Prof. V. Ya. Mostovich.

Gold Refining by Miller’s Process

(From the Metallurgical Laboratory of the I Tomsk Technological Institute).

(The article is accompanied by 1 drawing and 2 diagrams).

The refining and affining of low-grade alloys containing from 500 to 800 finenesses of gold—for example, gold scrap of 530–550 fineness, gold from the cyanide process and amalgamation of 600–800 fineness, etc.—on a small scale, in quantities not exceeding one or several kilograms, presents considerable difficulties. For this purpose the usual methods, such as the acid and electrolytic methods, are not directly suitable for reasons of a technical and economic nature.

These methods impose certain requirements on the composition of the gold to be refined, and for their successful and economical execution the low-grade gold must first be purified of the greater part of the base metals (ligature); moreover, the quantities of silver and gold in the alloy must be in a definite ratio, different for the individual methods².

Thus, when affining by means of concentrated sulfuric acid, the total content of ligature in the alloy must not be more than 10%; in particular, copper not more than \(6\text{–}7\tfrac{1}{2}\%\) (60–75 fineness), lead not more than 5% (50 fineness), and the most suitable gold content is considered to be 18–25% (180–250 fineness).

When affining by means of nitric acid, or so-called quartation, even considerable quantities of base metals in the alloy are not harmful to the process, provided that these metals dissolve readily in nitric acid. But usually for this method as well, in view of the high cost of nitric acid, the gold is first purified and the ligature is brought to a minimum, since base metals require considerably more acid for their dissolution than silver.

Thus, theoretically, for dissolution:

1 kg of silver requires 2.4 kg of nitric acid, sp. gr. 1.2
(containing 32% anhydrous \(\mathrm{HNO_3}\))

1 kg of copper requires 8.3 kg of nitric acid, sp. gr. 1.2

1 kg of zinc requires 7.6 kg of nitric acid, sp. gr. 1.2, etc.

The ratio of gold to silver in the alloy, for the possibility of their complete separation, must be from \(1:2\tfrac{1}{2}\) to \(1:2\), i.e., the gold content in the alloy must not exceed 33%.

For refining by electrolysis according to Wohlwill’s method[^2], the content of silver and lead in the anodes of crude gold must not exceed 15% in total, and in particular lead—not more than 5%. When an asymmetric alternating current, or so-called pulsating current, is used, the silver content in the anodes may reach 17% without harmful consequences for proper operation. But the greatest drawback of electrolytic refining of gold is the excessively long duration of the process and, consequently, the loss of interest on the gold tied up in the process.

Thus, refining a low-grade alloy to pure gold consists of the following seven operations: 1) remelting and bringing up the alloy, 2) casting the anodes, 3) electrolysis in a silver bath, 4) washing the anodic residues (slimes), 5) remelting them into gold anodes, 6) electrolysis in a gold bath, and, finally, 7) remelting the cathode gold.

All these seven operations, when working with pulsating current (in the sixth operation), require 10–12 days. When using direct current and working only during the daytime, electrolysis in the silver bath requires 16 days, in the gold bath 5 days, and the total duration of the process from the raw alloy to pure gold is 21 days.

To accelerate the process, efforts are made to eliminate electrolysis in the silver bath, as the most protracted operation, by bringing the raw alloy, by preliminary refining—for example, by Miller’s method—to a fineness of 970–980, and subjecting only such gold to electrolytic refining to a fineness of 999.9. All the methods considered for the refining and affining of gold alloys require for their execution, besides a considerable amount of time, also special, expensive apparatus, and therefore are economically expedient only when work is carried out on a large scale. For these reasons, the refining of gold alloys is usually performed at special plants and mints.

At the present time, on a large scale, refining is carried out only by three methods: with sulfuric acid, by electrolysis, and with chlorine. Of these methods, the predominant one in Europe is sulfuric acid; in America, electrolysis; in Australia, chlorine.

Approximately 10% of all gold is refined with chlorine, 15–20% by electrolysis, and about 70% with sulfuric acid.

At private affining plants they use almost exclusively the sulfuric-acid method, as the cheapest and, chiefly, the fastest; at mints, as government enterprises not especially interested in the percentages of gold tied up in the process, electrolysis predominates.

The simplest and fastest method of refining gold, especially suitable for work on a small scale and where acid is expensive and it is difficult to obtain the pure silver needed for quartation, is Miller’s method.

Miller’s method is based on the following fact.

When chlorine is passed through a liquid alloy consisting of gold, silver, and base metals, all the components of the alloy combine with chlorine in the order of the strength of their chemical affinity for chlorine. This affinity of metals for chlorine finds expression in the heats of formation of the corresponding metallic chlorides, and it is greater the greater the heat of formation of the chloride.

According to the strength of their affinity for chlorine, the components of the alloy are arranged in the order indicated in Table 1.

Table 1.

Thermal effect in kilogram-calories.

The cited heats of formation of metallic chlorides show that the metals, first, have a greater affinity for chlorine and, second, when the latter is passed through the alloy, they will burn out in the order indicated in the table; namely, tin, lead, iron, copper, silver, and only to an insignificant extent gold will combine successively with chlorine. So long as the alloy contains the indicated metals, they will absorb all the chlorine and thereby protect the gold from its action. We have a completely analogous phenomenon in the process of bessemerizing copper mattes, where ferrous sulfide, oxidizing and slagging in the form of iron silicate, protects the copper sulfides from oxidation.

But since gold predominates in the alloys being refined, along with the burning-out of the base metals and silver the formation of gold chlorides must inevitably take place. Of these, aurous chloride is highly volatile at all temperatures from 180° to 1100°, as was shown by Rose’s investigations¹. Thus, when pure gold was heated in a stream of chlorine under atmospheric pressure, Rose found that the loss of gold, owing to the formation of volatile aurous chloride, amounted at 805° to 0.5%, at 965° to 1.63%, and at 1100° to 1.93%. These data were confirmed by Prof. Christy¹, who studied the question of gold losses under the action of chlorine upon it. Consequently, in chlorinating gold alloys one might have expected considerable losses of gold; however, as the practice of refining with chlorine has shown, these losses are very small and do not exceed 0.01–0.02% of the weight of the gold. This discrepancy between the results obtained by Rose and Christy and the data from the practice of chlorination by Miller’s method is explained by the fact that, first, the investigations of Rose and Christy relate to pure gold, and not to its alloys with silver, copper, and other metals, and, second, by the difference between the conditions under which Rose and Christy carried out their experiments and those under which Miller’s process is conducted in practice. These conditions will be given below in the description of the process.

Thus, refining by Miller’s method consists, in brief, in passing chlorine under pressure through a molten alloy; the base metals and silver combine very vigorously with the chlorine, forming chlorides stable at high temperatures, while the gold remains in an unchanged state. Therefore chlorination is a very simple and reliable method of refining gold, by which both aims are achieved at the same time: the purification of gold from the ligature, i.e., refining proper, and the separation of silver from gold, or affination.

The first experiments in refining gold with chlorine were carried out on a small scale by Thompson and Aiken in 1838, but only in 1867 was this method developed and established for large-scale work by Miller at the mint in Sydney, Australia, where, under the special local conditions, the process proved very suitable. Imported acid was very expensive; Australian gold was of very high fineness and contained little silver; while in the local low-gold silver, needed for the parting of gold, there was a great shortage. Since then the method, which received the name of Miller’s process, began to be used at mints, and later also at some metallurgical plants in South Africa, Canada, the United States of North America, and England.

The reasons that led to the rapid spread of the process were the following great advantages over other methods of refining:

1) simplicity of execution, since the process requires only an inexpensive and universally available installation; it does not require silver, pure acids, or other reagents,

2) speed of execution,

3) very small losses of gold, not exceeding hundredths of a percent.

Practical execution of Miller’s process[^6].

The process consists of the following seven operations: melting the gold, chlorinating it, separating the refined gold from the chlorides, remelting the pure gold, de-golding the chlorides, reducing the silver from silver chloride, and remelting the metallic silver.

Melting the gold and chlorination.

For the work, an ordinary self-blowing coke or oil furnace is used, with a clear cross-section of \(10'' \times 10''\) or \(12'' \times 12''\) and a depth of \(20''\)—\(21''\). The gold is melted in a round fireclay Battersea crucible with a capacity of 15–20 kg of alloy (crucible height \(10 \tfrac{1}{2}''\) and internal diameter \(5''\)). To prevent loss of metal through accidental cracking or burning through of the fireclay crucible, it is placed inside an outer, somewhat wider graphite crucible, which serves as a safeguard and has the following dimensions: height \(8 \tfrac{1}{2}''\) and internal diameter—\(6''\). The graphite crucible is considered unsuitable for melting, since silver chloride, as it forms, is believed to be reduced by hydrogen evolved from the walls of the crucible (?). The fireclay crucible is first glazed on the inside with borax, to prevent absorption of silver chloride by its walls. The glazing is carried out by melting borax in the crucible, or in the following way. The crucible is filled with a boiling concentrated solution of borax and the latter is kept in the crucible for 10 minutes, after which the solution is poured out and the crucible is dried. On red heat the walls of the crucible become covered with a borax glaze and become impermeable to silver chloride.

The fireclay crucible is set in the furnace on a fireclay stand \(2 \tfrac{1}{2}\)—3 dm high, so that the upper rim of the crucible is level with, or only slightly below, the base of the flue. During operation the crucible is covered with a clay lid provided with two openings, one of which serves for introducing the tube supplying chlorine, the other for carrying off vapors and for observing the operation.

Chlorine is supplied through a refractory clay tube of internal diameter 3–5 mm. The lower end of the tube is cut wedge-shaped to facilitate the escape of gas when the tube rests on the bottom of the crucible. The alloy is charged into the crucible heated to red heat and, when it has completely melted, borax is melted onto it in a layer about 12–15 mm thick; for this, 60–90 g of fused borax is sufficient for 15–20 kg of alloy. The crucible is then covered with the lid, and the tube, the lower end of which has first been heated to red heat, is slowly introduced into the alloy. At the same time chlorine is started. When the tube has been lowered to the bottom of the crucible, the supply of chlorine is increased until a uniform stream of gas is established. Chlorination requires a continuous and uniform stream of chlorine under a pressure corresponding to the height of the layer of metal in the crucible. In alloys of 800–900 fineness, a column of metal in the crucible 1 dm high corresponds to a pressure of 16–18 dm of water column. With a charge of 20 kg, the height of the alloy layer in a crucible of the dimensions indicated above will be about 7 dm; therefore the chlorine pressure must amount to from 10 to 11 ft of water, or 22.3–25.5 cm of Hg column.

When chlorine is admitted, vapors of the chlorides of base metals at once begin to be evolved from the crucible; these vapors are especially dense if the alloy contains much zinc and lead. After some time, depending on the quantity of base metals in the alloy, the evolution of vapors almost ceases.

From this moment the chlorination of silver begins; the silver absorbs all the chlorine, and the latter, even with a strong jet, is not released from the crucible. The stronger the stream of chlorine, the faster the refining proceeds. Silver chloride forms, with the chlorides of the base metals, very low-melting, fluid alloys of low specific gravity, which collect above the gold beneath the layer of borax. Throughout the entire process the borax must cover the alloy; this is a very important detail of the process. In the vapors issuing from the crucible there is no silver chloride. Toward the end of the operation, when all the ligature and the greater part of the silver have passed into chlorides, splashing is observed in the crucible, and vapors again begin to be evolved from it, but denser ones, yellow-brown in color, consisting of free chlorine, colored by ferric chloride and sodium chloride ^7). The vapors become luminous and pass into tongues of red-brown flame. But all these phenomena by themselves do not serve as a reliable indication of the end of refining. The latter is recognized only by the special reddish or yellow-brown coating or precipitate which the vapors issuing from the crucible form on cold surfaces, for example on the cover of a porcelain crucible or on a piece of white tube brought into contact with them for several seconds. So long as coatings of another color are obtained, the gold has not been refined. This characteristic coating contains gold. Upon completion of the operation the stream of chlorine is stopped, the crucible is removed from the furnace and cooled on the furnace plate until the gold has completely solidified. The melting temperature of gold is 1064°; the melting temperature of silver chloride is 480°; cuprous chloride \((\mathrm{Cu}_2\mathrm{Cl}_2)\) is 434°; the alloy of chlorides formed in the process has an even lower melting temperature, and therefore, without fear of freezing the chlorides, the crucible may be cooled to a very low temperature, approximately to 400°.

Separation of the chlorides from the gold and remelting of the gold. The completely liquid chlorides together with the borax are poured into cast-iron molds. When the crucible is overturned, a cone of almost pure gold falls out of it; this is remelted in a graphite crucible and cast into ingots. In view of the poisonous properties of the vapors of metallic chlorides, the work should be carried out under a strong draft.

In this way, about 98% of all the gold that was in the alloy is obtained in refined form, of 995 fineness; the remaining 2% of the gold is contained in the chlorides, partly in the form of entrained prills of gold and partly in the form of double chlorides of silver and gold.

Degolding of the chlorides. To extract the gold from the chlorides, they are melted in a fire-clay crucible, covered with a layer of molten borax 5–6 mm thick, and the required quantity of anhydrous soda is sprinkled onto the borax in small portions, without mixing it with the contents of the crucible; from time to time, however, with the aid of an iron needle (rod), the soda is made to sink into the chlorides. The soda reduces silver chloride according to the reaction:

\[ 2\mathrm{AgCl} + \mathrm{Na}_2\mathrm{CO}_3 = 2\mathrm{Ag} + 2\mathrm{NaCl} + \mathrm{CO}_2 + \mathrm{O} \]

and the prills of metallic silver, falling to the bottom of the crucible, extract gold from the chlorides and degold them. After all the soda has been added and the reaction of silver reduction has ended, the crucible is covered with a lid, the melting is continued for another 10–20 minutes, and, when the contents of the crucible have become completely fluid, the crucible is removed from the furnace, cooled until the bead of gold-bearing silver has solidified, and the still-liquid chlorides are poured off from the latter. By this method it is possible to lower the gold content in the silver obtained in the subsequent renewed treatment of the chlorides to 0.5–2 fineness units. For complete degolding of the silver chloride and for obtaining from its subsequent reduction perfectly pure silver containing no more than traces of gold,

it is necessary to subject the chlorides to repeated degolding by means of a small addition of soda, according to the method described above. The gold content in the silver can then be lowered to 0.1–0.2 fineness units (0.01–0.02%).

Recovery of silver from chlorides.

The degolded chlorides consist chiefly of silver chloride, since by Miller’s process alloys are usually refined which contain not less than 88–90% gold, not more than 10–12% silver, and base metals; the quantity of the latter does not exceed 1–2%.

The silver chloride is cast into slabs \(12'' \times 10'' \times 5/4''\), which are washed in hot water to separate the copper chlorides.

The slabs of silver chloride are set in a vertical position between plates of metallic zinc measuring \(14 \times 12 \times 1/2''\), in such a way that each plate of silver chloride is placed between two zinc plates at a distance of about \(1/4\) in. from the latter. The slabs of silver chloride and the zinc plates are brought into metallic contact by means of silver ribbons. The whole system, resembling an electrolytic installation, is placed in a tank with dilute hydrochloric acid, to which a little zinc chloride has been added. Owing to the galvanic current arising in the system, the silver chloride is gradually reduced to spongy metallic silver, while the silver slabs retain their original form; the zinc, however, dissolves. If cuprous chloride has remained in the silver chloride, then both metals are reduced simultaneously and cuprous silver is obtained. After all reaction has ceased, the latter is boiled first in acidified water, then in pure water, dried, melted, and cast into ingots. In this operation the following reactions take place:

\[ 2\mathrm{AgCl} + \mathrm{Zn} = \mathrm{ZnCl}_2 + 2\mathrm{Ag} + 38640\ \text{cal.} \]

\[ \mathrm{Cu}_2\mathrm{Cl}_2 + \mathrm{Zn} = \mathrm{ZnCl}_2 + 2\mathrm{Cu} + 31650\ \text{”} \]

The reactions are accompanied by the liberation of a considerable quantity of heat, so that the solution becomes heated and no warming of the solution is required.

According to theory, 22.81 parts by weight of zinc are required for 100 parts by weight of silver chloride; in practice the consumption of zinc amounts to 25% of the weight of the silver chloride.

Instead of zinc, at some mints—for example, in Melbourne and Ottawa—sheet iron is used. The plates of silver chloride are placed in contact with iron plates in a bath of dilute hydrochloric acid. The bath is heated with steam. Cuprous chloride dissolves and, on coming into contact with the iron, is reduced by it to cement copper, which settles on the bottom of the bath; as a result, the recovered silver is obtained in a much purer form than when zinc is used, and on remelting it does not give off more poisonous vapors (\(\mathrm{Cu}_2\mathrm{Cl}_2\)).

Such, in general outline, is Miller’s process, which makes it possible to bring the fineness of the refined gold to 990–995 and to obtain recovered silver of fineness 995–998.

The process has found application at the mints in Sydney, Melbourne (Australia), Pretoria (Transvaal), and Ottawa (Canada), and various improvements have gradually been introduced into it.

It remains to dwell somewhat more fully on the details of the process.

Source of chlorine. Before liquefied chlorine in steel cylinders appeared on the market, chlorination was carried out with chlorine produced in special generator apparatus from common salt, hydrochloric acid, and manganese dioxide. Later, chlorine began to be used directly from cylinders with liquid chlorine, which simplified and cheapened the work.

Thus, for example, at the mint in Ottawa \(^{8}\) there is the following installation. A cylinder containing 100 kg of chlorine under a pressure of 97 lb per sq. in. (6.2 atmospheres) is located in a special room communicating, on one side, with an exhaust flue from the furnace leading to a condensation chamber and, on the other side, with the outside air by means of a special pipe. The chlorine from the cylinder is supplied to the crucible through a chain consisting of lead and rubber tubes and glass stopcocks. For chlorination a continuous and uniform stream of chlorine is required, under a pressure of 5–6 lb per sq. in. (25.7–31 cm Hg column), which corresponds to the resistance of a column of metal in the crucible 7 in. high. Since a reducing valve cannot be used for chlorine, the following pressure regulator is employed to attain this condition. The cylinder is connected with a lead vessel having the form of a Woulfe bottle. A glass tube is lowered through the middle neck to the bottom of the vessel, and outside the vessel it is connected with a lead tube. The latter, in turn, communicates with a lead bulb in the form of a large pipette. The Woulfe vessel contains enough water to give a liquid column 11 ft high, corresponding to a pressure of 5 lb per 1 sq. in. When the pressure of the chlorine exceeds this limit, all the water from the Woulfe vessel is forced by the gas pressure into the lead pipette, whereby the excess gas is freely removed into the chain and enters it continuously at a constant pressure equal to 11 ft of water column, or 5 lb per 1 sq. in.

The crucible. In all installations for the chlorination of gold, a fireclay crucible is used, glazed on the inside with borax and covered during operation with a lid. The lid is provided with two openings, each \(3/4\) in. in diameter, and is concave on the inner side in the form of a cup. For convenience, instead of a round opening for the chlorine-supply tube, a recess is made in the edge of the crucible lid, through which the tube can be lowered into and removed from the crucible; this makes it possible to remove the crucible lid without removing the tube.

At one time, from 10 to 22 kg are chlorinated in a single crucible, depending on the fineness of the alloy; in Pretoria, with a gold fineness of 840, the charge is 10 kg; in Sydney, with a gold fineness of 900, it is 22 kg. The length of the tube for supplying chlorine is 24 in., with an internal diameter from \(1/8\) to \(3/16\) in. The quantity of fused borax charged into the crucible is about 90 g per 18 kg of alloy.

The duration of chlorination needed to bring the gold to 995 fineness depends on the content of silver and base metals in it. When the latter are present in considerable quantity, the duration of the operation can be shortened by forcing the gas supply.

Thus, for example, in Sydney the chlorination of 18.6 kg of alloy containing 90% gold and 10% silver (i.e., fineness 900) to a fineness of 995 required 1.5 hours. In Melbourne the chlorination of 20–22.3 kg of gold with 2% silver and 0.3% base metals (gold fineness 975) to a fineness of 995 required 1.5 hours; the chlorination of the same quantity of gold with 3.5% silver and 1.5% ligature, i.e. of fineness 950, to a fineness of 995 required 2 hours. In Ottawa the chlorination of an ingot weighing 15.5 kg with 16% silver and 28% ligature (fineness 560) to a gold fineness of 995 continues for seven hours.

In any case, the sole criterion for determining the end of the operation should be the color of the deposit on a piece of clean tube brought for several seconds into contact with the vapors issuing from the crucible. The deposit should be yellow- or reddish-brown. It should be noted, however, that recognizing the end of the operation by this indication requires considerable experience, and this criterion is not reliable in all cases. The color of the deposit depends on the composition of the alloy being refined. If the amount of ligature in the initial alloy is insignificant and it consists only of gold with 10–15% silver, then toward the end of refining a clear reddish-brown deposit is obtained. With a large quantity of ligature in the alloy, the volatile chlorides of base metals, for example copper,

having dissolved in the silver chloride, are only slowly liberated from it during the whole operation, but toward the end of the process they are carried away by the chlorine that is no longer being absorbed. These chlorides strongly mask the color of the coating and make it difficult to determine the end of refining. Thus, Begeer ⁹), who introduced chlorination at the mint in Pretoria, says that he never succeeded in recognizing the end of the operation by the color of the coating.

When chlorinating high-standard gold containing about 90% gold and 10–12% silver, the chlorides formed consist chiefly of silver chloride with a small admixture of cuprous chloride.

For the de-golding of silver chloride, soda is used in an amount of 7–9% of its weight. Usually, in one operation, 7 kg of silver chloride are treated.

If less than 7% soda is taken, too much gold still remains in the silver chloride; with 9% soda, the gold obtained is too silvery.

Thus, with the addition of 480 g of soda to 6.9 kg of silver chloride, or 7% soda, 900–990 g of gold of fineness 940–950 were obtained; in the silver recovered from the silver chloride there remained from 1 to 2, and sometimes up to 6, parts of gold (0.1–0.6%). With 540 g, or 7.8%, of soda per 6.9 kg of silver chloride, 900–1050 g of gold of fineness 920–930 were obtained; the recovered silver contained 0.5–1 part of gold (0.05–0.1%).

With 600 g of soda per 6.9 kg of silver chloride, or 8.7%, 1050 g of gold of fineness 870–880 were obtained; the recovered silver contained 0.2–0.5 parts of gold (0.02–0.05%). Thus, the most suitable amount of soda is 480–600 g per 6900 g of silver chloride, or 7–9% of the weight of the silver chloride. For complete de-golding of the silver chloride, the operation of reducing the latter should be repeated, for which remelting the chloride with 1½% soda is sufficient.

As regards the applicability of Miller’s process, technically it is suitable for refining alloys with any gold content, but economically, in large-scale work, the process has advantages for refining only comparatively high-standard gold, containing not less than 850–880 parts of gold and not more than 120–150 parts of silver and base metals. With low-standard gold the process becomes prolonged and more expensive, since the productivity of the installation falls, the direct yield of refined gold decreases, gold losses increase, and thus Miller’s process loses its principal advantages over the ordinary sulfuric-acid method.

The first large-scale application of Miller’s process was found in the refining of Australian gold, both alluvial and from amalgamation. This gold is of very high standard. Thus, the gold refined in Sydney in 1895 contained 832.1 parts of gold, 135.5 parts of silver, and only 32.4 parts of alloying constituents. The average fineness of the refined gold for the period 1884–1895 was 995.9 parts of gold and 4.1 parts of silver. The yield of refined gold directly after chlorination was 98%, and 2% of the gold passed into the silver chloride, from which the gold was extracted by smelting with soda. The de-golded and recovered silver contained 970 parts of silver and only traces, up to 0.2 parts, of gold; the remainder was copper. The total loss of gold in the entire process amounted to 0.11 to 0.19 per 1000, or 0.011 to 0.019%; the loss of silver (1895) was 4.25%.

At the mint in Melbourne, during the period 1891–1895, gold having the following composition was refined: gold—937.8, silver—49.6, and alloying constituents—12.7 parts. The average fineness of the refined gold for the same period was 995.9, and its total loss 0.0170%.

In 1904 the gold had the following composition: gold 923, silver 42.5, and ligature 34.5 fineness. The gold was brought up to 994.9 fineness. In one operation, in one crucible, from 20.1 to 21.3 kg of gold were chlorinated. Melting lasted 1.5 hours, chlorination to 995 fineness—2 hours. After this the crucible was cooled on the furnace board for 7 minutes, the chlorides with borax were poured into molds, the crucible was broken, and the cone of gold was remelted in a protective graphite crucible and cast into ingots. The broken crucibles were ground in a Chilean mill, and the gold washed from them was added to the refined gold at the end of the day’s work. Per day, 2790 kg of gold containing 10% silver were refined. The chlorides, containing from 5 to 10% gold, were melted in a graphite crucible of 93 kg capacity, and the de-golding was carried out in the following manner. To the molten chlorides 7% sodium carbonate was added, without mixing with the chlorides. Since a single addition of soda did not succeed in achieving a sufficiently complete separation of the gold, a second addition of soda was made without removing the crucible from the furnace. After this the crucible was cooled until the precipitated auriferous silver had solidified, while the chlorides were poured into molds. The auriferous silver contained 40–60% gold and was subjected to refining. This silver and the purified gold contained 99.95% of all the gold put into the operation; the remaining gold remained on the walls of the crucible. The yield of refined gold immediately after chlorination was 97.6%; the amount of gold remaining in the twice de-golded and reduced silver was 0.005 to 0.01%. The chloride slabs, each weighing 7.75 kg, consisted of silver chloride, copper \((\mathrm{Cu}_2\mathrm{Cl}_2)\), and 7% sodium chloride, obtained from the reaction of silver chloride with soda. Each ingot was sewn into a flannel bag and boiled in water for 4–5 days. Under the action of air and water the slabs became covered with a green precipitate of cupric chloride \((\mathrm{Cu}_2\mathrm{Cl}_2)\) and copper oxychloride \((\mathrm{Cu}_2\mathrm{O}\cdot\mathrm{Cu}_2\mathrm{Cl}_2)\). On boiling in water, a solution of sodium chloride was formed, in which the indicated copper compounds dissolved rather easily. The slabs, purified of most of the copper, were placed between iron plates \(1/8\) dm thick, like electrodes, so that they did not touch one another but were connected by a metallic conductor. The entire system was immersed in an iron tank filled with an aqueous solution of ferric chloride. The solution was heated with live steam. The reduction of silver chloride under these conditions ended in 24 hours. Owing to the flannel bags, the copper was reduced not in the bags but outside them, did not adhere to the silver, and did not contaminate it. Upon completion of the reduction, the spongy silver was washed in hot water and remelted with 1–5% saltpeter. Silver of 990 fineness was obtained and contained no more than 0.05–0.1 fineness of gold.

At the mint in Pretoria, owing to the high cost of acid, Miller’s process was temporarily installed. Gold of three sorts arrived at the mint: from amalgamation, containing 840–910 fineness of gold and 50–110 fineness of silver, and from chlorination—930 fineness of gold with traces of silver. Only high-grade gold from amalgamation and chlorination was chlorinated. A charge of 10.3–12.4 kg of gold of 840–930 fineness was melted in a round Battersea fire-clay crucible, glazed inside with borax. The gold was covered with a layer of borax \(1/2\) dm thick and, after chlorination for 6 hours, yielded metal with a content of 990 fineness. After cooling the crucible until the gold froze, the chlorides were poured into iron molds and then remelted with an addition of soda. The resulting auriferous silver contained only a few finenesses of gold and, for separation, was sent to England. With two furnaces, about 300 kg of refined gold were obtained per month.

With still greater success, chlorination was used in order to make brittle mint gold (916 fineness) malleable. In one operation 37.3 kg were refined. The molten gold was covered with molten borax and chlorinated for 15 minutes, after which a sample was taken for testing.

In concluding his communication, Begeer says that Miller’s process is not suitable for refining low-grade gold from the cyanide process, for the following reasons:

1. The considerable content of silver and base metals necessitates an excessively long duration of chlorination, which is unprofitable and impractical.

2. Long before the end of refining, the chlorides of base metals formed in large quantity corrode and destroy the crucible.

In view of this, only high-grade gold containing no more than 11% silver should be subjected to chlorination.

Begeer gives no precise data on the magnitude of the losses of gold and silver, limiting himself merely to the statement that the losses of gold are very small, while those of silver are large.

Thus, from the foregoing it is evident that where Miller’s process was established, it was used for refining only high-grade gold of a fineness not below 850.

On the other hand, Pearson, describing the installation of Miller’s process at the mint in Ottawa (Canada), says that the process is the most suitable for separating from gold even large quantities of silver and base metals. And indeed, in Ottawa, 15.5 kg of gold of 560 fineness are refined in one operation, although chlorination up to 895 fineness lasts seven hours. Unfortunately, I did not have in my hands Pearson’s original article; in the brief abstract published in Metallurg. Chemic. Eng. for 1915 (vol. XIII, 508), there are no indications of the magnitude of the losses of gold and silver. In the list of apparatus for the process, a condensing chamber is noted. This compels one to suppose that there are losses of gold through volatilization owing to the great duration of chlorination, caused by the low fineness of the alloy.

The quantity of gold passing into the chlorides in Miller’s process depends on the fineness of the initial gold. When chlorinating Australian gold of 900 fineness and higher, 2% of the gold passes into the chlorides; with more argentiferous Californian gold, according to Price’s experiments, from 5 to 10%. After reduction of the silver chloride with zinc and its dissolution in nitric acid, the gold remains in the form of small flakes, and not rounded globules, which, according to Miller, would be the case if all the gold had been carried into the upper layer of chlorides only mechanically by the action of the chlorine jet. This shows that part of the gold is dissolved in the chlorides in the form of a double chloride with silver chloride. The latter, being only slightly volatile, retains the volatile gold chloride in solution and prevents losses of the latter. These losses, according to data from work in Sydney and Melbourne, do not exceed 0.01–0.02% of gold. In addition to the above, other reasons accounting for such small losses of gold are the following:

1. The chlorides containing gold in solution are, throughout the entire process, covered by a layer of borax, through which the vapors of the chlorides, in particular of gold, not possessing sufficient elasticity, are unable to penetrate.

2. Silver and base metals, as having a greater affinity for chlorine, absorb it and protect the gold from the action of chlorine.

3. Since chlorine passes through the layer of gold in the form of small bubbles, the surface of gold exposed to the action of chlorine is very small.

4. Free chlorine begins to leave the alloy only toward the end of refining, and throughout the entire operation the gold is covered by a layer of chlorides, and above them by a layer of molten and fairly viscous borax.

5. The process is stopped before all the silver has been removed, and in refined gold there always remains at least 5 fineness parts of silver (0.5%). It is difficult to say what the losses of gold would be if chlorination were carried to

complete removal of silver; in any case, given the great volatility of gold chloride, these losses would be very large, which is why in practice refining is everywhere carried out to no higher than 990–995 fineness.

All the circumstances indicated make it clear why, in the chlorination of gold-bearing alloys, the total losses of gold do not exceed hundredths of a percent and do not reach the considerable values that one would have expected on the basis of the investigations by Chisty and Rose on the volatility of gold in an atmosphere of chlorine.

The metallurgical laboratory has periodically received proposals to subject to refining low-grade gold containing no more than 550–600 fineness of gold, about 100–200 fineness of silver, and 300–350 fineness of copper. The gold was delivered in small quantities, from 1.5 to 2 kg, and it was required to bring its fineness up to 920–930. The use of acid and electrolytic methods for refining this gold was entirely excluded, both because the laboratory lacked the necessary apparatus and materials and because of the requirement for the gold to be realized as rapidly as possible, approximately within twenty-four hours, in ingots of a definite fineness. It was therefore decided to carry out the refining by Miller’s process, as requiring the simplest apparatus, an insignificant consumption of reagents, and being easily practicable.

And indeed, for the process only an ordinary assay furnace, crucibles, refractory tubes, borax, soda, and chlorine were necessary. The latter was available in the laboratory in compressed form in cylinders. About 20 ingots of various finenesses, from 910 to 990, were refined. In several operations, thermal measurements and a complete accounting of the gold were made, with the aim of determining the losses of the latter and certain details of the process².

As indicated above, Miller’s process has found preferential application for refining only high-grade gold, and in the literature there are not sufficient indications of the suitability of the process for highly cupriferous low-grade gold containing little silver and not less than 30% copper.

Therefore the publication of the results obtained and of certain observations will probably be of interest to persons engaged in gold refining. The apparatus shown in Fig. 1 served for the work. The gold was melted in a Morgan graphite crucible of No. 3 or 5. A crucible was selected of such capacity that the alloy occupied not more than one third of its volume, since during chlorination the volume of the alloy increased greatly. The crucible was covered with a fireclay lid, in which two round holes 10–15 mm in diameter were drilled, one for introducing the tube and the other for carrying off vapors. The crucible was placed on a fireclay support on the grate of a natural-draft furnace, which was fired with coke. Chlorine was drawn from the cylinder, and for supplying it to the crucible there served a lead tube with an internal diameter of 8 mm, opening into one neck of a Woulff bottle. Into the middle neck, down to the bottom of the bottle, a glass tube with an internal diameter of 4 mm and a height of 80 cm was lowered, serving as a manometer; from the third neck of the Woulff bottle the gas was led off through a glass tube and then through a rubber tube, one end of which was connected with a quartz tube with an internal diameter of 4 mm and a length of 65 cm. Concentrated sulfuric acid was contained in the Woulff bottle. The velocity of the chlorine jet was regulated by a simple outlet valve at the cylinder. The gas pressure was maintained at 20–22 cm of sulfuric-acid column in the manometer tube, which corresponded to 36.8–40.5 cm of water column and was balanced by a layer of metal in the crucible from 2½ to 3½ cm high.

Procedure.

The gold received for refining was remelted in a graphite crucible and stirred with a carbon stirrer. From the liquid alloy, by means of a graphite sampling spoon, a sample was taken for analysis; the alloy was cast into an ingot, which was accurately weighed, after which it was again charged into a crucible heated to red heat, on the bottom and walls of which borax had been melted. The whole operation was carried out in a graphite crucible, since the laboratory did not have a fireclay one which, in form and size, would correspond to the graphite crucible as a protective vessel. Experiments in chlorination directly in a graphite crucible showed that the use of the latter is associated with the following major inconveniences. The process is greatly slowed down; the crucible is rather quickly and unevenly corroded, as a result of which the walls of the crucible acquire an uneven surface, which promotes the adhesion to them of prills of metal. The graphite becomes entangled in the chlorides and greatly hampers their further treatment. In the subsequent de-golding and de-silvering, graphite completely excludes the possibility of obtaining high-standard gold and silver, since, upon precipitation with soda, along with silver, much copper is reduced, which strongly dilutes the alloy. In this case the following reactions take place: $\mathrm{Cu_2Cl_2 + Na_2CO_3 = Cu_2O + 2NaCl + CO_2}$; $\mathrm{Cu_2O + C}$ (graphite) $\mathrm{= 2Cu + CO}$.

Replacing soda with metallic zinc does not improve matters, since zinc also reduces $\mathrm{Cu_2Cl_2}$ and $\mathrm{AgCl}$. The electrolytic separation of silver from chlorides containing graphite is likewise greatly hampered. Plates cast from molten chlorides are very brittle and rapidly disintegrate in the bath. In addition, graphite noticeably weakens the action of the current and slows the reduction of silver. In view of these circumstances, for chlorination one should use exclusively fireclay crucibles glazed on the inside with borax.

The molten gold was covered with a layer of borax $1—1\tfrac{1}{2}$ cm thick, for which about 60—80 g of molten borax were charged into the crucible per 1.5 kg of alloy. Then, through the opening in the crucible lid, a quartz tube was introduced into the alloy until it touched the bottom of the crucible. The lower part of the tube had previously been heated to red heat². At the same time the introduction of chlorine was begun. From the very start of the operation, thick yellow vapors began to be evolved from the crucible, consisting mainly of cuprous chloride. Gradually the evolution of vapors weakened, but it did not cease during the whole operation. Toward the end of refining the quantity of vapors again increased; they became thicker and darker and, finally, passed into tongues of dark-red, strongly fuming flame. At this moment the following test was made. A porcelain plate (the crucible lid) or a tube of white clay, held in tongs, was introduced for several seconds into the vapors issuing from the crucible. On the plate there formed a dark-brown deposit with a reddish tint, which, after the plate cooled, turned the violet color of carnation. This deposit began to appear rather early, when the fineness of the alloy sample reached 910—920, and, as the fineness of the gold rose, the intensity of the coloration only increased. Thus, the color of the deposit by itself could not serve as a sufficiently reliable criterion either for determining the approximate fineness of the gold or for recognizing the end of its refining, i.e., the moment when the gold content in the alloy reached a fineness of 990—995. A considerable amount of copper in the vapors strongly altered the color of the deposit, giving it various shades, from brown to black. Since the purpose of refining was to bring the gold to a fineness of 910—930, subsequently only by experiment was the minimum duration of chlorination to the required fineness established, and the operation was carried out by time. At a chlorine pressure of 37—40

cm of water column, chlorination of 400 g (1 pound) of alloy from fineness 550–560 to gold of fineness 950–960 required thirty minutes.

In view of the absence of clear signs, it was impossible to chlorinate to a definite fineness, and at first the operation was ended at different gold finenesses, from 910 to 920; only later was the duration of chlorination to the required fineness established experimentally.

At the end of the operation the tube was removed from the crucible, and after this the supply of chlorine was stopped. The crucible was covered with a new and intact lid and cooled on the furnace board until the gold solidified, which can easily be determined by probing the bottom of the crucible with an iron needle (rod). The melt of chlorides above the solidified gold remains completely liquid for a long time. Therefore the crucible may be cooled without danger of solidification of the chlorides until the evolution of extremely unpleasant vapors ceases. The chlorides, together with borax, were poured into shallow cast-iron pans, where they solidified in the form of cakes. The crucible with the gold remaining in it was again placed in the furnace; the gold was remelted, poured into a mold, cleaned in dilute sulfuric acid, weighed, and assayed. The chlorides were charged into the same crucible and, after melting, were subjected to de-golding by means of anhydrous soda. The latter was charged into the crucible onto the chlorides in an amount equal to half that required for complete de-golding of the chlorides. The soda was not mixed with the chlorides, but was pushed through the layer of borax with the aid of an iron needle. After the first addition of soda the crucible was somewhat superheated and, after about ten minutes, the remaining soda was charged. Then the crucible was heated until its contents were completely melted and, after taking the crucible out of the furnace, they were poured into a mold. At the bottom of the mold there collected an ingot of highly copper-bearing silver, which had extracted practically all the gold from the chlorides. This ingot was added during refining to the next portion of gold. It would have been simpler to separate Au, Ag, and Cu by means of nitric acid, since the gold content in the copper-bearing silver did not exceed 20%, but this, in addition to the consumption of acid, would have delayed the delivery of the gold. The de-golded chlorides were not subjected to further treatment, since, apart from copper and sodium chloride, they contained too little silver.

For de-golding silver chloride, which constitutes the main mass of the chlorides obtained in refining high-grade gold, from 7 to 9% soda, based on the weight of the silver chloride, is used; one gram of soda gives about two grams of silver. But since in our case low-grade gold was refined, the chlorides consisted chiefly of cuprous chloride with a small amount of silver chloride. One gram of soda gave about 1 g of copper-bearing silver, according to the reaction:

\[ 2\mathrm{AgCl} + \mathrm{Na}_2\mathrm{CO}_3 = 2\mathrm{Ag} + 2\mathrm{NaCl} + \mathrm{CO}_2 + \mathrm{O}, \]

1 g of soda gives 2.7 g of silver. In view of the predominance of cuprous chloride in the chlorides, the greater part of the soda was consumed in converting \(\mathrm{Cu}_2\mathrm{Cl}_2\) into cuprous oxide according to the reaction:

\[ \mathrm{Cu}_2\mathrm{Cl}_2 + \mathrm{Na}_2\mathrm{CO}_3 = \mathrm{Cu}_2\mathrm{O} + 2\mathrm{NaCl} + \mathrm{CO}_2. \]

Cuprous oxide, separating from the melt, made it extremely viscous, foamy, and refractory; part of the cuprous oxide, by the action of carbon from the walls of the crucible, was reduced to metallic copper. The latter alloyed with the silver and made it highly copper-bearing. The reduction of silver chloride to metal and the conversion of cuprous chloride into cuprous oxide and then into metal—all these reactions proceeded in parallel; therefore, in order not to obtain silver that was too copper-bearing, it was necessary to take care to reduce only such an amount of it as would ensure sufficiently complete de-golding of the chlorides.

After assaying, the refined gold was alloyed with the calculated amount of copper or gold-bearing silver in order to obtain the required fineness.

Below are given examples of several chlorination operations.

Melt No. 47.

Duration of the individual operations.

1. Melting the gold, taking a sample, and casting the ingot . . . . . — hr. 30 min.
2. Remelting the gold, addition of auriferous silver from melt No. 46 and borax . . . . . . . . . . . . . . . . . . . . . . . . . — hr. 30 min.
3. Chlorination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 hr. 42 min.
4. Remelting the gold and decolorizing the chlorides . . . . . . . . . . 1 hr. — min.

Total duration of refining . . . . . . . . . . . . . . . . . . . . . . . 4 hr. — min.

77.5 g of auriferous silver from melt No. 46, fineness 175.2, was added.
For decolorizing the chlorides, 100 g of soda was used.
Weight of gold before remelting: 1523 g.
Weight “ after “: 1495.31 g.

Yield of gold:

Charged into the melt: 1495.31 g of gold of fineness . . . . . . . . . . . . . 550.2 = 822.7 g.
77.5 g of auriferous silver of fineness . . . . . . . . . . . . . . . . . . . . . 77.5 = 13.58 “

Total pure gold . . . 836.28 “

Obtained after chlorination:

refined gold 882.22 g, fineness . . . . . . . . . . 932.34 = 822.54 g, pure gold.
auriferous silver 105 g, “ . . . . . . . . . . . . . . 82.8 = 8.64 “ “ “

Total pure gold 831.23 g.

The total loss amounts to 836.28 − 831.23 = 5.05 g, or

\[ \frac{5.05}{836.28}\,100 = 0.6\%. \]

Distribution of the gold:

passed into refined gold . . . . . . . . . . . . . . . . . . . . . . . . . . . 98.35%
“ “ auriferous silver . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.06%
“ “ loss . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.6%

100,—

Melt No. 44.

Duration of the individual operations:

1. Melting the gold, taking a sample, and casting the ingot . . . . . — hr. 30 min.
2. Remelting the gold and addition of borax . . . . . . . . . . . . . — “ 25 “
3. Chlorination . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 “ 40 “
4. Remelting the gold and decolorizing the chlorides . . . . . . . . — “ 45 “

Total duration of refining . . . . . . . . . . . . . . . . . . . . . . 4 “ 20 “

For decolorizing the chlorides, 95 g of soda was used, which was added in two portions.

Weight of gold before remelting . . . . . . . . . . . . . . . . . . . . 2121.5 g.
“ “ after “ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2100.0 “

Yield of gold.

Charged into the melt: 2100 g of gold of fineness 535.8 = 1125 g pure gold.

Obtained after chlorination:

1144 g of gold of fineness 962.3 = . . . . . . . . . . . . . . . . . . . 1101 g.
100 “ auriferous silver of fineness 194.3 = . . . . . . . . . . . . . . . . 19.43 g.

Total . . . . . 1120.43 g.

The total loss amounts to 1125 − 1120.43 = 4.57 g of gold, or

\[ \frac{4.57}{1125}\cdot 100 = 0.406\%. \]

Distribution of the gold:

Melt No. 50.

Weight of gold after remelting—1580.7 gr.

Yield of gold:

• Obtained after chlorination:

Overall loss of gold—880.6 − 872.3 = 8.3 gr., or

\[ \frac{8.3}{880.6}\cdot 100 = 0.943\%. \]

Distribution of the gold:

Melt No. 46.

Weight of gold after remelting—1698.5 gr.

Duration of chlorination: 2 h. 40 min.

Yield of gold.

Obtained after chlorination:

Overall loss of gold = 931.73 − 926.46 = 5.27 gr., or

\[ \frac{5.27}{931.73}\cdot 100 = 0.566\%. \]

Distribution of the gold:

Melt No. 0.

Weight of gold after remelting: 1034 g.

Duration of chlorination: 1 h. 20 min.
During chlorination, 133 g of auriferous silver of fineness 131.5 were added.
For deaurizing the chlorides, 120 g of soda were used.

Gold yield.

Received after chlorination.

Overall loss of gold:

\[ 623.9 - 618.34 = 5.56 \text{ g, or } \frac{5.56}{623.9}\cdot 100 = 0.891\%. \]

Distribution of the gold.

Melt No. 38.

Weight of gold after remelting—2185 g.

Duration of chlorination—3 hours.
During chlorination, 121 g of auriferous silver of fineness 36.75 were added.
For deaurizing the chlorides, 80 g of soda were used.

Gold yield.

Received after chlorination.

Overall loss of gold:

\[ 1292.5 - 1290.02 = 2.48 \text{ g, or } \frac{2.48}{1292.5}\cdot 100 = 0.192\%. \]

Distribution of the gold.

Melt No. 41/42.

Weight of gold after remelting — 1433.3 g.

Gold yield:

Overall gold loss = 881.82 − 880.9 = 0.9 g, or

\[ \frac{0.9}{881.82}\cdot 100 = 0.104\%. \]

Distribution of the gold.

Melt No. 17.

Weight of gold after remelting — 1557.6 g.

The de-golding of the chlorides was carried out twice, yielding two ingots of silver. The first weighed 113.5 g, fineness 35, and the second weighed 112.5 g, fineness 3.81. The de-golded silver chloride, after complete reduction with soda, yielded 224 g of silver, fineness 0.46.

Gold yield.

Obtained after chlorination:

Overall gold loss = 873.8 − 872.8 = 1 g, or

\[ \frac{1}{873.8}\cdot 100 = 0.1144\%. \]

Distribution of the gold.

In an analogous manner, several more alloys of low-grade gold were refined, and Table II gives the results of melts for which, under the operating conditions, it was possible to carry out the necessary analyses.

Table II.

Results of refining with chlorine.

In the Miller process the most interesting and important point was to establish the yield of high-grade gold and its loss. In this respect the alloys investigated may be divided into two groups: the first eight, in which copper predominated, and the last two, in which silver predominated.

From the data for the first group of alloys it is evident that the percentage yield of high-grade gold is the lower, the higher the assay of gold to which chlorination is carried. If chlorination is carried to an assay of 960, the yield of high-grade gold is about 98%; at higher final assays the yield is considerably lower.

In chlorinating alloys of the second group, i.e., those containing little copper, the direct yield of gold is considerably greater and, up to a fineness of 920, amounts to not less than 99%. Unfortunately, I did not have at my disposal a larger number of analogous alloys, and it is difficult to state precisely what the yield of refined gold would have been if the chlorination of these alloys had been carried to the same finenesses as for the alloys of the first group. According to practical data, when silver-bearing gold is chlorinated (true, with a gold fineness not lower than 880–900) to a fineness of 995, about 2% of the gold passes into the chlorides. From the data given in the table it is evident that the more gold passes into the chlorides, the higher the final fineness of the latter, i.e., the longer the process. This proves that, along with copper and silver, gold is also chlorinated, and the chloride of the latter dissolves in cuprous and silver chlorides.

The next very important circumstance in the Miller process is the loss of gold. According to practical data, when alloys containing 880–900 finenesses of gold and not more than 100–120 finenesses of silver are chlorinated to a fineness of 990–995, the total losses of gold do not exceed 0.01–0.02%. According to our experiments, when low-grade coppery gold is chlorinated, the losses of the latter reach considerably higher values, even if refined to no higher than 900–960 finenesses. Gold losses may be divided into mechanical and chemical losses. The former are caused by the fact that globules of gold adhere to the walls of the crucible. These losses are not large, and the gold may be recovered from the used crucibles after crushing and washing them. Chemical losses are caused by the volatilization of gold chloride. The magnitude of these losses depends on the composition of the alloy, the temperature, and the duration of chlorination. The more copper there is in the alloy, the longer the operation, the higher the temperature developed by the combustion of copper into cuprous chloride, and the longer the period of high temperatures. All these circumstances contribute to strong evaporation of cuprous chloride and to its carrying away the gold dissolved in the chlorides. Cuprous chloride begins to volatilize already at so low a temperature as 340°¹⁰), which is why abundant evolution of its vapors is observed from the moment chlorine is introduced and continues throughout the entire operation, increasing as the temperature rises. The volatilization of gold is proved by the appearance of a characteristic violet coating on cold surfaces introduced into the vapors evolved from the crucible. This coating is strongly masked by the deposit of cuprous chloride, but it can already be observed at a fairly early stage of the process, when the gold has been brought to a fineness of 910–93°. Therefore the appearance of the characteristic coating during the chlorination of strongly coppery gold cannot serve as an indication of the end of refining. In chlorinating silver-bearing high-grade gold with a low copper content, different phenomena occur. Here gold chloride is also formed, but it is held in solution by an excess of slightly volatile silver chloride; moreover, owing to the small amount of silver and the small thermal effect of its chlorination, the temperature developed during the process is considerably lower than with coppery alloys. Therefore the losses of gold are very small. Only when almost all the silver has passed into the silver chloride does the absorption of chlorine cease, and the latter, penetrating through the melt of chlorides, carries gold chloride along with it. The losses of the latter through volatilization rapidly increase and could reach inadmissible limits if chlorination were not stopped in time—namely, when not all the silver has yet been removed. This is why, in order to avoid losses of gold, the latter is brought only to a fineness of 990–995, leaving in it still about 5–10 finenesses (0.5–1%) of silver.

Thus, in chlorinating alloys, silver protects gold from volatilization, while copper, because of the strong volatility of its chloride, promotes it.

If it is required to bring gold in refining to a fineness of 990–995, then a high silver content in the initial alloy is likewise undesira-

...undesirable, since this prolongs the process, as a result of which too much gold passes into the silver chloride and the direct yield of metal is reduced. Thus, in refining Australian gold with 10–11% silver, no more than 2% of all the gold passes into the silver chloride; in refining highly argentiferous Californian gold, from 5 to 10% of the latter remains in the silver chloride.

All these circumstances explain why Miller’s process has found almost exclusive application for refining high-grade gold with a low content of silver and base metals.

As for low-grade alloys with a considerable copper content, with respect to them, both in the economic sense (duration, crucible consumption) and from the standpoint of gold losses, the process is very expedient and practical only for partial refining, for example, to finenesses of 920–950.

When work is carried out on a large scale, total gold losses are reduced by periodically extracting it both from spent crucibles and from dust and vapors, for the collection and condensation of which dust-collecting chambers are installed. But on a small laboratory scale these methods are difficult to implement, and, especially in periodic work, one has to put up with losses many times greater than those at plants.

In two melts, Nos. 47 and 50, measurements were made of the temperatures of the alloys from the moment chlorine was admitted to the end of the operation. For this purpose, through a special opening in the crucible cover, a Le Chatelier pyrometer in a protective quartz tube was lowered into the liquid alloy, the tube resting on the bottom of the crucible. The pyrometer was connected to a Siemens-Halske millivoltmeter, and its readings were recorded every five minutes. At the same time, the moments of coke charging, changes in chlorine pressure, and the character of the vapors issuing from the crucible were noted.

Observations of temperature as a function of time, the material balance of the melts, and the details of the experiments are presented in diagrams 1 and 2.

From the diagrams we see that at the moment chlorine was admitted the temperature of the alloys was 1100–1120°, i.e., the alloys had been superheated by 40–50° above the melting temperature of pure gold (1064°). As soon as chlorine was admitted, the temperature of the alloys began to rise rapidly, owing to the strongly exothermic reactions of formation of the chlorides of copper and silver. For alloy No. 42 the temperature reached a maximum of 1332° after 20 minutes; for alloy No. 50, 1302° after 50 minutes from the start of the operation. Thus, the chlorination of copper and silver superheated the alloys by almost 240–270° above the melting temperature of gold.

As is known, a self-blast furnace is one of the least perfect melting devices, both in respect to utilization and to conservation of heat, which makes it impossible to maintain a more or less constant temperature in the furnace. The coefficient of heat utilization in the furnace does not exceed 2%, and such circumstances as the charging of a small amount of cold coke, a slight change in draft intensity, and the opening of the furnace cover rapidly and sharply change the temperature in the furnace, causing corresponding changes of temperature in the crucible. This circumstance also explains the irregular appearance of the temperature curves and the occurrence on them of sharp maxima and equally sharp minima. The temperature of the alloy at any given moment depends both on the temperature of the furnace space outside the crucible and on the number of calories evolved in the crucible during the chlorination of the components of the alloy. If the sharp maxima and minima are excluded from the temperature curves, then the average temperature of the alloy in melt No. 47 was 1270–1280°, and in melt No. 50, 1280–1290°; these temperatures were maintained during the greater part of the process. Thus, the average tem-

temperatures in both melts were almost identical. This circumstance indicates that in both melts there occurred the burning out of almost identical quantities of “fuel,” which in the alloys were copper and silver. From the data of the tables placed on the diagrams, the following calculations can be made.

Since, in the formation of cuprous chloride, 517.7 cal. are liberated per 1 g of copper, and in the formation of silver chloride, 272.0 cal. per 1 g of silver, the total heat input per minute was, for the melt:

\[ \begin{aligned} \text{No. 47}\quad & (517.7 \times 4.22) + (272 \times 2.13) = 2764.1\ \text{cal.}\\ \text{No. 50}\quad & (517.7 \times 4.16) + (272 \times 2.33) = 2787.8\ \text{cal.} \end{aligned} \]

A comparison of the quantities of heat liberated per unit time in melts No. 47 and 50 fully confirms the conclusion made on the basis of the temperature curves of both operations: almost identical quantities of heat developed during the process determined the similarity of the temperature curves.

After the appearance of dense yellow-brown vapors, the temperature of the alloys in both melts began to fall rather rapidly. By this moment the greater part of the copper and silver had burned out, and the end of the operation was near. Since in melt No. 47 chlorination was stopped at a higher temperature (1220°) than in melt No. 50 (1180°), consequently more copper and silver should have remained in alloy No. 47 than in alloy No. 50. The analysis of the final alloys shown on the diagrams confirms this conclusion as well, which can be drawn on the basis of the temperature curves.

All these circumstances show that the temperature curves are very characteristic of the chlorination process, and that temperature observations make it possible to control the process.

It is clear that the temperature curves would have had a more regular form if a gas or electric furnace had been used for heating the crucible. In that case, the temperature of the medium in which the crucible is heated would have been constant and would not have affected the temperature of the alloy during its chlorination.

Conclusions.

1) Miller’s process is a very simple and convenient method for refining low-grade gold alloys with a considerable copper content, suitable for work on a small scale, where the usual methods of refining and affining are not applicable for economic and technical reasons.

2) In the process, losses of gold occur, caused chiefly by the volatilization of gold chloride.

3) Losses from this cause are the greater, the lower the initial fineness of the alloy, the more copper it contains, and the higher its final fineness.

4) With an increase in the quantity of ligature, the duration of chlorination increases, as do the temperature developed during it and the formation of volatile chlorides, for example cuprous chloride, which carry gold with them.

5) Therefore, in order to avoid considerable losses of gold in the case of low-grade and highly cupriferous alloys, Miller’s process should be applied only for the partial refining of the latter, approximately to a fineness of 920–930.

6) The yield of refined gold is inversely proportional to the content of silver and ligature in the initial alloy. In refining to a fineness of 920–930, not more than 2% of the gold passes into the chlorides.

7) In chlorinating the above-mentioned alloys, the temperature reaches 1330°.

8) Chlorination should be carried out in a fireclay crucible, glazed on the inside with borax, and placed in an outer graphite crucible as a protective one. A graphite crucible slows the operation and causes great difficulties in de-silvering and de-golding the chlorides. For introducing chlorine, tubes made of refractory clay are more convenient, as being more resistant, and not quartz tubes, which are destroyed rather quickly.

Literature.

1. Rose. The Metallurgy of Gold, V Edition, 389.
   E. Kern. Metallurgical Chemical Eng. vol. IX, 1911, 343.
   Titus Ulke. Mineral Industry, vol. IV, 1895, 343.

2. Rose Trans. Inst. Min. Met. (London), vol. XXIV, 1915, p. XXXV.

3. I. Richards. Metallurgische Berechnungen, 21.
   M. Pavlov. Journal of the Russian Metallurgical Society, 1911, No. 4, 434.

4. See No. 1—p. 22.

5. Christy. Trans. Amer. Inst. Min. Eng. Vol. XVII, 1889, 19.

6. See No. 1, p. 407.
   J. Percy. Die Metallurgie des Silbers und Goldes.

7. Christy, see No. 5, p. 32.

8. R. Pearson. July Bull. of the Canadian Mining Institute. Abstract: Metallurg. Chemic. Eng. Vol. XIII, 1915, 508.

9. B. Begeer. The Metallurgy of Gold on the Rand, p. 113.

10. Kothny. Oester. Berg. u Hüttenmän. Jahrb. LVIII, 1910, 141.

to the article by V. Ya. Motovich: “Gold Refining by Miller’s Process”

Fig. 1.

Fig. 9. Ya. Ya. Mostovich. “Gold Refining by Miller’s Process.”

Diagram II.

Vertical scale: Temperature, °C.
Horizontal scale: Minutes.

Graph annotations visible on the diagram:

• Chlorine admission.
• Slag formation and melting.
• Beginning of gold boiling.
• Coke charging.
• Added 19–20 st. of \(H_2SO_4\), dense.
• Coke charging.
• Chlorine admission.
• Coke charging.
• Coke charging.
• Chlorine admission.
• Coke charging.
• Appearance of thick yellow-brown fumes.
• Dense yellow-brown fumes.
• Added 21–22 st. of \(H_2SO_4\).
• Coke charging.
• Pressure 18 st.
• Coke charging.
• Cessation of chlorine admission.

Duration of chlorination: 120 minutes.

To the article by V. Ya. Mostovich, “Gold Refining by Miller’s Method.”

Diagram I.

Temperature, °C

Minutes

• Chlorine admission.
• The lightest vapors.
• Loading of coke.
• Chlorinated gold.
• Loading of coke.
• Addition of silver and loading of coke.
• Appearance of brown vapors.
• Dense yellow-brown vapors.
• Loading of coke.
• Appearance of orange flame.
• Cessation of chlorine admission.

1. Since the process takes place at a temperature above 1000° C, copper forms only cuprous chloride \((\mathrm{Cu}_2\mathrm{Cl}_2)\); cupric chloride, beginning at 340°, dissociates: \(2\mathrm{CuCl}_2 = \mathrm{Cu}_2\mathrm{Cl}_2 + \mathrm{Cl}_2\). ↩↩

2. Quartz tubes are rapidly destroyed, and therefore it is better to use tubes of refractory clay; however, the author did not have such tubes at his disposal. ↩↩↩