Electro-reduction Process

The present invention relates to a method for removing oxygen from metal oxides by electrolysis in a fused salt to produce elements or alloys of high purity, and to the products of such processes. The invention relates in particular to the production of high purity silicon and other group IV elements from silica and other group IV oxides.

Electro-winning of elements from their ores has become well-established over the last two hundred years. In a. recent development, a new process has come to light in which a metal oxide is fabricated as an electrode with an electrolytic collector and is cathodically reduced to the metal in the presence of a fused electrolyte and applied direct current voltage. Such an "electro-deoxidation" or "electro-reduction" process, hereinafter generically referred to as an "EDO" process, is described in WO 99/64638 to Fray, and is generally- applicable to oxides of metals, metalloids and mixtures of oxides. WO 99/64638, the disclosure of which is incorporated herein by reference, is especially concerned with the extraction of titanium from its oxide, although it teacϊies that the process may also be applied to decompose electrolγtically oxides of U, Mg, Al, Zr, Hf, Nb, Mo and rare earths such as Nd and Sm, as well as oxides of metalloids such as germanium and silicon.

In WO 99/64638, it teaches that it is important for the potential of the cathode to be maintained and controlled potentiostatically so that only oxygen removal occurs and not the more usual decomposition of the electrolyte. Accoording to Fray, on the application of a voltage via a power source, a current will not start to flow until balancing reactions occur at both the anode and the cathode. At the cathode there are two possible reactions, the discharge of the cation from the salt or the removal of oxygen. The latter reaction occurs at a more positive potential than the discharge of the metal cation and, therefore, will occur first. However, for the reaction to

proceed, it is necessary for the oxygen to diffuse to the surface of the oxide and, depending on the temperature, this can be a slow process. For best results it is, therefore, important that the reaction is carried out at a suitably elevated temperature, and that the cell potential is controlled, to prevent the potential from rising above a threshold and chlorine evolution from the electrolyte occurring as a competing reaction to the oxygen being removed from the oxide. The benefit from reduction by an EDO process is that high yields can be achieved with less energy compared to other extraction processes.

Although an EDO process may also be used to remove dissolved (e.g. interstitial) oxygen or other similar elements such as sulphur, nitrogen and carbon, it does not of course address the problem of dissolved metallic impurities.

Elements that require a high purity level in respect of metallic impurities cannot be formed directly by an EDO process as described in WO 99/64638, due to the likely cathodic reduction and incorporation of said metallic impurities into the desired element. For elements- requiring a high state of purity in respect of metallic impurities, for example silicon in the electronics industry, this is a severe limitation.

High purity silicon is used widely in the electronics industry, for example in the production of semiconductors, integrated circuits and photovoltaic cells. The required purity level depends upon the ultimate application. Typically, the purity of photovoltaic grade silicon (>99.99999%) is lower than semiconductor ^• or integrated circuit grade silicon (>99.99999999%) . Critical impurities requiring removal include the first row transition elements such as Cr, Fe, Mn, Ni as well as Al, B, P, ^' Zr, Nb, Mo, Ta and W.

At present silicon is produced industrially by heating silicon dioxide with a reducing agent. The low cost, metallurgical grade silicon obtained from the extraction process is then further purified, for example by the Siemens

process. The purification stage adds considerably to the cost of the final silicon product. Further processing stages may

, also be required to incorporate dopants into the silicon. At present there is no specific production process for photovoltaic grade silicon and solar cells are produced from expensive semiconductor grade silicon made by the Siemens process. The availability of moderately pure silicon for solar cell fabrication is currently limited by this bottleneck, and hence the market in solar cells is waiting to exploit high volumes of photovoltaic grade silicon made by an alternative, lower cost process.

Accordingly, a first aspect of the present invention provides a method for producing a metal, metalloid or alloy from a metal oxide, metalloid oxide or mixture of oxides of alloying elements forming a working electrode, by electrolysis in a fused salt or mixture of salts in the presence of a counter-electrode, the method comprising applying to the working electrode a plurality of electrolysis cycles, wherein at least one part of the electrolysis cycle is conducted at a cathodic potential relative to the counter-electrode under conditions so as to remove oxygen from the metal, metalloid or mixed oxide; and wherein at least one other part of the electrolysis cycle is conducted at an anodic voltage relative to the counter-electrode such that ionisation of metallic impurities from the oxide rather than removal of oxygen occurs and said impurities are released into the electrolyte.

Since the electrodes in this EDO cell are caused to reverse their polarity, the oxide electrode and secondary electrode are hereinafter referred to as the 'working electrode' and ^λ counter-electrode' , respectively.

Either the working electrode is formed from the oxide or it is in electrical contact with the oxide, for example by threading oxide tiles onto a stainless steel rod. The oxide typically has a porous or micro-porous structure produced, for example, by sintering oxide particles of suitable size. The micro-porosity decreases the length of diffusion paths for ions

and other mobile species within the oxide, thus improving the diffusion efficiency of such species and aiding their removal from the electrode.

We have discovered that a high purity metal, metalloid or alloy can be produced directly from an oxide or mixed oxide by- applying a plurality of electrolysis cycles to a working electrode fabricated from, or contiguous with, said oxide or mixed oxide compound, wherein each electrolysis cycle comprises at least one cathodic electro-reduction phase in accordance with a typical EDO method and at least one anodic electro- oxidation phase for the ionisation and dissolution of metallic impurities. Thus the present invention provides a combined extraction and purification method for an element from its oxide with significant advantages over prior art systems. During an EDO process, the oxide lattice rearranges into the crystal structure of the associated metal, metalloid or alloy. The phase change accompanying the recrystallisation process presents ideal conditions for the out-diffusion of other species present in the oxide. Thus, we have discovered that in an EDO process the growing surface is especially- susceptible to the gettering out of metallic impurities.

The present invention has particular application to the production of silicon, especially photovoltaic grade silicon from silicon dioxide. However, it will be understood that the method of the current invention is applicable to any element that can be extracted from its oxide by an EDO process and such elements include, for example, V, U, Mg, Al, Zr, Hf, Nb, Mo and rare earths such as Nd and Sm, as well as metalloids such as Ge and Si. For any particular element, the type, rate and magnitude of electrode perturbation is chosen to achieve the reduction of the working electrode to the target element, whilst removing particular impurities according to their standard redox potentials in the particular electrolyte. It will be understood that, when two or more elements with differing redox potentials exist in the same electrode material, it is possible to oxidise

and dissolve one into an electrolyte without affecting the other. In other words, if the redox potential of the required material is greater than the redox potential (s) of the impurity or impurities, preferential dissolution of the impurity or impurities can take place. Thus by cycling the electrode potential it is possible to remove unwanted elements from the surface of an electrode containing the desired element. The present invention may be applied to any element or salt able to undergo this process and includes Group IV metalloids such as Si and Ge, which elements have redox potentials higher than typical metallic impurities such as, for example, transition elements.

The present method also applies to the extraction of high purity alloys from mixed oxides of alloying elements. The cathodic voltage is set at a suitable level for an EDO process to take place, which level is sufficiently negative such that oxygen is removed from the cathode, but below the potential at which chlorine is evolved from the electrolyte. For electro-deoxidation using a carbon anode and calcium chloride electrolyte at 700-1000 ⁰ C, the cathodic voltage suitably lies in the range 1.5V to 3.2V.

Suitably the anodic voltage applied to the working electrode is in the range 0.0 to 2.0V, referenced to a standard calomel electrode, and is sufficient to oxidise metallic impurities at the surface of the working electrode without re- oxidising the element undergoing extraction. Preferably the anodic voltage applied to the working electrode is in the range 0.1 to 1.5V SCE, which voltage will remove a range of transition metal impurities, including, first row elements such as, for example, Cu, Pe, ,Cr ₇ Mn, and Ni. In practice, a reference electrode may not be used and the cell voltage may be selected having regard to the known redox potentials of target metallic impurities.

As mentioned above, the use of cyclic polarisation of the working electrode causes the target element to form under cathodic polarisation and the metallic impurities to be removed

from the reduced surface of the element under anodic polarisation. The cathodic polarisation is set so that the element at the growing surface is momentarily formed. At this point the electrode is made to undergo a change in polarisation by altering the applied potential. The removal of cathodic protection enables noble elements such as copper and iron, and other transition metals, to be ionised and thus be able to diffuse away from the growing surface of the silicon. By momentarily cycling the potential in this way, re-adsorption and deposition of said metallic impurities can be significantly curtailed . The re-adsorption of impurities into the working electrode will be governed, in part, by the concentration set up between the background concentration of the impurity in the electrolyte and those contained in the Helmholtz sphere at the electrode surface.

The cathodic voltage will usually be applied to the metal oxide, metalloid oxide or mixed oxide electrode for up to 100 seconds in each electrolysis cycle so as to reduce oxides at the electrode surface, but may be applied for only 10 to 100 milliseconds. The cathodic deoxidation stage electrolytically reduces metallic impurities present at the working electrode as oxides or other salts, in addition to the target extractant, where their redox potentials are less cathodic.

The anodic voltage may be applied to the metal oxide, metalloid, oxide or mixed oxide electrode for up to 100 seconds, or for as little as 10 to 100 milliseconds, and may be of sufficient duration to deplete the double layer surrounding the working electrode of cationic impurities.

The plurality of electrolysis cycles conveniently comprises an alternating waveform, for example a sinusoidal, sawtooth or square waveform. Alternatively, the plurality of electrolysis cycles comprises a pulsed or stepped waveform. The plurality of electrolysis cycles preferably includes the condition of open circuit, which condition may aid the diffusion of impurities from the electrode bulk to the electrode surface.

The plurality of electrolysis cycles is applied until the required purity of the metal, metalloid or alloy product is obtained. Th.e cycles may be of identical duration, or may vary, for example, getting progressively longer or shorter in duration. Usually the cycle lengths will get progressively longer with, time because, as the bulk concentration of impurities ±n the working electrode decreases, the impurity concentration at the surface of the working electrode takes longer to build up. Thus, the cycle rate may be directly proportional to the cell current and may be high initially. The relative proportions of the anodic and cathodic phases may also be constant or vary, and the anodic phases may become shorter with successive cycles as fewer impurities require removal.

The plurality of electrolysis cycles may be followed by a non-cyclic stage where the working electrode is maintained at a cathodic potential until substantially all of the oxygen has been removed from the oxide, for example, where a selected minimum impurity concentration threshold has been reached.

For best results the temperature of the electrolysis cell should be maintained at 700-1000°C throughout the plurality of electrolysis cycles, thus aiding diffusion processes.

Ideally, the electrolyte comprises a fused salt, or mixture of salts, which is more stable than the equivalent salt of the element that is being extracted. Furthermore, the salt, or mixture of salts should have a wide difference between its melting and boiling point and a high temperature of operation to improve the diffusion of species in the electrode and elsewhere. Suitable electrolytes include the chlorides of barium, calcium, caesium, lithium, strontium and yttrium. CaCl ₂ is particularly suitable as an electrolyte, which electrolyte decomposes at approximately 3.3V at 700-1000 ⁰ C.

The effective removal of impurities from the working electrode _Ls dependent upon two out-diffusion, or gettering, processes, specifically diffusion of metals and metal oxides through the electrode bulk and diffusion of ionic impurities away from the electrode surface into the Helmholtz double

layer. Once occupying this phase they are subject to removal along with other reduced species associated with the de- oxidation process. In the case of a calcium chloride electrolyte, the reduced species will be elemental calcium produced by the reduction of calcium oxide present in the electrolyte as an impmrity, which species getters oxygen and other impurities from the electrode surface. Once reoxidised through electrode cycling, these are removed into the electrolyte. Both the <de-oxidation and gettering processes are governed by the respective diffusion rates of the various species at the process temperature, and the concentration gradients set up within the diffusion sink, in this case the electrolyte. Electrode surface area, process temperature and the purity of the electrolyte all contribute to the efficiency of the out-diffusion processes, and electrolyte purity, in particular, provides a means of controlling concentration gradients and creating conditions suitable for gettering unwanted impurities from the electrode.

Accordingly, in. parallel with the plurality of electrolysis cycles, steps are preferably taken, on a continuous, semi-contzLnuous or batch basis, to maintain the purity of the electroDLyte in respect of target impurities. The combination of electrode cycling and electrolyte purification can achieve high purity levels and preferably, purification of • the electrolyte occurs for at least the first two thirds, and preferably, the whole reaction time. One method of doing this is to replace the electrolyte within the cell with pre-purified electrolyte from an erxternal supply. More advantageously, the electrolyte is, purified, using an additional purification means, which means may be Located within the electrolysis cell or externally. In one preferred aspect of the current invention, the electrolyte is recycled through a purification means such as an ion exchange medium, for example a high melting point ion exchange resin such as Zeolite, located adjacent to the electrolysis cell. Alternatively, a purification means such as a secondary plating electrode may be located within the cell

and used to deposit metallic impurities from the electrolyte in situ. A sacrificial electroactive material such as zinc or sodium may also be used to replace impurities in the electrolyte, which element will suitably be more electropositive than the impurities that it replaces, and have no impact (e.g. electrical impact in the case of Si) on the target element. A suitable electroactive material is Zirconia loaded with sodium.

Other electrolyte purification means will suggest themselves to one skilled in the art. The electrolyte purification means may optional.Iy comprise a filter to remove deposited salts from the electrolyte.

An EDO process is usually conducted using an anode comprising carbon. However, the use of a carbon counter electrode in the present invention may introduce disadvantageous additional impiαrities to the system, in which case an inert counter electrode may be used. Such a counter electrode may comprise a pure metal, metal oxide or oxide- coated metal and more preferably the counter electrode comprises "Ebonex" , a conducting titanium oxide, or rhenium oxide. The cell potential will need to be adjusted to account for the overpotential at an inert anode and suitable materials may be added to the electrolyte to improve the cell efficiency, as would be familiar to the skilled person. Where necessary, the catriode material may be doped with elements such as phosphorous or boron, by the addition of suitable materials to the electrolyte. The dopant-containing material is added to the electirolyte in excess concentration so that, for example, any phosphorous or boron removed from the working electrode during the anodic phase is more than replaced.

It will be understood that the purification method of the present invention proceeds via. surface mechanisms. Thus, in addition to the micro-porosity of the oxide, the working electrode is preferably fabricated with a high surface area. Suitable electrode forms comprise flat, tiled, folded or

cellular structures, for example pellets, foams, or single or multiple tiles threaded onto a steel conductor. Advantageously, a thin layer of the oxide or mixed oxide is positioned on a conducting substrate such as a metal sheet and conveyed through and out of the reaction cell to speed throughput of material. This may be in the form of . a conducting belt, from which elemental material is subsequently removed, or a substrate coated with the element for permanency.

Preferably the working electrode is fabricated in the near net shape of the required metal, metalloid or alloy product. Such electrodes are described j_n WO 99/64638.

A further aspect of the current invention provides a shaped component for a photovoltaic cell obtained by the process described above. A final aspect of the current invention provides a method of doping a semiconductor.

The following Examples illustrate the invention. Example 1

A sample of compressed silicon dioxide on a steel cathode was electro-reduced to silicon in an electrochemical cell at 800° C using a fused impure CaCl ₂ electrolyte and metal anode by the application of 3 volts dc. A second sample of silicon was produced using high purity CaCl ₂ electrolyte. The first sample exhibited a higher level of bulk impurities than the second.

Example 2

A sample of compressed silicon dioxide on a steel cathode was electro-reduced to silicon in an electrochemical cell at 800 ⁰ C using a fused pure CaCl ₂ electrolyte and metal anode by the application of 3 volts dc. The applied electrode potential was periodically switched off and ttie sample left at open circuit. The impurities were found to have out diffused and accumulated at the surface of the silicon with the electrolyte. Example 3 A sample of compressed silicon di.oxide on a steel cathode was electro-reduced to silicon in an electrochemical cell at

800° C using a fused pure CaCl ₂ electrolyte and metal anode by the application of 3 volts dc. The electrode was periodically cycled between cathodic and anodic potentials. The impurities found to have accumulated at the surface of the silicon in example 2 had been removed from the bulk and surface of the silicon. Example 4

A sample of compressed silicon dioxide on a steel cathode was electro-reduced to silicon in an electrochemical cell at 800° C using a fused pure CaCl ₂ electrolyte and metal anode by the application of 3 volts dc. A second metal cathode was added to the cell and 3 volts dc applied to it whilst the electro- reduction is kept at open circuit. Impurites were found to have deposited on the second cathode, tlius purifying the electrolyte. Example 5

Powdered zinc was added to a fused and impure CaCl ₂ electrolyte at 800° C. The molten material was able to scavenge impurities from more noble metals and. thus purify the electrolyte whilst increasing its concentration in the CaCl ₂ . Example 6

Powdered sodium ion-containing zircσnia was added to a fused and impure CaCl ₂ electrolyte at 800 ⁰ C. Impurities in the electrolyte were removed and the concentration of sodium ions increased. Example 7

A sample of silicon dioxide was compressed with a small amount of sodium borate to form an electrode pellet on a steel cathode. The sample was electro-reduced and found to contain mostly silicon but with additional boron species through its bulk suitable for further processing into various p-type semi¬ conducting phases.

The above Examples are not to be regarded as limiting in that modifications to the above would be olovious to one skilled in the art. In particular, their teaching would be transferable to other elements of the periodic table, and the form of the

working electrode may be selected depending on the final application or end use. As described earlier, one or more of the processes for purifying the electrolyte may additionally be employed, depending on the required purity level of the extracted element.