentadithiophene-benzothiadiazole derivatives [41].This research group cografted the dyes having poor light absorption abilities, by which they successfully removed the penetration channels for Co(III) redox ions and thereby restricted the charge recombination at the titania surface.The dyes they used contain triphenylamine and benzoic acid moieties on each end of the cyclopentadithiophenebenzothiadiazole group, where different orders of the moieties generate two different dyes.The high PCE of 12.8% is even comparable to the Ru-based dyes, which proves that metal-free organic DSSCs can be as efficient as conventional DSSCs and sets a new standard for organic DSSCs for peer researchers.
Another approach taken by Qu et al. studied the diketopyrrolopyrrole (DPP) dyes and their photovoltaic characteristics as shown in Figure 4 [42].These dyes commonly had a triphenylamine donor, carboxylic acid acceptor, and DPP pi-bridge.The introduction of DPP moieties improved the spectral absorption at longer wavelengths.The PCE was 3.44%, relatively high, and was attributed by the authors to high dye aggregation and electron recombination.Adding 1.5 mM of chenodeoxycholic acid (CDCA) partly resolved the aggregation issue and raised the PCE to 4.14%, with  sc = 9.78 mA cm −2 ,  oc = 0.605 V, and a fill factor of 0.7 under AM1.5 solar light and with an iodine electrolyte.
Another approach to DSSC dyes involves extracting the natural dyes and structurally adjusting them for solar cell applications [49].None of them so far has led to a PCE significantly higher than those of other organic dye sensitizers.One reason for their low PCEs is thought to be the low interaction between the dyes and TiO 2 .High adsorption of dyes onto the TiO 2 structure is vital for high  oc , because electron recombination can occur.In fact, most of the natural dye-based DSSCs suffer from low  oc .Low stability of natural dyes is another issue.Therefore, improvements either in  oc or in stability are essential in order to justify further studies using this approach.

Metal Oxide
Metal oxide nanostructures have also been integrated into DSSCs to maximize the number of adsorbed dyes and thereby maximize the photocurrent.O'Regan and Grätzel used a mesoporous layer of TiO 2 to produce a 1000-fold increase in the surface area [7].TiO 2 NPs are generally preferred for DSSC applications due to their high stability.The group also studied various alternatives, such as ZnO, SnO 2 , or Nb 2 O 5 [50], which were chosen because their energy band structures are similar to that of TiO 2 .However, these alternatives rarely yielded PCEs comparable to those of TiO 2 -based cells [51].
4.1.TiO 2 Nanoparticle Films Based DSSCs.TiO 2 has three types of crystallinities: anatase, rutile, and brookite.A comparative study between anatase and rutile TiO 2 nanoparticle films was conducted by Park et al. [52].This study concluded that the anatase TiO 2 nanoparticles have better electron transport than do rutile TiO 2 nanoparticles in DSSCs.
While nanoparticle films have shown significantly higher PCEs compared to those of other nanostructures, due to their excessive number of adsorbed dyes, they have a clear disadvantage in charge transport.In nanoparticle films, the generated charge has to pass several nanoparticles before it reaches the substrate.Since the particles slightly differ in their energy states, the combined energy band might contain several trap states, and the charge may be trapped in such states, losing the energy and delaying the charge transfer process [53].Moreover, the grain boundary between nanoparticles is not effective for electron transport.
In TiO 2 nanotubes, the electron transport is better than TiO 2 nanoparticles due to the absence of a grain boundary.However, the small surface area for sensitizers is one of the main obstacles to achieving a high PCE in DSSCs.To retain the advantage in the surface area while improving the charge transport, numerous studies have been conducted to test DSSCs based on TiO 2 nanotubes.
Gong et al. reported on TiO 2 nanotube arrays in a 0.5 wt% HF aqueous solution at room temperature using different anodizing voltages [68].The film thickness could not be increased above 400-500 nm when an HF-based electrolyte was used.The fluoride solution dissolves TiO 2 by forming TiF 6 2− anions.The acidity of HF electrolytes is too high and results in a rapid dissolution of TiO 2 .To overcome this problem, Mor et al. reported that the addition of acetic acid to a 0.5 wt% HF electrolyte in a 1 : 7 ratio resulted in more mechanically robust TiO 2 nanotube arrays without changing their shape or size [57,69].The surface morphology of TiO 2 nanotube arrays anodized in an electrolyte containing 2.5% HNO 3 and 1% HF at 20 V for 4 hours exhibited a uniform, clean, and regular TiO 2 nanotube array structure with a length of about 400 nm as reported by Ruan.The anodization with an electrolyte composed of 0.5 M H 3 BO 3 , 2.5% HNO 3 , and 1% HF at 20 V for 4 h led to irregular pore of TiO 2 nanotube arrays but a long length of about 560 nm [70].To increase the length of the TiO 2 nanotube arrays, Ti was anodized with KF or NaF in the electrolyte [71].The acidity of the electrolyte could be controlled by adding HF, H 2 SO 4 , or Na 2 SO 4 to adjust the balance of the dissociation of TiO 2 at the electrolyte/oxide interface and the oxidation of TiO 2 at the oxide/metal interface [72,73].
The key to constructing longer TiO 2 nanotube arrays is to reduce the water content in the anodization bath to less than 5%.In organic electrolytes, a little water content reduces the dissociation of the oxide in the fluorine-containing electrolytes.
The chemical reaction of titanium anodization should be the same as that of aluminum anodization [44,74,77,78].The key processes are the growth of an anodic TiO 2 when the surface of Ti is oxidized with oxide anion and the dissociation of oxide anions through dissociation and fieldassisted dissolution in water, as shown in Figure 5 [43].At the electrolyte/oxide interface, the reaction is written as The cations move to the cathode and undergo the reaction: The oxide ions, OH − or O 2− , migrate through the oxide layer to form TiO 2 : where  is yet to be determined.Titanium hydroxide decomposes to form TiO 2 at the oxide/hydroxide interface.F − in the electrolyte moves to the anode and undergoes the reaction: The overall reaction of TiO 2 nanotube arrays is where  was introduced to indicate the ratio of the dissociation rate of water to the dissolution rate of TiO 2 .
Pores develop from pits on the Ti plate surface, and a schematic diagram for the equifield strength model is shown in Figure 6.
The initiation and growth of pores are associated with accelerated dissolution of TiO 2 due to the influence of an electric field [79].Many defects, such as impurities, dislocation, grain boundaries, or nonmetallic inclusions, could cause a faster dissolution rate and lead to pit growth [80,81].When Ti 4+ cations escape from the oxide surface because of an applied field, Ti 4+ cation vacancies can appear and accumulate to form high-density voids in the oxide layer, which can aid in the propagation of pits [82,83].
e −   This investigation was the first attempt to construct a DSSC of TiO 2 nanotube arrays.However, it used backside illumination.Therefore, incident sunlight had to pass a Pt-coated counter electrode and electrolyte that absorbed light near 400 nm.To increase the energy conversion efficiency using TiO 2 nanotube arrays in DSSCs, the structure should be designed with front illumination.Several approaches using TiO 2 nanotube arrays in DSSC designs have incorporated a thin film consisting of TiO 2 nanotube arrays with the photoanode of the DSSCs, improving the scattering-driven light harvesting and suppressing the charge recombination.Park et al. reported a simple method for preparing TiO 2 nanotube arrays on the FTO glass [16].In this method, a thin film of TiO 2 was anodized with NH 4 F to form an array of TiO 2 nanotubes.The nanotube array was then detached from the metallic Ti substrate by the use of an HCl solution and was annealed onto a sheet of FTO glass at 500 ∘ C for 30 min.The photoanode fabricated with this method was equipped with a Pt counter electrode, Ru(II)LL  (NCS) 2 dye (L = 2,2-bipyridyl-4,4-dicarboxylic acid; L  = 2,2-bipyridyl-4,4-ditetrabutylammonium carboxylate), and an iodine electrolyte to yield 7.6% power conversion efficiency under 1 sun illumination.
Chen and Xu produced a crystalline TiO 2 nanotube array thin film [17].The crystalline TiO 2 nanotube arrays were usually obtained by annealing the array at a high temperature, and the high-temperature annealing caused the TiO 2 nanotube arrays to curl.The researchers circumvented this problem by introducing an additional step of anodization after annealing, which results in a double-layer structure of TiO 2 nanotube arrays, in which the lower layer is later dissolved in an H 2 O 2 solution.The resulting free-standing TiO 2 nanotube arrays did not need any support to maintain their crystal structure, enabling them to be attached to FTO glass and to fabricate a front-illuminated DSSC.The front illumination of the DSSC enhanced the irradiation intensity by 11% compared to the conventional back-illuminated method.Additionally, the crystalline free-standing TiO 2 nanotube arrays provided a longer electron lifetime, which caused the PCE to increase from 2.8% with a back-illuminated DSSC equipped with thick TiO 2 nanotube arrays on a Ti substrate to 5.5% with a frontilluminated, crystal-structured, TiO 2 nanotube array-based DSSC.
Lin et al. fabricated open-ended TiO 2 nanotube arrays, which differed from the previously mentioned nanotube structures in that there was no blocking layer of TiO 2 at the junction point between the FTO glass and the TiO 2 nanotube arrays [18] the enhanced interaction between the electrolyte and NP-TiO 2 underlayer due to the absence of the bottom caps.The open-ended nanotube array showed a strong enhancement in capturing incident photons with longer wavelengths, which also indicated that opening the tube bottom enabled the DSSC device to collect the excited electrons over the entire length of the nanotubes.Li et al. reported on the fabrication of crystalline TiO 2 nanotube arrays using the secondary anodization step in a similar manner to Chen et al. [17,19].With these freestanding crystalline TiO 2 nanotube arrays, they used a dry-etching method with a plasma reactor under BCl 3 /Cl 2 to make the nanotube bottom open-ended and compared this array to a close-ended nanotube array fabricated with the same methodology, but without the etching step.The resulting PCE was higher for the open-ended nanotube arrays (6.24%) than it was for the close-ended nanotube arrays (4.84%).The authors explained this difference as the result of the interface between nanotube arrays and FTO being more compactly adhered, leading to improved electron diffusion and charge collection.This conclusion is similar to that of Lin et al.
The open-ended TiO 2 nanotube arrays generally show a better photoelectric performance than do the close-ended nanotube arrays.Rho et al. studied the effect of the thickness of the bottom cap (barrier layer) of the nanotube arrays, as shown in Figure 7 [45].The authors fabricated the TiO 2 nanotube arrays with two anodization steps, followed by ion milling with argon ions.The thickness of the barrier layer was controlled by varying the time for the ion-milling step.The resulting PCEs varied from 2.5% to 3.7%.As the barrier layer was removed, the PCE increased.When the DSSC  was designed with back-side illumination, the efficiency was lower than when it was designed with front-side illumination.In back-side illumination, the existence of the barrier layer does not affect the incident light intensity, as the light reaches the barrier layer after it is absorbed by the dye molecules.Rather, in this case, the more decisive factor influencing the conversion efficiency was the efficiency of electron diffusion, as the barrier layer may significantly hinder the electron transport.
The energy conversion efficiency of DSSCs based on open-end TiO 2 nanotube arrays was much higher than the efficiency based on closed-end ones.This result means that the barrier layer, corresponding to the under layer of the detached TiO 2 nanotube arrays, has a marked effect on the conversion efficiency, since the barrier layer has been removed in open-end TiO 2 nanotube arrays but left intact in closed-end TiO 2 nanotube arrays.The barrier layer could affect the conversion efficiency by preventing the diffusion of materials like electrolytes and dye molecules, by reducing the transmittance of light, and by hindering the electron transport.The electrons generated by excitation of the dyes adsorbed on the TiO 2 nanotube arrays could transfer to the electrode through the TiO 2 nanotube arrays.In this case, they must pass the barrier layer to reach the electrode.(DSSCs with free-standing TiO 2 nanotube array were summarized in Table 1.)

Development of TiO 2 Nanotube Arrays for DSSC.
The various approaches mentioned in the earlier sections exploited TiO 2 nanotube arrays to achieve DSSC systems with higher efficiencies.Generally, the techniques to increase the efficiencies of TiO 2 nanoparticle systems can be also applied to TiO 2 nanotube systems.For example, combining plasmonic enhancement and/or TiCl 4 could benefit TiO 2 nanotube arrays, which have large spaces in them.
(1) Plasmonic TiO 2 Nanotube Arrays for DSSCs.Plasmonic solar cells are photovoltaic devices that use noble metal surface plasmons.Most solar cells have weak absorbers.To absorb larger amounts and longer wavelengths of incident light, the morphology of the substrate is etched with pyramids with sizes of 2-10 m to obtain a wavelength-scale texture.This structure improves the light trapping for solar cells but increases the surface recombination and reduces the material quality [84,85].
There is another approach involving the use of noble metal nanoparticles in solar cells.The light can be absorbed and scattered from the noble metal nanoparticles that are excited at the surface plasmon resonance.
The mechanism of plasmonic solar cells can be used to explain the photocurrent enhancement by metal nanoparticles incorporated into or on solar cells, as shown in Figure 8 [46].One is scattering from the metal nanoparticles, which can be used as subwavelength scattering elements to couple and trap plane waves propagating from the Sun into the semiconductor thin film, by folding the light into a thin absorber layer as shown in Figure 8(a).
Light is preferentially scattered and trapped in the semiconductor thin film by multiple high-angle scatterings,  effectively increasing the optical path length in the cell.In inorganic plasmonic solar cells, the photocurrent enhancement is increased by scattering from metal nanoparticles.Another approach makes use of the near-field enhancement from metal nanoparticles that can be used as subwavelength antennas in which the plasmonic near-field is coupled to the semiconductor, increasing its effective absorption crosssection, as shown in Figure 8(b).In organic plasmonic solar cells, the photocurrent enhancement is increased by near-field enhancement.The direct generation of charge carriers in the semiconductor substrate has also been studied.A corrugated metallic film on the back surface of a thin photovoltaic absorber layer can couple sunlight into surface plasmon polariton (SPP) modes supported at the metal/semiconductor interface as well as guided modes in the semiconductor slab, whereupon the light is converted to photocarriers in the semiconductor as shown in Figure 8(c) [86][87][88].
Stuart and Hall reported an 18-fold photocurrent enhancement for light-sensitive devices with a 165-nm-thick siliconon-insulator photodetector at a wavelength of 800 nm by using silver nanoparticles on the surface of the device [89].The gold nanoparticles used for scattering and absorption of light on highly doped wafer-based solar cells were reported on by Schaadt et al. [90].Gold nanoparticles on thin-film amorphous silicon solar cells gained an 8% overall increase in conversion efficiency as reported by Derkacs et al. [91].Silver nanoparticles on 1.25-m-thick silicon-on-insulator solar cells and planar wafer-based cells obtained 33% and 19% photocurrent increases, respectively, as reported by Pillai et al. [92].Enhancement in the photocurrent by up to a factor of 2.7 for ITO-copper phthalocyanine-indium structures was reported by Stenzel.Enhancement for silver clusters incorporated in ITO and zinc phthalocyanine solar cells was reported by Westphalen et al. [93].Enhanced efficiencies for ultrathin film organic solar cells due to 5 nm diameter silver nanoparticles were reported by Rand.An increase in efficiency by a factor of 1.7 for organic bulk heterojunction solar cells was reported by Morfa et al. [94].Increased photocurrents for CdSe/Si heterostructures [95] and enhanced carrier generation in dye-sensitized solar cells have also been reported [96].
Rho et al. attached TiO 2 nanotube arrays, which were prepared by the electrochemical anodization method, onto a TCO glass to fabricate a transparent photoanode and a frontilluminated DSSC as shown in Figure 9 [47].With this cell design, the N719 dyes were sensitized more directly by the incident light compared to the conventional back-illuminated DSSCs in which light has to pass through the transparent counter electrode in order to reach the dyes.Ag nanoparticles with diameters smaller than the light wavelength can improve the light absorption dramatically due to the surface plasmon resonance effect.The authors varied the time of exposure to irradiation to examine the Ag nanoparticle deposition kinetics and the existence of the optimal irradiation.The Ag nanoparticles were deposited not only on the outside of the TiO 2 nanotube arrays but also on the inside.The photovoltaic properties measured for DSSCs without Ag nanoparticles and DSSCs with nanoparticles and varying irradiation times indicated the same tendency.The highest energy conversion efficiency achieved was 6.14%, around 32% improvement of the energy conversion efficiency for DSSCs without Ag nanoparticles, 4.64%.Comparison was also made between front-illuminated and back-illuminated DSSCs.The highest efficiency reported for the back-illuminated DSSCs was 3.077%, which was lower than the front-illuminated DSSCs' highest efficiency.Notably, the effect of incorporating Ag nanoparticles was also lower for back-illuminated DSSCs.The back-illuminated DSSCs showed only a 9.46% increase in the efficiency after the deposition of Ag nanoparticles, which was lower than the 32% increase in the front-illuminated DSSCs.The authors attributed this difference to the reduced intensity of the light passing through the counter electrode and electrolyte layer in the back-illuminated DSSCs, which caused the surface plasmon radiation to occur only at the top part of the TiO 2 nanotube arrays.
(2) TiCl 4 with TiO 2 Nanotube Arrays for DSSC Solar Cells.TiO 2 nanoparticles have been deposited to fill the microcracks formed in TiO 2 nanotube arrays using an electrophoretic method [97].By depositing nanoparticles, the conversion efficiency was improved from 4% to 5.2%.Nanoparticles were found in the cracks and on the surface of the nanotubes.The total inner space of the nanotubes might be much larger than the space of the microcracks.However, nanoparticles might not be deposited deep inside the pores.Since TiO 2 nanotube arrays are semiconductors, electrochemical depositions take place at or near the entrances of their pores.There is no driving force to push the deposited nanoparticles into the deep parts of the pores.In vacuum filtering, the solvent provides a driving force, and nanoparticles can be filled throughout the long channels of the TiO 2 nanotube arrays.The TiO 2 nanoparticles that fill the channels of TiO 2 nanotube arrays can make contact with the tubes directly or through other nanoparticles.Since nanotubes are more efficient in electron transfer than are nanoparticles or composites of nanoparticles and nanotubes, electron transfer to the electrode could also be more efficient.
DSSCs based on TiO 2 nanotube arrays filled with TiO 2 nanoparticles have also been fabricated as shown in Figure 10 [48].The energy conversion efficiency was enhanced by about 21% by filling the channels of the membrane with TiO 2 nanoparticles.By treating with TiCl 4 after filling with nanoparticles, the efficiency was enhanced significantly by about 40% due to an increase in the total available surface area for dye adsorption and better electron transfer to the electrode.The method of filling the channels of the TiO 2 membrane with TiO 2 nanoparticles and then treating with TiCl 4 was used to improve the energy conversion efficiency of DSSCs based on TiO 2 nanotube arrays.
(3) Additional Layer Technique.Introducing a scattering layer such as TiO 2 , ZrO 2 , or SiO 2 can increase the total energy conversion efficiency of DSSCs [98].TiO 2 is a good material to use for a scattering layer due to its chemical stability and

Conclusions and Future Work
In the future, photovoltaic cells will be used in many fields, such as mobile commerce, building integrated photovoltaics (BIPVs), and vehicles.Moreover, photovoltaic cells are essential in the Smart Grid, which is utilized in our daily lives.To apply solar energy in the Smart Grid, photovoltaic cells are required to have transparency, flexibility, light weight, low cost, and high energy conversion efficiency.In terms of low cost and light weight, organics, inorganics, and hybrid materials have brighter prospects than the semiconductors.In hybrid materials, photovoltaic cells have been prepared with the advantages of organics or inorganics selectively.However, it is not easy to increase the energy conversion efficiency.One potential solution is to use 3-dimensional nanostructures, such as nanotubes, nanowires, composite films with metal, or other types of nanostructures.As of today, 0-dimensional nanoparticles in photovoltaic cells still yield the higher energy conversion efficiency by supplying larger surface areas for the sensitizer adsorption.However, the electron transport and flexibility are sufficiently high in 3-dimensional nanostructures.To improve the energy conversion efficiency of 3-dimensional nanostructure based solar cells and take both advantages of 0-dimensional nanoparticles and 3-dimensional nanostructures, the light harvesting ability must be reinforced.One way of achieving the better light harvesting ability is to fill the 0-dimensional nanoparticles in the cavities of 3-dimensional nanostructure, which is another currently active field of study [99].

Figure 6 :
Figure 6: Schematic diagram of the evolution of a nanotube array at constant anodization voltage: (a) oxide layer formation, (b) pit formation on the oxide layer, (c) growth of the pits into scallop-shaped pores, (d) metallic part between the pores undergoing oxidation and field-assisted dissolution, and (e) fully developed nanotube array with a corresponding top view.Reprinted with permission from [44].Copyright 2006, Elsevier.

Figure 8 :
Figure 8: Plasmonic light-trapping geometries for thin-film solar cells.(a) Light trapping by scattering from metal nanoparticles at the surface of the solar cell.(b) Light trapping by the excitation of localized surface plasmons in metal nanoparticles embedded in the semiconductor.(c) Light trapping by the excitation of surface plasmon polaritons at the metal/semiconductor interface.Reprinted with permission from [46].Copyright 2010, Nature Publishing Group.

Figure 9 :
Figure 9: (a) Schematic of plasmon DSSC based on method reported by Rho et al. (A) Ti anodization in organic electrolyte, (B) freestanding TiO 2 nanotube arrays, (C) formation of Ag nanoparticles on TiO 2 nanotube arrays by UV irradiation at 254 nm.(b) Schematic of an assembled cell.Reprinted with permission from [47].Copyright 2014, Elsevier.

Table 1 :
Summary of DSSCs with free-standing TiO 2 nanotube arrays.