The most recent wave of thin-film excitement came to an end in 2009, when the price of silicon dropped from great heights to its present, more sustainable level. As a result, German company CSG Solar was one of the many start-ups which struggled to survive. At the time, CSG Solar’s approach, which was based on Australian research, was to deposit 1–2 micrometre (µm) layers of silicon on borosilicate glass and then crystallise it through the solid phase, by a long oven anneal, thus combining the quality of crystalline silicon with the low cost of thin-film.
With its cost per watt no longer competitive in the new market conditions, the company recognised that it was time to get out of manufacturing and back into research. With the resources it had left, CSG Solar searched for a fundamental change that could overcome efficiency limitations of around 10 per cent, and also simplify the process sequence.
There was no question of moving away from silicon for the company. While cadmium telluride and copper indium gallium selenide were looking promising, the relative scarcity of tellurium and indium meant that they could only be short term solutions. On the other hand, the abundance and non-toxicity of silicon positions it as a material that could meet energy demands almost indefinitely.Article continues below…
An Australian Research Council (ARC) linkage grant with the University of New South Wales (UNSW) assisted CSG Solar’s search for an improvement on solid-phase crystallisation, and out of this came a laser crystallisation technique which produces crystal grains of up to millimetres in size which are virtually defect-free. It was enough to impress Suntech Power, which bought the Australian research subsidiary of CSG Solar, renaming it Suntech R&D Australia. The ARC collaboration with UNSW continued and has now been strengthened with a new Australian Solar Institute project grant.
The laser crystallisation technique
The new process sequence begins with deposition of intermediate ‘buffer’ layers onto a sheet of borosilicate glass. Around 10 µm of silicon is deposited by electron beam evaporation. A line-focused, diode laser beam is then scanned across the surface, melting the silicon, which recrystallises upon solidification, forming the absorber layer.
The emitter is then formed with a spin-on dopant source and thermal diffusion process. The material quality is further improved with plasma hydrogen passivation. All metallisation is done on the rear side to create a superstrate configured solar cell, with sunlight entering through the glass. Light that is not absorbed in the first pass can be reflected off the rear layers back into the silicon.
The results so far are very promising. Efficiencies over 8 per cent have already been achieved, setting a new benchmark for a laser crystallised solar cell. Projected increases from improved metallisation are expected to take the efficiency up to over 10 per cent in the near future, challenging the records for single-junction thin-film silicon.
Most impressive is the open-circuit voltage. With values of over 550 millivolts already measured, the limitations of previous work in thin-film polysilicon, including the CSG Solar solid-phase crystallised technology, appear to have been overcome. Modelling, based on present material quality, suggests that efficiencies towards 14 per cent are possible, merely with improvements to metallisation and light-trapping. Further work on the laser crystallisation, passivation and junction formation processes are expected to allow increases even beyond that.
It is not only the efficiency that makes this technology promising. Fabrication techniques suitable with large-scale manufacturing are an important part of the development process. While CSG Solar’s approach used slow plasma-enhanced chemical vapor deposition tools designed for amorphous silicon modules, electron-beam evaporation has been the basis of this new technology, allowing silicon deposition rates up to 1 µm per minute.
The laser crystallisation process is also much faster and cheaper than the solid-phase crystallisation process previously used, which required oven annealing for around 24 hours. The laser process is also scalable. While a 12 mm long laser beam is presently used to crystallise small research samples in just a few seconds, LIMO GmbH, the company that supplied the research prototype, has laser systems currently available with beams up to meters in length, which are capable of processing several panels per minute.
This will be an important point of research in coming years – to understand how the beam length and shape affects the crystallisation process, and how that can be optimised to ensure cell efficiencies are just as good or even better when the system is scaled-up.
While other aspects of the process development plan are simply waiting to be ‘ticked off’, the road to commercialisation still has one significant unknown. The borosilicate glass used presently is far too expensive to compete with established photovoltaic cells. At present costs, any saving due to silicon reduction would be wiped out in glass costs.
It is a clear necessity to move to cheaper, soda-lime solar glass. This, combined with efficiencies of around 15–16 per cent, would likely make laser crystallised silicon competitive with standard modules in mainstream markets.
A few ‘toe in the water’ experiments have been conducted, which show that the process also works with soda-lime glass and focus will likely shift to this work in the near future. For the time being, however, research will continue with borosilicate in order to determine the performance potential as quickly as possible.
There is still is some time left for research. With balance of systems currently dominating system costs, the present premium on efficiency is a disadvantage for thin-film. The challenge is to be ready when the pendulum swings back to dollar per watt advantages. Then we may see thin-film crystalline silicon finally realise its potential.
Jonathon Dore worked for CSG Solar five years in Australia and Germany across several areas, including production engineering, process development engineering, performance monitoring and optimisation, and research and development. Mr Dore began full-time PhD studies at UNSW in 2011 while working part-time for Suntech R&D Australia.