US 20040238372 A1
Electrolyte for the manufacture or refining of silicon at high temperatures, particularly suited for the manufacture of high grade silicon. The electrolyte is mainly formed from a salt melt of CaCl2 and CaO. The invention further concerns a method for the manufacture of silicon in a salt melt at a high temperature, in which quartz with a low content of phosphorus and boron is subjected to electrolysis in such a melt, and a method for the refining of silicon where the silicon to be refined is used as an alloy element for the anode used in an electrolytic cell including the melt defined above.
1. Electrolyte for the manufacture of or refining of silicon at high temperatures, characterized in that it mainly comprises a salt melt of CaCl2 and CaO.
2. Electrolyte as claimed in
3. Electrolyte as claimed in
4. Method for the manufacture of silicon in a salt melt at high temperatures, characterized in that quartz with a low content of phosphorous and boron is subjected to electrolysis in a melt ofCaCl2 and CaO.
5. Method as claimed in
6. Method as claimed in
7. Method as claimed in
8. Method as claimed in
9. Method as claimed in
10. Method as claimed in
11. Method for the refining of silicon at a high temperature, characterized in that the silicon to be refined is used as an alloy element in the anode in an electrolytic cell comprising a melt of CaCl2 and CaO.
12. Method as claimed in
13. Method as claimed in
14. Method as claimed in
15. Method as claimed in
 The present invention relates to an electrolyte. The invention further relates to a method for the manufacture of or refining of silicon, by which the electrolyte is utilized.
 Silicon may in terms of quality be divided into three categories, silicon for metallurgical purposes, high-grade silicon for solar cell production (SoG-Si) and extremely high-grade silicon for electronic purposes. The different quality levels of silicon are manufactured by different methods and varying conditions. With respect to the SoG-Si for solar cells, this quality has predominantly been produced from scrap resulting from the manufacture of the even higher grade silicon for electronic purposes. As long as this source for solar cells is sufficiently large to cover the market need, the price has been at an acceptable level. This is partly due to the fact that the price tolerance of the silicon for electronic purposes is very high in relation to the silicon for use in solar cells.
 The demand for solar cell silicon is increasing, however, and already within 2001 access to raw material in the form of scrap from the electronics industry may be too small to cover demand.
 Electrochemical production of Si from quartz dissolved in a cryolite melt is described e.g. in PCT patent publication No. WO 95/33870. The cryolite melt has the advantageous property that it dissolves silicon dioxide well and that it is inexpensive. It is thus convenient for the manufacture of metallurgic grade silicon with a typical purity of 99.5-99.7%, provided there is no requirement for absence of particular kinds of impurities. This method has, however, severe drawbacks when it comes to the manufacture of high grade silicon. The cryolite melt is extremely corrosive, particularly at high temperatures, as a consequence of its fluoride content. Therefore the range of allowable electrode materials in such a melt is very limited. In practice only carbon electrodes have been able to be utilized for this purpose. Carbon electrodes are encumbered with certain disadvantages related to manufacture of very pure silicon as they tend to contaminate the melt, and thus the silicon produced therefrom, with traces of boron and phosphorous. These elements, which are found as trace elements in carbon, are not removable from the product by any known purification method, which renders silicon manufactured this way very difficult to use for the purposes of solar cells.
 In addition substantial amounts of the cryolite melt deposits with the product, and this contamination is also very difficult to remove from the silicon by any known refining or purification method.
 It is thus an object of the present invention to provide a method for the manufacture of solar cell quality silicon (SoG-Si), i.e. with a maximum allowable content of B and P in the magnitude of 1 ppm.
 It is a further object of the invention to provide a method of this type that is based on electrolysis and where contaminations included in the product are of a kind that are easily removable in a subsequent purification step.
 It is a still further object of the invention to provide a method of this type that is simple and inexpensive, so that the resulting product may be produced at a reasonable cost.
 These and other objects are achieved by means of the electrolyte and the method according to the invention.
 The invention thus concerns an electrolyte as defined by claim 1.
 The invention also concerns a method for the manufacture of silicon as defined by claim 4.
 Finally the invention concerns a method for the refinery of silicon as defined by claim 10.
 Preferred embodiments of the invention are disclosed by the dependent claims.
 According to phase diagrams a melt of CaCl2 will be able to dissolve SiO2 in an amount sufficient for the salt to serve as an electrolyte in a process of the kind mentioned above, and more precisely in the magnitude of 5%. It was, however, discovered during the work leading to the present invention, that said phase diagrams are incorrect. Pure CaCl2 dissolves SiO2 only to a very limited degree, namely in the magnitude of 0.1%. CaCl2 is highly hygroscopic, and a possible source of error for the known phase diagram(s) may be that the measurements have been conducted with a not completely pure CaCl2, which means that oxygen in the form of water may have been included in the melt.
 It has been found as part of the present invention that addition of comparatively modest amounts of CaO to CaCl2 provides a melt that dissolves SiO2 to an extent that is fully satisfactory. Already at a 5% content of CaO the solubility is in the range of 3-4%, which is more than sufficient for the purpose. Acceptable solubilities were found already at concentrations lower than this. The underlying chemistry is not fully understood for this melt more than for other melts, but there is reason to believe that the SiO2 combines with CaO in an unknown stoichiometry. Such compounds complicate the deposition of metal from the melt. For the process according to the invention, this has, however, proved in practice to not be a problem.
 A possible disadvantage of chloride based melts for electrolysis of dissolved oxides, is that significant amounts of chlorine gas develop at the anode. Thermodynamically oxygen should form before chloride, but kinetical relations will in practice decide the relative amounts of these gases. Therefore it is convenient to use an anode material that promotes the development of oxygen and inhibits the development of chloride gas. Carbon is an example of an anode material that is well suited for this purpose, but as mentioned above it has the disadvantage that it (usually) contains phosphorous and boron that easily transfer to the product. Some carbon sources may, however, be well suited as anode material for the method according to the invention.
 In addition to carbon with a particularly low content of phosphorous and boron, modified nickel ferrite, doped tin oxide or an oxidation resistant metal alloy chosen among the metals, tungsten, silver, gold, platinum and palladium, may be used as anode material. Generally it is convenient to use inert, i.e. non-consumable, anodes.
 As cathode material, e.g. silicon or alloys containing silicon are well suited. Silicon alone has, however, an inconveniently low electrical conductivity , which is why some amount of metal, e.g. calcium, is preferably added. The amount of calcium in such an alloy may vary typically from a few per cent to e.g. 30%. The amount need not exceed a few per cent, typically 5% or less. Other materials that by experience may be included in such alloys, are tungsten, nickel and iron. Iron or tungsten have proved to be particularly advantageous, as well as alloys including silicon and said metals, as these metals/alloys do not react with metals that deposit in the process, particularly calcium.
 It is important that contaminations that occur in the silicon deposited on the cathode may easily be removed by simple methods to a comparatively high degree of purity. The process may—with less purification—be competitive also with production processes of silicon of a lower quality than the solar cell quality.
 According to one aspect, the invention may be utilized for refining silicon of an arbitrary degree of purity. According to this embodiment of the invention an alloy of the “impure” silicon together with another metal is used as anode in an electrolyte of said type. Particularly preferred is an alloy of copper and silicon. When such a process is run, nearly pure silicon will migrate through the electrolyte from the anode to the cathode and be deposited on the latter electrode as part of the production process. The contaminations will almost exclusively remain in the anode.
 Two simple examples were conducted without any particular optimization of the electrodes, in order to get an indication of the electrolyte's suitability for the process. The test conditions are listed in table 1.
 Silicon manufactured during the test runs were analysed with respect to phosphorus and boron according to test method (Ar-ICP-AE). The measurements were controlled against a pure electronic grade silicon. The results are shown in table 2.
 The results of tests 1a and 1b indicates a certain content of phosphorous, which mainly origins from the raw materials. Typically the CaO would contain some phosphorous. With respect to boron, the very low levels found in the test runs 1a and 1b indicate that the method works according to the expectations with platinum anode and CaSi cathode.
 Test 2 shows an undesired high content of boron with respect to the use of the material for electronic purposes. This content mainly originates from the electrodes.
 In general these tests show that the electrolyte and the method as such work according to expectations, but that the choice of electrodes needs to be adapted and optimized according to the desired purity of the end product.