Preparation of Aluminium

The oxide, sulphide, and halides of aluminium are very stable compounds, and the preparation of the metal by chemical processes is somewhat difficult. From thermochemical considerations, it may be deduced that the most successful methods would be those in which an aluminium halide is reduced by fusion with an alkali metal.

Equilibrium diagrams for the systems I. alumina - cryolite, II. alumina - fluorspar.

It is possible that aluminium was obtained by Oersted in 1842, but it is generally acknowledged that the merit of having first prepared aluminium and studied its properties belong to Wohler, who, by heating anhydrous aluminium chloride with potassium, obtained the metal first in the form of a grey powder which became brilliant when burnished, and subsequently as fused metallic globules.

In 1854 both Bunsen and Deville succeeded in preparing aluminium by electrolysing the fused anhydron chloride, and in the same year Deville commenced his classic work on the manufacture of aluminium by reducing aluminium sodium chloride with metallic sodium. The following year saw the introduction of another process, namely, the reduction of cryolite by fusion with sodium. This method was worked out by Percy in England and by Rose in Germany. Later, Grabeau reduced anhydrous aluminium fluoride by means of sodium, and obtained aluminium of remarkable purity.

At the present time the only method by which aluminium is produced on a commercial scale is an electrolytic method, developed independently and almost simultaneously in 1886 by C. M. Hall in America and P. V. Heroult in France. Alumina is dissolved in molten cryolite and electrolysed, when aluminium separates out at the cathode. Attempts have been made to electrolyse aluminium sulphide dissolved in fused sodium sulphide, but they have not yet been commercially successful.

The necessity for finding a solvent for the alumina arises from the fact that its melting-point is 2010° to 2050°. The melting-point of cryolite is a little below 1000°. In order to produce a more fluid bath, other fluorides are added to the cryolite, e.g. aluminium fluoride and calcium floride.

Equilibrium diagram for the system cryolite-fluorspar

The system cryolite - alumina - fluorspar. Projection of " liquidus." Contour lines for each 50° difference in temperature.

The equilibrium diagrams for the systems (i.) alumina - cryolite, (ii.) alumina - fluorspar, (iii.) cryolite - fluorspar, (iv.) cryolite - alumina - fluorspar, and (v.) aluminium fluoride - sodium fluoride have been partly or wholly determined. The first three systems have been studied by Pascal and Jouniaux, whose results are shown graphically in figs. In each system, two series of mixed crystals are formed, and the " liquidus" curve consists of two branches meeting at a eutectic point. The ternary system as worked out by the same experimenters is shown graphically in figs., which represent the "liquidus" and "solidus" respectively. The isotbermals are given for a series of temperatures separated by intervals of 50°. The ternary eutectic point is 868° and corresponds to a mixture of the following composition: cryolite, 59.3 per cent.; fluorspar, 23.0 per cent.; alumina, 17.7 per cent. In actual practice, according to Pascal and Jouniaux, alumina is added so as to constitute 10 to 25 per cent, of the alumina-cryolite mixture, and fluorspar is added up to 36 per cent, of the weight of the cryolite present. The working compositions are therefore included in the trapezium bounded by the lines F6, Fc, be, and A a. This trapezium includes the eutectic mixture, and no mixture in it has a melting-point higher than 980°. The temperature limits adopted in practice are given by Pascal and Jouniaux as 875° to 950°.

The system cryolite - alumina - fluorspar. Projection of "solidus." Contour lines for each 50° difference in temperature.

Equilibrium diagram for the system equilibrium sodium fluoride - aluminium fluoride.

The system aluminium fluoride - sodium fluoride has been studied by Fedoteev and Iljinsky, whose results are shown graphically in fig. The branches AB, BCD, and DE correspond to the solid phases NaF, 3NaF.AlF3, and 5NaF.3AlF3 respectively, and the latter compound breaks up at 725° into 3NaF.AlF3 and A1F3. It will be seen that the addition of aluminium fluoride to cryolite produces a readily fusible mixture. Fedoteev and Iljinsky recommend that the solvent for alumina should be prepared by adding sufficient aluminium fluoride to cryolite to give the mixture the composition G (fig.); that the temperature should be maintained at 900°, and that the quantity of alumina added should not exceed 7.5 per cent, of the weight of the solvent. If pure cryolite be used as solvent, Fedoteev and Iljinsky recommend that not more than 10 per cent, of alumina be added.

Density of aluminium, etc., at high temperatures.

It is essential that the molten metal shall be liberated at the cathode in a medium of lower specific gravity than the metal itself. At the ordinary temperature, aluminium is less dense than cryolite; fortunately, however, the reverse is true at high temperatures, as will be seen from the results tabulated below and represented diagrammatically in fig. It will be seen that molten cryolite, like water, has a point of maximum density, the density then being 2.22 and the temperature 995°. The density of cryolite is diminished by the addition of silica and increased by the addition of calcium fluoride (see fig.); the influence of alumina on the density is rather complex, but unless a large quantity (over 20 per cent.) is added, the density is diminished (see fig.). The densities of the commercially important ternary mixtures of cryolite, alumina, and fluorspar at 950° are shown graphically in fig.; it will be seen that they are less than 2.40, and in actual practice the presence of a little silica in the cryolite makes them all a little lower than the values given in the diagram.

Densities of mixture of cryolite, fluorspar, and alumina at 950° C.

The Hall and Heroult processes only differ slightly in the nature of the plant employed, and perhaps to some extent in the composition of the electrolyte adopted. Each Hall cell consists of a rectangular cast-iron box, 6'×3'×3', thickly lined with carbon. This lining forms the cathode. The actual depth of electrolyte in the bath is 6 inches. As anode, a large number of carbon rods are used, which dip into the electrolyte and end about 1 inch above the level of the aluminium at the bottom. It is therefore clear why the density of the bath must be such as to allow the aluminium, as it is produced, to sink rapidly to the bottom; otherwise, besides oxidation of the metal taking place, short circuiting will occur. About 15 to 20 parts of alumina are added for each 100 parts of solvent. The electrolyte is roughly covered with a layer of carbon, on the top of which alumina is placed. As electrolysis proceeds, this alumina is stirred into the bath and at intervals the separated aluminium is tapped off from the bottom.

The current consumed by each cell is 10,000 amperes at 5.5 volts, the current density being 100 amp./dcm.² at the cathode, and about 500 amp./dcm.² at the anode. The current efficiency does not exceed 70 per cent., largely owing to the formation of a " metal mist" of aluminium particles which diffuse to the surface and become oxidised. The least potential difference capable of producing continuous electrolysis is 2.1 to 2.2 volts, which value does not differ greatly from the approximate figure calculated from the heat of formation of alumina. The potential differences required for the electrolysis of aluminium chloride, aluminium fluoride, sodium fluoride, and calcium fluoride are approximately 2.3, 4.0, 4.7, and 4.7 volts respectively.

The cost of the carbon anodes, which are eaten away by the liberated oxygen with the formation mainly of carbon monoxide, is a very serious item. The anode rods must be evenly hard, very slightly porous, and leave very little ash when burnt. They are usually made from petroleum coke. Water- power is always employed in the production of the necessary electric current. The world's supply of aluminium is produced mainly by the Aluminium Company of America, at Niagara, Massena, and Shawingian Falls, the Societe Electrometallurgique Frangaise, at Froges, La Plaz, and St Michael, and the British Aluminium Company, at Kinlochleven in Scotland. The world's output of aluminium was about 8000 tons per annum from 1900-1905; since then it has steadily increased, and was about 30,000 tons in 1909.

Aluminium produced by electrolysis contains 99 per cent, or more of aluminium, the chief impurities being iron and silicon. A little carbon is also present, and in a sample examined spectroscopically by Hartley and Ramage, traces of sodium, potassium, calcium, copper, silver, manganese, lead, gallium, and indium were detected. It is extremely difficult to purify the aluminium after it has once been produced, and hence it is necessary to employ pure materials in its preparation. It is for this reason that the carbon anodes must be practically free from ash. The alumina is usually prepared from bauxite. The cryolite is not specially purified, but any foreign metals present are removed after it has been submitted to the action of the current for a short time.

Pure aluminium is best prepared by reducing pure, redistilled aluminium tribromide with a slight deficiency of sodium. The tribromide is mixed with sodium and potassium chlorides and heated with metallic sodium in a crucible lined with a mixture of alumina and sodium aluminate.