Royal Society Publishing

Nanostructures in photovoltaics

Kylie R Catchpole


The world has recently been waking up to the urgent need to move away from fossil fuels and towards a low-carbon economy. To achieve this, we need a way of producing electricity that is efficient, widely applicable and cheap. At the same time, there has recently been an appreciation of the tremendous scope for making entirely new types of devices, and even seeing new physics, by structuring matter at the nanoscale. Furthermore, the occurrence of self-assembly in nature suggests that a range of types of nanoscale structures could be made simply and cheaply. The application of nanostructures to photovoltaics combines a field of almost limitless possibilities with a problem of vital urgency. In this paper, some of the newer ideas emerging from this trend are described, along with how they challenge our ideas on what a solar cell looks like. We are at the beginning of a time of radically rethinking the design of the solar cell, which may lead to the exploitation of completely new physical ideas in achieving a sustainable energy future.


1. Introduction

It is now becoming widely accepted that in order to avoid dangerous climate change, we need to drastically cut our emissions of carbon dioxide. In order to achieve this goal at the least cost to society, we need to take action now, with the technology presently available; but we also need to develop cheaper and more efficient ways of supplying our energy. The potential for improvement is huge. One only has to look at a spider spinning a web to appreciate that it is not necessary to have a large polluting factory to produce a material stronger than steel, and the same applies for meeting our energy requirements.

One of the technologies with the greatest scope for reducing carbon dioxide emissions is solar photovoltaics, which uses sunlight incident on a semiconductor diode to excite an electron and directly generate electricity. Photovoltaics is particularly important as a future energy source owing to the massive amount of energy available directly from sunlight. For example, it would be theoretically possible to supply all of Australia's electricity needs from an array of solar panels of approximately 33×33 km—smaller than Sydney. Of course, the most practical and cost-effective way of reducing greenhouse gas emissions will be a combination of measures. It has been calculated that if the photovoltaics were to be one of 10 technologies used to reduce greenhouse gas emissions sufficiently to stabilize the climate, approximately 700 times the presently installed capacity would be required by 2050 (Pacala & Socolow 2004). While this may sound daunting, it represents an annual growth rate of less than 20%, and the industry has been growing at over 30% for the past 5 years.

Since the early silicon solar cells in the 1950s, which had efficiencies around 6%, there have been great improvements in the efficiency of converting sunlight to electricity, with the record efficiency of 24.7% for silicon solar cells produced at the University of New South Wales (Zhao et al. 1998). At the same time, costs of solar modules have steadily fallen, from a cost of US$100/W in 1970 (in today's dollars) to around US$5/W. The economies of scale associated with the increasing market for photovoltaics together with the development of new devices that use far less expensively purified silicon are expected to further reduce costs significantly. However, looking further ahead, if we want to produce the most efficient solar cell at the least possible cost, it is beneficial to put aside our ideas of what a solar cell looks like now and ask the question, what could a solar cell look like? The nanotechnology revolution presently underway is giving us unprecedented control over the structure of materials at the nanometre scale, resulting in dramatically different optical and electronic properties. We are also discovering that a self-assembly of elaborate structures is common in nature, not only in biological systems but also in much simpler physical and chemical systems, e.g. snowflakes. The nanotechnology revolution offers the possibility to make solar cells that are much more efficient and cheaper than present day cells (Green 2001). In order to realize this potential, we must radically rethink our ideas of solar cell design.

2. What is a nanostructured solar cell?

Present state-of-the-art high-efficiency silicon solar cells are based on silicon wafers approximately 300 μm thick, as shown in figure 1a. The property of semiconductors like silicon which allows them to be used for solar cells is the fact that they have a gap in their electronic levels, known as the electronic bandgap, which is similar to the energy of photons of visible light (figure 1b). The energy levels below the bandgap are known as the valence band, while the energy levels above this gap are called the conduction band. The key feature of conventional solar cells is the p–n junction. The junction is between silicon doped with small amounts of impurities such as boron (p-type) and silicon doped with impurities such as phosphorus (n-type). The incident light excites electrons from the valence band to the conduction band of a semiconductor, leaving behind spaces where electrons have been excited known as holes. The asymmetry of the p–n structure leads to the development of gradients in the concentrations of electrons and holes, and also an electric field. The field and the concentration gradients result in a net driving force when the device is illuminated, towards the n-type contact for electrons and towards the p-type contact for holes. The p- and n-type regions of the cell act as one-way membranes (Würfel 2005), separating the electrons on the n-type side and holes on the p-type side of the solar cell. Thus, the fundamental functions of a solar cell are to absorb light, to separate electrons and holes with a useful amount of potential energy (which can result in a voltage being produced across the cell) and to transfer the electrons and holes to the metal contacts of the device and hence through the external circuit (i.e. to produce an electric current). Novel nanostructures have been proposed recently to achieve each of these functions.

Figure 1

(a) The structure of a high efficiency silicon solar cell. The silicon wafer is mostly p-type, with a thin n-type region on top. The top surface is structured with inverted pyramids to trap light inside the cell and increase its chance of being absorbed. (b) The electronic band structure of a p–n junction solar cell showing the directions of travel of electrons and holes. The bending of the conduction and valence bands at the junction is due to an electric field in that region. The gradients in the concentrations of electrons and holes (not shown in the diagram) also contribute to the driving force for electrons and holes.

In a conventional solar cell, the interfaces between different parts of the device (e.g. the junction in a p–n solar cell) have a similar area to the whole device. In a nanostructured cell, the interfaces may have areas orders of magnitude higher. This offers the opportunity not only to separate electrons and holes very rapidly, but also leads to a large area where recombination of electrons and holes can potentially occur. Thus, nanostructured cells demand new cell designs for devices, where interfaces are inherently a large component of the device. Furthermore, in a conventional solar cell, the silicon performs the roles of absorbing the light, separating the electrons and holes and transporting them to the contacts. It may be an advantage in a nanostructured cell to have different regions of the cell performing different functions, to reduce the number of requirements on each region.

There are too many types of emerging nanostructured solar cells to do them all justice here. These include dye-sensitized cells and some types of organic solar cells, both of which have been the subject of recent reviews (Gratzel 2001; Peumans et al. 2003). Instead, I will focus on a few of the newer ideas to demonstrate some of the things that are possible and how they challenge our thinking on what a solar cell looks like.

3. Metal nanostructures for improving light absorption

In a conventional solar cell, metal plays the role of a low-resistance electrical contact, connecting the solar cell to an external circuit. The fundamental property, which makes metals good conductors, is that there are many energy levels closely spaced among which electrons can move freely. This property also makes it easy for electrons to fall into holes, known as recombination of electrons and holes. For this reason, the main attention that has been paid to metal contacts in solar cells has been to minimize their area (to reduce recombination and shading losses) while at the same time keeping resistive losses low.

With this in mind, the number of ways that metal nanoparticles can potentially improve the performance of a solar cell, and in particular, the light absorption, is surprising. The basis for all these effects, which will be discussed later, is the fact that light incident on metal nanoparticles can excite surface plasmons. A surface plasmon is a collective oscillation of electrons in a conductor. Surface plasmons can exist on continuous surfaces or isolated particles; when they are on isolated particles, they are also called localized surface plasmons (figure 2). We can imagine a localized surface plasmon as a bit like a balloon filled with water. We shake the balloon back and forth. At a certain frequency of oscillation, the water balloon will respond strongly. This is the resonant frequency. Similarly, there is a frequency at which the electrons in the metal nanoparticle respond strongly to the incident light; this is the surface plasmon resonance.

Figure 2

An incident light wave excites a localized surface plasmon on a spherical metal nanoparticle.

In order for a photon to contribute to the electrical current produced by a solar cell, it must be absorbed by the silicon. But silicon absorbs light relatively weakly. One way to overcome this limitation is to ensure that light passes several times across the device, to increase its chance of being absorbed. Conventional silicon solar cells achieve this through upright or inverted pyramids approximately 10 μm in diameter etched into their surface to trap light inside the cell by total internal reflection (figure 1a). This sort of surface texture is not an option for a nanostructured device, which may be only a few micrometres thick. Texturing the semiconductor surface with features approximately 300 nm in size is one way of trapping light for thin devices. But metal nanoparticles on the surface are another unexpected way of trapping the light in the solar cell. There are a number of ways to fabricate metal nanoparticles, but one simple method is to evaporate a very thin layer of metal and then heat it to approximately 200°C. Surface tension forms the metal into roughly spherical droplets. Metal nanoparticles formed in this way have been used to enhance the light-generated current in ultra-thin silicon solar cells (Stuart & Hall 1998; Pillai et al. 2006). The light hitting the solar cell excites a surface plasmon on the metal nanoparticle, which then re-radiates most of its energy into the silicon in such a way that the light is trapped inside the cell (figure 3a).

Figure 3

(a) Light excites a surface plasmon resonance on a metal nanoparticle which then re-radiates the light into a trapped waveguide mode in the silicon. (b) When an external voltage is applied to a solar cell, it operates as a light-emitting diode. One the left side of the device metal nanoparticles have been applied, while on the right, there are no metal nanoparticles. It can be seen that the metal islands substantially increase the amount of light emitted.

An important principle in optics is that the path of light is reversible if only elastic scattering (without energy loss) is involved. This is known as the reciprocity theorem, and its original formulation is due to Helmholtz (Helmholtz 1909; Greffet & Nieto-Vesperinas 1998). This means that if a structure can be used for trapping light inside a solar cell, it can also be used for extracting light trapped inside a device. We have used this idea to demonstrate a dramatic increase in light emitted from our thin solar cells when they are operated as light-emitting diodes, as shown in figure 3b (Catchpole & Pillai 2006; Pillai et al. 2006). Thus, the ideas described here are also relevant in improving the efficiency of light-emitting diodes. Indeed, light-emitting diodes also play a role in reducing carbon dioxide emissions, owing to their much greater efficiency than incandescent globes.

An advantage of using metal nanoparticles to scatter light is that they can be applied to any thin material, not just thin layers of silicon. This is important because with nanostructured materials, there may be limitations on layer thickness in order to get good performance. Light trapping could become even more important for nanostructured photovoltaic devices than for conventional crystalline silicon devices, because we cannot expect nanostructured materials to have the same excellent charge carrier transport properties as high-purity crystalline semiconductors.

The strong interaction between light and metal nanoparticles also leads to increases in the electromagnetic field around the particles (figure 4). The particles effectively concentrate the light into small regions. If a semiconductor is close to or surrounding the metal particles, this will increase the light absorbed by the semiconductor in that region. This effect can be used to increase the number of electron–hole pairs generated near the junction of a solar cell, where they have the best chance of being collected before recombining (Rand et al. 2004; Schaadt et al. 2005). A range of shapes of metal nanoparticles and interactions between the particles can lead to a broad surface plasmon resonance; therefore, absorption can be increased over a relatively wide range of photon energies. It is interesting to note that it is possible to get increases in the light-generated current even for metal nanoparticles, which are so small (approx. 10 nm) that it might be expected that absorption in the nanoparticle is the dominant process, which would be dissipated as heat. However, the enhanced current which has been observed shows that even for small nanoparticles, just because a surface plasmon has been excited does not mean that the energy is lost.

Figure 4

Intensity enhancement around 100 nm diameter silver cylinders on a quartz substrate illuminated from above. The maximum intensity enhancement here is a factor of 5; this can be increased by decreasing the spacing or changing the shape of the silver structures.

It has also been suggested that excitation of a surface plasmon can lead to direct electron emission (Westphalen et al. 2000). The mechanism here is that the collective oscillation of the electrons (the surface plasmon) in the metal nanoparticle transfers its energy to a single electron within the metal particle, creating an electron–hole pair. If there is an asymmetry in the local environment, for example, if the metal particles are located in a strong electric field, the electron and hole can be separated.

Surface plasmons on metal nanostructures can even change the optical properties of a nearby semiconductor. Near metal nanostructures, the number of states in which a photon can exist (the photonic mode density) is increased. This increases the number of ways an electron–hole pair can be excited by a photon. Hence, the ability of a semiconductor material to absorb light is increased near a metal nanostructure. It might be possible to use this effect to turn a weak absorber like silicon into a strong absorber (Biteen et al. 2005).

From these examples, we can see that there is tremendous potential to incorporate surface plasmons in new designs for solar cells. This section has focused on metal nanostructures for improving light absorption; in the following sections, I discuss how nanostructures can be used to make full use of the solar spectrum and some considerations on cell designs for nanostructured materials.

4. Quantum dots for making the most of the solar spectrum

In conventional solar cells, only one type of semiconductor is used. This means that the voltage produced from, e.g. blue light is the same as that produced by near infrared light, even though the blue light may have three times as much energy per photon. This fact limits the efficiency of single junction solar cells to 31% for unconcentrated sunlight. This limitation can be overcome through the use of a tandem stack, in which several different semiconductors of different bandgaps are used to make a series of interconnected cells that sit on top of one another. The blue light can then be absorbed by a high bandgap cell at the top of the stack to produce a large voltage, while the infrared light passes through the first cell and is absorbed by a lower bandgap semiconductor, and produces a lower voltage. Using this approach, the limiting efficiency for unconcentrated sunlight is increased to nearly 70% (Marti & Araujo 1996). The efficiency is not 100% because in order to be able to absorb radiation, a solar cell must also be capable of emitting radiation and this emitted radiation is lost. Tandem development has been limited by the availability of suitable materials, since there are few semiconductors which have the bandgaps required to form a tandem stack, and they tend to be expensive or toxic.

We can avoid being restricted to a few materials by using nanostructures. Semiconducting particles a few nanometres in size are called quantum dots. In quantum dots, the electronic bandgap is increased owing to restrictions on the wavelengths of the electrons that can fit within the dot. As the size of the quantum dot decreases, the bandgap increases. This means that we could use quantum dots of different sizes in a tandem stack, with the smallest dots absorbing the high-energy light and the largest dots absorbing the lower energy light (figure 5). At the University of New South Wales, our group is working on silicon quantum dots, owing to the abundance and non-toxicity of silicon, and the huge amount of knowledge about silicon that has been and continues to be generated by the semiconductor industry. The dots are fabricated by depositing thin films of silicon-rich silicon dioxide or silicon nitride. Heating the films precipitates out the silicon in dots with diameters controlled by the width of the initial thin films. Good control of dot size and evidence of a size-dependent bandgap, as well as promising early results of electrical measurements have been achieved so far (Green et al. 2005; Conibeer et al. 2006).

Figure 5

A schematic diagram of the bandgaps of a silicon quantum dot tandem solar cell. The upper cell is formed from a ‘superlattice’ of small quantum dots with a large effective bandgap. The middle cell is formed from larger quantum dots with a smaller effective bandgap, and the lower cell is ordinary bulk silicon.

5. Cell design for nanostructured cells

A conventional silicon solar cell uses a p–n junction to separate electrons and holes. When we design solar cells at the nanoscale, we need to broaden our thinking beyond the p–n junction. Würfel has pointed out that the fundamental feature of a solar cell is not an electric field used to separate electrons and holes, but more generally, is one-way ‘membranes’ that allow only one type of carrier to pass (Würfel 2005). The membranes can take the form of an energy barrier but can also take other forms. The membranes are essentially an asymmetry in the device that leads to differences in the directions in which the electrons and holes are transported.

Two recent examples illustrate the very different cell designs that can be produced with this approach. Interestingly, like the metal nanoparticles used to improve light absorption, both of these new structures use metals in new ways, suggesting that metals may play a much expanded part in the nanostructured solar cells of the future.

In the first device, an organic dye is used to absorb light and generate electron–hole pairs (Lenzmann et al. 2005). The dye is located between a gold electrical contact and semiconducting titanium dioxide, and electrons travel to the titanium dioxide while holes travel to the metal (figure 6a). It is surprising that this device works, because normally it would be expected that the electrons and holes would recombine efficiently at the metal surface. The reason that they do not is because injection of electrons into the titanium dioxide is extremely fast (Moser & Gratzel 1998). Thus, the asymmetry in this device comes from the difference in the transfer rates of electrons between the gold and the titanium dioxide. This ultra-fast injection of electrons is also important for standard dye solar cells, which have a similar cell design but with an electrolyte between the dye and the metal. Thus, there does not actually need to be an energy barrier to create a membrane, an asymmetry in the transfer rates can be sufficient.

Figure 6

New solar cell designs that use different asymmetries to the p–n junction. (a) Ultra-fast transport of electrons into the titanium dioxide, represented by the thick arrow, reduces the probability of electrons and holes recombining at the metal. (b) Ballistic transport for electrons and conventional transport for holes creates the asymmetry in this device.

The second device also involves gold, an organic dye and titanium dioxide, but this time in a different order, and its operation is just as surprising (McFarland & Tang 2003). Here, the gold is located between the organic dye and the titanium dioxide (figure 6b). As in the first device, the organic dye is used to absorb light and generate electron–hole pairs. However, in this device, the photo-excited electrons then travel ballistically (i.e. without scattering) across the thin gold film and into the semiconducting titanium dioxide layer. Holes travel through the gold film in the usual, non-ballistic way since there is no level in the titanium dioxide available at the energy of the holes. The asymmetry in this device comes from the fact that the electrons travel ballistically through the gold and subsequently travel through the titanium dioxide to the titanium contact, while the gold forms a conventional contact for the holes.

6. Discussion

In 1939, diode action was discovered in a crystal of germanium with an accidentally grown-in p–n junction, and up until recently, solar cell design has mostly been based on incremental improvements to this structure. While great gains in efficiency and reductions in cost have been achieved, the principles of device operation have not changed. It is only now we are realizing that the most fundamental property of a solar cell is its asymmetry in the transport of electrons and holes, and that this asymmetry can take a variety of forms. Having recognized this, we can go looking for asymmetries in structures that could be easy and cheap to fabricate. Furthermore, since solar cells up until now have had large-scale features, a geometrical treatment of how light travels through the device has been sufficient. In the various possible applications of surface plasmons described in this paper, I have only scratched the surface of how making use of the wave nature of light might be able to improve photovoltaics.

In recent years, our ideas on the properties that materials can have and even the fundamental requirements for a solar cell have been radically revised. The examples described above are just a few of the new nanostructures that have potential for photovoltaics; there are many others, but this gives a flavour for the new physics that we see and the new effects we can incorporate when designing devices at the nanoscale. These nanostructures can be fabricated by simple self-assembly techniques, making the structures potentially low cost. With the new nanomaterials available, the question of the optimum design for a solar cell becomes wide open. Progress in the field is very rapid, and we can expect more surprises as photovoltaics develop into a major contributor to a sustainable energy future.


K.R.C. acknowledges the support of an Australian Research Council fellowship, and also the support of the Centre of Excellence for Advanced Silicon Photovoltaics and Photonics, supported by the Australian Research Council.


  • One contribution of 23 to a Triennial Issue ‘Mathematics and physics’.


View Abstract