We have another substance at hand with a great tendency to form negative ions. This is the oxygen of the air. In the presence of water, the molecule of oxygen, O2, tends to assume the ionic state, combining with water to form four negatively electrified ions, thus: O2 (±) + 2H2O (±) = 4 (OH) (—). Or, as has been suggested by Traube, when metals dissolve in the presence of oxygen, a molecule of the latter combines directly with two atoms of potentially nascent hydrogen thus: O2 + 2H = H2O2. Later, the peroxide of hydrogen dissociates into two negative hydroxyl ions, which, entering the solution with their negative charges of electricity, tend to produce a current in the same direction as the positively electrified mercury ions do when they leave the solution. That is, oxygen can play the same part in causing the solution of the gold as the mercury ions did in the normal electrode above cited.
The controlling importance of an abundant supply of oxygen is well shown by the curves in Fig. 20. In curve c, although there is only one-fourth as much cyanide present as in curve b, the amount of gold dissolved is greater, except for the very dilute solutions. The evident reason is that the aeration is greater. The cyanide-supply being ample in both cases, the oxygen-supply determines the rate of solubility. For dilute solutions, the amount of dissolved oxygen being sufficient in b, the greater volume of cyanide is the determining factor, and the amount dissolved in b is in this case greater than in c.
Interesting confirmation of these views is found in Maclaurin’s experiments on the solubility of gold in a solution of cyanide of potassium saturated with oxygen. He conducted two sets of experiments with gold strips in solutions of different strengths. The first set was left at rest for three hours, the second set was agitated. The losses are given in the following table:
Maclaurin deems the results in the second table more reliable than those in the first. In both it will be seen that there is a rapid increase of the dissolving power up to about 5 or 10 per cent. KCy; then it gradually falls off till at 50 per cent, the solubility of the gold is less than at 1 per cent.
The importance of the remarkable relation thus discovered by Maclaurin has, I think, never before been appreciated. Is it not a little remarkable that the strong cyanide solution should dissolve less gold than a weak one, while the electromotive force of the gold goes on steadily increasing ?
But in the light of the new theory the reason is not far to seek, for at no time does the electromotive force of the gold rise high enough to displace without external aid any other positive ions, such as those of the potassium in the cyanide or the hydrogen in the water; and unless this be done, the gold ions cannot continue to form, nor the gold to dissolve. For this reason (as Maclaurin, myself and others have shown), in the absence of oxygen or some equivalent agency, gold does not dissolve in cyanide solutions. In other words, unless some negative ion like (OH) (—), (Cl) (—), or (Br) (—) is added, or some other positive ion as (K) (+) etc. is removed by some external source of energy, the action cannot go on. Ordinarily the oxygen of the air furnishes this energy; as we have seen above, it dissolves in the solution and furnishes the negative ions necessary to cause the solution of the gold.
Again, Maclaurin has found the key to the anomalous action of strong cyanide solutions. It is in the fact which he demonstrated, that oxygen is less soluble in strong than in weak cyanide solutions. The following results for the solubility-coefficient of oxygen in KCy are plotted from his curves by interpolation.
I have replotted the results of the above experiments of Maclaurin so as to make them more comparable with my own results. I have replotted both the gold losses of Maclaurin and the second of his oxygen solubility coefficients in Fig. 21; and I have also added the voltage-curve for gold from my own experiments.
It is plain now, for the first time, why there should be a maximum solubility somewhere between 5 and 10 per cent. There are two causes at work tending to dissolve the gold. First, the electromotive force of the gold itself, which alone is insufficient for the purpose; and second, that of the hydroxyl ions. If we suppose the latter proportional to the solubility of the oxygen, we see that the two forces operating to cause the solution of the gold tend to increase in inverse relation. Further, that the electromotive force of the gold rises very rapidly till it gets to between 5 and 10 per cent, and then rises very slowly after that, so that it has little effect on the solubility beyond that point. The solubility of the oxygen (and, as we have assumed, of the hydroxyl ions) is a maximum for pure water, and sinks as the gold-voltage rises. It is at between 5 and 10 per cent, that these two factors give their maximum effect. Beyond that point, the solubility-curves of the solution for oxygen and for gold run along nearly parallel.
Neither of these two factors alone is able to account for the maximum point in the solubility-curve. If the presence of oxygen were the only cause, the maximum solubility would be with dilute solutions. If it were alone due to the electromotive force of the gold, it would be greatest in strong solutions. As both act together, the maximum effect lies between these extremes.
As far as I am aware, this inverse relation between the electromotive force of gold, and that of oxygen in cyanide solutions of varying strength as a controlling factor in determining the solubility of gold in such solutions has never been brought out before. In a certain sense it is a turning-point in this discussion, and hence merits a little close attention.
The ionizing tendency of oxygen has been measured by a cell containing a platinum electrode made absorbent for oxygen by coating it with platinum sponge. When this is immersed in oxygen at atmospheric pressure, and the end of the wire is immersed in M/1 sulphuric acid, and the latter is connected with the normal electrode, the mercury dissolves, and a positive current flows through the solution from the mercury to the platinum with a potential of + 0.75 volt.
This current moves in the opposite direction to that due to the electromotive force of the mercury, viz.: —0.560 volt; consequently the electromotive force of the oxygen at atmospheric pressure in contact with platinum sponge in M/1 sulphuric acid is equal to the sum of these, or 1.31 volts. It is negative, or —1.31 volts, since negative ions are produced, and the solution is negatively electrified by them.
If the above determination is correct, it follows that if, instead of the normal electrode in the above combination, we place a vessel containing a gold electrode and a solution of cyanide of potassium so weak that the potential of the gold is not merely zero, but as low as that of the mercury, viz.: —0.560, it is plain that a similar voltage of + 0.75 should exist; but in this case the gold would dissolve instead of the mercury, and the positive current would flow through the solution from the gold to the platinum as before. In this case it would be, of course, necessary to interpose an M/1 solution of K2SO4, KCl, or some other neutral salt, between H2SO4 and the KCy, to prevent their direct action with each other from interfering with the mere transfer of electromotive forces at the end of the line which we wish to effect.
Now gold does not absorb and ionize oxygen as readily as platinum does, but it acts similarly, though to a much less extent. In order to test the correctness of these views, I took two small porcelain cups, B and O, Fig. 22, in which were immersed the two electrodes b and o. These were gold strips held in platinum-tipped forceps, connected in series with a reflecting galvanometer G of 3000 ohms resistance, including that of the cell, and a resistance R of 30,000 ohms. The solution in either vessel is connected electrically by the liquid in the siphon C.
It is very difficult to prepare, and impossible to keep, a cyanide solution entirely free from oxygen, unless it is hermetically sealed. But the following method was selected as giving an approximation to it. A liter of distilled water was boiled under a filter-pump, and when most of the dissolved oxygen had been removed, cyanide of potassium was added, and the boiling was continued a few minutes, to drive out the air
absorbed during the solution of the cyanide. A cork was provided with two tubes like those of an ordinary wash-bottle; and after inserting the long tube below the surface, a layer of paraffine oil was floated on to the surface to exclude the air. The tip of the discharge-tube was kept closed by a cork when not in use. It was easy, by blowing in through the short tube above the surface of the oil, to discharge any required amount of the solution as required, but of course each time this was done a small amount of air entered the solution. After cooling, the liquid was titrated and found to contain 0.62 per cent. KCy. A similar 0.621 per cent. KCy solution was prepared and nearly saturated with oxygen. Through the galvanometer G and the resistance R, a Latimer-Clark cell gave a deflection of 7 scale-divisions. In vessel B were placed 12 c.c. of boiled 0. 62 per cent. KCy solution and in O an equal volume of 0.621 per cent. KCy solution containing oxygen. On immersing the gold strips, the strip in B became negative, that is, the positive current flowed from B through the solution to O, with an EMF = +0.02 volt. When the liquid in both B and O was covered with paraffine oil to exclude the air, the EMF rose to + 0.108 volt. On gently shaking electrode O, the EMF rose to +0.185 volt; on gently shaking B it fell to +0.08 volt (owing to absorbed oxygen). On cutting out the 30,000 ohms resistance, leaving that of the galvanometer (3000 ohms), the deflection rose to 6.5 scale-divisions, coming back again on inserting the resistance R to 0.6 scale-division or +0.12 volt. This gradually fell to 0.2 scale-division or +0.04 volt, where it remained for two hours. At the end of that time the resistance was cut out and the deflection rose to 2.5 scale-divisions; then, on shaking, 0 to 12 divisions; and then sank again to 2.7, where it remained fairly steady for two hours longer. At the end of this time, four hours in all, the electrodes were removed and cleaned with gasoline and ether from the oil and solution ; and it was found that the electrodes had lost weight as follows:
b lost 1.28 mg. o lost 1.73 mg.
The solutions contained in the vessels B and O and in the siphon C were also assayed with the following results:
B contained 1.25 mg., O contained 1.68 mg., and C contained 0.06 mg. of gold.
The total loss of the electrodes was 3.01 mg., and that found was 2.99 mg. The difference of 0.02 mg. was probably lost in the washings of the electrodes, which were not saved.
This experiment, corroborated by many others, shows clearly that the positive current flows from the deoxygenated to the oxygenated cyanide, just as theory would indicate. The fact that more gold has dissolved in the oxygenated than in the de-oxygenated cyanide does not militate against the indication of the galvanometer.
The solution of the gold in the vessel O is evidently due to the well-known phenomenon of “ local action.” The current that flows through the siphon has to overcome a resistance of from 3000 to 33,000 ohms, while local action can go on in the vessel O wherever an OH (—) ion comes in contact with gold and KCy. Here it forms a “ short circuit,” and it completes itself on the gold strip o at any point free from oxygen, without having to pass through the entire external circuit.
It might be objected that the fact that 1.73 mg. of gold had dissolved in O as against 1.28 in B only went to prove that some oxygen had been contained in B, though less than in O, and that the solution in each had been simply in proportion to the oxygen present. But this does not account for the absolute verdict of the galvanometer, which shows that the positive current flowed during the entire experiment from strip b through the solution to the strip o. The only explanation that remains is the one which I have suggested. There is no doubt that considerable local action went on in cell O. That this was the case is also evidenced by the fact that the action was more uniformly distributed over the surface of b, while the strip o was not uniformly acted on, but was eaten into in a remarkable manner. These strips, and particularly some of those to be described later (with peroxide of hydrogen), were not corroded most upon the edges where one would naturally expect it, but along vertical lines running up and down the middle of the strip. In some cases they were eaten through along these lines in such a manner that nothing remained but a thin film like gold lace. It appeared that local action started in along these lines rather than at the edges, owing to differences of potential due to the distribution of the oxygen, and that when it had once set in, it was able to maintain itself.
It is probable that in all cases of the solution of gold in aerated cyanide solutions the process, as in the above case, is one of local electrolytic action, though, as it is impossible in such a case to apply the galvanometer, it would be difficult to prove this proposition except by inference.
In all such experiments it is important to be certain that the gold strips are in the same physical state, since the existence of microscopic films or unweighable traces of occluded gas cause an appreciable difference of potential in apparently similar gold strips. This is best tested by comparing the strips in the same solution. They react similarly if they are carefully cleaned with boiling acid, and are then washed with distilled water and ignited to redness side by side in the muffle or over a Bunsen flame in a small porcelain dish. But if they are heated in different parts of the same Bunsen flame, they frequently show quite appreciable differences of potential due to occluded gases.