Polysilicon: Supply, Demand and Implications for the PV Industry
Polysilicon 2008 is the latest in our bottom-up analysis of the polysilicon industry with forecasts updated through 2012. The results of our research reveal the product of much effort over the last few years by manufacturers, both old and new, in the polysilicon supply chain. 2008 will finally begin to see the substantial additions to virgin Siemens polysilicon p lants that were originally planned beginning in 2005, the year that the polysilicon shortage began to be widely recognized.
After 24 to 30 months of construction, many of these plants are nearing completion and will start ramping up their output in the second half of this year. Given the long lead time for the plants, the forecasts through 2009 are still quite solid and track our 2007 forecasts, but our survey and report have shown that substantial new capacities will come on line by 2010 at plants recently funded and commenced are completed by then. A summary of the results is outlined below.
Majors Expand Capacity
The big seven traditional polysilicon producers continue to expand their capacity at varying rates. Most of them plan to complete new plants or expansions in 2008 including REC, MEMC, Wacker, and Hemlock. Many of these will begin shipments in 2008, but the bulk of the polysilicon from these expansions will come up to full scale by early next year. Further expansions through 2012 will see the cumulative output of these companies grow by a minimum of three-fold from 2006 levels. Many of them have left the door open to further expansions based on customer demand.
The expansions have proceeded generally smoothly, but some have encountered minor difficulties. MEMC's expansion in Pasedena, Texas was beset with technical issues that required additional maintenance, while REC's new fluidized bed reactor plant was somewhat delayed by longer than expected building and higher costs. REC now expects the plant to be on-line next year. (NB: our 2007 forecast had anticipated this delay.)
New Siemens Producers Grow
Though faced with the difficult task of completing large-scale plants, some new producers of polysilicon that use traditional Siemens technology are close to completing their initial plants. Two, in particular, have shown tremendous ability in managing the complex process, one we have been following for some time and another new to our list. The first, DC Chemical in Korea, should have its 5,000 ton plant online and ramped this year. DC is already planning a second 10,000 ton plant for 2009-2010. LDK of China has come on very quickly with a similar phase expansion, and photos are available on the Web to witness their amazing progress.
Many more are progressing as well. China has dozens of stated plants, though most will likely not achieve the necessary scale and speed to be competitive. In fact, Trina Solar in China has already backed out of its plans to build a plant. Nitol in Russia has an initial 3,600 ton plant, but is having difficulty raising capital for further expansions, notwithstanding a recent minority investment by Suntech of China. Finally, a few others are in the advanced planning or building stage including M. Setek of Japan, Isofoton's consortium in Spain, and Hoku in the U.S. Dozens of others are claiming progress, but we predict few of them will ultimately be successful, both technically and financially.
Met-Grade Solutions Bloom
A real boost to the supply of polysilicon came from new producers using upgraded metallurgical feedstock. Three met-grade poly producers, Elkem in Norway, Dow Corning in Brazil, and recent rising star Timminco of Canada all have meaningful additions planned in coming months. While initial results from users indicate that these materials will be used in blends with virgin polysilicon, some are planning to quickly move to pure use of the feedstock. Timminco has had some surprisingly strong results recently, even though impurity levels are reported at the high end of the allowable range.
The great advantage of these companies will be the low amount of capital and shorter lead times to add meaningful capacity. Other new entrants like AE Polysilicon's FBR plant in the U.S. may not share those advantages, but many of these entrants will significantly add to global polysilicon capacity and production in the next few years.
2008 Guidance Remains Firm
Aggregating the results from our survey, we updated our forecast to include all planned production from existing producers. We also added in a percentage of the forecast production from the emerging producers using both traditional technologies and new technologies. The percentage used in 2008 was the same as in prior forecasts (60 percent), but was reduced in subsequent years (50 percent in 2009 and then 40 percent beyond). These de- rate factors should more closely approximate actual results as the difficulty financing and ramping new plants in a period of rising plant construction costs and looming new supply continues to rise. Finally, our methodology adjusts for polysilicon use in and returned from the semiconductor industry and some inventory effects to account for existing non-standard inventories and channel inventory throughout the supply chain.
The result is a growth in polysilicon-based cell, and module production will grow by 60 to 70 percent in each of the next couple of years; total module supply will grow by over 80 percent in each of those years when thin film is added in. This reinforces our belief that supply will grow faster than demand in each of those years and should result in a substantial price decline for modules through 2010. The precise degree and timing of that trend is subject to the remaining uncertainty surrounding the renewal of both the U.S. and Spanish national support programs for PV.
The major plants in 2008 that matter for our forecast include Hemlock, Wacker, and MEMC, as well as the plants being built by DC Chemical and Elkem among half a dozen other plants that are near completion. Using our conservative methodology and de-rating factors for new plants, we are highly confident in our forecast numbers for 2008 and 2009.
2010 and Beyond
It remains very difficult to gauge supply beyond 2010 due to the simple fact that the decisions to build new capacity for that time frame would not yet need to be made. We believe that the combination of additional supply in our forecast, the introduction of new technologies that can be scaled up more quickly, and the twin competing technologies of thin-film PV and large-scale concentrating solar will limit the growth of demand for Siemens-based polysilicon plants in that time frame. Only yet-to-be-revealed economics will determine the degree to which that forecast is accurate.