the necessary proportion, and melted in crucibles to give merchantable bronzes containing between 114 and 10% of aluminium. Although the copper took no part in the reaction, its employment was found indispensable, as otherwise the aluminium partly volatilized, and partly combined with the carbon to form a carbide. It was also necessary to give the fine charcoal a thin coating of calcium oxide by soaking it in lime-water, for the temperature was so high that unless it was thus protected it was gradually converted into graphite, losing its insulating power and diffusing the current through the lining and walls of the furnace. That this process did not depend upon electrolysis, but was simply an instance of electrical smelting or the decomposition of an oxide by means of carbon at the temperature of the electric arc, is shown by the fact that the Cowles furnace would work with an alternating current.
In 1883 R. Crätzel patented a useless electrolytic process with fused cryolite or the double chloride as the raw material, and in 1886 Dr E. Kleiner propounded a cryolite method which was worked for a time by the Aluminium Syndicate at Tyldesley near Manchester, but was abandoned in 1890. In 1887 A. Minet took out patents for electrolysing a mixture of sodium chloride with aluminium fluoride, or with natural or artificial cryolite. The operation was continuous, the metal being regularly run off from the bottom of the bath, while fresh alumina and fluoride were added as required. The process exhibited several disadvantages, the electrolyte had to be kept constant in composition lest either fluorine vapours should be evolved or sodium thrown down, and the raw materials had accordingly to be prepared in a pure state. After prolonged experiments in a factory owned by Messrs Bernard Frères at St Michel in Savoy, Minet’s process was given up, and at the close of the 19th century the Héroult-Hall method was alone being employed in the manufacture of aluminium throughout the world.
The original Deville process for obtaining pure alumina from bauxite was greatly simplified in 1889 by K. T. Bayer, whose improved process is exploited at Larne in Ireland and at Gardanne in France. New works on the same process have recently been erected near Marseilles. Crude bauxite is ground, lightly calcined to destroy organic matter, and agitated under a pressure of 70 or 80 ℔ per sq. in. with a solution of sodium hydroxide having the specific gravity 1·45. After two or three hours the liquid is diluted till its density falls to 1·23, when it is passed through filter-presses to remove the insoluble ferric oxide and silica. The solution of sodium aluminate, containing aluminium oxide and sodium oxide in the molecular proportion of 6 to 1, is next agitated for thirty-six hours with a small quantity of hydrated alumina previously obtained, which causes the liquor to decompose, and some 70% of the aluminium hydroxide to be thrown down. The filtrate, now containing roughly two molecules of alumina to one of soda, is concentrated to the original gravity of 1·45, and employed instead of fresh caustic for the attack of more bauxite; the precipitate is then collected, washed till free from soda, dried and ignited at about 1000° C. to convert it into a crystalline oxide which is less hygroscopic than the former amorphous variety.
The process of manufacture which now remains to be described was patented during 1886 and 1887 in the name of C. M. Hall in America, in that of P. T. L. Héroult in England and France. It would be idle to discuss to whom the credit of first imagining the method rightfully belongs, for probably this is only one of the many occasions when new ideas have been born in several brains at the same time. By 1888 Hall was at work on a commercial scale at Pittsburg, reducing German alumina; in 1891 the plant was removed to New Kensington for economy in fuel, and was gradually enlarged to 1500 h.p.; in 1894 a factory driven by water was erected at Niagara Falls, and subsequently works were established at Shawenegan in Canada and at Massena in the United States. In 1890 also the Hall process operated by steam power was installed at Patricroft, Lancashire, where the plant had a capacity of 300 ℔ per day, but by 1894 the turbines of the Swiss and French works ruined the enterprise. About 1897 the Bernard factory at St Michel passed into the hands of Messrs Péchiney, the machinery soon being increased, and there, under the control of a firm that has been concerned in the industry almost from its inception, aluminium is being manufactured by the Hall process on a large scale. In July 1888 the Société Métallurgique Suisse erected plant driven by a 500 h.p. turbine to carry out Héroult’s alloy process, and at the end of that year the Allgemeine Elektricitäts Gesellschaft united with the Swiss firm in organizing the Aluminium Industrie Actien Gesellschaft of Neuhasen, which has factories in Switzerland, Germany and Austria. The Société Electrométallurgique Française, started under the direction of Héroult in 1888 for the production of aluminium in France, began operations on a small scale at Froges in Isère; but soon after large works were erected in Savoy at La Praz, near Modane, and in 1905 another large factory was started in Savoy at St Michel. In 1895 the British Aluminium Company was founded to mine bauxite and manufacture alumina in Ireland, to prepare the necessary electrodes at Greenock, to reduce the aluminium by the aid of water-power at the Falls of Foyers, and to refine and work up the metal into marketable shapes at the old Milton factory of the Cowles Syndicate, remodelled to suit modern requirements. In 1905 this company began works for the utilization of another water-power at Loch Leven.
In 1907 a new company, The Aluminium Corporation, was started in England to carry out the production of the metal by the Héroult process, and new factories were constructed near Conway in North Wales and at Wallsend-on-Tyne, quite close to where, twenty years before, the Alliance Aluminium Co. had their works.
The Héroult cell consists of a square iron or steel box lined with carbon rammed and baked into a solid mass; at the bottom is a cast-iron plate connected with the negative pole of the dynamo, but the actual working cathode is undoubtedly the layer of already reduced and molten metal that lies in the bath. The anode is formed of a bundle of carbon rods suspended from overhead so as to be capable of vertical adjustment. The cell is filled up with cryolite, and the current is turned on till this is melted; then the pure powdered alumina is fed in continuously as long as the operation proceeds. The current is supplied at a tension of 3 to 5 volts per cell, passing through 10 or 12 in series; and it performs two distinct functions:—(1) it overcomes the chemical affinity of the aluminium oxide, (2) it overcomes the resistance of the electrolyte, heating the liquid at the same time. As a part of the voltage is consumed in the latter duty, only the residue can be converted into chemical work, and as the theoretical voltage of the aluminium fluoride in the cryolite is 4.0, provided the bath is kept properly supplied with alumina, the fluorides are not attacked. It follows, therefore, except for mechanical losses, that one charge of cryolite lasts indefinitely, that the sodium and other impurities in it are not liable to contaminate the product, and that only the alumina itself need be carefully purified. The operation is essentially a dissociation of alumina into aluminium, which collects at the cathode, and into oxygen, which combines with the anodes to form carbon monoxide, the latter escaping and being burnt to carbon dioxide outside. Theoretically 36 parts by weight of carbon are oxidized in the production of 54 parts of aluminium; practically the anodes waste at the same rate at which metal is deposited. The current density is about 700 ampères per sq. ft. of cathode surface, and the number of rods in the anode is such that each delivers 6 or 7 ampères per sq. in. of cross-sectional area. The working temperature lies between 750° and 850° C., and the actual yield is 1 ℔ of metal per 12 e.h.p. hours. The bath is heated internally with the current rather than by means of external fuel, because this arrangement permits the vessel itself to be kept comparatively cool; if it were fired from without, it would be hotter than the electrolyte, and no material suitable for the construction of the cell is competent to withstand the attack of nascent aluminium at high temperatures. Aluminium is so light that it is a matter requiring some ingenuity to select a convenient solvent through which it shall sink quickly, for if it does not sink, it short-circuits the electrolyte. The molten metal has a specific gravity of 2·54, that of molten
