Abstract

Since O’Regan and Grätzel’s first report in 1991, dye-sensitized solar cells (DSSCs) appeared immediately as a promising low-cost photovoltaic technology. In fact, though being far less efficient than conventional silicon-based photovoltaics (being the maximum, lab scale prototype reported efficiency around 13%), the simple design of the device and the absence of the strict and expensive manufacturing processes needed for conventional photovoltaics make them attractive in small-power applications especially in low-light conditions, where they outperform their silicon counterparts. Nanomaterials are at the very heart of DSSC, as the success of its design is due to the use of nanostructures at both the anode and the cathode. In this review, we present the state of the art for both -type and -type semiconductors used in the photoelectrodes of DSSCs, showing the evolution of the materials during the 25 years of history of this kind of devices. In the case of -type semiconductors, also some other energy conversion applications are touched upon.

1. Part I: -Type Semiconductors

1.1. Introduction

A dye-sensitized solar cell (DSSC) is a semiconductor-based photovoltaic device that directly converts both artificial and natural (solar) radiation into electric current. In contrast to the conventional systems, where the semiconductor assumes both the tasks of light absorption and charge carrier separation and transport, the two functions are separated in a DSSC. In a conventional DSSC, light is absorbed by a sensitizer, which is anchored to the surface of a wide band gap -type semiconductor [1]. Often referred to as a Grätzel cell, this hybrid device was first reported to the scientific world in 1991 by the seminal publication of Brian O’Regan and Michael Grätzel presenting a device made of sensitized nanocrystalline TiO2 with a power conversion efficiency of 7.1% [2]. DSSCs constitute probably the cheapest photovoltaic technology today available and their efficiency has been constantly improved in the last 25 years. The confirmed efficiency record is 11.9% [3], achieved by the Sharp company. However, it was Mathew et al. that reported the best performing DSSC in literature up to now. DSSCs fabricated utilizing [Co(bpy)3]2+/3+ as redox couple and the SM315 dye demonstrated panchromatic light-harvesting without the use of cosensitization, leading to a record efficiency value of 13% under one-sun illumination [4]. DSSC is an ecofriendly technology that can be used to produce electricity in a wide range of light conditions, indoors and outdoors. If one considers that dye-sensitized solar cells work properly also under low irradiation conditions, the efficiencies they reach make the technology extremely interesting with respect to conventional solar cells in such conditions. The DSSCs and their inventor received prestigious awards, including the Balzan Prize in 2009 and the 2010 Millennium Technology Prize, the most important technology prize in the world [5].

1.2. Device Structure and Operation Principles: Brief Discussion of Components and Their Role in DSSC

A simple scheme of the components and of the basic operating prinles of a DSSC is reported in Figure 1. The device is comprised of four components only:(i)Nanostructured -type semiconductor (wide band gap metal oxide) coated over transparent conducting substrate (TCO (ITO [6] or FTO [7]))(ii)Visible-light absorber dye (several organic dye can be used, such as N3 [8], N719 [9], N749 [10] (the so-called black dye), K8 [11], K19 [12], CYC-B11 [13], and C101 [2])(iii)Electrolyte [14, 15](iv)Counterelectrode (TCO coated with a platinum layer or other suitable catalyst)In the case of -type materials, current is generated when a photon absorbed by a dye molecule causes an electron injection into the conduction band of the semiconductor. The dye molecules are anchored to the surface of the semiconductor particles by a chemical bond (sensitization). In a nanostructured oxide, the semiconductor provides a large surface area for dye molecule anchoring. The photons are absorbed at the surface of the nanostructure by the dye, promoting its excitation and then the electron injection. The photogenerated electrons are injected into metal oxide and the dye is regenerated by a redox species present in the electrolyte. The electrons move through the semiconductor to a current collector and then to an external circuit. The redox mediator is regenerated at the cathode, and the process is cyclic. Each part of the device heavily determines the cost and the efficiency of DSSCs. The optimization of light absorption properties can be achieved by modifying the dye alone, while charge transport properties can be improved by the optimization of the semiconductor and of the electrolyte composition [16]. Thus, in the last years, almost all research efforts have been focused on the modification of each component for practical applications. Ye et al. [17] reported a scheme of the number of publications for sensitized solar cells. The major area of interest was represented by the fabrication of -type nanostructured semiconductor photoanodes with improved architectures for high dye loading and fast electron transport. Therefore, this section of the present paper aims at bringing together the various contributions brought about by researchers to improve the efficiency of the -type sensitized semiconductors for DSSCs. Moreover, the detailed discussion of the role of band gap, morphology, composition, and doping of -type semiconductors for the development of efficient photoanodes will be reported. Several contributions are summarized here with the description of nanostructures (0D, 1D, 2D, 3D, and mesoporous nanostructures) and of the effect of doping of photoanodes materials (mostly TiO2 and ZnO) on the performances of DSSCs.

1.3. Photoanode Materials for -Type DSSC

Semiconducting nanostructured films are the main core of DSSCs photoanodes. The main issues in DSSCs are the charge recombination processes and their relatively low light-harvesting capability. The photoanode performs a dual function as both the support for the sensitizer and carrier of photogenerated electrons from the sensitizer to the external circuit [17]. The photoanode material should not absorb visible light and should have sufficiently high surface area for optimum dye adsorption.

In semiconductor terminology, the top of the valence band and the bottom of the conduction band are called the valence band edge and the conduction band edge, respectively. The energy of the conduction band edge is denoted as ; similarly, the energy of the valence band edge is denoted as . The energy difference between these levels is called the band gap (). The size of this band gap is perhaps the most important property of a semiconductor, as it influences all of the most important electronic properties of the material [18]. The conduction band edge of the photoanode material should match with that of the excited dye molecules. To efficiently collect the photogenerated electrons, the photoanode should have high charge carrier mobility.

Desirable properties of the material are also the ease to preparation, stability, low-cost, and environmental friendliness. These properties are the defining characteristics of an ideal photoanode [19].

1.3.1. -Type Semiconductors

Probably it was Faraday [20] that made the first significant observation in the semiconductor science in 1833 when he discovered the negative temperature resistivity coefficient of silver sulfide [21]. Nowadays, semiconductors are defined by the unique behaviour of their electrical properties. As their name implies, semiconductors possess an electrical conductivity intermediate between conductors such as metals and insulators such as ceramics [22]. For high-density electron ensembles such a valence electrons in metals, Fermi statistic is applicable. The Fermi level, (defined 0 K as the energy at which the probability of finding an electron is ) can be regarded as the electrochemical potential of the electron in a particular phase (in this case, a solid). Thus, all electronic energy levels below are occupied and those above are likely to be empty. Electrons in semiconductor may be regarded as low-density particles ensembles such that their occupancy in the valence () and conduction bands () may be approximated by the Boltzmann function [23, 24]:Now, we come to another important distinction between metals and semiconductors in that two types of electronic carriers () are possible in the latter. Consider the thermal excitation of an electron from to . This gives rise to a free electron in the and a vacancy or hole in the . Thus, (1) becomeswhere and are the effective density of states (in cm−3) at the lower edge and top edge of and , respectively. These expressions can be combined with the recognition that to yieldTo provide a numerical sense of the situation, and are typically both approximately so that the constant in (1) is about .

The latter case refers to the semiconductor in its intrinsic state with very low carrier concentrations under ambient conditions. The Fermi level, , in this case lies approximately in the middle of the energy band gap [25]. This simply reflects the fact that the probability of the electron occupancy is very high in and very low in and does not imply an energy level at itself that can be occupied.

In extrinsic semiconductors, the carrier concentrations are perturbed such that . The analogy with the addition of an acid or base to water is quite instructive here. Consider the case when donor impurities are added to an intrinsic semiconductor. Since the intrinsic carrier concentrations are low (subparts per trillion), even additions in part per billion levels can have a profound electrical effect. This process is known as doping of the semiconductor. In this particular case, the Fermi level shifts towards the one of the bands’ edges. When the donor level is within a few in energy from the CB edge, appreciable electron concentrations are generated by donor ionization process (at ambient temperatures) such that now . This is termed doping, and the resultant (extrinsic) semiconductor is called By analogy, semiconductors have . The terms minority and majority carriers now become appropriate in these cases. For a semiconductor case, the Fermi level now lies close to edge. The Fermi level determines the tendency of the semiconductor to transfer charges to other phases. The movement of with dopant concentration can also be rationalized via the Nernst formalism [26].

Doping can be accomplished by adding aliovalent impurities to the intrinsic semiconductor. For example, P (a group 15 or VB element) will act as a donor in Si (a group 14 or IVB element). This can be rationalized on chemical terms by noting that P need only four valence electrons for tetrahedral bonding (as in Si lattice) and the fifth electron is available for donation by each P atom. The donor density, nominally is approximately . Thus, assuming that (complete ionization at 300 K), will be only approximately (recall the product ), bearing out the earlier qualitative assertion that .

Impurity addition, however, is not the only doping mechanism. Nonstoichiometry in compound semiconductors such as CdTe also gives rise to - or -type behaviour, depending on whether Cd or Te is in slight excess, respectively. For metal oxides, doping with suitable cation/anion modifies band gap and may modulate their electrical properties. The defect chemistry in these solid chalcogenides controls their conductivity and doping in a complex manner. Excellent treatises are available on this topic and on the solid-state chemistry of semiconductors in general [27]. The distinction between metal and semiconductor electrodes is important when we consider the electrostatics across the corresponding solid-liquid interfaces [metal oxide semiconductor (MOS)/electrolyte].

1.3.2. -Type Semiconductor-Electrolyte Interface

The electrochemical potential of electrons in a redox electrolyte is given by the Nernst expression:where and are the concentrations of the oxidized and reduced species, respectively, in the redox couple. The parameter ( as defined by this equation can be identified with the Fermi level ( in the electrolyte. When a semiconductor is immersed in this redox electrolyte, the electrochemical potential (Fermi level) is disparate across the interface. Equilibration of this interface thus necessitates the flow of charge from one phase to the other and a “band bending” ensues within the semiconductor phase. For an -type semiconductor (TiO2) electrode at open circuit, the Fermi level is typically higher than the redox potential of the electrolyte, and hence electrons will be transferred from the electrode into the solution. Therefore, there is a positive charge region, and this is reflected in an upward bending of the band edges [28] (see Figure 2). Since the majority charge carrier of the semiconductor has been removed from this region, the latter is also referred to as the depletion layer.

1.3.3. -Type Nanostructured Semiconductor for DSSCs

Nanostructured materials present a specific surface area significantly larger than that of the corresponding bulk materials. The nanoscale size may also affect the behaviour of electrons transport in nanostructures in view of a limit to the electron mean free path. This is called quantum confinement effect [2931]. The quantum confinement effect is observed when the size of the particle is too small to be comparable to the wavelength of the electron. On the optics side, by forming photonic band gap, periodic nanostructures (known as a photonic crystals) also are shown to be special in light manipulation and management through generating optical confinement or photonic localization [3235]. These unique properties of nanomaterials received considerable attention and have been extensively investigated for applications in electronic, optoelectronic, photovoltaic, photocatalytic, and sensing devices [3640]. The metal oxides in their nanoform can be synthesized under various morphologies with different shapes and sizes, thus offering the possibility for modulating their properties.

The crystallinity of the metal oxides should be high to prevent the recombination of and . Ohtani et al. clarified that the photocatalytic activity of TiO2 powders strongly depends on its physical properties such as crystal structure, surface area, particle size, and surface hydroxyls [41]. In literature, a variety of preparation techniques, such as sol-gel [42], hydrothermal/solvothermal [43, 44], electrochemical anodization [45], electrospinning [46, 47], spray pyrolysis [48], and atomic layer deposition [49], have been developed and applied to obtain different morphologies in photoanode materials. Asim et al. summarized the different preparation and deposition methods, which have been used for photoanode materials of DSSCs, emphasizing their advantages and disadvantages, in order to allow a researcher to carefully choose and optimize a given method [50]. Perez-Page et al. [51] presented several methods based on templates for shape-controlled nanostructuring. A variety of nanostructured materials are produced through template-based syntheses, including zero-dimensional (nanoparticles), one-dimensional (nanowires, nanotubes, etc.), and two dimensional (nanoflakes and nanosheets) structures [51]. In 1995, Ying and Antonelli for the first time reported the synthesis of mesoporous TiO2, which was accomplished through a modified sol-gel process involving TIP (titanium isopropoxide) as a precursor [52]. Later, various procedures have been developed for the synthesis of mesoporous materials. In 2001, Grosso et al. [53], Yun et al. [54], and Hwang et al. [55] reported syntheses of mesoporous titania films in the anatase phase. In the development of a simple and general methodology for the synthesis of highly organized mesoporous metal oxide, Sadatlu and Mozaffar proposed a novel and effective strategy [56].

1.3.4. Nanoparticles-Based Semiconductors (0D Nanostructures)

A typical DSSC photoanode is composed of nanocrystalline semiconductors. Sufficient light absorption is achieved by the nanocrystalline form of the semiconductor, because a large internal surface area increases the dye concentration in the film per unit device area. Wide band gap MOSs ( > 3 eV) [57], such as TiO2 [5860], ZnO [6167], SnO2 [60, 6871], and Nb2O5 [7274], have been studied more or less extensively and used as photoanode materials for DSSC devices. These MOSs present good stability against photocorrosion, transparency in the major part of the solar spectrum, and good electronic properties [7578]. Photocorrosion, which is caused by the oxidation by holes (generated through band gap excitation) of the redox species in the electrolytes, may affect the performance of the semiconductor.

TiO2 is a low-cost, widely available, nontoxic, and biocompatible material. It has been used in health care products as well as in domestic applications such as paint pigments [16]. On the other hand, ZnO, which has a similar conduction band edge and work function compared to TiO2, but with a higher carrier mobility than TiO2, was considered as a promising photoanode materials for DSSCs. However, the instability of ZnO in acidic environment and formation of dye aggregates on its surface deteriorate its performances [79]. This is probably due to its acidic surface and more positive (versus NHE) conduction band edge position compared to TiO2. Some other semiconductor materials, such as Zn2SnO4 [80], CeO2 [81], WO3 [82], SrTiO3 [83], In2O3 [84], and Al64Cu25Fe11 [85] were also studied. Alami et al. evaluated the suitability of the Al64Cu25Fe11 intermetallic compound to enhance spectral solar absorption when replacing the mesoporous layer material of DSSCs. Zheng et al. [86] presented for the first time a complete study on WO3 as photoanode material for DSSC.

The efficiency of DSSCs using these materials with a suitable redox electrolyte should be around 10–15% under one-sun irradiation. The main issues in DSSCs are charge recombination processes and their relatively low light-harvesting efficiency. As stated before, the photoanode performs a dual function as both the support for the sensitizer and carrier of photogenerated electrons from the sensitizer to external circuit [17]. In this regard, a major benefit of using nanostructured materials in DSSCs is that the resulting photoelectrode films are highly porous. Porous nanostructures offer a larger surface area for dye absorption compared to a bulk material. Among all semiconductor oxides, porous, anatase phase titanium dioxide (TiO2) is the best and most common choice for high-efficiency DSSCs due to its low-cost, easy synthesis, nontoxicity, and biocompatibility [87]. A photoanode based on nanosized TiO2 crystals with various geometries, such as nanoparticles [88], ordered mesostructured materials [89, 90], and one-dimensional structured materials, for example, nanorods, nanowires, and nanotubes [91, 92], has been extensively studied. Up to now, the most efficient DSSCs are based on TiO2 nanoparticles owing to their integrated advantage in porosity, dye absorption, charge transfer, and electron transport. Moreover, electron transport in nanoparticles films suffer from trapping/detrapping processes [93]. Trapping is an important effect affecting the electron lifetime in the semiconductor, significantly reducing the electron diffusion length, . In order to have a quantitative collection of electrons at the anode, condition should be satisfied, with being the semiconductor film thickness. Quaranta et al. [94] showed that can be increased by a given quantity of functionalized multiwalled carbon nanotubes (MWCNTs) coated by nanostructured TiO2-anatase. Moreover, other 0D structured photoanodes were used to prevent this drawback of nanoparticles. Chen et al. [95] prepared a core-shell structure that consisted of a nanoporous TiO2 film coated with oxides such as Nb2O5, ZnO, SrTiO3, ZrO, Al2O3, and SnO2. Moreover, Palomares et al. explained the growth of conformal metal oxide insulating overlayers (SiO2, Al2O3, and ZrO2 overlayers) on preformed nanocrystalline TiO2 films, focusing on the ability of such insulating layers to retard interfacial recombination dynamics and thereby modulate the performance of DSSCs fabricated using these films [96].

The best performance in DSSCs has been achieved with the anatase form of TiO2 [97, 98]. On the other hand, relatively little attention has been paid to the rutile form of TiO2. Park and Kim [99] studied a rutile TiO2-based DSSC [100]. Rutile is the thermodynamic stable polymorph of TiO2 that forms when less stable polymorphs are heated at temperatures over 700°C. Therefore, the synthesis temperature of rutile TiO2 is a potential issue for use in DSSCs since nanocrystalline morphology is hardly preserved at high temperatures. Moreover, the anatase form is preferred because of its higher band gap (3.2 eV) compared to rutile (~3 eV). High band gap energy makes anatase chemically more stable in DSSCs [101].

The bandgap values and the relative positions of the band edges of several commonly used semiconductors are shown in Figure 3 [102]. The positions of the redox levels of some of the most important couples are also shown.

The solid-state physics community has adopted the electron energy in vacuum as a reference, whereas chemists have traditionally used the normal hydrogen electrode (NHE) as energy reference. NHE lies at −4.5 eV with respect to the vacuum level. We can now relate the redox potential (as defined with reference to NHE) with the Fermi level : Parameters related to the band gap position are the flat-band voltage () and flat-band capacitance (). Material properties such as oxide charge, semiconductor carrier density, and the effective work function of the metal can be obtained directly from and [103, 104].

The standard method for determining the flat-band parameters is through Mott-Schottky plots.

Generally, a three-electrode single compartment configuration was adopted for capacitance analysis used to obtain Mott-Schottky plots. MOS coated on conducting glass was used as working electrode, while platinum and standard calomel electrode (SCE) were used as counter and reference electrodes, respectively. To measure the capacitance, oxide coated plates were immersed in 0.5 M Na2SO4 solution and the capacitances were measured as a function of the potential across the space charge layer at different frequencies [105]. Shifting of the flat-band potentials of semiconductors plays an important role in determining their photoelectrochemical properties as thin oxide films and powders. The shifting of   can be harnessed in solar energy conversion processes in several ways. In DSSCs, the open-circuit voltage () is determined by the difference between the quasi-Fermi level of electrons in the oxide film and the energy of the redox couple in the electrolyte [106]. Consequently, the shift of potential in the negative direction results in increase in barrier height and hence in the attainable of the solar cell. Similarly, the flat-band potential can be shifted negatively in order to electrolyze a redox couple that is otherwise too positive in the electrochemical series to be reactive at the illuminated semiconductor surface [107, 108]. Park and Kim also reported the relative flat-band potential and conduction band energies of some oxides such as TiO2, ZnO, SnO2, Nb2O5, FeTiO3, and Ta2O5 and sulfides that could be used for DSSCs when considering only conduction band energy positions (adapted from [99]) (Figure 4).

1.3.5. High Ordered Nanostructures (1D Nanostructures: Nanowires, Nanotubes, etc.)

A major feature that discriminates various types of nanostructures is their dimensionality. Low-dimensional nanostructures have an aligned structure that can act as a single crystal, thus facilitating the rapid electron transport in unidirectional manner, and show potentiality for obtaining high-performance devices [109]. Zhang and Cao [110] classify the nanostructures, presenting nanoparticles advantages (offer large surface area to photoelectrode film for dye absorption) and drawbacks (electron transport by trapping and detrapping processes, which may result in energy losses). In this regard, one-dimensional nanostructures such as nanowires and nanotubes provide direct pathways resulting in a much faster electron transport than in the nanoparticles films [110]. However, nanoparticles are advantageous in providing large surface area for dye absorption. Law et al. [111], for the first time in 2005, presented a comparison between a DSSC made with ZnO films made of nanoparticles and nanowires to demonstrate that one-dimensional nanostructures may provide direct pathways for electron transport in DSSCs. A high-performance nanowire photoanode must have a large surface area for dye adsorption, comparable to that of a nanoparticle film. At one-sun irradiation, the device presented an efficiency value within 1.2–1.5%. In the same year, Baxter and Adyl [112] first fabricated ZnO nanorod-based DSSC by using a hydrothermal method; they achieved an overall photoelectric conversion efficiency of 0.5%. Since then, 1D nanostructures of ZnO (nanorods and the like: Martinson et al., 2007 [113]; Schlur et al., 2013 [114]; Ameen et al., 2012 [115]; McCune et al., 2012 [116]; Guo et al., 2013 [117]) have been widely investigated as photoelectrodes for enhancing the DSSCs performances [118]. Other ZnO-based 1D structures have been investigated for photoanodes. Kim et al. studied the performance of ZnO nanofibers [119], Tan et al. constructed ZnO nanorods arrays on ITO glass [120], and Yang et al. prepared ZnO nanotips on rough Zn microtip foil [121].

ZnO-based DSSCs with delicately designed nanostructures, from irregular microrods to nanosheets with simply adjusting the dosage of capping agents in the hydrothermal synthesis process, were fabricated to study the photovoltaic performances versus morphology. In Figure 5, SEM micrographs of as-synthesized ZnO structures evolving from microrods to nanosheets are reported [122].

TiO2 nanowires in standard DSSC configuration gave an overall conversion efficiency of 5%, which is much higher than those reported for ZnO nanowires [66, 123].

In addition, the other TiO2 1D nanostructures gave higher efficiencies than ZnO-based ones. TiO2 nanorods, for example, were synthesized by Chen et al. via a microemulsion electrospinning technique. The authors observed a power conversion efficiency of 8.53%, which originated from the large structure of nanorods [124].

The most studied class of one-dimensional nanostructures is that of nanotubes. The recombination rate for TiO2 nanotubes is ~10 times lower than that of nanoparticles [125]. Their hollow structure usually gives a larger surface area than that nanowires or nanorods. Roy et al. presented a complete review of the current status of the use of TiO2 nanotubes in Grätzel cells [126]. Since the first results in 2005 with efficiencies of 0.04%, reproducible efficiency results in the order of 4-5% have been obtained. The best morphology reported is called bamboo-type rings. This may be achieved in various ways by modifying the tube walls with porous materials (double-walled tubes). During this last year, Momeni presented a study on Cr-doped TiO2 nanotubes [127]. The effect of chromium doping on the photovoltaic efficiency of dye-sensitized solar cells (DSSCs) with TiO2 nanotubes prepared by an anodization procedure followed by an annealing process was investigated. He reported an interesting comparison between different doped nanotubes. In Table 1 [128133], the obtained results are reported.

ZnO nanotubes have been also studied for application in DSSCs. The utility of high surface area Al-doped ZnO (AZO) nanotubes for DSSC application was studied by Martinson et al. [113]. Compared to similar ZnO-based DSSCs, the AZO nanotubes show superior photovoltage and fill factors with photoconversion efficiencies up to 1.6%. Xie et al. [134] prepared a novel type of coaxial TiO2/ZnO nanotube arrays for DSSCs. The obtained efficiency is 2.8%, due to enhanced charge separation effect in this structural arrangement [134]. A smaller portion of literature is devoted to other TiO2 one-dimensional nanostructures, for example, nanospindles [135, 136], electrospinning materials [137], and nanoembossed hollow structure (NeHS-TiO2) [138140]. In particular, one-dimensional hierarchical structures with hollow or porous interiors are considered promising structures due to the longitudinal pathways of the hollow or porous interior structures and the 1D interconnected nanocrystal network that allows for efficient electron transport [141]. Wang et al. reported 1D titania with tunable structural hierarchy that exhibits about 1.3–1.5 times higher power conversion efficiency than commercial P25 as photoanode material for DSSCs [142]. Zhu et al. synthesized ZnO nanorods-nanosheets (NR-NS) hierarchical architecture [143].

1.3.6. Combination of 1D Nanostructures (2D Nanostructures)

A combination of nanowires, nanotubes, and nanoparticles creates 2D nanostructures that are significant for their sufficiently high surface area available for dye loading. In the hybrid structures, the resulting photoelectrode film presents the advantages of the one-dimensional nanostructures and the exposed surface area could offer more space for dye absorption and electron conveyance. Most research was performed on the array films of ZnO and TiO2 nanowires or TiO2 nanotubes filled with ZnO or TiO2 nanoparticles [144]. In Table 2, some 2D hybrid nanostructures used as photoanode of DSSC are reported [145151].

1.3.7. 3D Nanostructures (Hierarchic Nanostructures, Composite, Multilayer, and Thin Film Materials)

As we have seen until now, electron transport in nanoparticle films is via a random route with multiple trapping and detrapping events. This mechanism seriously affects the electron diffusion coefficient and the electron collection efficiency. Moreover, small sized particles show inefficient light-scattering ability, which results in poorer light-harvesting efficiency. On the other hand, 1D and combination of 1D (2D) nanostructures attracted much attention in DSSCs owing to the improvement of the charge collection efficiency, which ascribes to a faster electron transport and slower recombination rate. However, the low surface area of 1D nanostructures hinders significant improvement of the photovoltaic performance owing to poor dye loading compared to nanoparticles. For a better balance of those three factors (surface area, light-scattering, and electron transport), bi-/trifunctional hierarchical structured materials consisting of nanoparticles/nanorods, have been widely investigated as photoanodes for DSSCs. These structures demonstrated enhanced photovoltaic performance due to their superior dye adsorption properties, light-scattering ability, and faster electron transport compared to traditional nanoparticle only based DSSCs [152]. Liao et al. in 2011 presented the effect of TiO2 morphology on photovoltaic performance of DSSCs with particular attention on hierarchical structures [153]. Chen and Yang [154] presented a series of works in which they demonstrated the viability of new photoanode architectures in DSSCs comprising building blocks of ZnO nanoplates [155] and TiO2 nanosplindles [135]. Recently, they presented a systematic study about DSSCs based on ZnO nanotetrapods-based photoanodes [156] and ZnO nanotetrapods/SnO2 nanoparticles composite photoanode [157, 158]. Other authors report ZnO 3D nanostructure (e.g., caterpillar-like nanorods () [116], nanodendrites () [159]) based photoanodes that show significant photoconversion efficiencies for DSSC application.

Sauvage et al. [160] introduced a novel photoanode with a nanoscale structure that resembled a forest, fabricated by pulsed laser deposition (PLD) under relatively high background gas pressures. This new nanostructure, which they named “nanoforest,” replaces the traditional random nanoparticle oxide network by vertically aligned bundles of TiO2 oxide nanocrystals. Grätzel speculated earlier that a structure combining the high surface area of nanoparticles with the electron transport directionality of vertical rods would be optimal for DSSCs [16].

Another approach to the improvement of DSSC is the enhancement of the conduction band level relative to the electrolyte redox/oxidation level that would result in higher photovoltage [161]. Nb2O5 is an example of one such -type transition metal oxide with a higher conduction band edge than TiO2. Ghosh et al., for the first time, reported the fabrication of a Nb2O5-based photoanode with a nanoforest structure and the results of investigations of the photovoltaic properties of this material under a variety of key growth parameters (layer thickness, background gas pressure, and composition) [73].

Other efficient TiO2 3D nanostructures producing good efficiency values are branching nanorods ) [162], dendritic hollow structures [163], bridge nanotubes ( [164], and so forth. Photoanodes made of two or more materials have attracted attention due to the obvious advantages of combining different materials. Yan et al. [165] combined the high electron transport rate of ZnO and the high electron injection efficiency of TiO2 with Ru-based dyes. Huu et al. [166] presented a hierarchical photoanode comprising a SnO2 nanoparticle underlayer and a ZnO nanorod overlayer and its photovoltaic performance was compared to photoanodes consisting of SnO2 nanoparticles only and ZnO nanorods only.

Satapathi et al. presented a DSSC fabricated using photoanodes made from graphene-TiO2 nanocomposites [167]. The relationship between the size of graphene sheet and the cell performance was explored. It was concluded that the cells loaded with the smaller graphene sheets yielded larger enhancement. The maximum efficiency was obtained by 184 nm sheets with .

Also composite TiO2 anatase materials with CaTiO3 [168] and BaTiO3 [169, 170] have been applied to electrodes for DSSCs. Among perovskite-type compound, BaTiO3 is a -type semiconductor with band gap energy of 3.3 eV. Such band gap value may increase the open-circuit voltage of a DSSC as Suzuki and Kijima reported in 2005 [171].

Manoharan and Venkatachalam investigated the photoelectrochemical performances of DSSCs based on aluminum doped titanium dioxide (Al-TiO2) nanoparticles/nanowires (TNPWs) composite photoanode layers, with different electrolytes. The maximum obtained efficiency is 7.26% [172].

An efficiency value of 3.44% has been reported by Song et al. [173] using as double scattering layer a ZnO film consisting of ZnO monodisperse light-scattering layer and a submicrometer-sized plate-like ZnO film as overlayer in the photoanode of a DSSC. Ghanbari Niaki et al. presented a new strategy for improving the efficiency of TiO2 DSSCs by design of a new double-layer film doped with Zn ions, with various morphologies [174].

1.3.8. Mesoporous Structures (3D Oxide Aggregate Nanostructures)

Among 3D nanostructures, mesoporous MOS-based materials (in particular titania-based) with a crystalline framework, high specific surface area and tunable pore size have received significant attention for energy conversion applications. Several morphologies, such as mesoporous TiO2 films [175], beads [176], monoliths [177], and networks [178], with controlled porosity have been prepared via different synthetic strategies. The most interesting are monodisperse TiO2 beads with a submicrometric-sized diameter. Chen et al. [179] reported the synthesis of monodisperse mesoporous anatase titania beads with high surface area and tunable pore size and grain diameter via a combination of sol-gel and solvothermal processes in presence of hexadecylamine (HDA) as a structure-directing agent. In this way, they enhanced the light-harvesting capability of TiO2 electrode without sacrificing the accessible surface for dye loading. These mesoporous TiO2 beads have a diameter of less than 1 μm and are composed of anatase TiO2 nanocrystals. The beads have been used in the preparation of photoanodes for DSSCs and an improved efficiency was observed when compared to analogous cells prepared using standard Degussa P25 TiO2 photoanodes of similar thickness. An overall light conversion efficiency of 7.20% (open-circuit voltage (Voc) 777 mV, short-circuit current density () 12.79 , and fill factor (FF) 0.72) was achieved using the mesoporous TiO2 bead electrodes [180]. The substantial improvement of and for the hierarchical sphere-based DSSC when compared to other nanostructure-based DSSCs is mainly due to the larger dye loading, higher light-scattering ability, faster charge transport, and longer electron lifetime [181, 182]. Submicrometric mesoporous TiO2 beads are also used to form a scattering layer on top of a transparent, 6 mm thick, nanocrystalline TiO2 film. Later, Sauvage et al. presented DSSCs with photoanodes made of mesoporous TiO2 beads that achieved a power conversion efficiency over 10% [183]. Kim et al. [184] reported a two-step method for the synthesis of TiO2 aggregates which first produced TiO2 spheres via a controlled hydrolysis and then etched the spheres under hydrothermal conditions. These aggregates, applied to DSSCs, yielded an efficiency as high as 10.5%.

In addition, ZnO spheres were used to enhance the performances of ZnO-based DSSCs. These later were synthesized by hydrolysis of a zinc salt in polyol medium [185]. A significant difference in the conversion efficiency values (2.4% and 5.4% for ZnO nanoparticles and aggregates, resp.) has been verified. The explanation of this difference is based on the consideration that ZnO is not stable in an acidic dye solution and the formation of a Zn2+/dye complex on the surface of ZnO may seriously hinder the electron injection process from the dye molecules to the semiconductor [186]. In the case of film made of ZnO nanoparticles, the formation of complex on the film surface would block the pores and result in an incomplete infiltration of dye molecules [187]. However, the situation is quite different in the case of aggregates. Due to the existence of large pores among the submicron-sized aggregates, the dye penetration can be accomplished in a very short time (e.g., 30 min) prior to the formation of the complex layer [185, 188]. Other mesoporous structures have been investigated for DSSCs applications. The use of preformed mesoporous solid as hard templates allows for the preparation of novel mesostructures of TiO2 with high crystallinity. Moreover, hard templates can provide an excellent support and confinement framework to prevent collapse of mesoporous structures, thus allowing for high crystallinity. All these characteristics are found in mesoporous single crystals (MSC) of anatase, which were first prepared by Crossland et al. [189]. The MSC anatase displays higher conductivity and electron mobility than those of conventionally used nanoparticles. In solid-state DSSCs they obtained a power conversion efficiency of 7.2%. Latini et al. [190] used MSC as photoanode for conventional DSSCs in a comparison between hard and soft template-based materials. Five different anatase phase mesoporous titanias were used as photoanodes in DSSCs: two materials were synthesized by using silica nanospheres (hard template) and the other three using two different organic polymeric templating agents, P123 and Brij 58 (soft template). Unexpectedly, MSCs are the least suitable for application in DSSCs because of their low specific surface area. Finally, Zhang et al. developed a simple surfactant-sulfuric acid carbonization method to synthesize ultrastable ordered mesoporous titania with high crystallinity [191].

1.3.9. Doping and Codoping Effect on Nanocrystalline and Mesoporous Structures

In addition to the optimization of the oxide nanostructure, the combined use of different metal oxides and/or the use of doped materials with higher conduction band energies should, in principle, allow further improvement of DSSCs performances. In metal oxides, the doping with suitable cation/anion modifies the band gap and may modulate their electrical proprieties. The properties of titania, including anatase to rutile phase transition as well as the photoactivity, greatly depend on the presence of dopants, that is, cations [192] and anions [193] and impurities, as well as on the crystallinity [194, 195], grain size, surface area, and so on, in addition to the presence of such substances as amorphous silica at the interfaces and/or grain boundaries. All these properties have a profound influence on the mass transfer rate, diffusion, and crystallite growth of titania in composite nanoparticles [196].

The ionic radii of niobium Nb5+, aluminum Al3+, gallium Ga3+, yttrium Y3+, and scandium Sc3+ are quite close to that of titanium Ti4+. Solid solutions of these cations in titania have been prepared and characterized [197]. For example, Chandiran et al. studied the effect of Ga3+ and Y3+ in mesoporous anatase and of Nb5+ in nanocrystalline anatase for DSSCs photoanodes [198, 199]. Latini et al. [105] reported the beneficial effect on the performances of DSSCs of Sc3+ doping of mesoporous anatase beads. At 0.2% of Sc atoms, a maximum efficiency of 9.6% was obtained, which was found to be 6.7% greater than the efficiency of DSSCs with pure anatase. Li et al. [200] reported a DSSC based on TiO2 photoanode that was modified by the Al-doped TiO2 layers using the chemical bath deposition method. The Al-doped TiO2 layer improved the photocurrent density. The effect of a codopant M (M = Ga, Al, Sc) on the formation, crystallite growth, optical band gap, photocatalytic activity, and phase stability of anatase-type TiO2 solid solutions (NbXMXO2) containing the same amount of dopant Nb that were directly formed as nanoparticles under mild hydrothermal conditions at 180°C for 5 h, was investigated by Hirano and Ito [201]. The effect of lanthanum ions (La3+) on charge trapping in dye-sensitized solar cell (DSSC) photoanodes has been investigated with doped and surface-treated TiO2 nanoparticles [202]. Including those mentioned before, transition metal dopants in general could extend the photoresponse in the visible-light region. This is due to low energy photon excitations of the corresponding metal oxide clusters with smaller band gap and partially from the excitations of the introduced localized states in the band gap of doped TiO2 [203].

On the other hand, nonmetal dopants (N, C, B, S, P, etc.) can exist as isolated atoms rather than clusters, which have greater potential for realizing visible-light photoactivity [204]. In anion doped TiO2, the formation in the oxide phase of new defect levels slightly above the valence band generally causes the red shift of absorption spectrum [205209]. Pan et al., for example, presented the enhanced efficiency of DSSCs by trace amount Ca-doping of TiO2 photoelectrodes. The best photovoltaic performance was obtained from 50 ppm Ca-doping with a conversion efficiency of 7.45% [210]. In Table 3 [211223], some useful dopants for anatase TiO2 have been grouped together as well as their respective photoconversion efficiencies as reported in literature.

The amount of papers dedicated to systematic studies of the effect of heteroatoms and their concentration in the anatase lattice on the performances of DSSCs are far less numerous. Cavallo et al. [224] have undertaken a systematic work of synthesis, characterization, and test of DSSC photoanodes containing mesoporous anatase bead doped with RE (rare earth) cations (Nd, Sm, Gd, Er, and Yb). The maximum efficiency has been obtained for the sample containing 0.2% Er metal atoms (8.7%).

Similarly, papers dealing with doping of ZnO with different cations and anions are also present in literature, as reported in Table 4 [225232].

2. Part II: -Type Semiconductors

2.1. Introduction

In the recent past, there has been a crescent interest towards the development and the realization of solar radiation conversion devices in which an electrochemical process of reduction is photoactivated via the absorption of light by a cathode either in the pristine [233237] or in a modified/sensitized state [238264]. These devices are photoelectrochemical cells (PECs) [40, 87] with photoactive cathodes consisting of -type semiconducting materials and include -type dye-sensitized solar cells (-DSCs) (Figure 6) [242, 243, 247, 248, 264] and tandem DSCs (t-DSCs) (Figure 7) [265269] cells of photoelectrolysis for nonfossil fuels production, namely, molecular H2 in the process of water splitting (Figure 8) [270281] and for carbon dioxide redox reduction (Figure 9) [282313]. The initial charge separation produced as a consequence of light absorption [312, 313] can occur either directly on the -type semiconductor (Figure 9) or on the electrically connected sensitizer which accomplishes successively the transfer of charge according to the vectoriality imparted by the relative positions of the energy levels of the electronic states involved and by the kinetics of the possible redox reactions (Figures 68).

Several requirements of thermodynamic as well as kinetics character must be fulfilled in order to realize an efficient process of photoconversion into the desired product. The latter can be the electrical power as in case of the -DSCs [314] and t-DSCs [268] (Figures 6 and 7) and fuels/chemicals in case of the photoelectrolytic cells based on the working principles schematized in Figures 8 and 9. The most crucial aspect generally limiting the performance of radiation conversion devices is the one related to the effectiveness of charge separation. This is because the separation of charge, once effectuated, can either evolve (a) to the displacement of the charges towards the sites/reactants for the realization of the desired redox transformations (reductions in the present context) or (b) lead to the occurrence of unwanted charge recombination/trapping phenomena depending on the relative rates of processes (a) and (b) and the chemical-physical features of the various interfaces on which the events of charge separation take place [315317]. For this reason is fundamental to envisage photoconversion devices with rapid mechanisms of photogenerated charge displacement within the semiconductor [318]. A relatively recent example of that has been the invention of the DSC by Grätzel [2]. The latter represents a photoconversion device in which the electrons photoinjected at a sensitized photoanode of TiO2 in the rutile form [319] are swiftly drifted to the metallic cathode for the occurrence of a fast interfacial et from the cathode to the anode-generated oxidized form of a redox couple. Fundamentally, this successful result is based on very fast charge transport within the photoactive semiconducting electrode. This would prevent recombination between the photocharge generated at the surface of the light-absorbing semiconducting electrode and the redox species in the electrolyte initially oxidized/reduced by the photoinjected charge [320]. The most recent advancements in the design and realization of PECs of -type have been achieved when the photoelectroactive -type semiconductors had nanostructured features (Figure 10) [238, 268, 321327]. This is motivated by the fact that a nanostructured surface increases enormously the effective area of contact for the interfaces created by the nanostructured semiconducting material with respect to the same system possessing a compact morphology. This difference improves the kinetics of those electrochemical processes occurring at an extended interface in favour to the nanostructured version [314].

Another important motive of interest in adopting nanostructured semiconductor electrodes resides in the possibility of modifying the optical properties of the pristine system through sensitization with very large amounts of dye-sensitizers anchored per unit surface [328, 329]. These colouring additives sometime represent the sole actual light-harvesting species in case the semiconductor has a wide band gap with respect to the energy of the luminous radiation [330]. On the other hand, the main concern that might arise by the utilization of nanostructured semiconductors regards the delineation of an opportune pattern of energy bands (or a set of discrete energy levels), which still allows the electrical connection through the nanostructured semiconductor [331, 332]. In fact, a nanostructured semiconductor represents a system characterized by not having an internal electric field because of the inexistence of a charge depletion layer [233235]. In this type of semiconductors, the typically discontinuous pattern of frontier energy levels is also associated with electronic states delocalized disorderly at a variable extent [236]. Consequently, the latter feature imposes a charge transport mechanism of variable range hopping between localized states [237340], through which the displacement of charge is originated by diffusion [341345].

In this review, we will report the most recent developments and progress achieved with PECs having photoactive cathodes (vide supra). Despite the fast advancement of the technology of perovskite solar cells [346380], in the present contribution, this type of device will be not reviewed since the photocurrent generated in perovskite based photoconversion devices is not associated with the primary event of photostimulated redox reactions but is rather a consequence of a light-driven process of charge separation at a p-i-n junction. As such, it does not introduce any redox states variations in the conducting materials during ordinary operative conditions of perovskite systems [381383].

2.2. Devices for the Conversion of the Electromagnetic Radiation into Electrical Power with Photoelectroactive -Type Semiconductors: Analysis of Performances Recently Reported in the Literature
2.2.1. DSCs of -Type and -Type

One of the first studies on sensitized -type semiconductors with nanostructured features for DSCs of -type was reported in 2000 [268]. In this seminal work, nanoporous nickel oxide (NiO) prepared via sol-gel was employed as cathode, while erythrosine B was the dye-sensitizer. This prototypical -DSC displayed a photovoltage of about 80 mV and a photocurrent density of ca. 0.2 mA cm−2, the photocurrent being originated by the associated photoinjection of mobile holes in the VB of NiO after photoexcitation of the chemisorbed colorant (Figure 6). Despite the low overall efficiency (less than 0.1%), these results represented in principle a quite remarkable step forward especially in comparison with the modest performance of one of the first -DSCs that utilized a semiconducting photocathode with compact morphology [330]. Anyhow, when compared to the analogous -type counterparts, -DSCs show their relative limits especially in terms of overall efficiencies since the most performing -DSCs reach at most 1-2% of overall conversion efficiency [384, 385], whereas there is a constant upgrade of efficiency records with -DSCs that reach values in the broad range of 10–15% [4, 386388]. What mainly motivates the research on -type photoelectrochemical devices is the realization of t-DSCs (Figure 7) [265269], which possess both electrodes with photoactive features due to their sensitization with colorants having complementary absorption properties [235].

Such a tandem configuration would allow the achievement of larger open-circuit photovoltages with respect to the corresponding DSCs having single photoactive electrodes (Figure 11), but the gain in photopotential is obtained at expenses of current density and fill factor, that are both controlled by the less performing photoactive electrode (typically the nanoporous photocathode) [235, 265269]. The expected improvement of the overall photoelectrochemical performance in t-DSC [389] in comparison to the DSC configurations with single photoactive electrode is possible only when - and -type devices generate separately photocurrents and reach FFs with comparable values, the sensitizing agents of cathode and anode being complementary in terms of optical absorption (Figure 12). The present work updates the most recent achievements in the fields, which were not covered by the works of Daeneke et al. on the general limits of -DSCs [242] and of Dini et al. [237, 241] about the influence of photocathode nature and preparation method on the relative photoelectrochemical properties in the corresponding -DSCs. Table 5 presents the most recent achievements obtained in the best performing -DSCs when the redox shuttle was the couple /. The comparison of the -DSCs regards PECs differing for the structure of the dye-sensitizer [238, 239, 265, 266, 314, 327, 390393] (Figure 12) and the nature of the nanostructured -type photocathode [394398].

As recognized in precedence [242], the main drawback of a -DSC is the attainment of relatively low fill factors, which generally never exceed 40% in the most efficient configurations (i.e., with , Table 5) [238, 239, 265, 266, 327, 392] when the redox couple / is present. Upon adoption of a redox mediator other than the one based on iodide, the -DSCs display performances with considerably ameliorated characteristics (Table 6) [240, 384, 399408]. These facts would indicate that the main routes of photocathode optimization and refinement of dye-sensitizer design (Figure 13) must be necessarily accompanied by a systematic research on the selection and eventually the definition of new redox mediators in order to improve the performance of -DSCs.

Closely related to the progress of -DSC is the one of t-DSC with electrodes both nanostructured and opportunely sensitized [235, 265269, 399, 409]. In terms of performance, the t-DSC represents a device still in its infancy, which strongly needs of a further stage of evolution for reaching and eventually surpassing the performances of the corresponding -DSCs in accordance with the predictions of thermodynamics [389]. Table 7 presents the list of the parameters characterizing the performance of the t-DSCs reported so far.

3. Conclusions

The use of -type semiconductors as photoanodes and -type semiconductors as photoactive cathodes of - and t-DSCs has been reviewed. Concerning the -type semiconductor, a substantial performance improvement has been obtained by optimizing the dimensions of the particles that can be achieved by tailoring the synthesis conditions. The role of band gap, morphology, composition, and doping of -type semiconductors for the development of efficient photoanodes has been reported. Several contributions were summarized here with the description of nanostructures (0D, 1D, 2D, 3D, and mesoporous nanostructures) and of the effect of doping of photoanodes materials (mostly TiO2 and ZnO) on the performances of DSSCs. The maximum photoconversion efficiency (13–15%) is reported using titania-based photoanodes. About photocathodes, we have posed our attention towards the most recent developments on -type semiconductors having nanostructured features and mesoporous morphology. Such features impart large surface areas and allow the anchoring of a large number of dye-sensitizers per unit area with favourable consequences on the efficaciousness with which the desired photoactivated electrochemical process occurs. An important issue related to the employment of nanostructured semiconducting cathodes is the determination of the position of the energy levels for the frontier states directly involved in the exchange of photocarriers with the dye-sensitizer in the excited state. Among nanostructured semiconducting cathodes, nickel oxide having nonstoichiometric features is the most important representative of -type photocathodes due to its intrinsic photoelectroactivity. Other examples of nanostructured photocathodes have been also mentioned in the present review, thus indicating new possible directions of research in the materials science of semiconducting electrodes of -type. Also the most important examples of dye-sensitizers for -type nanostructured electrodes have been reviewed. The aspects of electronic conjugation, presence of spacers and substituents with peculiar electronic effects, and state of surface immobilization have been briefly considered in the analysis of the most important structural factors of the dye-sensitizers, which control the process of charge photogeneration and injection in the photocathode. An aspect that has not been considered in depth is the feasibility of the synthesis and purification methods for purposely designed photo-/electrocatalytic systems for -type semiconductors. In terms of short-term goals, the main purpose of the development of increasingly efficient -DSCs is the achievement of efficiencies comparable to those of the -type counterparts in order to produce tandem devices with efficiencies approaching the theoretical limit.

Competing Interests

The authors declare that they have no competing interests.

Authors’ Contributions

Part I has been written and compiled by Alessandro Latini, Carmen Cavallo, and Francesco Di Pascasio who contributed equally to the completion of this first part of the review. Part II has been written and compiled by Danilo Dini and Matteo Bonomo who contributed equally to the completion of this second part of the review.

Acknowledgments

The authors wish to thank the Università di Roma “La Sapienza” for financial support through the Project Ateneo Sapienza (Protocol no. 2011/VG1-C26A11PKS2). Moreover, the authors gratefully acknowledge the financial support from Ministero dell’Istruzione dell’Università e della Ricerca-MIUR (Project Title: DSSCX; Protocol no. PRIN 2010-2011, 20104XET32). Danilo Dini wishes to thank Professor J. G. Vos at the School of Chemical Sciences of Dublin City University (DCU), Ireland, for profitable discussion.