There is no denying that things are tough for solar companies at the moment, as the recent bankruptcies of three well known players in the U.S. illustrates (Solyndra, Evergreen and SpectraWatt).
But despite the gloomy headlines it wasn't all bad news. The last quarter of 2011 saw a 23% rise in the European solar photovoltaic (PV) market, creating short-term optimism in the industry, according to NPD Solarbuzz. And more recently, it was announced by Bloomberg New Energy Finance that US clean energy investments surpassed those of China for the first time since 2008 (though important Federal support mechanisms are set to expire at the end of 2012).
And last year saw a plethora of company annoucements from around the world featuring all manner of technology developments, from casting elimination; to N-type silicon; thinner wafers; ion implantation; electro-deposition and developments in organic materials and plastics to name a few…
It's clear to see that despite the problems, innovation does goes on, and is plain to see right along the value chain.
1366 Technologies, for instance, is a startup that grew out of a group of Massachusetts Institute of Technology (MIT) techies – the obscure name highlights the geeky heritage – it has something to do with the amount of solar flux that penetrates the earth's atmosphere. The company has bold ambitions.
“We need to take solar up to the Terawatt scale,” said Craig Lund, vice president of Business Development at 1366 Technologies. “That has to mean a product that is cheap and a supply chain that is financeable.”
1366 is focusing on the development of a kerfless (no waste through cutting) wafer which requires no casting of cells. That eliminates what Lund called the “choke point” of silicon production. As well as time required, casting accounts for over 50% of the cost of silicon, 75% of plant CAPEX, and wastes over 40% of the material in the process. Gone would be all the fighting over silicon supply and the energy requirements of cell production would be considerably curtailed.
“Our approach would double silicon supply overnight,” he said.
He criticised the “primitive process” used to produce wafers, which has only minor differences between manufacturers and harnesses the same core tools from supplier to supplier: Place silicon into crucibles; drop it into the furnace; cast an ingot; saw it into bricks; and slice it into wafers. That amounts to 48 to 60 hours for casting, and another 12 to 14 hours for wafering.
Using the 1366 methodology, a much smaller furnace is used, to produce a 200 micron wafer every 20 seconds. The company has no immediate plans to invest in the development of thinner wafers.
“50 micron wafers sounds good but who buys them?” said Lund. “There is technological risk in thinner wafers.”
1366 is currently moving into commercial production. Lund claims that uniformity is better – within 1% compared to 1.5% for traditional wafering, consumes only 10 UScents per kWh and cuts silicon waste from over 40% to less than 5%. The company has no plans, though, to go into the PV module business.
“We view ourselves as a wafer manufacturer,” said Lund. (Editor's note - see 1366's cto Emanuel Sachs appear on one of our webinars a few years ago; time for an update maybe?)
The return of N-type silicon
N-type silicon was the original form of silicon in the early days, but it was soon replaced by P-type. Dr. Dengyuan Song, CTO of Yingli Solar, believes the tide is turning to N-type.
“There is great interest in re-investigating N-type Si because phosphorus doped n-type Si has many advantages over the standard boron doped Si substrate,” said Song.
Among the benefits, he said, were lack of oxygen introduced into the silicon (which helps reduce LID), higher efficiency cells compared to P-Si (due to less recombination of metal impurities) and greater annual yield of modules due to higher performance in weak light.
“Major metal impurities are positively charged in P-type, which are neutral or negatively charged in N-type Si,” said Song.
To date, three companies are in mass production of N-type silicon – Yingli, Sanyo and SunPower. Yingli, to date has managed efficiencies of just under 20%. It offers Yingli Panda Technology as a low-cost N-Si solar cell and module developed in collaboration with ECN and Amtech.
“We are using bifacial cells based on N-type silicon,” said Song. “This is compatible with standard cell and module manufacturing methods.”
Bifacial cells are employed to avoid full-area rear metallisation. There is also less cell bow with temperature variation due to no bending of the cell. Higher infrared light harvesting is another feature which is attributed to better backside passivation.
For now, the company is concentrating on a mono-crystalline N-Si substrate with a phosphorous diffused back surface field (BSF), a SiNx passivation coating and contacts of silver. The textured front surface of SiO2 and SiNOx provides an anti reflection coating. A boron diffused emitter is included.
The company currently provides several cells known as the Panda 325, 265 and 201 models. The Panda 325, for instance, provides 46.08 volts.
“These N-type units are very simple to manufacture and have enormous potential for high efficiency,” said Song. “While it is all mono right now, why not start developing multi-crystalline N-type Si?”
Dr. Doug Rose, vice president of technology strategy at SunPower, is another proponent of N-type silicon.
“N-type should reach 10% of the total silicon market by the end of this year and from there will be ramping up considerably,” said Rose.
He outlined his company's expansion plans as well as ongoing technological advancements in back contacts, backside mirrors, thinner wafering and more. This is being coupled with a ramp up of capacity that should see the company able to produce 1 GW per year by year end.
SunPower, of course, released the first production all back contact cell in 2005. The company makes use of screen printing, and copper metallisation, and that scored 20.6% efficiency for the first generation.
“A back contact architecture has the advantage of lower Si usage,” said Rose. “But our first generation had limitations due to lateral transport loss and the presence of dark areas.”
That led to a Generation 2 cell architecture to address these issues. Rose estimated that the cost is currently down to US$1.71 per Watt for silicon manufacturing. However, this newer design permits thinner cells – currently the fab thickness is down to 145 microns.
“Our modules are performing well in the field,” said Rose. “The back contact cell design has reliability advantages including low corrosion, and runs cooler than other systems.”
To achieve US$1 per Watt modules with 20% efficiency by 2014, SunPower is investing in a third generation approach, primarily through process simplification. Further down the road, it has a Gen 4 thin silicon target of 60 UScents per watt. But Rose doesn't envisage ultra-thin wafers any time soon.
“Once you get to 50 microns, efficiency drops sharply,” he said.
He ended with a call for integrity in the warranty arena. He claimed insider knowledge of how the original vendor to propose a 25-year warranty came to offer it. An engineer told the ceo of that firm that it could manage perhaps 10 to 12 years at best. The ceo thought 25 years sounded better and it was announced shortly thereafter. Within a year, the rest of the industry had aligned itself with that “standard.”
“The industry has been catching up to that 25-year promise for many years,” said Rose. “But while we couldn't deliver in the past, now we are getting there.”
Ion implantation of selective emitters
A different perspective on silicon doping was provided by Dr. Russell Low, director of technology, for Varian Semiconductor Equipment Associates. He described the process of ion implantation as a means of creating high efficiency selective emitters. The basic concept is that ions are implanted into a solid, to change the physical properties of the solid.
“Ion implantation is a proven doping technique for N-type and P-type high efficiency cells,” said Low. “It is a single-sided doping approach, which eliminates edge isolation.”
That means no dead layer or glass layer, good blue light response and uniformity. Implantation, however, does damage the crystal, so it needs to be annealed and oxidised prior to adding the emitter.
“A laser has a 0.1% loss in efficiency due to it cutting the surface,” said Low. “Ion implantation is more efficient and takes out one step.”
He claimed a 2X reduction in emitter saturation current at no extra cost. In terms of cell efficiency, he's talking about 18.6% with little variation. Suniva is an example of a vendor harnessing Varian's Solion system in this manner – its capacity is 1100 wafers per hour. Low stated that Suniva's POCI3 modules are 4.94% more efficient by using this technology.
“We can bring emitter doping with ion implantation up to 19.3% efficiency in the immediate future for patterned phosphorous P-type selective emitters,” he said. “For the N-type, the goal is 20%.”
A one-year old Silicon Valley startup known as Encore Solar showcased a technology known as electro-deposition or electroplating. Dr. Bulent Basol, founder and cto of Encore Solar, drew attention to the fact that this technology held the highest CdTe efficiency until recently. Electro-deposited CdTe made by BP was tested at 11% efficiency in the year 2000.
Encore Solar, therefore, is taking this method and adding innovation developed in the last decade. In particular, it has come up with an alternative to the CdTe vapour technologies, which dominate today. In essence, hot CdTe vapour is pushed through a nozzle to make the film on the substrate when it condenses. This vacuum-based system operates at 500 to 650 degrees Centigrade (C.) and produces a 3 to 4 micron thick film.
Encore Solar, on the other hand, is conducting low temperature electroplating. It deposits a layer of CdSi and another of CdTe, and treats them with chlorine at up to 550 degrees C., as a means of improving electronic quality. The formation of thin film is accomplished using a conductive electrolyte containing some ionic species.
“We can create uniform films on a large area of substrate of any desired thickness,” said Basol. “Vapour methods can't go to very small thicknesses as you end up with islands.”
As electroplating is carried out below 100 degrees C., it has better thickness and smoothness control. Material utilisation, he added, is almost 100%, an important factor with Te being so scarce. Another advantage is compositional control within the deposited layers.
That isn't to say that it is a simple approach. Basol admitted that there were abundant complexities in electro-deposition due to the chemical interactions involved, and that it requires a high level of in-house expertise. Further, the deposited films have small grain size typically. A treatment stage is inserted, therefore, to grow small grain size, which is done at around 350 to 400 degrees C. This boosts grain sizes up into the 1 to 3 micron range.
Film quality defects can also be picked up due to substrate surface problems. That is addressed typically by stringent quality control.
“Electro-deposition is the only technique that has excellent CdTe composition control at low temperature,” said Basol. “The technology can produce uniform layers in the 0.7 to 1.5 microns, whereas vapour techniques can only get down to 3 microns.”
The deposition process itself requires the useage of the optimum voltage. Te, it turns out, deposits first at small voltages, while Cd deposits at larger voltages. There is, however, an ideal CdTe deposition zone in the mid-voltage ranges. The result is 50% savings in Te usage. To date, modules with 11% efficiency have been demonstrated. Basol viewed 17% efficiency as being achievable.
From silicon to carbon
With so many silicon challenges, Howard Berke, ceo of Konarka Technologies, advocates what is known as Organic PV (OPV) – in other words switching from silicon and other elements, to simple carbon-based organic chemistry substances commonly used in plastics. One consequence could be to help integrate PV into urban landscapes.
“Does the USA really need 15,000 square miles of PV farms for its electricity,” said Berke? “PV should be able to be built into human structures, be comfortable with curved surfaces and be proximate to the consumer. It also needs to be more intelligent in partitioning the available light.”
The company is commercialising a semitransparent PV film with tunable colors. This makes it thinner and lighter than other thin film methods, while being rugged. As it is printed (just like printing paper) it has the lowest energy consumption in manufacturing. The end product is flexible, and printable in various widths. Berke also claimed it was more sensitive to low light than amorphous silicon.
In terms of usage, it can be built into tents, blinds, bus shelters and canopies. Konarka has demonstrated its workability by using it as curtains on south and east facing walls in one commercial property. Interestingly, the east wall had higher energy production.
“OPV has no issue with permanent shading and its off-angle sensitivity is high,” said Berke. “Efficiency is currently around 8.3%, we are working to push it over 10% and the practical limit single junction efficiency is 18%.”
The ink itself is based on P-type and N-type organic semiconductors. A high-speed printing process gives an output that is 6 nanometres in thickness and 1.5 metres wide. As it can be printed in long rolls, the company believes it can attain a 1 GW annual production capacity at full scale in its new facility. Lifetime is estimated at 8 to 10 years.
“By using the earth's most abundant elements we outperform other PV technologies in the real world,” said Berke:
“OPV produces more energy than CiGS and silicon if you take kWh divided by cost.”
In part two - streamlining production...
About: Drew Robb is a graduate of the University of Strathclyde in Glasgow. Currently living in Los Angeles, he is a freelance writer focusing on engineering and technology.