Sei sulla pagina 1di 18

CHAPTER 3 SOLAR PHOTOVOLTAICS

65

Chapter 3 Solar Photovoltaics


by Godfrey Boyle

66

RENEWABLE ENERGY

3.1 Introduction
In Chapter 2 we saw how solar energy can be used to generate electricity by producing high temperature heat to power an engine which then produces mechanical work to drive an electrical generator. This chapter is concerned with a more direct method of generating electricity from solar radiation, namely photovoltaics: the conversion of solar energy directly into electricity in a solid-state device. The chapter starts with a brief look at the history and basic principles of photovoltaic energy conversion, concentrating initially on devices using monocrystalline silicon. We then review various ways of reducing the currently high costs of energy from photovoltaics. The electrical characteristics of photovoltaic cells and modules are then described, followed by a discussion of the various roles of photovoltaic energy systems in supplying power in remote locations and in feeding power into local or national electricity grids. The concluding sections review the economics and the environmental impact of photovoltaic electricity, examine the resources available from photovoltaics and how it might be increasingly integrated into electricity supplies, and look at the world market and future prospects for photovoltaics.

3.2 Introducing photovoltaics


If you were asked to design the ideal energy conversion system, you would probably find it difficult to come up with something better than the solar photovoltaic (PV) cell. In this we have a device which harnesses an energy source that is by far the most abundant of those available on the planet. As earlier chapters have emphasised, the net solar power input to the earth is more than 10 000 times humanitys current rate of use of fossil and nuclear fuels. The PV cell itself is, in its most common form, made almost entirely from silicon, the second most abundant element in the earths crust. It has no moving parts and can therefore in principle, if not yet in practice, operate for an indefinite period without wearing out. And its output is electricity, probably the most useful of all energy forms.

A brief history of PV
The term photovoltaic is derived by combining the Greek word for light, photos, with volt, the name of the unit of electromotive force the force that causes the motion of electrons (i.e. an electric current). The volt was named after the Italian physicist Count Alessandro Volta, the inventor of the battery. Photovoltaics thus describes the generation of electricity from light.
Figure 3.1 Edmond Becquerel, who discovered the photovoltaic effect

The discovery of the photovoltaic effect is generally credited to the French physicist Edmond Becquerel (Figure 3.1), who in 1839 published a paper (Becquerel, 1839) describing his experiments with a wet cell battery, in

CHAPTER 3 SOLAR PHOTOVOLTAICS

67

the course of which he found that the battery voltage increased when its silver plates were exposed to sunlight. The first report of the PV effect in a solid substance appeared in 1877 when two Cambridge scientists, W. G. Adams and R. E. Day, described in a paper to the Royal Society the variations they observed in the electrical properties of selenium when exposed to light (Adams and Day, 1877). In 1883 Charles Edgar Fritts, a New York electrician, constructed a selenium solar cell that was in some respects similar to the silicon solar cells of today (Figure 3.2). It consisted of a thin wafer of selenium covered with a grid of very thin gold wires and a protective sheet of glass. But his cell was very inefficient. The efficiency of a solar cell is defined as the percentage of the solar energy falling on its surface that is converted into electrical energy. Less than 1% of the solar energy falling on these early cells was converted to electricity. Nevertheless, selenium cells eventually came into widespread use in photographic exposure meters. The underlying reasons for the inefficiency of these early devices were only to become apparent many years later, during the first half of the twentieth century, when physicists such as Planck and Einstein provided new insights into the nature of radiation and the fundamental properties of materials (see Section 3.3 below). It was not until the 1950s that the breakthrough occurred that set in motion the development of modern, high-efficiency solar cells. It took place at the Bell Telephone Laboratories (Bell Labs) in New Jersey, USA, where a number of scientists, including Darryl Chapin, Calvin Fuller and Gerald Pearson (Figure 3.3), were researching the effects of light on semiconductors . These are non-metallic materials, such as germanium and silicon, whose electrical characteristics lie between those of conductors, which offer little resistance to the flow of electric current, and insulators, which block the flow of current almost completely. Hence the term semiconductor. A few years before, in 1948, two other Bell Labs researchers, Bardeen and Brattain, had produced another revolutionary device using semiconductors the transistor. Transistors are made from semiconductors (usually silicon) in extremely pure crystalline form, into which tiny quantities of carefully selected impurities, such as boron or phosphorus, have been deliberately diffused. This process, known as doping , dramatically alters the electrical behaviour of the semiconductor in a very useful manner that will be described in detail later. In 1953 the ChapinFullerPearson team, building on earlier Bell Labs research on the PV effect in silicon (Ohl, 1941), produced doped silicon slices that were much more efficient than earlier devices in producing electricity from light.

Figure 3.2 Diagram from Charles Edgar Fritts 1884 US patent application for a solar cell

Figure 3.3 Bell Laboratories pioneering PV researchers Pearson, Chapin and Fuller measure the response of an early solar cell to light

68

RENEWABLE ENERGY

By the following year they had produced a paper on their work (Chapin et al., 1954) and had succeeded in increasing the conversion efficiency of their silicon solar cells to 6%. Bell Labs went on to demonstrate the practical uses of solar cells, for example in powering rural telephone amplifiers, but at that time they were too expensive to be an economic source of power in most applications. In 1958, however, solar cells were used to power a small radio transmitter in the second US space satellite, Vanguard I. Following this first successful demonstration, the use of PV as a power source for spacecraft has become almost universal (Figure 3.4). Rapid progress in increasing the efficiency and reducing the cost of PV cells has been made over the past few decades. Their terrestrial uses are now widespread, particularly in providing power for telecommunications, lighting and other electrical appliances in remote locations where a more conventional electricity supply would be too costly. A single conventional PV cell produces only about 1.5 watts, so to obtain more power, groups of cells are normally connected together to form rectangular modules. To obtain even more power, modules are in turn mounted side by side and connected together to form arrays. A growing number of domestic, commercial and industrial buildings now have PV arrays providing a substantial proportion of their energy needs. And a number of large, megawatt-sized PV power stations connected to electricity grids are now in operation in the USA, Germany, Italy, Spain and Switzerland.

Figure 3.4 The International Space Station is powered by large arrays of photovoltaic panels with a total output of 110 kilowatts

The efficiency of the best single-junction silicon solar cells has now reached 24% in laboratory test conditions (see Box 3.1 and Box 3.3). The best silicon PV modules now available commercially have an efficiency of over 17%, and it is expected that in about 10 years time module efficiencies will have risen to over 20% (Appleyard, 2003). Over the decade to 2002, the total installed capacity of PV systems increased approximately tenfold, module costs dropped to below $4 per peak watt and overall system costs fell to around $7 per peak watt (see Figures 3.5 and 3.6). As we shall see, improvements in the cost-effectiveness of PV are likely to continue.

3.3 PV in silicon: basic principles


Semiconductors and doping
PV cells consist, in essence, of a junction between two thin layers of dissimilar semiconducting materials, known respectively as p (positive)type semiconductors, and n (negative)-type semiconductors. These semiconductors are usually made from silicon, so for simplicity we shall

CHAPTER 3 SOLAR PHOTOVOLTAICS

69

1400

1200

grid-connected centralised grid-connected distributed

cumulative installed capacity/megawatts

1000 off-grid non-domestic 800 off-grid domestic

600

400

200

1992

1993 1994 1995 1996 1997 1998 1999 2000 2001 2002

Figure 3.5 Cumulative PV power installed by type of application in IEA reporting countries, 19922002 (source: International Energy Agency, 2003). Note: data in this figure are from 20 International Energy Agency reporting countries, mainly developed nations. There is additional significant PV module production and installed capacity in a number of other countries, particularly in the developing world

30 country 1 modules 25 country 2 modules


price/$US per watt

country 1 systems country 2 systems country 3 systems

20

country 3 modules

15

10

1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002

Figure 3.6 PV system and module price trends in three selected IEA reporting countries (source: International Energy Agency, 2003)

70

RENEWABLE ENERGY

BOX 3.1 Standard test conditions for PV cells and modules There is widespread international agreement that the performance of PV cells and modules should be measured under a set of standard test conditions. Essentially, these specify that the temperature of the cell or module should be 25 C and that the solar radiation incident on the cell should have a total power density of 1000 watts per square metre (W m2), with a spectral power distribution known as Air Mass 1.5 (AM 1.5). The spectral power distribution is a graph describing the way in which the power contained in the solar radiation varies across the spectrum of wavelengths. The concept of Air Mass describes the way in which the spectral power distribution of radiation from the sun is affected by the distance the suns rays have to travel though the atmosphere before reaching an observer (or a PV module). Just outside the earths atmosphere, the suns radiation has a power density of approximately 1365 W m2. The characteristic spectral power distribution of solar radiation as measured before it enters the atmosphere is described as the Air Mass 0 (AM0) distribution. At the earths surface, the various gases of which the atmosphere is composed (oxygen, nitrogen, ozone, water vapour, carbon dioxide, etc.) attenuate the solar radiation selectively at different wavelengths. This attenuation increases as the distance over which the suns rays have to travel through the atmosphere increases. When the sun is at its zenith (i.e. directly overhead), the distance over which the suns rays have to travel through the atmosphere to a PV module is at a minimum. The characteristic spectral power distribution of solar radiation that is observed under these conditions is known as the Air Mass 1 (AM1) distribution. When the sun is at a given angle to the zenith (as perceived by an observer at sea level), the Air Mass is defined as the ratio of the path length of the suns rays under these conditions to the path length when the sun is at its zenith. By simple trigonometry (see Figure 3.7), this leads to the definition: 1 Air Mass cos
sun at angle to zenith sun at zenith

e her osp m at

earths surfa ce SO Air Mass = ZO


O

Figure 3.7 Air Mass is the ratio of the path length of the suns rays through the atmosphere when the sun is at a given angle () to the zenith, to the path length when the sun is at its zenith

An Air Mass distribution of 1.5, as specified in the standard test conditions, therefore corresponds to the spectral power distribution observed when the suns radiation is coming from an angle to overhead of about 48 degrees, since cos 48 = 0.67 and the reciprocal of this is 1.5. The approximate spectral power distributions for Air Masses 0 and 1.5 are shown in Figure 3.8. In practice, the power rating in peak watts (Wp) of a cell or module is determined by measuring the maximum power it will supply when exposed to radiation from lamps designed to reproduce the AM1.5 spectral distribution at a total power density of 1000 W m2 .

2.5
1

2.0

6000 K black body AM0 radiation

energy distribution/kW m

1.5

AM1.5 radiation

1.0

0.5

0.2

0.4

0.6

0.8 1.0 1.2 wavelength/m

1.4

1.6

1.8

2.0

Figure 3.8 The spectral power distributions of solar radiation corresponding to Air Mass 0 and Air Mass 1.5. Also shown is the theoretical spectral power distribution that would be expected, in space, if the sun were a perfect radiator (a black body) at 6000 C

CHAPTER 3 SOLAR PHOTOVOLTAICS

71

initially consider only silicon-based semiconductors although, as we shall see later, PV cells can be made from other materials. n-type semiconductors are made from crystalline silicon that has been doped with tiny quantities of an impurity (usually phosphorus) in such a way that the doped material possesses a surplus of free electrons. Electrons are sub-atomic particles with a negative electrical charge, so silicon doped in this way is known as an n (negative)-type semiconductor. p-type semiconductors are also made from crystalline silicon, but are doped with very small amounts of a different impurity (usually boron) which causes the material to have a deficit of free electrons. These missing electrons are called holes. Since the absence of a negatively charged electron can be considered equivalent to a positively charged particle, silicon doped in this way is known as a p (positive)-type semiconductor (see Box 3.2).

The pn junction
We can create what is known as a pn junction by joining these dissimilar semiconductors. This sets up an electric field in the region of the junction. This electric field is like the electrostatic field you can generate by rubbing a plastic comb against a sweater. It will cause negatively charged particles to move in one direction, and positively charged particles to move in the opposite direction. However, a pn junction in practice is not a simple mechanical junction: the characteristics change from p to n gradually, not abruptly, across the junction.

The PV effect
What happens when light falls on the pn junction at the heart of a solar cell? Light can be considered to consist of a stream of tiny particles of energy, called photons. When photons from light of a suitable wavelength fall within the pn junction, they can transfer their energy to some of the electrons in the material, so promoting them to a higher energy level. Normally, these electrons help to hold the material together by forming so-called valence bonds with adjoining atoms, and cannot move. In their excited state, however, the electrons become free to conduct electric current by moving through the material. In addition, when electrons move they leave behind holes in the material, which can also move (Figure 3.9 in Box 3.2). The car parking analogy shown in Figure 3.10 may be helpful in visualizing the processes involved. When the pn junction is formed, some of the electrons in the immediate vicinity of the junction are attracted from the n-side to combine with holes on the nearby p-side. Similarly, holes on the p-side near the junction are attracted to combine with electrons on the nearby n-side. The net effect of this is to set up around the junction a layer on the n-side that is more positively charged than it would otherwise be, and, on the pside, a layer that is more negatively charged than it would otherwise be. In effect, this means that a reverse electric field is set up around the junction: negative on the p-side and positive on the n-side. The region around the junction is also depleted of charge carriers (electrons and holes) and is therefore known as the depletion region.
(a)

(b)

Figure 3.10 Car parking analogy of conduction processes in a semiconductor. (a) The ground floor of the car park is full the cars there (representing electrons in the valence band) cannot move around. The first floor is empty. (b) A car (electron) is promoted to the first floor (representing the conduction band), where it can move around freely. This leaves behind a hole that also allows cars on the ground floor (valence band) to move around (source: Green, 1982)

72

RENEWABLE ENERGY

BOX 3.2 Crystalline silicon, doping, pn junctions and PV cells


Figure 3.9 (a) Crystal of pure silicon has a cubic structure, shown here in two dimensions for simplicity. The silicon atom has four valence electrons. Each atom is firmly held in the crystal lattice by sharing two electrons (small dots) with each of four neighbours at equal distances from it. Occasionally thermal vibrations or a photon of light will spontaneously provide enough energy to promote one of the electrons into the energy level known as the conduction band, where the electron (small dot with arrow) is free to travel through the crystal and conduct electricity. When the electron moves from its bonding site, it leaves a hole (small open circle), a local region of net positive charge (b) Crystal of n-type silicon can be created by doping the silicon with trace amounts of phosphorus. Each phosphorus atom (large coloured circle) has five valence electrons, so that not all of them are taken up in the crystal lattice. Hence n-type crystal has an excess of free electrons (small dots with arrows) (c) Crystal of p-type silicon can be created by doping the silicon with trace amounts of boron. Each boron atom (large coloured circle) has only three valence electrons, so that it shares two electrons with three of its silicon neighbours and one electron with the fourth. Hence the p-type crystal contains more holes than conduction electrons (d) A silicon solar cell is a wafer of p-type silicon with a thin layer of n-type silicon on one side. When a photon of light with the appropriate amount of energy (labelled a) penetrates the cell near the junction of the two types of crystal and encounters a silicon atom, it dislodges one of the electrons, which leaves behind a hole. The energy required to promote the electron into the conduction band is known as the band gap. The electron thus promoted tends to migrate into the layer of n-type silicon, and the hole tends to migrate into the layer of p-type silicon. The electron then travels to a current collector on the front surface of the cell, generates an electric current in the external circuit and then reappears in the layer of p-type silicon, where it can recombine with waiting holes. If a photon with an amount of energy greater than the band gap (labelled b) strikes a silicon atom, it again gives rise to an electronhole pair, and the excess energy is converted into heat. A photon with an amount of energy smaller than the band gap (c) will pass right through the cell, so that it gives up virtually no energy along the way. Moreover, some photons (d) are reflected from the front surface of the cell even when it has an antireflection coating. Still other photons are lost because they are blocked from reaching the crystal by the current collectors that cover part of the front surface (source for text and figures: Chalmers, 1976)

(a)

(b)

(c) front metal contacts d c a b antireflection coating electrons drift to n-region (front contacts) current flows in external circuit ammeter electron-hole pairs formed holes drift current to p-region collector (back contact)

antireflection coating n-type crystal p-type crystal rear metal contact (d)

CHAPTER 3 SOLAR PHOTOVOLTAICS

73

When an electron in the junction region is stimulated by an incoming photon to jump into the conduction band, it leaves behind a hole in the valence band. Two charge carriers (an electronhole pair) are thus generated. Under the influence of the reverse electric field around the junction, the electrons will tend to move into the n-region and the holes into the p-region. The process can be envisaged (Figure 3.11) in terms of the energy levels in the material. The electrons that have been stimulated by incoming photons to enter the conduction band can be thought of as rolling downwards, under the influence of the electric field at the junction, into the n-region; similarly, the holes can be thought of as floating upwards, under the influence of the junction field, into the p-region. The flow of electrons to the n-region is, by definition, an electric current. If there is an external circuit for the current to flow through, the moving electrons will eventually flow out of the semiconductor via one of the metallic contacts on the top of the cell. The holes, meanwhile, will flow in the opposite direction through the material until they reach another metallic contact on the bottom of the cell, where they are then filled by electrons entering from the other half of the external circuit. In order to produce power, the PV cell must generate voltage as well as the current provided by the flow of electrons. This voltage is, in effect, provided

energy level conduction band energy gap valence band

energy level electron

incoming energy

electron

hole

(a) energy level

(b) depletion region conduction band electrons roll downwards


energy gap

n oto ph

energy level
n oto ph

valence band (c) n p (d) holes float upwards n p

distance from front of cell


Figure 3.11 (a) Energy bands in a normal (intrinsic) semiconductor; (b) An electron can be promoted to the conduction band when it absorbs energy from light (or heat), leaving behind a hole in the valence band; (c) When the n-type and p-type semiconductors are combined into a pn junction, their different energy bands combine to give a new distribution, as shown, and a built-in electric field is created; (d) In the pn junction, photons of light can excite electrons from the valence band to the conduction band. The electrons roll downwards to the n-region, and the holes float upwards to the p-region

74

RENEWABLE ENERGY

BOX 3.3 Band gaps and PV cell efficiency According to the quantum theory of matter, the quantity of energy possessed by any given electron in a material will lie within one of several levels or bands. Those electrons that normally hold the atoms of a material together (by being shared between adjoining atoms, as we saw in Figure 3.9) are described by physicists as occupying a lower-energy state known as the valence band. In certain circumstances, some electrons may acquire enough energy to move into a higher-energy state, known as the conduction band, in which they can move around within the material and thus conduct electricity (Figure 3.11). There is a so-called energy gap or band gap between these bands, the magnitude of which varies from material to material, and which is measured using an extremely small energy unit: the electron volt (eV) (see Appendix A2). Metals, which conduct electricity well, have many electrons in the conduction band. Insulators, which hardly conduct electricity at all, have virtually no electrons in the conduction band. Pure (or intrinsic) semiconductors have some electrons in the conduction band, but not as many as in a metal. But doping pure semiconductors with very small quantities of certain impurities can greatly improve their conductivity. If a photon incident on a doped, n-type semiconductor in a PV cell is to succeed in transferring its energy to an electron and exciting it from the valence band to the conduction band, it must possess an energy at least equal to the band gap. Photons with energy less than the band gap do not excite valence electrons to enter the conduction band and are wasted. Photons with energies significantly greater than the band gap do succeed in promoting an electron into the conduction band, but any excess energy is dissipated as heat. This wasted energy is one of the reasons why PV cells are not 100% efficient in converting solar radiation into electricity. (Another is that not all photons incident on a cell are absorbed: a small proportion are reflected.) Because the energy of a photon is directly proportional to the frequency of the light associated with it, photons associated with shorter wavelengths (i.e. higher frequencies) of light, near the blue end of the visible spectrum, have a greater energy than those of longer wavelength near the red end of the visible spectrum. The spectral distribution of sunlight varies considerably according to weather conditions and the elevation of the sun in the sky (see Box 3.1). For maximum efficiency of conversion of light into electric power, it is clearly important that the band gap energy of the material used for a PV cell is reasonably well matched to the spectrum of the light incident upon it. For example, if the majority of the energy in the incoming solar spectrum is in the yellowgreen range (corresponding to photons with energy of around 1.5 eV), then a semiconductor with a band gap of around 1.5 eV will be most efficient. In general, semiconductor materials with band gaps between 1.0 and 1.5 eV are reasonably well suited to PV use. Silicon has a band gap of 1.1 eV. The maximum theoretical conversion efficiency attainable in a single-junction silicon PV cell has been calculated to be about 30%, if full advantage is taken of light trapping techniques to ensure that as many of the photons as possible are usefully absorbed (Green, 1993). However, multi-junction cells have also been designed in which each junction is tailored to absorb a particular portion of the incident spectrum. Theoretically, such cells should have a much higher efficiency, possibly as high as 66% for an infinite number of junctions though the efficiencies so far achieved by multi-junction cells in practice have been very much lower than this (see Section 3.6). In practice, the highest efficiency achieved in commercially available single-junction monocrystalline silicon PV modules (as distinct from individual PV cells) is currently around 17%. The efficiency of PV modules is usually lower than that achieved by cells in the laboratory because:

it is difficult to achieve as high an efficiency consistently in mass-produced devices as in one-off laboratory cells under optimum conditions; laboratory cells are not usually glazed or encapsulated; in a PV module there are usually inactive areas, both between cells and due to the surrounding module frame, that are not available to produce power; there are small resistive losses in the wiring between cells and in the diodes used to protect cells from short circuiting; there are losses due to mismatching between cells of slightly differing electrical characteristics connected in series.

by the internal electric field set up at the pn junction. As we shall see, a single crystalline silicon PV cell typically produces a voltage of about 0.5 V at a current of up to around 3 A that is, a peak power of about 1.5 W. (Depending on their detailed design, some PV cells produce more power than this, some less.)

CHAPTER 3 SOLAR PHOTOVOLTAICS

75

Monocrystalline silicon cells


Until fairly recently, the majority of solar cells were made from extremely pure monocrystalline silicon that is, silicon with a single, continuous crystal lattice structure (Figure 3.9) having virtually no defects or impurities. Monocrystalline silicon is usually grown from a small seed crystal that is slowly pulled out of a molten mass, or melt, of polycrystalline silicon (see below), in the sophisticated but expensive Czochralski process developed initially for the electronics industry. The entire process of
metallurgical grade silicon coke reduction arc furnace sand (SiO2 ) reduction with H2 (900 C) polycrystalline silicon heat (1500 C) Czochralski process distillation dissolve in HCl

high purity trichlorosilane

chlorosilanes

silicon wafers diamond sawing silicon crystal doping to form pn junction chem/mech polishing

polished wafers formation of front contact antireflection coating testing

interconnection, testing, encapsulation and assembly into modules

assembly of modules into array

Figure 3.12 The overall process of monocrystalline silicon solar cell and module production

76

RENEWABLE ENERGY

oxide plated metal front contacts (in laser-cut grooves) metal back contact n+ layer

p layer p+ layer

Figure 3.13 Main features of the laser-grooved buried-grid monocrystalline PV cell structure, developed at the University of New South Wales (Green, 1993) and used in many high-efficiency PV modules. A pyramid-shaped texture on the top surface increases the amount of light trapped. Buried electrical contacts give very low electrical resistance whilst minimizing losses due to overshadowing. The symbols p+ and n+ denote heavily doped layers that reduce electrical resistance in the contact areas

monocrystalline silicon solar cell and module production is summarized in Figure 3.12. Many of the most efficient single-junction monocrystalline PV modules currently available use a laser-grooved buried-grid cell structure as shown in Figure 3.13.

3.4 Crystalline PV: reducing costs and raising efficiency


Although the latest monocrystalline silicon PV modules are highly efficient, they are also expensive because the manufacturing processes are slow, require highly skilled operators, and are labour- and energy-intensive. Another reason for their high cost is that most high-efficiency cells are fabricated from extremely pure electronic-grade silicon. However, PV cells can be made from slightly less pure, but less costly, solar-grade silicon, with only a small reduction in conversion efficiency. A number of approaches to reducing the cost of crystalline PV cells and modules, or increasing their efficiency, have been under development during the past 20 years or so. These include cells using polycrystalline rather than single-crystal material; growing silicon in ribbon or sheet form; and the use of other crystalline PV materials such as gallium arsenide.

Polycrystalline silicon
Figure 3.14 Polycrystalline silicon consists of randomly packed grains of monocrystalline silicon

Polycrystalline silicon essentially consists of small grains of monocrystalline silicon (Figure 3.14). Solar cell wafers can be made directly from polycrystalline silicon in various ways. These include the controlled casting of molten polycrystalline silicon into cube-shaped ingots which

CHAPTER 3 SOLAR PHOTOVOLTAICS

77

are then cut, using fine wire saws, into thin square wafers and fabricated into complete cells in the same way as monocrystalline cells. Polycrystalline PV cells are easier and cheaper to manufacture than their monocrystalline counterparts. But they tend to be less efficient because light-generated charge carriers (i.e. electrons and holes) can recombine at the boundaries between the grains within polycrystalline silicon. However, it has been found that by processing the material in such a way that the grains are relatively large in size, and oriented in a top-to-bottom direction to allow light to penetrate deeply into each grain, their efficiency can be substantially increased. These and other improvements have enabled commercially available polycrystalline PV modules (sometimes called multi-crystalline or semi-crystalline) to reach efficiencies of over 14%. Companies such as Pacific Solar in Australia and Astropower in the USA are developing ways of depositing polycrystalline films on to ceramic or glass substrates, forming PV modules with an efficiency of around 10%. The silicon films used are somewhat thicker than in other thin film PV cells (see below), so cells made in this way are sometimes known as thick film polycrystalline cells.

Silicon ribbons and sheets


This approach involves drawing thin ribbons or sheets of multicrystalline silicon from a silicon melt. One of the main processes is known as edgedefined, film-fed growth (EFG), and was originally developed by the US company Mobil Solar. It is described in Figure 3.15. EFG cells are manufactured by the German company RWE Schott Solar (under the brand name ASE) and silicon ribbon cells by Evergreen Solar in the USA.

crystalline silicon die

nonagon tube pulled from melt 1 2

doping and processing 3

die

silicon melt

nine-sided die silicon melt at 1400 C

nonagon tube cut by laser

finished silicon cell

Figure 3.15 Edge-defined, film-fed growth process for PV production.Thin, hollow polygonal tubes of polycrystalline silicon up to 6 m long are slowly pulled through a die from a melt of pure silicon, then cut by laser into individual cells

Gallium arsenide
Silicon is not the only crystalline material suitable for PV. Another is gallium arsenide (GaAs), a so-called compound semiconductor. GaAs has a crystal structure similar to that of silicon (see Figure 3.9), but consisting of

78

RENEWABLE ENERGY

alternating gallium and arsenic atoms. In principle it is highly suitable for use in PV applications because it has a high light absorption coefficient, so only a thin layer of material is required. GaAs cells also have a band gap wider than that of silicon, one close to the theoretical optimum for absorbing the energy in the terrestrial solar spectrum (see Box 3.3). Cells made from GaAs are therefore more efficient than those made from monocrystalline silicon. They can also operate at relatively high temperatures without the appreciable reduction in efficiency that affects PV cells made from silicon when their temperatures rise. This makes them well suited to use in concentrating PV systems (see Section 3.6 below). On the other hand, cells made from GaAs are substantially more expensive than silicon cells, partly because the production process is not so well developed, and partly because gallium and arsenic are not abundant materials. GaAs cells have often been used when very high efficiency, regardless of cost, is required as in many space applications. They have also powered many of the winning cars in solar car races (see Figure 3.16).

Figure 3.16 The Dutch Nuna solar car, winner of the 2001 World Solar Challenge, arriving at Adelaide, South Australia after completing the 3000 km journey from Darwin in just over 32 hours at an average speed (during daylight hours) of 91 km per hour. The ultralightweight car was produced by students from the universities of Delft and Rotterdam, sponsored by the utility Nuon. It is driven by electric motors powered by arrays of high-efficiency, dual-junction and triple-junction gallium arsenide PV cells developed by the European Space Agency for satellite use. (The 2003 solar challenge winner was Nuna II, also from the Netherlands, which completed the journey at an average speed of 97 km per hour)

3.5 Thin film PV


Amorphous silicon
Solar cells can be made from very thin films of silicon in a form known as amorphous silicon (a-Si), in which the silicon atoms are much less ordered than in the crystalline forms described above. In a-Si, not every silicon atom is fully bonded to its neighbours, which leaves so-called dangling bonds that can absorb any additional electrons introduced by doping, so rendering any pn junction ineffective.

However, this problem is largely overcome in the process by which a-Si cells are normally manufactured. A gas containing silicon and hydrogen (such as silane, SiH4), and a small quantity of dopant (such as boron), is decomposed electrically in such a way that it deposits a thin film of amorphous silicon on a suitable substrate (backing material). The hydrogen in the gas has the effect of providing additional electrons that combine with the dangling silicon bonds to form, in effect, an alloy of silicon and hydrogen. The dopant that is also present in the gas can then have its usual effect of contributing charge carriers to enhance the conductivity of the material. Solar cells using a-Si have a somewhat different form of junction between the p- and the n-type material. A so-called pin junction is usually formed, consisting of an extremely thin layer of p-type a-Si on top, followed by a thicker intrinsic (i) layer made of undoped a-Si, and then a very thin layer of n-type a-Si. The structure is as shown in Figure 3.17. The operation

CHAPTER 3 SOLAR PHOTOVOLTAICS

79

of the PV effect in a-Si is generally similar to that in crystalline silicon, except that in a-Si the band gap, although wider, is less clearly defined. Amorphous silicon cells are much cheaper to produce than those made from crystalline silicon. a-Si is also a much better absorber of light, so much thinner (and therefore cheaper) films can be used. The manufacturing process operates at a much lower temperature than that for crystalline silicon, so less energy is required; it is suited to continuous production; and it allows quite large areas of cell to be deposited on to a wide variety of both rigid and flexible substrates, including steel, glass and plastics. But a-Si cells are currently much less efficient than their single-crystal or polycrystalline silicon counterparts: maximum efficiencies achieved with small, single-junction cells in the laboratory are currently around 12%. Moreover, the efficiency of many currently available singlejunction a-Si modules degrades, within a few months of exposure to sunlight, from an initial 610%, stabilizing at around 48%. Strenuous attempts have been made by manufacturers to improve the efficiency of a-Si cells and to solve the degradation problem. One approach involves the development of multiple-junction a-Si devices (see Section 3.6 below.) Amorphous silicon cells are already widely used as power sources for a variety of consumer products such as calculators, where the requirement is not so much for high efficiency as for low cost.

light

glass O2) e (Si d i x o n di er silico g lay uctin 2) d n o top c ide (SnO x tin o e p-typ licon us si o h p r amo sic intrin
e n-typ

Other thin film PV technologies


Amorphous silicon is by no means the only material suited to thin film PV, however. Among the many other possible thin film technologies some of the most promising are those based on compound semiconductors, in particular copper indium diselenide (CuInSe2, usually abbreviated to CIS), copper indium gallium diselenide (CIGS) and cadmium telluride (CdTe). Modules based on all of these technologies have reached the production stage, but production volumes are small. Thin film CIGS cells have attained the highest laboratory efficiencies of all thin film devices, around 17%, and CIGS modules with stable efficiencies of around 10% are produced by Shell Solar in the USA and Wrth Solar in Germany. Cadmium telluride modules can be made using a relatively simple and inexpensive electroplating-type process. The band gap of CdTe is close to the optimum, and module efficiencies of over 10% are claimed, without the initial performance degradation that occurs in a-Si cells. However, the modules contain cadmium, a highly toxic substance, so stringent precautions need to be taken during their manufacture, use and eventual disposal (see Section 3.11). BP Solar, a subsidiary of the oil company BP, was involved in CdTe module manufacture but withdrew from production in 2002, explaining that although it did not consider the presence of cadmium in its modules to be a problem, customers appeared to believe otherwise. The Japanese firm Matsushita also withdrew from CdTe production in 2002, but First Solar Inc. in the USA continues to produce CdTe modules.

alum

inium

back contact

Figure 3.17 Structure of an amorphous silicon cell. The top electrical contact is made of an electrically conducting, but transparent, layer of tin oxide deposited on the glass. Silicon dioxide forms a thin barrier layer between the glass and the tin oxide. The bottom contact is made of aluminium. In between are layers of p-type, intrinsic and n-type amorphous silicon

80

RENEWABLE ENERGY

3.6 Other innovative PV technologies


Multi-junction PV cells
light

One way of improving the overall conversion efficiency of PV cells and modules is the stacked or multi-junction approach, in which two (or more) PV junctions are layered one on top of the other, each layer extracting energy from a particular portion of the spectrum of the incoming light. A cell with two layers is often called a tandem device. The band gap of amorphous silicon, for example, can be increased by alloying the material with carbon, so that the resulting material responds better to light at the blue end of the spectrum. Alloying with germanium, on the other hand, decreases the band gap so the material responds better to light at the red end of the spectrum. Typically, a wide band gap a-Si junction would be on top, absorbing the higher-energy photons at the blue end of the spectrum, followed by other thin film a-Si junctions, each having a band gap designed to absorb a portion of the lower light frequencies, nearer the red end of the spectrum (Figure 3.18). Multi-junction modules using amorphous silicon are available with stable efficiencies of around 8% from companies such as Unisolar and RWE. Cells of different types can also be used, as in the Sanyo Hybrid HIT module, in which a thin monocrystalline layer is sandwiched between two amorphous silicon layers. This enables very high conversion efficiencies to be achieved with low material and manufacturing energy requirements. Another example is the tandem module produced by Kaneka of Japan, which has a layer of amorphous silicon on top of a thin layer of microcrystalline silicon (i.e. silicon in the form of extremely small crystals less than one micrometre in diameter).

glass silico xide n dio

) (SiO 2

layer ting c u d on e top c tin oxid e p-typ icon l us si th rpho loyed wi o m a l a ) sic (intrin carbon e n-typ e p-typ n silico ous h p r amo intrinsic e n-typ e p-typ licon s u si ed y rp h o amo insic) allo um (intr germani with pe n-ty e -oxid m-tin u i d in r silve back contact

Concentrating PV systems
Another way of getting more energy out of a given number of PV cells is to use mirrors or lenses to concentrate the incoming solar radiation on to the cells, an approach similar to that described in Chapter 2, Section 2.10, on solar thermal engines. This has the advantage that substantially fewer cells are required to an extent depending on the concentration ratio, which can vary from as little as two to several hundred or even thousand times. The concentrating system must have an aperture equal to that of an equivalent flat plate array to collect the same amount of incoming energy. In concentrating PV systems the cells usually need to be cooled, either passively or actively, to prevent overheating. Systems with the highest concentration ratios use complex sensors, motors and controls to allow them to track the sun on two axes azimuth (horizontal orientation) and elevation (tilt) ensuring that the cells always receive the maximum amount of solar radiation. Systems with lower concentration ratios often track the sun on only one axis and can have simpler tracking mechanisms.

Figure 3. 18 Structure of a multijunction amorphous silicon cell

CHAPTER 3 SOLAR PHOTOVOLTAICS

81

Most concentrators can only utilize direct solar radiation. This is a problem in countries like the UK where nearly half the solar radiation is diffuse. However, some unconventional designs of concentrator allow some of the diffuse radiation, as well as direct radiation, to be concentrated (see Boes and Luque, 1993).

Silicon spheres
The US firm Texas Instruments has developed an ingenious way of making PV cells using tiny, millimetre-sized, spheres of polycrystalline silicon embedded at regular intervals between thin sheets of aluminium foil. Among the advantages of this approach are that impurities in the silicon tend to diffuse out to the surface of the spheres, where they can be ground off as part of the manufacturing process. This allows relatively cheap, low-grade silicon to be used as a starting material. The resulting sheets of PV material are very flexible (see Figure 3.19), which can be an advantage in some applications. The technology is being commercialized by Automation Tooling Systems Inc. in Canada, where a 20 MWp (megawatts-peak) plant is under construction.

Photoelectrochemical cells
A radically different, photoelectrochemical, approach to producing cheap electricity from solar energy has been pioneered by researchers at the Swiss Federal Institute of Technology in Lausanne. (Strictly, photoelectrochemical devices are not photovoltaic: this term implies a solid-state device. Photoelectrochemical devices, however, use liquids.) The idea of harnessing photoelectrochemical effects to produce electricity from sunlight is not new indeed, Becquerels pioneering PV experiments were with liquidbased devices. The Swiss researchers, led by Professor Michael Grtzel, have achieved much higher efficiencies than before, in a device that could be extremely cheap to manufacture. It consists essentially of two thin glass plates, both covered with a thin, transparent, electrically conducting tin oxide layer (Figure 3.20, overleaf). To one plate is added a thin layer of titanium dioxide (TiO2), which is a semiconductor. The surface of the TiO 2 has been treated to give it exceptionally high roughness, enhancing its light-absorbing properties. Immediately next to the roughened surface of the TiO2 is a layer of sensitizer dye only one molecule thick, made of a proprietary transition metal complex based on ruthenium or osmium. Between this sensitized TiO2 and the other glass plate is a thicker layer of iodine-based electrolyte. On absorption of a photon of suitable wavelength, the sensitizer layer injects an electron into the conduction band of the titanium dioxide. Electrons so generated then move to the bottom electrically conducting layer (electrode) and pass out into an external circuit where they can do work. They then re-enter through the top electrode, where they drive a reductionoxidation process in the iodine solution. This then supplies electrons to the sensitized TiO2 layer in order to allow the process to continue.
Figure 3.19 Silicon spheres PV technology

82

RENEWABLE ENERGY

light glass tin oxide iodine (I)-based electrolyte e + I I reduction e I + e oxidation


e

current flows e

titanium dioxide tin oxide glass

sensitiser layer e e

Figure 3.20 Principles of operation of the photoelectrochemical Grtzel Cell, developed at the Swiss Federal Institute of Technology, Lausanne

The Swiss team claims to have achieved efficiencies of 10% in full (AM 1.5) sunlight and that its cells are stable over long periods, though some researchers are not fully convinced of this. PV cells based on this technology are being manufactured on a small scale by STI in Australia and by Greatcell Solar SA in Switzerland. (See Grtzel et al., 1989, Grtzel, 2001 and ORegan et al., 1991, 1993.)

Third generation PV cells


Still at the frontiers of research is a range of new photovoltaic technologies known as third generation PV (crystalline PV is considered first generation and thin film PV second generation). These devices are generally based on nanotechnology that is, technology which aims to manipulate molecules and atoms at extremely small scales, measured in billionths of a metre, or nanometres (nm). These tiny particles and structures are called nanoparticles and nanostructures; crystals of such sizes are termed nanocrystals. Nanoparticles consisting of extremely small collections of atoms of semiconducting material are called quantum dots. As Dresselhaus and Thomas (2001) observe, a whole new class of materials is emerging, such as films containing nanocrystalline structures, quantum dots and nanostructured conducting polymers. These films typically contain nanoparticles with a size distribution showing a wide range of electronic band gaps (it is this gap that determines which wavelengths can be absorbed) so that much of the solar spectrum can be absorbed by the cell. If such films can be made cheaply, with an optimised distribution of nanoparticle diameters, and can be properly aligned, one might have an ideal solar collector. This field is still very young, and is moving very rapidly. Eventually, if research on third generation PV proves successful, it could lead to PV cells made, for example, from extremely thin stacked plastic sheets, converting solar energy to electricity with very high efficiency at very low cost.

Potrebbero piacerti anche