Extraction of Scandia

The method employed by Crookes in extracting scandia from wiikite consisted in first isolating the scandia and rare earths, and then separating the scandia from this mixture by the fractional decomposition of the nitrates. To obtain the mixture of scandia and rare earths, the mineral (1 pt.) was finely powdered and fused in a clay crucible with potassium hydrogen sulphate (5 pts.). The melt was cooled, powdered, and mixed with five times its weight of water, stirred for twelve hours, and then filtered through a linen filter. The filtrate was boiled with an excess of ammonia when all the earths were precipitated together with zirconia, titanic acid, ferric hydroxide, etc. The well-washed precipitate was heated with a slight excess of oxalic acid, and when cold the oxalates were filtered and washed. The oxalates were next converted into anhydrous sulphates, dissolved in cold water, filtered, and the ammonia and oxalic acid processes repeated. The oxalates were then well washed and calcined to oxides.

The extraction of scandia from wolframite has been very thoroughly studied by K. J. Meyer and others. Only the wolframite from Zinnwald (Z.) and Sadisof (S.) is said to be worth working up for scandium; analyses of the carefully picked mineral yield the following results: -

	WO3	FeO	MnO	PbO + SnO2	TiO2 + Ta2O5	CaO	Rare Earth
(Z.)	75.41	9.34	14.00	0.18	0.50	0.55	0.15
(S.)	73.47	15.13	9.81	0.47	0.63	0.54	0.20

Rather more than half of the "rare earths" present consists of scandia.

The finely powdered wolframite is fused with about 2.3 times its weight of sodium carbonate, a little potassium nitrate being also added. The melt is boiled with water and a little alcohol, and the brown mixture of oxides left (which contains all the scandium) is washed, dissolved in hydrochloric acid, and the silica present removed in the usual manner.

The oxide residues from tungsten factories, if good Zinnwald wolframite has been worked up, are more convenient to use as raw material than wolframite itself, and contain 0.30 per cent, of rare earths. One kilogram of the oxide residues is dissolved in 3 litres of raw hydrochloric acid, the powder being introduced into the boiling acid. After drawing off the deposit of silica and boiling it with dilute hydrochloric acid, the filtrates are united.

To separate the scandium, the preceding solution is boiled, and 40 grams of sodium silicofluoride are added slowly with stirring. The boiling is continued for half an hour, when the scandium separates completely as a white, pasty precipitate of scandium fluoride (produced by decomposition of the silicofluoride). The precipitate is filtered on a Nutschen filter and boiled with very dilute hydrochloric acid. The fluoride is converted into sulphate in a platinum dish, the scandium precipitated as hydroxide with ammonia, and converted into oxalate by boiling with oxalic acid. The oxalate when washed and ignited leaves a white residue of nearly pure scandium oxide, which should weigh 3.0 to 3.3 grams.

The small amount of yttrium earths present in the scandia is easily removed. The scandia is dissolved in hydrochloric acid, excess of the acid removed, and the scandium precipitated at 100° as the basic thiosulphate. The washed precipitate is decomposed by hydrochloric acid, and the precipitation as thiosulphate repeated. The washed precipitate, now quite free from yttrium earths, is decomposed by hydrochloric acid, and the scandium in the filtered liquid precipitated as oxalate. The only impurity now remaining is thorium.

It is not difficult to reduce the amount of thoria in scandia to 0.5 per cent. This may be effected in several ways, one of the most satisfactory being to pour the neutral solution of the chlorides of scandium and thorium into a large excess of 20 per cent, sodium carbonate solution, and boil the solution, when scandium sodium carbonate is precipitated, the bulk of the thorium remaining in solution. The final 0.5 per cent, of thoria, however, is difficult to eliminate. It cannot, for instance, be precipitated as the peroxide, and, moreover, it cannot be detected in the scandia by its arc spectrum. A number of methods for its elimination are known, the first being to drop slowly a neutral solution of the scandium salt (contaminated with thorium) into a concentrated (10 to 20 per cent.) neutral solution of ammonium tartrate, and boil the clear solution so obtained with ammonia. Scandium ammonium tartrate is thus precipitated free from thorium, and is washed with dilute ammonium tartrate solution. The second method can only be advantageously employed on a small scale. It consists in precipitating the thorium from a nitrate solution containing much free nitric acid by the addition of an excess of potassium iodate sufficiently great to co-precipitate a portion of the scandium. The liquid is then quite free from thorium. The third method consists in pouring a neutral chloride solution of the scandium chloride into an excess of aqueous ammonium fluoride contained in a platinum dish and vigorously stirred. One gram of scandia requires 8 grams of ammonium fluoride. The liquid is evaporated at 100°, and the thorium separates out as the insoluble fluoride; any scandium ammonium fluoride that crystallises out may be redissolved by warming with more water, and the liquid then filtered. A fourth method, which is not, however, very convenient, is to separate the anhydrous chlorides of thorium and scandium by fractional sublimation, the former being the more volatile.