Structural Engineering on Pt-Free Electrocatalysts for Dye-Sensitized Solar Cells

In recent decades, plenty of nanomaterials have been investigated as electrocatalysts for the replacement of the expensive platinum (Pt) counter electrode in dye-sensitized solar cells (DSSCs). The key function of the electrocatalyst is to reduce tri-iodide ions to iodide ions at the electrolyte/counter electrode interface. The performance of the electrocatalyst is usually determined by two key factors, i.e., the intrinsic heterogeneous rate constant and the effective electrocatalytic surface area of the electrocatalyst. The intrinsic heterogeneous rate constant of the electrocatalyst varies by different types of materials, which can be roughly divided into five groups: non-Pt metals, carbons, conducting polymers, transition metal compounds, and their composites. The effective electrocatalytic surface area is determined by the nanostructure of the electrocatalyst. In this chapter, the nanostructural design and engineering on different types of Pt-free electrocatalysts will be systematically introduced. Also, the relationship between various nanostructures of electrocatalysts and the pertinent physical/electrochemical properties will be discussed.


Dye-sensitized solar cells (DSSCs)
Fossil fuel, as a limiting energy source, may be run out in the upcoming centuries. However, the consumption of energy increases every year [1,2]. As a result, finding and developing renewable energy sources is an urgent problem. Due to the unlimitedness of renewable energy resources, they are candidates to be reliable replacement for sustainable usage in the future. Among them, the Sun has been considered as one of the most promising renewable energy sources. It provides about 120,000 terawatts to the earth, which equals thousand times of the current energy consumption rate. The solar cells can utilize the sunshine and transform to electricity [3][4][5]. Generally, solar cells can be classified to four generations: the first generation is silicon-based solar cells; the second generation is CIGS (CuInGaSe), CZTS (CuZnTiSe), and CdTe solar cells; the third generation is organic photovoltaics (OPVs) and dye-sensitized solar cells (DSSCs); and the fourth generation is perovskite solar cells (PSCs). The first and second generations have been widely explored for decades, and they are the most common solar cells at present. However, they are fabricated through expensive, toxic, energy-intensive, high-temperature, and high-vacuum processes. Therefore, DSSCs are very competitive to the first and second generations of solar cells due to numerous advantages including easy fabrication, low cost (Figure 1) [3], and high performance at dim-light condition. Moreover, DSSCs can be used in indoor ambient applications [5][6][7][8][9].
A DSSC is composed of a photoanode, electrolyte, and counter electrode (CE), as shown in Figure 2. When a photoanode is excited by the sun or photon, it will release the electron to the external circuit. At the same time, the iodide/triiodide (I − /I 3 − ) redox couple will relax the photoanode to its ground state. Then the CE will reduce the redox couple to regenerate the DSSC. Among them, the CE plays an important role to determine the DSSC performance [10]. At the CE/electrolyte interface, the electrochemical mechanism goes through I 3 − decomposition (Eq. (1)) → adsorption (Eq. (2)) → catalytic reduction reaction (Eq. (3)) → desorption (Eq. (4)), and the overall reaction shows as Eq. (5) [11]. Among these reaction steps, Eq. (3) is found to be the slowest step, which means the rate-determining step to decide the DSSC performance.
To date, there have been a very limited number of non-Pt nanomaterials that could have a comparable intrinsic heterogeneous rate constant to that of Pt. However, the specific structural designs of nanomaterials would largely increase the effective electrocatalytic surface area so as to provide better overall electrocatalytic ability than Pt. Moreover, the nanostructured electrocatalysts could have appropriate interfacial affinity, good electrochemical stability, or specific self-assembly natures; these properties may influence the DSSC performance as well. A typical nanostructured material can be defined if any dimension of the material is lower than 100 nm. The nanostructured material can be classified into three groups: zerodimensional (0D, e.g., nanoparticle, nanocube, etc.), one-dimensional (1D, e.g., nanorod, nanotube, nanoneedle, etc.), and two-dimensional (2D, e.g., nanosheet, nanopental, etc.) structures, as shown in Figure 3. Generally, 0D structure is  expected to supply the high electrochemical surface area, and 1D/2D structures are claimed to have directional electron transfer pathways. In this chapter, different strategies of designing nanostructured carbon materials, conductive polymers, and transition metal compounds to increase their active surface area/charge transfer route will be systematically discussed. The corresponding DSSC performance is also included.

Carbon materials
Carbon materials, composed of carbon atoms having sp 2 -hybridization (e.g., graphite, graphene, graphene oxide, and carbon nanotube) or sp 3 -hybridization (e.g., carbon black and activated carbon), have been widely investigated as the CEs due to their material specialties such as low cost, high electrical conductivity, high thermal stability, and good corrosion resistance [3,4,[80][81][82][83][84]. The carbon materials with sp 2 -hybridization normally exhibit a 1D or 2D structure, and those with sp 3hybridization have a 0D structure. However, carbon materials often exhibit serious aggregation that reduces the material conductivity and limits the electrochemical surface area. Here we include several studies which provided practical strategies to overcome this problem and reach comparable/better cell efficiencies than the Pt-incorporated DSSCs, as listed in Table 1.
For example, Fan et al. used a small 0D porous carbon nanoball (diameter = 20 ± 3 nm) to assemble a large 0D hollow nanoball (diameter = 100 ± 10 nm), as shown in Figure 4(a) [12]. Tseng et al. introduced a one-step synthetic method to make tens of 2D nitrogen-doped graphene with a thickness of ~3.5 nm stacking together to form a building block as a 0D hollow nanoball, as shown in Figure 4(b) [22]. Fan et al. used a small 0D porous carbon nanoball (diameter = 20 ± 3 nm) to assemble a large 0D hollow nanoball (diameter = 100 ± 10 nm), as shown in Figure 4(b) [12]. The hollow nanoballs consisting of nitrogen-doped graphene and porous carbon gave their DSSCs ηs of 7.53 and 8.67%, respectively, which were comparable to the Pt-based cells. Besides, nitrogen-doped graphene (Figure 4(c), 7.07%) [13] and the wrinkled 2D graphene (Figure 4(d), 7.80%) [14] nanosheets were used to form a honeycomb-like structure having extra surface area and vertically aligned subnanosheets as the additional electron transfer routes. The combination of few kinds of carbon materials was reported to form a hierarchical structure, which could not only create a high surface area but also a directional electron transfer pathway. Dong et al. made 1D few-walled carbon nanotubes (CNTs, tens of microns long) vertically fuse onto the 2D graphene nanosheet (<1 nm thick), as shown in Figure 4(e) [15]. The red seven-membered rings at the neck seamlessly fuse the tubular CNTs to the planar graphene without obvious CNT aggregation (Figure 5(a)). Even though the CNTs only covered few parts of the electrode surface, they still benefited the electrolyte wetting and electron transfer rate within the counter electrode, leading to a better η (8.2%) than the Pt-based cell (6.4%). Ma et al. reported a similar hierarchical structure, where the single-walled carbon nanotube (SWCNT) was located on a flat N-doped graphene (N-doped GN) nanosheet by the z-axis direction (Figure 4(f)) [16]. As a result, composite SWCNT@N-doped GN reached higher η (8.31%) than Pt (7.56%). Yeh et al. prepared a hybrid heterostructure of multiwalled carbon nanotube (MWCNT, diameter = 15 nm) and reduced graphene oxide (rGO, thickness = 25 nm) nanosheet, where the 2D rGO nanosheet-like shell was wrapped around the 1D CNT core (Figure 4(g)) [27]. In the MWCNT@rGO composite material, the tubular MWCNTs functioned as the 1D heterogeneous electron transfer pathways, which provided sufficient electrons to the electrochemical reaction; at the same time, the 2D rGO nanosheet supplied multiple edges as the active sites to reduce I 3 − to I − (Figure 5(b)). The hybrid heterostructure of MWCNT@rGO was found to avoid  [12,22], (c) nanosheet [13], (d) honeycomb [14], and (e-g) nanotube with nanosheet [15,16,27].
The scheme of hierarchical structures of (a and b) nanotube with nanosheet [15,27]. the aggregations among the MWCNTs or among the rGOs. Thus, the MWCNT@ rGO rendered its DSSC an η of 6.91%, which is close to the Pt-based cell (7.26%).
The hierarchical nanosphere with PPy (denoted PPy-HNS) has the hierarchical nanospherical structure with an average diameter of 100-200 nm, as shown in Figure 7(a) [44]. The PPy-HNS has the following photovoltaic parameters: a V OC of 0.70 V, a J SC of 16.49 mA cm −2 , a FF of 0.58, and an η of 6.71%. The nanopatterning process is one of the methods to obtain the specific structure [42]. The nanopatterned PEDOT CE shows uniform hole patterns with ~100 nm diameter (Figure 7(b)) and exhibits an η of 6.71%. Regardless of the nanoparticle or nanosize hole, their corresponding η values are still lower than the η of Pt CE. Therefore, there are other structures, which were synthesized to overcome the challenge.

Transition metal composites
Transition metal composites (TMC) possess high potential to replace Pt CE in DSSCs because of the similar electronic structures between TMCs and Pt. Metal compounds, including carbides, nitrides, chalcogenides, oxides, phosphides, and so on, have been applied as an electrocatalyst in DSSCs to replace expensive Pt. It is still a challenge to replace Pt with TMCs due to the relatively low conductivity of TMCs. Accordingly, various TMC structures, including nanoparticle, hollow sphere, nanorod array, nanowall, hierarchical nanorod, etc., are investigated to improve the performance of TMC-based CEs, as shown in Figure 8. The corresponding ηs of DSSCs with various structures are listed in Table 3.
α-NiS has a sphere-like morphology with a diameter of 50-80 nm, as shown in Figure 8(a) [78]. The other NiS (β-NiS) has a nanorod 2-5 μm in length and 1000 nm in diameter. The DSSC of α-NiS CE has a better η (5.20%) than the β-NiS (η of 4.20%). The particle size of α-NiS is much smaller than nanorods of β-NiS. The smaller the size of a particle, the larger specific surface area it possesses. With the increase of specific surface area of α-NiS, the conversion efficiency reaches a higher value. The nanoparticle of CoSe 2 /CoSeO 3 (CoSe 2 /CoSeO 3 -NP) has a diameter of 50-60 nm, as shown in Figure 8(b) [74]. And CoSe 2 /CoSeO 3 -NP has a larger reaction area than the nanorod and nanocube of CoSe 2 /CoSeO 3 , confirmed by the electrochemical double-layer capacitance, which is positively related to the  [78], (c) double-shelled ball-in-ball hollow sphere [70,74], (d) hollow spherical particle [71], (e) acicular nanorod array [49], ( f-h) nanorod [53,54,75], (i and j) nanosheet [55,66], (k and l) nanowall [64,72], and (m) hierarchical nanosphere with nanorod [77]. reaction area. The DSSC of CoSe 2 /CoSeO 3 -NP CE has an η of 9.27%, which is better than those of the nanorod and nanocube of CoSe 2 /CoSeO 3 and is higher than that of the Pt CE (7.91%). The double-shelled hollow sphere (BHSs) structure exists in NiCo 2 S 4 BHSs with the separation of hollow and solid parts. In Figure 8(c) [70], the diameter of the inner shell is 300 nm and that of the outer shell is approximately 550 nm. The thickness of the outer thin shell is 10-30 nm, which is quite less than that of the inner shell. The DSSC of NiCo 2 S 4 BHSs CE exhibits an η of 9.49%, which is higher than that of the Pt CE (η of 8.30%). From a broken NiCo 0.2 @C microsphere shown in Figure 8(d) [71], the well-defined hollow structure with a shell thickness of around 200 nm can be observed. Meaningfully, the hollow spherical space can greatly shorten the diffusion paths within the electrode and serves as a robust reservoir for ions. The NiCo 0.2 @C exhibits an η of 9.30%, which is higher than that of the Pt CE (η of 8.04%). Most of the nanoparticles, double-shelled ballin-ball hollow sphere, and hollow spherical particle structures have better ηs than the Pt CE.
Although TMCs present good electrocatalytic ability, the electrons may be insufficient at active sites. The rod structure is claimed to provide the specific electron transfer. It can supply sufficient electrons to keep consistent electrocatalytic reaction. From Figure 8(e), it can be observed that CoS has 1D acicular nanorod arrays with the relatively rough surface of the nanorods (noted CoS ANRAs-24h) [49]. It is vertical to the FTO substrate and has a height of about 7 μm. The DSSC with CoS ANRAs-24h CE shows an η of 7.67%, which is virtually the same as the sputtered Pt-CE (η of 7.70%). The MoN nanorod (NR) on the Ti substrate has a one-dimensional structure with a diameter of 40-100 nm and a length of 0.5-2 mm, as shown in Figure 8(f) [54]. The electrode structure is expected to trigger positive effects on the electrochemical processes occurring in the electrode films. The MoN NR-Ti CE shows comparable performance to that using a Pt-FTO glass electrode with a V OC of 0.740 V, a J SC of 15.26 mA cm −2 , a FF of 0.65, and an η of 7.29%. The single-crystal CoSe 2 has nanorods 50-800 nm in length and 20-150 nm in width, as shown in Figure 8(g), and possesses a lattice spacing of 3.71 ± 0.01 Å, corresponding to the   (110) planes of orthorhombic CoSe 2 [53]. Impressively, the single-crystal CoSe 2 CE produces an η of 10.20% with a V OC of 0.753 V, a J SC of 18.55 mA/cm −2 , and a FF of 0.73, which is better than the Pt CE (8.17%). The Ni 3 S 4 -PtFe heteronanorods are highly monodispersed with an average length of ∼34.0 nm and an average diameter of 9.0 nm, as shown in Figure 8(h) [75]. The DSSCs using Ni 3 S 4 -PtFe produce an η of 8.79%, which is higher than that of the Pt CE (7.83%). The 1-D structure is obviously promoting the electrocatalytic ability of TMCs and most of the 1-D structures for TMCs exhibit a better η than Pt CE. It can be claimed that the 1-D structures of TMCs could replace the Pt CE. The 2D structure of TMCs also has a specific electron pathway and it could be vertical to the substrate to offer sufficient electrons on active sites. Moreover, the hierarchical structure has both the advantages of a large reaction area and vertical electron pathway. For example, the direction of the fractured NbSe 2 sheet shows a structure with the [001] crystallographic orientation and revealed a very thick (>100 mm), disordered network arrangement of 2D sheets, as shown in Figure 8(i); in comparison, the ground materials were very thin, separated nanosheets [55]. The NbSe2 sheet CE has an η of 7.73%, which reveals the potential to replace Pt CE (7.01%). The WSe 2 is composed of several interlaced nanosheets with an average thickness of approximately 15 nm and a width between 60 and 100 nm, as shown in Figure 8(j) [66]. The WSe 2 CE shows good electrical conductivity, subsequent energy band calculation results, and large reaction area that exhibits an η of 7.48%.
Vertically-aligned structures of electrocatalysts were reported to facilitate faster charge transport from the substrate through the electrocatalysts to the electrolyte [64,72,77], as shown in Figure 9. This structure is expected to have better electrocatalytic ability. The nanowall and the hierarchical nanorod are used with TMCs. The CoSe 2 nanoclimbing wall (CoSe 2 /C-NCW) reveals arrays of vertically-aligned nanowalls with sharp edges, as shown in Figure 8(k) [64]. In addition, the nanowalls are covered with dot-matrix-like projections; these projections are expected to provide a large surface area to the film. On account of direct electron transfer and large surface area, the CoSe 2 /C-NCW film, on the whole, could be a better electrocatalyst for the reduction of I 3 − to I − , as shown in Figure 9(a). The cell with CoSe 2 /C-NCW CE reaches the highest efficiency of 8.92%, with a V OC of 0.73 V, a J SC of 18.03 mA cm −2 , and an FF of 0.67; this efficiency is even higher than that of the cell with Pt (8.25%). The CZTS nanowall electrodes (NWD) on Mo substrate show nanowalls with a width of ~500 nm, a thickness of nearly 15 nm, and a height of ~1.5 μm, which were adequately aligned in a densely packed array, which was nearly perpendicular to the surface of the Mo substrate, as shown in Figure 8(l) [72]. In this case, CZTS-NWD demonstrates a concept of "nano-geogrid"-reinforced CZTS nanowall electrode by synthesizing a thin layer of a porous CZTS nanostructure mimicking a geogrid on a substrate and then fabricating a CZTS nanowall on top of the nanostructure, as shown in Figure 9(b). The η of the NWD device is 7.44%, which is comparable to the Pt device (7.21%). TiO 1.1 Se 0.9 wrapping around a carbon fiber, and many nanospheres and nanorods of TiO 1.1 Se 0.9 are grown on the TiO 1.1 Se 0.9 under-layer [77]. The TiO 1.1 Se 0.9 nanorods are perpendicular to the surface of the carbon fiber; TiO 1.1 Se 0.9 nanorods, therefore, are expected to facilitate a fast and 1D electron transport from the CC substrate to the interface with the electrolyte, where I 3 − reduction occurs The correctional images show the hierarchical electron transfer route electrode, where the carbon fiber in the CC serves for transporting the main stream of electrons through its 1D direction, and the deposited electrocatalysts provide secondary channels of electrons for executing the reduction of I 3 − , as shown in Figure 9(c). The DSSC with the TiO 1.1 Se 0.9 -3/CC shows the best performance and gives the best η of 9.47%. Furthermore, the DSSC is used for further comparison with those cells with the Pt/CC (7.75%). In conclusion, the TMCs have good electrocatalytic ability and possess vertically aligned structures. They exhibit great ηs, which are better than that of the Pt CE.

Summary and future prospects
The counter electrode is a paramount part of DSSCS and has a significant influence on both the photovoltaic performance and the device cost of DSSCs. As a counter electrode, it must possess high conductivity and good catalytic activity toward electrolyte regeneration, as well as good stability. The DSSC devices employing CEs of different materials including carbon materials, conductive polymers, and transition metal composites have been summarized and discussed. One key point is that the CE performance can be optimized by combining special nanostructures into CE films to promote the industrialization of Pt-free CE catalysts. The nanostructure can briefly be classified into 0D, 1D, and 2D, which have different properties. The different materials with various nanostructures can overcome the problem of the material.
The carbon materials have numerous advantages including low cost, plasticity, simple fabrication procedures, high electrical conductivity, high thermal stability, and good corrosion resistance. The η of carbon materials has been improved by the hierarchal structures (nanotube with nanosheet and nanotube with nanoribbon) and most of the carbon materials with hierarchal structure CE have a better value of η than the traditional Pt CE. However, most of the performances of the DSSCs with carbon material CEs are still slightly lower than those DSSCs with Pt CEs. This mostly results from various resistances associated with the structurally complex carbon electrodes, such as bulk resistance through the comparatively thick carbon CE, contact resistance to the TCO substrate, the diffusion resistance in the pores of the CE, etc.
The conductive polymer materials possess outstanding electron conductivity, good adhesion, and easy fabrication. According to the literature above, it can be concluded that the 1D structure conductive polymer material-based CE can provide better η than the particles, nanosphere, and nanosheet structures. Although conductive polymer materials have larger reaction area and specific electron pathway, most of the conductive polymer material-based CEs still have a lower η than the Pt-based CEs. Only a few examples show better performance than Pt CE. It means that the conductive polymer materials need a hybrid with other electrocatalysts to obtain better electrocatalytic ability.
The TMCs exhibit great electrocatalytic ability, easy preparation, and modification. However, the poor conductivity needs to be solved in order to replace Pt CE. By synthesizing nanostructures, including nanoparticle, double-shelled ballin-ball hollow sphere, hollow spherical particle, acicular nanorod array, nanorod, nanosheet, nanoclimbing wall, hierarchical nanorod, etc., TMCs reveal better performances than the Pt CE. It can be said that the TMCs with nanostructure successfully replace Pt CE.
© 2019 The Author(s). Licensee IntechOpen. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/ by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Moreover, changing DSSC electrolyte toward the Cu, Co, Fe, etc. redox couples is another important research topic. Furthermore, the dim light condition application is another prospect of DSSCs. Among them, the matching material of CE is a key point to promise the η of DSSC. To get the point, it is dependent on the nanostructure and the hybrid materials, such as carbon materials with TMCs, carbon materials with conductive polymer materials, conductive polymer materials with TMCs, and TMCs with both conductive polymer materials and carbon material. Further development should focus on these main requirements: conductivity, catalytic activity, stability, efficiency, cost, and environmental friendliness. Also, the regulation mechanism for photo-induced charge carrier generation, evolution, and transportation should be of concern.