The high-temperature deposition of silicon is mainly accomplished by two methods, liquid phase epitaxy or chemical vapour deposition (CVD), the latter technique is explained in detail chapter 3. Silicon deposition performed at temperatures above 1000°C typically produces good crystal quality. On substrates with a similar lattice match to silicon, epitaxial2 growth occurs and results in poly-, multi- or even monocrystalline silicon films, depending on the
2 Epitaxy is deviated from ‘epi’ and ‘taxis’, which means in greek ‘upon’ and ‘ordered’, respectively. Epitaxy denotes the growth of crystals of a material on the crystal base of another material, such that the crystalline substrates of both materials have the same structural orientation [23].
substrate structure (Figure 2-1). Despite the high temperature, if a foreign material substrate is used, a polycrystalline film is deposited and grain enlargement becomes necessary to reach acceptable efficiencies. As a rule of thumb for good efficiencies, the diffusion length in cSiTF solar cells should be more than 2 times longer that the thickness of the layer. Due to the fact that the substrate is heat-resistant, high-temperature recrystallisation techniques can be applied, e.g. rapid-thermal annealing, laser, liquid-phase or zone-melting recrystallisation. However, the requirements for the substrate are quite demanding and there are less suitable substrates available compared to the low-temperature approach. Furthermore, the purity of the substrate has to be high enough so that only few impurities can diffuse into the active layer during the deposition. For substrates with a high number of impurities, an intermediate layer is necessary to act as a diffusion barrier. Three main concepts within the high-temperature approach are presented in this section:
- the epitaxial thin-film silicon solar cell on a low cost silicon substrate;
- the recrystallised thin-film silicon solar cell on foreign, conductive and non-conductive substrates;
- and the lift-off thin-film silicon solar cell, where the silicon substrate is re-used.
2.4.1 Epitaxial wafer-equivalents
The simplest form of a crystalline silicon thin-film (cSiTF) solar cell is an epitaxial active layer deposited on a highly-doped inactive silicon substrate (Figure 2-2-A). Due to its resemblance to a silicon wafer solar cell it is also called the epitaxial wafer-equivalent (EpiWE). Standard industrial solar cell processes can be applied to this wafer-equivalent.
CSiTF solar cells were first fabricated in the 1970s. Chu et al. presented epitaxial silicon thin-films deposited on purified and uni-directionally solidified mg-Si substrates. They reached efficiencies up to 9.7% measured under an AM1 spectrum [24]. Despite this effort, little progress was made with EpiWEs until years later. In the 1990s, high-efficiencies were investigated with a best result from a 45 µm thick epitaxy layer on an Fz substrate with an ion-implanted SiO2 layer beneath the surface. With this sophisticated SIMOX3 approach and a high-efficiency solar cell process, a 19.2% cell high-efficiency was reached [25].
3 SIMOX: Separation by IMplanted OXygen.
Nowadays, more cost-effective substrates and solar cell processes are used. An attractive option to reduce the substrate costs while keeping the impurity level of the substrate low is the recycling of waste silicon. Such materials are highly-doped reclaimed wafers from the microelectronics industry or ‘tops and tails’
from Czochralski-grown silicon, which are cast to multicrystalline silicon (off-spec cast mc). More abundant, but also more impure substrates are metallurgical grade silicon (mg-Si) or up-graded mg-Si (umg-Si). The impurities diffuse through the material easily at high temperatures and it is difficult to develop a process that prevents the formation of recombination centres.
As the substrate and the deposited layer in the epitaxial Si solar cell have the same refractive index, no internal reflection occurs at this interface. In order to introduce optical confinement, an intermediate layer with a different refractive index must be identified that also allows good quality crystal growth. This could be achieved by using e.g. porous silicon, of which the porosity and therefore the refractive index can be adjusted [26, 27]. The advantage is that epitaxy is possible on porous silicon without sacrificing the quality of the epitaxial layer.
This method was also part of this thesis project and is described in more detail in Section 6.3. Another option is to deposit a perforated SiO2 layer, which is then overgrown by the epitxay. A selective epitaxial growth through a pattern of openings with liquid phase epitaxy was already developed by [28]. The oxide layer has a perfect refractive index for this purpose and acts additionally as a diffusion barrier.
At present, research with optimised solar cell concepts is underway. E.g. the emitter wrap-through concept is known to reduce the grip shadowing [29]. A concept combining the advantages of the wafer-equivalent approach and the emitter wrap-through concept is currently under investigation [30].
n+emitter 0.4 - 1 µm
Figure 2-2: Thin-film concepts pursued at Fraunhofer ISE: Epitaxial wafer-equivalent (A), Laser-fired rear access (LFA) (B), recrystallised wafer-equivalent on ceramics (C).
2.4.2 Zone-melted crystalline films
By zone-melting recrystallisation (ZMR), microcrystalline silicon films are melted and cooled in a controlled manner so that the grain size is enlarged to over 10 mm2. A long, narrow silicon strip is melted and this liquid zone is scanned across the sample. Thereby, elongated grains of up to 10 cm in length and several millimetres in width are formed [31, 32]. The dislocation density varies widely with crystal orientation [24, 32] and much effort has been invested into achieving large and homogeneous recrystallisations [33].
A new concept called laser-fired rear access (LFA) is currently under development at Fraunhofer ISE. A scheme of these concepts are shown in Figure 2-2-B [34]. On a low-cost and impure silicon substrate, an SiO2 diffusion barrier is deposited and subsequently a silicon layer. The microcrystalline silicon is then zone-melted. A subsequent laser processing creates holes, which penetrate the insulating SiO2 layer. The holes are filled up by an epitaxially deposited base, establishing the connection to the rear side contact (hence the name). This approach combines the advantages of an intermediate SiO2 layer (light trapping and diffusion barrier) with a simple two-sided cell metallisation process. Recently, efficiencies up to 8.4% have been reached at Fraunhofer ISE [34]. Mitsubishi electric corporation presented a 16.5% efficient cell with an SiO2 intermediate layer and zone-melted film, but more complex cell concept [32]. This proves the potential of this concept.
When using foreign substrates, such as ceramic or graphite, a recrystallisation of the microcrystalline silicon film becomes necessary. On non-conductive substrates, sophisticated cell structures are applied, such as front side contacts.
These solar cell processes are too complex for a low-cost industrial application and therefore efforts are channelled to the development of conductive substrates and intermediate layers [35]. Figure 2-2-C shows the scheme of a cSiTF solar cell on a ceramic substrate with two-sided contacts, an SiC intermediate layer and ZMR recrystallisation. Efficiencies up to 7.2% were reached so far [35].
2.4.3 Transfer techniques
A compromise between high-temperature epitaxial deposition and low cost substrates can be reached with transfer techniques. A high-quality cSiTF is deposited on a host substrate by epitaxy. The emitter and front contact formation are done while the cSiTF is still attached to the substrate and only after completion of the front-side cell structure the cSiTF is lifted-off and transferred
to a low-cost substrate. The host substrate can be electrochemically pre-etched to form a highly porous silicon layer (PSI). The porous structure of the layer changes during thermal annealing such that large voids are formed and separation is easily possible. On top of this layer a layer of low porosity is etched so that an epilayer of good quality can be grown. The subsequent solar cell process can be simplified, e.g. by autodiffusion of the emitter when using a highly-doped n-type substrate [36]. The cSiTF is separated mechanically and glued to a textured glass superstrate for mechanical protection. The host-substrate can be reused several times; however, the quality of the epitaxial layer worsens [28].
Another transfer technique is the Epilift concept, where the epitaxy is grown by LPE on a monocrystalline substrate with a mesh-like SiO2 layer. The cSiTF has then a waffle-grid structure, which can be contacted using an interdigitated grid. Efficiencies up to 13% have been reached [37].