double chloride, alumina must be ignited with carbon in a current of chlorine, and to exclude iron from the finished metal, either the alumina must be pure or the chloride be submitted to purification. This preparation of a chlorine compound suited for electrolysis becomes more costly and more troublesome than that of the oxide, and in addition four times as much raw material must be handled.
At different times propositions have been made to win the metal from its sulphide. This compound possesses a heat of formation so much lower that electrically it needs but a voltage of 0·9 to decompose it, and it is easily soluble in the fused sulphides of the alkali metals. It can also be reduced metallurgically by the action of molten iron. Various considerations, however, tend to show that there cannot be so much advantage in employing it as would appear at first sight. As it is easier to reduce than any other compound, so it is more difficult to produce. Therefore while less energy is absorbed in its final reduction, more is needed in its initial preparation, and it is questionable whether the economy possible in the second stage would not be neutralized by the greater cost of the first stage in the whole operation of winning the metal from bauxite with the sulphide as the intermediary.
The Deville process as gradually elaborated between 1855 and 1859 exhibited three distinct phases:—Production of metallic sodium, formation of the pure double chloride of sodium and aluminium, and preparation of the metal by the interaction of the two former substances. To produce the alkali metal, a calcined mixture of sodium carbonate, coal and Chemical reduction.chalk was strongly ignited in flat retorts made of boiler-plate; the sodium distilled over into condensers and was preserved under heavy petroleum. In order to prepare pure alumina, bauxite and sodium carbonate were heated in a furnace until the reaction was complete; the product was then extracted with water to dissolve the sodium aluminate, the solution treated with carbon dioxide, and the precipitate removed and dried. This purified oxide, mixed with sodium chloride and coal tar, was carbonized at a red heat, and ignited in a current of dry chlorine as long as vapours of the double chloride were given off, these being condensed in suitable chambers. For the production of the final aluminium, 100 parts of the chloride and 45 parts of cryolite to serve as a flux were powdered together and mixed with 35 parts of sodium cut into small pieces. The whole was thrown in several portions on to the hearth of a furnace previously heated to low redness and was stirred at intervals for three hours. At length when the furnace was tapped a white slag was drawn off from the top, and the liquid metal beneath was received into a ladle and poured into cast-iron moulds. The process was worked out by Deville in his laboratory at the École Normale in Paris. Early in 1855 he conducted large-scale experiments at ]avel in a factory lent him for the purpose, where he produced sufficient to show at the French Exhibition of 1855. In the spring of 1856 a complete plant was erected at La Glaciére, a suburb of Paris, but becoming a nuisance to the neighbours, it was removed to Nanterre in the following year. Later it was again transferred to Salindres, where the manufacture was continued by Messrs. Péchiney till the advent of the present electrolytic process rendered it no longer profitable.
When Deville quitted the Javel works, two brothers C. and A. Tissier, formerly his assistants, who had devised an improved sodium furnace and had acquired a thorough knowledge of their leader’s experiments, also left, and erected a factory at Amfreville, near Rouen, to work the cryolite process. It consisted simply in reducing cryolite with metallic sodium exactly as in Deville’s chloride method, and it was claimed to possess various mythical advantages over its rival. Two grave disadvantages were soon obvious—the limited supply of ore, and, what was even more serious, the large proportion of silicon in the reduced metal. The Amfreville works existed some eight or ten years, but achieved no permanent prosperity. In 1858 or 1859 a small factory, the first in England, was built by F. W. Gerhard at Battersea, who also employed cryolite, made his own sodium, and was able to sell the product at 3s. 9d. per oz. This enterprise only lasted about four years. Between 1860 and 1874 Messrs Bell Brothers manufactured the metal at Washington, near Newcastle, under Deville’s supervision, producing nearly 2 cwt. per year. They took part in the International Exhibition of 1862, quoting a price of 40s. per ℔ troy.
In 1881 J. Webster patented an improved process for making alumina, and the following year he organized the Aluminium Crown Metal Co. of Hollywood to exploit it in conjunction with Deville’s method of reduction. Potash-alum and pitch were calcined together, and the mass was treated with hydrochloric acid; charcoal and water to form a paste were next added, and the whole was dried and ignited in a current of air and steam. The residue, consisting of alumina and potassium sulphate, was leached with water to separate the insoluble matter which was dried as usual. All the by-products, potassium sulphate, sulphur and aluminate of iron, were capable of recovery, and were claimed to reduce the cost of the oxide materially. From this alumina the double chloride was prepared in essentially the same manner as practised at Salindres, but sundry economies accrued in the process, owing to the larger scale of working and to the adoption of W. Weldon’s method of regenerating the spent chlorine liquors. In 1886 H. Y. Castner’s sodium patents appeared, and The Aluminium Co. of Oldbury was promoted to combine the advantages of Webster’s alumina and Castner’s sodium. Castner had long been interested in aluminium, and was desirous of lowering its price. Seeing that sodium was the only possible reducing agent, he set himself to cheapen its cost, and deliberately rejecting sodium carbonate for the more expensive sodium hydroxide (caustic soda), and replacing carbon by a mixture of iron and carbon—the so-called carbide of iron—he invented the highly scientific method of winning the alkali metal which has remained in existence almost to the present day. In 1872 sodium prepared by Deville’s process cost about 4s: per ℔, the greater part of the expense being due to the constant failure of the retorts; in 1887 Castner’s sodium cost less than 1s. per ℔, for his cast-iron pots survived 125 distillations.
In the same year L. Grabau patented a method of reducing the simple fluoride of aluminium with sodium, and his process was operated at Trotha in Germany. It was distinguished by the unusual purity of the metal obtained, some of his samples containing 99·5 to 99·8%. In 1888 the Alliance Aluminium Co., organized to work certain patents for winning the metal from cryolite by means of sodium, erected plant in London, Hebburn and Wallsend, and by 1889 were selling the metal at 11s. to 15s. per ℔. The Aluminium Company’s price in 1888 was 20s. per ℔ and the output about 250 ℔ per day. In 1889 the price was 16s., but by 1891 the electricians commenced to offer metal at 4s. per ℔, and aluminium reduced with sodium became a thing of the past.
About 1879 dynamos began to be introduced into metallurgical practice, and from that date onwards numerous schemes for utilizing this cheaper source of energy were brought before the public. The first electrical method worthy of notice is that patented by E. H. and A. H. Cowles in 1885, which was worked both at Lockport, New York, U.S.A., Electrical reduction.and at Milton, Staffordshire. The furnace consisted of a flat, rectangular, firebrick box, packed with a layer of finely-powdered charcoal 2 in. thick. Through stuffing-boxes at the ends passed the two electrodes, made after the fashion of arc-light carbons, and capable of being approached together according to the requirements of the operation. The central space of the furnace was filled with a mixture of corundum, coarsely-powdered charcoal and copper; and an iron lid lined with firebrick was luted in its place to exclude air. The charge was reduced by means of a 50-volt current from a 300-kilowatt dynamo, which was passed through the furnace for 112 hours till decomposition was complete. About 100 ℔ of bronze, containing from 15 to 20 ℔ of aluminium, were obtained from each run, the yield of the alloy being reported at about 1 ℔ per 18 e.h.p-hours. The composition of the alloys thus produced could not be predetermined with exactitude; each batch was therefore analysed, a number of them were bulked together or mixed with copper in
