A third-generation photovoltaic cell known as a Gratzel or Dye-Sensitized Solar Cell (DSSC) has been gaining attention due to its high lab efficiencies, ease of manufacture, and earth-abundant, nontoxic composition. These cells promise to be marketable in the near future, but several hurdles must still be overcome before Gratzel Cells can be mass-produced. This chapter outlines the technological theory behind Gratzel cells, summarizes the state-of-the-art of DSSC research, and identifies the major challenges that still must be addressed to reduce recombination and increase voltage, current density, and cell lifetime.
2. Grätzel cell overview
Some of the most promising photovoltaic technologies are the emerging third-generation technologies. These range from polymer cells to cells made from proteins extracted from jellyfish (Chiragwandi et al). One variety in particular, known as the Grätzel cell, has gained prominence due to its very low cost and ease of manufacture. Developed in 1992 by Michael Grätzel at the Ecole Polytechnique in Switzerland (O’Regan & Grätzel, 1991), these biomimetic devices represent a significant departure from typical silicon solar cells. Unlike p-n junction cells, Grätzel cells, also known as dye-sensitized solar cells (DSSCs), work much like a photosynthetic plant cell (O’Regan & Grätzel, 1991).
The heart of a DSSC is the dye, which operates much like chlorophyll in a photosynthetic plant cell. This dye, usually a ruthenium-based organometallic complex, is responsible for harvesting photons. When a photon hits the dye, an electron is excited from the ground-state (HOMO) into the first excited state (LUMO), leaving behind a hole. In other words, the dye is responsible for generating an exciton (O’Regan & Grätzel, 1991).
Just as with a silicon cell, the hole and excited electron will recombine if they are not separated. Thus, in order to ensure an efficient design the electron and hole must be quickly separated from each other to prevent recombination. This separation has two parts: first, the excited electron must be brought to one side of the cell – the anode. Second, the hole must be transported to the opposite side of the cell – the cathode.
The first part – transporting the electron – is typically performed by a titania (TiO2) nanostructure and is known as the electron transport layer (ETL). This titania nanostructure is mesoporous, meaning that it has a pore structure on the order of 2-50 nanometers, and is annealed directly onto the anode. The dye molecules are introduced into this titania matrix via immersion in a dye solution, where they chelate to the TiO2 directly. When electrons are excited in the dye molecule, they are able to flow very quickly from the dye molecule LUMO into the lower-energy TiO2 conduction band, where they flow down to the anode’s conduction band, which is still lower in energy (Duffy et al., 2000). The importance of the nanoporous semiconductor layers, not only on electron transport, but also on increasing photon absorption are described in section 3.
The second task – transporting the hole – is performed by the hole transport layer (HTL) and can be accomplished in a variety of ways. The first dye-sensitized cells created by Grätzel et al. utilized an iodide / triiodide redox couple, where iodide regenerates the oxidized dye molecules. The resulting oxidized species, mostly triiodide, are transported to the cathode via diffusion, where they are reduced back to iodide. In this way, positive charge carriers (holes) are transported from dye molecules to the cathode (O’Regan & Grätzel, 1991). A variety of new HTL materials have emerged since the first Grätzel cell, ranging from new liquid electrolytes through solid-state conjugated polymers to quasi-solid polymer-electrolyte mixes. The function of the HTL and ways to improve it are described in section 4.
Regardless of materials, the electrical characteristics of these cells are still not completely understood. The voltage, specifically, is not well explained in the literature(Li et al., 2011, Johansson et al., 2005). Voltage is, ultimately, the difference in work function under illumination of the two electrodes from which voltage measurements are taken. In other words, this voltage is a result of difference in electrochemical potential of mobile electrons under illumination at the electrodes(Cahen et al., 2000). In a DSSC, the electrochemical potential at the electrodes is primarily determined by differences in energy between the HTL and ETL - the maximum possible photovoltage is widely agreed to be the difference in energy between the Fermi energy of electrons in illuminated TiO2 (roughly approximated as the conduction band edge) (Duffy et al., 2000), and the redox potential of the liquid electrolyte (O’Regan & Grätzel, 1991). In the case of a solid-state device, the difference between TiO2 conduction band and the HOMO level of the HTL material determines the maximum voltage instead (Johansson et al., 2005).
However, the observed photovoltage is rarely this maximum value – for example, similar cells using Poly(3,4-ethylenedioxythiophene) poly(styrenesulfonate) (PEDOT:PSS) and poly(4-undecyl-2,20-bithiophene) (P3PUT) differ in photovoltage by 0.7 V, even though HOMO levels of these polymers lie 1 eV apart(Johansson et al 2005, Smestad et al., 2003). One of the major detractors of voltage is recombination, the process by which excited electrons and holes recombine, giving off energy as photons. This not only detracts from current, since it means fewer electrons doing useful work, but it also lowers the observed voltage (Ghadiri et al., 2010, Ito et al., 2008, Saito et al., 2004). Because recombination causes fewer electrons to reach the anode, the Fermi energy of electrons at the anode will be decreased. Similarly, fewer holes reaching the cathode will increase the energy of electrons at the cathode. Photovoltage, which is determined by the difference between the two electrodes, is therefore decreased. This and other factors can seriously affect the photovoltage, thus complicating the mechanism for establishing a photovoltage.
Though these cells are nascent and their properties still being explored and optimized, DSSCs have achieved moderate efficiencies of over 11%, which is encouraging given the extremely low cost of these cells. Although these cells are not yet marketable, improvements in each of the three main components (Dye, ETL and HTL), as well as improved reliability and lifetime, will help Grätzel cells to achieve their full industrial potential.
3. The sensitizer
The dye in a Dye Sensitized Solar Cell has one of the most fundamental tasks. This molecule is responsible for photon harvesting and generation of an exciton. The dye, in other words, uses light to create the charge separation that is the goal of any electrochemical cell. This dye, though a complex molecule, is very simple functionally: it must only satisfy a few key requirements. The first of these is that it must absorb radiation across a broad visible spectrum. The second is that it must be small enough to enter the pores in the nanoporous semiconductor lattice. The third is that once inside the lattice, it must be able to chelate to the TiO2 or other semiconductor. The last, and potentially the most challenging requirement, is that the dye must be able to accomplish the other three requirements while being cheap and environmentally benign.
The majority of incoming solar power is due to visible wavelengths. Thus, a dye must be able to absorb across these wavelengths – it must appear black. Porphyrin-based sensitizers have been shown to exhibit such effects. Figure 2 demonstrates the broad-spectrum absorption properties of N-719 and Z-907. Note that the wavelengths at which these dyes absorb photons are determined by the band gap properties of the dye, while the magnitude of the absorption is due to the molar absorption coefficient, ε, of the dye. In order for a dye to be effective, it must have a high value for ε in addition to exciting over a wide spectrum.
The second requirement of the dyes, that they must be able to enter the anatase structure, is relatively easy to satisfy. Though in the original Grätzel article the large size of the dye molecule prevented complete saturation of chelation sites (O’Regan & Grätzel, 1991), dye molecules are typically on the order of 1 nmin diameter (Katoh et al., 2010), and thus are easily able to penetrate into the 2-50 nm pores characteristic of mesoporous materials. Moreover, further work is being done into alternative semiconductor morphologies such as anatase nanotubes that will increase dye access to adsorption sites. This research is further discussed in section 3.
Third, the dye must be able to chelate to the semiconductor structure. This is usually accomplished via carboxylate or carboxylic acid groups, with the deprotonated species being far more soluble (N3 and N719 dyes, two of the most common, are identical except for this difference). Chelation can also be accomplished via phosphate and sulfonate groups (Johannsson et al., 2005).
3.1. Ruthenium dyes
Ruthenium-based dyes have met the described requirements (i.e., photon absorption, spreading in porous media and chelation to the semiconductor) superbly. Dyes made of ruthenium-based organo-metallic complexeshave achieved overall conversion efficiencies of greater than 11% (O’Regan & Grätzel, 2004). The carboxyl moieties on the molecules allow them to chelate to the TiO2 surface, they have broad-spectrum absorption, and exhibit high molar extinction coefficients. N3 (fig. 3), for instance, has absorption maxima at 518 and 380 nm with a high value of ε = 1.35*104 M-1 cm-1 in alkaline solution at 500nm (Katoh et al., 2010). N719, the partially deprotonated salt of N3, has near-identical properties (ε=1.42*10-4 M-1 cm-1) (G.W. Lee et al., 2010) plus enhanced solubility, while Z907 is a hydrophobic variant with two long photoactive carbon chains that shows broad absorption across the visible range. The structures of these dyes, along with that of N749, are shown in figure 3.
Further advancements in dye engineering have produced even more functional compounds. Many dyes are unable to absorb in the red wavelengths, but black dye (N749), with its three thiocyanate functional groups, absorbs wavelengths in the red and near IR range, giving it a dark brown-black color. Figure 4 shows that the photocurrent response of black dye extends to about 900 nm while that of N3 is attenuated in the range of 750-900 nm. Despite a lower value for ε of 7.7*103 M-1 cm-1 (G.W. Lee et al., 2010), cells using black dye have achieved a conversion efficiency of 10.4% in full sunlight, approaching the maximum efficiencies of N3 (Grätzel, 2004), because of this spectral advantage. As well, Chiba et al. have built cells with black dye that yielded efficiencies of 11.1% when scattering particles were added to TiO2 to increase haze (ratio of diffused transmittance to total optical transmittance) (Chiba et al, 2006). Other emerging ruthenium sensitizers include HY2, which has absorption peaks at 437 and 550nm with a molar absorption coefficient of 1.98 x 104 M-1cm-1 at the latter wavelength, significantly higher than that of N3 at its peak wavelength of 530nm. DSSCs with HY2 have attained efficiencies of 8.07% (Chen et al, 2009). A ruthenium sensitizer called K20 has recently been submitted for a patent with the remarkably high molar absorption coefficient of 2.2 x 104 M-1cm-1 at 520nm, a significant improvement over N3, N719, and black dye. Characterization of DSSCs sensitized with K20 has not yet been pursued (Ryan, 2009).
All of these dyes have several beneficial shared properties. Ruthenium-based complexes exhibit metal-to-ligand charge transfer, a process in which the excited electron in a d-orbital of Ru is transferred to the π* orbital of the carboxyl ligand, from which it is injected into the conduction band of TiO2.This transfer is fast and irreversible, and will be discussed further in section 3. Moreover, they are remarkably stable over long periods of time. In the 1991 article by Grätzel and O’Regan, dyes sustained 5*106 turnovers without serious decomposition (O’Regan & Grätzel, 1991). More modern dyes are even more stable - in its solid state, N3 can withstand temperatures up to 280ºC, and maintains high performance over 108 redox cycles, the equivalent of 20 years of use in sunlight (Grätzel, 2004).
If pure functionality were the only requirement for these dyes, then perhaps alternatives to these existing dyes would not be needed. However, the last and most difficult requirement is that these dyes must be cheap, nontoxic, and environmentally benign. The greatest barrier to such a dye is the use of ruthenium as a metal center in these metal complex dye molecules. World production of ruthenium is approximately 12 metric tons (Lenntech), and though it is found in the rare minerals laurite, ruarsite, and ruthenarsenite, its commercial recovery is limited almost entirely to trace elemental amounts in nickel deposits in Africa and the Americas. Due to its rarity in difficulty to obtain, Ruthenium is expensive even in small quantities. For instance, a 10 mg sample of N-3 from Solaronix currently costs approximately $244USD (Solaronix.com as Ruthenizer 535). Moreover, ruthenium complexes are potentially hazardous to health and the environment. Though not yet fully studied for toxicity in their own right, when heated in air ruthenium complexes form ruthenium tetroxide, a highly volatile and toxic compound that damages the eyes and the upper respiratory system (Dierks, 2006). For all these reasons, a variety of entirely organic dyes as well as dyes using different metal centers are being developed for use in DSSCs.
3.2. Ru-free porphyrin dyes
Several main approaches have emerged to bypass the need for ruthenium dyes. One of the most common is to modify the dye structure to allow for a different metal center such as zinc or magnesium. Porphyrins, a class of naturally occurring biological compounds that include chlorophyll and hemoglobin, are known to absorb radiation in the visible range due to a conjugated system of pyrolles, frequently complexed around a metal ion center (iron in the case of hemoglobin, and magnesium in the case of chlorophyll). As chlorophyll is the biological inspiration for many of the dyes in DSSCs today, much work is being done currently to produce porphyrin-based, ruthenium-free organometallic dyes. The Grätzel lab recently developed a porphyrin dye called YD-2 that achieved an 11% power conversion efficiency in a 16μm TiO2 layer, the highest efficiency demonstrated by a ruthenium-free sensitizer. The structure of YD-2 is shown in figure 5.
Moreover, chlorophyll itself is relatively simple to modify, which allows researchers to create highly biomimetic devices while manipulating properties such as excited-state lifetime and LUMO energy level as well as enhance adsorption to TiO2 (Wang et al., 2010b). Chlorophyll (Chl) comes in several distinct forms, most commonly Chl a, Chl b, and Chl c; the latter two are typically found as accessory pigments in light-harvesting complexes, which serve to absorb additional solar radiation and transport excited electrons to the reaction center, which contains Chl a. (Figure 6a). In an effort to mimic the multimolecular approach used by plants to harvest light, researchers constructed DSSCs with co-sensitized Chl derivatives, which were labeled a-type, b-type, or c-type depending on which Chl they structurally resembled. Combinations of a-type Chl sensitizers with c-type Chl sensitizers resulted in augmented power conversion efficiencies of 5.4% (Wang et al., 2010a). This group subsequently developed a Chl-a derivative, chlorin-3, with a dodecyl ester group at C17, where a carboxyl group would normally be. DSSCs with chlorin-3 obtained a conversion efficiency of 8% with an AcCN redox electrolyte HTL, the highest efficiency for any chlorophyllous sensitizer to date. (Wang et al., 2010b).
3.3. Fully organic dyes
Another fast-growing method for doing away with ruthenium in sensitizers is to create completely organic sensitizers without the need for a metal center at all. Indoline sensitizers such as D149 show promise for their extremely high molar absorption coefficient of 6.87 x 104 M-1cm-1 at 526 nm. DSSCs with D149 have achieved efficiencies of 9.0% and 6.7% in cells with acetonitrile-based electrolytes and cells with ionic liquid (IL) electrolytes, respectively(Ito et al., 2006). Subsequent work using the indoline sensitizer D205 with IL electrolytes achieved an efficiency of 7.2%, the highest efficiency obtained in DSSCs with organic dyes and an IL HTL (Kuang et al., 2008).
Yum et al have demonstrated efficiencies of 7.43% and respectable absorbances in the red/near IR region using cells cosensitized with JK2 and SQ1, both organic, non-porphyrin molecules (Figure 8). JK2 and SQ1 have impressive molar absorption coefficients of 4.2 x 104 M-1cm-1 at 452 nm and 15.9 x 104 M-1cm-1 at 636 nm, respectively (Yum et al., 2007).
Advanced dye engineering has produced superior organic dyes. However, the simplest of these molecules can be found as the colorful dyes in berries and flowers. Anthocyanins and betalains are two classes of biological molecules being investigated for their photoactive properties. Anthocyanins are colorful flavonoids that are found in fruits such as blackberries and raspberries. Though they can be modified - their energy absorption is pH sensitive with acidic solutions turning many anthocyanins into a red dye (Calogero et al., 2009) – their overall conversion efficiencies are quite low. DSSCs sensitized with juice from red Sicilian oranges (Citrus Sinensis) achieved an efficiency of 0.66%(Calogero & Di Marco, 2008). Nevertheless, anthocyanins are easily extracted from plants and are both abundant and diverse. Betalains are found in Caryophyllales plants, which include red turnip, wild purple sicilian prickly pear, and bougainvillea flowers. Betalain pigments include betanine, which has a red-purple coloration, and betaxanthins, which are yellow-orange; both compounds have carboxyl functional groups, allowing them to readily chelate to TiO2. DSSCs sensitized with betalains from red turnip yielded an overall conversion efficiency of 1.75%, a remarkable value for nonsynthetic, biological dyes (Calogero et al., 2009). These biological sensitizers are in the initial stages of development and may be a nontoxic and inexpensive option for DSSCs of the future.
Long-term stability is a factor to consider for all dyes, and the stability of ruthenium complexes is the main reason that they are still the standard sensitizers in high-performance DSSCs. However organic dyes are also gaining the potential to exhibit long-term stability; a sensitizer called D21L6 sustained 1000 hours of light soaking at 60ºC and maintained an efficiency of 90% of the initial value, an unprecedented degree of stability for organic sensitizers. Furthermore, D21L6 has a molar absorption coefficient of 3.7 x 104 M-1cm-1 at 458 nm, superior to ruthenium-based dyes, and have achieved overall conversion efficiencies of 7.25% and 4.44% in cells with AcCN electrolytes and solid-state HTL material, respectively (Yum et al., 2009). This combination of stability and efficiency indicates that D21L6 and other organic dyes may be able to replace ruthenium complexes as the preferred dyes in DSSCs.
The high price and limited availability of ruthenium dyes is one of the main drawbacks to DSSCs. However, new advancements in dye fabrication promise to take the ruthenium out of Grätzel cells and significantly improve their watt-to-cost ratio. Though ruthenium dyes are still the most commonly used dyes, both organic and porphyrin-based organometallic dyes are poised to become the cheap and nontoxic alternatives for future cells.
4. The Electron Transport Layer (ETL)
Sensitizers generate excitons, thus creating the initial charge separation that is vital to the function of any solar cell. However, these excitons are very short lived, and excited electrons tend to recombine with the holes to which they are still coulombically (i.e. electrostatically) bound. Thus, an effective semiconductor electron-transport matrix is essential in a DSSC to efficiently separate charge.
Optically transparent microfilms of anatase TiO2 (and some other metal oxide semiconductors such as ZnO) have been found to effectively transport charge from an excited dye to an electrode. However, charge transport alone is insufficient. A monolayer film of dye adsorbed onto a one-dimensional TiO2 layer yields very low-efficiency solar cells, well below 1% (O’Regan & Grätzel, 1991). This low efficiency is due to the poor absorption properties of dyes in a monolayer – less than 1% of incident light is absorbed by a monolayer of dye on a flat TiO2 surface (O’Regan & Grätzel, 1991).
Though Grätzel cells differed from previous dye-sensitized photoelectrical cells by introducing a more efficient dye, the true innovation that made the first Grätzel cell so significant was the use of a highly mesoporous (pores in the range of 2-50 nm) semiconductor layer. This layer, fabricated by the deposition of TiO2 or another metal oxide semiconductor in a colloidal solution followed by drying and sintering, allows dye to be adsorbed in the pores of the semiconductor, thus vastly increasing the amount of dye adsorbed onto the material. Figure 9 shows scans of a typical anatase mesoporous film.
To give an idea of the impact of such a mesoporous layer, the 10 micrometer thick layer of 15 nm average diameter TiO2 particles used in the original Grätzel cells, packed cubically, is expected to have a roughness factor of 2000. In other words, these films are expected to produce a 2000-fold increase in surface area as opposed to a perfectly flat film. In reality, Grätzel and O’Regan found that this first attempt at a mesoporous TiO2 anode only produced a roughness factor of 780(O’Regan & Grätzel, 1991). This discrepancy was attributed to “necking” – inter-particle fusions that decrease surface area – as well as the fact that some parts of the porous structure were too small for the relatively large dye molecules to penetrate. Regardless of these imperfections, a new method had been demonstrated for extremely high adsorption of dye molecules into a small area while still allowing contact between dye and liquid electrolyte. With this remarkably simple mechanism, which allowed for 7% efficiencies in ambient sunlight, and 46% harvesting of incident photons (O’Regan & Grätzel, 1991), dye-sensitized photoelectrical cells began to show real promise of future commercial viability.
One great advantage to a mesoporous TiO2 structure aside from higher dye packing is its ability to trap light within its pore structure. Instead of photons hitting a flat surface (and either being absorbed or reflecting away), unused photons are scattered by TiO2 nanoparticles until they either escape or are absorbed by a dye molecule. This process repeats, immensely increasing a photon’s chance of interacting with dye. As opposed to the 1% light harvesting efficiency seen on a flat ETL, recent papers have achieved over 95% harvesting of incoming photons (Ghadiri et al., 2010).
Mesoporous TiO2 layers are effective because they allow for very dense dye adsorption, and increase the path length of light inside their porous structure. These factors contribute to a sufficiently high rate of exciton generation upon illumination – in other words, photons have a high probability of reflecting around inside the crystal TiO2 structure until they hit one of the densely adsorbed dye molecules. However once this occurs, the TiO2 must then accept this excited electron from the dye and also transport the excited electron to the anode before it has a chance to recombine with holes in the oxidized dye or in the HTL.
The driving force behind electron injection into the TiO2 from the dye is twofold. Firstly, there is an enthalpic impetus to electron injection from dye LUMO to the TiO2 conduction band. Because the conduction band lies below the dye LUMO in energy, it is energetically favorable for an electron to transport from dye to TiO2 immediately upon excitation (Cahen et al., 2000; Liu et al., 2002). The energy gap ELUMO – ECB between the two materials determines the strength of the enthalpic driving force. For a typical dye such as N3, which has a LUMO energy of -3.74 eV vs. vacuum (Nazeeruddin et al., 1993, Zafer et al., 2005), injecting electrons into mesoporous anatase TiO2 with conduction band edge at -4.1 eV vs vacuum (Cahen et al., 2000, Liu et al., 2002, Tang et al., 1994), this energy difference is ~0.34 eV.
The second impetus to electron injection from dye to TiO2 is entropic. Per unit area, anatase TiO2 has a much higher density of possible electronic states than the dye molecule. Thus, an electron diffusing from dye to TiO2 causes a gain in entropy, much like a gas molecule flowing from a small container to a large container. This entropy gain can result in an additional, substantial driving force of up to ~0.1 eV (Cahen et al., 2000).
The advantage of these two driving forces – entropic and enthalpic favorability – is that they are essentially irreversible. Reverse transport of electrons from the TiO2 conduction band edge to dye LUMO is unlikely (O’Regan & Grätzel, 1991; Cahen et al., 2000). Moreover, since there are essentially no minority charge carriers in TiO2 (i.e. there are no free holes in the TiO2), recombination within the semiconductor is unlikely (O’Regan & Grätzel, 1991). The implication is that rough or broken TiO2 films can still produce perfectly functional Grätzel cells. Limiting recombination is key for a solar cell – as discussed in the introduction, though open-circuit voltage (Voc) is primarily determined by the energy difference between ETL Fermi energy upon illumination and HTL oxidation potential (Nernst potential in the case of a liquid electrolyte redox couple), recombination detracts from this value significantly (Ghadiri et al., 2010, Li et al., 2011, Peng et al., 2004;).
Thus, the majority of recombination in a Grätzel cell is due to interactions between electrons in the TiO2 semiconductor and anode material, and holes in the HTL. For a liquid iodine redox couple, this means that oxidized species of iodine, such as I2 - and I3 -, can absorb electrons in the TiO2 matrix. However, this recombination requires electrons in the TiO2 nanostructure to be within tunneling distance of the HTL (Cahen et al., 2000), or within approximately 3 nm. In a system where particles have an average diameter of 20 nm, this means that only electrons on the surface of TiO2 particles have the possibility of recombining at all. Moreover, this implies that the recombination rate is inversely proportional to particle size, and directly proportional to the specific surface area.
There are other mechanisms that prevent recombination in Grätzel cells. Among the most important of these is ionic screening in a liquid electrolyte cell. Iodine and counterions shield holes in the HTL and electrons in the ETL from their coulombic attraction, thus decreasing the likelihood that these particles will recombine (Cahen et al., 2000).
These and other lesser mechanisms make for a very low rate of recombination in Grätzel cells, especially those employing a liquid electrolyte (O’Regan & Grätzel, 1991, Cahen et al., 2000,Peng 2004). This low recombination rate results in a relatively high fill factor (i.e. the ratio of observed efficiency to theoretical max efficiency). The fill factor for the first DSSC was 0.684 (O’Regan & Grätzel, 1991), and recent cells attain values of up to 0.75 (Ghadiri et al., 2010). To give a comparison, the fill factor for an average commercial silicon cell is ~0.83, while some advanced inorganic thin-film cells have achieved almost 0.90 (Green, 1981; PVEducation.org). Grätzel cells therefore compare very favorably with current commercial cells in this respect.
Thus, four important characteristics have emerged that work to make a TiO2 mesoporous semiconductor layer more effective than a flat electrode. First, the TiO2 nanostructure provides a very high effective surface area for dye molecule adsorption due to its very high roughness factor. Second, the TiO2 provides very favorable and effectively irreversible entropic and enthalpic driving forces for charge injection. Third, low recombination within the semiconductor matrix as well as between the semiconductor and the HTL allows for very efficient electron transport to the anode. Fourth, light scattering within the porous crystalline structure increases the radiation path-length and thus improves photon harvesting. And, because of the porous nature of the ETL, this all happens while simultaneously allowing the HTL to come into direct contact with every single dye molecule just like for a flat electrode (O’Regan & Grätzel, 1991).
4.1. Improving the ETL through new morphologies
These four main requirements have guided efforts to improve ETL function further. If dense adsorption of dye onto TiO2 is responsible for the efficiency of the original Grätzel cells, then increased-surface-area materials and more favorable dye-semiconductor adsorption interactions can make these cells even more effective. Likewise, finding ways to prevent recombination, increase light scattering, and increase the driving force for charge injection will further increase cell performance.
The use of light-scattering semiconductor particles is one of the simplest and easiest ways to improve device efficiency. Ito et al. used larger TiO2 particles on top of a mesoporous TiO2 layer to scatter photons into the semiconductor pores, increasing path-length of the photons and increasing cell efficiency substantially (Ito et al., 2007).
Other methods have emerged to prevent recombination rates in Grätzel cells. One major problem is contact between the hole-rich HTL and the electron-rich photoanode. In solid-state devices, which use a solid HTL, this is an even larger problem. In the original Grätzel cells, liquid electrolyte provided ionic screening to prevent charge recombination. In a solid device, this phenomenon does not occur. Moreover, some solid HTLs cause a short circuit by providing direct electrical contact between anode and cathode, thus rendering the device non-functional (Peng et al., 2004). A simple solution to this problem is the use of a nonporous semiconductor layer in between the anode and the mesoporous semiconductor layer, shown in figure 10. While providing electron transport in the same fashion as the mesoporous layer, this nonporous “blocking” layer prevents HTL material, whether liquid electrolyte or solid material, from directly contacting the anode (Peng et al., 2004, Saito et al, 2004). Work by Peng et al. has shown that a blocking layer of TiO2 can significantly prevent charge recombination at the anode, and that the ideal thickness for a blocking TiO2 layer is between 120 and 200 nm (Peng et al., 2004). Since the porous layer of titania is on the microscale – typically 10 or more micrometers, this study shows that a significant increase of efficiency is possible using only minimal extra amounts of material.
4.2. Semiconductor nanotubes as ETL
One ETL technology that has been gathering interest recently is the use of semiconductor nanotubes. One of the problems with mesoporous films is that cracks and gaps between TiO2 particles in the anatase layer create electron traps (Ghadiri et al., 2010, Li et al., 2011). In other words, gaps between nanoparticles in the mesoporous structure make it difficult for electrons to tunnel from particle to particle, thus retarding electron movement through the semiconductor (Mor et al., 2006). This slowing of electrons gives rise to a higher probability of recombination with holes in the HTL. One-dimensional electron transport via semiconductor nanotubes seeks to remedy this problem. Adachi et al. reported efficiencies of 4.88% using disordered nanotube arrays in 2003 (Adachi et al., 2003). Mor et al. reported the use of highly-ordered TiO2 nanotubes for improved electron transport in 2006. Though their devices exhibited only 2.9% efficiencies, this was for a nanotube layer only 360 nm thick, or under 1/30 the thickness of the original 10 micrometer films used by Grätzel and O’Regan (Mor et al., 2006, O’Regan & Grätzel, 1991). They noted that efficiency scaled linearly with nanotube length, meaning that dye adsorption appears to be the limiting factor. They hypothesized that with several-micrometer-long nanotubes this system has an ideal limit of ~31% efficiency (Mor et al., 2006). SEM scans of nanotube systems are shown in figure 11.
The reason for this high ideal efficiency with nanotubes is due to their simple linear architecture. A nanotube, unlike a mesoporous TiO2 layer, provides very direct and one-dimensional electron transport (Bendall et al., 2011, Ghadiri et al., 2010, Li et al., 2011). The small boundary between TiO2 particles in highly-ordered nanotubes largely bypasses the problem of electron traps caused by gaps in mesoporous structures. Because of this, the average length an electron can travel in these tubes before recombining, known as the diffusion length, is much longer (Ghadiri et al., 2010; Li et al., 2011). It has been estimated by Jennings et al. that the diffusion length in a nanotube cell is approximately 100 micrometers (Jennings et al., 2008). As many nanotube layers in this kind of study have only reached 20 micrometers or less, there is great potential for scaling up tube size, and thus effective surface area, without significantly increasing the rate of recombination.
Additionally, nanotubes are more “open” than mesoporous layers. This means that oxidized species in the liquid electrolyte can quickly escape from the tube structure as opposed to being trapped in the pore structure. Once the electrolyte accepts a hole from the dye, it can quickly move to the cathode and be reduced, and thus has less chance of recombining with electrons in the ETL (Ghadiri et al., 2010, Li et al., 2011).
One of the downsides to carbon nanotubes is their high manufacturing cost. They usually require much more energy input than mesoporous layers, and are more time-consuming to fabricate (Bendall et al., 2011, Li et al., 2011). Thus, efforts are being made to decrease cost of these cells while increasing efficiency.
Hollow TiO2 fibers were recently fabricated by Ghadiri et al. through a nanotemplating method, shown in figure 12. By creating cellulose nanofibers, coating them in TiO2, and then removing the fibers via heat treatment, this group produced hollow TiO2 fibers similar to nanotubes via a very cheap and effective method(Ghadiri et al., 2010), outlined in figure 13. Moreover, these structures were shown to have 2-3 times faster electron transport than mesoporous films, higher voltage due to lower recombination (despite nanofiber energy actually being closer to electrolyte Nernst potential than mesoporous films), and high efficiencies. In fact, cells made with fibers only 20 micrometers long exhibited 7.2% efficiencies, nearly identical to similar devices made with TiO2 nanoparticles(Ghadiri et al., 2010). There are downsides to this technology, including lower roughness factors and less light trapping than mesoporous TiO2. These problems can be mitigated by more efficient dyes (thus reducing the need for high-density adsorption), and decreasing fiber diameter to increase surface area. Nevertheless, despite these drawbacks nanotube cells performed similarly to mesoporous ETL cells. Thus, this technology has great potential for improvement and has promise to play a role in future generations of DSSCs.
Nanowires are a variation on the theme of highly-ordered nanostructures for the ETL. ZnO nanowire and nanotube fabrication require much milder conditions than for TiO2 nanowires and nanotubes, and thus are an attractive option for their low cost of manufacture(Bendall et al., 2011). ZnO also has improved electron mobility as compared to TiO2, which would lead to potentially longer diffusion length. Generally, ZnO exhibits lower affinity for dye adsorption, and tubes made from ZnO are less stable. However, Bendall et al. reported 2.53% efficiency using a mix of ZnO and TiO2 nanowires, and reported effective dye adsorption with use of a novel organic dye(Bendall et al., 2011). Thus, ZnO remains a potentially attractive alternative to TiO2 as the ETL material.
The ETL of a Grätzel cell, though usually made of just one material, is as vital as it is simple. Superior surface area allows for hundreds of times more dye to adsorb to surfaces than a flat ETL. Light trapping increases the likelihood that photons will interact with dyes. Low recombination rates due to unfavorable back-reactions combined with protective surface morphologies result in very high fill factors. And while performing all these tasks, anatase and ZnO semiconductor matrices allow light and HTL materials to contact the dye fairly unhindered. New developments in nanotubes and nanowires take these properties to the next level. They allow ultrafast, one-dimensional electron transport from dye to anode while still providing very efficient dye packing and easy access to HTL materials. New breakthroughs in low-cost nanotubes, higher-efficiency mesoporous layers, and nanowires, along with alternative ETL materials such as ZnO, continue to raise the efficiency of Grätzel cells, lower their cost, and further elucidate their inner workings.
5. The Hole Transport Layer (HTL)
The HTL in a DSSC plays a critical role in determining the device’s efficiency, closed-circuit voltage, photocurrent and physical architecture. An effective HTL must be able to transfer charge rapidly so as to be able to regenerate the dye quickly and mitigate losses due to charge recombination between oxidized dyes and injected electrons (Hao et al., 2011). In addition, the material must be stable over long periods of time and elevated temperatures in the cell’s operating range, around 80-85 C (Wang et al., 2003). Though the most common and efficient HTL materials to this date are liquid electrolytes, a wealth of new and innovative research is being done on a wide variety of materials to make hole transport cheaper, more effective, and more robust (Helgesen et al., 2009).
5.1. Electrolyte redox couples as HTL
In the original DSSC design described in Grätzel’s seminal 1991 publication, the cell utilized an I-/I3 - redox couple composed of tetrapropylammonium iodide, iodine, ethylene carbonate and acetonitrile. While the iodine is the active component of the redox shuttle, the other components are necessary to create a liquid electrolyte. Using this formulation, even without any of the significant improvements that have been developed in recent years, the cell exhibited an impressive efficiency in the range of 7-8% (O’Regan & Grätzel, 1991). This relatively high efficiency is due to the excellent mobility of the liquid electrolyte and its ability to make contact with, and rapidly regenerate, the dye inside the ETL. In a follow-up paper, Grätzel also demonstrated the significance of the electrolyte composition by showing that the I-/I3 - counterion (e.g. Li+, tetrapropylammonium) had a significant effect on the photocurrent of the system (Kay & Grätzel, 1996) due to the improved ionic screening provided by these counterions. Among the iodide salts that form room-temperature ionic liquids, 1-propyl-3-methylimidazolium iodide (PMII) has the lowest viscosity, making it the most effective and widely used liquid electrolyte HTL (vide infra) (Bai et al., 2008).
The role of iodide in the electrolyte is to regenerate oxidized dye according to the following reduction reactions:
The resulting triiodide must diffuse away to the cathode where it is reduced back to iodide according to the equation
Thus, the diffusion coefficients of iodine and triiodide are very important in determining the rate at which the electrolyte can regenerate the dye and prevent recombination (Hao et al., 2011). The diffusion coefficients are a function of physical diffusion and exchange reaction constants. Physical diffusion is dictated by viscosity, which increases with electrolyte concentration due to ionic and van Der Waals interactions. A more viscous solution inhibits I-/I3 - diffusion, which reduces its ability to regenerate the dye. However, a high electrolyte concentration is necessary to yield sufficient ionic conductivity and enhances the rate of exchange reactions, so there is some optimal electrolyte concentration that balances these factors. Using a binary electrolyte system consisting of 1-ethyl-3-methylimidazolium dicyanamide (EMIDCA) / PMII with a fixed iodine concentration and a varying EMIDCA volume fraction, Lin et. al. concluded that a volume fraction of 40 vol% EMIDCA produced the maximal balance between physical diffusion and exchange reactions, leading to the optimal electrolyte composition. Indeed, when fabricated into a working cell, this electrolyte led to the highest efficiency of 5.20% (Figure 13) (Hao et al., 2011).
Since 1991, numerous liquid electrolytes have been tested in DSSCs (often by varying the counterion), with the most efficient yielding cells with about 11% efficiency(W. Peng et al., 2003, Hao et al., 2011). However, Grätzel noted that the presence of a liquid solvent for the electrolyte requires “hermetic” sealing of the module in order to prevent evaporation of the volatile solvent (Kay et al., 1996). Such processes are difficult and costly to perform, and mitigate the ease of fabrication that makes DSSCs an attractive alternative to silicon solar cells. The necessity to create a vacuum within the cell and backfill with liquid electrolyte requires slow heating and cooling in order to avoid cracking the glass, and large modules would require significant mechanical strength in order to hold up against bending under wind load. It has also been shown that the I-/I3 - redox couple is highly reactive and corrosive with the other components of the DSSCs, especially the sealing material (I. Lee et al., 2010). Finally, the disproportionation reaction of I2 - has a redox energy that is much higher than the reaction between dye and iodide – this reaction lies closer in energy to the TiO2 conduction band edge. Thus, the energy difference between HTL and ETL is made smaller by this reaction, and it in fact accounts for an approximately 0.75V loss in maximum photovoltage (Boschloo et al., 2009).
5.2. Ionic liquids as HTL
To address these problems, non-volatile, room-temperature ionic liquid electrolytes have been investigated as hole transporters in DSSCs. These electrolytes have virtually zero vapor pressure in the temperature range experienced inside the cell, although sealing is still required to contain the liquid (Welton et al., 1999). Although these non-volatile liquid electrolytes obviate the need for hermetic sealing, they are by nature very viscous and incapable of achieving the rapid physical diffusion rates of the solvent-based liquid electrolytes. Using 1-alkyl-3-methylimidazolium iodide as a room-temperature ionic liquid electrolyte (ILE), Yanagida et. al. were able to achieve a device efficiency of 5.0%, as compared to 7.9% when using the redox couple and organic liquids (OGEs). In order to achieve a quasi-solid state device, which is more robust and resistant toleakage over time, they added a gelator to form an ionic gel electrolyte (IGE) (Figure 14), and they observed the same device efficiency of 5.0%. Furthermore, they found that at high temperatures, the IGE cells outperformed the OGEs, due to the thermal stability of the ionic gel electrolyte (Kubo et al., 2003). Finally, it was determined that the viscosity of the liquid could be manipulated by varying the alkyl chain length. Longer chains lowered the viscosity of the solution (by interrupting ionic interactions) and led to the most efficient device. Along similar lines, Zakeeruddin et. al. used a mixture of carbon black and PMII-based ionic liquid and obtained a device efficiency of 6.37% (Lei et al., 2010). Lianos et. al. cleverly designed a sol-gel process whereby 1-methylimidazole is reacted with 3-iodopropyltrimethoxysilane to form the alkoxysilane derivitized electrolyte and I-, giving an almost solid state electrolyte (Figure 15). By virtue of its solidity, albeit desirable, this electrolyte resulted in a mediocre device efficiency of only 3.2% (Jovanovski et al. 2006).
5.3. Quasi-solid state HTL
Besides adding gelators to ionic liquids, it is possible to create quasi-solid state electrolytes by incorporating polymers into the electrolyte in a number of creative ways. One such method is to use a polymer to form a host matrix that traps the liquid electrolyte. These devices maintain the high ionic conductivities of liquid electrolytes, but protect against leakage and physical stress from thermal expansion. This concept was pioneered in 1995 by Searson et. al., who simply polymerized polyacrylonitrile in a mixture containing ethylene carbonate, propylene carbonate, acetonitrile and the desired concentration of NaI. The resulting gel was applied to the counter-electrode of a DSSC and efficiencies of 3-5% were achieved (Cao et al., 1995). Since then and continuing today, a variety of polymers and electrolyte combinations have been explored, with reported efficiencies ranging from 2% to 8.2% (Bai et al., 2008, Jin et al., 2010). This remarkably high efficiency was achieved by Grätzel et. al., who reasoned that a eutectic melt of different polymers would have a lower viscosity than a binary melt or other simple formulations. Yet, despite this result, more recently published papers continue to report much lower efficiencies for polymer-based quasi-solid state devices, suggesting that there is a relatively poor understanding of the factors that lead to a highly effective electrolyte. Continued research in this field looks to be promising, and may eventually lead to devices that surpass the efficiency of liquid electrolyte systems at operating temperatures.
5.4. Solid state HTL
Even in quasi-solid-state devices, sealing is necessary to contain the liquid electrolyte. In order to completely preclude this necessity, it is possible to create solid-state DSSCs (ss-DSSCs) with completely solid-state materials, namely small molecules and conjugated polymers. Small molecules can be inorganic, such as the copper salts CuI and CuSCN, or entirely organic, such as SpiroMeOTAD (2,2’,7,7’-tetrakis(N,N-di-p-methoxyphenyl-amine)9,9’-spirobifluorene) (Bach et al., 1998).Solutions of these molecules are given time to permeate the TiO2, and then the solvent is evaporated, leaving only the solute molecules. However, device efficiencies with these materials are generally poor – devices using SpiroMeOTAD hover around 5% efficiency (Ding et al., 2009). Although hole transport in ss-DSSCs is not fully understood, it is postulated that these materials function poorly because they transport charge electronically, as opposed to ionically. Studies suggest that the ionic nature of liquid electrolytes is critical to their high efficiency (Nogueira et al., 2001).
Polyelectrolytes are a relatively recent innovation, in principle able to exploit both the efficiency advantages of ionic conduction and the stability of solid-state materials. The polyelectrolytes used are typically non-conjugated polymers containing covalently bound charged groups. Like other solid-state devices, they still suffer from diminished efficiencies because charge transport, while ionic, relies entirely on exchange reactions and not diffusion. Unfortunately, polymers are large and often unable to penetrate the TiO2 matrix and contact the dye, so they must often be synthesized in situ for contact with the dye, and afterwards there is little or no further diffusion into the pore spaces. In 2001, De Paoli et. al. introduced the first solid state device using polyelectrolytes, achieving a maximum efficiency of only 2.6% under low-light conditions (Nogueira et al., 2001). Since then, despite numerous attempts, the efficiency of such devices has generally remained low. In 2010, Jin et. al. achieved a device efficiency of only 2.42% (Jin et al., 2010). These materials, however, still provide much promise for future improvement. Their performance is limited primarily by the decreased diffusion rates of the electrolytes they contain. With the discovery of new, more intensely absorbing dyes, the required thickness of DSSCs will decrease, mitigating the significance of electrolyte viscosity (Kubo et al., 2003). As this happens, quasi-solid state electrolytes may become increasingly competitive with liquid electrolytes.
Conjugated polymers are a widely-used material in solid-state hole transport devices, though like all solid-state devices they remain fairly inefficient compared to liquid electrolyte cells. However, they hold promise for being stable, efficient solid-state hole transport-devices. These materials, which are characterized by a long, delocalized pi-bond system along their carbon backbone, are able to transport holes through their valence band. Many polymers have emerged, most notably polythiophenes such as Poly(3,4-ethylenedioxythiophene) poly(styrenesulfonate) (PEDOT:PSS). Recent work with graphene / PEDOT:PSS - coated cathodes has yielded cells with 4.5% efficiencies (Hong et al., 2008). As well, these materials are often able to catalyze hole transport in liquid devices by providing an extra electronic step in between liquid electrolyte redox potential and cathode. In 2009, Cheng et. al. fabricated a counter-electrode from PEDOT:PSS,which catalyzed the regeneration of iodide at the counter-electrode and achieved an efficiency of 5.25% (Sirimanne et al., 2010).
Though still a relatively low-efficiency component in DSSCs compared to liquid electrolytes, conjugated polymers provide a wealth of opportunities for improvement and experimentation. Work has been done in the Johal lab at Pomona College on the tuning of photoactive conjugated polyelectrolytes. These materials are adept at transferring holes through their HOMO, but are also excited in the visible range, and have charged side-groups that allow them to be tuned via surfactants. Work by Treger et al has shown that adding surfactant lowers the optical band gap of these polymers (presumably by increasing the effective conjugation length), improves internal quantum efficiency, and decreases photooxidation and degradation (Treger et al, 2008a, 2008b).
Recent work by Polizzotti, DeJong, Schual-Berke, and Johal has attempted to examine the effects on efficiency of adding the conjugated polyelectrolyte Poly[5-methoxy-2-(3-sulfopropoxy)-1,4-phenylenevinylene] (MPS-PPV) to a liquid electrolyte-based cell. Though a marked increase in current and cell performance is observed (especially upon the addition of surfactant), voltammetric and spectroscopic analyses of these materials show hole transport from iodine redox couple to polymer to be forbidden (see figure 16a). To rectify this discrepancy, a mechanism in which the photoactive polyelectrolyte is generating a second exciton to facilitate hole transport is proposed and shown in figure 16b. As of yet, little to no work has been done on the subject of dual exciton generation in DSSCs. It is hoped, however, that this study will inspire further investigation into this matter, and prove to be yet another way of tuning and optimizing hole transport in DSSCs.
Liquid electrolytes remain the most efficient HTL to date. However, from a mechanical standpoint, solid-state electrolytes are the ideal option for DSSCs due to their improved stability. The main drawback of these materials is that they have yet to prove their viability as effective charge transporters, and a wealth of new research is based on creation of an optimal solid-state DSSC. Conjugated polymers, small-molecule solid HTLs, and others all show promise to provide the next breakthrough in ss-DSSCs, while quasi-solid-state gel electrolyte systems combine the benefits of polymeric materials and liquid electrolytes, also provide promise of high-efficiency, long-lived DSSCs. It is unclear which type of HTL will become the dominant technology, but it will surely have a drastic impact on the efficiency of DSSCs.
Grätzel cells are poised to become an important player in photovoltaic energy generation. They have already achieved over 11% efficiencies in prototype champion cells. Their watt-to-cost ratio promises to be very favorable, especially with the recent introduction of ruthenium-free dyes. As advances are made in more effective semiconductor morphologies, lower-cost, higher-efficiency dyes, and more effective and long-lasting hole conduction materials, these devices grow ever closer to commercial viability. However, there is still work to be done before these cells can start producing cheap, clean energy on a larger scale. In this chapter we described the functions and challenges for each of the components of a Grätzel cell, and outlined research needed for new and innovative ways of separating and transporting charges within these devices. By having a full understanding of the drawbacks facing DSSCs and the ways that current research is attempting to address them, can future experimenters be poised to discover the next groundbreaking solutions.
The authors would like to thank the Pomona College chemistry department for resources to conduct our research as well as their unending support. They would also like to thank the other members of the Johal lab for advice and assistance in and out of the lab.