METHOD AND APPARATUS FOR SOLAR GRADE SILICON PRODUCTION CROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims the benefit of, and incorporates herein in its entirety,

U.S. Provisional Patent Application Serial No. 62/246,520 filed on October 26, 2015 and entitled "SOLAR GRADE SILICON PRODUCTION BY THE SIF4 MG PROCESS."

BACKGROUND

[0002] The field of the invention relates to a method and an apparatus for producing silicon with high purity and low cost for solar cell manufacture, and more specifically, for producing solar grade silicon with a multi-step process including reducing silicon tetrafluoride (S1F4) gas with pure elemental magnesium.

[0003] More than 90% of the annual solar cell production is based on crystalline silicon, and half of the cost of solar cells results from fabrication of the Si substrate. Currently, there are two main industrial paths for the production of solar-grade Si. The primary path is the chemical routes, which include several variants of the Siemens process originally developed for semiconductor-grade Si production. However, the rate of production of this type of gas/solid reaction is significantly limited by the small surface area available and heat transfer capabilities of the system. In practice, many reactors (> 30) have to be operated in parallel for mass production. Generally, the large energy requirements for this type of process (over lOOkWh/kg Si) and the complicated process/reactor results in high costs.

[0004] Another industrial production path is the selective purification of metallurgical- grade Si. With this particular application, the need to combine several steps to achieve a minimum desired purity has resulted in a high cost-to-purity ratio process. Therefore, a need exists to produce solar grade silicon with a highly efficient (consuming less energy), low cost- to-purity ratio, reliable, and high yield process. In addition, a new design for a high throughput apparatus for producing solar grade silicon is desired.

SUMMARY OF THE INVENTION

[0005] Embodiments of Applicants' disclosure describes a method for making high purity elemental silicon, comprising: pressurizing a first reacting container with silicon tetrafluoride (S1F4); adding elemental magnesium (Mg°) to the container to generate Si ⁰ and MgF2; and separating the Si ⁰ and the MgF2.

[0006] Further, the pressure of SiF4 in the first reacting container could be between about 0.1 standard atmosphere and about 10 standard atmospheres; the pressure of S1F4 in the first reacting container could also be between about 0.5 standard atmosphere and about 1.5 standard atmospheres; and the pressure of S1F4 in the first reacting container could be between about 0.8 standard atmosphere and about 1.2 standard atmospheres.

[0007] The method further comprises heating the first reacting container to a first temperature, wherein the first temperature could be between about 800°C and about 2000°C; the first temperature could be between about 1200°C and about 1600°C; and the first temperature could be between about 1250°C and about 1450°C.

[0008] The method further comprises injecting elemental magnesium in a form that is selected from the group consisting of elemental magnesium particles, elemental magnesium bricks, elemental magnesium sponge, and liquid elemental magnesium. Moreover, the liquid elemental magnesium is added at a temperature between about 662°C and about 725°C. In addition, the elemental magnesium is added in amounts such that when melted a pool of molten elemental magnesium comprises a height less than about 10 cm. The elemental magnesium can also be added in amounts such that when melted a pool of molten elemental magnesium comprises a height less than about 3 cm.

[0009] Moreover, the method comprises providing a first reacting container comprising a reaction zone and one or more first reacting container walls; and adding the elemental magnesium at a rate to control a temperature of the reaction zone and a temperature of one or more first reacting container walls. The providing further comprises providing a first reacting container comprising one or more first reacting container walls and providing the first reacting container comprising an insulating layer, which may have a heating unit disposed therein, disposed between the one or more external walls and the one or more internal walls. In addition, the one or more first reacting container walls comprises an external portion comprising Inconel and the one or more first reacting container walls an internal portion comprising Graphite. The insulation material is selected from the group consisting of silicon dioxide (S1O2), silicon nitride (Si3N ₄ ), and silicon carbide (SiC). One or more internal walls can be coated with pure MgF2.

[0010] The method further comprises removing unreacted S1F4 gas from the first reacting container; filling the first reacting container with an inert gas; opening the first reacting container; and discharging the elemental Si and the MgF2. Moreover, the method comprises providing a first reacting container formed to include one or more apertures extending through a bottom portion thereof; disposing a second container adjacent to the first reacting container; and discharging MgF2 from the first reacting container into a second container. In addition, the method comprises melting the elemental Si and discharging molten elemental Si into the second container and forming the one or more apertures to comprise a diameter of between about 0.5 mm and about 1 cm. Further, the method comprises after discharging the elemental Si from the first reacting container, crystallizing the molten element Si to form an elemental Si ingot.

[0011] Embodiments of Applicants' disclosure further describe an apparatus for preparing high purity elemental Si. The apparatus comprises a container having an outer container surface and defining an enclosed space, wherein the container is capable of withstanding temperature greater than about 1400°C; and a lid removably attached to the container. The apparatus further comprises a cooling coil comprises a spiral wound, tubular member to circulate a coolant therethrough.

[0012] Further, the container comprises one or more container walls. The one or more container walls comprise an external portion comprising Inconel. The one or more container walls comprise an internal portion comprising Graphite and the one or more container walls comprise an insulating layer disposed between the external portion of the one or more external walls and the internal portion of the one or more internal walls.

[0013] The apparatus further comprises a thermocouple capable of monitoring a temperature of the reaction zone and the one or more container walls; a pressure sensor in fluid communication with the enclosed space; a pressure-gauging port extending through the one or more container walls to communicate with the pressure sensor; a vacuum port extending through the one or more container walls to communicate with the enclosed space; a material input tube extending through the one or more container walls to communicate with the enclosed space; and a gas inlet extending through the one or more container walls to communicate with the enclosed space.

[0014] In addition, the apparatus comprises one or more second containers, wherein at least one second container is located below or adjacent to the first reacting container such that a reaction product can be disposed into the second container from the first reacting container.

BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The patent or application file contains at least one drawing executed in color.

Copies of this patent or patent application publication with color drawing(s) will be provided by the Office upon request and payment of the necessary fee.

[0016] FIG. 1 A illustrates an embodiment of a reacting container 100 used to carry out the reaction of elemental magnesium (Mg) and S1F4;

[0017] FIGS. IB and 1C illustrate reacting container 101; [0018] FIG. ID shows yet another embodiment of the reacting container 101;

[0019] FIG. IE shows different layers of one or more walls of the reacting container

100/101 ;

[0020] FIG. IF illustrates reacting container 102;

[0021] FIGS. 2A-2C show XRD phase analysis of the reaction product after S1F4 and

Mg reaction at a series of temperatures;

[0022] FIG. 3 A is a top view of an open container 130 during Mg ignition showing a bright color;

[0023] FIG. 3B is the reaction product showing both brown particles and fracture areas with metallic appearance;

[0024] FIG. 4 is phase diagram of Mg-Silicon;

[0025] FIGS. 5A-5D are flowcharts summarizing a method for making high purity elemental silicon (Si ⁰ );

[0026] FIGS. 6A and 6B show cross-section views of the reaction product in a graphite crucible under optical microscope and the product in the crucible after melt separation;

[0027] FIGS. 7 A and 7B illustrate Si ⁰ after the melt and separation process; and

[0028] FIG. 8 is a diagram of solubility of Mg in silicon.

[0029] The foregoing and other aspects and advantages of the invention will appear from the following description. In the description, reference is made to the accompanying drawings that form a part hereof, and in which there is shown by way of illustration preferred embodiments of the invention. Such embodiments do not necessarily represent the full scope of the invention, however, and reference is made therefore to the claims and herein for interpreting the scope of the invention.

DETAILED DESCRIPTION

[0030] Embodiments of the disclosure solve the issue of manufacturing solar grade silicon with high purity at a large scale with a lost cost. In order to be suitable for use in the photovoltaic industry, the photovoltaic industry generally requires that metallurgical grade silicon that has a purity level of about 98-99% by weight, be further purified to a purity level of 99.99-99.9999% by weight.

[0031] Embodiments of disclosure are described with examples referring to the

Figures, in which like numbers represent the same or similar elements. Reference throughout this specification to "one embodiment," "an embodiment," or similar language means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the present invention. Thus, appearances of the phrases "in one embodiment," "in an embodiment," and similar language throughout this specification may, but do not necessarily, all refer to the same embodiment.

[0032] The described features, structures, or characteristics of the invention may be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are recited to provide a thorough understanding of embodiments of the invention. One skilled in the relevant art will recognize, however, that the invention may be practiced without one or more of the specific details, or with other methods, components, materials, and so forth. In other instances, well-known structures, materials, or operations are not shown or described in detail to avoid obscuring aspects of the invention.

[0033] For the terms "for example" and "such as," and grammatical equivalents thereof, the phrase "and without limitation" is understood to follow unless explicitly stated otherwise. As used herein, the term "about" is meant to account for variations due to experimental error and to reflect a plus or minus 10% difference in any measurement unless stated otherwise. Numerical result of any measurement disclosed below is understood to be modified by the term "about," whether or not the term is explicitly used and unless explicitly stated otherwise.

[0034] Referring to FIGS. 1A-1D and IE, various embodiments of a reacting container

100 for preparing high purity elemental Si (Si ⁰ ) are illustrated. In certain embodiments, referring to FIG. 1A, the reacting container 100 comprises an outer container surface 1 10, which defines a first enclosed space 120; an induction coil 140, which wraps around the outer reacting container surface 110; and an open cup 130, which is disposed within the first enclosed space 120. Further, the outer reactor surface 110 comprises a first wall 102 and a second wall 104. In certain embodiments, the first and the second walls 102 and 104 comprise Quartz with cooling coolant running in between. In other embodiments, the walls 102 and 104 comprise high temperature enduring and corrosion resistant alloys, such as, molybdenum, stainless steel, tungsten, tantalum, titanium, or nickel.

[0035] The first and the second walls 102 and 104 define a second enclosed space 106.

The second enclosed space 106 can be used to circulate liquid coolant, for example, water, to cool the outer reactor surface 1 10 and the reacting container 100. The first wall 102 further comprises a water inlet 108 and a water outlet 109, which are used to inject, circulate, and remove liquid coolant within the second enclosed space 106 between the two walls 102 and

104. The first wall 102 also comprises a vacuum port 180 and a pressure-gauging port 190, which can be connected to a pressure sensor 170 to monitor the pressure within the reactor 100. Further, the vacuum port 180 ensures the first reacting container 100 is able to remain gas tight during any reaction. In other embodiments, both the walls 102 and 104 can be made of materials that are suitable to withstand high temperature up to 2000°C.

[0036] In certain embodiments, the induction coil 140 comprises a radio frequency

(RF) induction coil. The induction coil 140 is in physical contact with the outer reacting container surface 110 in some embodiments, while the induction coil 140 is not in physical contact in other embodiments. RF induction is the use of a radio frequency magnetic field to transfer energy by means of electromagnetic induction in the near field. A radio-frequency alternating current is passed through a coil of wire that acts as the transmitter, and a second coil or conducting object, magnetically coupled to the first coil, acts as the receiver. To a person skilled in the art, other suitable methods for heat induction and energy transferring can be employed here to heat up the reacting container 100 to a certain temperature.

[0037] The open cup 130 comprises a plurality of layers. In certain embodiments, the open cup 130 comprises three layers: an outer graphite felt insulation layer 132, a middle graphite layer 134, and an inside graphite layer 136. In other embodiments, the open container 130 can comprise additional 1 or 2 insulation layers made of suitable materials that can withstand high temperature and are able to insulate. The graphite used here possess certain chemical and physical properties that make it suitable for high-temperature applications, such as thermal radiation shielding and exothermic reaction. Similarly, other types of material that possess the same properties can be used to form the plurality of layers of the open container 130.

[0038] The reacting container 100 further comprises a quartz stand 150 and a thermocouple 160. The quartz stand comprises a first surface 151, which touches the bottom of the open cup 130. The thermocouple 160 extends through an opposing surface 152 of the quartz stand 150 and the first surface 151 of the quartz stand 150. Further, the thermocouple 160 is in physical contact with the open cup 130. Thus, the thermocouple 160 is able to monitor the open cup 130's temperature.

[0039] Referring to FIGS. IB- ID, different embodiments of the reacting container are illustrated. These embodiments are designed for industrial scale-up production of Si with a high purity. The reacting container 101 can have a volume of 10 liters to 1,000 liters. While specific values chosen for this embodiment are recited, it is to be understood that, within the scope of the invention, the values of all of parameters may vary over wide ranges to suit different applications. The reacting container 101 comprises a top lid 210, a bottom pot 220, an elemental Mg injection tube 212, and a gas injection port 214. The top lid 210 is removably attached to the bottom pot 220. The top lid 210 can be secured to the bottom pot 220 and keep the bottom pot 220 gas tight by methods known to person skilled in the art. For example, in this embodiment, a rim 216 of the top lid 210 is screwed to a rim 224 of the of the bottom pot 220. Further, materials that seal and keep a container air tight can be disposed between the rim 216 and the rim 224. The U.S. Patent No. 4,781,565 is incorporated by reference in its entirety to describe the elemental Mg injection tube 212 and the gas injection port 214 in certain embodiments.

[0040] According to different designs and needs, in certain embodiments, the bottom pot 220 can be a cylindrically shaped with a substantially flat bottom 226. In other embodiments, an outer wall 228 of the bottom pot 220 can be pentagonal, hexagonal, heptagonal, octagonal, nonagonal, or decagonal; and the bottom 226 can comprises a curvature. In addition, metal fins encircles the bottom pot 220 to strengthen the pot and dissipate heat during reaction process. A cooling coil 230 wraps around the outer wall 228 and is disposed at the top of the bottom pot 220 to keep the rim 216 and rim 224 cool. In certain embodiments, the cooling coil 230 is in physical contact with the outer wall 228 and comprises a tubular member to circulate a coolant therethrough.

[0041] Referring to FIG. ID, the reacting container 101 further comprises handles

240a, b, which are used to facilitate tipping the reaction container 101 to discharge metal or salt molten formed in the enclosed space 120 after removing the top lid 210.

[0042] Further, FIG. IE illustrates the structure of the outer wall 228. In certain embodiments, the outer wall 228 comprises an external portion 240, which comprises Inconel.

Optionally, the external portion 240 is lined with a Nickel lining-layer 242. In other embodiments, the external portion 240 comprises high temperature enduring and corrosion resistant alloys, such as, molybdenum, stainless steel, tungsten, tantalum, titanium, or nickel.

Further, the outer wall 228 comprises an internal portion 246, which comprises GRAFOIL with the same chemical and physical properties described above. In certain embodiments, an insulation layer 244 is disposed between the external portion 240 and the internal portion 246.

The insulation layer 244 comprises flexible graphite felt in certain embodiments; while the insulation layer 244 comprises a material that is selected from a group consisting of silicon dioxide (SiC ), silicon nitride (S13N4), silicon carbide (SiC), other non-contaminating materials, and any combination thereof. The insulation layer 244 materials can be porous or in the form of sponges, felts, beds of particles, and/or any combination thereof. In certain embodiments, an internal heating unit is disposed between the external portion 240 and the insulation layer 244. In other embodiments, the internal heating unit is disposed within the insulation layer 244. Further, in some embodiments, the internal heating unit comprises a RF coil, which does not physically contact the external portion 240 or the insulation layer 244. In other embodiments, the heating unit employs resistance heating method or any other methods known in the art to heat the reacting container 101. In addition, in certain embodiments, the reacting container 101 can be heated by an external heating unit. In some embodiments, the external heating unit comprises heating tapes and/or hot plate.

[0043] Moreover, FIG. IF illustrates an embodiment of the reacting container 102 that can be employed in a semi-continuous or a continuous multi-stage system for reaction and melt coalescence and separation. U. S. Patent No. 8,475,540 is incorporated herein by reference in its entirety to describe the multi-stage system for reaction and melt coalescence and separation. In certain embodiments, the reacting container 102 comprises a first orifice 250 and a second orifice 260 extending through a bottom 270 of the reacting container. In other embodiments, the first orifice 250 and the second orifices extend through the outer wall 228. The first and the second orifices 250 and 260 allow removal of liquefied reaction product generated in the reacting container 102. In certain embodiments, a plurality of orifices can be disposed in the bottom 270 and/or the outer wall 228. With regards to the size of the orifice, in certain embodiments, the diameters of the orifices 250 and 260 are less than 1 cm; in other embodiments, the diameters of the orifices 250 and 260 are less than 0.5 cm; and in yet other embodiments, the diameters of the orifices 250 and 260 are less than 2 mm. While specific values chosen for the number and the size of the orifices are recited, it is to be understood that, within the scope of the invention, the values of the number and the size of the orifices can vary over wide ranges to suit different applications.

[0044] FIGS. 5A-5D illustrate Applicants' method for making high purity elemental silicon (Si ⁰ ) using different embodiments of the reacting container 100, 101, or 102. In step

505, elemental magnesium (Mg°) is weighed for each reaction and preloaded into a material input tube 180 connected to the reactor 100. In certain embodiments, Mg° comprises a Log5 purity (herein after as 5N Mg°), which is 99.999% pure of Mg° and comprises less than about

1 parts per million by weight (ppmw) of boron (B). In other embodiments, Mg° comprises a

Log4 purity (herein after as 4N Mg°), which is 99.99% by weight pure of Mg° and comprises less than 1.3 ppmw of B. In certain embodiments, Mg° is divided into cubes, each of which weighs about 2 g to 4 g. In other embodiments, Mg° can be injected into the reacting container

101 via injection tube 212 in the form of Mg° particles, Mg° bricks, Mg° sponge, and/or liquid

Mg°. Magnesium particles are typically in the range of 50 to 120 micron in diameter and are also available as nanoparticle (< 1 micron in diameter). In yet other embodiments, liquid Mg° is added at a temperature of the reacting container 101 from about 662°C and about 725°C. Preferably, Mg° is added at a temperature below the temperature of the liquid Mg°'s reaction or ignition point in S1F4 atmosphere.

[0045] Now referring to step 510, before heating in step 525, the reacting container

100, 101, or 102 is evacuated through the vacuum port 180 (FIG.1A). Then the reacting container 100, 101, or 102 is back-filled with inert gas in step 515. In certain embodiments, the back-filling of inset gas in step 515 is repeated one more time. In other embodiments, the back-filling of inert gas in step 515 is repeated two more times. The examples of how many times to back fill the reacting container 100, 101, or 102 with inert gas are not limiting, the reacting container 100, 101, or 102 can be back-filled with inert gas followed with evacuation for several times as long as the residual air in reacting container 100, 101, or 102 is sufficiently low. In certain embodiments, the inert gas is argon. In other embodiments, the inert gas is from the Periodic Table Group 0 gases, such as helium, argon and the like.

[0046] In step 520, silicon tetrafluoride (S1F4) gas is added through a gas injection port

214 (FIG. IB). In certain embodiments, during this filling process, the pressure within the reacting container 100, 101, or 102 is constantly monitored by the pressure sensor 170. Once the interior pressure of the reacting container 100, 101, or 102 reaches about a range of about 0.1 standard atmospheric pressure (atm) to 10 atm. One standard atmospheric pressure is about 101325 Pascal (Pa). Once a desired interior pressure of the S1F4 gas, the gas injection port 214 is shut off to prevent over-filling of more S1F4 gas.

[0047] After the reacting container 100, 101, or 102 is filled with S1F4, in step 525, the reacting container 100, 101, or 102 is heated to a certain temperature by supplying power to the induction coil 140. In certain embodiments, the temperature of the reacting container 100,

101, or 102 is monitored via the thermocouple 160. In other embodiments, the temperature of the reacting container 100, 101, or 102 is monitored by an optical pyrometer focused on the reacting container 100, 101, or 102. In some embodiments, the reacting container 100, 101, or 102 is heated to between about 800°C and about 2000°C. In other embodiments, the reacting container 100, 101, or 102 is heated to between about 1200°C and about 1600°C. In yet other embodiments, the reacting container 100, 101, or 102 is heated to between about 1250°C and about 1450°C. Heating temperatures may vary throughout the heating step, and particularly throughout the S1F4 and Mg exothermic reaction process. Thus, where it is said that the reacting container 100, 101, or 102 is heated to a nominal temperate, the actual temperature may fluctuate and may not be held constant at said temperature. [0048] When a reaction zone 280 (FIG. ID) reaches the predetermined temperature, the Mg° in any selected form is dropped into first reacting container 100, 101 , or 102 at a controlled rate via the material injection tube 212 in step 530. In certain embodiments, the controlled rate of injecting Mg° is determined by a desired rate of reaction. If a faster rate of reaction is desired, a faster controlled rate of Mg° injection is used. If a slower rate of reaction is desired, a slower controlled rate of Mg° injection is used. In other embodiments, the controlled rate of injecting Mg° is also determined by a height of a pool of molten Mg° in the reaction zone 280 (FIG. ID). For example, the Mg° is injected at a rate such that when melted a pool of molten elemental magnesium comprises a height less than about 10 cm; the Mg° is injected at another rate such that when melted a pool of molten elemental magnesium comprises a height less than about 5 cm; and the Mg° is injected at yet another rate such that when melted a pool of molten elemental magnesium comprises a height less than about 3 cm.

[0049] As the first Mg° get sufficiently hot, it reacts vigorously with surrounding S1F4 gas through the reaction:

Basic thermochemical estimates indicate the reaction is very exothermic. Table 1 shows the enthalpy and Gibbs energy change of this reaction in a range of temperatures. Even at 1373 K, the heat released is still well over 600 kJ. Based on the heat capacity data and estimations using the Outokumpu HSC® Chemistry program and database, the enthalpy of reaction should be enough to melt all the MgF2 produced (1260°C), and the adiabatic temperature of the system is above the melting point of Si (1412°C), which could result in melt coalescence of the silicon powder product, a desirable characteristic for the separation of product from byproduct. Moreover, the temperature of the reaction zone 280 is maintained by a combination of

Table 1. Enthalpy (ΔΗ) and Gibbs energy change (AG-), of the reaction SiF ₄ (g)+2Mg;(l)=Si-)-2MgF ₂ at a series of: temperatures. Data provided by Outokumpu HSC ^® .

Temperature (K) Mi (kJ) AG (Id) Temperature (K) ΔΙΙ (kJ) AG (kJ)

873 -646.1 -408 1573 -514.8 -223.8

973 -644.5 -380.8 1673* -510.3 -205.4

1073 -642.9 -353.7 1773 -455.9 -190.0

1173 -64L 1 -326.9 1873 -451.6 -175.1

1273 -639.2 -300.2 1973 -447.4 -160.4

1373 -637.3 -273.6

* ote: Si melts at 1685K exothermic reaction, external heating, and the insulation lay [0050] In certain embodiments, when Mg° is sequentially added in solid form, the next

Mg° piece is fed in the first reacting container 100 when reaction turbulence subsides.

[0051] During step 535, the temperature of the reaction zone 280 and the S1F4 pressure inside the reacting container 100, 101, or 102 are monitored continuously. The S1F4 gas should always be in excess of Mg° to ensure complete reaction of Mg°, therefore, the pressure of S1F4 gas in the reaction zone 280 is maintained at about 0.1 to 10 atm in certain embodiments. In other embodiments, the pressure of S1F4 gas in the reaction zone 280 is maintained at about 0.5 atm to 1.5 atm. In yet other embodiments, the pressure of the S1F4 gas in the reaction zone 280 is maintained at about 0.8 atm to 1.2 atm. In still yet other embodiments, the pressure of the S1F4 gas in the reaction zone 280 is maintained at about 1 atm. The pressure of the S1F4 gas may vary throughout step 520, and particularly throughout the S1F4 and Mg exothermic reaction process. Thus, where it is said that the first reacting container 100 is pressurized to a nominal pressure, the actual pressure of the S1F4 gas may fluctuate and may not be held constant at the pressure.

[0052] The separation of Si ⁰ from the reaction product between S1F4 and Mg exothermic reaction process can be performed in a batch mode, a semi-continuous mode, or a continuous mode. If the separation of Si ⁰ is performed in a batch mode, the step 536 transitions to step 540 in FIG.5B. In step 540, when there are more Mg°, the steps 530 and 535 will be repeated until there is no more Mg° in the material injection tube 214. When there is no more Mg°, step 545 will be performed.

[0053] In step 545, after the reaction for the last piece/amount of Mg° added subsides, the reacting container 100, 101, or 102 is allowed to cool by filling in and circulating one or more Periodic Table Group 0 gas in the reacting container 100, 101, or 102. In certain embodiments, argon gas is used to cool the reacting container 100, 101, or 102. In other embodiments, the cooling coil 230 is used to cool the reacting container 100, 101, or 102. After the reacting container 100, 101, or 102 cools down, the reaction product is collected and analyzed in step 550. For example, the morphology of the reaction product is studied by optical microscopy and the constituent phases are analyzed using X-ray powder diffraction (XRD).

[0054] Referring to FIGS. 2A-2C, in certain embodiments, when the first reacting container 100 is heated to 650°C in step 525, FIG.2C illustrates that the reaction product was brownish, and XRD detected crystalline phases consisting mainly of MgF2, Mg2Si, and small amounts of Si and unreacted Mg. Thermochemical analysis using the HSC program and data indicates that in this temperature range— Mg2Si is more stable than Mg or Si at this temperature, and its stability decreases with increasing temperature. In other embodiments, when the first reacting container 100 is heated to 820°C in step 525, FIG.2B illustrates the XRD spectrum and the peaks and amount of Si ⁰ increased and that of Mg2Si formed decrease significantly to trace levels. Further, both brown lava-like globular formations and metallic powders can be seen in the reaction product, as shown in FIG. 3B. In yet other embodiments, when the first reacting container 100 is heated to 1020°C in step 525, FIG.2A illustrates the XRD spectrum and shows that the main reaction product are silicon crystallites dispersed in a matrix of MgF2, with only a trace amount of Mg2Si left. The Mg is completely reacted based on the weight balance analysis.

[0055] Referring to FIG. 4, a phase diagram of Mg-Si shows that Mg2Si might be present with Si and Mg at temperatures below about 1081.4°C. Therefore, the reacting container temperature needs to be above about 1081°C (not necessarily initially during the beginning of the reaction between S1F4 and Mg) in order to completely eliminate Mg2Si. For example, when the reaction product is heated in S1F4 at temperatures above 1200°C, a complete reaction could be achieved and no Mg2Si peaks are detected.

[0056] Referring to FIG. 5B, after the completion of step 560, the reaction product generated in step 535 will go through steps 560-585 to separate Si from the reaction product. In certain embodiments, the method of melting of both Si and MgF2 phases to form two immiscible liquids that can be discharged separately. In step 560, the reaction product generated from step 535 is loaded in a second container 200. During heating, the second container 200 can be heated to any temperature sufficient to melt silicon. In certain embodiments, the second container 200 is heated inside a RF field to about 1420°C in step 565. In other embodiments, the second container 200 can be heated to a temperature of about 1400°C to about 1700°C or from about 1500°C to about 1600°C. Further, where it is said that the second container 200 is heated to a certain temperature, the actual temperature can fluctuate and may not be held constant at the temperature. In certain embodiments, most of the silicon powders coalesced and sunk to the bottom as large Si balls (> 1 mm in diameter) in step 570, i.e., the melt coalescence and separation step, as shown in FIG.7A. This is expected since molten Si has a density of 2.47 g/cm ³ and MgF2 has a density of 2.34 g/cm ³ at around 1420°C; Si has a higher surface tension. Further, FIG.6B shows the spherical frozen shape of Si (bright circles) surrounded by the darker MgF2. Further, FIG. 7A shows a product of Si in a fluoride after melt coalescence and separation step and the product is collected in step 575. FIG.7B illustrates a purified Si ingot after cleaning the fluoride by known methods in the art.

[0057] In step 580, the second container 200 is allowed to cool by filling in and circulating one or more Periodic Table Group 0 gas in the reacting container 100, 101, or 102.

In certain embodiments, argon gas is used to cool down the second container 200. After the second container 200 cools, the reaction product is analyzed in step 580. The purity of the final silicon product is critical to produce solar-grade silicon. Especially, the level of impurities that have relatively low separation coefficients, such as boron and phosphorus, are great concerns to the purity of the final silicon product because other metallic impurities that have large segregation coefficient between solid and liquid Si phases during crystal pulling are expected to accumulate in the liquid phase, thus are removed from solid crystal Si phase. Therefore, the levels of impurities having low segregate coefficient in the as-produced Si product are critical and need to be at or below 1 part per million (ppm) level. In certain embodiments, the phosphorus level of the Si product obtained using 5N Mg is only at about 0.1 ppmw and the boron level is only at about 0.77 ppmw. This Si product yields a P-type silicon after crystal growth. In other embodiments, the phosphorus level of the Si product obtained using 4N Mg is only at about 2.6 ppmw and the boron level is only at about 1.3 ppmw.

[0058] If the separation of Si ⁰ is performed in a semi-continuous mode, step 537 transitions to step 605 in FIG. 5C. When the reaction in step 535 runs continuously, part of the molten MgF2 is discharged continuously into a second container 200 via the orifice 260 (FIG.

IF) in step 610. In certain embodiments, the receiving second container 200 is disposed below the reacting container 100, 101, or 102 and the molten MgF2 flows into the receiving second container 200 via the orifice 260. In certain embodiments, the diameter of the orifice 260 is less than 1 cm. In other embodiments, the diameter of the orifice 260 is less than 0.5 cm. In yet other embodiments, the diameter of the orifice 260 is less than 2 mm.

[0059] In step 615, if the reacting container 100, 101, or 102 has more capacity to generate more reaction product in step 535, the reaction in step 535 will continue and the discharging of molten MgF2 in step 610 will continue as well. When the reacting container 100,

101, or 102 has no capacity left to generate more reaction product, i.e., the reacting container

100, 101, or 102 is filled with Si and the internal portion 246 of the wall 228 is wet with MgF2, the reaction in step 535 will stop. In certain embodiments, the temperature of the reacting container 100, 101, or 102 is increased to above the melting point of Si so that both remaining

MgF2 and the generated Si can be discharged into a second receiving container 200 in step 620.

Then the second receiving container 200 can be cooled to let molten Si to slowly solidify to obtain a further purified obtain poly crystalline or single crystal Si ingot in fluoride (FIG. 7A).

In other embodiments, after discharging of byproduct MgF2, generated molten Si can be directly cooled in the reacting container 100, 101, or 102 to slowly solidify to obtain a further purified obtain poly crystalline or single crystal Si ingot in fluoride (FIG. 7A). Further purification methods known in the art can be used in step 630 to remove fluoride (FIG. 7B).

[0060] If the separation of Si ⁰ is performed in a continuous mode, step 538 transitions to step 705 in FIG.5D. In this continuous mode, the reacting container 100, 101, or 102 does not need to stop the reaction process in step 535 since byproduct MgF2 and Si are both discharged into the second receiving container 200 and a third receiving container 300 respectively from steps 710, 715, and 720. When the third receiving container 300 is filled with Si, the third receiving container 300 will be cooled to collect polycrystalline or single crystal Si ingot in step 725. Similarly, further purification methods known in the art can be used in step 730 to obtain polycrystalline or single crystal Si ingot (FIG. 7B). At the meantime, an empty receiving container 300 will be used to collect Si from the first reacting container 100 continuously. The continuous mode is largely useful in solar cell industry because the production capacity of Si with high purity is increased compared to semi-continuous mode or batch mode.

[0061] Moreover, any residual of Si and MgF2 is left in either the first reacting container

100, the second receiving container 200, or the third receiving container 300 can be melted and separated by using a centrally located heat source such as those provided by a plasma torch, arc, or electrode.

[0062] The following examples are presented to further illustrate to persons skilled in the art how to make and use the invention. These examples are not intended as a limitation, however, upon the scope of the invention, which is defined by claims herein. All publications, patents, and patent applications disclosed herein are hereby incorporated by reference as if set forth in their entirety.

EXAMPLE 1

Temperature Effects, Reaction Kinetics and Product Analysis

[0063] Inventors first explored the effect of initial temperature of Mg on ignition time, reaction rates, and the morphology and composition of the reaction products. Pieces of Mg (a few grams each) were fed into a pre-heated graphite crucible under an atmosphere of S1F4. Inventors observed whether or not any reaction was occurring by monitoring the surface of the Mg and the pressure drop.

[0064] No significant reaction occurred when the crucible was pre-heated below

650°C, which is the melting point of Mg. Based on these findings, inventors performed experiments in which the graphite crucible was preheated to 650°C, above 800°C, and above 1000°C, respectively, in 1 atm pressure of S1F4 before inventors drop-fed pieces of Mg metal into the hot crucibles.

[0065] Run 1. In the first run, the graphite crucible was pre-heated to 650°C (just above the melting point of Mg), and the pressure of S1F4 was slightly higher than 1 atm. When the Mg was dropped into the hot crucible, it did not ignite after 5 minutes. Inventors think the delay was probably due to heat transfer from the hot crucible to the cold Mg piece. After heating power was increased, the Mg "ignited" and gave off a bright flame, the temperature of the reactor increased to above 700°C, and simultaneously the pressure of S1F4 dropped. The peak reaction rate was relatively slow, the flame took ~ 1 minute to diminish, and no spark and a minimum amount of dust flew out of the crucible.

[0066] The reaction product was brownish, and XRD detected crystalline phases consisting mainly of MgF2, Mg2Si, and small amounts of Si and unreacted Mg, as shown in FIG. 2A. The crucible was lined with GRAFOIL, and some of the reaction product stuck to it, so peaks of carbon were also observed in the XRD spectrum as well. Thermochemical analysis using the HSC program and data indicates that in this temperature range— Mg2Si is more stable than Mg or Si at this temperature, and its stability decreases with increasing temperature.

[0067] Run 2. The crucible was preheated to a temperature of 820°C, and the Mg ignited

~ 1 minute after it was dropped into hot crucible (FIG. 3A) at which point the S1F4 pressure dropped rapidly due to gas consumption. The flames above the reaction zone were accompanied by a generation of bursts of parti cles/dust, which occasionally reached the top of the reactor and coated the reactor walls with a fluffy brown powder. The crucible temperature increased to 862°C because of the heat released from the reaction. The total reaction took ~ 2 minutes to finish. The ignition and reaction of subsequently added Mg chunks became more violent, faster (~ 1 minute of burning), and more intense as the temperature of the crucible increased as heat was released. The crucible reached a maximum temperature of 909°C. FIG. 3B shows a photo of the reaction product retrieved after Run 2. Both brown lava-like globular formations and metallic powders can be seen in the reaction product, and its XRD spectrum is shown in FIG. 2B. As the system temperature increased, the peaks and amount of Si increased and that of Mg2Si formed decreased significantly to trace levels.

[0068] Note that at 859°C, the temperature experimentally measured at the bottom of the crucible, the vapor pressure of pure Mg should be only 0.1 atm. But the formation of a flame above the solid-gas reaction zone indicated that the Mg must be vaporizing at pressures above 1 atm and reacting with S1F4 in a homogeneous gas-gas reaction. Qualitatively, the brightness of the gas-solid reaction zone also indicated the temperature was well above 1000°C. Since Mg boils at 1088°C (at 1 atm vapor pressure), the temperature in the reaction zone was probably above this level. Furthermore, there was evidence of melting and flowing of MgF2, indicating a temperature above 1260°C may have been reached locally.

[0069] Run 3. The starting temperature of the crucible was further increased to 1020°C to minimize the formation of Mg2Si and to facilitate the consolidation of the Si product. The reaction took off 30 seconds after Mg chunks were dropped into crucible, and more violent bursts of dark powder and fire balls were observed. This lasted about 35 seconds. A flat, dense reaction product was also obtained at the bottom of the crucible, indicating that the temperature at the reaction zone clearly reached over 1260°C, the melting point of MgF2. The Mg was completely reacted based on the weight balance analysis. XRD analysis (FIG. 2C) showed the main reaction products were MgF2 and Si, with only a trace amount of Mg2Si left (note that XRD can detect phases as low as 1% by weight).

[0070] In another experiment, inventors observed that if the reaction product was heated in S1F4 at temperatures above 1200°C, a complete reaction could be achieved and no Mg2Si peaks were detected, which confirms our hypothesis.

EXAMPLE 2

Reaction Mechanism(s) and Discussion

[0071] When very stable SiF ₄ tetrahedral molecules (AG° =-1572.6 kJ/mol or -375

Kcal/mol) arrive at the surface of Mg metal, inventors expect a reaction driven by a very negative, free energy of formation of product MgF2 (AG°= -1071 kJ/mol or - 256 Kcal/mol) according to reaction (1), which should be very exothermic based on the thermochemical data presented in Table 1. At low temperatures, S1F4 may remain chemisorbed on the Mg surface. When the temperature is high enough to allow for the breakage of the first Si-F bond and formation of Mg-F species, a protective film seems to form on the Mg surface. Above the melting point of Mg (650°C), the reaction proceeds to a certain extent and a thicker yellow-to- brown film is formed. Based on the observations in this work and our earlier experiments with SiF4-Na system, in which this type of film was isolated and analyzed, inventors interpret this film to consist of Si and MgF2 byproduct. The Mg liquid likely wicks or diffuses across the Si- MgF2 film to meet the gas. At this stage of the reaction, the product film becomes wrinkled. In some cases, inventors observed an effect similar to lava bubbling across the product surface before the reaction took off and the dust product impeded direct observation of the reaction zone. [0072] Because of the exothermic nature of the reaction, heat released is partially absorbed by Mg and the rest of the reaction product. Consequently, the remaining liquid Mg pool expands and generates bulges and/or cracks along the reaction product film. These stressed areas provide the path for the exit of Mg liquid, which formed bulges, and eventually for Mg vapor that reacted with S1F4 (g) and released more heat so that the reactor temperature kept increasing. This allowed more Mg vapor to come out at higher pressure, and more paths were generated, giving rise to flames and a runaway reaction. Eventually, the depletion of Mg and growth of the product crust slowed the reaction.

[0073] Note that Mg solubility in MgF2 increases with temperature and therefore, the permeability of Mg through the product film increases with temperature. This is a likely cause of the bursts of reaction and formation of powders from the homogeneous gas-gas phase reaction. As the temperature increased further, the MgF2 melted. Mg has a solubility of about 0.3 to 0.6 mole% in molten MgF2, so Mg can diffuse through the liquid product layer to reach and react with the S1F4 generating flames. Eventually the pool of Mg (1) was depleted and left a hole inside the reaction product, which was visible in the reaction products, as shown in FIG. 3B. Eventually, the reaction decayed, the temporary high temperatures reached locally (close to but below the adiabatic temperature) dropped, and the pressure of S1F4 remained constant indicating the reaction had finished.

[0074] Because the reaction products reached local temperatures above 1420°C

(typically for a few seconds), the Si powders melted and coalesced (in the molten MgF2) and grew into larger droplets— several mm-sized metallic beads were observed in the reaction product. FIG. 6A shows a cross section of the reaction product in a graphite crucible. The starting temperature for this run was 1020°C. Several mm-sized Si beads (metallic and shinning) were clearly observable to be embedded in the MgF2 media (darker area in photo). This clearly demonstrates that the temperature at the reaction zone reached the melting point of Si (1412°C) so that Si micro-droplets could migrate through liquid MgF2 and coalesce into larger droplets, a consequence of its high surface tension. Large Si droplets are also visible on the wall of the graphite crucible. These could be a result of the violent reaction that may have splashed Mg (1) and MgF2 (1) onto the wall and the reaction taking place in situ thereafter. The product appears compact without voids, indicating that MgF2 was molten and flowed to fill all the pores.

[0075] Inventors conclude that at starting reactor temperatures (600°C to 800°C), Mg can break through or diffuse out and react with Si to form stable Mg2Si, which has a relatively low melting point, thus explaining why metallic Mg2Si beads were seen in Run 1. As the reactor temperature increased, Mg2Si was no longer the most stable phase, so Mg further diffused out to react with S1F4 to form more Si and MgF2.

[0076] Although a significant part of the silicon is originally in the form of micron- and submicron-sized powders (from microscopy and the powder obtained after leaching in aqueous acid environments), most of the final silicon ranged in size from 1 micron to 100 microns when the initial reactor temperature was 650°C, but ranged from 1 micron to 1 mm when the initial temperature was over 800°C.

EXAMPLE 3

Melt Separation of Si from MgF2

[0077] FIG. 6B shows the cross section of the product in the crucible after the melt- separation process. Most of the silicon powders coalesced and sunk to the bottom as large Si balls (> 1 mm in diameter). This was expected since molten Si has a density of 2.47 g/cm3 and MgF2 has a density of 2.34 g/cm3 at around 1420°C; Si has a higher surface tension. Note that the crucible was heated inside a RF field and, therefore, the smaller silicon particles could have been maintained and dispersed in the melt by electromagnetically induced force. In addition, those particles that were smaller than 10 microns were observed by high-magnification microscopy and identified by the grey color of the salt surrounding them. Inventors estimate that the total weight of the small-yet-uncoalesced particles was less than 1%.

[0078] In an inert gas, the surface tension of Si is 0.82 J/m ² , whereas that of the molten

MgF2 is only 0.22 J/m ² . As such, inventors expected that Si would coalesce surrounded by MgF2 to minimize the system energy. FIG. 6B shows the spherical frozen shape of Si (bright circles) surrounded by the darker MgF2. It was estimated that the surface tension of Si in molten MgF2 was approximately 0.31 J/cm ² , significantly smaller than that in inert air. Also, the meniscus of the molten MgF2 at the graphite crucible wall indicated that MgF2 wet the crucible walls. Since molten MgF2 does not wet pure graphite, inventors suspected (and later confirmed experimentally) that this wetting behavior change indicated the graphite wall changed in chemistry.

[0079] Inventors know that at this high temperature, equilibrium exist between S1F4 and Si according to:

S1F4 (g) + Si = S1F2 (g) (2)

At the wall, the disproportionation of S1F2 back to Si will produce a silicon carbide layer according to:

Si + C = SiC (3) Inventors think that a thin film of carbide was mainly responsible for the change in wetting behavior.

EXAMPLE 4

Purity of Melt Separated Silicon

[0080] The purity of the final silicon product was the main focus of this work since the goal was to produce solar-grade silicon. In particular, inventors were interested in the levels of residual Mg and those of dopant elements B and P. Table 2 shows the impurity distribution and levels in the Si product after melt separation, with 4N and 5N Mg as the starting reducing agent, respectively. The amount of Mg detected is probably due to the dissolution of Mg into molten silicon. The value of 2300 ppm is close enough to what FIG. 8 shows for the solubility of Mg in Si, which goes through a maximum (typical in Si) as the temperature drops from its melting point to about 1250°C where it reaches levels around 1000 of ppmw (0.1%). The value of 25,000 ppm is abnormally high, and it is most probably due to incomplete removing of the MgF2, but it could be also due to the presence of trace amounts of Mg2Si not detectable by X- ray diffraction. In any case, this high level of Mg can be reduced to acceptable levels by evaporation during the melting of Si or by reactive blowing with Ch or Ch on the silicon melt as is commonly practiced in the silicon industry.

[0081] The level of most other metallic impurities, particularly the transition metals, is relatively low in the Si product. Moreover, because these metallic impurities generally have large segregation coefficient between solid and liquid Si phases during crystal pulling, they are expected to accumulate in the liquid state, thus yielding solar-grade silicon with the pulled solid crystal.

[0082] By contrast, B and P have relatively low separation coefficients, and cannot be removed easily by solidification processes. Therefore, their levels in the as-produced Si product are critical and need to be at or below 1-ppm level. The Si product obtained using 5N Mg has very encouraging B and P levels, both are sub-ppm. Since P is much lower and only at 0.1 ppmw, this product should yield a P-type silicon after crystal growth. However, Si obtained using 4N Mg has borderline levels of B and P. This product will still be P type after crystal growth, but the cell efficiency obtained from this silicon may be lower than that with good solar-grade silicon. Since the original concentrations of B and P in the 4N and 5N Mg did not vary much (< 0.05 ppmw B and 0.6 ppmw P in 4N Mg and < 0.05 ppmw B and 0.13 ppmw P in 5N Mg), it is not clear why the concentrations of these two impurities are so different in the final Si product. The most probable source of external impurities was the graphite crucibles used. Nevertheless, the fact that solar-grade quality was clearly obtained in some cases is encouraging and indicates that this process could be used for solar cell manufacture.

Table 2. impurity ^' distribution and levels in Various M and silicon products.

[0083] While the preferred embodiments of the present invention have been illustrated in detail, it should be apparent that modifications and adaptions to those embodiments may occur to one skilled in the art without departing from the scope of the present invention.