Solar Cells and Photovoltaics

  • Stuart Irvine
Part of the Springer Handbooks book series (SPRINGERHAND)


Photovoltaic solar cells are gaining wide acceptance for producing clean, renewable electricity. This has been based on more than 40 years of research that has benefited from the revolution in silicon electronics and compound semiconductors in optoelectronics. This chapter gives an introduction into the basic science of photovoltaic solar cells and the challenge of extracting the maximum amount of electrical energy from the available solar energy. In addition to the constraints of the basic physics of these devices, there are considerable challenges in materials synthesis. The latter has become more prominent with the need to reduce the cost of solar module manufacture as it enters mainstream energy production. The chapter is divided into sections dealing with the fundamentals of solar cells and then considering six very different materials systems, from crystalline silicon through to polycrystalline thin films and perovskites. These materials have been chosen because they are all either in production or have the prospect of being in production over the next few years. Many more materials are being considered in research and some of the more exciting, excitonic cells and nanomaterials are mentioned. However, there is insufficient space to give these very active areas of research the justice they deserve. I hope the reader will feel sufficiently inspired by this topic to read further and explore one of the most exciting areas of semiconductor science. The need for high-volume production at low cost has taken the researcher along paths not normally considered in semiconductor devices and it is this that provides an exciting challenge.

Photovoltaic (PV ) devices are a method of converting radiant solar energy into electrical energy. Most of our energy sources, including fossil fuels, hydroelectric and wind power actually come from solar radiation but are indirect conversions into electricity. Another class of solar energy conversion is the heating of water in solar thermal panels. Although the conversion efficiency can be high, they do not generate the thermal energy necessarily where and when it is needed, so storage is required. Direct generation of electric energy is attractive because it is a versatile energy form, rapidly converted into heat, mechanical or light energy. Photovoltaic energy is the main source of energy in the rapidly expanding satellite market with high-efficiency photovoltaic modules producing more than 1 kW of power. Terrestrial applications are also rapidly growing with an estimated total installed capacity worldwide of over 177 GW at the end of 2014 [43.1], and this has been increasing annually by 30%. This, however, is small compared with the total amount of electrical energy consumed each year, but is now supplying more than 1% of primary energy demand. Solar energy is very attractive as it is completely nonpolluting and can help to reduce the amount of fossil fuels that we burn to generate electricity. World CO2 emissions grew by 4% a year up to 2012. Any contribution from non-fossil-fuel alternatives such as solar energy will help to reduce this annual burden of CO2 emissions that is now a widely accepted cause of global warming. So how much solar energy is available for conversion to electricity? The total solar energy falling on the Earth’s surface each year is huge and 10000 times the current consumption of energy. We only need to capture a tiny fraction of this to make a major contribution to our electricity supply but this will mean incorporating solar electric panels into most buildings; to achieve this it will need to be much cheaper to compete with current fossil-fuel electricity generation.

The range of applications outlined here place different demands on the design of the photovoltaic module and hence the materials solutions may well be different. Cost has already been mentioned as paramount for terrestrial power requirements, but for space applications resistance to ionising radiation is important. This section will consider the materials, the implications of competing technologies and how well they meet the criteria for different applications. The first section will deal with how a photovoltaic cell operates and introduces the figures of merit that are used to compare photovoltaic cells and relate performance to theoretical performance. Solar cell technology is promoted as a clean fuel technology that does not introduce CO2 and pollutants into the atmosphere. However, for this to be a truly environmentally friendly technology, the manufacturing and eventual disposal will also have to be environmentally safe and this depends on the materials used and the fabrication technologies. These factors will be considered for each material system.

43.1 Figures of Merit for Solar Cells

In this section we will consider the operation of a photovoltaic cell and the significant parameters that characterize its performance. The objective of a photovoltaic cell is to capture as much of the solar energy as possible and convert this into electrical energy. The solar energy reaching the Earth’s atmosphere fits a black-body distribution for a body at 5800 K. This spectrum becomes highly structured, particularly in the infrared part of the spectrum, by absorption bands due to atmospheric gases. By the time the solar radiation reaches the Earth’s surface, it no longer fits a black-body distribution [43.2]. The shift in the spectral distribution will obviously affect the efficiency of absorption of the solar radiation, particularly considering that all semiconductor materials will display a cutoff wavelength dictated by the bandgap of the semiconductor. It is necessary to specify the atmospheric absorption when quoting efficiency, as the depth of atmosphere that the solar radiation has passed through will affect both the spectral distribution and the total amount of energy. The measurement used is air mass (AM) and is defined as 0 for solar radiation outside the atmosphere and 1 for radiation reaching the ground when the Sun is at its zenith. For shallower angles the solar radiation has to penetrate a larger depth of atmosphere and the AM is therefore going to be greater than 1. For space applications AM 0 is the appropriate condition and for terrestrial applications (depending on the position on the Earth) AM 1.5 is a typical value quoted, which corresponds to a solar angle of 48. All this assumes that there is no cloud cover, which will further reduce the integrated intensity and modify the solar spectrum. Although cloud cover will reduce the amount of solar energy available for conversion, useful amounts of electricity can still be generated.

The available solar energy also decreases from 1367 W ∕ m2 outside the atmosphere (AM 0) to 1040 W ∕ m2 for AM 1, when the sun is directly overhead. In practice, for terrestrial applications the available solar energy is considerably less than this where, in general, AM > 1 and is further reduced by cloud cover.

A photovoltaic cell is basically a diode with a photogenerated current. A band diagram is shown schematically in Fig. 43.1. Absorption of radiation can occur on both sides of the junction, creating minority carriers that can diffuse towards the junction. A photocurrent is generated if the minority carriers can drift across the junction without recombination. In practice, the junction is shallow and absorption will occur predominantly on one side where there is a greater depth of absorbing material. There are also heterojunction p-n devices where less absorption will occur in the wider-bandgap layer, as shown in Fig. 43.2. So, one side of the junction is considered to be the absorber layer and it is the spectral absorption characteristics of this layer that will determine the maximum absorption possible from the available solar radiation.
Fig. 43.1

Schematic of an energy-band diagram for a p-n junction solar cell showing photoabsorption to create an electron-hole pair and diffusion of the electron towards the junction

Fig. 43.2

Band diagram for a heterojunction PV cell showing illumination through the wide-bandgap window layer

An ideal diode characteristic for an illuminated cell is shown in Fig. 43.3 . In the dark the JV plot would go through the origin and, as the light intensity increases, the short-circuit current becomes increasingly negative, indicating the presence of a photogenerated current. The equation for the JV characteristic of this ideal device is
where Js is the saturation current in reverse bias under zero illumination, q is the charge on the carrier, V is the applied voltage, kB is Boltzmann’s constant, T is the temperature of the cell and JL is the photogenerated current. In the ideal cell this is equal to the short-circuit current, indicated as Jsc on the JV curve in Fig. 43.3. Power can be extracted from the device in the +V, −J quadrant of the JV plot and the load will determine the operating point in this quadrant. The power is determined from the product JV at this operating point, shown graphically in Fig. 43.3. The maximum power will correspond to the operating point that will give the largest JV area on this graph. For the ideal diode characteristics given in (43.1), this maximum power is given by the following (43.2)
$$\begin{aligned}\displaystyle P_{\mathrm{m}}&\displaystyle=J_{\mathrm{m}}V_{\mathrm{m}}\\ \displaystyle&\displaystyle=J_{\mathrm{m}}\left[V_{\text{oc}}-\frac{k_{\mathrm{B}}T}{q}\ln\left(1+\frac{qV_{\mathrm{m}}}{k_{\mathrm{B}}T}\right)-\frac{k_{\mathrm{B}}T}{q}\right]\\ \displaystyle&\displaystyle=J_{\mathrm{m}}\left(\frac{E_{\mathrm{m}}}{q}\right),\end{aligned}$$
where Em is the maximum energy that can be extracted per photon and depends on the band parameters for the semiconductor absorber layer, which determine Voc and Vm. These parameters are marked on the JV plot in Fig. 43.3. We now have two fundamental parameters which will limit the efficiency of the cell:
  • The fraction of solar photons absorbed in the cell

  • The electrical energy created per photon.

The first factor can be calculated by integrating over the solar spectrum for the appropriate AM number and including the cutoff wavelength of the semiconductor absorber layer.
This is shown graphically in Fig. 43.4 for AM 1.5. Any photons with energy less than the bandgap will not be absorbed and will not contribute to the photocurrent.
Fig. 43.3

Ideal JV characteristic for a photovoltaic cell, according to (43.1) with Js equal to 30 mA ∕ cm2. The parameters Jsc and Voc are indicated on the graph. The maximum extracted power is shown as the shaded area

Fig. 43.4

Graphical representation of the maximum energy that can be extracted from a CdTe solar cell with a bandgap energy of 1.45 eV

The second efficiency factor mentioned above implies that not all the photon energy will be converted into electrical energy, even if one photon absorbed constitutes one minority carrier crossing the junction. The electrical energy per carrier is given by the factor Em in (43.2), so the maximum power of the device is the product of the absorption rate of photons and the mean electrical energy created per photon. This product is represented by the inner shaded area of the solar spectrum shown in Fig. 43.4. The difference between curve 1 and curve 2 is simply the energy lost per photon because not all the photon energy is converted into electrical energy. Different semiconductor materials will have different efficiencies primarily because of different values for the bandgap. The ideal value for Em will track the bandgap, so for narrower-bandgap materials there will be a larger proportion of photons absorbed but less electrical energy per photon. The function of efficiency for semiconductors with different bandgaps, taken from curve 2 in Fig. 43.4, is plotted in Fig. 43.5 and shows that the optimum efficiency occurs for semiconductors with a bandgap in the near-infrared region, around 1.5 eV. This represents the best compromise between absorption of solar radiation and transferring the optimum amount of energy per photon into electrical energy.
Fig. 43.5

Plot of ideal efficiency against bandgap energy for a single-junction cell for AM 1.5 illumination conditions. (After [43.2])

The maximum efficiency is predicted for Si, InP, GaAs and CdTe, which are in the region of 30% for AM 1.5 irradiation (1 sun illumination) and is called the Shockley–Queisser limit  [43.3]. This means that the very best conversion efficiency for a single-junction cell is approximately one third of the available solar energy. In practice, photovoltaic cells have efficiencies considerably less than this due to optical reflections, poor junction characteristics and carrier recombination. These are materials issues that are often traded off against production costs, for example using polycrystalline rather than single-crystal material. Conversely, higher conversion efficiencies can be achieved using multiple junctions, which are more expensive but attractive for space applications and when used in combination with solar concentrators, so less surface area of the expensive multijunction cell is required. Other factors that can influence the choice of photovoltaic materials include the following:
  • Absorption coefficient

  • Contact resistance

  • Abundance of raw materials

  • Toxicity of materials

  • Stability of materials and junctions

  • Radiation resistance.

These factors will be considered in the following examples of different photovoltaic systems to assess the merits of different materials. It is probably true to say that there is not one ideal material but different applications can make one material more attractive than another.

43.2 Crystalline Silicon

Over 85% of the current world production of solar modules are made from either single-crystal or multigrain silicon. This is the most mature of the photovoltaic materials and has benefited considerably from the size of the silicon semiconductor industry. This has ensured a ready supply of material and processing tools suitable for large-scale production. However, crystalline silicon does suffer from a fundamental disadvantage in that the bandgap of silicon is indirect, which means that the absorption coefficient is much lower than a direct-bandgap semiconductor such as CdTe. In practice this means that a much thicker piece of material is needed to absorb all the solar radiation greater than the bandgap energy. This requires wafers of silicon thicker than 100 μm and means that this material is not suitable for thin-film technology. The highest performance solar electric modules are made from wafers of single-crystal silicon cut from Czochralski-grown crystals, up to 30 cm in diameter. This was originally developed on the back of the silicon electronics industry, but the rapid increase in production volume of PVs is now driving production of silicon crystals. The conversion efficiency for single-crystal PV modules is around 20% with the potential for increases in the near future to 22%.

Multicrystalline silicon involves the relatively cheap path of casting silicon ingots that are not seeded but produce a random grain size of the order of 1 cm across. The ingots are cast in blocks, larger than 100 kg, and then sliced into wafers 300 μm thick and with an area of up to 20 cm2. The grain boundaries cause these wafers to be mechanically weaker than the single-crystal wafers and they will typically be thicker. This causes some trade off in price. Each cell can be expected to contain grain boundaries, so loss of photogenerated charge at the grain boundary can cause some loss of efficiency. Typical module efficiencies for multicrystalline silicon are currently around 17%.

The junction is formed by phosphorus implantation to form a p-n homojunction and contacted by printing of a metal grid, usually of a Ni-Au alloy. Patterning of the surface prior to implantation and contacting has achieved the highest efficiency cells by improving light collection [43.4]. Another recent innovation that has contributed to higher efficiencies is the V-grove or buried junction. This is shown schematically in Fig. 43.6 and entails the grooving of the p-type silicon substrate, implantation of phosphorus to create a buried junction and filling with metal contact alloy. This helps to improve the collection efficiency of the cell, particularly in the blue part of the spectrum. These manufacturing steps have to handle large volumes and be cheap. Cheaper alternatives to ion implantation involve thermal diffusion from a screen-printed paste or spin-on glass [43.5].
Fig. 43.6

Schematic of buried-contact technology for Si solar cells

The main disadvantage of silicon is that the absorption coefficient is low because it is an indirect-bandgap semiconductor (\({\mathrm{2\times 10^{3}}}\,{\mathrm{cm^{-1}}}\) for Si compared with \({\mathrm{1\times 10^{5}}}\,{\mathrm{cm^{-1}}}\) for CdTe in the green part of the spectrum). This means that the amount of material needed to absorb the solar radiation is greater than for a direct-bandgap semiconductor. The absorption can be improved by having a reflecting back contact that, if it is roughened, will increase the mean path length of the back reflected light in the silicon cell. This is particularly important for thin-film silicon devices where the amount of material is more of an issue, although the thickness of 30–100 μm can be compared with the much thinner layers used for direct-bandgap semiconductors (see CdTe and GaAs).

Thin-film polycrystalline silicon is grown by ribbon casting techniques. The pulling speeds are in the range of 10–1800 cm ∕ min, and this represents a cheap method for the production of solar cell material but the conversion efficiencies are currently low. One way of increasing the absorption coefficient is to deposit a film of amorphous silicon (a-Si), which has a larger bandgap than crystalline silicon but is a direct-bandgap material. Amorphous silicon has led the way in cheap thin-film solar cells but suffers the disadvantages of low efficiency (< 10%) and poor long-term stability. These cells will be discussed in greater detail in the next section.

Crystalline silicon solar cells, over the last ten years, have approximately doubled in efficiency and this has been achieved by a combination of the following:
  • Improved material quality, leading to improved minority-carrier diffusion length.

  • Improved Voc and fill factor through emitter and base doping and contact optimization.

  • Improved Jsc through diffusion length improvement using phosphorus gettering, hydrogen passivation and buried contacts.

  • Surface passivation and contact grid optimization.

All these aspects involve materials issues, either directly associated with the quality of the silicon or with contacting and passivation. It is also important to avoid degradation of the cell and the final encapsulation must avoid exposure of the cell to water. Multicrystalline silicon can be passivated with silicon nitride deposited by plasma-assisted CVD (CVD ) from SiH4 and NH3 to reduce surface recombination [43.6].

Silicon solar modules are becoming cheaper to produce, nontoxic and stability in a nonradiation environment is good. Multicrystalline silicon is not so suitable for space applications because of the combination of thick absorber layers, requiring greater weight per unit area and the sensitivity of the cell to degradation in a high-radiation environment. However, single-crystal silicon has dominated the space market, but it has now been replaced by higher efficiency III-V multijunction cells. Single-crystal modules for terrestrial applications produce the highest available output power with over 300 W (AM1.5) per panel.

43.3 Amorphous Silicon

Amorphous silicon offers the potential for a cheap production technology for terrestrial photovoltaic solar cells. The amorphous state displays different physical properties to the crystalline with a modified band structure. One consequence of this is that the absorption coefficient in the green part of the visible spectrum is a factor of ten higher at \({\mathrm{2\times 10^{4}}}\,{\mathrm{cm^{-1}}}\), which makes it more suitable for thin-film applications. The amorphous structure leaves dangling bonds, which pin the Fermi level and would normally prevent doping of the material. This is overcome by the inclusion of hydrogen, which passivates these dangling bonds, so the material is referred to as a-Si:H. The Si can also be alloyed with Ge, C, and N in a glow-discharge evaporator. These alloys are particularly useful for multijunction devices used for increasing the solar collection efficiency. In the laboratory single-junction cells have an efficiency around 10% [43.7] with multijunction cells going up to over 13%, where the a-Si junction is combined with nanocrystalline Si junctions [43.8]. In production an a-Si module has an efficiency of 7–10%, considerably less than the crystalline materials.

The most common method for producing a-Si:H for photovoltaics is by plasma-enhanced CVD from SiH4 mixtures. The films are deposited onto a textured conducting oxide such as indium-tin oxide (ITO ), which provides the electrical contact and increases the average light path in the absorber layer to increase absorption. The device structure is a p-i-n with absorption taking place in the middle (insulator) layer, which is only 0.5 μm thick. The p and n layers can be deposited by adding B2H6 or PH3 respectively to the plasma.

One of the major disadvantages of a-Si:H is the instability and long-term deterioration under light illumination. This is caused by the rearrangement of dangling bonds, often associated with rearrangement of hydrogen atoms close to weak or dangling bonds. The energy comes from nonradiative bimolecular reactions and hence depends on the illumination intensity (for further details see the review by Bloss et al. [43.9]). In practice, this causes a downward drift in efficiency with time, which can be as much as 2% in 100 h. The cost of a-Si thin-film modules ten years ago was cheaper than crystalline silicon (3 US$/Wp) but has not kept track with falling prices and larger manufacturing volumes of crystalline silicon modules [43.10, 43.11]. One of the main technical challenges is to maintain stabilized efficiency above 10%. The relatively high cost of PV solar modules compared with more conventional sources of energy can only become attractive if it can operate efficiently over a long period of time. This means that the modules must remain efficient over a period of 20 years. Shorter operating lifetimes would have to be mitigated by much lower production cost.

In conclusion, a-Si:H is a low-cost technology for terrestrial applications and has found its way into low-power and other more bespoke applications such as small-scale standalone systems. The temperatures used in processing these modules is lower than the high temperatures needed to melt silicon for the crystalline silicon modules, which means that the energy payback times and carbon footprint are less. However, longer-term stability will affect the amortized cost of electricity compared with crystalline silicon PV modules.

43.4 GaAs Solar Cells

The III-V semiconductors have advantages over silicon due to their direct-bandgap photon absorption, with an absorption coefficient in the green of \({\mathrm{8\times 10^{4}}}\,{\mathrm{cm^{-1}}}\). This means that, for example, GaAs can theoretically yield over 30% efficiency (AM 1.5) for an absorber layer thickness of just 1–2 μm, compared with a hundred times this for crystalline silicon. The efficiency, stability and thin-film deposition technology make these cells attractive for space applications. From Fig. 43.5 it is clear that the bandgap of GaAs is well matched to the optimum for maximum efficiency. It is also noteworthy that the highest efficiency single-junction solar cell is epitaxial GaAs with 28.8% AM1.5 efficiency [43.12]. The III-V semiconductors also offer the flexibility of alloying to change the bandgap and tune the response of the photovoltaic junction. In addition, heterojunctions can be formed and multijunction solar cells can be produced to convert more of the solar spectrum into electricity and thus exceed the theoretical limits set by the Shockley–Queisser limit. For example, in the laboratory a GaAs/GaSb tandem solar cell was reported with 35.6% efficiency as long ago as 1990 [43.13]. Higher efficiencies are possible with three and even four junctions to capture more of the infrared that would otherwise not be absorbed. The Ge substrate, which is closely lattice matched with GaAs, is commonly used as an infrared absorber to capture the radiation that passes through the GaInP2/GaAs cell [43.14, 43.15]. The top cell has an absorber of GaInP2, which has a bandgap of 1.9–2 eV, and will capture the visible part of the spectrum without too much voltage loss. The next cell is GaAs, which has a bandgap of 1.42 eV, and will capture the red to infrared part of the spectrum. The bottom junction is formed in the Ge substrate, which has a bandgap of 0.67 eV and will capture the light further into the infrared. The best triple-junction solar cell has achieved 37.9% efficiency under AM1.5 illumination [43.16]. The triple-junction cell, in production for space applications, can yield 29.5% (AM 0) conversion efficiency [43.17].

The layers of these multijunction cells are grown epitaxially onto single-crystal Ge substrates. As with crystalline silicon, high crystalline quality is needed to obtain high efficiency cells; polycrystalline GaAs does not work as well due to grain-boundary conduction that reduces the available photocurrent [43.18]. This restricts the photovoltaic applications to high-quality epitaxial material as GaAs is not suitable for cheap thin-film photovoltaics on glass substrates. The topic of thin-film polycrystalline photovoltaics will be considered in the next section on CdTe solar cells. Epitaxial growth avoids minority-carrier recombination at grain boundaries, by avoiding their formation, but the cost of the substrates is inherently higher than the glass or ceramic substrates used for thin-film devices. However, defects must also be avoided at the junction and interfaces between layers of different composition and this imposes the constraint that the heterostructures should be lattice matched. This restricts the choice of alloys to those that are either lattice matched, or so that the thickness and mismatch are such that the film is strained. The simplest structures use AlGaAs, which has a close lattice match to GaAs over the entire composition range, as a window and passivation layer. GaAs is also closely matched to Ge, offering a choice of substrates. The GaAs or Ge substrate is the narrowest bandgap part of the structure, so the device has to be front-side illuminated with a grid of metal contacts on the top surface to contact the cells. The design of the cell structure needs to have the wider-bandgap layers further from the substrate (last to be grown) to allow the longer wavelengths to penetrate to the narrow-bandgap absorber layer. This places further constraints on the design of these epitaxial structures.

The Ge and GaAs substrates have a similar lattice parameter to AlGaAs, which is used as a passivation layer for both n-type and p-type GaAs layers. Other III-V compounds that can be used include GaSb, which is sensitive to the near infrared out to 1.8 μm and will therefore absorb the radiation that is transmitted through the GaAs cell, as an alternative to the Ge-bottom cell. The alloy GaInP also lattice matches for the 50% Ga/In mix and is useful as a wider-bandgap junction. GaSb is also attractive for thermo-photovoltaics where the solar radiation is converted into heat, which is then absorbed by the narrow-bandgap GaSb device. The narrow-bandgap quaternary GaInNAs has been investigated [43.19] as a narrow-bandgap absorber layer and can be inserted between the GaAs layer and the Ge substrate to give a small boost in voltage and overcome the disadvantage of the excess current that would be produced by the Ge bottom cell.

The push to achieving further increases in multijunction solar cell efficiency is driving new approaches to increasing the number of cells without being constrained by lattice mismatch. One of these approaches is the inverted metamorphic structure (IMM ) where lattice mismatch is graded out using a buffer layer but to avoid compromising the top GaInP junction with threading dislocations this is grown as an inverted structure onto lattice matched GaAs and then the substrate is removed [43.20]. By this way the highly mismatched bottom junction of In0.27Ga0.73 As is grown last but can provide better a bandgap match, giving a boost in efficiency. This is opening the way for better current matching and optimizing the bandgap for increasing the efficiency of multijunction cells and moving towards achieving greater than 50% conversion efficiency.

The III-V photovoltaic structures are grown by metalorganic vapor-phase epitaxy (MOVPE ), which gives excellent control over alloy composition and doping concentration. An example of a multijunction cell is shown in Fig. 43.7. The structures tend to be complex because a tunnel junction is needed for current to flow between the two (series) photovoltaic cells, otherwise the connection between the two would be a high-impedance reverse-bias junction. It is also important for the device design to match the currents between the two (or more) junctions so that they can both operate at near-optimum conditions.
Fig. 43.7

Schematic of a triple-junction GaAs solar cell, showing the positions of three junctions. (After [43.13])

GaAs-based solar cells are now dominating the space market to power satellites. Their efficiency is high and stability is good, but the single-crystal Ge substrates and complex epitaxial layer growth leads to high cost. There may be toxicity issues for the disposal of solar modules if GaAs was used widely for terrestrial applications. The most attractive terrestrial application would be in concentrators where the solar radiation is optically concentrated onto the PV cells, so the collection area can be much greater than the area of expensive PV modules. This is looking like an attractive and cost-effective option for regions of the world with a high proportion of direct sunlight [43.21].

43.5 CdTe Thin-Film Solar Cells

The bandgap of CdTe of 1.45 eV at room temperature makes it another semiconductor that is close to the theoretical optimum for efficient conversion of solar radiation (Fig. 43.5) into electrical power. The absorption coefficient is \(> {\mathrm{5\times 10^{4}}}\,{\mathrm{cm^{-1}}}\) for photon energy greater than the bandgap, allowing efficient collection for only 2 μm-thick films, similar to GaAs. CdTe, like GaAs is also a direct-bandgap semiconductor but is from the II-VI family of semiconductor compounds. The theoretical efficiency for the CdS/CdTe photovoltaic cell is ≈ 30% at AM 1.5, the world record efficiency has increased from nearly half this maximum, in 2010 to 21.5% in 2015 [43.22] and for modules in production the efficiency has increased to over 14%. The attraction of CdTe compared with GaAs is that these cells can be made from polycrystalline thin films on glass substrates, thus avoiding the need for expensive single-crystal substrates. A further advantage in using glass substrates is that illumination of the photovoltaic cell occurs through the substrate rather than from the top face, so the substrate becomes the window for the cell. The front contact is made with a transparent conducting oxide (TCO ) such as ITO, as shown in Fig. 43.8, and this avoids the need for an opaque grid of metal contacts. This approach of using a TCO on glass substrate as the window is called a superstrate. The TCO has to have high optical transmission as well as a metal-like electrical conductivity. A typical thin-film resistivity would be 10 Ω per square.
Fig. 43.8

Schematic of a CdTe device structure based on a glass superstrate

The CdTe solar cell is made from a heterojunction between the wider-bandgap CdS and the CdTe absorber layer. CdS is often referred to as the window layer; it has a bandgap of 2.4 eV allowing most of the visible spectrum to pass with very little absorption. The most common method for depositing CdS is by chemical bath deposition (CBD ). The layer is n-type, forming a junction with the p-type CdTe. The thickness is kept to a minimum (around 100–200 nm) in order to minimize absorption at the blue end of the spectrum. A schematic of the CdTe PV structure is shown in Fig. 43.8.

Deposition techniques for CdTe (electrodeposition or close-space sublimation (CSS )) require postdeposition treatment for doping, grain growth and resistivity reduction, at temperatures above 400C. The exact processes are not clear but have a role in providing grain boundary passivation, improving efficiency from a few % to > 10% if this anneal is carried out in air in the presence of CdCl2 [43.23]. There are variations in this process in terms of temperature, time and ambient gas but all involve the diffusion of Cl into the CdTe and some interdiffusion of the CdS/CdTe interface. This process improves grain size to ≈ 1 μm if the starting material has a much smaller grain size. This is further complicated by the variations in grain size from the CdS interface up to the back surface, where the junction region displays smaller grains. The best efficiency in the laboratory, which stood for ten years since 2001, was 16.5% reported by Wu [43.24] and in 2015 rose to 21.5% [43.22].

A typical CdTe module with an area of 0.72 m2 has a peak output power of 110 W; this compares favorably with other thin-film photovoltaics and approaches that of multicrystalline modules [43.25]. The advantage for CdTe thin-film PVs over the maturer crystalline silicon PVs is the smaller temperature coefficient where conversion efficiency decreases with increasing module temperature by −0.5% for crystalline silicon and −0.25% for CdTe modules [43.26]. Future prospects for CdTe can be seen in current research activities with encouraging results for CdTe solar cells on flexible plastic substrates [43.27] and more recently both superstrate on thin polyimide and substrate devices on flexible metal foils [43.28]. Using these substrates places tighter constraints on deposition and annealing temperatures and alternatives to the CdCl2 anneal would be attractive, particularly considering the toxic nature of CdCl2. A nontoxic approach has been proposed by Major et al. [43.29] using MgCl2, which achieves the same device improvement as CdCl2 but is nontoxic and costs less than 1% of the cost of CdCl2.

Another area for improvement in CdTe solar cells is in the p-type doping, which has relied on native defects or a controlled Cu diffusion [43.30]. An alternative approach is in situ doping of CdTe with arsenic using metalorganic chemical vapor deposition (MOCVD ) [43.31, 43.32]. Active acceptor doping concentrations as high as \({\mathrm{5\times 10^{15}}}\,{\mathrm{cm^{-3}}}\) have been demonstrated and these devices show excellent stability. MOCVD offers greater flexibility over deposition conditions and this has been utilized with CdTe solar cell structures on ultra-thin glass superstrates for ultra-lightweight PV solar cells for space [43.33].

A concern with CdTe is the abundance of tellurium and the volume of available supply to feed a large expansion in the manufacturing of CdTe PV modules. The abundance figures look low but up to 5000 t of Te could be produced each year from copper ore and 20000 t of Cd is produced each year as a by-product of zinc production. An important environmental and resource issue may be the recycling of these materials. It is important to note that although Cd is a very toxic metal it is relatively benign in the form of CdTe as it is a very stable compound and is not soluble in water. The use of Cd in solar cells has the added benefit of making use of the Cd that is produced as a by-product of zinc production in an environmentally friendly way. Jager-Waldau [43.10] states that Every energy source or product may present some environmental, health and safety hazard, and those of CdTe should by no means be considered barriers to technology scaling. This technology is less well developed than those in the other case studies but the thin-film technology on glass combined with good stability makes this a very attractive proposition for both terrestrial and space applications.

43.6 CuInGaSe2 (CIGS2) Thin-Film Solar Cells

This is an alternative material system for achieving an optimum bandgap for solar energy conversion and one that has attracted a lot of research effort in the past 20 years with impressive improvements in conversion efficiency. The highest efficiency reported for laboratory solar cells back in 2003 was 19.2%, which then was the best of the thin-film PV technologies [43.34]. This has risen to the current world record in 2015 of 22.3% by the leading CIGS module manufacturer, Solar Frontier [43.35]. In common with other thin-film PV technologies, these films can be deposited onto cheap substrates, at relatively low temperature, and the potential for processing in large volumes. The absorber layer is based on the Chalcopyrite-phase CuInSe2 (CIS). This has a bandgap of 1.04 eV, which is lower than the bandgap for optimum efficiency in the Shockley–Queisser limit. However, cells made from CIS have achieved > 10% efficiency. Alloying with Ga to form CuIn1−xGa x Se2 (CIGS) increases the bandgap to 1.7 eV for x = 1. This gives a range of alloy composition across the important optimum range for efficient conversion. An increase in bandgap will increase the voltage of the cell (Voc) but decrease the number of absorbed photons, thus decreasing Jsc. In practice the alloy is not uniform and thus a greater proportion of the photon flux is absorbed by the single-junction device than would occur for a single wide-bandgap absorber layer [43.36]. The other benefit of grading of the alloy composition is that the resultant bandgap grading creates a built-in electric field that can drift the electrons towards the junction.

Unlike a-Si and CdTe, the preferred arrangement is the substrate configuration where the cell is illuminated from the top surface as with the crystalline GaAs cells. The typical layer structure is shown in Fig. 43.9. The substrate is soda-lime glass with a sputtered coating of molybdenum, which acts as the back contact. The next layer is the CuIn1−xGa x Se2 alloy, which is the p-type absorber layer. The junction is formed with a thin layer of CdS (as with CdTe). The front contact is formed by aluminum-doped ZnO, which is highly conducting but transparent, allowing solar radiation to pass without significant absorption. One disadvantage of the substrate approach is that the top surface is protected with another glass sheet, or other suitable transparent material, when it is fabricated into a solar PV module. This step is not necessary in the superstrate approach. However, the substrate configuration for CIGS has shown much higher maximum efficiency than the superstrate approach (favored by CdTe and a-Si) and provides the opportunity to deposit the CIGS layers onto flexible metal foils. The structure for CIGS on metal foil substrates is similar to those on the Mo/glass substrates except for a barrier layer of SiO2 between the metal foil and the Mo back contact. This is needed to reduce the diffusion of impurities into the active semiconductor layer. However, one impurity, sodium, is needed to give the desired doping properties for the CIGS absorber layer [43.37]. This impurity occurs naturally when cheap soda-lime substrates are used, as the sodium will diffuse from the glass substrate into the CIGS layer during film deposition and annealing. Using metal foil substrates makes the PV cells flexible and enables their manufacture in a continuous reel-to-reel process.
Fig. 43.9

Schematic of CIGS device structure based on a glass substrate

There are various deposition methods that can be used for the deposition of CIGS and this can lead to low-cost production routes. The early results were obtained by co-evaporation from separate elemental sources. More recently a range of techniques have been used from e-beam evaporation of the metal sources followed by selenisation and annealing to electrodeposition. The characteristic of CIGS synthesis approaches is a separation of the deposition process and alloy formation, allowing cheap and potentially high-throughput techniques to be used. This also enables control of the alloy composition and stoichiometry that will affect the doping [43.38]. The subsequently deposited junction layer, CdS, can be deposited by either chemical bath deposition, sputtering or CVD. The transparent contact layer, ZnO, is typically deposited by sputtering. All these techniques can be operated on a large industrial scale and lend themselves to cheap production methodology.

CIGS solar cells have moved rapidly into production and a number of manufacturers are offering modules with efficiencies in the region of of 14%. For example, Avancis produces a CIGS panel with a peak output power of 140 W for a module area of 1.05 m2 and Soler Frontier produces modules with a peak power output of 170 W for 1.228 m2 modules. The potential for further price reduction is similar to other thin-film technologies and competitive with crystalline silicon.

As with CdTe solar cells, the abundance of the elements In and Se appears low, but for a thin-film photovoltaic, where each module only needs a few grams of material, the total amount needed is not huge. However, the current price of In is high as it is not readily produced as a by-product of other mining processes. Price volatility with indium supply could significantly affect the future price of these modules and the cost of the active materials could contribute nearly a half of the overall module production cost [43.39]. Concerns about materials cost and abundance of materials has stimulated research into alternative chalcogenide absorber materials to replace CIGS in the longer term.

One candidate is the Kesterites copper zinc tin sulfide/selenide (CZTS) with the formula unit Cu2ZnSnS4 [43.40]. With a bandgap of 1.45 eV, CZTS has the right characteristics for an efficient single junction absorber material. The synthesis routes are similar to those for CIGS with a preference for formation of the metal precursor through evaporation followed by sulphidisation [43.41]. To achieve the highest conversion efficiencies it has been necessary to alloy with selenium, which reduces the bandgap towards 1.2 eV. The best laboratory efficiency was reported by Wang et al. [43.42] with 12.6%, which is well below the CIGS record but shows the potential for high efficiency solar cells using Kesterites. This was achieved through alloying with selenium but it would be advantageous to avoid the less abundance selenium altogether. The best result for selenium-free CZTS is 8.4% reported by Shin et al. [43.43] and shows the potential for future development of this material towards low-cost thin-film PV module production.

43.7 Excitonic PV

The drive to find cheaper solutions to the conversion of solar energy has taken research down some interesting and unusual avenues. One of the more successful of the alternative approaches is the dye-sensitized cell, known as the Grätzel cell after its inventor Michael Grätzel [43.44]. This uses organic dyes to absorb the solar radiation and transfer electrons to a porous TiO2 surface, which is effectively the junction. Other approaches have looked at conducting polymers and polymer blends to form photovoltaic junctions. The generic term, excitonic PV, refers to the relatively strong excitonic binding energy of the excited state following photon absorption that binds the electron to the hole. For polymer blend solar cells the dissociation of the exciton is achieved at the interface with a high electron affinity material such as fullerene. The hole is then free to be transported via a hole collector, which is a doped conducting polymer such as PEDOT:PSS [43.45]. The conversion efficiency of this type of solar cell has risen sharply in recent years with the best laboratory efficiency over 10% [43.46]. Organic PV offers the potential for very low-cost production where low temperature processing will also lead to the benefit of a very low carbon footprint.

An exciting recent development with excitonic PV has been the rise of the perovskite absorbers, which were initially used to replace the organic dye in dye-sensitized solar cells [43.14]. A characteristic of excitonic PV is the short excitonic diffusion lengths, so efficient cells require a nanostructured material where an electron collecting interface is never far away. A remarkable property of the methylammonia lead tri-halogen perovskites was demonstrated by Liu et al. who showed that planar structures could be made, similar to conventional thin-film PV, and achieve high conversion efficiency of over 15% [43.47]. The world record (2015) currently stands at 21%, which was achieved by Michael Grätzel’s EPFL team [43.48]. There is still much work to be done on the perovskite solar cells, one of which is to overcome the moisture sensitivity and long term stability. Some of the research is aimed at replacing Pb as the metal center with Sn, which removes any toxicity issues. The potential is for the perovskite solar cells to be used in new applications such as color neutral, partially transparent PV glazing in building-integrated PV [43.49] and many other applications that can take advantage of their unique properties.

43.8 Conclusions

This chapter has introduced some basic concepts about photovoltaic solar cells and examples of some of the more common materials being used for solar-cell production. The area of research is huge with a number of specialist journals dedicated to solar energy and large international conferences held annually. Recent developments with crystalline silicon, III-V heterojunctions, thin-film PV and exciton solar cells have been summarized. These examples show the diversity of materials that can be used for solar PV conversion and different stages of maturity. These approaches will enable very cheap solar PV modules to become a reality in the next five to ten years and also diversify the applications of solar PV.

The largest and most mature production facilities are based on crystalline or multicrystalline silicon. These modules are also the most efficient for single-junction cells. The most efficient cells are made from multijunction GaAs on Ge substrates and are the most complex and most expensive. Most of the multijunction solar cell production is for the space market but increasingly is showing potential for use with concentrators for terrestrial applications. This is acceptable for standalone solar power generation but the largest challenge for large-scale terrestrial power generation is to integrate photovoltaic modules into buildings. The equivalent of a gigawatt power station would require, at least, of the order of 107 m2 of solar modules. In reality it would have to be much larger to account for the average power production being much less than the peak production. A modest installation would have approximately 100 m2 of solar panels so we would need more than 105 buildings. So, the goal has to be that every house and building has a PV façade or roof.

Large-scale implementation will require the modules to be architecturally acceptable, in appearance and size, and probably serve multiple functions such as keeping the rain out and heat in, and so on. This need is opening the opportunity for a range of materials and designs and ultimately flexible panels. On top of this the cost per watt of electricity produced needs to come down, probably by a factor of three based on current installed costs. The thin-film technologies are inherently cheaper but much of this cost benefit is not realized at present because the efficiencies and production volumes are lower than crystalline silicon. The current growth in PV solar energy after decades of research has become very rapid and this expansion is set to continue well into the future.


  1. 43.1
    S. Nowak: Report IEA-PVPS T1-27: (International Energy Agency, Paris 2015)Google Scholar
  2. 43.2
    C.H. Henry: J. Appl. Phys. B 51, 4494 (1980)CrossRefGoogle Scholar
  3. 43.3
    W. Shockley, H.J. Queisser: J. Appl. Phys. 32, 510 (1961)CrossRefGoogle Scholar
  4. 43.4
    J. Zhao, A. Wang, M.A. Green: Appl. Phys. Lett. 73, 1991 (1998)CrossRefGoogle Scholar
  5. 43.5
    L. Pirozzi, G. Arabito, F. Artuso, V. Barbarossa, U. Besi-Vetrella, S. Loreti, P. Mangiapane, E. Salza: Sol. Energy Mater. Sol. Cells 65, 287 (2001)CrossRefGoogle Scholar
  6. 43.6
    A.G. Aberle: Sol. Energy Mater. Sol. Cells 65, 239 (2001)CrossRefGoogle Scholar
  7. 43.7
    T. Matsui, H. Sai, T. Suezaki, M. Matsumoto, K. Saito, I. Yoshida, M. Kondo: Development of highly stable and efficient amorphous silicon based solar cells, Proc. 28th Eur. Photovolt. Solar Energy Conf., 2213, Paris (2013)Google Scholar
  8. 43.8
    S.W. Ahn, S.E. Lee, H.M. Lee: Toward commercialization of triple-junction thin-film silicon solar panel with > 12% efficiency, Proc. 27th Eur. Photovolt. Solar Energy Conf., 3AO5.1, Frankfurt (2012)Google Scholar
  9. 43.9
    N. Bernhard, G.H. Bauer, W.H. Bloss: Prog. Photovolt. Res. Appl. 3, 3 (1995)CrossRefGoogle Scholar
  10. 43.10
    A. Jager-Waldau: Sol. Energy 77, 667 (2004)CrossRefGoogle Scholar
  11. 43.11
    A. Goodrich, P. Hacke, Q. Wang, B. Sopori, R. Marg, M. Woodhouse: Solar Energy Mater. Solar Cells 114, 110 (2013)CrossRefGoogle Scholar
  12. 43.12
    B.M. Kayes, H. Nie, R. Twist, S.G. Spruytte, F. Reinhardt, I.C. Kizilyalli, G.S. Higashi: 27.6% conversion efficiency, a new record for single-junction solar cells under sun illumination, Proc. 37th IEEE Photovolt. Specialists Conf. (2011)Google Scholar
  13. 43.13
    L.M. Fraas, J.E. Avery: Optoelectron. Dev. Technol. 5, 297 (1990)Google Scholar
  14. 43.14
    A. Kojima, K. Teshima, Y. Shirai, T. Miyasaka: J. Am. Chem. Soc. 131, 6050 (2009)CrossRefGoogle Scholar
  15. 43.15
    N.H. Karam, R.R. King, M. Haddad, J.H. Ermer, H. Yoon, H.L. Cotal, R. Sudharsanan, J.W. Eldredge, K. Edmondson, D.E. Joslin, D.D. Krut, M. Takahashi, W. Nishikawa, M. Gillanders, J. Granata, P. Hebert, B.T. Cavicchi, D.R. Lillington: Sol. Energy Mater. Sol. Cells 66, 453 (2001)CrossRefGoogle Scholar
  16. 43.16
    K. Sasaki, T. Agui, K. Nakaido, N. Takahashi, R. Onitsuka, T. Takamoto: AIP Procs. 1556, 22 (2013)CrossRefGoogle Scholar
  17. 43.17
    B.C. Richards, Y. Lin, P. Patel, D. Chumney, P.I.R. Sharps, C. Kerestes, D. Forbes, K. Driscoll, A. Podell, S. Hubbard: Performance and radiation resistance of quantum dot multi-junction solar cells, Proc. 38th IEEE Photovolt. Specialists Conf. (PVSC) (2012) p. 158Google Scholar
  18. 43.18
    J.R. Sites, J.E. Granata, J.F. Hiltner: Solar Energy Mater. Solar Cells 55, 43 (1998)CrossRefGoogle Scholar
  19. 43.19
    J.F. Geisz, D.J. Friedman, J.M. Olson, S.R. Kurtz, B.M. Keyes: J. Cryst. Growth 195, 401 (1998)CrossRefGoogle Scholar
  20. 43.20
    J.F. Geisz, D.J. Friedman, J.S. Ward, A. Duda, W.J. Olavarria, T.E. Moriarty, J.T. Kiehl, M.J. Romero, A.G. Norman, K.M. Jones: Appl. Phys. Lett. 93, 123505 (2008)CrossRefGoogle Scholar
  21. 43.21
    P. Perez-Higueras, E. Munoz, G. Almonacid, P.G. Vidal: Renew. Sustain. Energy Rev. 15, 1810 (2011)CrossRefGoogle Scholar
  22. 43.22
  23. 43.23
    X. Wu, J.C. Keane, R.G. Dhere, C. Dehert, D.S. Albin, A. Dude, T.A. Gessert, S. Asher, D.H. Levi, P. Sheldon: Proc. 17th Eur. Photovolt. Sol. Energy Conf. II (2001) p. 995Google Scholar
  24. 43.24
    X. Wu: Sol. Energy 77, 803 (2004)CrossRefGoogle Scholar
  25. 43.25
    S. Hegedus: WIREs: Energy Environ. 2, 218 (2012)Google Scholar
  26. 43.26
    N. Strevel, M. Gloeckler: Photovoltaics Int. 17, 1 (2012)Google Scholar
  27. 43.27
    X. Mathew, J.P. Enriquez, A. Romeo, A.N. Tiwari: Sol. Energy 77, 831 (2004)CrossRefGoogle Scholar
  28. 43.28
    L. Kranz, S. Buecheler, A.N. Tiwari: Solar Energy Mater. Solar Cells 119, 278 (2013)CrossRefGoogle Scholar
  29. 43.29
    J.D. Major, R.E. Treharne, L.J. Phillips, K. Durose: Nature 511, 334 (2014)CrossRefGoogle Scholar
  30. 43.30
    L. Kranz, C. Gretner, J. Perrenoud, R. Schmitt, F. Pianezzi, F. La Mattina, P. Blosch, E. Cheah, A. Chirila, C.M. Fella, H. Hagendorfer, T. Jager, S. Nishiwaki, A.R. Uhl, S. Nuecheler, A.N. Tiwari: Nature Commun. 4, 2306 (2013)CrossRefGoogle Scholar
  31. 43.31
    S.J.C. Irvine, V. Barrioz, D. Lamb, E.W. Jones, R.L. Rowlands-Jones: J. Crystal Growth 310, 5198 (2008)CrossRefGoogle Scholar
  32. 43.32
    G. Kartopu, L.J. Phillips, V. Barrioz, S.J.C. Irvine, S.D. Hodgson, E. Tejedor, D. Dupin, A.J. Clayton, S.L. Rugen-Hankey, K. Durose: Prog. Photovolt. (2015), doi: 10.1002/pip.2668
  33. 43.33
    S.J.C. Irvine, D.A. Lamb, A.J. Clayton, G. Kartopu, V. Barrioz: J. Electron. Mater. 43, 2818 (2014)CrossRefGoogle Scholar
  34. 43.34
    K. Ramanathan, M.A. Contreras, C.L. Perkins, S. Asher, F.S. Hasoon, J. Keane, D. Young, M. Romero, W. Metzger, R. Noufi, J. Ward, A. Duda: Prog. Photovolt. Res. Appl. 11, 225 (2003)CrossRefGoogle Scholar
  35. 43.35
    T. Kamada, T. Yagioka, S. Adachi, A. Handa, K. Fai Tai, T. Kato, H. Sugimoto: Proc. Photovoltaic Specialists Conference (PVSC), 2016 IEEE 43rd, 1287 (2016)Google Scholar
  36. 43.36
    I.M. Kotschau, G. Bilger, H.W. Schock: Mater. Res. Soc. Symp. Proc. 763, 263 (2003)CrossRefGoogle Scholar
  37. 43.37
    D. Hermann, F. Kessler, K. Hertz, M. Powalla, A. Schulz, J. Schneider, U. Schumacher: Mater. Res. Soc. Symp. Proc. 763, 287 (2003)CrossRefGoogle Scholar
  38. 43.38
    S. Hishikawa, T. Satoh, S. Hayashi, Y. Hashimoto, S. Shimakawa, T. Megami, T. Wada: Sol. Energy Mater. Sol. Cells 67, 217 (2001)CrossRefGoogle Scholar
  39. 43.39
    C. Candelise, M. Winskel, R. Gross: Prog. Photovolt. 20, 816 (2012)CrossRefGoogle Scholar
  40. 43.40
    K. Ito, T. Nakazawa: J. Appl. Phys. 27, 4 (1988)Google Scholar
  41. 43.41
    H. Katagiri, K. Jimbo, S. Yamada, T. Kamimura, W.S. Maw, T. Fukano, T. Ito, T. Motohiro: Appl. Phys. Express. 1, 2 (2008)CrossRefGoogle Scholar
  42. 43.42
    W. Wang, M.T. Winkler, O. Gunawan, T. Gokman, T.K. Todorov, Y. Zhu, D.B. Mitzi: Adv. Energy Mater. 4, 1301465 (2014)CrossRefGoogle Scholar
  43. 43.43
    B. Shin, O. Gunawan, Y. Zhu, N.A. Bojarczuk, S. Jay Chey, S. Guha: Prog. Photovolt. Res. Appl. 21, 72 (2011)CrossRefGoogle Scholar
  44. 43.44
    M. Gratzel: MRS Bull. 30, 23 (2005)CrossRefGoogle Scholar
  45. 43.45
    J. Nelson: Mater. Today 14, 462 (2011)CrossRefGoogle Scholar
  46. 43.46
    T. Ameri, P. Khoram, J. Min, C.J. Brabec: Adv. Mater. 25, 4245 (2013)CrossRefGoogle Scholar
  47. 43.47
    M. Liu, M.B. Johnston, H.J. Snaith: Nature 501, 395 (2013)CrossRefGoogle Scholar
  48. 43.48
  49. 43.49
    G.E. Eperon, V.M. Burlakov, A. Goriely, H.J. Snaith: ACS Nano 8, 591 (2013)CrossRefGoogle Scholar

Copyright information

© Springer International Publishing AG 2017

Authors and Affiliations

  1. 1.College of EngineeringSwansea UniversitySwanseaUK

Personalised recommendations