The Potential of Thin-Film Crystalline Silicon Solar Cells
If the efficiency and cost targets can be met, thin-film crystalline silicon solar cells have the potential to become a solid alternative to the bulk multicrystalline silicon solar cells that currently dominate the photovoltaics market.
Koen Snoeckx, Guy Beaucarne, Filip Duerinckx, Ivan Gordon and Jef Poortmans, IMEC, Leuven, Belgium -- PV Society, 6/1/2007
The cost of silicon accounts for about half of the production cost of current industrial silicon solar cells. In order to reduce the amount of consumed silicon, the photovoltaics (PV) industry is counting on a number of options being developed in research. The most obvious one is to move to thinner silicon substrates. Presently, the thickness of silicon substrates for solar cell production is slightly above 200 µm, but technologies to cope with substrate thicknesses to slightly &100 µm are under development. To go for active silicon thickness as low as 5-20 µm, a layer of active silicon can be deposited on top of a lower-cost substrate. This approach is described by the term thin-film crystalline silicon solar cells. In order to be industrially viable, the challenge is to find the ideal trade-off between efficiency and reduced cost in a process suited for large-scale production. To manufacture the active silicon layer, several possibilities exist.1 We are investigating three such alternatives.
Thin-film PV basics
The first route is to create epitaxial thin-film solar cells (Fig. 1 ) by starting with highly doped crystalline silicon wafers (e.g., from upgraded metallurgical grade silicon or scrap material), and depositing an epi layer by chemical vapor deposition (CVD). Next to the cost and availability advantages, this approach enables a gradual transition from a wafer-based to a thin-film technology. As the process is similar to a classical bulk silicon process, this technology is easier to implement in existing lines than any other thin-film technology.
Secondly, in thin-film solar cells based on layer transfer, one could deposit a monocrystalline layer epitaxially on a porous silicon film that is — at some point in the process — separated from the substrate. The idea is to reuse the parent substrate many times, so that the final wafer cost per solar cell is low. An interesting option under investigation is the possibility of processing the freestanding film by separating the porous silicon film prior to epitaxy.
|
|
1. The epitaxial thin-film silicon solar cell approach features the advantages of substrate availability and likeness to existing wafer processes.
|
Finally, for thin-film polysilicon solar cells, a layer of only a few microns of crystalline silicon is deposited on a cheap foreign substrate, such as a ceramic (Fig. 2 ) or high-temperature glass. Polycrystalline silicon films with grain sizes between 1-100 µm appear to be particularly good candidates. We have demonstrated that good polycrystalline silicon solar cells can be obtained using aluminum-induced crystallization of amorphous silicon. This process leads to very thin layers with an average grain size around 5 µm. These seed layers are then epitaxially thickened into absorber layers several microns in thickness using high-temperature CVD with a deposition rate exceeding 1 µm/min. Ceramic alumina and glass ceramics are used as substrates. We selected thermal CVD because of the high growth rates and achievable crystalline quality. This choice, however, imposes the use of a heat-resistant substrate material such as a ceramic. This technology is not yet as mature as the other thin-film technologies, but shows high cost-reduction potential.
|
|
2. The thin-film polycrystalline silicon solar cell approach uses a ceramic substrate for lower overall cost.
|
Results in thin-film PV technology have already led to solar cells with high efficiency or processing ease, leading to lower cost. Combining both aspects is something no one has succeeded in at this point. Some recent results, however, take another of the many necessary steps in the right direction.
Improved epitaxial cells
For epi thin-film silicon solar cells, their moderate efficiency (around 12% for semi-industrial screen-printed cells) currently limits the attention that is paid to this cell type by the PV community. Compared with bulk silicon solar cells, similar levels of open-circuit voltage and fill factor (±77.8% for monocrystalline silicon solar cells) can be obtained. The short-circuit current (Jsc), however, is held back by the optically thin active layer (&20 µm). Light that traverses the epi layer is lost for collection in the highly doped, low-quality substrate. As a consequence, a difference in short-circuit current of 7 mA/cm² between both cell technologies is not uncommon. Bulk crystalline solar cells typically show Jsc values ~33 mA/cm², whereas epi thin-film cells average at ~26 mA/cm².
However, two separate developments at the cell level have positively changed this picture.2 By increasing the optical path length in the thin active layer, we reported screen-printed epi cells with Jsc approaching 30 mA/cm² and efficiencies of 13.8%.
|
| 3. Plasma texturing of the surface of epitaxial thin-film silicon solar cells leads to optical path-length enhancement and, therefore, to higher efficiencies. |
The first improvement that contributes to these results is an adaptation of the surface light scattering by fluorine-based plasma texturing (Fig. 3 ). Ideally, the textured surface of the active layer should be 100% diffusive (i.e., a Lambertian refractor). This would lead to photons moving through the active layer at an average angle of 60°, resulting in an optical path-length enhancement of two. In other words, a 20 µm layer would optically behave as if it were 40 µm thick. It was found that this complete light scattering could be achieved for a silicon removal of only 1.75 µm. The plasma texture brings a lot of advantages at the cell level: a lower reflection (down to 10% starting from 35% before texturing), oblique light coupling and a lower contact resistance (caused by the larger contact area between the silicon substrate and silver electrodes). An absolute improvement of 1.0-1.5 in Jsc is observed, together with an efficiency increase of 0.5-1.0%.
A second improvement was achieved by incorporating a porous silicon Bragg reflector for internal light trapping. To decrease the transmittance of long wavelength light into the substrate, an intermediate reflector was positioned at the interface between the substrate and epi layer. Photons reaching this interface can now be reflected and pass a second time through the active layer. Because the light is diffused from the moment it enters the cell (due to the Lambertian nature of the plasma texture), a large percentage of the photons will strike the front surface outside the escape angle. Therefore, most of the photons will be reflected internally for a third pass. The story repeats itself from that moment on so that multiple passes through the epi layer become possible (Fig. 1 ).
|
| 4. TEM of a reorganized porous silicon stack that serves as an embedded reflector for internal light trapping in epi thin-film solar cells. |
In practice, the reflector is made by electrochemical growth of a porous silicon stack of alternating layers with high and low porosity (a multiple Bragg reflector). During the epi growth of the active layer, the porous silicon stack automatically transforms into alternating layers consisting of small and large voids (Fig. 4 ). This structure has been proven to be ideal for the concept of a reflector based on constructive interference. For a 15-layer porous silicon stack, calculations lead to an optical path-length enhancement of 14. In other words, a 15 µm layer would optically behave as a 210-µm-thick silicon slab.
To validate both approaches, epi cells were processed on three different carrier substrates with cell surfaces of 18 cm². On monocrystalline silicon, serving as proof-of-concept, cell efficiencies increased up to 13.8%, with a fill factor of 77.8%, indicating that there is no conductance problem through the reorganized porous silicon stack. On low-quality silicon substrates, the results are slightly lower, with efficiencies up to 13.5% and a fill factor of 77.7%. The lower performance can be explained by the fact that the epi growth is of lower quality in the case of porous silicon on multicrystalline substrates. The optimization of this process is ongoing, and will most likely lead to higher efficiency gains in the near future.
Improvement in poly thin films
For another type of solar cells, namely polycrystalline thin-film solar cells based on aluminum-induced crystallization, we recently obtained a record efficiency of 7% for this type of cell. The cells were made on high-temperature substrates using seed layers based on aluminum-induced crystallization of amorphous silicon that were epitaxially thickened into absorber layers at 1130°C. It is important to note that this process does not make use of remelting of silicon. Remelting silicon on the ceramic substrate is an alternative approach to obtain polycrystalline solar cells. However, because of the extremely high temperatures that are needed for this approach (over 1400°C), it requires substrates with outstanding thermal stability and yields a high risk for contamination. The secret behind the achievement was a dedicated design and implementation of the cell contacts combined with plasma texturing of the surface.
|
| 5. A dedicated interdigitated contact design results in better electrical characteristics of polycrystalline thin-film solar cells. |
Since most high-temperature substrates suited for polycrystalline silicon solar cells are insulators, new metal contact schemes have to be developed, avoiding the use of back contacts. In view of low-cost manufacturing of modules, the most convenient approach would be to integrate the cell interconnection process with the cell fabrication. Our approach uses a monolithic module process in which the cell interconnection is combined with the cell contacting. All contacts are made on top of the cells in an interdigitated pattern (Fig. 5 ). Different process sequences can be used to achieve the novel contact structure. Presently, a simple two-step laboratory process is used, combining photolithography with metal evaporation. In view of mass production, the metallization should be performed in a single step by either screenprinting or evaporation through a shadow mask.
The dedicated contact design was implemented at the cell level on cells with an active area of 1 cm² and compared with cells that had base contacts at the periphery (Fig. 6 ). The open circuit voltages (Voc) were comparable for both types, but the interdigitated cells scored remarkably better in terms of short-circuit densities (Jsc) and fill factors. Depending on the grain size and thickness of the layers, efficiencies up to 5.6% were obtained.3
|
|
6. Polycrystalline silicon solar cells with contacts at the periphery (left) and a dedicated interdigitated contact design (right) leads to higher efficiencies. (Note: Layers are not drawn to scale.)
|
To further enhance the current density and, hence, the efficiency of the cells, we applied an innovative cell concept using plasma texturing. So far, polycrystalline silicon solar cells were made in substrate configuration with the substrate acting as a back reflector. Texturing the front side of the cells generates more effective light trapping by lowering the front reflectivity of the cells and obtaining better coupling of light into the cell. Plasma texturing was done in a reactor using a fluorine-based chemistry. As a result, the current density was increased by ~15% (results obtained for an alumina substrate). This increase in current density boosted the cell efficiency toward the record of 7.0%.4
However, while the obtained Voc (506 mV) and fill factor (71%) can be called state-of-the-art, current densities (19.7 mA/cm²) and cell efficiencies are still too low for commercialization. By optimizing the plasma texturing process and lowering the thickness of the back surface field layer of the cells, we expect to reach efficiencies well above 7.0% in the near future.
Conclusion
While the cell efficiencies of bulk crystalline solar cells are still in another league (up to 18% on 130 µm thin cells recently reported), the substantially lower cost potential of thin-film cells justifies the continuous and increasing research effort of the international PV community. By 2010, thin-film crystalline silicon solar cells on non-silicon substrates are predicted to reach a fabrication cost below 1 euro per Watt peak. Which fabrication route will finally lead to a manufacturable process is hard to predict at this stage. All have their pros and cons. For epi thin-film silicon solar cells, the most recent results are promising and encouraging, but efficiencies in the range of 14-15% should be attained before considering mass production. In addition, scale up of the equipment for growing epi layers, as well as the manufacturing of low-cost silicon substrates, are requisites for successful introduction into the industry. Also for polycrystalline thin-film solar cells, envisaged efficiencies over 7% are going in the direction of what is needed for commercialization, certainly because of the very low cost of the substrate. Major obstacles to overtake remain the optimization of the production processes and their conversion toward more standard solar-cell production steps.
Acknowledgements
This work was partially funded by the European research programs Athlet, Meteor and Latecs, as well as the FP5 and FP6 programs Epimetsi, Sweet and Crystal Clear.
R&D Pushes c-Si Solar Cells Through Hurdles
08/31/2009Si and Ge Solar Cell Efficiency Records Set
10/11/2007Schott Solar Joins IMEC Silicon PV Program
06/03/2009Solar Cell Technology Trends
08/20/2009
