The photovoltaic industry has changed dramatically in the last few years. China has become the major manufacturing country for solar cells and modules, followed by Taiwan, Germany and Japan. Amongst the 20 biggest photovoltaic manufacturers in 2012, only three had production facilities in Europe, namely First Solar (USA, Germany, Malaysia), Hanwha Q Cells (Germany and Malaysia) and REC (Norway and Singapore). However, REC closed down its production in Norway in early 2012 and First Solar closed its factory in Germany later in the year.
It is important to remember that the PV industry consists of more than just the cell and module production, and that the whole PV value chain is relevant. Besides the information in this chapter, the whole upstream industry (e. g. materials, equipment manufacturing), as well as the downstream industry (e. g. inverters, BOS components, system development, installations) has to be looked at as well.
The implementation of the 100,000 roofs programme in Germany in 1990, and the Japanese long-term strategy set in 1994, with a 2010 horizon, were the beginning of an extraordinary PV market growth. Before the start of the Japanese market implementation programme in 1997, annual growth rates of the PV markets were in the range of 10%, mainly driven by communication, industrial and standalone systems. Since 1990, PV production has increased by almost three orders of magnitude, from 46MW to about 38.5GW in 2012. Statistically documented cumulative installations worldwide accounted for almost 100 GW in 2012. This represents mostly the grid-connected photovoltaic market. To what extent the off – grid and consumer product markets are included is not clear, but it is believed that a substantial part of these markets are not included in these figures as it is very difficult to track them.
Even with the current economic difficulties, overall installations are increasing, due to overall rising energy prices and the pressure to stabilise the climate. In the long term, growth rates for photovoltaics will continue to be high, even if economic conditions vary and lead to short-term slowdowns.
This view is shared by an increasing number of financial institutions, which are turning towards renewables as a sustainable and stable long-term investment. Increasing demand for energy is pushing the prices for fossil energy resources higher and higher. In 2007, a number of analysts predicted that oil prices could well hit 100 $/bbl by the end of that year or early 2008 (Bloomberg, 2007). After the spike of oil prices in July 2008, at close to 150 $/bbl, prices decreased due to the worldwide financial crisis and hit a low of around 37$/bbl in December 2008. Since then, the oil price has rebounded and the IEA reported average prices for oil imports around 110 $/bbl since the second quarter of 2011, with peaks around 120 $/bbl between February and April 2012 and September 2013. Oil demand has increased from about 84 million bbl/day in Q1 2009 to around 91 million bbl/day in Q3 2013, and a moderate further increase to 93 million bbl/day by Q4 2014 is expected. Even though no significant changes are currently forecast by analysts for 2014, the fundamental trend, that increasing demand for oil will drive its price higher again, is intact and will be evident as soon as the global economy recovers.
Over the last 20 years, numerous studies of the potential growth of the PV industry and the implementation of PV electricity generation systems have been produced. In 1996, the Directorate-General for Energy of the European Commission (EC) published a study entitled Photovoltaics in 2010 (European Commission, 1996). The medium scenario of this study was used to formulate the EC’s White Paper target of 1997, which was to have a cumulative installed capacity of 3 GW in theEU by 2010 (European Commission, 1997). The most aggressive scenario in this report predicted a cumulative installed PV capacity of 27.3 GW worldwide and
8.7 GW in the EU for 2010. This scenario was called ‘Extreme scenario’ and it was assumed that a number of breakthroughs in technology and costs, as well as continuous market stimulation and elimination of market barriers, would be required to achieve it. The reality check reveals that even the most aggressive scenario is lower than what we expect from the current developments.
According to investment analysts and industry prognoses, solar energy systems will continue to grow at high rates in the coming years. The different photovoltaic industry associations, as well as Greenpeace, the European Renewable Energy Council (EREC) and the IEA, have developed new scenarios for the future growth of PV. Table 13.1 shows the various scenarios of the Greenpeace/EREC study, as well as the different 2013 IEA World Energy Outlook scenarios and the IEA PV Technology Roadmap of 2010. It is interesting to note that the 2015 capacity values of only two scenarios — the Greenpeace (revolution) and IEA 450ppm Scenarios — were not reached at the end of 2013. With forecast new installations of between 93 and 106GW in 2014 and 2015, even the Greenpeace revolution scenario is no longer fictional thinking (Bloomberg New Energy Finance, 2014).
The above-mentioned solar photovoltaic scenarios will only be possible if solar cell and module manufacturing are continuously improved and novel design concepts are realised, as with current technology the demand for some materials,
Table 13.1 Predicted cumulative solar electrical capacities until 2035. Data sources: Bloomberg New Energy Finance, 2014; Greenpeace, 2012; IEA, 2010, 2013d.
Actual installations 100 138
*2035 values are extrapolated, as only 2030 and 2040 values are given
**2015 and 2030 values are extrapolated, as only 2011, 2020 and 2035 values are given
***2015 and 2035 values are extrapolated, as only 2010, 2020, 2030 and 2040 values are given
like silver, would dramatically increase the costs for these resources within the next 30 years. Research to avoid such problems is underway and it can be expected that these bottle-necks will be avoided. With 100 GW cumulative installed photovoltaic electricity generation capacity installed worldwide by the end of 2012, photovoltaics still is a small contributor to the electricity supply, but its importance for our future energy mix is finally acknowledged.
 The power output of a PV cell or module is rated in peak watts (Wp), meaning the power output at 25°C under standard AM1.5 solar radiation of global irradiance 1kWm-2. PV ratings in ‘watts’ invariably mean peak watts. To convert from peak watt rating to 24-hour average power output in a
sunny location, divide by a factor of ~5.
 Becquerel’s observation was strictly speaking a photoelectrochemical effect, but its basis—the rectifying junction formed between two dissimilar electric conductors — is the same as that of the photovoltaic effect in purely solid-state devices.
Zhores Alferov won the 2000 Nobel Prize for Physics for his development of heterostructures and is probably the only solid-state physicist to have an asteroid named after him.
Cadmium sulphide also lives on in the paintings of impressionists such as Monet, whose favourite yellow pigment it was.
c-Si cells are usually configured n-on-p because this best suits the properties of silicon, but some other p-n cells are configured p-on-n. These cells are also quite thick, because c-Si absorbs light relatively weakly. Most other cells are much thinner.
 The Fermi level is the energy for which the probability of a state being occupied by an electron is exactly one-half. In an intrinsic (undoped) semiconductor, the Fermi level falls in the middle of the forbidden gap. In a lightly doped semiconductor, the Fermi level remains within the forbidden gap but is near the majority-carrier band edge. In a heavily doped semiconductor, the Fermi level lies within the majority-carrier band.
 All the currents given the symbol i in Figs. 1.6—1.8 are strictly speaking current densities.
Forward biasing a junction means applying a voltage across the device that lowers the band-bending barrier. Reverse biasing means applying a voltage in the opposite direction.
If the cell has significant internal resistance, the output voltage V drops below the junction voltage Vj, and a small forward bias remains across the junction when the cell is short-circuited.
12In lightly doped Si cells, Voc can exceed Vbo at the junction, with the bulk going into high injection and an additional voltage across the rear back-surface field: SunPower cells effectively work in this way.
An ideal isotropic cell is one in which electrons and holes are thermalised to the band edge, the only decay channel for excited states is radiative recombination, and light can enter the cell at all forward
Rather than in accumulation, which would create a photovoltaically inactive ohmic contact.
 Included for reference.
and possibly beyond 30% in the longer term. For wafer-based technology, the figures would be correspondingly higher.
Several options for obtaining such efficiency improvements have been reviewed. Tandem cell stacks such as those discussed in Chapter 7of the current volume are clearly the most promising route based on experience to date. Hot-carrier cells have the attraction of offering higher efficiency than a 5-cell tandem stack in a simple 2-terminal configuration, although considerable theoretical and experimental progress will be required before any performance improvement is likely to be demonstrated with this approach. Up – and down-conversion offer the prospects for ‘supercharging’ the performance of existing cell technologies, although improvements in the performance for both types of converters are required also before any performance gain would be expected. Intermediate-band and multiple exciton generation devices offer improvements for specific material systems.
The ‘thermal approaches’ — thermophotovoltaics, thermionics, thermoelectric or hot lattice ‘hot-carrier’cells — offer tantalisingly high limiting efficiencies. However, the gap between what is likely to be obtained in practice and the theoretical potential is expected to be much larger with such approaches, because of the much higher energy losses that are likely to occur in practice. Such approaches may have potential in concentrating systems, but this is where tandem cell stacks have had the most impact terrestrially and are showing unabated potential for further development.
As is already becoming apparent as the photovoltaic industry matures, efficiency will be a key driver to future cost reduction. Evidence for this is the emphasis given by thin-film companies such as First Solar to quarterly improvements in module efficiency (from under 8% in 2003 to over 13% in 2013) and the increasing
 Figure 4.3 shows that the bandgap of amorphous silicon is between 1.7 to
1.8 eV; this is generally considered to be too high for efficient absorption of the solar spectrum. Attempts were therefore made to lower the bandgap by alloying with germanium or tin, or to find new materials with lower gaps for the construction of advanced tandem and multijunction devices.
• While the demonstration of doping was surprising because an amorphous network should easily incorporate elements with lower or higher valence, it was soon noted that the achievable doping efficiency is limited; films with better
 Historically, the first tandem cells combined two purely amorphous silicon absorbers; while they do not extend the spectral response compared with a single-junction cell, they are nevertheless advantageous because of reduced degradation with respect to a single-junction cell of comparable thickness (Hanak and Korsun, 1982; Bennett and Rajan, 1988; Lechner et al., 2008). Figure 4.11 shows the repartition of 15mAcm-2 between a thin a-Si:H top and a relatively thick a-Si:H bottom cell.
• The combination of amorphous and microcrystalline cells in a tandem with high total current appears unfavourable because the top cell sees only a single
 It must have sufficient transverse (out-of-plane) conductivity to avoid electrical losses caused by its series resistance.
• In order to reflect, it must be fabricated from a low refractive index material. Fresnel reflection occurs due to the index contrast between films; the refractive index of silicon nrSi is approximately 3.8 at 600 nm, and nrIRL should therefore ideally be below 2 at this wavelength.
• In addition to the refractive index, the thickness of the IRL controls the intensity and spectral selectivity of back reflection and forward transmission into the bottom cell via interference effects (see Figure 4.20).
The first efficient IRLs were realised using TCO layers such as ZnO (Fischer etal., 1996). More recently, phosphorus-doped silicon oxide materials (n-SiOx :P) with nriRL = 1.6-2 at 600 nm and with sufficient transverse conductivity were synthesised by PECVD, giving an in situ IRL alternative (Buehlmann et al., 2007;
 The BL allows thinner CdS layers to be used. One proposed reason for this is that the space charge on the n-type side of the junction can expand into the BL when the CdS becomes too thin to balance the charge needed to maintain the space-charge width in the p-type side of the junction. This enables the space charge in the p-CdTe region to remain sufficiently wide as the CdS layer is thinned, thereby avoiding device functionality dominated by voltage – dependent collection (i. e. low fill factor arising from loss of minority-carrier collection at high forward bias). A different explanation is that the BL reduces surface recombination at the CdS/TCO interface. The suggested benefit is that reduced recombination at this interface may improve device performance if thinner CdS layers position this interface closer to the CdS/CdTe junction region.
• The BL provides tolerance to shunt (or short) paths from the back contact to the front contact. These pathways will be present if the CdTe layer contains
 The film quality has been substantially improved by the crystallisation mechanism induced by the presence of CuySe (у < 2). This process is further
 The valence band-edge energy Ev lies above the surface-Fermi level EF by about 1.1 eV for CuInSe2 films (Schmid et al., 1993). This energy is larger than the bandgap energy Ega of the bulk of the absorber material. This was taken as an indication of a widening of bandgap at the surface of the film. For the surfaces of Cu(In1-xGax)Se2 thin films it was found that EF – Ev = 0.8eV (almost independent of the Ga content if x > 0) (Schmid et al., 1996b).
• The surface composition of Cu-poor CuInSe2, as well Cu(In, Ga)Se2 films, corresponds to a surface composition of (Ga + In)/(Ga + In + Cu) of about 0.75
 The bulk defect at activation energy Ea = 300 meV displays a minimum defect density D for the composition x = 0.26. Higher bandgap material (x = 0.56), as well as lower bandgap material (x = 0), has higher defect densities. It might not be incidental that the superior electronic quality of the material is found at that composition which is used to produce the highest-efficiency Cu(In, Ga)Se2 devices.
• The interface-related peak at lower activation energies for the compositions x = 0 and 0.26 is no longer visible at x = 0.56. This is just what we would expect from the band diagram shown in Fig. 6.28b: the Fermi level at the
 Selection and high-quality growth of InGaP as a top-cell material.
• Use of double-hetero structures and wide-bandgap tunnel junctions for cell interconnection.
• Precise lattice matching of the InGaP top cell and the InGaAs middle cell with the Ge substrate.
• Use of AlInP as a back-surface-field layer for the InGaP top cell.
• Use of InGaP-Ge heteroface structure bottom cells.
 Varying the relative solubility of the blended materials — either by altering the molecules themselves (Park et al., 2006), changing the type or temperature of the solvent (Liu et al., 2001), or adding co-solvents (Lee et al., 2008).
 The DSSC efficiency is practically temperature-independent in the range 25-65°C while that of monocrystalline and pc-Si declines by ~20% over the same range.
• Outdoor measurements indicate that light capture by the DSSC is less sensitive to the angle of incidence, although this needs to be further assessed. The DSSC
Sadly, SolFocus later fell victim to the fall in price of flat-panel silicon PV and in September 2013 was reportedly in negotiations to sell its assets and intellectual property (IP).
 All production and installation figures cited in W in this chapter are actually Wp.
Solar-cell production capacities mean: in the case of wafer silicon-based solar cells, only the cells; in the case of thin-films, the complete integrated module. Only those companies that actually produce the active circuits (solar cells) are counted; companies that purchase these circuits and make cells are not counted.
Exchange rate: l€ = 1.30 US$.
Average exchange rate 2010 and 2013: l€ = 1.40 AUD.
 Average exchange rate 2011: 1 €= 1.30 AUD.
 The Indian financial year ends 31 March.
Exchange rate: l€ = 130 ¥.
 This has been updated in the meantime.
Exchange rate 2012: l€ = 8.1 CNY.
Exchange rate: l€ = 1,420 KRW.
Exchange rate: l€ = 44 THB.
Exchange rate: l€ = 4.0 MRY.
Exchange rate 2011: l€ = 60 PHP.
Exchange rate 2012: l€ = 53 PHP.
Exchange rate: l€ = 7.46 DKK.
Exchange rate: l€ = 10.5 ZAR.
High concentration > 300 Suns (HCPV), medium concentration 5 < x < 300 Suns (MCPV), low concentration < 5 Suns (LCPV).
Solar cell production capacities mean: in the case of wafer silicon-based solar cells, the cell; in the case of thin-films, the complete integrated module. Only those companies that actually produce the active circuit (solar cell) are counted. Companies that purchase these circuits and make cells are not counted.