Growth in the solar PV sector should be accompanied – industry tells us – by a reduction in price. And while cells and modules will only ever account for a proportion of the total installed cost of a PV system, lower manufacturing costs are an essential for the much more widespread adoption of solar PV. So what does the industry say about its future? How can it shave costs? And how can the manufacturing equipment sector contribute? These were the subjects of discussion at SEMI’s PV Fab Managers’ Forum held in Leipzig in March.According to the European Photovoltaic Industry Association, the dramatic growth experienced by the photovoltaic sector over the past five years (averaging 40% per year) is expected to continue until 2010, after which it will enter a period of stabilization, with growth nearer 26% per year. Altogether, the global photovoltaic industry is expected to invest €2.6 billion in new production capacities this year, and EPIA estimates that by 2010 investments needed to meet increasing demand will reach €14 billion. With further growth should come further price reductions, which in turn should spur further demand.
So, what is needed next to achieve this growth, especially from the manufacturing sector?
Anton Milner, CEO of Q-Cells, was one of the first to address the Fab Managers’ Forum in Leipzig, stressing – like other speakers – the importance of improvements in equipment and manufacturing to bringing down the costs of solar PV in the future. As a PV manufacturer, his dream is ‘the same as it ever was’. It’s to see terawatts of photovoltaics installed, with PV cost-competitive and available to the world – and that includes off-grid markets. He reminded the audience that an IEA report anticipates that by 2030 no fewer than 3.4 billion people are expected to be without access to electricity, and that PV can provide the ideal solution.
Meanwhile, he said, the industry continues to be characterized by growth, doubling in size every second year. In 2006 statistics will show that cell manufacture reached 2.3-2.5 GW, the great majority of which is crystalline technology. By 2010 he expects this to reach 10 or even 11 GW – and he anticipates that thin-film will make up 20%-25% of total production. The market will start to look very different.
What’s more, this growth is taking place when the industry is experiencing the most severe supply-side constraint it has ever known – and it is still doubling, with a silicon shortage! ‘We’ll really start moving afterwards’.
What about markets? Up to now, said Milner, the industry has lived in supported markets – and rightly so in order to stimulate growth. But this is no longer sustainable – the industry needs to develop technology further, and bring down costs so that it can access the electricity supply markets and compete on costs. ‘In California we are already there’, he said, but for mainstream markets there is still some way to go, needing to reduce costs by 40%-50 % in the next few years. Yet most of the innovations within the industry are not fundamental, but rather, are more a matter of making things happen. And sometimes quickly, as in the case of thin film.
In terms of costs per megawatt, we haven’t seen too much of a step down in the past few years, says Milner. ‘We need to reduce costs by 40%-50% in the next few years. It needs a radical rethink’.
So what is his view in terms of specifics? First, cell efficiency has to increase by 2%-3%. And wafers must become significantly thinner. New generations of silicon cells are on their way with efficiencies of over 18% (this could be as high as 22%-24% for monocrystalline cells).
In order to radically increase the production volumes of PV, new manufacturing methods will be required sharp
Many improvements will be made in the manufacturing process, such as making modules out of back-contact ‘pick and place’ cells. Steps like this could take a lot of the cost out of the chain, says Milner.
The main job is to get costs down across the board and ramp up the scale of production. Milner told the Fab Managers’ Forum in Leipzig that ‘we should expect to see a lot more variety in the future, with more new processes and efficiency concepts … this is becoming a multi-technology environment’. He does not believe that there will be one winning PV technology. ‘The main job,’ said Milner, ‘is get costs down across the board and ramp up production.’
Other speakers concurred – several of the large manufacturers who grew their businesses around crystalline silicon also prefer to back more than one horse, and are diversifying into a wider range of PV technologies. For example, ErSol is currently setting up its new 40 MWp production facility for amorphous thin film modules in Erfurt, while Q-Cells is also broadening its offering – initially through its collaboration with Evergreen Solar on the Ever-Q product.Delivering gigawatts
Milner told the audience that Q-Cells uses the same process logic and equipment setup as when it was designed back in 2000. Others concurred that today’s manufacturing is essentially a scaled-up laboratory processes. Now it’s time to rethink the processes, in order to gear up for ‘immense volume growth’. Milner believes the fab of the future will produce hundreds of megawatts, then gigawatts. ‘The future won’t just require more manufacturing lines, it needs new dedicated technology, an expanded way of production.’
Ingo Schwirtlich of Schott Solar presented a ‘shopping list’ of step-changes that the industry needs to implement in order to achieve this kind of growth. Production line capacity, he said, needs to move from today’s 30-60 MWp/year to 100-200 MWp/year. He does not believe this can be achieved simply by adding more lines of a similar nature of those that already exist, but it requires use of in-line and cluster systems (see ‘two concepts’, below). Furthermore, the lead time for new manufacturing plants needs to come down from 9-12 months to under six months; new plant ramp-up time needs to be reduced from 3-6 months to under one month – with more equipment manufacturers offering piloting facilities.
One specific area for improvement, says Schwirtlich, is in standardization of the interfaces across the industry – (electrical, mechanical, human-machine, standardization of warning lights, and so on). At present there is no such standardization in place. (This is precisely the type of area that the broader semiconductor industry has the experience to help put in place.)
Next, the cost of new manufacturing investment needs to go down, said Schwirtlich – at present investment is in the range of €0.4-0.6 million/MW, and this needs to drop €0.2 million/MWp/year.
As a step towards this, Schwirtlich anticipates a reduction in the number of people employed in manufacture (in terms of MWp produced), as the manufacturing systems become more sophisticated. At present, he estimates that manufacture requires 1-2 employees/MWp/year, a figure that with fully automated process sequences and advanced central process control, recording and visualization and improved reliability of the manufacturing equipment could be below 0.5 employees/MWp/year.
He also envisages slicker processes, with a decrease in system downtime – uptime could be more than 95%, compared with today’s typical figure of 85%-90%,
Proper software integration is key to the efficient operation of a PV manufacturing plant solar integrated technology
Like other industry representatives, Schwirtlich says that the 100-150 µm wafers that the industry is headed towards, in its quest to minimize use of expensive silicon, will need low-stress wafer handling and transport systems, and adapted processing equipment. (At the conference, one equipment-handling supplier got a round of supportive applause when he asked to be given samples of 180 µm thin wafers to experiment with ‘so we can try to deliver’ the right equipment.)
On the thin cell front, Lars Podlowski – of module manufacturer Solon – outlined some new technological concepts that could be used with thinner wafers to help avoid breakage – such as ‘soft’ soldering using laser, or soldering via induction, which appears to reduce breakage compared with use of infra red or hot air. And when it comes to handling, he believes pneumatics will no longer be the answer, as they simply can’t be gentle enough. His company is already switching to 166 µm thick cells from one supplier.What industry wants
ErSol’s Technical Director, Rüdiger Schulz, believes that suppliers of manufacturing equipment to the industry need to focus much more closely on their customers’ needs. His complaint is not against any one supplier – rather, it is an observation he has made over many years of working with different companies.
He believes strongly that suppliers of manufacturing equipment need to have a training centre on-site. ‘You can’t train within a production environment’, he says, ‘you can’t wait until a tool is broken to show how to fix it!’
Too often, companies try to sell their solution, rather than focusing on customer needs, he said. When it comes to delivery, the industry really needs reliable delivery dates – it is less whether delivery time is two weeks or five months than that the stated date should be adhered to.
And after-sales can be very poor, not really offering solutions when software fails to function, or not always offering a service hotline. Among his frustrations is the apparent reluctance of some suppliers to accept customer feedback: ‘some assume that the PV engineers are being stupid’. ErSol expects that the next version of any tool they buy will incorporate their suggestions or requirements.Two concepts
Schulz also outlined two different manufacturing concepts that the industry can adopt to help it rapidly increase the scale of its operations. The first – inline manufacturing – connects all the manufacturing steps into one process, with no human handling, and with smart automation. Smart automation is essential, because if one piece of the chain goes down, it is difficult to deal with.
The alternative concept is called a process cluster. In this case the tools are not really connected, leading Schulz to believe that this solution offers much more flexibility in periods of down time, or if a tool needs to be exchanged or upgraded.
In both these scenarios there will also be a great need for more advanced software, and equipment interfaces, particularly with regards to increasing the viability of remote access.
Another of the key issues under debate in Leipzig was whether or not to track the progress of individual cells – tracking by means of unique serial number is common practice in the semiconductor industry. The debate is whether the benefits of tracking are worth the extra expense involved.Consensus
There appears to be consensus on parts of the ‘radical rethink’ required for PV manufacturing. First, solutions will come across a range of PV technologies. Second, crystalline silicon manufacture will require systems that can handle very thin wafers. Many manufacturers are looking for flexibility and upgradeability of their manufacturing kit, and there is general agreement that the time has come to introduce much more interface standardization in order to achieve this. There’s a general agreement that it’s time for photovoltaics manufacturing to complete its transition from the provision of bespoke ‘ad hoc’ solutions that are effectively an upscaling of lab technology, to a more mature industry that benefits from the advantages of greater standardization. Underpinning all this are the reductions in the price of silicon which the industry is expecting – crucial, since silicon accounts for around 70% of material costs. And throughout, maintaining quality to allow 25-year warranties remains an essential prerequisite.
Written by members of the Renewable Energy World team
e-mail: [email protected]
Hands-on in China
Suntech Power’s COO Graham Artes told the conference about a different system – how his company’s ‘hands-on’ approach works in China. Working on a 24-hour, seven-day-a-week basis, his company is currently producing 4500 modules per day – 0.8 MW – filling seven shipping containers daily, from its base in Wuxi. (The company has a second base in Luoyang.) The process involves 2000 employees, no fewer than 300 of whom work in quality control.
Artes insists that ‘the human touch’ (which is not the approach adopted throughout that country) ensures far lower breakage rates for cells than automated systems. ‘With 0.018% breakage rate there’s no need to buy equipment’, he argues.
Artes explained the very strong team culture, with good training and supervision. Staff are trained in depth and continuously for four weeks before they go into ‘live’ production. This enables them to familiarize themselves with the material and with handling it.
Standards are high, with, for example, the soldering recalibrated every day, after every run.
A strong sense of process ownership helps to reinforce this. Artes explained that each manufacturing stage is controlled by a ‘process owner’ who is responsible for that process alone – no multitasking. Each of those process owners ensures that the equipment is inline to the processes that follow. Downstream of the process owner is a maintenance group, and then individual operators.
Currently Suntech tracks its wafers manually but is introducing process control next year ‘it’s very important to be able to track’, said Artes.
The staff – mostly female and young, with 200 working per shift – work quickly and efficiently. A worker can tab a cell in 15 seconds, then passes it on to next colleague. Flexibility is built in: if you want to change the product in some way ‘just tell people,’ says Artes, ‘and they will do it’.
The R&D department plays an integral part in the regular production process, having responsibility for its own 30 MW production line. When a new batch of wafers come in ‘whatever thickness’, explains Graham Artes, the R&D department sets up the process and runs it on its line, passing on the relevant know-how when that batch of cells goes into live production.
He acknowledges that such a high-energy approach – in terms of human input – can probably only work in China, and that there is probably an optimum operation size above which it probably cannot function optimally.