Sep 01, 2009 Predictions of long-term growth in the photovoltaics market are not just good news for triple-junction solar cell makers. SiC diode and transistor manufacturers are also set to benefit, because these chips can drive up the efficiencies of converters that transform the DC output from cells into the AC form needed for the grid. Richard Stevenson reports.

 

When power electronics manufacturers launched their SiC Schottky barrier diodes onto the market at the beginning of this decade they had one application in mind – switch mode power supplies (SMPS). Existing supplies converted AC mains into DC outputs for driving computer hardware employed silicon diodes, and higher efficiencies could be realized by switching to SiC equivalents.

 

These gains in efficiency stem from the very low switching losses associated with SiC. The reverse recovery charges and stored charges are incredibly small, and the forward current and the frequency of the switching speed has very little impact on switching behavior. In short, SiC diodes possess an ideal set of characteristics. 

 
Sales of SiC diodes to builders of SMPS continue to grow. However, SiC power electronic device manufacturers are not complacent - they are looking to new markets to generate additional revenue growth. One opportunity is in the hybrid electric vehicle, which currently employs silicon chips to convert the DC output from the battery into an AC form that is suitable for driving the car. SiC electronics can operate at higher temperatures, eliminating the need for an independent, space-consuming cooling system for the electronics that takes up valuable space. Freeing-up space is a benefit that is highly valued by carmakers, but qualification times of eighteen months to two years mean that it will a while time before SiC can make an impact on the automobile sector.



 
 
Where SiC diodes are already making inroads is in solar inverters, which convert the DC output from solar panels into an AC form that is suitable for connection to the grid. These inverters operate at powers ranging from one kilowatt up to 500 kW, a span that reflects the variations in solar installations, which go from a handful of panels on the roof of a house to massive arrays on a solar farm covering many hectares.

 


 
Like SMPS, the first solar inverters employed silicon devices, and again a switch to SiC products could increase efficiency. Although SiC devices are more expensive than their silicon rivals, Brice Le Gouic, a market and technology analyst at Yole Développement that specializes in power electronics, believes that producers of inverters are willing to pay the additional cost. “The return on investment is the first word that people are talking about,” says Le Gouic, and higher costs are acceptable, so long as they are outweighed by increased efficiency. Improving a solar inverter’s efficiency by just 1 percent in a 30 kW system can yield an additional 3000 kWh over ten years, equating to a cost saving of hundreds of dollars.

 


 
Le Gouic says that the SiC market for solar is still in its infancy. It will be worth $3 million next year, but will grow to tens of millions dollars over the next decade. However, sales to SMPS builders will continue to dominate this SiC market. They will be worth $55 million in 2010, and increase at a compound annual grow rate of 60 percent between then and 2015. But the rocketing growth of shipments of SiC products to solar manufacturers will gnaw away at this sector, with sales to SMPS makers declining from 88 percent of the market in 2010 to 55 percent in 2015.

 


 
Emerging markets, such as solar inverters and HEVs, also offer new product opportunities for SiC. “SMPS manufacturers are mainly interested in SiC diodes,” says Le Gouic “but all the other markets will be driven by the inverter application and have two SiC devices in each of the systems - transistors and diodes.” Single-phase inverters employ four transistors and four diodes, while three-phase variants use six transistors and six diodes.





 
Inverters take the DC input from solar cells, possibly boost it, and then chop it at very high frequencies, before the signal is filtered to produce a rectified sine wave. The higher the frequency, the cleaner the output. However, higher frequencies mean higher switching losses, primarily through higher switching losses in the power transistors, which are far lower for SiC devices than their silicon equivalents.





 
Another major benefit of switching to SiC transistors is that it can lead to a reduction in the inverter’s bill of materials. Today’s silicon inverters typically operate at 16-18 kHz and include large inductors and capacitors. Replacing the silicon power electronics with SiC chips allows a transition to even higher operating frequencies, without sacrificing efficiency, and also cuts the costs of the inverter thanks to smaller inductors and capacitors that can be used at these higher frequencies.



 
 
SiC transistors can also operate at higher temperatures, cutting the cooling requirements for the power electronics in the inverter. Opportunities for savings are greatest in the high-power, water-cooled inverters used in solar farms, and far smaller in the air-cooled versions used for small roof-top installations.



 
 
A handful of SiC device makers have been developing various forms of SiC transistor over the last few years, and these products are winning the attraction of solar inverter manufacturers. For example, ISET (Institut für Solare Energieversorgungstechnik, Kassel, Germany), has built a unit with a SiC MOSFET that has an efficiency of 99.08 percent, according to Le Gouic.



 
 
Developers of SiC transistors include Infineon Technologies, which is working on a JFET that will complement its well-established range of SiC diodes. Interest in these established products continues to grow, thanks to the emergence of voluntary energy efficiency standards for power supplies, according to Jan-Willem Reynaerts, who combines the role of Business Segment Manager for high-voltage metal-oxide semiconductors with worldwide responsibility for SiC Schottky diodes. “On top of that, the solar industry is also extremely interested in SiC, whether it is high voltage diodes or switches.” He is also seeing interest in SiC diodes for HEV, and in lighting applications.

 


 
Infineon launched the world’s first SiC diode to market in 2001, and it came out with a second-generation product in 2005. “The biggest change from generation one to generation two was the added surge current capability, “explains Reynaerts. This addressed the thermal runaway issue that plagued the initial product - a cycle of rising temperatures leading to an increase in resistance, a higher forward voltage, and greater power dissipation that drove up the device’s temperature once more. 

 


 
Early this year, Infineon launched a third generation of SiC diodes to address three issues: the customer’s desire for a less expensive product; better switching performance at higher frequencies; and improved efficiency when the inverter operates at a relative light load, such as 20 percent of its maximum. “The third generation is an answer to all those points,” says Reynaerts.



 
 
Development focused on improvements to device performance at high current densities. This held the key to cutting chip sizes, which would reduce capacitance and improve high-frequency performance, while simultaneously lowering production costs. “What we did was find a way to get the heat out of the package,” explains Reynaerts. Engineers developed a “diffusion soldering” process that almost eliminates the solder layer between the chip and the lead frame, and produces a very good thermal connection for cooling. 

 


 
The new approach gives third generation diodes about a 20 percent cost advantage over second-generation equivalents, if products are ordered in similar volumes. “We are not talking small beer,” says Reynaerts. “This is really important for our customers. We are the leading supplier in power semiconductors, and it is for us to lead in developing the SiC market.”

 


 
These SiC products are still much more expensive than their silicon equivalents, but the gap is closing fast, partly due to reduction in the cost of 4 inch substrates at higher volumes. “There used to be a time when SiC diodes were nearly an order of magnitude more expensive, but this is coming down to roughly a factor of three, product for product,” says Reynaerts. Closing this gap is encouraging the SMPS and inverter manufacturers to think seriously about SiC. “This message is starting to sink in that at the system cost level, in a lot of applications where energy efficiency is important, the added cost of SiC over silicon is more than paid back.”

 


 
Infineon faces competition from a handful of other SiC diode manufacturers, including SemiSouth Laboratories, which released its product to the market in 2008. Jeffrey Casady, the company’s chief technical officer and vice-president of business development, says that its diode is similar to Infineon’s. However, the company’s primary focus is on the promotion of its JFET. “We made the first normally-off JFET in the summer of 2007, we started ramping in 2008, and now in 2009 we have got reliability and qualification data that looks really good,” says Casady.

 


 
SemiSouth’s normally-offJFETs have recently been deployed in inverters for photovoltaic systems that were built by Fraunhofer Institute for Solar Energy Systems in Freiberg, Germany. A conversion efficiency of 99.03 percent was announced this July, with the German team stating that the SiC JFETs were the key to behind their latest efficiency improvements. More recently, they have shown that a SiC-based inverter operating at 48 kHz with SiC products can outperform a silicon-based equivalent running at 16 kHz. This means that it is possible for SiC-based inverters to not only deliver higher efficiencies than their silicon-based rivals, but also realize this while using smaller, lower-cost inductors and capacitors. 

 


 
When SemiSouth began its SiC transistor development, it could have decided to invest in a MOSFET, JFET, or a bipolar device. “Every devices has its pros and cons,” says Casady, but SemiSouth selected the JFET because of its ruggedness, low cost, and avoidance of an oxide that can lead to reliability issues. “It behaves in many ways like a MOSFET would, with some differences in the drive. It has a very familiar feel to circuit designers.” 

 


 
Casady says that the JFET is now in low-volume production – SemiSouth produces hundreds of these wafers every month. Production is currently performed on 3-inch material, but the company plans to convert to 4-inch in the next year or two, when the higher volumes justify this move. By then the current blip in the solar market will be a thing of the past, and Semisouth, along with other SiC transistor and diode makers, will be battling it out for substantial contracts with the builders of solar inverters.

    Email this article to a friend   Connotea       Cite-U-Like        Del.icio.us        Digg       Facebook

  Google       Live Spaces       MySpace       Newsvine       Reddit    

  Stumbleupon       Technorati       Twitter       Yahoo! My Web