In metallurgy, the operation of roasting, as a preliminary to chlorination, has for its object the expulsion of the sulphur, arsenic, antimony and other volatile substances existing in the ore, and the oxidation of the metals left behind, so as to leave nothing (except metallic gold) which can combine with chlorine when the ore is subsequently treated with it in aqueous solution. For this purpose the ore is heated in a furnace, through which a current of air is passed, salt being added if oxide of copper, lime, magnesia, are present. Ores containing much pyrites might be freed from most of their sulphur by pile roasting, and then subjected to fine crushing and a dead roast in a reverberatory furnace, but the extra cost of handling would probably exceed the saving due to the smaller consumption of fuel. This system has not been tried in chlorination mills on an extensive scale. The ordinary reverberatory furnace, worked by hand labour, is still in use, especially where only a few tons, or less, of concentrates are to be treated per day. Various mechanical furnaces, capable of handling large quantities of ore, have been devised to supersede the old-fashioned contrivance, and some of these will be described in the sequel.

Chemistry of Oxidising Roasting

Sir Wm. Roberts- Austen discusses as follows the roasting of a “mixture consisting of sulphides mainly of iron and copper, with some sulphide of lead, small quantities of arsenic and antimony as arsenides, antimonides, and sulpho-salts, usually with copper as a base. The temperature of the furnace in which the operation is to be performed is gradually raised, the atmosphere being an oxidising one. The first effect of the elevation of the temperature is to distil off sulphur, reducing the sulphides to a lower stage of sulphurisation. This sulphur burns in the furnace atmosphere to sulphurous anhydride (SO2), and coming in contact with the material undergoing oxidation is converted into sulphuric anhydride (SO3). It should be noted that the material of the brickwork does not intervene in the reactions, except by its presence as a hot porous mass, but its influence is, nevertheless, considerable. The roasting of these sulphides presents a good case for the study of chemical equilibrium. As soon as the sulphurous anhydride reaches a certain tension, the oxidation of the sulphide is arrested, even though an excess of oxygen be present, and the oxidation is not resumed until the actions of the draught change the conditions of the atmosphere of the furnace, when the lower sulphides remaining are slowly oxidised, the copper sulphide being converted into copper sulphate, mainly by the intervention of the sulphuric anhydride, formed as indicated. Probably by far the greater part of the iron sulphide only becomes sulphate for a very brief period, being decomposed into the oxides of iron, mainly ferric oxide, the sulphur passing off. Any silver sulphide that is present would have been converted into metallic silver at the outset were it not for the simultaneous presence of other sulphides, notably those of copper and of iron, which enables the silver sulphide to become converted into sulphate. The lead sulphide is also converted into sulphate at this low temperature (about 500°). The heat is now raised still further with a view to split up the sulphate of copper, the decomposition of which leaves oxide of copper. If, as in this case, the bases are weak, the sulphuric anhydride escapes mainly as such; but when the sulphates of stronger bases are decomposed the sulphuric anhydride is to a great extent decomposed into a mixture of sulphurous anhydride and oxygen. The sulphuric anhydride, resulting from the decomposition of this copper sulphate, converts the silver into sulphate, and maintains it as such, just as, in turn, at a lower temperature, the copper itself had been maintained in the form of sulphate by the sulphuric anhydride eliminated from the iron sulphide. When only a little of the copper sulphate remains undecomposed, the silver sulphate begins to split up (at about 700°) partly by the direct action of heat alone, and partly by reactions such as those shown in the following equations:

Ag2SO4 + 4Fe3O4 = 2Ag + 6Fe2O3 + SO2
Ag2SO4 + Cu2O = 2Ag + CuSO4 + CuO

The charge still contains lead sulphate, which cannot be completely decomposed at any temperature attainable in the roasting furnace except in the presence of silica. The elimination of arsenic and antimony gives rise to problems of much interest, and again confronts the smelter with a case of chemical equilibrium. For the sake of brevity it will be well for the present to limit the consideration to the removal of antimony, which may be supposed to be present as sulphide. Some sulphide of antimony is distilled off, but this is not its only mode of escape. An attempt to remove antimony by rapid oxidation would be attended with the danger of converting it into insoluble antimoniates of the metals present in the charge. In the early stages of the roasting it is, therefore, necessary to employ a very low temperature, and the presence of steam is found to be useful as a source of hydrogen, which removes sulphur as hydrogen sulphide, the gas being freely evolved. The reaction

Sb2S3 + 3H2 = 3H2S + 2Sb

between hydrogen and sulphide of antimony is, however, endothermic, and could not, therefore, take place without the aid which is afforded by external heat. The facts appear to be as follows :—Sulphide of antimony, when heated, dissociates, and the tension of the sulphur vapour would produce a state of equilibrium if the sulphur,thus liberated were not seized by the hydrogen, and removed from the system. The equilibrium is thus destroyed, and fresh sulphide is dissociated. The general result being that the equilibrium is continually restored and destroyed until the sulphide is decomposed. The antimony combines with oxygen and escapes as volatile oxide, as does also the arsenic, a portion of which is volatilised as sulphide.

“ The main object of the process which has been considered is the formation of soluble sulphate of silver.” The reactions, however, are precisely similar in an ordinary oxidising roast.

The following remarks on the decomposition of the various minerals present in complex ores may be of use in assisting the student to understand the reactions which proceed in the roasting furnace:

Iron Pyrites, FeS2

On heating this compound sulphur is volatilised, the reactions being probably expressed thus:

3FeS2 = Fe3S4 + S2
7FeS2 = Fe7S8 + 3S2

The sulphur burns to SO2, which is partly converted by the heated quartz, &c., into SO3, uniting with the free oxygen present. The ferrous sulphate formed by this sulphuric acid is split up by the heat and the ferrous oxide (FeO) converted into ferric oxide (Fe2O3) which gives the ore a red colour when cold. Some basic sulphates always remain undecomposed. If the temperature of the part of the charge next the fire-bridge has been too high, or if the charge is kept too long in the furnace, especially when not freely exposed to the air, some magnetic oxide is formed, thus:

3Fe2O3 = 2Fe3O4 + O

The presence of magnetic oxide makes the ore darker in colour. This is an undesirable change, as the magnetic oxide is acted on by chlorine far more readily than the sesquioxide.

Donald Clark states that magnetic pyrites, Fe3S4, produced at an early stage is oxidised direct to Fe3O4 if the temperature is moderately high and the supply of air ample. When this is brought into the hotter part of the furnace, the magnetic oxide is slowly converted to ferric sesquioxide, Fe2O3. He gives no evidence in support of these statements, which can only be true of a small portion of the charge, if any.

W. E. Greenawalt states that the dark magnetic oxide can be reconverted to the red sesquioxide by subjection to a lower temperature and an abundant supply of air. For this reason he advocates finishing at a lower temperature than is used for the previous stages of roasting. In four successive tests on 100-ton lots he found that a higher percentage of extraction by chlorination was obtainable by roasting at a high initial heat and a low finishing heat than by roasting at a lower initial heat and a higher finishing heat, although, in the latter case, the elimination of the sulphur was more complete.

Roasting Copper Pyrites

The decomposition of the copper sulphate formed in the furnace leaves a mixture of cuprous and cupric oxides, both soluble in chlorine.

Galena, PbS – Roasting

The presence of this mineral in any but small quantities is very detrimental, as both lead sulphate and lead silicate (formed by its decomposition in the presence of silica) are very fusible, and, at the temperature required to split up copper sulphate, cause the ore to become pasty and form lumps. Roasting must be performed very slowly and cautiously to avoid this effect.

Roasting Arsenical Pyrites, FeAsS

Arseniates of iron, copper, lead, when formed are not easily decomposed, as they resist a high temperature, and are only slowly converted into sulphates by sulphuric acid at a red heat. It is, therefore, desirable to avoid their formation, and with this end in view the precautions which have been already mentioned above are taken.

Antimonial Sulphide Roasting

Antimonial Sulphides are still more difficult to deal with, the antimoniates formed being less easily decomposed than arseniates. Their formation is avoided in the manner already described.

Blende, ZnS Roasting

Blende, ZnS, forms oxide and sulphate of zinc, of which the latter can only be split up by a very high temperature. At a bright red heat a basic sulphate is formed which is converted to oxide at a white heat. If blende is roasted at a high temperature and with a plentiful supply of air, sulphate of zinc is not formed to a large extent.

Carbonate of Lime, CaCO3

Carbonate of Lime, CaCO3, is decomposed at a red heat, CO2 being given off and caustic lime, CaO, left in the charge. The change is slow at 600° (low red heat) and rapid at 800° (full red heat). Magnesium carbonate is similarly decomposed.

Tellurides Roasting

Tellurides containing gold are fusible far below a red heat, and heavy losses of gold may occur through absorption by the furnace bottom. The melted tellurides of gold may remain in the ore, and in that case at a red heat the tellurium is partly volatilised and partly oxidised to the oxide TeO2, which also sublimes. Spherical beads of gold remain behind and are difficult to dissolve. Selenides behave similarly.

Metallic Gold, Silver Roasting

Metallic Gold, Silver, are fused at high temperatures, forming spherical beads which are difficult to dissolve. The temperature of the ore should never be allowed to exceed 1,000°, for this reason.

Roasting to Eliminate Arsenic and Antimony

H. M. Howe, in explaining how this is effected, distinguishes three horizontal zones in the ore:

1. the upper surface, where oxidation is only slightly hindered by sulphurous and sulphuric acids and by the products of combustion of the fuel;
2. the middle layers where oxidation proceeds to a very limited extent;
3. the lowest layers, where “a pellet of ore is simply exposed to the action of the other pellets with which it is in contact, of volatilised sulphur, and of sulphurous and sulphuric anhydrides generated by the action of sulphur on previously formed metallic oxides.”

He proceeds—“The expulsion of arsenic and antimony as sulphides is favoured in the middle and lower zones by the presence of volatilised sulphur, mixed with sulphurous acid and at most a very limited supply of free oxygen and sulphuric acid. In the upper part of the middle layer, to which a small amount of free oxygen penetrates, we have the gently oxidising conditions favourable to the formation of arsenious acid and trioxide of antimony. In the upper zone the stronger oxidising conditions rather favour the formation of fixed arseniates and antimoniates, though, even here, part of the arsenic and antimony may volatilise and escape while passing through their intermediate volatile condition of arsenious acid and trioxide of antimony.” On stirring the mass, these arseniates and antimoniates, being exposed to the reducing action of volatilised sulphur and undecomposed sulphides in the lower zones, may again be converted into volatile oxides. Protoxide of iron, suboxide of copper, and sulphurous acid are also efficacious in reducing arsenic acid, higher oxides of iron and copper, and sulphuric acid being formed. “ Thus, every individual atom of arsenic may travel forth and back many times through the volatile condition, being oxidised at the surface and reduced below the surface, and every time it arrives at this volatile condition an opportunity is offered it to volatilise and escape.” If a small quantity of coal or coke dust is mixed with the ore, after it has been completely oxidised, and the air excluded, the arseniates and antimoniates are again reduced to the lower oxides, and, if they are “carried past the volatile state,” i.e., reduced to metals, they may be again passed through it by an oxidising atmosphere. “Of course the expulsion of arsenic and antimony is favoured by the presence of a large proportion of pyrites, both because the sulphur distilled from the pyrites tends to drag them off as sulphides, and because the presence of the pyrites prolongs the roasting, and thus increases the number of times which the arsenic and antimony pass back and forth past their volatile conditions; hence, it is sometimes desirable to mix pyrites with impure ores to further the expulsion of their impurities.”

The Use of Salt in Roasting

Certain ores require the addition of salt in roasting in order to chloridise material which would otherwise absorb chlorine when the ore came to be “gassed,” and so cause additional expense as well as inconvenience. If silver as well as gold is to be extracted from the ore, the addition of salt is necessary in order to form chloride of silver in the furnace, since metallic silver is not attacked by chlorine at the highest temperature ever employed in the leaching vat. The silver chloride is then dissolved out by hyposulphite of soda or some other solvent either before or after the extraction of the gold.

Even if no silver is present, an ore must be roasted with salt if it contains much copper (as sulphide, or as an oxidised salt), lime, magnesia, or other substance which, after being subjected to an oxidising roasting, is rapidly attacked by chlorine at ordinary temperatures. The salt is usually added towards the end of the operation, when no sulphides and only a small percentage of sulphates are left undecomposed; sometimes, however, the ore and salt are mixed before charging-in. To some sulphides only 5 pounds of salt per ton of ore are added, but others require as much as 90 pounds per ton. The weight of salt added must be at least six to eight times that of the silver present in the ore. If a large amount of salt is used, it is desirable to leach the roasted ore with water, before treating it with chlorine gas, in order to remove the coating of soluble sulphates and chlorides remaining on the surface of the granules  of ore.

The chemical action of the salt is due to a double decomposition between it and the sulphates of the heavy metals, by which sulphate of soda and the chlorides of the heavy metals are produced. The following general equation approximately represents the reaction :—

2NaCl + RSO4 = RCl2 + Na2SO4

Chlorine is also set free by the action of sulphuric anhydride on salt, and the presence of water vapour induces the formation of much hydrochloric acid. These gases act directly on the several constituents in the ore, forming chlorides and oxychlorides. The metallic chlorides and oxychlorides formed are in many cases volatile (e.g., the compounds of copper, iron, lead, arsenic, antimony, &c.), and, in passing off, the volatile compounds carry away with them varying proportions of gold and silver, which, as a rule, are not recoverable in the dust chambers. The chloride of copper is especially active in causing these losses.

Other reactions which probably take place are as follows:

1. Ferrous sulphate, acted on by salt at a red heat in presence of air, yields hydrochloric acid and chlorine, which act on the gold and silver, while ferric sesquioxide and sodic sulphate are produced.
2. Ferric chloride, Fe2Cl6, is also produced at the same time. This is volatile, and chloridises silver with great energy at a red heat, sesquioxide of iron being produced.
3. Cupric chloride, CuCl2, is easily decomposed into cuprous chloride, Cu2Cl2, and free chlorine, or into the oxychloride, Cu2O . Cl2, and free chlorine. The vapours of CuCl2 thus give rise to further supplies of nascent chlorine available for the chlorination of the silver.
4. Arsenic and antimony form volatile chlorides which are decomposed by means of oxygen and water vapour, yielding arsenious and antimonious acids and nascent chlorine or hydrochloric acid.

It is thus obvious that the presence of base minerals is advantageous in that they may cause nascent chlorine to be set free in the presence of silver in all parts of the furnace. On the other hand, the loss of gold is increased by any increase in the quantities either of silver or of the base metals, since in the former case the time of the roasting is prolonged. The best chloridising effect is obtained in a highly oxidising atmosphere, so that very little sulphur is required in the ore, and, if much is present, the practice of eliminating the greater part before adding the salt is not likely to be attended with any diminution in the percentage of silver chloride formed. Moreover, the water vapour in the furnace gases promotes the formation of hydrochloric acid. When salt is used in roasting, the ore is often allowed to cool slowly in heaps after being withdrawn from the furnace. If treated in this way, a higher percentage of the silver, &c., is found to be chloridised than if the ore is wetted down at once, or even spread out to cool in a thin layer. On the other hand, the losses of gold by volatilisation are increased by slow cooling. Chlorine continues to be evolved for a long time after the withdrawal of the charge has taken place, the heaps smelling strongly of the gas.

Losses of Gold in Roasting

Plattner proved in 1856 that in the oxidising roasting of ordinary auriferous pyrites, a loss of gold can take place only when the operation is carried on so rapidly that fine particles are carried off mechanically by the draught. This conclusion, as far as sulphides and arsenides are concerned, has been confirmed by Kustel, and by Prof. S. B. Christy, but the latter adds that it is extremely difficult to prevent all mechanical loss by dusting, which is caused by even a moderate draught. Kustel records the loss of 20 per cent, of the gold present during the oxidising roasting of certain tellurides of gold and silver, and states that this is not a mechanical loss, but is due to volatilisation. This seems to be a mistake, see p. 7.

The losses of gold which are sustained when salt is added to the furnace charge may be very great. Kustel found that a telluride ore, on being roasted with 4 per cent. of salt, lost 8 per cent. of its gold before the ore was red hot. Aaron found that certain ores, consisting of simple pyrites, suffered great loss of gold in roasting with salt which had been added at the commencement of the operation; only a small part of this gold was condensed in the flue, in which was found a yellowish fluffy precipitate, consisting largely of chlorides of copper and iron, and containing nearly 30 ozs. of gold to the ton. He found that the loss was greatly reduced by diminishing the quantity of salt, and by reserving it until the dead roasting was nearly complete.

In the chloridising roasting of a Mexican ore, consisting mainly of magnetite and pyrites with 3.5 to 7 per cent, of chalcopyrite, C. A. Stetefeldt found the losses of gold to be from 42.8 to 93 per cent. of the total gold contained. He states that “ there is no doubt that the volatilisation of the gold takes place with that of the copper chlorides. The loss increased with the quantity of these chlorides formed and volatilised.” He further shows, however, that the presence of copper chloride is not the only possible cause of loss, since an ore consisting of hard white quartz, intimately mixed with about 7 per cent, of calcite and a little pyrites, lost 70 to 80 per cent, of its silver, and 68 to 85 per cent. of its gold, when roasted with 5 per cent, of salt. When subjected to an oxidising roast, no loss of gold took place. The reason for the extraordinary behaviour of this ore was not discovered.

Prof. Christy found that, in the ores on which he experimented on a small scale in a muffle furnace, a greater loss was sustained by adding the salt near the end of the roasting operation than by mixing the same weight of salt with the ore at the start. He explained that this is due to the fact that the amount of gold volatilised varies with the amount of chlorine which comes in contact with it. When the salt is added at the start, the chlorine is at first removed by the sulphur as fast as it is formed, escaping as chloride of sulphur, and thus the gold is protected from attack. When the salt is added after a long oxidising roast, the chlorine is rapidly generated (the ore being red hot and containing large quantities of sulphates), and the gold is no longer protected from attack by the sulphur. The loss of gold is also in all cases increased by working at a higher temperature, owing to the large amount of chlorine generated, and to the increase in the volatility of the gold. It is apparent from the results given on p. 23 that the temperature used in chloridising roasting must be very carefully regulated, the loss of gold being increased far more by high temperature than by a lengthening of the time in the furnace. Moreover, the salt must be reduced to the least possible quantity. It must, however, be remembered that the maximum volatility of gold chloride is at about 250°, or far below a red heat.

The advantage found to be gained in practice by adding salt near the end of the operation is due to the fact that, in the continuous roasting of ores in long-bedded furnaces, the gases given off from the finishing floor pass over a great length of comparatively cold, unsalted, and unoxidised ore before reaching the flue. The quantity of gold chloride mixed with the chlorine which is evolved from the red-hot ore as soon as the salt is added is no doubt large, but the SO2 from the colder ore, and the steam from the fuel, “offer excellent means for the reduction of the chloride of gold right within the furnace, while the most efficient means probably is the pyrites themselves,” which have been proved to be readily capable of condensing gold on their surface. If all the salt is added at the start, there is a continued volatilisation of chloride of gold throughout the furnace, and a less favourable opportunity for it to condense. The difference between the results in the muffle and in the reverberatory furnace is thus explained.

At Nevada City, at the Merrifield Mine, and in other works in the neighbourhood, the old-fashioned long furnace, with a single step separating the finishing hearth from the rest of the furnace, was still used in 1888. These furnaces are from 55 to 65 feet long, holding from 6 to 9 tons, and producing about 3 tons of roasted ore per day, so that the ore remains in the furnace from two to three days. The custom there was to give the ore a long oxidising roast at a low red heat, ending in a low cherry-red heat, and then, when the ore reached the finishing floor, the temperature was slightly lowered, and the salt added. The salt was stirred thoroughly into the ore, and as soon as it was “ dissolved” by the roasted ore—i.e., in about half an hour— the charge was drawn into the cooling pit. This lowering of the temperature is evidently of great importance in reducing the loss, while the duration of the roasting is regarded as less material, so long as no salt is present. These mills were on custom work, charging $15 to $17 per ton of ore for treatment, and guaranteeing a yield of 90 per cent, of the gold and 60 per cent, of the silver. Their method of roasting seems to be considered in California as that best suited to concentrates containing a high percentage of sulphur, but their loss in roasting has not been ascertained. The best method of roasting any particular ore, however, cannot be determined by any general rule, and exhaustive experiments must be made in every case before a definite course of procedure is finally adopted.

At one of the Californian chlorination mills it was found by experiment in 1882 that nearly 50 per cent, of the gold and 28 per cent, of the silver were being lost by volatilisation. In this case the pyrites was roasted on two hearths for thirty-six hours, 1 per cent, of salt being added four hours before the charge was drawn. The reason for the great loss was thought by Professor Christy to be the high temperature of roasting, particularly on the charging-in floor.

The variation of the loss in different ores which are treated precisely alike is doubtless due partly to the presence or absence of metals forming volatile chlorides which carry off the gold, and partly to the physical condition of the latter, the volatilisation being greater if it is in a state of minute subdivision.

Croasdale found, from experiments on a number of ores, that from 80 to 99 per cent, of the gold contained in them could be volatilised by roasting with salt.

Cooling Roasted Ore

This was formerly effected by spreading the roasted ore on sheet-iron floors, and either leaving it for twenty-four hours, or damping it down by a hose-pipe. More recently it has been passed through plain or multitubular cylinders cooled with water, or passed back by automatic rabbles above or below the stationary hearth furnaces, or allowed to fall from shelf to shelf through air in towers built on the Hasenclever principle.