It may be asserted that almost every known method of desulphurization was examined and experimented upon on a large scale. Gas firing was exclusively used on certain classes of ores for several years with considerable success, and revolving furnaces of the Brückner type—gas fired—were also tried. Although varying degrees of success were obtained, no really great progress was made in actual desulphurization; methods were cheapened and larger quantities handled at a time, but the final product—whether sintered or in a pulverulent state—seldom averaged much under 5 per cent. sulphur, while the days of the old “gray slags” (1 per cent. to 2 per cent. sulphur) from the reverberatories totally disappeared, together with the class of ores which produced them.
During the long period of these experiments in desulphurization various facts were established:
(1) That sulphide of lead—especially in a pulverulent state—could not be desulphurized in the same way as other sulphides, such as sulphides of iron, copper, zinc, etc., because if roasted in a mechanical furnace the temperature had to be kept low enough to avoid premature sintering, which would choke the stirrers and cause trouble by the ore clogging on the sides and bottom of the furnace. If, however, the ore was roasted in a “dry state” at low temperature, a great deal of sulphur remained in the product as sulphate of lead, which was as bad for the subsequent blast-furnace work as the sulphide of lead itself. When air was pressed through molten galena—in the same way as through molten copper matte—a very heavy volatilization of lead took place, while portions of it were reduced to metal or were contained as sulphide in the molten matte, so that a good product was not obtained.
(2) That no complete dead roast of lead ores could be obtained unless the final product was thoroughly smelted and agglomerated.
(3) That a well roasted lead ore could be obtained by oxidizing the PbS with compressed air, after the ore had been suitably prepared.
(4) That metal losses were mainly due to the excessive heat produced in the oxidation of PbS to PbO, and other sulphides present in the ore.
It was by making use of these facts that the H.-H. roasting process was finally evolved, and by carefully applying its principles it is possible to desulphurize completely the ore to a practically dead roast of under 1 per cent. sulphur; in practice, however, such perfection is unnecessary and a well agglomerated product with from 2 to 2.5 per cent. sulphur is all that is required. During some trials in Australia, where a great degree of perfection was aimed at, a block of over 2000 tons of agglomerated, roasted ore was produced containing 1 per cent. sulphur (as sulphide); as the ores contained an average of about 10 per cent. Zn, this was a very fine result from a desulphurization point of view, but it was not found that this 1 per cent. product gave any better results in the subsequent smelting in the blast furnace than later on a less carefully prepared material containing 2.5 per cent. sulphur.
In the early stages of experiment the great difficulty was to obtain agglomeration without first fusing the sulphides in the ore, and turning out a half-roasted product full of leady matte. Simple as the thing now is, it seemed at times impossible to avoid this defect, and it was only by a careful study of the effects of an addition of lime, Fe2O3 or Mn2O3, and their properties that the right path was struck. Before the introduction of the H.-H. process lime was only used in the reverberatory process (Flintshire and Tarnowitz) to stiffen the charge, but as Percy tells us that after its addition the charge was glowing, it must have had a chemical as well as a mechanical effect. In recognition of this fact fine caustic lime or crushed limestone was mixed with the ore before charging it into the furnace and exposing it to an oxidizing heat.
It was thought probable that a dioxide of lime might be temporarily formed, which in contact with PbS would be decomposed immediately after its formation, or that the CaO served as Contactsubstanz in the same way as spongy platinum, metallic silver, or oxide of iron. As CaSO4 and not CaSO3 is always found in the roasted ore, this may prove that CaO is really a contact substance for oxygen (see W. M. Hutchings, Engineering and Mining Journal, Oct. 21, 1905, Vol. LXXX, p. 726). The fact that the process works equally well with Fe2O3 instead of CaO speaks against the theory of plumbate of lime. Whatever theory may be correct, the fact remains that CaO assists the roasting process and that by its use the premature agglomeration of the sulphide ore is avoided. A further advantage of lime is that it keeps the charge more porous and thus facilitates the passage of the air.
The shape and size of the blowing apparatus best adapted for the purpose in view occupied many months; starting from very shallow pans or rectangular boxes several feet square with a few inches of material over a perforated plate, it gradually resolved itself into the cone-shaped receptacle—holding about a ton of ore—as first introduced together with the process. In later years and in treating larger quantities a more hemispherical form has been adopted, containing up to 15 tons of ore.