Silicon - a question of purity

"It already seemed likely that solar cells might perform very well even when made out of somewhat less pure silicon," says Ton Veltkamp of ECN Solar Energy. "But our results have surprised everybody."

Solar power is a very rapidly growing industry, increasing in size by more than 50 percent in some years. And that has been going on for more than a decade. Note also that electronic equipment contains grams of silicon, while there are kilo's of it inside the solar panels on the rooftops of ordinary homes. One megawatt of peak power - comparable to the capacity of a small wind turbine - requires about ten tonnes of silicon.
"Before 2003 the PV industry bought cheap waste material, crystalline silicon which wasn't quite pure enough to be turned into chips for computers and other electronic devices," says Veltkamp. But since about 2005, PV demand for silicon has been crowding out the needs of the semiconductor industry. Prices on the spot markets used to be fairly stable around $30 a kilo. When the supply failed to keep up, they occasionally exceeded 500 dollars!

The silicon for this ingot was purified by distillation in the gas phase. An ECN study proves that much less expensive material can be used, without hurting the solar cell's efficiency (Photo: Warut Roonguthai).

A less expensive feedstock
Since then, much work has been done to develop cheaper 'solar grade' material. Veltkamp: "A still common method for the purification of silicon is distillation in the gas phase. That takes a very large initial investment, and the process is extremely energy-intensive as well. The justification was mainly based on older studies, which showed decreasing solar cell efficiency when the concentration of impurities like iron and chrome increased beyond 0.05 ppm." (0.05 parts per million.)

To measure is to know
ECN's researchers weren't the only ones who doubted whether modern solar cells actually require such high purity levels. The manufacturing processes which turn silicon wafers into solar cells have been changed radically in the past ten years. "But," says Veltkamp, "the research needed to prove the validity of new figures requires the right facilities and plenty of time. Our lab is very suitable for it. CrystalClear gave us the opportunity. My colleague Gianluca Coletti coordinated the feedstock part of that European project."
ECN Solar Energy provided the specifications used by SINTEF in Norway to carefully prepare a series of crystalline ingots based on extremely pure silicon, each contaminated with a precise amount of a single impurity; for instance iron, chrome or nickel. They were shipped to Petten and turned into solar cells. Veltkamp: "We used our standard process, which closely resembles the way in which the most common commercial solar cells are manufactured. That makes our results valid for most industrial processes. Which explains why our graph gets so much attention. It shows that most of the ingot is perfectly usable, even when an impurity in the feedstock reaches 10 ppm - up to 200 times the usual limit!"  

Making money and energy
There is no need for highly pure silicon (electronic grade) in order to make efficient solar cells. Upgraded metallurgical grade silicon (UMG) will do and is much less expensive in terms of both money and energy. How much less, exactly? The cost of a finished solar panel can be split into five categories:

The 14 percent right at the beginning can be significantly reduced. Manufacturers believe the feedstock can become half as expensive, but Veltkamp hedges his bet: "They'll first have to prove that." Reducing the purity requirements may also decrease the cost of the wafer, the slice of silicon which finally becomes a solar cell. Currently, the raw material is crystallized by growing an ingot, which is sliced into wafers using a saw. About half the silicon is lost in the slicing process. Veltkamp: "We'll soon be able  to continuously pour thin layers of liquid silicon on a suitable substrate. That wastes very little material, because there are no sawing losses."
Modest demands for purity cause a greater variety of processes to be good enough for the production of solar grade feedstock, processes which can turn out to be less expensive for a variety of reasons. "Consider the effect of increased competition," says Veltkamp. "Accepting UMG silicon brings more players on the market."
Finally, it isn't just about money. The time it takes for a solar panel to pay back the energy which was invested in its production also matters a great deal. "With current technology, that takes about three years. Soon one year will be enough. And solar panels last a long time, require no maintenance, make no noise... It's a wonderful source of energy."

Contact
Ton Veltkamp
ECN Solar Energy
Tel. 022 456 4251
E-mail: Ton Veltkamp

Additional information
Impact of iron, nickel and chromium in feedstock on multicrystalline silicon solar cell properties
Impact of transition metals in feedstock on multicrystalline silicon solar cell properties
The carbon footprint of PECVD chamber cleaning using fluorinated gases
An overview of MWT cells and evolution to the ASPIRE concept: a new integrated mc-Si cell and module design for high-efficiencies

Effects of impurities on efficiency
The graph shows how four different impurities affect the efficiency of solar cells made from wafers sliced from specific parts of a silicon ingot. The black dots mark the performance when commercial pure silicon is used. When the feedstock has been contaminated with iron, chrome or nickel at 40 to 50 ppm (parts per million), the part of the ingot between 40 and 80 percent (measured from the bottom upwards) still yields solar cells which perform as well as the reference. Below and above that section, efficiency suffers, "but the data allow us to conclude that most of the ingot is perfectly usable when the impurity concentrations are limited to about 10 ppm," says Veltkamp.

Oxygen
ECN Solar Energy is also studying oxygen as a contaminant. More than 10 to 12 ppm is currently thought to be harmful, but recent experiments show that oxygen does not affect solar cell efficiency at concentrations up to 20 ppm. And a factor two is also quite significant.

Text: Steven Bolt

This ECN Newsletter article may be reproduced without permission, provided that www.ecn.nl/nl/nieuws/newsletter-en/ is acknowledged as the source of the material.

The effect of four different impurities on the efficiency of solar cells made from wafers sliced from specific parts of a silicon ingot. The black dots mark the performance when commercial pure silicon is used.