07 April 2007

Organic Photovoltaics

Introduction

Photovoltaic cells are large-scale electronic junctions that absorb sunlight and generate electricity. The most typical type under commercial production today is the polycrystalline Silicon cell. It is typically a 300 μm thick wafer of polycrystalline silicon that has been doped to have an excess of positive electrical charge carriers (positive or p-type) on the back surface and an excess of negative charge carriers (negative or n-type) on the facing surface. [Reminder: a hole is a pseuodo-particle commonly seen in semiconductor materials that represents an absence of an electron and behaves very similarily to a positively charged electron.] An aluminium surface acts as the cathode while a transparent conductive oxide layer such as Indium Tin Oxide (ITO) is applied to the sunward face as the anode for the purposes of collecting holes and electrons, respectively. Commercial Silicon solar cells typically convert incoming solar radiation (insolation) to electricity with an external efficiency of 12 % under Air Mass 1.5 Standard illumination.

The cost of generating electricity via photovoltaics is generally not competitive with conventional thermal power generation except for sites isolated from the existing grid. The high cost of photovoltaics is generally not due to the low efficiency but rather the cost of semiconductor-grade silicon and the extensive material processing required. The high capital cost of photovoltaics also constitutes a barrier to their entry to market as they must be amortized over their forty year lifespan. One of the main hopes of organic photovoltaics is that they can be produced using simple, low-tech, low-energy input techniques.

In an effort to reduce the quantity of zone-refined material required for photovoltaic production a number of thin-film technologies have been proposed. These include: hydrogenated amorphous and microcrystalline silicon, Cadmium Telluride, Copper Indium Gallium Selenide (CIGS) and various organic semiconductor concepts. The organic photovoltaic family includes systems based on short-chain polymers (Yoo et al., 2004) and/or conjugated polymers paired with carbon fullerenes (also known as Buckeyballs) (Yu et al., 1995) or dye-sensitized inorganic nanocrystals (McDonald et al., 2005; Petrella et al., 2004). [Aside: Some of the early literature on short-chain conductive polymers such as pentacene (Schon et al., 2000) was later retracted (Schon et al., 2003). Due to the difficulty of disentangling legitimate research from the tainted results I will not discuss those materials in this review. ]

Conventional photovoltaics absorb photons in order to generate electron-hole pairs and then collect the generated charges at their appropriate terminals in order to generate a photocurrent. Hence, fabrication of photovoltaic cells is generally an act of optimizing the thickness to maximize absorbance (which increases with thickness) and conductance (which decreases with thickness). In organic semiconductors the photogeneration of electron-hole pairs is complicated by the fact that they are generated in a bound electron-hole pair state known as an exciton. Thus organic photovoltaics need to manage the additional step of charge separation via some sort of charge transfer mechanism. Fortunately the absorption coefficient for organic photovoltaics is typically very high (~ 105 cm−1) (Hoppe and Sariciftci, 2004) which allows very thin active areas. In comparison (indirect band-gap) Silicon is an order of magnitude worse while (direct band-gap) GaAs has a similar absorbance.

In organic photovoltaics, the site that participates in photogeneration of charge carriers and that which transports charge to the terminals is often not of the same material. The material which transports holes may also be different from that which transports electrons.

Conduction in Polymers

Conjugate polymers consist of a long-chain alternating single and double-bond saturated hydrocarbons. A common example used in organic photovoltaics is polyphenylenevinylene (PPV) (Yu et al., 1995) and its derivative methoxy-ethylhexyloxy-phenylenevinylene (MEH-PPV). The simpliest possible example would be the allyl-chain shown in Fig.1.


Figure 1: Schematic representation of conjugate allyl chain (N = 3) showing pz-orbitals only (taken from Salem, 1966). The

Conduction is accomplished by the series of π-bonds at lie along the backbone of the carbon chain. π-bonds consist of parallel pz-shell electrons that have bonded. When electrons bond their orbitals merge to form a delocalized electron cloud, and either electron may freely move about inside. When a series of electrons form a continuous chain of π-orbitals, they form a conductive pathway along which conduction may occur. The sp2-hybridized σ-orbitals that are involved in hydrogen bonding are not considered to take part in conduction. Typically the π-orbitals are not completely delocalized as rotations in the backbone cause breaks in the conjugation on the order of 100 cm−1 (Scholes and Rumbles, 2006). Charge carriers require some sort of assistance, such as phonon-assisted band hopping, to jump from one localized state to another. The conductivity in conjugate polymers is obviously anisotropic (not the same in every direction), being highest along the direction of the backbone. As a result, the macroscopic conductivity (or charge carrier mobility) of a disordered conjugate polymer is quite poor compared to bulk inorganic semiconductors. Conjugate polymers have much poorer mobility than inorganic semiconductors and are typically better at conducting holes than electrons. Typical hole mobilities (μ) range from 10−5 to 0.1 cm2V -1s-1 (Dimitrakopoulos and Mascaro, 2001).
Figure 2: Molecular ’band’ structure of a two-phase organic photovoltaic cell with the donor acting as the site of exciton generation. In a dye-sensitized system the sensitizer lies between the donor and acceptor while acting as the generation site.

The analogues to the valence and conduction band in conjugate polymers are known as lowest unoccupied molecular orbital (LUMO) and highest occupied molecular orbital (HOMO) (depicted in Fig. 2). The literature also uses 'π-orbital' as a synonym for HOMO and 'π*-orbital' for LUMO. The difference in energy between LUMO and HOMO can be considered to be the bandgap analogue in conjugate polymers. The bandgap in conjugate polymers is typically around 2 eV (Hoppe and Sariciftci, 2004), which corresponds to a wavelength of 620nm. This figure is considerably far away from the optimal value of 1.4 eV for photovoltaics (Shockley and Queisser, 1961) and means that a large portion of red and infrared radiation will go unutilized. The poor overlap between photoconductive polymers and the radiation spectrum produced by the sun (as shown in Fig. 3) remains a significant problem for organic-based photovoltaics.

Figure 3: Absorption coefficients of common conjugate polymers used in organic photovoltaic cells incomparison to the Air-Mass 1.5 solar radiation spectrum(taken fromHoppe and Sariciftci, 2004). Note the poor absorption in the red and infrared regions of the spectrum. Human eyesight is sensitive from about 700 nm (red) to 400 nm (blue).

Conjugated polymers can be doped, although by a different process compared to inorganic semiconductors. Typically p-type materials are enhanced by oxidizing the material with iodine or some other agent (Chiang et al., 1977). The oxidizing agent accepts an electron from the polymer, rendering its net charge more positive. Application of n-type doping is more difficult as it requires reduction by hydrogen. Since the atmosphere is a net oxidizing environment, n-type doping needs to be ruggedly encapsulated to prevent reversal of the doping over the lifetime of the solar cell.

Excitons and Charge Transfer

When a photon is absorbed in a semiconductor it can be generally said that it promotes an electron from the valence band to the conduction band. This creates a mobile negative charge in the conduction band—the electron—as well a corresponding lack of an electron—known as a hole—in the valence band which is also mobile. Initially there is a static electric (Coloumb) attraction between the electron and hole which creates a quasi-particle known as an exciton. In inorganic semiconductors the exciton binding energy is of the same order as room-temperature vibrations so dissociation is typically extremely swift. Conjugate polymers, in comparison, have very high exciton dissociation energies, varying from 0.3 to 1.0 eV depending on the material and chiral orientation (Scholes and Rumbles, 2006). These tightly bound excitons are sometimes known as Frenkl excitons. This implies that charge seperation in organic semiconductors is a non-trivial event and functional optoelectronic devices based on conjugate polymers should be designed to optimize the process.

For conjugate polymers and other nanoscale materials the size and orientation of the polymer chains themselves influences the opto-electronic properties of the material. This is due the role quantum confinement plays in the static electric interaction of excitons (Scholes and Rumbles, 2006). The higher exciton binding energy in conjugate polymers is thought to be largely due to the one dimensional confinement imposed by them compared to bulk inorganic semiconductors (Kohler et al., 1998). This effect also manifests itself as peaks in the relationship of absorption coefficient as a function of wavelength.

Dispersed Heterojunction Devices

The first organic photovoltaic devices used a homojunction consisting of a thin organic layer between two electrical contacts (Nelson, 2002). The inherent electronic field of the device was used to separate charge carriers and drive them to the appropriate terminal. However, the efficiency of these devices was extremely low (<< 1%) due to the combination of poor mobility and fast exciton recombination time. Ultrafast photoluminescence studies of PPV found a diffusion rate of3·10−3 cm2 s−1 with a corresponding migration radius of 6 nm (Haugeneder et al., 1999; Markov et al., 2005). Haugeneder et al. reported that excitons could hope from state to state every 1 ps and conducted a total of 45 hopes on average before recombining.

A breakthrough came when researchers figured out how to create devices where ’acceptor’ and ’donor’ material was randomly dispersed in the bulk between the two electrical contacts (Halls et al., 1995). The device considered of a mix of PPV and a solvented fullerene which was then spin-coated to form a thin-film. This greatly increased the surface area between the two phases, thereby increasing the likelihood that a photogenerated exciton would be able to diffuse to a boundary and be separated before it decayed and recombined. The morphology of the material is critical as it determines if an exciton created at any potential location will be able to diffuse to the boundary and undergo charge separation (Moons, 2002). As such it is highly desirable that clusters as in Fig. (4) have a radius similar to that of the exciton diffusion length. An experiment using a bilayer C60/PPV cell found an optimal thickness of 23 ± 4 nm for the PPV layer and 27 ± 2 nm for the C60 layer (Stubinger and Brutting, 2001). The thickness dependence is thought to be an optical interference effect. This result suggests that the morphology of organic solar cells effects not only their electronic properties but also the optical properties. Characterization of organic thin film morphology is complicated by the fact that they are typically susceptible to damage when irradiated by an electron beam used in scanning electron microscopy or transmission electron microscopy. It is more common in the literature for scanning probe techniques to be used, although they can only provide information regarding the surface properties and cannot provide information about the bulk of a material.

Figure 4: SEM images of conjugate polymer (polyphenylenevinylene) : fullerene (propylphenyl-C61) ratios (by weight) spin-cast from toluene. The clusters of fullerene material becomes visible as the proportion of fullerene is double that of the conjugate polymer. The dimensions of the nanoclusters is critical as they should be less than the exciton diffusion length to ensure a high probability of charge carrier seperation (taken from Hoppe et al., 2004).

The concept of having two phases co-currently led to the concept where the hole and electron conductor are composed of two different materials. The most common variant is composed of MEH-PPV and the polymer-fullerene complex propylphenyl-C61 (abbreviated PCBM). The fullerene is highly electronegative relative to the conjugate polymer matrix and hence acts as an electron acceptor. The addition of the phenyl chain acts to make the fullerene soluble so that it may be spin cast.

Dispersed heterojunction organic photovoltaics are typically produced by spin coating with the two phases being suspended in a solvent such as toluene (Hoppe and Sariciftci, 2004). Alternatively they can be evaporated at low temperatures, relying on the low density of such films to encourage diffusion between donor and acceptor materials. Such devices are of course disordered and rather amorphous; as such quality control is quite difficult. One hope for organic photovoltaics is to use the potential for chemical and biological nanostructuring techniques to achieve a more ordered result with superior electrical properties. One active area of research is the use of self-assembly to orient the backbones parallel to the direction of charge carrier transport (Nelson, 2002).

The issue of relative electron affinities is also relevant in designing interfaces between the active layers and the electric contacts. For the contact to be nearly ohmic the work function of the anode should be close to but less than the LUMO of the electron acceptor. Similarly the cathode should be close to but greater than the HOMO of the electron donor to allow conduction of holes. The difference between the work function of the electrical contacts is a good indication of the potential generated by the cell (Gregg and Hanna, 2003). An overly large potential change at the interface between the contact and active material can not only lead to a loss of power through charge carrier thermalization but also form a Shockley barrier.

In a typical homojunction solar cell the difference in work function between the anode and cathode generates an electric field that causes the charge carriers to drift towards their appropriate terminal. When the cell is illuminated, charge carriers collect at their appropriate terminals and reinforce this built in electric field. Logically in a Silicon pn−junction the material adjacent to the cathode is p−doped and that adjacent to the anode is n−doped to assist in the sweeping out of charge carriers. However, in a dispersed polymer heterojunction donor material may be in contact with the cathode in places. One potential solution towards improving the electrical characteristics of the cell is to use buffer layers between the electrical contacts and the active dispersed layer (Peumans et al., 2000). These layers are not strongly optically absorbing but are effective conductors of their associated charge carrier.

Complementary Acceptor Materials

The difficulties associated with n−doping of conjugate polymer has lead to the development of a variety of alternative materials to use as the electron acceptor in organic photovoltaic devices. These materials are often inorganic in nature and not distributed homogenously throughout the active layer with the result that electrons must use vibration assisted hopping or tunnelling to move from one site to another as they progress towards the electrode. These processes tend to be slow and hence lead to a greater likelihood of
recombination and poor conductivity.

The orginal acceptor material was a the fullerene based propylphenyl-C61 (Yu et al., 1995). The phenyl chain is able to bond itself to the polymer matrix while the fullerene is electronegative compared to the polymer. Fullerene based cells hold most of the records with regards to performance, with external efficiency in excess of 2.5 % being reported (Shaheen et al., 2001) and the current certified record of 3.0 % (Green et al., 2007). Note that literature claims of higher efficiency are often under short wavelength monochromatic light. The improved efficiency was concluded to be due to the finer morphology generated by the use of chlorobenzene solvent in the spin casting process. Toulene cast films had feature sizes on the order of 0.5 μm whereas the chlorobenzene cast films had features on the order of 0.1 μm. The chlorobenzene also suppressed cluster formation of the fullerene, which would reduce electron conductivity in the bulk due to segregation of conductive volumes.

A similar approach was taken with TiO2 nanocrystals embedded in a matrix of PPV (Salafsky, 1999). The TiO2 nanocrystallites act to improve the absorbance of the material in short wavelengths. However, the bandgap of TiO2 is quite high (3.2 eV) which limits the wavelengths the nanocrystals will interact with. In this case the crystallites are on the scale of 20nm in diameter which is generally considered to be outside of the quantum confinement regime. At sufficiently high concentrations the TiO2-PPV cells acts in a similar fashion to a dispersed heterojunction with the TiO2crystallites providing an alternative conduction pathway for electrons. Devices of this type showed considerable improvement in performance over and above what would be expected from increased absorbance. This provided evidence that the heterogenous interface assisted charge separation and lead to more greater interest in other nanocrystalline material that could be used as a complementary acceptor.

One of the more recent and sophisticated attempts to use an acceptor material to complement the material properties of conjugate polymers involved the use of PbS quantum dots embedded in MEH-PPV (McDonald et al., 2005). Due to quantum confinement effects, the range of wavelengths that quantum dots will interact with can be controlled by varying their size. In particular, this allows conjugate polymer based devices to interact with infrared light that they would otherwise be insensitive to. Bulk PbS only has a bandgap of 0.4 eV but the quantum dots used by McDonald et al. had absorption peaks at 1.3, 1.03, and 0.92 eV . The efficiency of these early devices was poor (~ 3% internal quantum efficiency and very low external efficiency). Still this experiment was considered important as it offered the potential to allow organic photovoltaics to be better matched to the solar spectrum.

Conclusions

A review of the literature shows that the most predominate strategy for modern organic photovoltaic schemes involves the use of conjugate polymers acting as the exciton generation site and electron donor paired with a complimentary electron acceptor. The most common conjugate polymer seen in the literature is polyphenylenevinylene (PPV) and its derivative methoxy-ethylhexyloxy-phenylenevinylene (MEH-PPV). The fullerene propylphenyl-C61 (PCBM) complex is the most commonly used acceptor, although there is also considerable interest surrounding the use of inorganic nanocrystallites.

Scientifically organic photovoltaics is a very diverse field that offers a researcher a great many research paths in order to optimize cell performance. The nanomorphology of the donor and acceptor materials is critical in determining whether an exciton will be able to diffuse to a donor/acceptor boundary and undergo charge separation. The poor matching of conjugate polymer’s absorption to the sun’s radiation spectrum and the relatively poor photoconductivity of organic semiconductors remain important scientific problems to be addressed. Many nanoscience techniques such as self assembly offer promising paths towards achieving superior photovoltaic performance.

The greatest engineering challenges surrounding deploying organic photovoltaic modules to the marketplace are regarding lifetime and efficiency. Organic materials are typically vulnerable to photodegradation and have lifespans measured in hundreds of hours (Chambon et al., 2007). Economically, efficiency becomes relevant due to the limited roof-space available for photovoltaic installations and the associated costs of protective covering glass, mounts, and labour. Whether organic photovoltaics will become cost competitive with conventional polycrystalline Silicon or emergent inorganic thin film technologies remains to be seen.

References
  • Chambon, S., A. Rivaton, J.-L. Gardette, and M. Firon, 2007, Solar Energy Materials and SolarCells 91(5), 394.
  • Chiang, C. K., C. R. Fincher, Y. W. Park, A. J. Heeger, H. Shirakawa, E. J. Louis, S. C. Gau, and A. G. MacDiarmid, 1977, Physical Review Letters 39(17), 1098.
  • Dimitrakopoulos, C. D., and D. J. Mascaro, 2001, IBM Journal of Research and Development 45(1), 11.
  • Green, M. A., K. Emery, D. L. King, Y. Hishikawa, and W. Wart, 2007, Progress in Photovoltaics:
    Research and Applications 15(1), 35.
  • Gregg, B. A., and M. C. Hanna, 2003, Journal of Applied Physics 93(6), 3605.
  • Halls, J. J. M., C. A. Walsh, N. C. Greenham, E. A. Marseglia, R. H. Friend, S. C. Moratti, and A. B. Holmes, 1995, Nature 376(6540), 498.
  • Haugeneder, A., M. Neges, C. Kallinger, W. Spirkl, U. Lemmer, J. Feldmann, U. Scherf, E. Harth, A. Gugel, and K. Mullen, 1999, Physical Review B 59(23), 15346.
  • Hoppe, H., M. Niggemann, C. Winder, J. Kraut, R. Hiesgen, A. Hinsch, D. Meissner, and N. S. Sariciftci, 2004, Advanced Functional Materials 14(10), 1005.
  • Hoppe, H., and N. S. Sariciftci, 2004, Journal of Materials Research 19(7), 1924.
  • Kohler, A., D. A. dos Santos, D. Beljonne, Z. Shuai, J. L. Bredas, A. B. Holmes, A. Kraus,
  • K. Mullen, and R. H. Friend, 1998, Nature 392(6679), 903.
  • Markov, D. E., J. C. Hummelen, P. W. M. Blom, and A. B. Sieval, 2005, Physical Review B (Condensed Matter and Materials Physics) 72(4), 045216.
  • McDonald, S. A., G. Konstantatos, S. Zhang, P. W. Cyr, E. J. D. Klem, L. Levin, and E. H. Sargent, 2005, Nature Materials 4, 138.
  • Moons, E., 2002, Journal of Physics: Condensed Matter 14(47), 12235.
  • Nelson, J., 2002, Current Opinion in Solid State and Materials Science 6(1), 87.
  • Nelson, J., 2003, The Physics of Solar Cells (Imperial College Press, London).
  • Petrella, A., M. Tamborra, P. D. Cozzoli, M. L. Curri, M. Striccoli, P. Cosma, G. M. Farinola, F. Babudri, F. Naso, and A. Agostiano, 2004, Thin Solid Films 451-452, 64.
  • Peumans, P., V. Bulovic, and S. R. Forrest, 2000, Applied Physics Letters 76(19), 2650.
  • Peumans, P., A. Yakimov, and S. R. Forrest, 2003, Journal of Applied Physics 93(7), 3693.
  • Salafsky, J. S., 1999, Physical Review B 59(16), 10885.
  • Salem, L., 1966, The Molecular Orbital Theory of Conjugated Systems (W.A. Benjamin Inc., New
    York).
  • Sariciftci, N. S., L. Smilowitz, A. J. Heeger, and F. Wudl, 1992, Science 258(5087), 1474.
  • Scholes, G. D., and G. Rumbles, 2006, Nat Mater 5(9), 683.
  • Schon, J. H., C. Kloc, E. Bucher, and B. Batlogg, 2000, Nature 403(6768), 408.
  • Schon, J. H., C. Kloc, E. Bucher, and B. Batlogg, 2003, Nature 422(6927), 93.
  • Shaheen, S. E., C. J. Brabec, N. S. Sariciftci, F. Padinger, T. Fromherz, and J. C. Hummelen,
    2001, Applied Physics Letters 78(6), 841.
  • Shockley, W., and H. J. Queisser, 1961, Journal of Applied Physics 32(3), 510.
  • Strietwieser, A., 1961, Molecular Orbital Theory for Organic Chemists (Wiley, New York).
  • Stubinger, T., and W. Brutting, 2001, Journal of Applied Physics 90(7), 3632.
  • Yoo, S., B. Domercq, and B. Kippelen, 2004, Applied Physics Letters 85(22), 5427.
  • Yu, G., J. Gao, J. C. Hummelen, F. Wudl, and A. J. Heeger, 1995, Science 270(5243), 1789.
    14

1 comment:

Harald said...

Robert,
Great exlanations. We are selling fullerenes but I was never really certain how fullerenes create value for solar cells.

Thank you.
Harald