Mining Engineer S. M. Anisimov.

Sampling of Tailings from Cyanide Plants.

From the Metallurgical Laboratory of the Tomsk Technological Institute.

(With 1 diagram).

In the practice of the cyanide process there are two methods for determining the percentage extraction of gold from gold-bearing materials (ores, sands, and slimes):

1. sampling of cyanide solutions entering the extractors for gold precipitation, and
2. sampling of tailings.

Owing to the technical difficulty of applying the first method, in most cases sand plants use the second method. In work on a plant scale, sampling of solutions requires a sample of solution corresponding, in quantity and gold content, to all the solutions obtained from the treatment of only one vat. It is impossible to isolate the work of a vat without detriment to the operation of the whole plant; therefore sand plants sample solutions before and after the extractors exclusively for control of the operation of the extractors. If the method of sampling solutions cannot be applied in the sand process, then in the slime process, where one deals with a finely ground and homogeneous pulp, carrying out such sampling presents no special difficulties. Sampling of solutions does not exclude simultaneous sampling of the tailings of slime treatment; on the contrary, comparison of the results of both samplings is the best control, excluding the possibility of errors. Thus, both methods are equally applicable in the practice of the cyanide process.

In simplicity of execution and in the attainable accuracy of determination, solution sampling leaves nothing better to be desired. The matter stands somewhat differently with tailings sampling. In sampling tailings, the ordinary method of dry fire assay is used for determining gold in gold-bearing ores, in which the gold is found in the majority of cases in metallic form.

In cyanide treatment of gold-bearing ores, the extraction of gold is never complete; the tailings always still contain some quantity of it. The reasons for insufficient extraction of gold, in most cases, consist either in insufficiently fine grinding, necessary in order to expose particles of gold, or in the limited duration of treatment, during which the larger particles of gold do not have time to dissolve in the potassium cyanide in the time allotted to the process. The final purpose of the treatment is to wash out the residue of dissolved gold. The quantity of wash waters, which is determined solely by economic considerations, is usually insufficient to remove completely all the dissolved gold. Thus, in the tailings, along with metallic gold in the form of particles enclosed by barren rock, or else in the form of coarse particles that have not had time to dissolve in the potassium cyanide, we shall have gold in the form of the complex KAu(CN)^2, insufficiently completely washed out of the tailings, or adsorbed by barren rock.

Thus, Johnston¹) gives an analysis of tailings in which, per one ton, there is:

*) 1 dwt equals 1.56 g.

The behavior of the complex $K\mathrm{Au}(\mathrm{CN})_2$ during the crucible fusion of tailings may be somewhat different from that of metallic gold.

In our laboratory it was observed that the percentage of gold extraction from sands and slimes, determined from solutions and tailings, never coincided. Usually the extraction according to the tailings proved greater than the extraction according to the solutions, by 2–6%, i.e., in the tailings not all the gold was amenable to accurate determination. This recurring discrepancy in all operations led to the thought that the usual method of sampling tailings is insufficiently accurate.

In the literature on this question there are various, often indefinite, indications. Thus, Bugbee, in his book Text-Book of Fire Assaying[^2], says that the gold complex $K\mathrm{Au}(\mathrm{CN})_2$ is volatile even at moderate temperatures. If one agrees with this supposition, then during the crucible fusion of tailings a loss of gold must occur, and the insufficient accuracy of the method is quite understandable.

According to Hamilton[^3], the difference in assay results has been established by laboratory and plant practice. The assay of tailings always shows a higher extraction than is actually obtained in the process. But, unfortunately, he does not indicate how great this difference is. Caldecott[^4] gives the following explanation of this phenomenon. The tailings contain the gold complex $K\mathrm{Au}(\mathrm{CN})_2$, which is dispersed in the finest particles throughout the whole mass of the tailings, so that in its state it approaches the molecular; therefore, in crucible fusion it is almost impossible to collect in the lead button all the gold present in such a form, since there are not enough reduced droplets of lead for them to come into contact with every molecule of gold. In order to verify these assumptions, work was undertaken in the laboratory to determine the volatility of the gold complex¹. In brief, the essence of the work is as follows.

A solution of the cyanide complex was evaporated on a water bath. The residue was dried at 110° and thoroughly ground in a mortar. The amount of gold was determined in part of the dry substance. From the remaining part, weighed portions of 100 mg each were taken. Each weighed portion was placed in a porcelain boat, which was heated in a Heraeus electric furnace within the range from 130° to 1012°. The temperature was measured with a calibrated Le Chatelier pyrometer. After heating at the indicated temperatures, the boat was removed from the furnace and weighed to determine the losses in weight arising from decomposition or volatilization. The beginning of the decomposition of the substance, as indicated by a change in its color, was noted at 150°–160°. At 180° the evolution of dense white vapors began, and this evolution proceeded most intensely at 700°–800°. Qualitative analysis showed that the vapors contained potassium cyanide. With a gradual rise in temperature, the appearance and color of the substance changed: gradual melting was observed and the appearance of a ruby coloration characteristic of metallic gold. At 700° the separation of metallic gold began, which at 800° and above was completely separated in the form of a metallic powder. Thus the course of decomposition of the gold complex with change of temperature was clarified. To determine the volatility of $K\mathrm{Au}(\mathrm{CN})_2$, weighed portions of 100 mg of the dry salt were taken and placed in a boat, on the bottom of which pure sea sand had been poured to prevent fusion of the substance with the boat. The boat with the sample was ignited at 972° for half an hour, then removed and weighed. All the gold separated out in metallic form. The contents of the boat were poured into a previously prepared charge and assayed by the crucible-fusion method. As a result of the determinations it was established that there is no volatilization of gold in the form of the complex $K\mathrm{Au}(\mathrm{CN})_2$. The results of the experiments are placed in Tables 1 and 2 and in Diagram I.

Table 1.

Table 2.

In view of the non-volatility of gold in the form of its complex, the explanation given by Caldecott appears the more probable. In favor of this explanation is also the fact that at Rand, where a difference has been established in the results of assaying tailings by different methods, in order to determine gold in the tailings they first precipitate or fix it by the method proposed by Christy for the analysis of cyanide solutions. Hamilton³) gives a description of this method, which was used by Whitby⁵).

The tailings are placed in a porcelain dish, and a little potassium cyanide solution is added to them; then, successively, solutions of CuSO₄, Na₂SO₃, and H₂SO₄ are added. The entire contents of the dish are mixed, and the solution is evaporated. The dried sample is passed through a “100” sieve and subjected to crucible fusion. At Rand it became customary to take exceptionally large quantities for analysis, from 5 to 10 assay tons, i.e., from 145.85 to 291.70 g.*) This precaution is of particular importance in the case of poor tailings, since discrepancies in the results of analysis may lead to serious discrepancies in the amount of gold obtained during a month’s treatment. Hamilton³) gives a comparison of the results of analyses of slime pulp (over 26 days) with preliminary fixation and without it. The average difference in the analyses for the daily sample was 1.4 dollars per 1 ton of ore. By the ordinary method a content of 13.90 dollars was obtained, while with preliminary fixation it was 15.31 dollars per 1 ton. At the same time Hamilton points out that the cited difference is exceptional. Unfortunately, none of the authors indicates to what extent the fixation method fulfills its purpose in the sense of accuracy in determining gold. For tailings, even a small error in the analysis gives a serious difference between the actual and theoretical extraction. How significant the discrepancy can be is shown by the following approximate calculation.

If the difference in extraction from tailings and solutions is taken on average as 3%, then, in the case of Uralian sands containing on average 1.5 zolotniks per 100 poods, with gold extraction of 60–70%, the indicated difference over an annual campaign (600,000 poods of sands) will amount to 1360 rubles.

From the foregoing it is clear that the rather important question of sampling tailings from cyanide plants has not been sufficiently fully clarified, and the aim of the present work was, as far as possible, to elucidate this question. In accordance with the explanation given by Caldecott, the following methods of sampling tailings were planned. Since the ordinary method of sampling does not give a sufficient quantity of lead globules, it seemed possible to increase the weight of the lead button either by increasing the quantity of reducing agent during fusion, or by introducing additional lead by washing the molten sample. Finally, a method is possible for preliminary fixation of gold in the tailings before their subsequent fusion. In accordance with this, the following plan of work was outlined:

1. preparation of artificial gold-bearing tailings and assaying them by the indicated methods;

2. obtaining natural tailings and testing the methods developed.

Preparation of Artificial Gold-Bearing Tailings.

Spongy gold precipitated from a solution of gold chloride, AuCl₃, by oxalic acid served for the experiments. In order to achieve a more rapid dissolution of the gold in potassium cyanide, the following procedure was used. 40 mg of gold in the form of sponge were placed in a cone of filter paper inserted into a small glass funnel. The cyanide solution from a glass bottle with a tubulure and a rubber tube with a clamp flowed drop by drop onto the gold sponge. The rate of fall of the drops was regulated by the clamp. The drops fell from

*) Each charge is fused in two or more crucibles; the resulting lead buttons are combined.

at such a rate that each preceding drop had time to wet the sponge, drain off completely, and expose the gold, so as to bring it into contact with oxygen, and only after that did the next drop fall. Under this method, conditions were obtained that were very favorable for the rapid dissolution of the gold. 40 mgr. of the latter dissolved completely in the course of 15 hours. The resulting solution of $KAu(CN)_2$ was diluted with water to two liters. For greater accuracy, the gold in it was determined by three methods:

1. by evaporation in a lead cup;

2. by precipitation with zinc dust;

3. by precipitation according to Christy’s method.

The last method consists in the following. A definite quantity of the solution under investigation is placed in a beaker; to it is added a solution of $CuSO_4$ in slight excess (until a faint blue coloration appears), a little solution of $Na_2SO_3$ and $H_2SO_4$. If the solution contains too little potassium cyanide, then before precipitating it must be strengthened to 0.1%. For 10 assay tons of solution = $291^{2}/_3$ cu. cm. Clennell⁷) recommends the following quantities of reagents:

$$ \begin{aligned} CuSO_4\ (10\%\ CuSO_4 \cdot 5H_2O) \ldots\ldots &— 20\ \text{cu. cm.}\\ Na_2SO_3\ (15\%) \ldots\ldots\ldots\ldots &— 20\ \text{cu. cm.}\\ H_2SO_4\ (10\%) \ldots\ldots\ldots\ldots &— 10\ \text{cu. cm.} \end{aligned} $$

The solutions are added in the indicated order, shaking after each addition. The mixture is placed on a sand bath and boiled until the solution above the precipitate becomes clear. To hasten clarification it is recommended to add several drops of a 5% solution of $K_4Fe(CN)_6$. The precipitate is filtered off, making sure that the solution is completely transparent; if the solution is not transparent, the latter is refiltered through the same filter. The precipitate is washed with water and dried. A portion of the fluxes listed below is placed in a small crucible; the dried precipitate is transferred together with the other portion of the fluxes and melted. The resulting lead button is cupelled.

Composition of the charge:

$$ \begin{aligned} &50\ \text{gr. litharge}\\ &30\ \text{gr. borax}\\ &3\ \text{gr. tartar}\\ &30\ \text{mgr. silver} \end{aligned} $$

W. H. Virgoe⁸) indicates that the dried precipitate may be scorified directly with assay lead. The results of this method do not differ from the results obtained by evaporation in a lead cup. The whole determination can be completed within two hours. A. Cross⁹) recommends Christy’s method as accurate, but prefers to fuse the precipitate with fluxes, scorify the lead button, and then cupel it. The theory of Christy’s method may be represented by the following equations⁹):

$$ 1.\quad 2CuSO_4 + Na_2SO_3 + H_2SO_4 + H_2O = Cu_2SO_4 + 2H_2SO_4 + Na_2SO_4 $$

$$ 2.\quad Cu_2SO_4 + 4KCN = Cu_2(CN)_2 \cdot 2KCN + K_2SO_4 $$

$$ 3.\quad 2KAu(CN)_2 + Cu_2(CN)_2 \cdot 2KCN + 2H_2SO_4 = $$

$$ = Au_2Cu_2(CN)_4^{*} + 2K_2SO_4 + 4HCN. $$

By the first method, evaporation in a lead cup, use was made for control, since the evaporation method is the most accurate of all known methods, with which all the others are compared. Among the few shortcomings of this method must be counted its great duration. Ras—

*) The precipitate $AuCu(CN)_2$ is soluble in excess $KCN$, but insoluble in $H_2SO_4$.

solutions containing free acid and salts of other metals, besides gold and silver, cannot be assayed by this method. On the contrary, the method with zinc dust is less limited in its application. It is equally suitable for rich and poor, contaminated and pure solutions; it permits taking different quantities of solution for analysis; it has sufficient speed; and it gives results very close to those of the evaporation method. Plazusin\(^9\) and Domenov\(^ {10}\), in comparing the methods of evaporation and precipitation with zinc dust, indicate that the method of precipitation with zinc dust gives results not inferior in accuracy to the evaporation method. Determinations of gold in solution by the methods indicated above gave the results presented in Table 3.

Table 3.

From the figures in the table it is evident that the evaporation and Christy methods give sufficiently close results. The small deviation of the results obtained by precipitation with zinc dust can be explained by the fact that the zinc dust used for precipitation contained much zinc oxide. The 1.5 g of zinc dust taken for the analysis produced only a very small lead button and, probably, did not completely precipitate the gold from the solution. Subsequently, therefore, 3 g of zinc dust were taken for precipitation.

As the average of the determinations, it was accepted that 50 cu. cm of solution contain 0.965 mg of gold, or 1 cu. cm contains 0.019 mg of gold. For experiments with artificial tailings, pure washed quartz was ground in a ball mill. To characterize the material, its sieve analysis and its SiO\(_2\) content are given below.

The sieve analysis was carried out by a combined method, eliminating the shortcomings of both the purely dry and the purely wet method. In the dry method, usually a considerable part of the coarser grains is incrusted with the finest particles, silt; therefore the separation of particles from one another according to grain size is insufficiently complete. In the purely wet method this shortcoming is absent, but the method requires much time for its execution. Frequent transfer of the material from one sieve to another in the wet method leads to unavoidable mechanical losses, which, of course, must affect the results of the analysis. In the combined method, the class “−200” is completely washed through the “200” sieve; the residue of the material +“200” is subjected to dry sieve analysis.

Three charges of the average sample, 100–200 g each, were subjected to sieve analysis. One by the manual method, the second by the mechanical method*) for 15 minutes, and the third by the mechanical method for 30 minutes. The results of the sieve analyses are given in Tables 4, 5, 6.

*) For mechanical screening, an apparatus manufactured by the firm W. S. Tyler, in Cleveland, was used.

Table 4. Manual sieve analysis.

Table 5. Mechanical sieve analysis over 15 min.

Table 6. Mechanical sieve analysis over 30 min.

The difference between mechanical and manual sieving is apparent only in the final size classes. In mechanical analysis the separation of grains into classes is more complete. The duration of the analysis is three times shorter than for manual analysis. The figures in the tables for mechanical sieving for 15 minutes and for 30 minutes are very close to one another; therefore all subsequent analyses were carried out for only 15 minutes. The content of \(\mathrm{SiO_2}\) in the sand, according to analysis, was 98.01%.

According to the results of the sieve analysis and the content of \(\mathrm{SiO_2}\), the sand corresponds very closely to pure quartz tailings. This material was used for all subsequent experiments. Since the experiments had to be conducted with the greatest possible accuracy, the materials intended for use were tested for the presence of gold in them. It turned out that the sand contained a visible, but unweighable, amount of gold. It was assumed that the gold was introduced into the sand sample by the litharge; therefore the litharge was also sampled. Gold was found in it, visible only with the aid of a lens. Analyses established that the amount of gold introduced into the sample could not affect the results of the determination. The foil, assay lead, and other reagents, in the quantities used for the analyses, contained no gold.

Sampling of artificial tailings containing from 30 dolyas to 2 zolotniks of gold in 100 poods.

Artificial tailings were prepared by mixing definite quantities of sand and a cyanide solution of gold.

1) Ordinary method of sampling.

Into a porcelain dish, a definite number of cubic centimeters of a cyanide solution of gold was added from an accurate burette to a small quantity of water, and 50–100 g of the average sand sample was poured into the solution. The sand gradually absorbed the solution, and the latter was uniformly distributed throughout the whole mass. Such an amount of water was taken that the total quantity of liquid in the sample corresponded to the moisture retained by the sand after percolation in a laboratory percolator. The sand with the solution was thoroughly mixed with a glass rod. The sample was carefully evaporated to dryness on a water bath. The dried tailings were mixed with fluxes and fused. If the lead button obtained was more than 25 g, it was first reduced to 10–15 g and cupelled.

Composition of fluxes*) for one sample, in percentages of the weight of sand:

150% litharge,
100–120% soda,
10–20% borax.

Reducing agent, \(\mathrm{C_4H_5O_6K}\)—quantity sufficient to obtain a lead button of not less than 28 g for a 50 g charge and 40 g for a 100 g charge of tailings**).

Variant of the ordinary method—increasing the weight of the lead button by increasing the addition of reducing agent.

The difference from the ordinary method consisted in the fact that the weight of the reduced lead was somewhat increased by using a larger quantity of reducing agent.

*) Composition of fluxes for normal quartz tailings.
**) Clennell\(^{7}\) recommends obtaining a lead button of 18–23 g for a sample of 1 assay ton and 25–30 g for 2 assay tons (pp. 58, 34), in order to ensure complete extraction of the gold.

2) Method with subsequent washing of slags.

The tailings were prepared and fused under conditions analogous to the preceding ones. After the fusion had been completely completed and the surface of the slags had become even and calm, the crucible was taken out of the furnace, cooled somewhat, and the top of the still uncooled assay charge was covered with a mixture of litharge, PbO, and reducing agent, \(C_4H_5O_6K\). The crucible was returned to the furnace. The charge was overheated and poured into a mold. The resulting lead button was scorified and cupelled. In the first experiments the crucible was not taken out of the furnace; the mixture of litharge with reducing agent for washing was added to the overheated charge; but since the temperature of the furnace was very high, the cream of tartar began to burn before it entered the charge, and considerably less lead was reduced for washing than was required by calculation; therefore the washing method had to be changed after the very first experiment. During fusion the crucible, from sharp temperature fluctuations, cracked and became wholly unfit after two fusions.

3) Method with preliminary fixation of the gold.

The sample of prepared tailings was evaporated; after removal of the greater part of the moisture, the following solutions were added to the sample in succession:

\[ \begin{aligned} &\mathrm{CuSO_4}\text{—}10\%\text{ solution } &&10\ \text{cu. cm.}\\ &\mathrm{Na_2SO_3}\text{—}15\%\text{ } &&10\ \text{”}\\ &\mathrm{H_2SO_4}\text{—}10\%\text{ } &&5\ \text{”} \end{aligned} \]

After each addition the solutions were mixed with the tailings. Then the sample was evaporated on a water bath to dryness. The dried tailings were poured into a previously prepared charge. The residue was collected with a spatula and a piece of filter paper, which was added to the sample. The tailings were mixed with fluxes and fused. The resulting work lead was scorified and cupelled. Each two samples tested by one method were fused in a separate crucible. The slags from all the samples were quite fluid and, in the cold state, separated well from the work lead. The results obtained for tailings with a content from 30 dolyas to 2 zolotniks per 100 poods are given in Tables 7, 8, 9, 10, 11.

Table 7. Results of assaying tailings with a content of 30 dolyas/100 poods.

*) Each sample of tailings was fused in two crucibles; the resulting work leads were combined and scorified down to 10–15 g, which were cupelled.

Table 8. Results of sampling tailings with a content of 47.5 zolotniks/100 poods.

Table 9. Results of sampling tailings with a content of 78 zolotniks/100 poods.

Table 10. Results of sampling tailings with a content of 0.99 zol./100 pud.

Table 11. Results of sampling tailings with a content of 2 zol./100 pud.

From the data in the tables it is evident that ordinary fusion of the tailings in all cases gives 3–7% less gold than was introduced into the assay. With an increase in the weight of litharge (Table 11), the percentage of gold extraction rises. From the data in the tables it is also evident that the more gold there is in the assay, the greater must be the weight of additionally introduced lead (the greater the weight of the wash) in order to achieve complete extraction of the gold, since with an increase in the weight of litharge the quantity of gold extracted by it increases; and when additional lead is introduced in an amount equal to the weight of the litharge obtained in ordinary fusion, the extraction of gold reaches 100%. The method of preliminary fixation of the gold in all cases gives complete extraction of the latter. From Tables 8, 9, and 10 it is evident that, in fixation and in washing, the extraction of gold exceeded 100%. These deviations are quite insignificant; they can be attributed only to the insufficient sensitivity of the gold balances (0.02 mg). If one recalls Caldecott’s explanation, it becomes clear why, at two zolotniks’ gold content, it was not possible to obtain complete extraction of the gold from the tailings. The gold complex is “molecularly” dispersed throughout the entire mass of the tailings—therefore not only the litharge from ordinary fusion, but even an additional washing with lead weighing 100% more, could not completely extract all the gold. The question naturally arises: by what method is it most advantageous to increase the quantity of lead in order to achieve complete extraction of the gold—by an additional washing, or by increasing the weight of litharge reduced directly in the simple fusion of the assay? From Table 11 it is evident that, with one and the same weight of litharge obtained by ordinary fusion with an additional quantity of reducing agent (56 g of lead) and by fusion with washing (52 g of lead), the percentage of extraction in the latter case is greater (by 2%). This argues that the subsequent washing of the slags, in which the “rain” of reduced lead passes through the entire mass of slags from the surface to the bottom of the crucible, ensures more complete contact of each drop of lead with each particle of gold.

In view of the absence of literature data on the behavior of the cyanide copper complex, $\mathrm{Cu_2(CN)_2 \cdot 2KCN}$, at high temperatures in mixtures with $\mathrm{PbO}$, $\mathrm{Na_2CO_3}$, $\mathrm{Na_2B_4O_7}$, and a reducing agent, it may be assumed that, in the fusion of tailings with previously fixed gold, the following processes take place.

When gold is precipitated from a solution of its complex, $\mathrm{KAu(CN)_2}$, by an excess quantity of copper*) beyond what is required for the compound $\mathrm{CuAu(CN)_2}$, the complex formed, $\mathrm{Cu_2(CN)_2 \cdot 2KCN}$, is uniformly distributed throughout the whole mass of the tailings, surrounding the particles of $\mathrm{CuAu(CN)_2}$. During subsequent fusion, $\mathrm{CuAu(CN)_2}$, decomposing with increasing temperature, gives an alloy $\mathrm{CuAu}$; at the same time $\mathrm{Cu_2(CN)_2 \cdot 2KCN}$ liberates metallic copper, which immediately alloys with the formed $\mathrm{CuAu}$ alloy. In this way the copper that separates out plays the role of a collector of gold. The particles of the alloy of gold with copper that are formed, coalescing, acquire considerable size, which greatly facilitates their extraction by the quantity of lead that is obtained in the ordinary method and, thus, the need for washing with an additional quantity of lead disappears. Comparing the methods of fixation and washing, it must be noted that by both methods complete extraction of gold is achieved and both methods, for their execution, require almost the same duration of time. In the method with preliminary fixation, for materials containing up to 1 zolotnik of gold per 100 poods, it is sufficient to add to the material under investigation about 20 cubic cm of a 5% solution of $\mathrm{CuSO_4}$; for materials richer in gold, it is necessary to add 40 cubic cm of a 5% solution, or

*) 10 cubic cm of a 10% solution of $\mathrm{CuSO_4 \cdot 5H_2O}$ contain 1 g of metallic copper.

the corresponding amount of a 10% solution of CuSO₄. In the method with washing, it is necessary to wash the slags with an amount of lead approximately equal to the weight of a normal verblyud. For rich tailings (more than 1 zol. per 100 puds), this amount may be insufficient; therefore, before using this method for sampling tailings with the indicated content, it is necessary first to determine the minimum amount of lead required for washing. Among the disadvantages of the method one must also include the fact that the service life of crucibles in this method is very limited: from sharp temperature fluctuations the crucible cracks, and frequent loss of samples is possible.

In order to make sure that the developed methods satisfy the stated purpose, it was decided to test them on tailings from laboratory percolation of plant sands. The sands were treated with potassium cyanide solution by the percolation method. The percentage of gold extraction was determined from the solutions and tailings by various methods, and the results were compared. For this purpose the following were taken:

1. sands from the Mariinsky mine, Miass district,
2. sands from the Avenirovsky mine, Northern Mariinsky taiga. For characterization of the sands, their sieve analyses and some physical and chemical data are given in Tables 12, 13, 14, and 15.

Table 12. Sieve analysis of Avenirovsky sands.

Table 13. Sieve analysis of Mariinsky sands.

Table 14. Consumption of reagents during percolation of tailings.

Table 15.

Preparation of natural tailings.

Mariinsk tailings.

2 kg of the average sample of tailings were mixed with 4.5 g of lime (the amount of CaO being taken on the basis of the useful consumption of alkali) and loaded into a laboratory-type percolator. The tailings were treated with the following solutions of potassium cyanide:

I solution of strength 0.25% KCN, 40% of the weight of the tailings.
II “ “ 0.1% “ 40% “ “ “
III “ “ 0.05% “ 35% “ “ “
IV wash water “ “ 30% “ “ “
V “ “ “ “ 30% “ “ “

The amount of the first solution was taken with the intention of obtaining a certain excess over the moisture retained by the tailings during percolation (33.85%). The strength of the solution was determined from the useful consumption of potassium cyanide plus a certain excess. The first solution was introduced from below by means of a separatory funnel with a tube connected to the lower end of the percolator by a long rubber tube. With this method of charging, the loaded material is washed by the solution much more uniformly; no voids or channels are formed in it that would disturb the regularity of percolation. The solution remained in contact with the tailings for 24 hours. After this time had elapsed, the clamp was opened to discharge the solution, and the tailings were drained for 24 hours. During this period the moist tailings became saturated with air. With the second solution the tailings remained in contact for 12 hours and were drained for 12 hours. The following solutions were introduced after 12 hours of drainage, in the order indicated above. Each subsequent solution was introduced after the preceding one had completely run off, in an amount equal to that introduced. The first run-off solution contained free potassium cyanide (0.03%) and alkali, which indicates that the percolation proceeded under normal conditions. All solutions after treatment were poured together and evaporated to a volume of two liters, from which samples were taken

for determining the percentage of extraction. After percolation, the tailings were washed with water, dried, screened through a “10” sieve, reduced with a Jones splitter to 1 kg, and ground to 120–150 openings. From the ground portion, charges were taken for assay.

In view of the fact that a total of one pood of assay lead had been used for the experiments, a new lead was taken and assayed. It was found to contain a larger amount of gold than the lead already used. Therefore the lead was purified of gold; for this purpose 2 kg of lead were cupelled with the following charge:

The reduced lead collected all the gold from the lead. The purified lead was ground to 150 openings and assayed. Traces of gold were found, which could not affect the results of the analyses. Gold in solutions was determined by evaporation in a lead cup and precipitation with zinc dust. The tailings were assayed by the previously developed methods under conditions identical with those for artificial tailings. Composition of fluxes per 100 g of tailings:

The results of all determinations are given in Table 16.

Table 16.

*) More borax was taken than usual because the ore contained a large amount of Al₂O₃.

Verification of the results obtained.

In 100 g of slime there is . . . . . . . . . . . 0.325 mg of gold
Extracted in the solutions . . . . . . . . . . . 0.256 » »
————————————————————
In 100 g of tailings there should remain . . . . . 0.069 » » 100%
Extracted from the tailings by: 1. Ordinary fusion . . 0.05 » » 72.4%
» » » 2. Fusion with subsequent washing . . . 0.063 » » 91.3%
» » » 3. Fusion with preliminary fixation . . . 0.07 » » 100.0%

Avenir slime.

The slime was treated under the same conditions as the preceding ones. The extraction was determined from the solutions and tailings. The results and details of the determinations made are given in Table 17.

Table 17.

Verification:

In 100 g of slime there is . . . . . . . . . . . 0.71 mg of gold
Extracted in the solutions . . . . . . . . . . . 0.58 » »
————————————————————
In 100 g of tailings there should remain . . . . . 0.13 » » 100%
Extracted from the tailings by: 1. Ordinary fusion . . 0.095 » » 73.1%
» » » 2. Fusion with subsequent washing . . . 0.115 » » 88.5%
» » » 3. Fusion with preliminary fixation . . . 0.13 » » 100%

As was to be expected, the difference in recovery by solutions and by tailings, when the latter were assayed by ordinary fusion, proved considerable: for the Avenirovsky sands, 5.8%; for the Mariinsky sands, 4.9%. Turning to the consideration of the figures for the recovery of gold from tailings obtained by the fusion method with fixation, we find in them confirmation of the conclusions drawn on the basis of experiments with artificial tailings. The recovery of gold by solutions and by tailings agrees completely. As regards the degree of accuracy of the method with subsequent washing of the slags, as we see, the results obtained were somewhat lower. This is explained by the same causes as in the case of artificial tailings. The quantity of lead that was taken for washing (50% of the weight of the lead reduced by the ordinary method) proved insufficient to collect all the gold. The method can be applied only in the case when the minimum quantity of lead for washing has first been established, which will ensure complete recovery of the gold.

On the basis of all the foregoing, the practical performance of methods of assaying tailings from cyanide plants, worked out according to the literature indications and according to the results of the experiments carried out, is presented in the following form.

A sample of tailings, sufficiently dried, is ground to 120–150 mesh. The finer the grinding, the closer will be the contact with the fluxes and the more rapidly and completely will slagging and the extraction of gold take place*). From the ground tailings, average samples weighing 100–150 g are taken.

Method with preliminary fixation.

The sample is placed in a porcelain dish; to it are added about one gram of potassium cyanide in solution, 20–30 cu. cm of a 5% solution of CuSO₄, 10–15 cu. cm of a 15% solution of Na₂SO₃, and 10–15 cu. cm of 10% H₂SO₄. After each addition of solution, the contents of the dish are thoroughly stirred with a glass rod. The sample is evaporated to dryness on a water bath and fused in one or two crucibles with fluxes taken in the following proportion:

The reducing agent should be taken in such quantity as to obtain a button of lead of not less than 45 g. The regulus obtained is scorified and cupelled.

Method of assaying with washing.

The fusion is carried out under conditions identical with the ordinary method of assaying gold-bearing ores. After the fusion has been completed and the slags have acquired an even, quiet surface, the crucible is removed from the furnace, cooled somewhat, and a mixture of litharge with reducing agent is poured on top. The quantity of lead for washing must be taken as not less than 100% of the weight of the regulus obtained by the ordinary assay method. The crucible is returned to the furnace, the sample is overheated and poured into the mold. The resulting button of lead is scorified and cupelled.

Conclusions.

In the present work the principal methods of assaying tailings from cyanide plants were tested. The investigation gave the following results.

1. In tailings from the cyanidation of ores, sands, and slimes, part of the gold is enclosed in the form of the complex KAu(CN)₂, not washed from the material or im—

*) Hamilton³) recommends finely grinding the litharge as well.

adsorbed. As a result of the “molecular” distribution within the mass of tailings, such gold, when the tailings are assayed by ordinary crucible fusion, is not completely collected in the lead button, and the gold content in the tailings proves to be lower than the actual value. As a consequence of this, for the percentage recovery of gold from pulps and other gold-bearing materials, the figure obtained from the tailings assay is 2–6% higher than the true one.

1. The most accurate methods for assaying tailings should be considered to be fusion of the tailings with subsequent washing, and the method with preliminary fixation of the gold according to Christy.

2. Of the two methods indicated, preference should be given to the method with preliminary fixation of the gold, as the simpler and quicker one.

3. In ordinary crucible fusion of tailings, the quantity of reduced lead globules is insufficient to bring about complete contact with all the molecules of the gold complex, or with the gold particles separated out during its decomposition, dispersed through the mass of the charge, as a result of which some of them escape determination. With preliminary fixation, the gold is converted from the molecular state into an alloy with copper; the particles of gold receive, so to speak, an additional load, increase in weight and volume, and thanks to this are completely extracted even by that quantity of metallic lead which is reduced in ordinary crucible fusion. In addition to lead, as a collector of gold, copper also plays a very substantial role in crucible fusion, being precipitated in the charge itself during its fusion, owing to the decomposition of the complex, $\mathrm{Cu}_2(\mathrm{CN})_2 \cdot 2\mathrm{KCN}$, introduced into the charge in an amount many times exceeding the quantity of the analogous gold complex. Copper, as is known, along with lead, is one of the best collectors of gold.

In conclusion, I consider it my pleasant duty to express my profound gratitude to Prof. V. Ya. Mostovich for his guidance, advice, and instructions during the performance of the present work.

Literature

1. A. M. Ar. Johnston. A Text-Book of Rand Metallurgical Practice (1913), 308.
2. Ed. Bugbee. A Text-Book of Fire Assaying (1922), 234.
3. E. M. Hamilton. Manual of Cyanidation (1920), 154, 155, 156, 52.
4. Proceedings of Chemical, Metall. and Mining Soc. of S. A., vol. II, 372.
5. M. Whitby. Journal Chem. Metal. and Min. Soc. of S. A., March, (1906)—the original work could not be obtained on site.
6. Trans. Amer. Inst. Min. Eng. Vol. XXVI, (1897), 735.
7. J. Clennell. Cyanide Hand.-Book (1910), 443, 335.
8. J. Clennell. The Chemistry of Cyanide Solutions (1910), 117.
9. V. Pazukhin. On the action of sulfides on solutions of gold in potassium cyanide. Zh. R. M. O. 1916.
10. V. Domenov, “Ural Technician” (1910) No. 6–7, 25.
11. V. Ya. Mostovich and V. A. Pazukhin. Investigation of gold-bearing ores in the Metallurgical Laboratory of T. T. I. (1918), 11.
12. V. Ya. Mostovich. The Assay Art (1913), 91.

Diagram

Decomposition curves of a mixture of $\mathrm{KAu(CN)_2}$, $\mathrm{KCN}$, and $\mathrm{K_2CO_3}$ during ignition from 130° to 1012°

For the article by S. Anisimov: “Sampling of tailings from cyanide plants.”

Vertical axis: Loss in mg.

Horizontal axis: °C →
130°, 160°, 180°, 200°, 260°, 300°, 360°, 400°, 510°, 615°, 722°, 820°, 921°, 1012°.

Curve labels shown on the diagram: 0.2, 0.4, 0.9, 1.1, 1.0, 0.8, 0.4, 2.3, 1.5, 1.2, 1.8, 1.0, 2.4, 6.2, 21.6, 19.4, 12.0, 11.4, 0.2.

1. The work was carried out by engineer Grekhov under the direction of Prof. V. Ya. Mostovich. ↩