Advances in Nanocrystalline Semiconductor Solar Cells

Advances in Nanocrystalline Semiconductor Solar Cells Graeme Williams* Organic Optoelectronic Materials & Devices Laboratory Electrical and Computer Engineering University of Waterloo. Waterloo, ON Canada. N2L 3G1. *Correspondence should be addressed to g3willia@uwaterloo.ca

Abstract. Silicon photovoltaics have long been considered a feasible approach to meet the inevitable energy demands imposed by increased worldwide energy use and limited oil reserves. However, their cost of manufacturing and installation has slowed their widespread implementation. Nanocrystalline or quantum dot solar cells have since been identified as a potential high-efficiency, cost-effective alternative to traditional single crystal silicon p-i-n solar cells. In this regard, quantum dots have been successfully applied as strongly absorbing materials in various cell configurations, including both liquid-junction and solid-state solar cells. The unique properties of quantum dots allow for tuning of their optical absorption based on their size, potentially granting better solar spectrum matching and infrared-absorption capabilities. Furthermore, quantum dots have been shown to have extended hot carrier lifetimes and to be capable of multiple exciton generation as a consequence of their quantum confinement properties, which could allow for greatly enhanced solar cell efficiencies. This review serves to examine the experimental work of quantum dot solar cells to date, with a focus on solar cell architectures for efficient carrier extraction. From this base of knowledge, the review will discuss the inherent limitations of the current research, with suggestions for future areas of experimentation and study. 1

Introduction. From a simple evaluation of energy production versus energy consumption, it has been established that an alternative energy source to fossil fuels will be required by the end of this century [1]. Silicon photovoltaics have been considered a serious contender in this regard; however, they are economically limited by their high manufacturing costs and technologically limited by their absorption matching with the solar spectrum. For the purpose of comparison, the most pertinent solar cell parameter is the power conversion efficiency value, which refers to the amount of electrical power obtained from a solar cell divided by the total input optical power from the sun or an equivalent light source. Single crystal silicon solar cells generally achieve power conversion efficiency values of ~15%-19.3% commercially [1-3], with experimental cells developed in lab environments capable of 24.7% to 25% efficiency [4, 5]. Recent advances are primarily attributed to improvements in light in-coupling by optimizing the silicon surface geometry and have allowed the cells to near the ~31% efficiency upper limit (as defined by Shockley and Quessier in 1961 [6]). In order to address the need for cheaper solar cells, second generation solar cells have been developed. These solar cells include more cost-effective materials, such as thin films of polycrystalline and amorphous semiconductor materials, as well as solution-deposited organic and dye-sensitized films. However, in order to become competitive with silicon technology, it is necessary that these solar cells first improve their efficiencies. Perhaps the most competitive technology in this regime makes use of copper indium gallium diselenide composites that allow for lab-scale efficiencies up to 19.9% [7]. Unfortunately, attaining this level of efficiency 2

requires significant bandgap engineering through fine-tuned alloying, which sharply increases the cost of the cell. Further, integration into larger panels (as large as 90x60 cm 2 ) drops the efficiency value significantly, with reported efficiencies of 13.1% [8]. Alternative polycrystalline CdTe-based solar modules have also proven to be reasonably successful, with panel conversion efficiencies of 11.1% [3, 9]. Third generation cells based on nanocrystalline (quantum dot) materials are now being examined to attain both low cost and high efficiency solar cells. The size-tunable absorption characteristics of quantum dots are a large driving force for this research because they allow for better matching with the solar spectrum compared to single crystal silicon. For simple CdSe quantum dot solar cells, one might consider separating the cell into various size regimes to develop a rainbow solar cell, which has the capacity to absorb efficiently at all visible wavelengths [10]. Beyond matching the visible spectrum, PbS and PbSe quantum dots offer the capacity to absorb infrared (IR) photons [11], an impossible task with silicon solar cells due to silicon s bandgap of 1.1eV and corresponding absorption cut-off wavelength of ~1.13 m. Since the IR wavelengths account for ~25% of the energy of the sun s incident photons [12], IR quantum dot solar cells could allow for significant gains in solar cell efficiency. The carrier confinement property of quantum dots leads to additional desirable properties such as extended hot carrier lifetimes and multiple exciton generation. A hot carrier is generated when a quantum dot absorbs light with energy greater than its bandgap. In the case of an electron, light absorption promotes the electron to a vibrational state well above the edge of the conduction band. In a bulk single crystal semiconductor solar cell, hot carriers are 3

generated and then rapidly undergo thermal relaxation to the band edges by interactions with phonons. However, it has been shown that coupling a quantum dot to either a hole or electron accepting material can extend the lifetimes of hot carriers into the few to tens of picoseconds [13-16]. After undergoing fast electron or hole transport to the accepting material, the remaining carrier in the quantum dot is spatially confined and is therefore less likely to interact with any species that may result in the loss of energy (including phonons, opposite polarity carriers, etc.). If these hot carriers are collected before thermalization, this property may allow for very high efficiency solar cells. Alternatively, if a hot carrier is able to obtain thermal energy greater than its bandgap, it may instead undergo impact ionization, producing multiple excitons per incident photon [17, 18]. From the above discussion, it is clear that quantum dot solar cells have the potential to address many of the issues present in current technology photovoltaics. Unfortunately, as will be discussed throughout the body of this review, the most critical advantages of quantum dots have yet to be used to achieve the goal of cheap, high efficiency solar cells. This review article addresses the prevalent quantum dot solar cell configurations in literature. The most pertinent configurations leading to the realization of cheap, competitive photovoltaics are examined at depth. Very recent and promising work on the depleted planar heterojunction structure has allowed for power conversion efficiencies on the order of ~5%, which is a significant improvement over previous devices [19]. However, in spite of this outstanding leap in device efficiency, there remains an underlying need to develop novel device architectures to take full advantage of the unique properties of quantum dots for their use in photovoltaics. 4

Background. Quantum dots are defined as semiconductor particles where the exciton Bohr radius is smaller than the radius of the particle itself, leading to carrier confinement in all three spatial dimensions. The small size of quantum dots implies a smaller number of atoms contributing to molecular orbitals in the formation of energy bands. As a consequence, the bandgap energy and position are a function of the size of the dot, leading to size-tunable absorption and emission characteristics. Furthermore, in very small quantum dots, the conduction bands and valence bands are discontinuous and electrons occupy discrete, quantized states, analogous to the particle in a box model. This is shown in Figure 1 below. Figure 1 - Illustration of bandgap variation for bulk vs. nanocrystalline/quantum dot semiconductors. Also shown is the associated quantization of energy states for nanocrystalline semiconductors. (Adapted from Ref [20]). The existence of nanocrystalline materials and films produced by both physical and chemical vapour deposition methods has been known for decades they may be most simply considered polycrystalline materials with nanoscale grains. Today, semiconductor quantum dots fabricated through high vacuum thin film deposition techniques are primarily applied to very high efficiency intermediate band solar cells based on III-V semiconductors [21]. The true usefulness of quantum dots, however, only became apparent after their proven facile, solutionbased synthesis. The development of the reliable synthesis of quantum dots is largely thanks to 5

the work of Bawendi and coworkers, Alivisatos and coworkers and subsequent developments by Peng & Peng (who suggested use of the much safer CdO precursor) [22-24]. In a colloidal solution, quantum dots are capable of being spincoated, spraycoated, dropcast and dipcoated to form very thin (tens of nanometres) quantum dot/nanocrystalline films. Furthermore, as a solution-based material, they may be feasibly applied to large-scale, flexible manufacturing efforts such as reel-to-reel processing. In addition to their promise of cheaper manufacturing costs, third generation solar cells have been researched with the hope of improved efficiency. Efficiency values are typically reported in terms of a standard exposure of 100mW/cm 2 (one-sun) with an air-mass 1.5 (AM1.5) solar spectrum. This spectrum is analogous to light impingent on the earth after it has traveled through the earth s atmosphere a distance of 1.5 atmosphere thicknesses. Solar cells may also be described in terms of several key parameters, including the open circuit voltage (V oc ), short circuit current (I sc ) and fill factor. While under illumination, the open circuit voltage is the voltage across the cell when the cell current is zero and the short circuit current is the current across the cell when the cell voltage is zero. The fill factor may then be defined as:, where V m is the maximum power point voltage and I m is the maximum power point current. As such, the fill factor is a measure of the closeness to an ideal solar cell, which would have a rectangular shape in the fourth quadrant of an I-V output graph for a given light exposure. The power conversion efficiency may then be determined as: 6

, where P light-source is the incident light source power measured in Watts. The external quantum efficiency (EQE) value is also of significance, as it measures the number of carriers collected per number of photons directed to the solar cell at a given wavelength of interest. This value is frequently referred to as the incident-photon-to-carrier-efficiency (IPCE), and may be found as:, where I SC ( ) is the wavelength-dependent short-circuit photocurrent density in A/cm 2, is the exposure light wavelength in nm and P( ) is the wavelength-dependent light intensity in W/cm 2. Early implementations of quantum dots into photovoltaic applications were unsuccessful due to their high level of carrier confinement. For example, one may consider a simple CdSe quantum dot film deposited on a transparent, tin-doped indium oxide (ITO) electrode and integrated into a simple liquid-junction solar cell. Under light exposure, an electron is promoted to a higher state within the quantum dot; however, since the exciton Bohr radius is larger than the radius of the particle, strong attractive Coulombic forces prevent the exciton from breaking apart into an electron and hole. In this simple implementation, there is minimal driving force to separate the exciton into its constituents and therefore no photocurrent may be generated. Indeed, it is far more probable for the exciton to undergo direct recombination and emit a photon through photoluminescence. 7

One of the major breakthroughs for these third generation photovoltaics was the coupling of quantum dots with both organic and large bandgap inorganic semiconductor materials. In this manner, the excited state of the quantum dot is very quickly deactivated by a fast electron or hole transfer to the adjacent material, as shown in Figure 2. Gerischer and Lüebke may be considered the pioneers in this regard, as they were the first to witness TiO 2 sensitized by colloidal CdS [25]; however, Grätzel and coworkers work on dye-sensitized nanostructured TiO 2 films undoubtedly aided in the relative success of quantum dots when applied in this fashion [26]. Figure 2 - Illustration of the excited state deactivation mechanism for a CdS quantum dot through fast electron transfer to a TiO 2 semiconductor particle. (Adapted from Ref [27]). Efforts are now underway to improve upon the quantum dot-semiconductor composite film. Early work in this area, even prior to the Grätzel solar cell [26], focused on understanding the fast electron transfer from the sensitizer quantum dot to the wide bandgap colloidal semiconductor [28-30]. While the individual components of this system are all very promising, the earliest recorded power conversion efficiencies for liquid-junction solar cells based on these materials were quite low [29, 31] 1. These low efficiencies have been attributed to both band 1 As a warning when examining the efficiency values listed in these papers: neither reference uses a standard onesun, AM1.5 illumination source, and Vogel and coworkers use a single wavelength (monochromatic) light source that artificially inflates the reported power conversion value. 8

mismatch and, more importantly, high levels of carrier recombination due to the large number of interfaces. The interfaces generally have a higher number of deep trap states and therefore act as efficient recombination centres. As a consequence, the most crucial area of research in quantum dot solar cells pertains to the need for device architectures that can reduce interfacial recombination and drive the solar cell efficiencies to values on par with current technology. In order to surpass first generation silicon technology, such architectures must also allow for either hot carrier extraction or multiple exciton generation, as will be noted throughout this review. This variation in experimental method when measuring and reporting efficiency values is a common problem in photovoltaics research. However, in most recent research, researchers strive to make use of the standard onesun, AM1.5 power conversion values to make comparisons across literature simpler and more relevant. 9

Literature Review and Analysis. Early work in quantum dot solar cells can be broken down into three general configurations, as detailed by Nozik in his 2002 review, Quantum Dot Solar Cells [16]. In the past three years, an additional fourth configuration has also been identified, which shows high power conversion efficiencies and the capacity to absorb light in the IR region. These four configurations are examined below with a focus on the most pertinent devices that show promise in competing with current solar cell technologies. The extended hot carrier lifetime and multiple exciton generation effects will also be very briefly discussed in terms of the experimental results that have been achieved to date. First Configuration: Vacuum-Deposited Solid State Quantum Dot Solar Cells The first configuration makes use of vacuum deposition techniques to form quantum dots in solid state films, primarily to act as interband dopants [21]. Alternatively, it has been suggested that a quantum dot array be placed in the intrinsic region of a p+-i-n+ structure to potentially make use of hot carrier and multiple exciton generation effects [16]. These solar cells typically use expensive, epitaxially grown III-V semiconductors and are studied for high efficiency and extraterrestrial applications that do not directly address the goal of cheap photovoltaics. The introduction of solar concentrators to decrease costs further complicates their use, as the optics used in concentrators can also be costly. The economics of such a system are beyond the scope of this review. As such, this category of quantum dot solar cells will not be discussed at depth throughout this review the reader is encouraged to look at review articles [16, 21] for further information on these topics. 10

Second Configuration: Liquid-Junction Semiconductor-Quantum Dot Solar Cells The second configuration makes use of quantum dots coupled to either wide bandgap inorganic semiconductors or carbon nanostructures within a liquid junction cell. In this configuration, the quantum dot-sensitized film is deposited from solution onto a transparent electrode (typically ITO) defined as the working electrode. Opposite to the working electrode is a high surface area counter electrode. Excitons are generated within the quantum dot and then separated into holes and electrons at the quantum dot-semiconductor interface. Electrons are then collected on the working electrode while holes are scavenged by the electrolyte solution and transported to the counter electrode to complete the circuit. This experimental setup is shown in Figure 3 below. Figure 3 - Illustration of a liquid junction quantum dot-sensitized solar cell A significant amount of research has been dedicated to novel architectures based on nanomaterials in order to improve the efficiency of liquid junction cells. The reader is encouraged to examine references [2, 32] by Kamat and coworkers for detailed reviews on the subject. Both semiconducting and carbon-based 1-D and 2-D nanostructures have been examined to facilitate carrier transport with some level of success. However, efficiency values 11

have remained relatively low as a consequence of the high level of interfacial carrier backtransfer and recombination [32]. It is also worthwhile to critically examine both the practicality and stability of solution-based solar cells, especially in large-scale solar harvesting efforts. The presence of a liquid species in the solar cell greatly complicates manufacturing efforts. Leakage-free encapsulation and increased toxicity effects due to the use of a liquid electrolyte become serious areas of concern that will inevitably drive up solar cell costs. Also, similar to liquid electrolyte batteries, the stability and corrosion-resistant properties of electrode surfaces is of great concern for solar cell lifetime. The use of chemically inert counter electrodes, such as platinum gauze electrodes, partially addresses this concern, but would greatly increase the cost of the solar cell. The quantum dot-coated working electrode would inevitably suffer from degradation as the electrolyte dissolves the active layer over time. Solar heating of the electrolyte during operation will also cause variation in device behaviour, the effects of which have not yet been studied. For these reasons, it is unlikely that this configuration of solar cell will become competitive with current technologies in the near future. As such, this configuration will also not be examined at depth in this review. Third Configuration: Organic-Quantum Dot Hybrid Solar Cells The third configuration of quantum dot solar cells, termed hybrid solar cells, makes use of quantum dots coupled to an organic semiconducting material. By offsetting the energy bands and energy levels of the quantum dot and the organic semiconductor, it is feasible to break apart photogenerated excitons in the quantum dot through fast hole transfer from the quantum dot to the neighbouring polymer matrix, where the hole may be transported and 12

collected at the appropriate electrode. The remaining electron diffuses along the nanocrystalline phase to the opposite electrode, as dictated by the multiple trapping model typically used with nanostructured semiconductor films [33, 34] 2. Alternatively, a photogenerated exciton in the organic phase may undergo fast electron transfer to the quantum dot with the same effect. The conduction of holes in the organic phase and electrons in the nanocrystalline phase is preferable, as most polymeric semiconducting materials have higher hole mobility than electron mobility. In 1996, Alivisatos and coworkers developed preliminary solar cells based on CdS and CdSe coupled to poly(2-methoxy,5-(2 -ethyl-hexyloxy)-p-phenylenevinylene) (MEH-PPV) polymer, following nascent results from Wang et. al. that showed efficient electron transfer from CdS quantum dots to a polyvinylcarbazole polymer [35, 36]. In the early results from Wang et. al., the CdS quantum dots were used to merely sensitize the film and it was observed that only hole transport occurred throughout the film. Alivisatos and coworkers increased the relative amount of CdS/CdSe quantum dots in the composite above the percolation threshold in an attempt to form a discrete, separate nanocrystalline phase within the polymer. In this manner, it was shown that efficient bipolar carrier collection could occur, with electron transport occurring in the nanocrystalline phase and hole transport occurring in the organic phase. This model is analogous to the bulk heterojunction organic solar cell, with the high electron affinity polymer replaced with the nanocrystalline phase. 2 The multiple trapping model is reasonably well understood and generally accepted for TiO 2 nanostructured films. This model is very similar to the dispersive transport model observed for amorphous materials. The second reference noted here shows some early experimental data on CdSe QDs and CdSe/CdS core-shell quantum dots that appear to show similar carrier transport behaviour, with film mobilities up to 0.1 cm 2 /(Vs). 13

In this study, both trioctylphosphineoxide (TOPO)-coated quantum dots and naked quantum dots were mixed with the polymer material prior to spincoating. Only CdSe quantum dots were examined for their photovoltaic properties. Both systems exhibited very poor device behaviour, with peak quantum efficiencies of 12% and 0.005% for the naked and TOPO-coated systems respectively (at optimal quantum dot loadings with an excitation wavelength of 514nm). An Air-Mass 1.5, one-sun power conversion efficiency of 0.1% was achieved for the naked CdSe device. The difference in the efficiencies of naked and TOPO-coated quantum dots highlights one of the most critical findings in this report: charge transfer may only occur at the quantum dot-polymer interface. It is therefore necessary to either exchange or remove the quantum dot ligand (which is generally present as a consequence of the synthesis method detailed by Bawendi and coworkers [23]) prior to integration in any photovoltaic system. The authors attribute the overall poor device behaviour to electrons becoming trapped at dead ends within the nanocrystalline phase. A more likely explanation is that the majority of photogenerated carriers recombined at the deep trap states present at the numerous quantum dot-polymer interfaces. Sargent and coworkers later continued this work with relative success using PbS quantum dots coupled with MEH-PPV polymer in a bulk heterojunction [37, 38]. Qi et. al. simultaneously developed PbSe quantum dots coupled with MEH-PPV for photodetector applications [39]. MEH-PPV has a relatively low ionization potential (~4.9eV to 5.1eV), which matches well with the ionization potential of PbS quantum dots (~4.95eV) allowing for favourable hole transport to the organic material [37]. In this work, Sargent and coworkers used PbS quantum dots capped with octylamine ligands, which are shown to have superior 14

performance over the as-produced PbS quantum dots capped with oleic acid ligands [38]. This variation is shown in Figure 4A below, and follows as a consequence of the relative sizes of the carbon chains in each ligand (8 carbon atoms for octylamine and 18 carbon atoms for oleic acid). As with the naked vs. capped CdSe data above, the longer ligand separates the quantum dot from the polymer and therefore hinders exciton dissociation. Naked PbS quantum dots are not feasible in this case as a consequence of their poor uncapped stability. Figure 4 - A. Short (octylamine) vs. long (oleic acid) ligand PbS quantum dot:meh-ppv solar cell photoresponse IV characteristics. Note that only the short ligand sample shows any appreciable photocurrent. B. Photocurrent spectral response vs. absorption for octylamine-capped PbS quantum dot:meh-ppv solar cells (three diameters of quantum dots shown). B-Inset: Extended spectral response & absorption curves of the same (single diameter sample shown). (Adapted from refs [37, 38]). The performance of these organic-pbs/pbse hybrid solar cells were very poor, with peak external quantum efficiencies of 0.0064%, on the same level as the capped CdSe samples in the earlier study by Alivisatos and coworkers [35, 37]. Annealing of the samples improved the peak external quantum efficiency to 0.15% and allowed for a measurable power conversion efficiency of 0.001% at low light exposure powers, which is underwhelming compared to competing technologies [38]. In all samples, increasing the power of the light source decreased the quantum efficiency, which is a clear sign of bimolecular or bipolar recombination. In general, bipolar recombination becomes more prevalent at high light intensities when there are a larger number of photogenerated free carriers. Sargent and coworkers were, however, 15

successful in sensitizing the MEH-PPV polymer to absorb in the IR region, as shown in Figure 4B, where the spectral response matches well with the absorbance of the PbS quantum dots. Günes et. al. had slightly more success with PbS quantum dots coupled to poly (3- hexylthiophene) (P3HT) in a bilayer heterojunction. The authors were able to achieve I sc, V oc, fill factor and one-sun, AM1.5 power conversion efficiency values of 0.3mA/cm 2, 0.35V, 0.35 and 0.04% respectively [40]. In more recent work, Grätzel and coworkers have achieved a much higher power conversion efficiency of 1.46% for a significantly modified version of the PbS quantum dot hybrid cell [41]. In this work, the PbS quantum dots are formed directly on a TiO 2 nanoparticle (NP) layer through a process known as successive ionic adsorption and reaction (SILAR), where a substrate is successively and repeatedly dipped into Pb 2+ and S 2- solutions. The benefit of this approach is that stable PbS quantum dots can be formed without the ligand capping monolayer. Grätzel and coworkers completed the cell using by forming a junction with the organic hole transport material 2,2(,7,7(-tetrakis(N,N-di-p-methoxyphenylamine)-9,9(- spirobifluorene (spiro-ometad), to give the final structure of: ITO/TiO 2 nanoparticle film/pbs quantum dot film/spiro-ometad/gold electrode. The pertinent IV photoresponse and external quantum efficiency characteristics for this structure are shown in Figure 5 below. Both Larramona and coworkers as well as Grätzel and coworkers had similar levels of success applying this structure to CdS-sensitized hybrid solar cells [41, 42]. 16

Figure 5 - IV photoresponse and external quantum efficiency characteristics for a hybrid quantum dot solar cell with the structure: ITO/TiO 2 NP/PbS QD/spiro-OMeTAD/Au. (Adapted from ref [41]) In 2002, Alivisatos and coworkers published results showing that CdSe nanorods instead of CdS quantum dots significantly enhance the properties of quantum dot hybrid solar cells [43]. The improvement in device properties was attributed to three major effects: directed carrier motion through the 1-D structure, enhanced percolation of the nanorods to minimize dead ends (shown in Figure 6 below) and decreased hopping transport of electrons, which may now travel continuously along the length of the nanorods. In this initial work, Alivisatos and coworkers used a 90% by weight solution of 7nm diameter by 60nm length CdSe nanorods in P3HT as the active layer in the bulk heterojunction solar cell. Their final solar cell structure of ITO/PEDOT:PSS (smoothing layer)/p3ht:cdse nanorods/al achieved open circuit voltage, fill factor and one-sun, AM1.5 power conversion values of 0.7V, 0.4 and 1.7% respectively [43]. While this solar cell is a significant improvement over the 0.1% power conversion efficiency in Alivisatos and coworkers 1996 work [35], these values are still far below the levels required to be competitive with alternative technologies. 17

Figure 6 - Illustration of percolation of CdSe quantum dots (C) and CdSe nanorods (D) embedded in P3HT polymer. C was spin-cast from 60 wt % CdSe in P3HT to a thickness of 110nm. D was spin-cast from a 40 wt % CdSe in P3HT solution to a thickness of 100nm. Note that the rods are largely aligned along the substrate. (Adapted from ref [43]). Since this initial work on CdSe nanorods, research on quantum dot hybrid solar cells has radically shifted in focus toward the development of new geometrical shapes to enhance percolation and thereby improve electron transport. This research is not discussed in this review for reasons outlined immediately below. However, the reader may wish to examine references [44, 45] for further details on CdSe nanorod hybrid solar cells, reference [46] for details on CdTe nanorod hybrid solar cells, reference [47] for details on branched CdSe hybrid solar cells and reference [48] for details on hyperbranched nanocrystal hybrid solar cells. From the initial discussion on the characteristics of quantum dots and their relevance in photovoltaics, the desire to alter the geometric shape of the nanocrystalline materials seems counterintuitive. The exciton Bohr radii of CdSe and CdTe are 5.7nm and 7.3nm respectively. In the nanorods and branched structures, at least one dimension is no longer smaller than the exciton Bohr radius and, as a consequence, carriers are no longer confined in three dimensions. It is therefore unlikely that such structures will exhibit the extended hot carrier lifetimes and multiple exciton generation capabilities that are unique in third generation photovoltaics. These unique properties were desired at the beginning of this research effort to allow quantum dot solar cells to achieve higher efficiencies than single crystal silicon solar cells. It is now worth considering the role of quantum dots in photovoltaics, as well as some directly competing technologies. As a third generation technology, quantum dot solar cells appeal to both increased efficiency and lowered manufacturing costs. When used solely as sensitizers or simple absorbing species, quantum dot solar cells may instead be classified as 18

second generation solar cells and are therefore competing with all thin film photovoltaics. For simplicity, let us consider one such competitor, organic solar cells, which are at a similar level of development and research. While quantum dot hybrid solar cells have struggled to reach just under 2% power conversion efficiency, purely organic solar cells have, in a similar period of time, managed to reach 8.13% power conversion efficiency with the likely capacity to reach 10% power conversion efficiency by the end of 2011 [49]. Given the large number of interfaces and the huge amount of carrier trapping & recombination in hybrid solar cells, it seems unlikely that either quantum dots or differently shaped nanocrystals, when used only as simple absorbing species, will ever compete with purely organic solar cells. Furthermore, organic solar cells have maintained their relative ease of fabrication and do not require processing of toxic cadmium and lead, as is the case with quantum dot hybrid solar cells. In order to make a real impact in the realm of hybrid solar cells, research efforts would be better spent addressing the initial goals of third generation photovoltaics: developing materials and structures to achieve hot carrier extraction or efficient multiple exciton generation in an economically viable manner. Fourth Configuration: Schottky-Quantum Dot & Depleted Heterojunction Solar Cells In both the second and third configurations of quantum dot solar cells, a substantial amount of research has been dedicated to the development of more complex nanostructures and systems. These systems generally have more interfacial regions in which carrier trapping and recombination can occur. Furthermore, by increasing the complexity of device fabrication, it is difficult to foresee the development of feasible large-scale manufacturing efforts. In 19

contrast, the fourth configuration greatly simplifies the solar cell architecture and yet still manages to achieve power conversion efficiencies far superior to the quantum dot solar cells examined previously. This configuration was first developed by Sargent and coworkers in 2007 and initially relied on a simple ITO/PbS quantum dot/al structure. The quantum dots form a Schottky contact with the Al electrode to create a depletion layer to aid with exciton dissociation [50-52]. It has been previously established that nanocrystals generally do not exhibit significant band bending [32, 53, 54]. In liquid-junction cells, this is further emphasized by the fact that the electrolyte scavenges excess charges to hinder space charge layer formation [32, 55]. This scavenging effect may be expected to occur to some degree with the organic species in organicquantum dot hybrid solar cells. In these studies, single nanocrystals are treated as individual systems with negligible electronic wave function overlap with neighbouring nanocrystals. However, for very densely packed thin films of pure nanocrystalline material, such as a spincoated film of short-ligand PbS quantum dots, the film may be assumed to have homogenous macroscopic electrical properties [50]. This PbS quantum dot film may therefore be regarded instead as a rather poor quality semiconducting film. In contrast to the studies for isolated quantum dots, a densely packed PbS quantum dot layer coupled to an appropriate metal layer should exhibit band bending, form an associated space charge layer and display simple Schottky contact properties. This stipulation holds true for ITO/PbS nanocrystalline film/al devices, with initial studies showing device rectification, very good fit to the simple diode model and a measurable 20

depletion width of 90-150nm [50]. As a point of note, similar films made from the related PbSe nanocrystals have been shown to have mobilities on the order of 0.1-1 cm 2 /(Vs) after chemical treatment, which supports the possibility of fast and efficient carrier extraction [56]. Surprisingly, it has also been shown that the quantum dots in these Schottky barrier structures retain their individual quantum confinement effects, despite strong inter-particle carrier transport [57]. This property is important for the future possibility of hot carrier extraction and multiple exciton generation. Initial work on these planar Schottky-quantum dot solar cells showed a one-sun, AM1.5 power conversion efficiency of 1.8%, with an IR power conversion efficiency of 4.2% a 3-times improvement in IR efficiency over previous hybrid devices [51, 52]. The final structure for the optimized device was: ITO/PbS nanocrystalline film/lif/al/ag. LiF was used to improve the metal junction a common procedure in top cathodes used in organic solar cells [58]. Densely packed films were feasible because the authors exchanged the ~2.5nm oleate ligands on the quantum dots with ~0.6nm 4-carbon n-butylamine ligands. The band structure and output characteristics for this system are shown in Figure 7. Figure 7 - A. Band diagram for the simple ITO/PbS/Al Schottky-quantum dot solar cell shows device rectification. Electrons may diffuse to the depletion region where they are accelerated by the electric field to the Al electrode. Similarly, holes drift to the quasi-neutral region, where they may then diffuse to the ITO electrode. B. Output characteristics of the same solar cell under varying illumination schemes. Dark refers to no illumination while Solar refers to one-sun, AM1.5 illumination 21

from a solar simulator. The a-si filter only allows the passage of light where >640nm. The GaAs filter only allows the passage of light where >910nm. (Adapted from ref [51]). From the band diagram in Figure 7, it is clear that the diffusion transport of carriers across the quasi-neutral region is a limiting factor in carrier extraction. In thin films of PbS quantum dots, the diffusion lengths have been estimated to be 0.4 m and 0.1 m for holes and electrons respectively [52]. As such, for active layer films on the order of 200-300nm, carrier recombination due to slow diffusion should be reasonably small. Unfortunately, the use of thin films of quantum dots hinders device performance due to the greater number of photons transmitted through the film. Thicker active layers allow for the absorption of a larger number of impingent photons. Nozik and coworkers observed this problem and found that 250-nm to 400-nm PbSe nanocrystalline films exhibited lower quantum efficiencies for high energy photons [57]. This reduction in efficiency follows from the fact that high energy photons are absorbed near the surface of the cell and in the quasi-neutral region. As such, these absorbed photons receive no benefit for either exciton dissociation or subsequent carrier transport from the internal electric field provided by the Schottky contact. In addition to decreasing the quantum efficiency, thicker films have also been shown to lower the open circuit voltage and increase the series resistance of the solar cell. Sargent and coworkers examined the possibility of using a rough, microporous ITO electrode as a matrix for the PbS quantum dots in order to decrease the total length of the quasi-neutral region and allow for a thicker quantum dot film [59]. Through careful annealing, the microporous ITO/PbS/Mg cells were able to attain an IR power conversion efficiency of 2%, a slight improvement over the work described above. The authors did not report a standard 22

one-sun, AM1.5 power conversion efficiency. While such a device structure may offer a shorter path for free carriers, it should be noted that a rough, microporous ITO electrode also introduces a number of potentially deep-trap interfaces. The added presence of these deep trap states may serve to hinder instead of help device performance by further increasing carrier recombination. In addition, the conductivity across the porous ITO is likely poorer than the bulk film and may contribute to significant increases in the solar cell series resistance. Following the success of PbS-based Schottky-quantum dot solar cells, Nozik and coworkers as well as Sargent and coworkers investigated the use PbSe quantum dots in the Schottky solar cell architecture [57, 60]. Sargent and coworkers used a multiple spincoating approach combined with a benzenedithiol cross-linking ligand that allowed for the formation of insoluble thin films of tightly-packed PbSe quantum dots. In this work, the authors used an ITO/PbSe nanocrystalline film/mg/ag structure to achieve a one-sun, AM1.5 power conversion efficiency of 1.1%. PbSe-based Schottky-quantum dot solar cells appeared to rely much more strongly on diffusion transport of carriers when compared to PbS-based cells due to their smaller depletion layer width and larger quasi-neutral region [60]. Nozik and coworkers avoided the need for the complicated ligand exchange by using a much simpler and more effective layer-by-layer (LBL) deposition method. In this manner, the researchers formed a thin layer of quantum dots on a substrate through simple dip-coating and subsequently removed the oleate ligands by washing the substrates with 1,2-ethanedithiol (EDT). This process was repeated until suitably thick film was formed. In their ITO/PbSe nanocrystalline film/ca/al device, Nozik and coworkers were able to achieve a one-sun, AM1.5 23

power conversion efficiency of 2.1% [57]. Ailivisatos later employed Nozik and coworkers method of cell fabrication with ternary quantum dots composed of PbS 0.7 Se 0.3, allowing for a significant increase in the one-sun, AM1.5 power conversion efficiency to 3.3% [61]. The authors attribute this improvement to two major factors: Increased short circuit current due to a larger exciton Bohr radius (as a consequence of the PbSe contribution to the ternary structure), which provides better electronic coupling among neighbouring quantum dots Increased open circuit voltage due to a redistribution of trap states, following from the different surface energies of a ternary compared to a binary quantum dot Due to Fermi-level pinning, Schottky-quantum dot solar cells are inherently limited to maximum open circuit voltages of half of the bandgap of the quantum dot [62]. Bandgap widening due to quantization effects allows for reasonable open circuit voltages for PbS- and PbSe-based systems. However, in order to further improve device efficiency, it is desired to achieve higher open circuit voltages. For this reason, research has recently shifted to the formation of planar heterojunctions with wide bandgap films adjacent to nanocrystalline films. Leschkies et. al. recently examined ZnO-PbSe nanocrystalline planar heterojunctions for this reason and successfully achieved open circuit voltages on the order of 0.45V [63]. The full device structure along with a cross-sectional scanning electron microscope (SEM) image and the associated energy band diagram are shown in Figure 8. By separating excitons into free carriers at the PbSe/ZnO interface, this structure also reduces both geminate and bipolar recombination. This is due to the fact that electrons are isolated to the ZnO layer whereas the 24

holes are isolated to the PbSe layer. The ZnO layer was formed by sputtering and the PbSe nanocrystalline layer was formed through the LBL deposition method. Figure 8 - A. Illustrative drawing of a planar heterojunction quantum dot solar cell based on thin film ZnO and PbSe quantum dots. B. SEM micrograph of the same. C. Energy band diagram of the same. Note that excitons may favourably separate at the ZnO-PbSe and the a-npd interfaces to produce electrons, which travel toward the ITO electrode, and holes, which travel toward the Au electrode. (Adapted from ref [63]). The solar cell shown in Figure 8 makes use of a 15-nm small-molecule hole-transport layer that was thermally evaporated prior to the deposition of the gold electrode. This layer is used as an electron-blocking layer to prevent the transfer of electrons to gold that would oppose the short circuit current. In their best devices, the authors were able to achieve short circuit current, open circuit voltage and one-sun, AM1.5 power conversion efficiency values of 15.7 ma/cm 2, 0.39V and 1.6% respectively. Interestingly, the authors also observed a linear increase in the open circuit voltage of the solar cell with increasing PbSe bandgap (by decreasing the size of the PbSe quantum dots). Such behaviour has been witnessed for organic solar cells and supports the donor-acceptor, also known as the excitonic, solar cell model, where excitons must diffuse to the heterojunction to undergo dissociation [63]. Sargent and coworkers also examined a quantum dot planar heterojunction structure based on a TiO 2 film coupled with a PbS nanocrystalline film formed by a LBL method [19]. In contrast to the above study, a gold electrode was used directly without an electron blocking 25

layer. The optimized device for this structure was able to achieve the most impressive data for quantum dot solar cells to date, with short circuit current, open circuit voltage, fill factor and one-sun, AM1.5 power conversion efficiency values of 16.2mA/cm 2, 0.51V, 0.58 and 5.1% respectively [19]. Toward the goal of cheap photovoltaics, Sargent and coworkers also showed that this system could be made by using an LiF/Ni electrode instead of a gold electrode, while still achieving a one-sun, AM1.5 power conversion efficiency value of 3.5% [64]. Opposite to the simple donor-acceptor model offered by Leschkies et. al., Sargent and coworkers provide convincing evidence that their TiO 2 -PbS system follows a depleted heterojunction model that exhibits an internal electric field [19, 63]. Sargent and coworkers found that larger PbS quantum dots with no conduction band offset to the TiO 2 layer provided a solar cell that exhibited nearly the same photocurrent as smaller PbS quantum dots with significant offset. As such, an additional driving force for exciton separation beyond a simple heterojunction must be present, such as an internal electric field. Furthermore, with an excitonic solar cell model and primarily Förster energy transfer (a valid assumption given that Dexter electron transfer only occurs over very short ranges), excitons would have to hop an unreasonably large number of times to reach the heterojunction. Such diffusion time scales are much greater than the exciton lifetime [19]. As a point of note, depletion widths for both p-n junctions and Schottky contacts are strongly dependent on built-in voltage and free carrier density. Extending this knowledge to the current system, the variation in device behaviour between the ZnO-PbSe heterojunction in [63] and the TiO 2 -PbS heterojunction in [19] could be due to materials quality impacting free 26

carrier density or due to the relative band offsets of the constituent materials. It is likely that the ZnO-PbSe heterojunction exhibits a thin interfacial depletion layer that does not fully deplete the device active layer. In order to better understand the discrepancy observed in these results, additional data regarding the electron affinities of the different layer materials in addition to depletion layer widths and positions in the active layers would be required. The former may be acquired by photoelectron spectroscopy, whereas the latter can be obtained through simple capacitance-voltage measurements and calculations. While this fourth configuration of quantum dot solar cells allows for entirely inorganic structures, all studies have shown the solar cells to exhibit fast device degradation in oxygen and ambient environments. This instability has been attributed to both the surface oxidation of the quantum dots, resulting in the formation of deep trap states, and to the oxidation of the low work function top electrode. Sargent and coworkers examined a ligand exchange encapsulation method to address the prior issue [65]. This work allowed for stable Schottkyquantum dot solar cells with 3.6% one-sun, AM1.5 power conversion efficiency for 0.5h in air. In order to address the latter issue, the authors made use of a LiF passivation layer prior to the deposition of the Al top electrode, granting stability on par with organic photovoltaics [66]. X- ray photoelectron spectroscopy (XPS) studies indicated that the LiF layer slowed the oxidation of the Al electrode and reduced diffusion of oxygen into the nanocrystalline film. However, as is the case with organic photovoltaics, it is necessary to further improve the solar cell stability in order to become competitive with current technologies. Taking Advantage of Hot Carriers and Multiple Exciton Generation Effects 27

While there have been reports on PbS, PbSe, PbTe, Si and InAs quantum dots exhibiting multiple exciton generation effects [18, 67-72], many authors have been critical of the inability of these researchers to present these effects in experimental solar cells [73]. Indeed, most of this preliminary research is based on pump probe laser experiments with lasers exhibiting powers much greater than realistic one-sun, AM1.5 conditions. No researcher has successfully demonstrated either external quantum efficiency or even back-calculated internal quantum efficiency values in excess of 100%. Even more recent work on the promising Schottkyquantum dot solar cells have exhibited internal quantum efficiencies that fall short of 100% [57, 74]. Similarly, while there has been some recent success in experimentally showing hot carrier transfer [75], it is uncertain if structures can be designed to efficiently harvest these carriers under realistic exposure conditions. Given the above considerations, there has yet to be any direct proof that multiple exciton generation or hot carriers can provide additional current or cell potential that would improve the efficiency of third generation solar cells under realistic exposure conditions. However, given that quantum dot solar cells have only begun to see significant improvements in their power conversion efficiencies in the past three years, it is too early to dismiss this technology. With the development of depleted heterojunction solar cells, significant improvements in quantum dot solar cell characteristics are expected in the very near future. 28

Conclusions. Quantum dot solar cells fall into the third generation of photovoltaics, which can allow for both greatly enhanced efficiencies and cheap manufacturing. The decreased manufacturing costs arise from the facile, solution-based fabrication of quantum dots. Enhanced efficiency can be attributed to better solar cell absorption spectral matching with the sun s emission, including the capacity to harvest IR photons. Additional effects, such as hot carrier extraction and multiple exciton generation could potentially allow for quantum dot solar cell efficiency values greater than the Shockley-Quessier limit of ~31%. Unfortunately, neither of these effects has been measurable in the external and internal quantum efficiency values reported for any solar cells to date. Four configurations of quantum dot solar cells have been presented. Two such configurations, vacuum deposited solar cells and liquid junction solar cells, are interesting from a research standpoint; however, challenges in their complicated and costly fabrication must be overcome before they can compete with silicon photovoltaics. Hybrid organic-quantum dot solar cells have undergone extensive research, but have achieved only modest power conversion efficiencies. Fully inorganic, solution-processed quantum dot solar cells, such as the Schottky-quantum dot and depleted heterojunction structures, have shown significant promise with the highest quantum dot solar cell efficiency of 5.1%. In all configurations, the large number of deep interfacial traps limits device efficiencies. In order to become competitive with current photovoltaic technology, it is now necessary to investigate new device architectures to take advantage of hot carriers and multiple exciton generation effects. 29