The structure of a single junction thin-film solar cell is indicated in the left panel of figure 4.1. This thin-film solar cell has an active part of typically less than a few micrometres thick, consisting of one or more semiconductor materials. Semiconduc-tor materials frequently used in thin-film solar cells are amorphous silicon (a-Si), an amorphous silicon/germanium (a-Si/Ge) alloy, micro crystalline silicon (µc-Si), cop-per indium gallium diselenide (CIGS), cadmium sulphide (CdS) and cadmium tel-luride (CdTe). As indicated, the electrons and holes generated in the semiconductor are collected as good as possible by the back contact and the transparent conductive oxide (TCO) respectively.
Besides the single junction solar cell, multi-junction thin-film solar cells have been developed. In the right panel of figure 4.1, the structure of a so-called tandem solar cell is shown. This tandem cell has two semiconductor layers. Generally two differ-ent semiconductor materials are used, each one characterised by a differdiffer-ent bandgap energy Eg. As indicated in the figure, the semiconductor with the larger bandgap Eg1 is used as top layer and the semiconductor with the smaller bandgap Eg2(Eg2< Eg1) is used as bottom layer. In this way, short wavelength photons with a photon energy Eph > Eg1, are mainly absorbed in the top layer. The longer wavelength photons with Eg2 < Eph < Eg1 are transmitted by this top layer and mainly absorbed by the bottom layer. Because this tandem cell is a series connection of two solar cells
utilising different parts of the solar spectrum, a higher electrical efficiency can be obtained.
Because thin-film solar cells can consist of many layers, they are very complex optically. Also the fact that layer thicknesses or interface roughnesses can have di-mensions comparable to or smaller than the wavelength of solar irradiance, makes it very challenging to model the optical behaviour of thin-film solar cells accurately.
Nonetheless, optical models for thin-film solar cells have been developed to deter-mine the absorption profile for calculating the electrical performance [6, 33, 63, 64].
However, because sub-bandgap solar irradiance does not play a role in the genera-tion of electricity, the absorpgenera-tion of this part of the solar spectrum has not yet been studied extensively.
The optical model developed and described in chapter 2 is very suitable to ana-lyse optical systems containing many layers. Because the optical model allows each interface to be modelled individually, the model is flexible enough to capture the complex optical behaviour of thin-film solar cells. The optical model has been val-idated and used for c-Si cells in chapter 3. The same approach is used for thin-film solar cells as well. However, some specific aspects being relevant specifically for thin-film solar cells, will be highlighted first. In section 4.2.1 the optical properties of various semiconductor materials used in thin-film solar cells are summarised. Sec-tion 4.2.2 is devoted to the textured transparent conductive oxide. Both the optical properties and the way of modelling are presented.
4.2.1 Optical properties of semiconductor materials used in thin-film solar cells
Amorphous silicon
Among the commercially available thin-film solar cell technologies, the amorphous silicon (a-Si) technology is the most important one in terms of production and instal-lation [8]. The word ‘amorphous’ indicates that, as opposed to c-Si, the atoms in a-Si show no long range order. Because of the different arrangement of atoms, the elec-trical and optical properties of a-Si are very different from those of c-Si. For example the dangling bonds, present in high concentrations in a-Si, act as recombination cen-tres, reducing the minority carrier lifetime. In solar cell applications hydrogenated amorphous silicon (a-Si:H) is used because hydrogen atoms passivate the dangling bonds. Note that often the term ‘a-Si’ is used when ‘a-Si:H’ is meant, also in this thesis.
In the left panel of figure 4.2 the refractive index n and in the right panel the absorption coefficient α of a-Si are shown that are used in the optical model. a-Si has a high bandgap energy of 1.7 eV (slightly dependent on the hydrogen content) [6]
corresponding to a small threshold wavelength of 730 nm. It can be seen from the rapidly decreasing absorption coefficient, that irradiance with a wavelength larger
0.4 0.6 0.8 1 1.2 1.4 1.6
Figure 4.2:The refractive index n (left) and absorption coefficient α (right) of the semiconductors used in thin-film solar cells (a-Si [65], µc-Si [65], a-Si/Ge alloy [65], and CuInGaSe2[66]). The properties of c-Si [47] are given for comparison.
than this threshold wavelength is hardly absorbed. Because of the relatively large bandgap and the corresponding low threshold wavelength, a relatively large part of the AM1.5 solar spectrum is not absorbed by a-Si.
Amorphous silicon/germanium alloy
A-Si is deposited using plasma enhanced chemical vapour deposition (PECVD). By introducing germanium (Ge) during deposition, an amorphous silicon/germanium (a-Si/Ge) alloy is formed. By increasing the Ge content, the bandgap of the a-Si/Ge alloy can be decreased. In figure 4.2 the optical properties of an a-Si/Ge alloy with a bandgap of 1.5 eV are shown. With respect to a-Si, the graph of the absorption coefficient has shifted to higher wavelengths and the refractive index has increased somewhat.
Micro-crystalline silicon
If during deposition of a-Si, conditions such as H2/SiH4-dilution ratio, substrate temperature and layer thickness are favourable, silicon crystallites with dimensions of tens of nanometres are formed [67]. This form of silicon is called micro-crystalline silicon (µc-Si). In figure 4.2 the refractive index and the absorption coefficient of µc-Si are shown. It can be seen that optically µc-Si is more similar to c-Si than to a-Si.
Copper indium gallium diselenide
An alternative semiconductor material, not based on Si, is copper indium gallium diselenide (Cu(InGa)Se2or CIGS). Because indium can be substituted with gallium
0.4 0.6 0.8 1 1.2 1.4 1.6
Figure 4.3:The refractive index n (left) and absorption coefficient α (right) of alu-minium doped zinc oxide as derived from measurements by Berginski [68], as pro-vided by Zeman [65] and as derived from measurements of sample 1 (which will be introduced in section 4.3).
and vice versa, the composition can be expressed as Cu(In1−xGax)Se2, where x is between 0 and 1 and represents the relative gallium content. The bandgap of CIGS increases with increasing relative gallium content x. For solar cell applications, CIGS with x = 0.2 is used because this gives the highest electrical efficiencies. Therefore this composition will be considered here as well. The optical properties obtained from Alonso [66] are shown in figure 4.2. Because of the relatively small bandgap of 1.2 eV, this semiconductor has a high absorption coefficient remaining larger than 104cm−1up to a wavelength of 1.1 µm.
4.2.2 Transparent conductive oxides
As shown in figure 4.1, at the front and the back of the semiconductor layer, contact layers are present to collect the holes and electrons respectively. To prevent ohmic losses as much as possible, these layers require a high conductivity. At the back side typically a metal layer is used, which serves both as electrical contact and as an optical reflector. As front contact, a layer is required which is not only conduc-tive, but transparent as well. Transparent conductive oxides (TCO’s) exist, such as (indium) tin oxide and zinc oxide, meeting both requirements. The conductivity of these TCO’s can be improved by adding a suitable dopant.
Optical properties
For solar cell applications aluminium doped zinc oxide (ZnO:Al) is the state-of-the-art TCO. Berginski [68] has studied the conductivity, the transparency and the sur-face topography of magnetron sputtered ZnO:Al. In this study it was found that an
increased doping level improves the TCO’s conductivity, but at the same time re-duces the transparency for the near infrared part of the solar spectrum. Berginski has measured the transmittance of ZnO:Al layers having doping concentrations of 1.1 · 1020cm−3, 2.7· 1020cm−3and 3.8· 1020cm−3. From these transmittance curves the corresponding absorption coefficients were derived. The curve corresponding to the doping concentration of 3.8· 1020cm−3is shown in the right panel of figure 4.3 and corresponds nicely with the absorption coefficient provided by Zeman [65]. The mechanism responsible for absorption in the near infrared is free-carrier absorption.
Characteristic for free-carrier absorption is the increase of its strength with increas-ing wavelength. This explains the increase in α that can be observed for λ > 0.7 µm in the right panel of figure 4.3.
Zeman has provided the data corresponding to the refractive index n, shown in the left panel of figure 4.3 as well. In section 4.3, sample 1 will be introduced containing a ZnO:Al layer. Using this sample the optical properties (n and α) of ZnO:Al were determined. These results are also shown in figure 4.3 and agree quite well with the existing data. In the numerical simulations that will be described in section 4.3 the optical properties derived from sample 1 will be used.
Surface topography
After deposition, the initially smooth TCO film is textured by wet-chemical etch-ing. Berginski [68] reports that depending on doping level and substrate temper-ature, different postetching surface topographies appear. Using scanning electron microscopy and atomic force microscopy, Berginski analysed the surface topogra-phies. A root-mean-squared (rms) surface roughnesses σ of approximately 100 nm can be obtained. At the same time this texture can have a steepness γ of 17◦to 32◦.
Because the rms roughness σ is comparable to the wavelength λ of solar irra-diance, the optical model uses haze parameter H when modelling the rough TCO surface. H is defined as the fraction of incident light that is scattered and is a func-tion of σ/λ. As described in secfunc-tion 2.5.4, when modelling textured TCO surfaces, haze parameter H is used as weighting parameter for scattered light. Texture steep-ness γ is a parameter determining the angular distribution of the scattered light. In section 4.3.5 the optical effects of the parameters σ and γ are studied in more detail.