This section is from "Scientific American Supplement Volumes 275, 286, 288, 299, 303, 312, 315, 324, 344 and 358". Also available from Amazon: Scientific American Reference Book.
The quantitative separation of iron from aluminum, which presented many difficulties according to the older methods, may be easily performed by electrolysis. If a solution of iron ammonium oxalate and aluminum oxalate, to which an excess of ammonium oxalate has been added, be submitted to the action of the electric current, the iron will be deposited as a firmly adhering coat on the negative electrode, while the aluminum oxide remains in solution, just so long as the quantity of ammonium oxalate is in excess of the quantity of ammonium carbonate produced. When, finally, a precipitation of aluminum oxide takes place the liquid is almost free from iron. From time to time, the solution, in which the aluminum oxide is suspended, is tested for iron by ammonium sulphide, and the current is interrupted when no further reaction is observed. The best method of procedure is to add ammonium oxalate in excess to a neutral, a slightly acid solution, or to one which has been neutralized by the addition of ammonium hydrate (a hydrochloric acid solution is not well adapted for this purpose); then as much more solid ammonium oxalate is added until for every 0.1 gramme there is 2 to 3 grammes of the oxalate present. The hot solution is then directly submitted to the action of the electric current. After the iron has been precipitated, it is best to stop the action of the current before all the aluminum oxide is thrown down, for otherwise a portion of the latter may adhere firmly to the iron, and be difficult to remove.
In such a case, as was mentioned previously in the separation of iron from manganese, it is necessary to redissolve the iron (after previously having poured off the liquid) in oxalic acid, and then the electrolysis is continued.
In order to effect the complete precipitation of the aluminum oxide from the solution which was poured from the iron, ammonium hydrate is added, and the solution boiled for some time, and then the aluminum oxide is determined in the usual manner. When the quantity of aluminum is less than that of iron, this method may be relied upon to give exact results. With the reverse (i. e., an excess of iron) the precipitate of aluminum oxide must be dissolved in oxalic acid (without the interruption of the current), and the electrolysis continued.--Berichte der Deutschen Chemischen Gesellschaft, 14, 1662.