ARTICLE IN PRESS

Solar Energy Materials & Solar Cells ] (]]]]) ]]]–]]] www.elsevier.com/locate/solmat

New processes for the production of solar-grade polycrystalline silicon: A review A.F.B. Braga, S.P. Moreira, P.R. Zampieri, J.M.G. Bacchin, P.R. Mei Faculdade de Engenharia Mecaˆnica, Departamento de Materiais, UNICAMP—Universidade Estadual de Campinas, C.P. 6122, CEP 13083-970, Campinas, SP, Brazil Received 12 June 2007; received in revised form 27 September 2007; accepted 15 October 2007

Abstract The global energy consumption is predicted to grow dramatically every year. Higher energy prices and public awareness for the global warming problem have opened up the market for solar cells. The generation of electricity with solar cells is considered to be one of the key technologies of the new century. The impressive growth is mainly based on solar cells made from polycrystalline silicon. This paper reviews the recent advances in chemical and metallurgical routes for photovoltaic (PV) silicon production. r 2007 Elsevier B.V. All rights reserved. Keywords: Silicon purification, solar-grade silicon

1. Introduction The photovoltaic (PV) industry was limited to aerospace applications up to the early 1970s, at the time of the first oil crisis, when a more in-depth investigation began for terrestrial applications [1]. One of the alternatives proposed was the development of low-cost polycrystalline silicon cells. However, the advance of this technology was inhibited by the low efficiency of conversion of polycrystalline silicon.

  

In 1980, the efficiency of conversion in 100 cm2 cells was of the order of 8%. The year 1984 saw a 4% increment, with the conversion efficiency rising to 12%. In 1985, this efficiency had risen to 13% in laboratory cells (2 cm2) [2].

Although this advance was not enough to drive investments in this type of cell, it encouraged some groups to remain in this area. During this period, the greatest interest focused on monocrystalline silicon, amorphous silicon and other semiconductor materials. It was only in Corresponding author. Tel.: +55 19 3521 3334; fax: +55 19 3289 3722.

E-mail address: [email protected] (S.P. Moreira).

1990, with the announcement of the results of a laboratoryscale cell with a conversion efficiency of 35% (in areas of 5 mm2), that the polycrystalline silicon cell manufacturing technology became really interesting. This advance led to renewed investments in research to produce low-cost polycrystalline silicon [3]. The consequences of these research efforts are illustrated in Fig. 1, which shows that, up to 1996, the market was dominated by the production of monocrystalline silicon panels. The advances in polycrystalline silicon cell technology resulted in an inversion in the tendency of the curve in 1997, led, for example, by the 1996 publication presenting a panel with 15% conversion efficiency [4]. Basically, until 1997, the silicon employed in the production of polycrystalline solar cells originated mostly from waste produced by the microelectronics industry. Considering the magnitude-scale differences, differences in silicon specification requirements for application in microelectronics and in the area of PV, and the costs involved in the process, special interest has focused on the search for more economic routes for the production of PV silicon. Moreover, in the last 5 years, largely due to the rising price of crude oil and to growing awareness of the need to protect the environment, major investments began to be made in this technology [6].

0927-0248/$ - see front matter r 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.solmat.2007.10.003 Please cite this article as: A.F.B. Braga, et al., Sol. Energy Mater. Sol. Cells (2007), doi:10.1016/j.solmat.2007.10.003

ARTICLE IN PRESS 2

A.F.B. Braga et al. / Solar Energy Materials & Solar Cells ] (]]]]) ]]]–]]]

Russia, Norway and the USA. Brazil’s production of metallurgical silicon in 2005 was 230,000 metric tons. Its production volume in 2004 was the same, and in this year it exported 200 thousand tons of metallic silicon at approximately US $1.19 kg1 [12]. Considering the current picture, this paper reviews the recent advances in the chemical and metallurgical routes for the production of PV silicon. 2. New routes for solar-grade polycrystalline silicon production

Fig. 1. Production of cells by type [5].

Metallurgical-grade silicon is obtained from the reduction of silicon in the presence of carbon [13]: SiO2 þ 2C ! Si þ 2CO

In 2004, the PV market showed a 62% growth over 2003. The offer of PV energy in 2004 was 927 MW. Of the total installed PV industry, the two largest markets are Germany and Japan, which together account for 69% of the world market. The consolidated production of solar cells increased to 1146 MW in 2004, with the Japanese contribution representing 48%. The lack of silicon substrate to supply this market is well known, and this lack presages restricted growth in coming years [7]. The demand for solar-grade silicon has grown rapidly and an average growth of 30% per year is estimated for the next 10 years [8]. Table 1 presents the worldwide polycrystalline silicon production capacity and the projected growth in offer for 2010 [9]. Table 2 shows the relation between production and demand for polycrystalline silicon between 2003 and 2010 [10]. These estimates reinforce the concern of the international market to meet the demand for this

Demand for silicon metal comes primarily from the aluminum and chemical industries. The quantity of silicon metal that is refined into semiconductor-grade metal is less than 5% of total silicon metal demand [14]. In Brazil, the process for obtaining metallurgical silicon uses not only high-quality quartz but also charcoal as a reducing agent. This is reflected in the quality of Brazilian metallurgical silicon, which, under well-controlled processing conditions and raw material, can reach a purity of up to 99.88% [15]. 2.1. Obtaining polycrystalline solar-grade silicon: chemical versus metallurgical routes Currently, the process for obtaining polycrystalline solar-grade silicon is divided into two categories. The first, called the chemical route, is related to the purification of silicon by means of the Siemens process, consisting of decomposing trichlorosilane by CVD on inverse U-shape hot filament [16]:

SiðsÞ þ 3HClðgÞ ) HSiCl3 ðgÞ 2HSiCl3 ðgÞ þ H2 ðgÞ ) Si þ SiCl4 þ 2HCl

important technology over the next four decades. Much effort has focused on expanding silicon production plants, as well as on investing in new low-cost routes for the production of polycrystalline silicon. For Brazil, the mastery of a PV silicon production route may represent an important advance in the export segment of products with high added value. Moreover, several factors render the mastery of a technology of this type highly attractive. 1. Brazil has the largest worldwide reserve of quartz (the raw material for producing metallurgical silicon). 2. Brazil is the world’s fifth largest manufacturer of metallurgical silicon [11], preceded only by China,

) ðHot filamentFCVD 1000  CÞ

The alternative route, known as the metallurgical route, involves obtaining solar-grade silicon directly from metallurgical silicon. This route for production can be five times more energy efficient than the conventional Siemens process that uses more than 200 kWh/kg [8]. Researches involving the chemical route (associated with the Siemens process) are more advanced and are already operating on a pilot scale. The so-called metallurgical route, which proposes the purification of metallurgical silicon without the stages that involve the formation of chlorosilanes, is still in the research phase. However, Elkem of Norway developed a process for polycrystalline solar-grade silicon production and is building a 5000 metric tons plant [9].

Please cite this article as: A.F.B. Braga, et al., Sol. Energy Mater. Sol. Cells (2007), doi:10.1016/j.solmat.2007.10.003

ARTICLE IN PRESS A.F.B. Braga et al. / Solar Energy Materials & Solar Cells ] (]]]]) ]]]–]]] Table 1 Worldwide PV silicon production capacity from 2004 to 2010 (in metric tons) [9] Company Hemlock MEMC Mitsubishi Mat Wacker Tokuyama REC DeGussa/ SolarWorld Sumitomo China Endesa/Isofoton Elkem Korea Other Total (tons)

2004 7000 2550 2200 5200 4800 5300

2005

2007

2008

2009

2010

7500 10,000 10,500 14,500 19,000 27,000 3700 4400 6500 8000 8400 8820 2800 2800 2800 3200 3360 3528 5200 6500 7500 9000 13,500 14,175 5200 5400 5940 6500 7500 8400 5300 5500 5500 8125 13,000 13,650 100 200 850 850 700

1300

2006

1300

700 400

1400

750 800

800 1600

1500

1250 1000 2000

840 882 2700 3800 2500 2625 5000 10,000 2500 3000 3000 4250

28,350 31,700 37,100 41,390 56,175 82,150 100,980

Table 2 Analysis of silicon production and demand from 2003 to 2010 (in metric tons) [10] Year

Poly-Si capacity

Poly-Si demand (CIsemiconductor)

Poly-Si demand (PV)

Available PV polySi

PolySi stocka

2003 2004 2005 2006 2007 2008 2009 2010

26,700 28,800 30,200 34,500 38,050 48,550 53,800 58,800

17,000 19,350 20,085 21,166 23,071 26,301 26,837 27,632

9000 14,032 18,181 16,705 17,435 24,089 28,233 32,108

9700 9450 10,115 13,334 14,979 22,249 26,973 31,168

+700 4582 8066 3371 2456 1840 1260 940

a Lack () or excess (+) of polycrystalline silicon in the worldwide market.

The major problem of the chemical route is that it involves the production of chlorosilanes and reactions with hydrochloric acid. In addition to being toxic, these compounds are corrosive, causing irritations of the skin and mucous membranes [17]. Two silane compounds (trichlorosilane and silicon tetrachloride) are intermediates in the solar-grade silicon production process via the chemical route. Not only are they highly volatile, corrosive and toxic but their handling also requires the utmost care, since they are explosive in the presence of water and hydrochloric acid [18]. Chlorine emissions in polycrystalline silicon production by the chemical route are estimated to amount to 0.002 kg of chlorine per square meter of cell. Controlling this emission is important because chlorine is denser than air, which accelerates the poisoning process. Its odor can be perceived at atmospheric concentrations of 0.3–0.5 ppm, but the symptoms of chlorine poisoning often appear before we become aware of the contamination [19].

3

Fig. 2 shows a general scheme of the two routes, identifying the suppression of the use of hydrochloric acid in the metallurgical route, with the resulting reduction of one phase of environmental control. Another attractive aspect of the metallurgical route is that the total consumption of energy in this process is lower; indeed, the consumption of energy is estimated to be 25% lower than that of the Siemens process [20]. The metallurgical route also makes more sense from the environmental standpoint, since the high growth of the PV industry is related directly to the quest for renewable and clean forms of generating energy. The production of solar cells by the chemical route also requires considerable consumption of energy and the handling and emission of toxic chemical compounds, leading to the need for coherent and responsible research for solutions enabling the energy generated to be really clean and low cost [20]. The following sections discuss the state of the art in the development of silicon purification technologies to obtain low-cost material. 2.2. Chemical routes 2.2.1. Wacker Chemie AG This German company foresees an expansion of its polycrystalline solar-grade silicon production in Burghausen, Germany, where its production capacity was increased in two stages: in 2006, the annual production will be increased by 500 metric tons and in 2007 by 1000 metric tons. After this expansion, its annual polycrystalline production capacity will be 6500 metric tons/year. The investments for this expansion lie in the order of 75 million euros [21]. The traditional Siemens process is employed to supply the microelectronics industry, as briefly mentioned earlier. The process developed for the production of low-cost silicon in Burghausen involves deposition in fluid-bed reactor, which offers the following advantages: shorter deposition time; does not involve etching; and growth of polycrystalline silicon in Sio300 mm grains (seeds). This process begins with a gaseous mixture of trichlorosilane and hydrogen, which flows through a bed containing silicon seed grains (with sizes of less than 300 mm). A constant flow of hydrogen is fed from the lower to the upper portion of the tube. The hydrogen flow agitate the silicon seeds, leaving them in suspension and favoring the deposition of polycrystalline silicon on the surface of the grains, thereby increasing their size, as illustrated in Fig. 3. This process is continuous and presents zones with distinct temperatures, allowing the material to be packaged immediately for use [21]. 2.2.2. Tokuyama Corporation The Japanese company, Tokuyama Corporation, is the one of the largest worldwide manufacturers of solar-grade silicon, with an annual production of 5940 metric tons/ year. Actually, Hemlock Semiconductor Corporation

Please cite this article as: A.F.B. Braga, et al., Sol. Energy Mater. Sol. Cells (2007), doi:10.1016/j.solmat.2007.10.003

ARTICLE IN PRESS 4

A.F.B. Braga et al. / Solar Energy Materials & Solar Cells ] (]]]]) ]]]–]]]

METALLURGICAL ROUTE

CHEMICAL ROUTE

Quartz

Quartz

Reduction in the presence of carbon

Emission of CO and silica REQUIRES CONTROL

Si - Mettalurgical

Si - Metallurgical

Purification: reactions with HC1 and H2

Reduction in the presence of carbon

Emission of Cl2 and production of chlorosilanes

Purification: controlled fusion, refusion and solidification

REQUIRES CONTROL

Production of solar cells

Handling of toxic elements

Production of solar cells

REQUIRES CONTROL

Solar Panels and Modules

Use of toxic compounds in the manufacture of panels, Environmental policy for discarding batteries,

Solar Panels and Modules

Fig. 2. Stages of the production of solar cells and panels that require research in the area of environmental control [20].

The process is based on the production of trichlorosilane from metallurgical-grade silicon in a VLD reactor, where the liquid silicon is produced and deposited at much higher rates than in the conventional process [22].

Fig. 3. Silicon seeds: (a) grains grown in the process and (b) of silicon purified by Wacker’s new process (these photographs are illustrative and are not on the same scale [21]).

produces 10,000 metric tons/year, Wacker produces 7500 metric tons/year, MEMC Electronic Material 6500 metric tons/year and REC 5500 metric tons/year [9]. This company uses the Siemens process, but in 2006 it began activities in a new production plant using an alternative route called VLD (vapor-to-liquid deposition), which is based on the chlorosilane decomposition on a silicon liquid film and allows a 10-fold higher deposition rate than in the Siemens process, generating the product at a lower production cost. This project was financed by Japan’s New Energy and Industrial Technology Development Organization (NEDO), and US $28.5 million were invested in the construction of the new plant.

2.2.3. REC Group—REC silicon The company proposes an alternative to the Siemens process, whereby solar-grade silicon is produced in an inverse U-shape hot filament CVD reactor from thermal decomposition of SiH4. This proposal involves a continuous process, which recycles chlorides and hydrogen. Similarly to the aforementioned companies, the proposed cost reduction is based on the increase of the solar-grade polycrystalline silicon production rate. This company is currently producing solar-grade and electronic-grade silicon with annual production capacity currently amounting to approximately 6000 metric tons [23]. At the plant in Moses Lake REC has increased the production capacity by approximately 6500 metric tons, only solar-grade silicon qualities are produced here [23,24]. Moreover, this company is also introducing the fluidized bed reactor process based on silane (SiH4) [23]. 2.2.4. Chisso Corporation—Chisso solar-grade silicon (CSS) CSS transacts business in chlorination, reduction and electrolytic reaction. In this technology, silicon tetrachloride produced by the chlorination reaction of metal silicon is

Please cite this article as: A.F.B. Braga, et al., Sol. Energy Mater. Sol. Cells (2007), doi:10.1016/j.solmat.2007.10.003

ARTICLE IN PRESS A.F.B. Braga et al. / Solar Energy Materials & Solar Cells ] (]]]]) ]]]–]]]

reduced by zinc to produce 6N grade polysilicon (99.9999%). The polysilicon specified for the PV generation purpose is used as the raw material of crystalline silicon solar cells. Chisso Corporation and NEDO have been engaged with the research and development of CSS since 2002. Upgrade from the bench-scale laboratory to a pilot plant is currently scheduled at Minamata Research Center, Minamata City, Kumamoto. The chlorosilane production technology that is applied in the chloridation process is currently operated by Chisso. It follows the legacy of Japan’s first high-purity silicon technology for the semiconductor purpose that was offered by Chisso in the 1960s. The core technology developed by Chisso will be fused with the electrolytic technology for the metal titanium production that Toho Titanium has accumulated over the years and the high-purity metal technology of Nippon Mining Holdings Group. CSS also employs a closed loop system. It cyclically utilizes zinc chloride, which is a side effect of the reduction process, to reduce costs and to enhance the production of high-quality polysilicon while greatly reducing the byproducts [25]. 2.3. Metallurgical routes 2.3.1. SOLSILC project In Europe, the SOLSILC and SPURT projects propose the development of solar-grade silicon by carbothermal reduction of silicon, based on the use of very pure raw materials. The project is the result of a partnership between the SINTEF Materials and Chemistry, ECN—Energy research Centre of the Netherlands, ScanArc Plasma Technologies AB and Sunergy Investco BV. This proposal can be classified as a metallurgical route to produce silicon and foresees the use of plasma as the heating source to reduce the silicon, followed by a unidirectional solidification process [26]. The raw materials for this process are ultrapure quartz and carbon black. On the one hand, this differential favors the production of a product with low boron content (controlled in the selection of the quartz) and a low concentration of phosphorus (the phosphorus may come from charcoal, depending on the cultivation process of the wood). On the other hand, however, it may be limited due to the exhaustion of quartz with this characteristic. A limitation of the process is the residual carbon originating from the reduction process. An investment of 1.1 million euros was made and solargrade polycrystalline silicon cells with 10% conversion efficiency were obtained. The advantage of the proposal is that it reduces the consumption of energy to 25–30 kWh/kg of product obtained. The Siemens process uses four times more energy (120 kWh/kg of product). A pilot-scale plant is already operating with a production capacity of 100 tons/year. This process is expected to reduce the cost of silicon to 15h/kg.

5

There is a group in Kazakhstan, at the Kazakh National Technical University, which is also working on the carbothermal reduction of quartz using deposits of ultrapure quartz in this country. Their results are promising and the best material obtained presents a purity of 99.96% in mass [27]. 2.3.2. Elkem ASA This company of Norwegian capital is today the largest worldwide manufacturer of metallurgical-grade silicon. In view of its technological vocation, Elkem opted for the development of a metallurgical route based on pyrometallurgical refinement [28] and on chemical treatment using acid solutions [29]. The consumption of energy by this process is similar to the SOLSILC route (25–30 kWh/kg), which is still at laboratory level. Solar cells with efficiencies of 15–16% have been obtained [30]. Elkem Solar has commenced construction of the first industrial-scale metallurgical solar silicon plant scheduled to start mid2008. The investment is 2.5 billions Norwegian crowns (300 million euros) and will produce 2500 metric tons and will be doubled to 5000 metric tons by 2010 [31]. 2.3.3. Kawasaki Steel Corporation The Japanese corporation Kawasaki uses the smelting technique in an electron beam furnace allied to plasma fusion with the objective of eliminating P and B, respectively. The starting material utilized is metallurgical-grade silicon. The first stage foresees the use of an electron beam to eliminate impurities. This is followed by the plasma process. The results of Kawasaki’s purification process are given in Table 3. The plasma process for the purification of silicon has been known for over 10 years. It has been stated [33] that boron can be reduced from 35.7 ppm in mass to 0.4 ppm in mass. Boron oxides (BO, B2O, B2O3, etc.) can be formed at temperatures exceeding 2027 1C. These oxides present a relatively high vapor pressure. For example, the estimated vapor pressure of BO at 2000 1C is 74 Pa, while that of BO2 is 0.15 Pa and that of B2O3 is 0.056 Pa, and B in its elemental form is 104 Pa at this temperature. Therefore, Kawasaki’s process proposes the removal of boron from the cast silicon matrix at 2027 1C (the boiling point of silicon is 2355 1C) in the form of BO, using oxidizing conditions for the plasma. The plasma torch is composed of a mixture of argon and water [33].

Table 3 Behavior of P, Al and Ca in the process developed by Hazanawa [32] Impurity

Concentration (ppmw)

P Al Ca

0.05 18 o0.1

Please cite this article as: A.F.B. Braga, et al., Sol. Energy Mater. Sol. Cells (2007), doi:10.1016/j.solmat.2007.10.003

ARTICLE IN PRESS A.F.B. Braga et al. / Solar Energy Materials & Solar Cells ] (]]]]) ]]]–]]]

6

2.3.4. Apollon Solar—PHOTOSIL project PHOTOSIL is a project which includes partners from industry, R&D institutes and equipment manufacturers. The objectives of this project are the production of solargrade silicon at costs o15h/kg and of multi-crystalline ingots at costs o35h/kg, starting with metallurgical silicon and using a combination of innovative up-grading and purification techniques. On the basis of encouraging results on laboratory level, the PHOTOSIL consortium has obtained the funding for the construction of an industrial-scale pilot line. This line became fully operational in October 2006 and serves to demonstrate the industrial viability of the PHOTOSIL technology by up-scaling the different laboratory-scale processes to an industrial level. In the first stage, the pilot line operates with batch sizes of 60 kg which will be doubled to 120 kg in the second stage, arriving at a nominal capacity of 200 tons/year. The PHOTOSIL process includes metallurgical and plasma purification techniques, giving rise to a complete vertical integration from the metallurgical silicon production to the fabrication of exploitable multi-crystalline silicon ingots for the PV industry, of either p- or n-type [34]. 2.3.5. UNICAMP Developing a study using electron beam melting principle. The main advantages of an electron beam melting furnace are: (1) high vacuum processing, which allows the elimination of elements whose vapor pressures are higher than that of silicon and (2) the use of a refrigerated copper crucible, which does not contaminate the silicon. As the conventional chemical process (trichlorosilane) is not used, large amounts of chemical wastes are not produced and there is no aggression to the environment. The principle of this technique is the generation of a beam of free electrons that are accelerated towards a target conductor such as a metal. An interaction occurs at the point of action of the beam with the atoms of the material, converting the electron beam’s kinetic energy into other forms of excitation energy. A 99.9995% purity silicon is obtained (Table 4) [15]. 2.4. Other routes 2.4.1. NTNU and SINTEF The Department of Electrochemical Technology of Norwegian University of Science and Technology proposes the use of the electrochemical process based on the dissolution of quartz in fluoride. The silicon is deposited on an electrode, after which it is ground and washed in an acid solution. The material is then cast into ingots (there is a possibility that the plasma technology may be associated in this stage). The process is considered promising for several reasons, i.e., (1) most of the contaminants of metallurgical silicon derive from the raw materials (quartz, carbon and processing equipment); (2) carbon can be eliminated in the electrochemical process; (3) aluminum

Table 4 Chemical analysis before and after silicon melting, final purity and efficiency of extraction [15] Impurities

Before melting (ppmw)

After melting (ppmw)

Efficiency of extraction (%)

Al Ba B Ca Cu Fe K Mg Mn Na P Ti Others

110.00 0.04 10.00 26.00 6.50 790.00 0.10 4.20 75.00 0.33 38.00 42.00 16.32

0.44 o0.01 7.3 0.31 0.29 0.70 0.10 0.02 0.027 0.05 0.39 0.087 2.74

99.60 75.61 27.00 98.81 95.54 99.91 0.00 99.50 99.96 83.94 98.97 99.80 83.21

Total

1108.49

5.165

99.53

Final purity (%)

99.88

99.9995

–

Fig. 4. Electrochemical cell [35].

salt processes occur at temperatures below 1000 1C; (4) very pure silicon can be deposited from dissolved silicon [35]. This academic project is involving NTNU (Norwegian University of Science and Technology) and the SINTEF (Foundation for Scientific and Industrial Research). Characteristics of the process (Fig. 4):

   

Anode material: Pt, graphite Anode product: Oxygen Cathode material: Si-alloy, another conducting metal or refractory (nitride, carbide) Cathode product: Si(s), Ca

3. Conclusions The photovoltaic (PV) industry has a growth of 30%, a situation that many other industries can only dream about.

Please cite this article as: A.F.B. Braga, et al., Sol. Energy Mater. Sol. Cells (2007), doi:10.1016/j.solmat.2007.10.003

ARTICLE IN PRESS A.F.B. Braga et al. / Solar Energy Materials & Solar Cells ] (]]]]) ]]]–]]]

Today, the majority of solar cells are made of silicon, and experts believe that it will take at least a decade before any other PV technology based on other materials will become competitive. The dramatic growth in the PV industry has, however, caused a lack of solar-grade silicon (SoG-Si), i.e., silicon with the required chemical purity for PV applications, resulting in increased prices for such material. Presently, the shortage of low-cost SoG-Si is the main factor preventing environmentally friendly solar energy from becoming a giant in the energy market in a generation or two. A direct metallurgical route for production of SoGSi can be five times more energy efficient than the conventional Siemens process that uses more than 200 kWh/kg. Several companies (Wacker, REC Silicon, etc.) are making a big effort to economize the chemical route further with the aim to produce SoG-Si instead of semiconductor grade. Tokuyama is testing a strongly modified filament reactor for their ‘‘vapour-to-liquid deposition (VLD)’’ process. Several companies and research institutes are working towards employing metallurgical processes in order to produce SoG-Si. Elkem Solar has shown that their SoG-Si may be used to produce cells of an efficiency of 15–16%. Acknowledgments The authors acknowledge FAPESP, Brazil, for funding this research. We are also indebted to technician Emı´ lcio Cardoso for his assistance. References [1] J. Perlin, Photovoltaics, /http://www.californiasolarcenter.org/history_ pv.htmlS (accessed 21.02.06). [2] J. Fally, E. Fabre, B. Chabot, Revue de l’e´nergie 37 (385) (1986) 761. [3] J. Li, Appl. Phys. Lett. 60 (18) (1992) 2240. [4] W.K. Schubert, D.L. King, T.D. Hund, J.M. Gee, Sol. Energy Mater. Sol. Cells 41/42 (1996) 137. [5] D. Carlson, The status and outlook for the photovoltaics industry, BP Solar—presentation, 2006 [6] A. D’Angelo, Enfrentar o futuro—O sol esta´ a nascer na indu´stria fotovoltaica, Empresa Europa no. 18, January–March 2005, /http:// europa.eu.int/comm/enterprise/library/enterprise-europe/issue18/ articles/pt/topic2_pt.htmS (accessed 21.02.06). [7] Solarbuzz, 2004 WORLD PV MARKET REPORT HIGHLIGHTS: world PV market up 62% in 2004, Silicon feedstock to restrict 2005 market growth, /http://www.solarbuzz.com/Marketbuzz2005-intro. htmS (accessed 14.12.05). [8] A. Waernes, Solar grade silicon by direct metallurgical process, SINTEF/ECN/SCANARC/SUNERGY, Silicon for the chemical industry VIII, June 2006 (accessed 21.08.07). [9] J.W. Bencik, M. McNamara, Clean Technology—Solar Energy Primer—Report, Jefferies & Company Inc., 2007, /http://www. jefferies.com/pdfs/confs/0507clean/CleanTechPrimer.pdfS (accessed 14.08.07).

7

[10] W.J. Pichel, M.R. Yang, Solar powered: an emerging growth industry facing severe supply constraints, Piper Jaffray, 2005. [11] Mineral Commodity Summaries, Minerals USGS, 2006, pp. 150–151. [12] Anua´rio ABRAFE, Anua´rio das Indu´strias Brasileiras de Ferroligas e de Silı´ cio Meta´lico, 2004, Cells, in: Proceedings of the Second Photovoltaic World Conference, July 1998, Vienna [13] E.C. Koch, Propellants, Explosives, Pyrotechnics 32 (3) (2007) 1. [14] A. Lisa, SILICON, Corathers-Minerals Yearbook, Vol. I, Metals and Minerals, US Government, 2005. [15] J.C.S. Pires, J. Otubo, A.F.B. Braga, P.R. Mei, J. Mater. Process. Technol. 169 (2005) 16. [16] P. Woditscha, Sol. Energy Mater. Sol. Cells 72 (2002) 11. [17] B. Sorensen, Life-cycle analysis of present and future Si-based solar, 1998. [18] E. Williams, Global Production Chains and Sustainability, United Nation University, Tokyo, Japan, 2000, 147pp. [19] E. Segal, 2004, Toxicity, chlorine gas, /www.emedicine.comS (accessed 2.02.06). [20] B. Kindembe, 2004, Thematic Research Report on Environmental Issues, NNE5/2001/264, /http://www.pvnord.org/results/downloads/ Task_3.2_Environmental_issues_Final%20report.pdfS (accessed 14.08.07). [21] E. Dornberger, Tiny spheres with a big effect, Innovations, Wacker Co. internal newsletter, 2005. [22] Tokuyama Corp., Responsible Care & Eco Management Department Tokuyama Corporation—Responsible Care Report 2005, /http:// www.tokuyama.co.jp/eng/enviro/report/pdf/2005rc_report_e.pdfS (accessed 31.07.07). [23] REC Group, Market and customers, /http://www.recgroup.com/ default.asp?V_ITEM_ID=452S (accessed 31.07.07). [24] REC Silicon Announces Expansion in Moses Lake—Internal Report, /http://www.grantedc.com/files/newsletters/61_REC_Announcement. pdfS (accessed 31.07.07). [25] Chisso Corporation, Production technology co-development of Chisso solar-grade silicon, 20 September 2006, /http://www.chisso.co. jp/english/news/pres060920.htmlS (accessed 9.08.07). [26] L.J. Geerligs, et al., Solar-grade silicon by a direct route based on carbothermic reduction of silica: requirements and production technology, ECN Report, The Netherlands, 2002. [27] Schmela, M., Mapping the silicon world, Photon International, May 2005, /http://www.photon-magazine.com/akademie/Silicon_ Conference_2005.pdfS. [28] J. Smith, Silicon refining process, US Patent 5.820.842, October 1998. [29] B. Ceccaroli, Refining of metallurgical grade silicon, US Patent 6.861.040, March 2005. [30] A. Waernes, New feedstock materials, SINTEF/ECN/SUNERGY Report, 2005. [31] Elkem Solar-investment decision-Skøyen, 27 October 2006, Internal Report, /http://hugin.info/111/R/1084266/188768.pdfS (accessed 31.07.07). [32] K. Hazanawa, 2001, Process and apparatus for refining silicon, US Patent 6.231.826 B1. [33] N. Yuge, Sol. Energy Mater. Sol. Cells 34 (1994) 243. [34] PHOTOSIL—Abstract at the European Photovoltaic Conference 2006 in Dresden, /http://www.apollonsolar.com/admin/uploads/ PHOTOSILl%20confe´rence%20Dresden%202006.pdfS (accessed 31.07.07). [35] G. Hagen, 2002, Silicon deposition from a chloride-based melt, NTNU, Department of Materials Technology and Electrochemistry, in: Proceedings of the EUCHEM Conference on Molten Salts, Oxford, /http://www.sintef.no/static/mt/2423/solar_cell_technology/ euchem2002/presentation.pptS (accessed 25.08.07).

Please cite this article as: A.F.B. Braga, et al., Sol. Energy Mater. Sol. Cells (2007), doi:10.1016/j.solmat.2007.10.003