properties, and prepared several compounds hitherto unknown. Early in 1854, H. St Claire Deville, accidentally and in ignorance of Wöhler's later results, imitated the 1845 experiment. At once observing the reduction of the chloride, he realized the importance of his discovery and immediately began to study the commercial production of the metal. His attention was at first divided between two processes—the chemical method of reducing the chloride with potassium, and an electrolytic method of decomposing it with a carbon anode and a platinum cathode, which was simultaneously imagined by himself and R. Bunsen. Both schemes appeared practically impossible; potassium cost about £17 per ℔, gave a very small yield and was dangerous to manipulate, while on the other hand, the only source of electric current then available was the primary battery, and zinc as a store of industrial energy was utterly out of the question. Deville accordingly returned to pure chemistry and invented a practicable method of preparing sodium which, having a lower atomic weight than potassium, reduced a larger proportion. He next devised a plan for manufacturing pure alumina from the natural ores, and finally elaborated a process and plant which held the field for almost thirty years. Only the discovery of dynamo-electric machines and their application to metallurgical processes rendered it possible for E. H. and A. H. Cowles to remove the industry from the hands of chemists, till the time when P. T. L. Héroult and C. M. Hall, by devising the electrolytic method now in use, inaugurated the present era of industrial electrolysis.
The chief natural compounds of aluminium are four in number: oxide, hydroxide (hydrated oxide), silicate and fluoride. Corundum, the only important native oxide (Al2O3), occurs in large deposits in southern India and the United States. Although it contains a higher percentage of metal Ores.(52·9%) than any other natural compound, it is not at present employed as an ore, not only because it is so hard as to be crushed with difficulty, but also because its very hardness makes it valuable as an abrasive. Cryolite (AlF3·5NaF) is a double fluoride of aluminium and sodium, which is scarcely known except on the west coast of Greenland. Formerly it was used for the preparation of the metal, but the inaccessibility of its source, and the fact that it is not sufficiently pure to be employed without some preliminary treatment, caused it to be abandoned in favour of other salts. When required in the Héroult-Hall process as a solvent, it is sometimes made artificially. Aluminium silicate is the chemical body of which all clays are nominally composed. Kaolin or China clay is essentially a pure disilicate (Al2O3·2SiO2·2H2O), occurring in large beds almost throughout the world, and containing in its anhydrous state 24·4% of the metal, which, however, in common clays is more or less replaced by calcium, magnesium, and the alkalis, the proportion of silica sometimes reaching 70%. Kaolin thus seems to be the best ore, and it would undoubtedly be used were it not for the fatal objection that no satisfactory process has yet been discovered for preparing pure alumina from any mineral silicate. If, according to the present method of winning the metal, a bath containing silica as well as alumina is submitted to electrolysis, both oxides are dissociated, and as silicon is a very undesirable impurity, an alumina contaminated with silica is not suited for reduction. Bauxite is a hydrated oxide of aluminium of the ideal composition, Al2O3·2H2O. It is a somewhat Widely distributed mineral, being met with in Styria, Austria, Hesse, French Guiana, India and Italy; but the most important beds are in the south of France, the north of Ireland, and in Alabama, Georgia and Arkansas in North America. The chief Irish deposits are in the neighbourhood of Glenravel, Co. Antrim, and have the advantage of being near the coast, so that the alumina can be transported by water-carriage. After being dried at 100° C., Antrim bauxite contains from 33 to 60% of alumina, from 2 to 30% of ferric oxide, and from 7 to 24% of silica, the balance being titanic acid and water of combination. The American bauxites contain from 38 to 67% of alumina, from 1 to 23% of ferric oxide, and from 1 to 32% of silica. The French bauxites are of fairly constant composition, containing usually from 58 to 70% of alumina, 3 to 15% of foreign matter, and 27% made up of silica, iron oxide and water in proportions that vary with the colour and the situation of the beds.
Before the application of electricity, only two compounds were found suitable for reduction to the metallic state. Alumina itself is so refractory that it cannot be melted save by the oxy-hydrogen blowpipe or the electric arc, and except in the molten state it is not susceptible of decomposition by any chemical reagent. Deville first selected the chloride as his raw material, but observing it to be volatile and extremely deliquescent, he soon substituted in its place a double chloride of aluminium and sodium. Early in 1855 John Percy suggested that cryolite should be more convenient, as it was a natural mineral and might not require purification, and at the end of March in that year, Faraday exhibited before the Royal Institution samples of the metal reduced from its fluoride by Dick and Smith. H. Rose also carried out experiments on the decomposition of cryolite, and expressed an opinion that it was the best of all compounds for reduction; but, finding the yield of metal to be low, receiving a report of the difficulties experienced in mining the ore, and fearing to cripple his new industry by basing it upon the employment of a mineral of such uncertain supply, Deville decided to keep to his chlorides. With the advent of the dynamo, the position of affairs was wholly changed. The first successful idea of using electricity depended on the enormous heating powers of the arc. The infusibility of alumina was no longer prohibitive, for the molten oxide is easily reduced by carbon. Nevertheless, it was found impracticable to smelt alumina electrically except in presence of copper, so that the Cowles furnace yielded, not the pure metal, but an alloy. So long as the metal was principally regarded as a necessary ingredient of aluminium-bronze, the Cowles process was popular, but when the advantages of aluminium itself became more apparent, there arose a fresh demand for some chief method of obtaining it unalloyed. It was soon discovered that the faculty of inducing dissociation possessed by the current might now be utilized with some hope of pecuniary success, but as electrolytic currents are of lower voltage than those required in electric furnaces, molten alumina again became impossible. Many metals, of which copper, silver and nickel are types, can be readily won or purified by the electrolysis of aqueous solutions, and theoretically it may be feasible to treat aluminium in an identical, manner. In practice, however, it cannot be thrown down electrolytically with a dissimilar anode so as to win the metal, and certain difficulties are still met with in the analogous operation of plating by means of a similar anode. Of the simple compounds, only the fluoride is amenable to electrolysis in the fused state, since the chloride begins to volatilize below its melting-point, and the latter is only 5° below its boiling-point. Cryolite is not a safe body to electrolyse, because the minimum voltage needed to break up the aluminium fluoride is 4·0, whereas the sodium fluoride requires only 4·7 volts; if, therefore, the current rises in tension, the alkali is reduced, and the final product consists of an alloy with sodium. The corresponding double chloride is a far better material; first, because it melts at about 180° C., and does not volatilize below a red heat, and second, because the voltage of aluminium chloride is 2·3 and that of sodium chloride 4·3, so that there is a much wider margin of safety to cover irregularities in the electric pressure. It has been found, however, that molten cryolite and the analogous double fluoride represented by the formula Al3F5·2NaF are very efficient solvents of alumina, and that these solutions can be easily electrolysed at about 800° C. by means of a current that completely decomposes the oxide but leaves the haloid salts unaffected. Molten Cryolite dissolves roughly 30% of its weight of pure alumina, so that when ready for treatment the solution contains about the same proportion of what may be termed “available” aluminium as does the fused double chloride of aluminium and sodium. The advantages lie with the oxide because of its easier preparation. Alumina dissolves readily enough in aqueous hydrochloric acid to yield a solution of the chloride, but neither this solution, nor that containing sodium chloride, can be evaporated to dryness without decomposition. To obtain the anhydrous single or
