A rapid and ongoing expansion of polysilicon production capacity will likely generate an oversupply for the next several years, driving polysilicon manufacturers to further innovate processes and reduce operating costs in preparation for an impending wave of new demand. Drastic price drops experienced in 2009 are expected to continue at a slower pace in 2010, yet polysilicon producers remain poised to supply high-purity product to a market increasingly dominated by solar photovoltaic (PV) cells.
By all estimates, demand for silicon for the solar-PV market has shot upward over the last decade. In 2000, demand for silicon for solar energy use was around 3,800 metric tons (m.t.), a total that grew to 42,000 m.t. in 2008. Forecasts put the number well above 100,000 m.t. by 2014. Analysts expect that while the semiconductor industry should grow at around 5–9%/yr over the next several years, growth in the solar market could exceed 40%/yr for multiple years, although the exact trajectory is uncertain.
For the time being, though, recent production-capacity increases appear to have overshot demand. Industry analyst Richard Winegarner of Sage Concepts Inc. (Healdsburg, Calif.; http://www.sageconceptsonline.com) is among those who think the capacity expansion, combined with the entry of new companies into the silicon production industry to serve the growing solar-PV market, will give rise to an over-capacity for silicon at least for the next several years.
Silicon expansion projects
The capacity increases include major players in the polysilicon production area, along with some smaller companies. Among the announcements was a 2009 expansion by leading silicon producer Hemlock Semiconductor Corp. (Hemlock, Mich.; http://www.hscpoly.com). The company began operations of a $1-billion, two-phase expansion at its Hemlock, Mich. headquarters that raised its capacity to around 36,000 m.t./yr Meanwhile, Hemlock is constructing a new polysilicon facility in Tennessee. Hemlock sales and marketing vice president Jim Stutelberg reports that construction is progressing on schedule for production to begin in 2012. Paralleling Hemlock’s activities has been German specialty chemicals firm Wacker Chemie AG (Munich, Germany; http://www.wacker.com). Wacker also plans to open a polysilicon production site in Tennessee in a few years. Last year, Norwegian company Renewable Energy Corp. ASA (REC; Sandvika, Norway; http://www.recgroup) began operating a newly expanded polysilicon plant in Moses Lake, Wash., and Chinese solar wafer company LDK Solar (Xinyu City, China, http://www.ldksolar.com) began operating a 15,000 m.t./yr polysilicon plant in Xinyu.
Along with growing demand, tax credits have helped spur expansion of production capacity. In January, the U.S. government handed out $1.0 billion in tax credits to the solar industry for job creation investments. Hemlock and its majority owner, Dow Corning Corp. (Midland, Mich.; http://www.dowcorning.com), received a total of $169 million in tax credits for expansion projects, while Germany’s Wacker received $128.4 million for its new production facility in Tennessee. The U.S. arm of REC received tax credits worth $155 million for its recent expansion project at Moses Lake. Meanwhile, engineering companies such as Fluor Corp. (Irving, Tex., http://www.fluor.com) are having success constructing polysilicon sites. Fluor was involved with the construction of the facilities for LDK Solar, REC Group and others.
Price drops, but a sunny future
The fast-changing solar energy industry and the equally fast production ramp-up has impacted prices significantly. Prices for purified silicon rose sharply in 2007 and 2008 due to a shortage of the material — in 2008, prices for polysilicon peaked above $400/kg. Prices plummeted throughout 2009, however, falling to about $50–55/kg on average by year’s end. Prices are expected to continue to drop over the next three years, although not as precipitously as in 2009. By 2012, some industry watchers think prices could hover around $40/kg.
Although the expansion in production capacity likely means a silicon oversupply for the next few years, the situation may change after that if solar-PV succeeds in establishing itself as an economically viable alternative to conventionally derived energy. Hemlock is among those taking the long-view regarding the capacity increases. Hemlock’s Stutelberg views the growing capacity as essential to move solar energy toward “grid parity.” Depending on a host of factors, grid parity for solar energy could be reached at electricity production costs of around $0.15/kWh.
There appears little doubt that polysilicon production will be dominated by the solar PV market in years to come. While the need for high-purity silicon in its traditional market — the semiconductor industry — remains, most silicon produced today is destined for the solar photovoltaic (PV) market. By 2008, the size of the market for silicon for the solar energy industry had overtaken that of the electronics industry. Industry analyst Winegarner estimates that, in 2010, 70% of the purified polysilicon produced will enter the solar market, versus 30% for semiconductors. The balance is likely to move to 90:10% in favor of solar in the next few years, he says.
Silicon production processes
To meet the needs of a solar-dominated future, high-purity silicon companies are exploring process improvements mainly for two chemical vapor deposition (CVD) approaches — an established production approach known as the Siemens process, and a manufacturing scheme based on fluidized bed (FB) reactors. It appears likely that improved versions of the two types of processes will be the workhorses of the polysilicon production industry for the near future.
Siemens process — The Siemens reactor was developed in the late 1950s and has been the dominant production route historically. In 2009, about 80% of the total polysilicon manufactured was made through a Siemens-type process. The Siemens approach involves deposition of silicon from a mixture of purified trichlorosilane or silane gas, plus excess hydrogen, onto hairpin-shaped filaments of high-purity polysilicon crystals. Silicon growth occurs inside an insulated “bell jar,” which contains the gases. The filaments, which are assembled as electric circuits in series, are heated to the vapor deposition temperature by an external direct current. As the gases enters the bell jar, the high temperature (1,100–1,175 ºC) on the surface of the silicon seed filaments, with the help of the hydrogen, causes trichlorosilane to reduce to elemental silicon and deposit as a thin-layer film onto the hot seed filaments. HCl is formed as a by-product.
Temperature control is critical to the process — the temperature of the gas and filaments must be high enough for silicon from the gas to deposit onto the solid surface of the filament, but the temperature cannot be so high that the filament starts to melt. Further, the deposition rate must not be too rapid, or the silicon will not deposit in a uniform, polycrystalline manner, rendering the material of little use for semiconductor and solar applications.
Hemlock Semiconductor has a highly proprietary Siemens-type process that is capable of producing silicon of “11-nines” purity for the semiconductor industry and “9-nines” purity for the solar-PV market (see sidebar, below). Hemlock is working hard to develop “a portfolio of innovations” on the process, Stutelberg says.
Fluidized bed process— Several companies are developing polysilicon production processes based on fluidized bed (FB) reactors. The FB approach to polysilicon production has its origins in a 1980s-era program sponsored by the U.S. Department of Energy whose goal was to devise less energy-intensive methods for making silicon. FB approaches to polysilicon production offer the ability for continuous production, as opposed to the batch production of the Siemens route. In addition, FB polysilicon reactors consume less energy. REC Solar, for example, says their fluidized-bed polysilicon process consumes only around 10% of the energy required to run a Siemens-type process.
In an FB process, tetrahydrosilane or trichlorosilane and hydrogen gases are continuously introduced to the bottom of the reactor at moderately elevated temperatures and pressures. High-purity silicon particles are inserted from the top and are suspended by the upward flow of gases. When the reactor operates at high temperatures (750 ºC), the silane gas reduces to elemental silicon, which deposits on the surface of the seed particles. As the seed crystals grow, they fall to the bottom of the reactor, where they are removed continuously. To compensate for the removal of silicon granules, fresh seed crystals are injected into the top of the reactor.
MEMC Electronic Materials (St. Peters, Mo.; http://www.memc.com), a silicon wafer manufacturer, has been producing granular silicon from silane feedstock using a fluidized bed approach for over a decade. Several new facilities will also feature variations of the FB. Wacker Chemie is expected to announce the operation of a fluidized bed reactor facility using trichlorosilane as the working fluid in mid-2010. The plant, located in Burghausen, Germany, is designed specifically to make solar-PV material. Several major players in the polysilicon space, including Wacker and Hemlock, are developing FB processes, while at the same time continuing to produce silicon using the Siemens process as well.
New feedstock for FB process
A development-stage company called Peak Sun Silicon Corp. (Albany, Ore., http://www.peaksunsilicon.com) is working on an FB process using tribromosilane (TBS) as the feedstock, rather than either silane, or trichlorosilane. While TBS costs more than its bretheren, Peak Sun’s process uses less, so the feedstock costs are even. The inclusion of massive bromine atoms in the feedstock compound offers several key advantages.
“Using tribromosilane in the process allows different operating parameters,” says Scott Schumacher, vice president of sales and marketing at Peak Sun. “We have a much wider tolerance of temperatures and pressures” than the tight control that must be exerted in other processes, says Schumacher. TBS processes also operate at lower temperatures, which enables energy savings.
The TBS-FB approach also helps avoid formation of submicron-diameter amorphous silicon dust through a homogeneous nucleation route, a significant challenge encountered by silicon reactor operations that use silane or trichlorosilane as the feed gas.
Homogeneous nucleation can occur when materials move from vapor to solid phase and reach a “critical nucleus” of solid-phase material. At certain values of “ n ” atoms, the thermodynamics of the system favors self-nucleation, rather than deposition on the surface of a seed particle. The phase-change energy barrier is overcome more readily at higher values of n, since surface area-to-volume ratio decreases as particle diameter increases. Silane and trichlorosilane have a natural propensity for self-nucleation when overcoming the phase-change energy barrier. By virtue of its relatively large molecular size, TBS requires a greater n value to achieve critical nucleus, and molecules perfer to transition to solid phase by nucleating on an existing silicon surface. The result is that TBS deposition favors growth on the surface of crystalline seed particles over formation of low-value amorphous silicon dust.
When completed, the Peak Sun project will be the first demonstration of a FB silicon unit using TBS. Peak Sun claims that its process will have lower capital costs and 25% lower operating costs than a Siemens process unit.
Other advantages to the Peak Sun TBS-FB process include the formation of dense metallic beads with a narrow particle size distribution. Silicon beads formed by the TBS-FB process tend to have less surface oxidation and less gas molecule inclusions, which gives the silicon better melt properties.
Another key process in the manufacture of high-purity silicon involves a recrystallization step that converts polycrystalline material to monocrystalline silicon. The process is more important at present for semiconductor-grade silicon, since polycrystalline silicon is suitable for PV cells.
Manufacturers typically use some variation of the so-called Czochralski process, in which a seed crystal is introduced to a silicon melt and slowly withdrawn to generate a long mass of monocrystalline silicon.
An alternative is a float-zone process, in which impurities are segregated during a transition that occurs as a mass of polysilicon passes through a radio-frequency heating coil, which creates a localized molten zone from which the pure crystal grows.
UMG may be on the outs
One technology approach to polysilicon production that appears to be waning is upgraded metallurgical grade (UMG) silicon. UMG is produced by melting metallurgical grade silicon and slowly and directionally recrystallizing it. The approach would offer a less expensive route to material, but the 5-nines or 6-nines purity UMG can achieve wouldn’t be viable in an environment where higher-purity methods are cost-competitive. For example, metals firm Timminco Ltd. (Toronto, Canada, http://www.timminco.com) reportedly suspended its UMG-silicon operations in March as a result of low prices and lack of demand for UMG material for solar cells.
Stutelberg explains that the solar industry as a whole seems to be gravitating toward higher purity silicon. “Next-generation solar components may demand higher levels of purity” than UMG can offer, he says.
The majority owner of Hemlock Semiconductor Group is Dow Corning (http://www.dowcorning.com), not Dow Chemical Co. as originally reported.
GEA Messo PT’s (Hertogenbosch, the Netherlands; www.gea-messo-pt.com) proprietary crystallization technology was chosen by SK Chemicals (Gyeonggi-do, Korea; www.skchemicals.com) for its…
Improved understanding of crystallizer equipment and operation can aid purification efforts in the CPI Crystallization is a key purification technique…
Syrris In March, this firm signed an exclusive agreement with Prosonix. By combining Prosonix’s ultrasound technology with the Atlas jacketed…
Crystallization is a key purification technique for various sectors of the chemical process industries (CPI). Several approaches for industrial crystallization…
A team from the School of Chemical Sciences and Engineering, at the University of New South Wales (Sydney, Australia; www.unsw.edu.au),…
New filtration technology for highly corrosive media
PTA production: Lowering OPEX without compromising on quality
Sure that zero means zero in your zero-liquid discharge (ZLD) process?
How separation processes profit from Industrial Internet of Things (IIoT) solutions