Action of potassium cyanide solutions on New Zealand gold and silver.

By J. S.^* Maclaurin, B.Sc., University College, Auckland, New Zealand.

The present contribution is a continuation of a paper printed in the Transactions for 1893 (63, 724), in which the action between gold and potassium cyanide, the basis of the now extensively applied Macarthur-Forrest or Cassel process, was investigated, and in which it was proved, 1st, that oxygen is necessary for the dissolution of gold in potassium cyanide solutions, and that it combines with the potassium of the potassium cyanide in the proportions required by the equation 4Au + 8KCN + 2OH₂ + O₂ = 4AuCN,KON + 4KOH. 2nd. That the rate of dissolution of gold in potassium cyanide solutions varies with the concentration; increasing as the solution becomes more dilute until it reaches a maximum at about 5 per cent., below which the rate of dissolution of the gold falls off, and that this remarkable variation may be explained by the fact that the solubility of oxygen in such solutions decreases as the concentration increases, whereby the solvent power of the strong solutions is rendered less than that of the weaker solutions which are capable of taking up more oxygen.

As these results appeared to warrant further investigation, I continued my experiments, as detailed in the following pages.

In the paper already referred to, an experiment to prove the necessity of oxygen for the dissolution of gold in potassium cyanide is described, and it is shown that when precautions were taken to exclude oxygen, a gold plate lost only 0.0002 gram in 24 hours, whilst, when the same solution was exposed to the air, the plate lost 0.00835 gram in the same time. In order to get more convincing proof on this point, I prepared gold paper by steeping filter paper in a solution of gold trichloride, containing 1/10 th per cent of gold, suspending the moist paper horizontally over ammonia, and reducing the oxide thus formed by immersion in a hot solution of oxalic acid (Skey, Trans. N.Z. Inst., 25, 383). After washing and drying, the paper had a uniform pink tint. A piece, ½ in. square, containing about 0.00002 gram of gold, was introduced into the limb of a Dumas bulb which had been two-thirds filled with a 5 per cent, solution of potassium cyanide. The end of the limb was then drawn out to a small diameter, and the cyanide solution boiled briskly during an hour, after which the boiling was considerably slackened and the point of the limb sealed with the blowpipe; when the solution had cooled, the gold paper was shaken into it. In the first experiment, the colour in the paper faded before the limb was sealed. I concluded that this was due to the combined action of oxygen and hydrocyanic acid, the latter being evolved in small quantity by the boiling solution, and condensing in the cold part of the limb where the gold paper was placed. In a second experiment, I therefore kept the limb hot enough to prevent condensation of the hydrocyanic acid; this proved entirely successful, as, after the limb had been sealed, the gold paper, when shaken into the cooled solution, appeared to have lost none of its colour. Next day, however, the colour had faded a little, but it required eight days to make the gold paper as colourless as a piece of plain filter paper which had been introduced along with it for comparison. Or, in other words, it required eight days to dissolve 0.00002 gram of gold. The point of the limb was now broken, the solution well shaken to saturate it with air, and a piece of gold paper, in every respect similar to that used in the first part of the experiment, introduced into the bulb. The colour of this paper faded completely in two minutes. This experiment can leave no doubt as to the absolute necessity of oxygen in order to bring about dissolution of gold.

As the results given in Tables V and VI (Trans., 1893, 63, 731) are insufficient to completely determine the law governing the dissolution of gold in solutions of cyanides of varying strengths, I made the following additional series of experiments.

In the former experiments, a single gold plate was used for each determination, and the results show a certain amount of irregularity.

Four plates were now taken and dealt with as follows. The plates numbered 1 and 2 were suspended by cotton in a small flask containing a 50 per cent, cyanide solution, those numbered 3 and 4 in a similar flask containing a 40 per cent, solution. Purified air was aspirated through the two flasks. The losses sustained by the plates are given in the following table.

When the plates 1 and 2 were placed in the 40 per cent, solution, and 3 and 4 in the 50 per cent, solution, the results were as follows.

From these numbers, it appears that there is considerable variation in the amount of gold dissolved from apparently similar plates. In order to get more concordant results, I remelted several of the plates, and dividing the button into four parts, rolled and stamped these into plates as already described. With these plates, and using two laxgo Woulff's bottles in place of the small flasks employed in the last experiments, a considerable number of determinations were made; 100 c.c. of cyanide solution was put into each of these bottles and kept at a constant temperature by immersion in a small tank of water. The results, although agreeing much more closely than those just given, still varied considerably. I therefore modified the experiment by bending the points of the inlet tubes at right angles, so as to keep the solutions more thoroughly agitated than before. Eight determinations were made, using this method, but with no better results.

As it seemed probable that the irregularity of the results was due to small bubbles of air being carried in varying numbers to the different plates, it was necessary to devise some process by which the cyanide solution could be thoroughly agitated and kept saturated with air without the possibility of air bubbles in suspension coming page 203 driven at 20 revolutions per minute by a carefully regulated supply of water. These plates were dropped into the solution in A, and the cotton suspenders being always made the same length, the plates were at the same depth in all the experiments, being about an inch from the bottom in their lowest position, and two inches in their highest. The preliminary experiments with this apparatus showed slight differences in the losses of the four plates in the same solution. Thus, in two hours, with 50 per cent, solution, the losses were 0.001025, 0.00115, 0.001075, 0.00095; with 45 per cent., 0.001175, 0.001575, 0.001625, 0.00125; and with 35 per cent., 0.00225, 0.0023, 0.002275, 0.002175.

In these experiments, No. 1 plate hung from No. 1 hook, No. 2 plate from No. 2 hook, and so on. Thinking there might be differences in the rate of motion of the solution in proximity to the respective plates, owing to the syphon coming in at one side of the vessel A, the positions of the plates were altered, putting No. 1 plate on No. 2 hook, and so on, these different arrangements being repeated many times, but it was not observed that a plate in one position sustained a greater loss than in any other. After exposure to the cyanide solution, the plates were rapidly washed under the tap, and then with distilled water, roughly dried with filter paper, and heated to low redness by holding them with iron forceps in a Bunsen flame. As the different degrees of heating and different rates of cooling to which the four plates were subjected might affect their rates of dissolution in cyanide, 12 experiments were made in which everything was conducted as before, except that the plates, after being washed with water and then with alcohol, were dried at a low temperature. The results were no more concordant than before.

In order to discover if these slight irregularities in the results were due to impurities in the plates, I prepared some gold by the method adopted by Professor Thorpe when determining the atomic weight of that metal. Two plates were made from this purified gold, in the manner already described, but these, when exposed to the action of cyanide solution under varying conditions, differed slightly in their losses just as the old plates had done. These plates, with two of the old ones, were subjected to the action of cyanide solutions in eight experiments, and sustained the following total losses. New plates, 0.0402, 0.0415 gram; old plates, 0.0392, 0.0409 gram.

Finally I abandoned the attempt to get absolute agreement among the plates, and resolved to make up for the want of this by multiplying the number of determinations and taking the mean of the results. The numbers so found are shown in Table III, and graphically on pp. 205 and 206, Plates I and II.

These results show that the rate of dissolution of gold in solutions page 202 in contact with the gold plates. The apparatus shown in Fig. 1 meets these requirements. A regular stream of air driven by a filter pump through several solutions of potassium hydrate up a tower containing coke saturated with that solution, and then through three flasks containing barium hydrate solution, issues from the drawn-out point of the tube C, and passing into the open end of the tube D, carries a stream of cyanide solution into the bottle B, where it is syphoned back by E, and thus a constant circulation of the solution is maintained. The vessels A and B stood in a small tank containing water at 18°. From preliminary experiments, it had been found that when there is little cyanide solution the oxygen is withdrawn from the solution by the gold more rapidly than it is absorbed, even when the stream of air is rapid. In the present series of experiments, I therefore used a large amount of solution (500 c.c.), and reduced the duration of the experiment to one hour.

Fig. 1.

Fig. 2.

The cyanide solution was saturated at 18° with air by shaking vigorously for five minutes in a 40-ounce stoppered bottle, removing the stopper several times to equalise the pressure. The bottle fitted into a small wool-lined box, and, with due precautions, it was an easy matter to have the temperature of the contained solution, 18°, at the close of the shaking. The gold plates were suspended by cotton from the hooks attached to the beam K, Fig. 2, which was itself suspended from the crank II fixed to the shaft O of a small water wheel page 204

Table III.

KCN. Grams per 100 c.c.	Gold dissolved.	KCN. Grams per 100 c.c.	Gold dissolved.	KCN. Grams per 100 c.c.	Gold dissolved.	KCN. Grams per 100 c.c.	Gold dissolved.
50	0.00050	20	0.00277	4	0.00600	0.1	0.00675
45	0.00064	15	0.00350	3	0.00613	0.05	0.00666
40	0.00091	10	0.00440	2	0.00627	0.02	0.00613
35	0.00124	8	0.00488	1	0.00650	0.01	0.00345
30	0.00163	6	0.00537	0.5	0.00670	0.005	0.00030
25	0.00210	5	0.00572	0.25	0.00684

of potassium cyanide gradually increases as the concentration of the solution decreases, reaches a maximum at 0.25 per cent, solution, and again decreases.

As it seemed desirable to determine whether another metal would show similar variations in its rate of dissolution in cyanide, the following experiments were made with silver plates. These were of the same size as the gold plates, and were prepared from the chloride. On exposing them to the action of cyanide solution, the losses of the four plates varied slightly; as in the case of the gold. Their positions were changed, some determinations were made without blowing air through the solution; washing with alcohol and drying at a low temperature, instead of heating to redness, was also tried; and, lastly, the four plates were connected by a silver wire, but through all these varying methods there were slight differences in the losses of the plates. Sometimes one plate would lose more than any of the others in each of several determinations, and then another would head the list for a few experiments, only to give place to a third; at other times, the plates varied in their losses from experiment to experiment.

As in the case of gold, I finally gave up the attempt to get absolute

Table IV.

KCN. Grams per 100 c.c.	Silver dissolved.	KCN. Grams per 100 c.c.	Silver dissolved.	KCN. Grams per 100 c.c.	Silver dissolved.	KCN. Grams per 100 c.c.	Silver dissolved.
50	0.00033	25	0.00125	2.5	0.00393	0.05	0.00380
45	0.00044	20	0.00159	1	0.00395	0.02	0.00285
40	0.00056	15	0.00205	0.5	0.00400	0.01	0.00213
35	0.00073	10	0.00257	0.25	0.00410	0.005	0.00040
30	0.00097	5	0.00356	0.10	0.00396

page 205 agreement among the plates, and was forced to make a large number of determinations, and to take the mean of these in compiling Table

Plate I.

IV, and in drawing the curve shown in Plates I and II. The results so obtained are comparable with those shown for gold, since, in both cases, the plates were the same size and the conditions of experiment exactly the same.

The results given in this table show that the rate of dissolution of silver in solutions of potassium cyanide gradually increases as the concentration of the solution decreases, reaches a maximum at

Plate II.

0.25 per cent. solution, and again decreases. These changes in the rate of dissolution are similar to those shown by the gold plates, and the point of maximum solubility is the same. Further, the ratio page 207 of the gold to the silver, dissolved by any particular strength of cyanide solution, is approximately that of their atomic weights.

These remarkable variations in the solubility of gold and silver may be explained as already stated (p. 199), by the fact that the solubility of oxygen in cyanide solutions decreases as the concentration increases, and that thus the solvent power of the strong solutions is rendered less than that of the weaker solutions, which are capable of taking up more oxygen.

In Plate III, I have redrawn the curve, showing the absorption coefficients of oxygen in cyanide solutions, and have placed beside it

Plate III.

the curves representing the solubility of gold and silver. In order to make these comparable, the scale on which gold is represented in page 208 Plates I and II has been increased four times, and that for silver six and a half times.

In Table VI (p. 211), the relations of the gold and silver to the oxygen are shown under the headings Au/O and Ag/O (found). On considering these, it is evident that the solubilities of the two metals are dependent on that of oxygen. Now, if the amount of gold or silver dissolved depends solely on the amount of oxygen in solution, the values Au/O and Ag/O should be constant; but in the results found it will be seen that these values differ considerably, gradually decreasing as the concentration increases. Therefore, in the more concentrated solutions there is less metal dissolved than the amount of oxygen in solution appears to demand. This points to some retarding action on the motion of the oxygen molecules. Now, it seemed probable that viscosity has such a retarding action on the motion of the oxygen molecules in solution, reducing their velocity, and consequently diminishing the number of impacts on the surfaces of the plates in a given time, and so decreasing the amount of gold or silver dissolved. In order to test the validity of this conclusion, the rates of dissolution of gold and silver were determined in cyanide solutions rendered more viscous by the addition of various substances such as sugar and glycerol, which might be assumed to exert no chemical influence on the solubility of these metals. The results are shown in Table V. As it was found that the coefficients of absorption of oxygen for solutions containing equivalent proportions of sugar and of potassium cyanide are approximately the same only a few of the coefficients shown in this table were determined, the majority being calculated from the curve showing the solubility of oxygen in potassium cyanide solutions.

The results in this table, and especially those in the column Au/O, prove very conclusively that the assumption as to the retarding action of viscosity was correct.

Suppose now that we consider the number of times in a second a given oxygen molecule strikes a surface. We may assume from the results just given that this will depend on the viscosity coefficient z, or, in other words, will be a function of z. So that if N be the number, we can write N = a + bz + cz² + &c., where a, b, and c are independent of z (Maclaurin's theorem); or since Au/O, Ag/O are dependent on the number of impacts in unit of time, we can write Au/O = a + bz + cz² + &c., and Ag/O = a₁ + b₁z + c₁z² + &c.

In order to ascertain if these relations hold good for the values of Au, Ag, and O, found, I determined the coefficients of viscosity of a number of cyanide solutions. The observations were made by Gartenmeister's method (Zeit. physik. Chem., 6, 524), using Finkener's

In this expression, r is the radius and l the length of a capillary tube through which a volume v of the liquid of sp. gr. s flows, under a pressure p in unit of time. In my apparatus, Fig. 3, the constants were as follows, r = 0.23477 mm.; p = (h + h₁)s = (HC + CB)s = (318.4 mm. + 31.34 mm.)s = 349.74 s; l = DH = 300 mm.; v = V/t (where V is volume of bulb in c.c. and t is time of flow in seconds) = 6.88832/t. The value of was found by measuring with a cathetometer the height BC when half the weight of water contained between A and C had flown out. The trustworthiness of the apparatus employed was tested by determining the value for water, which was found to be 0.108842 at 18°, a value which corresponds well with Gartenmeister's number when corrected for temperature.

[Note.—There is an error in Gartenmeister's calculation of the value of the second part of Finkener's formula. In his paper, he gives for his first pipette the following constants.

V = 8.1015 c.c.; r = 0.31842 mm.; l = 335.2 mm.; h = 354.4 mm.; and h₁ = 17.4 mm.; and calculates for z the value 0.0₃55267 s.t. whereas it should be 0.0₃552715 s.t.

This error makes his values slightly smaller than they should be; thus, for water at 20° be finds z =

Fig. 3.

0.1030, whereas from his determinations and a correct use of the formula he should have obtained z = 0.1035.]

The values at 18° of z—the coefficient of viscosity—for 5,10, 20, 30, 40, and 49 per cent, solutions are represented graphically in Plate IV, and, by the aid of the curve so obtained, the intermediate values were calculated, and placed together with those found by direct experiment under z in Table VI. In the same manner, the undetermined values for oxygen have been calculated from the curve representing the solubility of oxygen, and the results, together with the original determinations, are embodied in this table.

Under Au/O and Ag/O are shown the values found, and also those calculated by the aid of Maclaurin's theorem; that is to say, by the

formula Au/O = a + bz + &c., whore a = 0.33 and b = −0.9, and Ag/O = a₁ + b₁z + &c., where a₁ = 0.18 and b₁ = −0.4.

The close agreement of the values found by these two methods is sufficient to prove that the true explanation of the smaller solubility of the gold and silver relatively to the oxygen in the more concentrated solutions is to be found in the greater viscosity of these solutions.

The following is a summary of the results obtained in this and the former paper.

1.	Oxygen is necessary for the dissolution of gold in potassium cyanide, and no gold is dissolved in its absence.
2.	The ratio of the gold dissolved to the oxygen required for its dissolution is 196: 8 as demanded by the equation 4Au + 8KCN + O2 + 2OH2 = 4AuCN,KCN + 4KOH.
3.	The rate of dissolution of gold in potassium cyanide solutions varies with the strength of the solution, being small for concentrated solutions, increasing as the solution becomes more dilute, reaching a maximum at 0.25 per cent, of cyanide, and then again diminishing.
4.	The rate of dissolution of silver in potassium cyanide varies in the same way, and the maximum is reached at the same degree of dilution.
5.	The ratio of the amount of gold dissolved by any given cyanide solution to that of the silver dissolved by the same solution is nearly the ratio of their atomic weights.
6.	The variation in the rate of dissolution of gold in cyanide solutions is not directly influenced by the amount of cyanide in solution, except in the case of very dilute solutions, but is mainly page 212 due to the solubility of oxygen in these solutions, the amount of gold dissolved being nearly proportional to the absorption coefficients of oxygen in such solutions.
7.	The rate of dissolution of gold is, however, not exactly proportional to the above-mentioned coefficients, but is rather less than it should be for the more concentrated solutions.
8.	The explanation of this diminishing ratio of the gold dissolved to the oxygen available, as the concentration of the solution increases, is to be found in the increasing viscosity of the solutions as the quantity of cyanide augments.
9.	The explanations given in 6, 7, and 8 are equally applicable to the dissolution of silver in potassium cyanide solutions.