The peroxide of hydrogen used was Marchand’s medicinal, containing 3.3 per cent, of available peroxide, as determined by titration with permanganate of potassium. According to the new theory, the H2O2, (±) takes up from the gold strip o, which becomes positive, two units of negative electricity and dissociates into 2 (OH) (—).
In the first experiment a M/1 KCN solution containing the usual amount of absorbed oxygen was used, and 10 c.c. of this solution was placed both in B and in O. Gold strips b and o were then placed in B and O, and the siphon was inserted. Both strips showed themselves of the same potential. The siphon was removed and 5 c.c. of water was added to B and 5 c.c. of hydrogen peroxide to O. On inserting the siphon and the electrodes, b proved to be electronegative, that is, the solution in B was electropositive by + 0.66 volt; in other words, the positive current flowed through the solution from b to o.
Another experiment was made with boiled water with 0.62 per cent. KCN that had been kept under 1/8 in. of oil for a week. B and O were each filled with 10 c.c. of this solution, and the gold strips and siphon were inserted. The strips proved to be of the same potential. The siphon was then removed, and to B was added 2 c.c. of distilled water, and to O two c.c. of peroxide of hydrogen. After mixing, on replacing the siphon, the voltage rose to +0.57 volt. That is, the positive current flowed through the solution from b to o. To exclude the air, a layer of paraffine oil about 1/8-in. thick was floated over each solution before inserting the siphon.
The resistance of 30,000 ohms was then cut out, leaving only that of the galvanometer (3000 ohms), and the needle which had previously shown a deflection of 2.6 scale-divisions was thrown out of sight. (The limits of the scale used were 21.0 scale-divisions.) After being thus short-circuited for an hour and a half, on throwing in again the 30,000 ohms resistance, the voltage of the combination showed itself to be still in the same direction, + 0.63 volt. The 30,000 ohms were again cut out and the combination was again short-circuited overnight. In the morning some bubbles of gas from the action of the peroxide had collected in the upper part of the siphon, and had nearly cut off the current. But on removing and refilling the siphon the voltage still showed itself to be in the same direction, + 0.55 volt. The resistance of 30,000 ohms was again cut out and that of the galvanometer only left in, and after 5½ hours more the electrodes were taken out and cleaned and weighed. Total time, 23 hours.
The strip contained in B had lost 13.25 mg., while that in O had lost only 9.20 mg. Evidently, in spite of the local action that had taken place in the vessel O, more gold had dissolved in the vessel B in the absence of the oxidizing agent, than in O where the oxidizing agent was present.
In order to determine how much of the loss in B might be due to dissolved oxygen which had leaked through, or by, the oil-cover into the cyanide solution since it had been made, a week previously, 10 c.c. of the same solution as that used in B was placed in a similar vessel, and a gold strip was immersed in it half-way, and the liquid was then covered with the paraffine oil just as had been done in B and O. After 19¼ hours it had lost 4.28 mg. A similar strip entirely submerged below solution and oil lost, in 24 hours, 2.64 mg. These experiments prove that some air had leaked through, or by, the oil cover. It had been previously proved that if a thicker layer were used, it was possible practically to prevent altogether the ingress of oxygen and the solution of the gold. In this case it was inconvenient to use a layer thicker than 1/8-in. But the experiment also clearly shows that the amount of gold thus dissolved by absorbed oxygen is so much less than that shown by the b strip, that the solution must have been caused by the electromotive forces of the combination in the manner I have explained.
The same experiment was repeated exactly as before, except that to 10 c.c. of 0.62 per cent. KCN in B was added 1 c.c. of water and to 10 c.c. in O was added 1 c.c. of peroxide of hydrogen. At first the voltage was + 0.652 volt, rapidly falling to + 0.63 volt. After cutting out all but 3000 ohms resistance for 21 hours, the voltage, on adding the 30,000 ohms, proved to be still + 0.63 volt. After again cutting out the 30,000 ohms for 27 hours, it still showed, on inserting it again, + 0.434, rising after resting a few minutes to + 0.456 volt. At this point, after a total of 47 hours, the electrodes were cleaned and weighed, and b was found to have lost 24.06 mg. and o to have lost only 13.25 mg. Here, again, the positive current has moved through the solution from b to o, and more gold has dissolved in the vessel containing no oxidizer, than in the one containing the oxidizer.
In some other experiments with peroxide of hydrogen, there was more local action in O, and the o strip lost as much, and in some cases even twice as much, as the b strip. The exact conditions governing this local action are still under investigation. But in these cases, also, the galvanometer showed that the positive current was still flowing through the solution from the strip b to the strip o in contact with the cyanide containing the oxidizer, and thence back through the gold strip o back again to b, the place of beginning.
The course of the negative current may be traced from the gold strip o immersed in the oxygenated cyanide to the strip b immersed in the unoxygenated cyanide in two ways, as follows :
1. According to Ostwald the reaction O2 + H2 = 4OH produces 4 x 21,100 calories. Assuming this to be true, the oxygen molecule O2 forms with the water four negative hydroxyl ions, 4 (OH) (—); these, assuming a negative charge from the electrode o, cause that end of the gold electrode to be positively electrified. How these negative ions travel through the solution, displacing at the other end of the line four negatively electrified cyanogen ions, 4 (CN) (—), which give up their negative charge at the other gold electrode b, and thus enable four positive gold ions, 4 Au (+), to go into solution there, forming with eight cyanogen ions four complex negative ions, 4 (AuCy2) (—).
The water present may be regarded as not dissociated appreciably, and the dilute solution of cyanide of potassium as entirely so. Making these assumptions, the principal reactions may be expressed as follows :
O2 (±) + 2H2O (±) + 4Au (±) + 8K (+) + 8 (Cy) (—) = 8K (+) + 4OH (—) + 4 (AuCy2) (—).
But this is equivalent to the so-called Ellsner reaction:
O2 + 2H2O + 4Au + 8KCy = 4KAuCy2 + 4KHO,
which Maclaurin has proved to be quantitatively correct.
2. The other view, following Traube, has been urged by Bodlaender, of the Clausthal Bergakademie. He shows first, in agreement with Maclaurin and myself, that the reaction
2H2O + 2Au + 4KCy = 2KAuCy2 + 2KHO + H2,
proposed by MacArthur to explain the solution of cyanide of gold in cyanide solutions, is incorrect. Next, he claims that the so-called Ellsner reaction really proceeds in two stages:
(a) The hydrogen, which is not formed according to MacArthur’s reaction, is, in the presence of cyanide of potassium, water, gold and oxygen, potentially nascent; and a molecule of oxygen combines directly with two atoms of nascent hydrogen, forming hydrogen-peroxide, while two atoms of gold dissolve; —thus:
O2 + 2H2O+ 2Au + 4KCy = 2KAuCy2 + 2KHO + H2O2.
(b) Next, the hydrogen peroxide gradually dissociates into hydroxyl, and causes the solution of two more atoms of gold thus:
H2O2 + 2Au + 4KCy = 2KAuCy2 + 2KHO.
The sum of these two reactions is, of course, the same as that of the Ellsner reaction, which correctly expresses the end-result.
When gold was rapidly dissolved in an aerated cyanide solution, Bodlaender was able to detect as much as 72.3 per cent, of the hydrogen-peroxide required by reaction (a); and, as reaction (b) had probably already set in, this renders his explanation extremely probable.
Expressed in terms of the ions, reactions (a) and (b) become :
(a) O2(±) + 2H2O(±) + 2Au(±) + 4K(+) + 4Cy(—) = 4K(+) + 2AuCy2(—) + 2OH(—) + H2O2(±).
(b) H2O2(±) + 2Au(±)+ 4K(+) + 4Cy(—) = 4K(+) + 2AuCy2(—) + 2OH(—).
The flow of ions through the solution is the same as in the first case. On the whole, the second seems the more probable explanation, though either agrees with most of the facts.
According to either of these views the new theory agrees quantitatively with the results of experiment, but offers for the first time a consistent explanation of its occurrence. It is due to the superior electromotive force of the oxygen (or, in case they are present, to some other electronegative ions, as (OH) (—), Cl (—) Br (—), etc.), together with the capacity of the gold for forming complex ions with cyanogen.
If instead of having the two ends of the gold strip immersed in two separate cyanide solutions, the strip is immersed in the same solution containing some dissolved oxygen, the same electrolytic action can still go on as a case of “ local action; ” for the couple
is still possible if we regard the gold to be short-circuited on itself, and the explanation given above still applies.
When I began this investigation, I marked out for myself a much wider range of investigation than here outlined, and the course of its partial execution has suggested many other interesting questions, some of which are still under investigation ; but the constant and pressing interruptions of routine- work have made it impossible to carry the work further at the present time.