EP3368477A1

EP3368477A1 - Method for the enrichment and separation of silicon crystals from a molten metal for the purification of silicon

Info

Publication number: EP3368477A1
Application number: EP16816738.5A
Authority: EP
Inventors: György 55 % DR. KAPTAY; József 30 % SZABÓ; Gergely Bálint 15 % TÓTH
Original assignee: Bay Zoltan Alkalmazott Kutatasi Kozhasznu Nonprofi Kft Logisztikai Es Gyartastechnikai Intezet
Current assignee: Bay Zoltan Alkalmazott Kutatasi Kozhasznu Nonprofi Kft Logisztikai Es Gyartastechnikai Intezet
Priority date: 2015-10-29
Filing date: 2016-10-25
Publication date: 2018-09-05
Also published as: HUP1500509A1; WO2017072655A1

Abstract

The invention relates to a method for the enrichment and separation of silicon crystals from a molten metal for the purification of silicon, for which contaminated (metallurgical-grade) Si is dissolved in a suitable molten metal, preferably molten aluminium at a relatively high temperature, then the silicon crystals precipitated, recrystallized in a purer form upon cooling are in part or in full separated from the molten metal in a suitable manner. Finally, the surface of the purified Si crystals is cleaned mechanically and with a suitable solvent, while the remaining Al-Si alloy can be reused. From the purified and enriched silicon, by known post-treatment steps, solar grade silicon can be obtained at a lower cost.

Description

Method for the enrichment and separation of silicon

crystals from a molten metal for the purification of silicon

The invention relates to a method for the enrichment and separation of silicon (Si) crystals from a molten metal for the purification of silicon, for which contaminated (metallurgical-grade) Si is dissolved in a suitable molten metal, preferably molten aluminium at a relatively high temperature, then the silicon crystals precipitated, recrystallized in a purer form upon cooling are in part or in full separated from the molten metal in a suitable manner. Finally, the surface of the purified Si crystals is cleaned mechanically and with a suitable solvent, while the remaining Al-Si alloy can be reused. From the purified and enriched silicon, by known post-treatment steps, solar grade silicon can be obtained at a lower cost.

Silicon (Si) is produced in large quantities and cheaply by carbothermic reduction by the metallurgical industry (so called MG-Si = metallurgical grade Si), but only with a purity of 99 wt% (so called 2 nines). The electronics industry requires Si with a purity of 9-10 nines (so called SeG-Si = semiconductor grade silicon). The emerging Si-based solar cell production requires Si with a purity of 6-7 nines (so called SoG-Si = solar grade silicon) (B. Ceccaroli, O. Lohne: Solar grade silicon feedstock; Handbook of Photovoltaic Science and Engineering, A. Luque, S. Hegedus (eds) (John Wiley & Sons), 2003, 153-204; K. Morita, T. Miki: Thermodynamics of solar-grade- silicon refining; Intermetallics, 2003, vol. 11, 1111- 111; M.D. Johnston, L.T. Khajavi, M. Li, S. Sokhanvaran, M. Barati: High-temperature refining of metallurgical-grade silicon: a review; JOM, 2012, vol. 64, 935-945).

Historically, metallurgical grade Si production was first followed by the development of electronic grade Si production, producing SeG-Si in many steps and therefore at a high cost. For example, the production of polycrystalline Si with the Siemens method includes the following steps: the conversion of MG-Si into a volatile Si compound, e.g. S1HCI₃, the purification thereof by fractional distillation, from the decomposition of the Si-compound SeG-Si is obtained, and the by-products are recycled. The most common method of the electronics industry for the production of monocrystalline Si is the Czochralski method (B. Ceccaroli, O. Lohne: Solar grade silicon feedstock; Handbook of Photovoltaic Science and Engineering, A. Luque, S. Hegedus (eds) (John Wiley & Sons), 2003, 153-204 B; W.C. O'Mara ed.: Handbook of semiconductor silicon technology; Noyes Publications (1990), Park Ridge, New Jersey, U.S.A.). When the demand for solar grade Si arose, first it was satisfied from SeG-Si overproduction, or from the unsuccessful batches, production lots of SeG-Si (with a purity of less than 9 nines), and this is still typical (B. Ceccaroli, O. Lohne: Solar grade silicon feedstock; Handbook of Photovoltaic Science and Engineering, A. Luque, S. Hegedus (eds) (John Wiley & Sons), 2003, 153-204 B). This method, however, makes the solar cells and the electricity produced by them too expensive, and limits the volume of production ( . Morita, T. Miki: Thermodynamics of solar- grade- silicon refining; Intermetallics, 2003, vol. 11, 1111-1117; B. Ceccaroli, O. Lohne: Solar grade silicon feedstock; Handbook of Photovoltaic Science and Engineering, A. Luque, S. Hegedus (eds) (John Wiley & Sons), 2003, 153-204). The same is the case if a simplified version of the SeG- Si production is considered, by omitting the last steps.

The above necessitate a new method for the production of SoG-Si, or an intermediate product, starting from MG-Si in as few a number of steps and at as low a cost as possible, allowing an economically more viable production of SoG-Si quality than the currently known solutions.

There are attempts to produce pure Si by electrochemical methods, in which silicon dioxide is dissolved in molten fluoride, and during the electrochemical process high-purity silicon is deposited from the melt on a cathode (A.F.B. Braga, S.P. Moreira_, P.R. Zampieri, J.M.G. Bacchin, P.R. Mei: New processes for the production of solar-grade polycrystalline silicon: A review; Solar Energy Materials & Solar Cells, 2008, vol. 92, 418^424). In other works other Si-compounds (e.g. K₂SiF₆) are used as an Si-source, or a mixture of chlorides and oxides is applied during the electrolysis (Cai Jing, Luo Xue-tao, Lu Cheng-hao, G.M. Haarberg, A. Laurent, O.E. Kongstein, Wang Shu-Ian: Purification of metallurgical grade silicon by electrorefining in molten salts; Trans, nonferrous Met. Soc. China, 2012, vol. 22, 3103-3017; G.M. Haarberg, L. Famiyeh, A.M. Martinez, K.S. Osen: Electrodeposition of silicon from fluoride melts; Electrochimica Acta, 2013, vol. 100, 226-228; E. Ergul, I. Karakaya, M. Erdogan: Electrochemical decomposition of Si0₂ pellets to form silicon in molten salts; J Alloys Compounds, 2011, vol. 509, 899-903; Su-Chul Lee, Jin-Mok Hur, Chung-Seok Seo: Silicon powder production by electrochemical reduction of Si0₂ in molten LiCl-Li₂0. J Ind Eng Chem 2008, vol. 14, 651-654). The temperature used is typically between 550 and 1000 °C depending on the particular system, however these methods are unproductive. Although Si with a purity of 99.999% could be produced by molten salt electrolysis under laboratory conditions (J.M. Olson, K.L. Carleton: A semi-permeable anode for silicon refining. J Electrochem. Soc. 1981, vol.128, 2698-2699), the components containing toxic F pose a serious threat not only to people, but to the environment and the equipment as well, especially on an industrial scale.

In practice today, for the time being the production of solar grade polycrystalline silicon by a chemical process, basically based on the Siemens method, is more successful. In this silicon of sufficient purity is produced by the decomposition of trichloro-silane on a semi- industrial scale, however, this method is not productive enough either. As an alternative, some are experimenting with significantly more energy-efficient metallurgical solutions, however, these are still at an experimental level (A.F.B. Braga, S.P. Moreira_, P.R. Zampieri, J.M.G. Bacchin, P.R. Mei: New processes for the production of solar-grade polycrystalline silicon: A review; Solar Energy Materials & Solar Cells, 2008, vol. 92, 418-424; . Tomono, Y. Okamura, H. Furuya, M. Satoh, S. Miyamoto, R. Komatsu, M. Nakayama: Selective hydrobromination of metallurgical-grade silicon in a flow reactor system. J. Mater. Sci. 2012, vol. 47, 3227-3232).

According to the literature electron beam melting in vacuum (M.D. ohnston, L.T. Khajavi, M. Li, S. Sokhanvaran, M. Barati: High-temperature refining of metallurgical-grade silicon: a review; IOM, 2012, vol. 64, 935-945) is investment intensive and is not even productive. Plasma melting is similarly disadvantageous and the handling of the large amounts of reactive gases poses a further problem (D. Lynch: Winning the global race for solar silicon; IOM 2009, vol. 61, 41-48). SoG-Si purity can be produced by zone melting as well, however, this method is also unable to meet the volume of market demand, as disclosed in International Publication No. WO 2008/026931. The productive solutions (e.g. slag melting in vacuum, where the role of the slag is to collect the impurities of Si), however, typically are unable to achieve SoG-Si purity in one step, as the partition coefficients of boron (B) and phosphorus (P) are unfavourable (M.D. lohnston, L.T. Khajavi, M. Li, S. Sokhanvaran, M. Barati: High- temperature refining of metallurgical-grade silicon: a review; IOM, 2012, vol. 64, 935-945; Wu li-jun, Ma Wen-hui, Yang Bin, Dai Yong-nian, K. Morita: Boron removal from metallurgical grade silicon by oxidizing refining. Trans. Nonferrous Met. Soc. China, 2009, vol. 19, 463-467). Therefore, other methods are studied for the removal of these impurities. Slag melting offers a productive solution, but its application needs to be carefully considered, because the high working temperature (T=1500 °C) requires the use of investment-intensive materials. The working temperature of the slag phase can be influenced by adding salts, for example the addition of CaF₂ to a CaO or CaO-Si0₂ system, or Na₂C0₃ to a slag forming Si0₂ has been previously studied. Furthermore, the production of SoG-Si from metallurgical grade silicon requires the use of slag/silicon in a weight ratio of 5: 1, as according to the literature no molten slag could provide the appropriate partition coefficient for the removal of the elements boron and phosphorus at a lower weight ratio. And that calls into question its economic viability (M.D. Johnston, M. Barati: Distribution of impurity elements in slag- silicon equilibria for oxidative refining of metallurgical silicon for solar cell applications. Solar Energy Materials & Solar Cells 94 (2010) 2085-2090; M.D. Johnston, L.T. Khajavi, M. Li, S. Sokhanvaran, M. Barati: High-temperature refining of metallurgical-grade silicon: a review; JOM, 2012, vol. 64, 935-945).

A promising group of methods is solvent refining, in which MG-Si is dissolved in a suitable solvent from which Si can be recrystallized to a purer form (M.D. Johnston, L.T. Khajavi, M. Li, S. Sokhanvaran, M. Barati: High-temperature refining of metallurgical-grade silicon: a review; JOM, 2012, vol. 64, 935-945). If the solvent is selected properly, then during this process most of the elements contaminating the original MG-Si remain in the solvent, resulting in the purification of Si. The solvents are typically molten metals, from which aluminium (Al), iron (Fe), copper (Cu), tin (Sn), etc. are suitable for the process (J. Gumaste, B. Mohanty, R. Galgali, U. Syamaprasad, B. Nayak, S. Singh, P. Jena, Solvent refining of metallurgical grade silicon. Sol. Energy Mater. 1987, vol. 16, 289-296; S. Esfahani, M. Barati: Purification of metallurgical silicon using iron as an impurity getter. Part I: Growth and separation of Si. Metals Mater. Int. vol. 17 (2011), pp. 823-829; J.M. Juneja, T.K. Mukherjee. A study of the purification of metallurgical grade silicon. Hydrometallurgy, 1986, vol. 16, 69-75; E. Bonnier, H. Pastor, J. Driole. Sur une preparation de silicum de haute purets. Metallurgie, 1965, vol. 7, 299-307). The degree of purification of Si depends on the properties of the solvent-molten metal. In this respect one of the most promising solvents is molten Al. According to the known method, MG-Si is dissolved in molten Al at a higher temperature, then the solubility of Si is reduced by decreasing the temperature in a controlled manner. Thus an Al-Si based melt supersaturated with Si is formed, from which, upon cooling, Si crystals precipitate that are significantly purer than the starting MG-Si. Most of the impurities, especially B and P, the removal of which causes problems for other methods, prefer the Al-based melt. With this method SoG-Si purity could be achieved under laboratory conditions (T. Yoshikawa, K. Morita: An evolving method for solar-grade silicon production: solvent refining; JOM, 2012, vol. 64, 946-951). The real problem is the separation of the solid Si crystals from the remaining Al-Si melt. According to the current solutions (T. Yoshikawa, K. Morita: Refining of silicon during its solidification from a Si-Al melt. J Crystal Growth, 2009, vol. 311, 776-779; K. Morita, T. Yoshikawa: Thermodynamic evaluation of new metallurgical refining processes for SOG-silicon production; Trans. Nonferrous Met. Soc. China, 2011, vol. 21, 685-690; T. Yoshikawa, . Morita: An evolving method for solar-grade silicon production: solvent refining; JOM, 2012, vol. 64, 946-951; M.D. Johnston, L.T. Khajavi, M. Li, S. Sokhanvaran, M. Barati: High-temperature refining of metallurgical-grade silicon: a review; JOM, 2012, vol. 64, 935-945) the purified Si can be separated from the metal matrix by acid leaching, however, this process generates significant amounts of acid waste solutions containing light and heavy metals.

One reason for the separation problem of an aluminium-silicone system is that the density of the Al-Si based melt and the Si crystals is practically the same, thus the sedimentation techniques are not efficient. This can be solved either by changing the density of the molten Al with additives, or by placing the two-phase Al-Si melt/Si crystal system in a suitable external force field (e.g. a centrifugal, or electromagnetic force field, an induction furnace) to enrich the Si crystals in one part of the melt (T. Yoshikawa, K. Morita: Refining of silicon during its solidification from a Si-Al melt. J Crystal Growth, 2009, vol. 311, 776- 779; K. Morita, T. Yoshikawa: Thermodynamic evaluation of new metallurgical refining processes for SOG-silicon production; Trans. Nonferrous Met. Soc. China, 2011, vol. 21, 685-690).

Another reason for the difficulty of separation is that, apart from the enrichment described above, the Si crystals dispersed in the Al-Si based melt remain inside the Al-Si melt, they do not separate spontaneously. There is also no known method suitable for transferring the Si crystals from the Al-Si based melt to the gas phase or an intermediate phase, where they would become readily accessible (M.D. Johnston, L.T. Khajavi, M. Li, S. Sokhanvaran, M. Barati: High-temperature refining of metallurgical-grade silicon: a review; JOM, 2012, vol. 64, 935-945; T. Yoshikawa, K. Morita: An evolving method for solar-grade silicon production: solvent refining; JOM, 2012, vol. 64, 946-951).

Instead, today the practice is to cool/crystallize all the Al-Si melt, then to break it, and to selectively leach the aluminium rich in the impurities of Si from the Si crystals. This, however, makes the process significantly more complicated and expensive, as it requires the use of process equipment and energy, as well as concentrated, environment polluting chemicals (e.g. aqua regia, hydrogen fluoride), moreover the heat energy used for melting the Al is lost, and any other energy (mainly electrical) used for the production of the Al is also lost, as the Al changes from a metallic state into an aqueous ionic solution. The same is the case when another molten metal is used as the solvent instead of the Al. There is no known method in the literature for using together a molten metal and a molten salt that are suitable and compatible with each other for the purification of Si, where the role of the molten metal is to purify the Si, while the role of the molten salt is to separate the purified Si crystals from the molten metal in such a way that the Si crystals transferred to the molten salt can be readily removed/washed from it at the end of the process. There is also no known method in which the precipitated Si crystals are filtered or floated from the Al-Si melt.

The aim of the invention is to eliminate the disadvantages of the known solution using a molten metal solvent, and to develop a new, improved method capable of efficiently separating the purified Si crystals precipitated from the molten metal upon cooling from the molten metal in such a way that the purified Si crystals become readily accessible at the end of the process, and thus solar grade silicon can be produced at a low cost.

The idea of the invention is in part based on the recognition that the above aim can be achieved by using a molten salt as an intermediate phase. Therefore, the key element of this method is that a salt of suitably selected composition (or in a certain ratio slag) is added to a molten metal or molten metal alloy suitable for the purification of Si, which melts at the temperature of the process and functions as a melt in such a way that it pulls the Si crystals precipitated from the molten metal upon cooling through the molten metal/molten salt interface into the inside of the molten salt at least in part, and separates them from the inside of the molten metal.

The invention is furthermore based on the recognition that if relatively large Si crystals are formed and the density of the molten salt is lower than the density of the Si crystals and the molten metal, then in a static case the separation of the Si crystals from the molten metal is not perfect, as although the larger portion of each Si crystal reaching the surface passes to the molten salt, however, for reasons of gravity their smaller portion adheres to the molten metal. Then preferably the system is cooled, the salt is removed mechanically and with a suitable solvent, and the Si crystals protruding from the surface of the solidified metal are recovered by dissolving the surface layer of the solidified metal with a suitable solvent. In this case a significant part of the metal used for purification can be reused, and the specific solvent consumption per unit of Si crystals recovered is lower compared to the case when all the metal needs to be dissolved, moreover the specific metal consumption is also reduced. The invention is furthermore based on the recognition that the previous method can be made more efficient if the molten metal/Si/molten salt system is mixed/dispersed, and by this the passage of the Si crystals from the molten metal to the molten salt is accelerated, which is an advantage especially when a larger volume of molten metal is used.

The invention is furthermore based on the recognition that in the case of mixing and dispersion the efficiency of the separation of the Si crystals and the molten metal can be increased if the density of the molten salt is lower than the density of the molten metal and the Si crystals, as with the reduction of the mixing speed the larger molten metal droplets sediment more quickly than the smaller Si crystals, and the molten salt temporarily rich in Si crystals can be poured from the surface of the molten metal before the sedimentation of the Si crystals, from which after solidification the Si crystals can be readily removed mechanically and with a suitable solvent, and the material and heat energy of the remaining molten metal can be reused in full.

The invention is furthermore based on the recognition that the Si crystals precipitated from the molten metal can be floated from the system by introducing small bubbles of an inert gas into the lower part of the molten metal, on the surface of which the Si crystals gather and together with the bubbles rise to the top of the molten metal, then to the top of the molten salt, from where the purified Si crystals can be skimmed off in the form of a foam or scum. Similarly to the previous case, the Si crystals can be readily removed from the solidified salt in part mechanically, and in part with a suitable solvent, and the material and heat energy of the remaining molten metal can be reused.

The invention is furthermore based on the recognition that the method can be made even more efficient by selecting the density of the molten metal and the molten salt in such a way that the densities of the phases are characterized by the following relations: the density of the Si is the lowest, the density of the molten salt is medium, and the density of the molten metal is the highest. In this case the buoyant force will also help the separation of the Si and the molten metal (molten metal at the bottom, molten salt at the top), the Si crystals precipitated upon cooling are first driven to the top of the molten metal by the buoyant force, then they are pulled through the molten metal/molten salt interface by the molten salt, then they are driven to the top of that as well, and finally the purified Si crystals can be skimmed off from the top of the molten salt. Similarly to the previous case, the Si crystals can be readily removed from the solidified salt in part mechanically, and in part with a suitable solvent, and the material and heat energy of the remaining molten metal can be reused. The invention is furthermore based on the recognition that the efficiency of the separation of the Si crystals precipitated from the molten metal upon cooling can be increased and the specific metal consumption can be reduced by mechanically filtering out the Si crystals with a suitable ceramic filter.

The invention relates to a method for the enrichment and separation of silicon crystals from a molten metal, in the first step of which the contaminated Si is dissolved in a molten metal of suitable composition at a relatively high temperature, preferably in a furnace, in the second step upon controlled cooling a part of the dissolved Si crystallizes from the molten metal, in the third step by means of a suitable molten salt (that may be present in the first and second steps as well, but then its presence is not necessary) the Si crystals are transferred into the molten salt, in the fourth step the Si crystals are possibly concentrated in as small a volume of the molten salt as possible (this is not a process step, it is required only for economic reasons), in the fifth step the molten salt rich in Si crystals is removed (skimmed off or poured) from the molten metal and the furnace, and finally the molten salt rich in Si crystals is cooled, and after the solidification of the salt the adherent salt is cleaned from the surface of the Si crystals mechanically and with a suitable solution.

According to the invention the contaminated Si is dissolved in such an amount and at such a relatively high temperature in such a metal alloy/molten metal, firstly, which is stable in the molten state under the conditions of the process, secondly, which allows the complete dissolution of the added Si, thirdly, from which primary Si crystals precipitate upon cooling, fourthly, which is preferred by the elements contaminating the original Si over the Si crystals and thus allows the purification of the Si from these impurities, fifthly, which helps the floatation of the Si crystals to the molten metal/molten salt interface and from there the passage of the Si crystals into the molten salt phase, sixthly, which after the removal of the Si crystals, in the molten state, can be used for the production of other products, therefore both the material and the thermal energy can be reused.

According to the invention such a molten salt is selected for the implementation of the method, firstly, which is stable in the molten state under the conditions of the process, secondly, which does not react with either the molten metal selected above or the Si crystals, thirdly, which deoxidizes the molten metal/molten salt interface, fourthly, which allows the spontaneous passage of the Si crystals from the molten metal to the molten salt through the molten metal/molten salt interface, fifthly, which helps the best possible separation of the Si crystals from the molten metal, sixthly, which can be separated/washed from the purified Si crystals with minimal cost and environmental impact, seventhly, the remaining part of which can be recycled into the process the most times possible, eightly, the replacement of which has the least possible environmental impact.

According to the invention the crucible or the inner wall of the furnace (and all other equipment required during the process, such as the mixer or the equipment suitable for measuring the salt rich in Si) is made of (or covered with) such a material, firstly, which is stable in the solid state under the conditions of the process, secondly, which does not react chemically with either the molten metal or the molten salt and/or slag, thirdly, which does not contaminate the Si crystals, fourthly, which is sufficiently durable to allow the economically viable implementation of the Si purification, fifthly, the replacement of which has the least possible environmental impact.

According to the invention the process is performed under such a gas atmosphere, firstly, which does not react chemically with either the crucible, the molten metal, or the molten salt, secondly, which does not contaminate the Si crystals, thirdly, the use of which has the least possible financial and environmental impact.

According to the invention such a solvent is used for the secondary cleaning (following mechanical cleaning) of the Si crystals from the solidified salt and/or metal adhering to them, firstly, which well and quickly dissolves the salt and/or metal used in the process, secondly, which does not contaminate the Si crystals, thirdly, the use of which has the least possible financial and environmental impact. It is advisable to use sonication to accelerate the dissolution.

According to the invention a ceramic filter made of such a material and with such a pore volume is used for the concentration of the Si crystals precipitated from the Al-Si melt, which does not contaminate either the Al-Si melt or the Si crystals, which is suitable for filtering out most of the precipitated Si crystals, which can be readily separated from the suspension rich in Si crystals and therefore can be reused, and the replacement of which has a minimal environmental impact.

According to the invention at the start of the process the contaminated (from primary metallurgical processes, and/or waste) silicon is dissolved at a relatively high temperature (Tl) in a relatively high concentration (CI) in a molten metal. During the process the system is cooled at a suitable cooling rate, and the process is completed at a lower temperature (T2) and a lower Si concentration (C2) of the molten metal. The initial concentration CI is determined by the value of Tl and the composition of the primary Si-free molten metal. Similarly, the final concentration C2 is determined by the value of T2 and the composition of the Si-free molten metal. The greater the (C1-C2) difference, the more Si crystals per unit of primary Si-free molten metal can be produced, but the lower is the degree of Si purification, therefore the (C1-C2) difference has an optimum value. A greater (C1-C2) difference typically requires a greater (T1-T2) difference, although the relationship between the two is not linear. According to the above, therefore the (T1-T2) difference also has an optimum value. Increasing Tl is limited in part by the melting point of Si, in part by economic considerations, and in part by the stability of the molten metal, the molten salt, the ceramic filter and the crucible. Increasing Tl beyond the limit decreases the degree of Si purification. Reducing T2 increases the degree of Si purification, but it is limited by the fact that each particular composition of the primary Si-free alloy has an eutectic temperature T2* at which, in addition to the Si crystals, other crystals also precipitate from the molten metal. As it is advisable to avoid that in order to keep the Si crystals pure, it is advisable to keep the value of T2 above T2* during the process. In addition, T2 shall not be lower than the liquidus temperature of the molten salt (if a molten salt is also used) - therefore optimally the melting point of the molten salt shall be adjusted (through its composition) to a temperature below T2*, or at least below temperature T2.

Accordingly, temperature Tl shall be between the temperature of the metal-Si eutectic closest to pure Si and the melting point of Si, while temperature T2 shall be lower than temperature Tl, but higher than the eutectic temperature T2* closest to pure Si.

According to the invention the amount of purified Si crystals that can be extracted per unit of a primary Si-free molten metal of a particular composition at particular parameters Tl, T2, C 1 and C2 has a theoretical maximum. The Si extraction efficiency shall be measured by comparison to this theoretical maximum. In principle, molten salts of different compositions can provide different extraction efficiencies. In addition, by increasing the specific amount of molten salt added per unit of molten metal, the extraction efficiency can be increased, but this function reaches saturation: excessively increasing the specific amount of molten salt does not increase significantly the extraction efficiency, but increases the incidental costs, therefore the specific amount of molten salt used per unit of molten metal has an optimum value.

According to the invention the higher the cooling rate used during the process, the more productive the process. However, increasing the cooling rate can lead to increasing molten metal inclusions in the growing Si crystals, contaminating the formed Si crystals. Therefore, the cooling rate used during the process also has an optimum value. According to the invention the purer the used molten metal and molten salt (that is the lower the initial concentration of elements critical from the point of view of Si purification in them), the higher the degree of Si purification. However, this relationship is not linear, therefore the use of a purer and purer molten metal and molten salt increases less and less the Si purification efficiency. Moreover, the cost of a purer molten metal and molten salt is higher, therefore the purity of the used molten metal and molten salt also has an optimum value.

According to the invention the metal and salt (or their melts) remaining after the process can be reused in another Si purification cycle. By increasing the number of reuse cycles the degree of Si purification gradually decreases, but the cost of the process also decreases, therefore the number of reuse cycles also has an optimum value. According to the invention the contamination level of the molten metal and molten salt no longer reusable for the purification of Si is so low that they have a good chance of being usable in other processes. The remaining, slightly contaminated Al-Si melt is suitable, for example, for the production of Al-Si castings.

The process embodiment of the invention depends on the density ratio of the phases. In the method according to the invention, in terms of density ratios, three types of processes are used:

- the common characteristic of type 1 processes is that the density of the molten salt is lower than the density of the molten metal and the Si crystals (the density ratio of Si and the molten metal is insignificant). Then for reasons of gravity the molten metal is at the bottom and the molten salt is at the top, and the Si crystals transferred into the molten salt are, at least in part, pulled back by gravity into the molten metal over time.

The type 1 processes have three versions.

In the type 1A process only a molten salt is added on top of the molten metal, typically no other action (either mixing or bubbling) is used. Then the Si crystals precipitated from the molten metal are enriched at the molten metal/molten salt interface. After cooling the system, the Si crystals enriched at this interface are recovered using a small amount of solvent, with a relatively low loss of solvent and metal. Then the remaining metal (with the Si remaining in it) can be reused. The yield of the process can be improved by slow mixing.

In the type IB process the molten metal and molten salt (after the precipitation of the Si, at temperature T2) are mixed/dispersed with a suitable mixer at a suitably high speed, by this the passage of the Si crystals to the molten salt is accelerated and implemented, then by suitably reducing the mixing speed (and raising the mixer) the molten metal is left to sediment, but before the sedimentation of the Si crystals the molten salt rich in Si crystals is poured from the surface of the molten metal. After solidification the salt is removed from the Si crystals mechanically and with a suitable solvent. Then all the molten metal remains and can be reused.

In the type 1C process after reaching temperature T2 an inert gas is introduced into the molten metal from below, with the smallest possible bubble size, and the Si crystals are floated. The Si crystals will be enriched on the surface of the bubbles in such a way that the average density of the bubbles and the Si crystals adhering to them is lower than the density of the molten salt. Then due to the bubbles the Si crystals move to the top of the molten metal, then to the top of the molten salt, and from there they can be skimmed off. The solidified Si crystals can be separated from the salt mechanically and with a suitable solvent. The remaining molten metal and molten salt can be reused.

- the common characteristic of type 2 processes is that from the phases the density of the Si crystals is the lowest, the density of the molten salt is medium, while the density of the molten metal is the highest. Then the molten metal is at the bottom and the molten salt is at the top. The Si crystals precipitated in the molten metal upon cooling are driven to the molten metal/molten salt interface by the buoyant force, where the suitably selected molten salt pulls them from the molten metal, then the Si crystals are driven to the top of the molten salt by the buoyant force, and from there they can be removed together with the upper part of the molten salt. The solidified Si crystals can be separated from the salt mechanically and with a suitable solvent. The remaining molten metal and molten salt can be reused.

- in type 3 processes a ceramic filter is used instead of the molten salt, by filtering the Al-Si (liquid)/Si (solid) suspension, the precipitated Si crystals are significantly enriched in the part blocked by the ceramic filter, while the melt passing through the filter contains practically no Si crystals. The latter part can be reused, while the purified Si crystals can be leached from the part blocked by the filter in an economically viable manner.

Preferably, the type 1A, IB, 1C, 2 and 3 processes described above can be combined with one another.

The material of the Si-free molten metal can be Al, Ca-, Cu-, Fe-, In-, Mg-, Ni-, Sb-, Sn- Zn, preferably, aluminium is used. In the simplest case Al with a purity of e.g. 99.7 wt%, produced by primary metallurgical processes can be used, but a higher degree of Si purification can be achieved by using purer Al. The advantage of dissolving Si in aluminium is that no intermetallic phase is formed, and it can be managed at a relatively low temperature. If type 1 processes are used with molten Al, the molten salt is preferably a mixture of sodium chloride (NaCl), potassium chloride (KC1) and sodium fluoride (NaF), but alkali metal halides of other composition and their mixture can also be used, such as for example a mixture of potassium chloride (KC1), potassium fluoride (KF) and potassium hexasilicofluoride (K^SiFg). Then the density of the molten salt is lower than the density of the molten metal, while the density of the Si crystals is similar to the density of the molten metal. The material of the crucible (and if needed, the mixer) is preferably corundum, although crucibles made of other materials can also be used. The material of the gas (and if needed, the bubbles) in the simplest case is preferably air, but for higher purity it is advisable to use inert gases (e.g. argon). Preferably water, or an aqueous solution (preferably with dissolved aluminium chloride: A1C1₃) can be used for dissolving the salt, while a concentrated acid, preferably hydrochloric acid (10 to 37 wt% HC1), sulphuric acid (40 to 98 wt% H₂S0₄), nitric acid (20 to 63 wt HNO₃), hydrofluoric acid (10 to 48 wt% HF), or a mixture of these is used for partially dissolving the Al. By selecting the preferable materials, the partial purification of Si can be achieved at the lowest possible material cost.

According to the type 1A process the system is cooled gradually, the purified Si crystals can be removed from the interface of the solidified Al and salt mechanically, and by dissolving a small amount of salt and Al. In this case most of the Al-Si remains and can be reused either in this or in other processes.

According to the type IB process mixing is applied at temperature T2. This can be external (electromagnetic) mixing, but a mechanical mixer can also be used. In this case the volume of the molten salt shall be larger than the volume of the molten metal, as only then can the dispersion of the molten metal droplets in the molten salt be achieved. Mixing shall be performed by lowering the mixer below the molten metal/molten salt level, at a high mixing speed. Then molten metal droplets are dispersed in the molten salt, as a result of which the specific surface area of the molten metal increases, through which the Si crystals quickly and fully pass from the molten metal into the molten salt. As a result of reducing the mixing speed and raising the mixer above the molten metal/molten salt level, the approximately spherical molten metal droplets sediment relatively quickly, while the smaller and non-spherical Si crystals sediment only more slowly, thus after the sedimentation of the molten Al the molten salt containing the Si crystals can be poured from the system. The solidified salt can be removed from the Si crystals mechanically and by leaching with an aqueous solution, the efficiency of the latter can be increased by sonication. Then the material and heat of the remaining Al-Si melt can be reused in full either in this or in other processes. According to the type 1C process small gas bubbles are introduced at the bottom of the molten Al at temperature T2. Due to the buoyant force the bubbles start to rise, together with the Si crystals adhering to their surface. Finally the bubbles raise (float) the Si crystals to the top of the molten metal and the molten salt, from where they can be skimmed off. The solidified salt can be removed from the Si crystals mechanically and by leaching with an aqueous solution, the efficiency of the latter can be increased by sonication. Then the material and heat of the remaining Al-Si melt can be reused in full either in this or in other processes.

In the type 2 process firstly the density of the Al-based melt shall be increased, secondly a molten salt of higher density shall be selected. Preferably, a copper (Cu) additive is used to increase the density of the Al-based melt, and thus the Si-free molten metal becomes an Al-Cu alloy. Sodium iodide (Nal) is used as the main component of the molten salt, in which a small amount of sodium fluoride (NaF), and cryolite (Na3AlFe) is dissolved. In this case the purified Si crystals are driven to the top of the molten metal and the molten salt by the buoyant force, and from there they can be removed together with a small amount molten salt, then the salt can be cleaned from the Si crystals mechanically and with a suitable aqueous solution. Then the material and heat of the remaining Al-Cu-Si melt can be reused in full either in this or in other processes.

In the type 3 process the Si crystals precipitated at temperature T2 are in part or in full filtered from the melt rich in Al by means of a ceramic filter. The ceramic filter can be the same as the one used nowadays by aluminium foundries for the filtration/purification of the melt. Its material is preferably aluminium oxide (corundum), or an appropriately surface treated version thereof, with a pore size between 10 micrometers and 1 mm. Filtration is assisted by gravity, as well as by any pressure difference created on the two sides of the filter or the melt. The process can be implemented in a batch or a continuous manner.

The method according to the invention and the products produced by the method will now be described in detail with reference to the following figures:

Figure 1 is a binary phase diagram of Al-Si, showing temperatures Tl and T2 (and T2*), and concentrations C 1 and C2,

Figure 2 is a schematic representation of the two steps of the type 1A process,

Figure 3 is a schematic representation of the five steps of the type IB process,

Figure 4 is a schematic representation of the three steps of the type 1C process,

Figure 5 is a schematic representation of the three steps the of the type 2 process,

Figure 6 is a schematic representation of the type 3 process, Figure 7 is an image of the cut Al-28Si experimental product according to Example 1, Figure 8 is an image of the end product according to the type 1A process,

Figure 9 is an image of the end product produced by slow mixing during cooling according to the type 1A process,

Figures 10a, 10b and 10c show the result of the type IB process, where 10a shows: Al-rich spheres bordered by Si crystals obtained in the salt matrix during the emulsification process, 10b shows: Si crystals obtained after the dissolution of the salt, and 10c shows: a piece of the Al-Si alloy stuck in the crucible, with many Si crystals protruding from its surface.

Figures 11a and l ib show the end product of the result of the type 1C process, where 11a shows: a specimen prepared by floatation, with Si crystals protruding from its lateral surface, and 1 lb shows: a part separated from it, rich in Si crystals.

Figure 12 shows a polished cross-section of the specimen prepared by means of the buoyant force according to the type 2 process,

Figures 13a and 13b show the result of the type 3 process, where 13a shows: an image of the specimen stuck to the ceramic filter and 13b shows: a polished cross-section of the specimen passing through the filter.

Description of the Experiments

The starting material for the experiments is a hypereutectic Al-Si alloy, which is either produced in a preliminary step, or the required amount of MG-Si is dissolved in the molten Al above the liquidus temperature in the first step of the experiment, and afterwards the produced starting material is treated according to the particular additional process. The material of the crucible is corundum. The starting alloy used for the experiments was Al-28Si, with a temperature Tl of 800 °C, and a temperature T2 of about 600 °C, making it a low temperature process among the Si purification processes.

Figure 1 shows the Al-Si equilibrium phase diagram [ASM-93]: a schematic representation of the principle of the purification of Si by means of molten Al. At a high temperature (Tl) the molten Al can dissolve a high concentration (CI) of Si. Upon cooling to a lower temperature (T2) the solubility of Si in the Al-Si melt decreases to C2, while purified Si crystals precipitate from the Al-Si melt. Temperature T2 shall be higher than the eutectic temperature T2* = 577 °C, to ensure that only purified Si crystals precipitate from the melt. This principle applies to the type IB and 1C processes (shown in Figures 3 and 4). To the type 2 process (shown in Figure 5) the Al-Cu-Si phase diagram applies, based on a similar principle, but with different details. In the type 1A process the system is cooled to room temperature TO (not indicated in the Figure), and although this results in the precipitation of almost the total Si content, only a part thereof can be dissolved from the metal/salt interface, because the other part crystallizes eutectically in the total volume.

According to the phase diagram shown in Figure 1, a temperature of 850 °C is sufficient for the preparation of the base alloy, in the melt of which upon cooling from Tl = 800 °C primary Si crystals precipitate, which are dispersed in the remaining Al-Si melt. This process is the purification process of the contaminated Si, as most of the impurities of Si remain in the Al-rich melt. The primary crystallization process of Si lasts until the eutectic temperature is reached (Teut = T2* = 577 + 1 °C, below that Al-Si eutectic crystallizes). In the maximum useful crystallization temperature range from Tl to T2*, in principle, maximum 17.6 wt% primary Si crystals can crystallize from the Al-28Si melt. In practice, it is not advisable to cool the system to temperature T2*, as the precipitation of the Al-Si eutectic would contaminate the crystallizing primary silicon, and near the eutectic temperature it is difficult to control the additional operations to be applied. Therefore temperature T2 = 600 °C was selected as the end of the cooling phase; then between 800 and 600 °C, in principle, 16.3 wt% Si crystallizes from the Al-28Si base melt. The cooling rate was about 1 °C/minute. The details discussed above are the same in the following examples, they only differ in the way in which the purified Si crystals are enriched/separated from the remaining Al-salt melt.

The method according to the invention is further described by way of the following examples:

Negative example 1: distribution of the recrystallized Si in the Al-28Si alloy

In this experiment everything was performed as described in the "Description of the Experiments". A polished cross-section was prepared from the middle part of the specimen, which is shown in Figure 7. The figure shows well the relatively large, dark grey, homogeneously distributed, long Si crystals grown to a size of 5-10 mm. The distribution of the crystals within the specimen is homogeneous, therefore the Si crystals could be recovered from the solidified alloy only by dissolving the total Al mass thereof. That process, however, is not economically viable and has a high environmental impact.

By comparison, in the following positive examples the Si crystals are more preferably enriched on the surface of the molten Al, or in the molten salt or the filter. Example 2: enrichment of the recrystallized silicon by means of a molten salt

Figure 2 shows a schematic representation of the type 1A process.

In the first step of this process, in a corundum crucible (a), at the bottom an Al-Si melt (d), above that a NaCl-KCl-NaF melt (c), and above that argon gas (b) is kept at a (high) temperature Tl. In the second step the temperature is gradually reduced from temperature Tl to room temperature TO. A part of the Si crystals (e) is at the metal/salt interface, and from there they can be recovered at a low specific solvent consumption. The other part of the Si crystals (f) remains dispersed in the aluminium.

In the experiment conducted according to the type 1A process, the Al-28Si alloy and a NaCl-KCl mixture of equimolar composition containing 10 wt% of NaF was melted in an aluminium oxide crucible (the density of the molten salt was lower than the density of the molten aluminium and the solid silicon). The system was maintained at a temperature of 850 °C for about 30 minutes, then it was crystallized by cooling it below the liquidus point of the molten metal. The molten metal containing the Si crystals was not mixed, furthermore the system was solidified in the crucible, without pouring it out. On the basis of the cross-section of the experimental product shown in Figure 8 it can be concluded that the Si crystals are enriched mainly along the lower interface of the alloy (in the direction of heat removal), at the molten metal/molten salt interface (the molten metal is separated from the crucible by a thin molten salt layer). A comparison of Figures 7 and 8 reveals that the molten salt had an influence on the shape of the Si crystals as well; Figure 8 shows well that the dark grey Si crystals are less needle-like and are enriched at the interface. From the end product shown in Figure 8 the Si crystals can be recovered by dissolving a smaller amount of Al-rich alloy then from the end product shown in Figure 7.

This experiment was repeated with mixing at a low mixing speed of 150 to 200 rpm (the mixer was made of graphite). The result of this experiment is shown in Figure 9. Slow mixing homogenizes the temperature field, and the distribution of the Si crystals becomes more homogeneous along the molten metal/molten salt interface compared to the product shown in Figure 8. Moreover, a part of the Si crystals was now in the molten salt. Thus slow mixing is preferable, as it reduces the specific amount of metal that needs to be dissolved for the purpose of recovering the Si crystals. Table 1 shows the chemical composition of the materials used and the Si crystals recovered in this experiment. Table 1: The results of ICP composition measurements made on the foundry aluminium, the metallurgical silicon and the purified Si crystals of the example shown in Figure 9

From Table 1 it can be seen that the Si is contaminated with Al (due to the metal) and Na (due to the salt). With respect to all the impurities the degree of purification was more than 5-fold, but Al accounted for more than 70 % of the impurities, and it can be removed by the directional solidification of Si (apart from the aluminium the degree of purification was 17- fold). Within this, the boron (B) content of silicon was reduced to a half and its phosphorus (P) content to a third. 1. Example 3 : enrichment of the recrystallized silicon by emulsification

This example corresponds to the type IB process, which is shown in Figure 3.

In the first step of the type IB process, in a corundum crucible (a), at the bottom an Al-Si melt (d), above that a NaCl-KCl-NaF melt (c), and above that argon gas (b) is kept at a (high) temperature Tl. In the second step the temperature is gradually reduced from temperature Tl to temperature T2, then purified Si crystals precipitate from the Al-Si melt (e). In the third step at temperature T2 this system is mixed with a mixer (f) in such a way that the mixer is lowered and a high mixing speed is used, then the Al-Si melt is dispersed in the molten salt, while the Si crystals pass into the molten salt. In the fourth step at temperature T2 the mixer is raised and the mixing speed is reduced, as a result of this the Al-Si melt sediments at the bottom, the Si crystals dispersed in the molten salt sediment more slowly. Finally, in the fifth step at temperature T2 the mixer (f) is removed and most of the molten salt containing the Si crystals is poured from the molten metal. Thus a mixture of salt/Si crystals cooled to room temperature TO is obtained in a separate crucible (g), from which the Si crystals can be recovered mechanically and by aqueous leaching. The remaining Al-Si melt of temperature T2 can be reused.

The Al-28Si melt + molten salt system was dispersed with a paddle mixer having a graphite mixing head, at a mixing speed of 800 rpm. Then both the molten salt and the molten metal were poured out, the molten metal rolled out from the crucible in the form of smaller or larger spherical droplets, the droplets were embedded in the salt matrix, as shown by the photo in Figure 10a. A part of the Si crystals recovered after dissolving the salt is shown in Figure 10b. Some Al-rich alloy remained in the crucible as well, with many Si crystals enriched on its surface, as shown in Figure 10c. Therefore the type IB process can separate the Si crystals from the remaining Al-Si melt more efficiently compared to the type 1A process.

Example 4: enrichment of the recrystallized silicon by floatation

This example corresponds to the type 1C process, which is shown in Figure 4.

In the first step of the process, in a corundum crucible (a), at the bottom an Al-Si melt (d), above that a NaCl-KCl-NaF melt (c), and above that argon gas (b) is kept at a (high) temperature Tl. In the second step the temperature is gradually reduced from temperature Tl to temperature T2, then purified Si crystals precipitate from the Al-Si melt (e). In the third step at temperature T2 an inert gas (g) (e.g. argon) is introduced at the bottom of Al-Si melt through a pipe (f). The bubbles rise and carry with them (float) the Si crystals, which can be skimmed off from the top of the molten metal or molten salt. Finally, a mixture of salt/Si crystals cooled to room temperature TO is obtained, from which the Si crystals can be recovered mechanically and by aqueous leaching (not shown in the Figure). The remaining Al-Si melt and NaCl-KCl-NaF melt of temperature T2 can be reused.

The essence of the process is that an inert gas is introduced into a hypereutectic Al-Si melt from below, whereby the gas bubbles carry with them the Si crystals precipitated during cooling, helping their passage from the molten metal alloy to the molten salt above it. The argon gas was bubbled through the system through a stainless steel pipe with an internal diameter of 2 mm, ending in a graphite capillary tube, over the temperature range from 710 to 610 °C for a total of 120 minutes, at a flow rate of about 25 - 35 cm3/min. The gas bubbles rising in the melt carried most of the Si crystals to the molten metal/molten salt interface, and a smaller part of them into the molten salt. The rising argon bubbles pushed aside the floating Si crystals, thus they are mainly not on the top of the alloy, but along its side walls, as shown in Figures 1 la and 1 lb. Figure 11a shows the Si crystals protruding form the lateral surface of the specimen, while Figure l ib shows a part separated from it, rich in Si crystals. The Si crystals formed an almost solid coating on the surface of the piece of alloy, which can be removed from there mechanically.

Example 5: enrichment of the recrystallized silicon by means of the buoyant force

This example corresponds to the type 2 process, which is shown in Figure 5.

In the first step of the process, in a corundum crucible (a), at the bottom an Al-Cu-Si melt (d), above that a NaI-NaF-Na3AlF₆ melt (c), and above that argon gas (b) is kept at a (high) temperature Tl. In the second step the temperature is gradually reduced from temperature Tl to temperature T2, then purified Si crystals precipitate from the Al-Cu-Si melt (e). As time passes, the Si crystals are gradually driven to the top of the molten metal/molten salt by the buoyant force. Finally, a mixture of salt/Si crystals cooled to room temperature TO is obtained, from which the Si crystals can be recovered mechanically and by aqueous leaching. The remaining Al-Si melt and NaCl-KCl-NaF melt of temperature T2 can be reused.

The essence of separation by means of the buoyant force is that the density relations of the solid Si/molten metal and molten salt system are adjusted by preferably changing the compositions in such a way that the density of the solid Si crystals is the lowest, it is exceeded by the density of the molten salt, and it is exceeded by the density of the molten metal. Then the Si crystals precipitated in the molten metal are driven upwards in the molten metal, and in part pass into the molten salt. The density of the molten metal was adjusted by alloying with copper, in fact an alloy of a composition of Al (60 wt%) + Si (20 wt ) + Cu (20 wt%) was used. The calculated density of the so produced alloy at 800°C is around 2.82 g/cm . The composition of the salt mixture: Nal (89.6 wt ) + NaF (7.9 wt%), Na₃AlF₆ (2.5 wt ), its calculated density at 800 °C is about 2.64 g/cm³. The density of the solid Si crystals at 800 °C is about 2.33 g/cm³. The result is shown in Figure 12. A large number of Si crystals cover the surface of the alloy having an increased density (not only its top, but its sides as well). Many of the surfacing Si crystals extend from the alloy into the molten salt.

Example 6: enrichment of the recrystallized silicon by filtration

This example corresponds to the type 3 process. Figure 6 shows the experimental arrangement for filtration. The crucible used in the process is divided into two parts by a filter. The material of the filter is preferably aluminium oxide (corundum), or an appropriately surface treated version of it, with a pore size between 10 μιη and 1 mm. In practice, the cooled starting material, containing the purified Si crystals and the remaining Al-Si melt, is placed in the upper part. Due to gravity, or any pressure difference created below and above the melt (below it reduced by a vacuum pump and/or above it increased by an inert gas), the Al-Si melt poor in Si passes through the pores of the filter, and can be poured/led out and reused. The purified Si crystals are highly enriched in the filter, or on the top thereof, from where the Si crystals purified from impurities can be leached in an economically viable manner. This process can be implemented in a continuous manner by using a vertical temperature gradient.

The Si crystals dispersed in an Al-Si melt (without a molten salt) were filtered from the melt with a corundum filter, using only gravity as the driving force. The filter and the alloy stuck in it are shown in Figure 13a; the Si content of this alloy was 52 ± 4 wt%. A polished cross-section of the Al-Si alloy passing through the filter is shown in Figure 13b; the Si content of this alloy was 17 + 2 wt%.

Claims

1) A method for the enrichment and separation of silicon crystals from a molten metal for the purification of silicon, characterized in that the contaminated silicon is dissolved in a molten metal at a temperature Tl between the temperature of the metal-Si eutectic closest to pure Si and the melting point of Si, then by reducing the temperature the Si crystals precipitating from the metal-Si melt are enriched in the molten metal, and/or are in part or in full separated from the molten metal by means of a molten salt, or a ceramic filter, or by gas floatation, and finally the Si crystals are recovered by solvent treatment.

2) The method according to claim 1, characterized in that the molten metal is selected from a group of metals consisting of Al, Ca, Cu, Fe, In, Mg, Ni, Sb, Sn, Zn.

3) The method according to claim 1, characterized in that aluminium (Al) is used as the molten metal, while a molten salt having a lower density than both the Al and the Si is used as the molten salt.

4) The method according to claims 1-3, characterized in that a mixture of sodium chloride (NaCl), potassium chloride (KCl) and sodium fluoride (NaF), or other alkali halides, such as a mixture of potassium chloride (KCl), potassium fluoride (KF) and potassium hexasilicofluoride (K₂SiF₆) are used as the molten salt.

5) The method according to claims 1-4, characterized in that the contaminated silicon is dissolved in the molten aluminium in the concentration range of 20 to 50 wt%, at a temperature between 700 and 950 °C, then the temperature is reduced to 580 °C, then the chemically purified Si crystals precipitated from the Al-Si melt are in part or in full separated from the molten metal by means of a molten salt, or a ceramic filter, or by gas floatation, and finally the Si crystals are recovered by solvent treatment.

6) The method according to claims 1-5, characterized in that the molten metal and molten salt system is cooled from temperature Tl to room temperature TO, then the molten salt is separated from the molten metal mechanically, the excess salt is removed with an aqueous solution, then a part of the purified Si crystals is leached from the surface layer of the metal with a low specific amount of acid.

7) The method according to claim 6, characterized in that the acid used is hydrochloric acid of a concentration of 10 to 37 wt%, sulphuric acid of a concentration of 40 to 98 wt%, nitric acid of a concentration of 20 to 63 wt%, hydrofluoric acid of a concentration of 10 to 48 , or a mixture of these.

8) The method according to claims 5-6, characterized in that the molten metal is mixed slowly, at 10 to 200 rpm, with a low speed mixer.

9) The method according to claim 1, characterized in that the molten metal and molten salt system is cooled to a temperature T2, where temperature T2 is lower than temperature Tl, but higher than the eutectic temperature closest to pure Si; it is dispersed with a lowered preheated mixer at a speed above 200 rpm, then after the passage of the purified Si crystals into the molten salt, the mixer is raised to a higher position and the speed is reduced below 200 rpm in such a way that the Al-Si melt is left to sediment, but before the sedimentation of the slower sedimenting Si crystals the Si-salt mixture is poured from the molten metal, finally the Si crystals are removed from the salt at room temperature TO by leaching with an aqueous solution.

10) The method according to claim 1, characterized in that the molten metal and molten salt system is cooled to a temperature T2, where temperature T2 is lower than temperature Tl, but higher than the eutectic temperature closest to pure Si, and bubbles of an inert gas are introduced into the molten metal, thereby most of the Si crystals are floated to the top of the molten metal and/or the molten salt, then the Si crystals are removed from the salt by leaching with an aqueous solution, and from the surface of the metal-Si alloy by acid leaching.

11) The method according to claim 1, characterized in that the remaining molten metal and molten salt can be reused. 12) The method according to claims 1-4, characterized in that an aluminium alloy with a density increased by a metal having a higher density than Al, preferably copper (Cu), is used as the starting molten metal (Al-Si-Cu), while the density of the molten salt is adjusted to be between the density values of the Al-Si-Cu melt and the Si crystals, preferably by using a mixture of sodium iodide (Nal), sodium fluoride (NaF), and cryolite (Na₃AlFe), then the system is cooled from temperature Tl to temperature T2, then the temperature is maintained for 30 to 180 minutes, then after the spontaneous rise of the Si crystals to the top of the molten metal/molten salt the purified Si crystals are removed from the salt by leaching with an aqueous solution, and from the surface of the Al-Si alloy by acid leaching.

13) The method according to claim 1, characterized in that the metal-Si melt system is cooled from temperature Tl to temperature T2, then it is filtered through a ceramic filter made of aluminium oxide with a pore size of 10 to 1000 microns by means of gravity and/or vacuum, or by a pressure difference, and the Si is extracted from the material of high Si content blocked by the filter.

14) The method according to claims 1-10, characterized in that, by a treatment using Al with a purity of 99.7 wt%, the total impurities content of metallurgical grade silicon, 70 % of which is made up of aluminium remaining from the Al-based alloy and can be removed by the directional solidification of the Si, is reduced to less than a fifth, furthermore the boron (B) content of the metallurgical grade silicon is reduced to a half, and its phosphorus (P) content to a third.

15) The method according to claims 1-14, characterized in that the different types of processes are freely combined with each other.