Concentration of the H² Isotope*

G. N. Lewis and R. T. Macdonald, Berkeley, California

Since the publication of the successes achieved by us in obtaining a high concentration of the hydrogen isotope has aroused some interest in the methods we used, we shall describe our procedure here, although we would have preferred first to carry out a more detailed scientific analysis of the various stages of the process, and although it is likely that more advanced methods for obtaining the H² isotope will soon be developed. The fact is that the difference in the properties of the two isotopes of hydrogen so greatly exceeds the difference in the properties of any other pair of isotopes that, despite the very small amount of H² in ordinary hydrogen, various methods will lead to an almost complete separation of H² and H¹, as we have convinced ourselves from some preliminary experiments.

Several months ago, while working on the separation of oxygen isotopes, we found that small changes in the isotope ratio in water can be detected by careful density measurements. The specific gravity of a water sample can be measured with an accuracy of up to one part in a million by determining the temperature at which a float of 10 cm³ will neither sink nor rise in ordinary pure water, and then determining the temperature at which this equilibrium will be restored in our water sample. At a temperature of about 16°C, a change in density of one part in a million corresponds to a change in the equilibrium temperature of the float of 0.006°. Of course, all water samples must be carefully distilled and freed from dissolved air.

At the time when we were occupied with the slow process of concentrating oxygen isotopes, we realized that it would be much more interesting and important to obtain the hydrogen isotope discovered last year by Urey, Brickwedde, and Murphy. Since the concentration of H² in ordinary hydrogen had recently been estimated at one part in thirty thousand or less, it seemed hopeless to achieve such a separation unless a fractionation method were used with an efficiency of an entirely different order of magnitude from that of the methods hitherto used in isotope separation. The more

* Journ. of Chem. Physics 1, 341, 1933. Translated by M. Davydkovsky.

The more we studied this question, the more it seemed to us that such a method could be found; this opinion seemed to be confirmed by the discovery by Washburn and Urey of the fact that water in an old electrolytic cell contains a noticeably larger amount of H² than ordinary water, although they did not determine the degree of concentration.

In our laboratory there is an old electrolytic cell which had been in operation for four years without any changes, except for the occasional addition of distilled water to replace that electrolyzed. We distilled part of the liquid taken from this cell and found that its specific gravity, relative to ordinary water at the same temperature, was 1.000034. This striking increase in density, if it is due to the concentration of the isotope H², would indicate the presence of H² in an amount of one part in 3000, or ten times greater than the amount believed to be present in ordinary water.

The theory of such fractionation is simple. If the cell is in operation for a sufficiently long time, gradually approaching a stationary state, and water is added throughout to maintain a constant volume, then the composition of the electrolyte must become constant. If \(a\) represents the percentage loss of H² divided by the percentage loss of H¹, then the ratio of H² to H¹ in the cell must ultimately approach a value \(1/a\) times greater than the initial value, and the composition of the evolved gases must approach the composition of the water being added. Therefore our first experiment would seem to indicate that even if our electrolytic cell has nearly reached equilibrium, the value of \(a\) cannot be greater than 0.1. All this assumes that the evolved gases contain no water vapor. The fact that water vapor is lost in a noticeable amount still further reduces the possible value of \(a\).

Instead of maintaining a constant volume of electrolyte by adding water, we may allow the total quantity of electrolyte to change. If \(x_1\) denotes the quantity of H¹ and \(x_2\) the quantity of H², then at each stage of the process:

\[ d \log x_2 = a\, d \log x_1 \tag{1} \]

or, denoting by \(x_1^0\) and \(x_2^0\) the initial quantities,

\[ \log \frac{x_2}{x_2^0} = a \log \frac{x_1}{x_1^0} \quad \text{or} \quad \frac{x_2}{x_2^0} = \left(\frac{x_1}{x_1^0}\right)^a . \tag{2} \]

Our next step consisted in carrying out such a method of electrolysis. A certain amount of liquid taken from the large electrolytic cell was placed in a vessel with two nickel plates as electrodes and was subjected to electrolysis with a current of approximately 15 A until ...

...until the volume had decreased to approximately \(2/3\) of the original. Then the electrolyte was redistilled: the increase in density relative to ordinary water proved to be 50% greater than before. It seemed that this once again testified that practically all the \(H^2\) remains in the cell.

These optimistic conclusions were suddenly overturned by a simple calculation. Prof. Giauque, who worked with a large electrolytic cell, was able to give a rough estimate of the total amount of water that had been added to the cell during the four years of its operation. It became clear that, if the ratio of the initial concentration of \(H^2\) to \(H^1\) is taken as 1 to 30,000, then no more than one third or one quarter of the amount of \(H^2\) that we found there had been added to the cell. The discrepancy can be explained in two ways. Either the concentration of \(H^2\) in ordinary water is much greater than had been assumed, or the observed increase in density was due to an increase in the concentration not only of the hydrogen isotope, but also of the heavy isotopes of oxygen, which had been present from the very beginning in a considerably larger amount.

It was soon possible to eliminate the supposition that a large accumulation of heavy oxygen was taking place. A certain quantity of water obtained by the electrolytic method described, and for which the equilibrium temperature of the float was \(0.31^\circ\) higher than the equilibrium temperature corresponding to ordinary water, was distilled in such a way that the vapor passed over heated iron wool, where part of the vapor was converted into hydrogen. The gases then passed into a condenser, from which the water returned to the distillation flask, while the hydrogen passed on into a tube containing heated copper oxide, where it was again converted into water and condensed. After this process was completed, the stream of hydrogen was passed through incandescent iron oxide, which yielded a second sample of water, all the oxygen of which came from the original water, as did all the hydrogen of the first sample. It turned out that the second sample had the density of ordinary water, whereas the first, all the hydrogen of which came from the water under investigation, again showed a difference of \(0.31^\circ\) for the equilibrium temperature of the float.

Thus we had to conclude that the concentration of \(H^2\) in ordinary water is much higher than the usual estimate and is closer to the value \(1:4500\), first predicted by Birge and Menzel from discrepancies in the atomic weight of hydrogen. Our electrolytic experiment can now be explained if it is assumed, first, that the ratio of \(H^2\) to \(H^1\) in the hydrogen released from the cell is higher than we had supposed, and second, that in calculating the concentration from the density it is necessary to introduce a correction for the appreciable difference between the densities of ordinary water and pure \(H^1H^1O\). Although

these new conclusions indicate a lower efficiency of electrolytic separation,—this is partially compensated by a higher initial concentration of H².

We therefore immediately set about reducing, by means of electrolysis, 10 l of water taken from the large electrolytic cell to 1 cm³ or less. Eight parallel nickel plates, each with an area of 55 cm², were placed in a large glass tube and connected alternately to the positive and negative leads, likewise made of nickel. A copper coil was placed in the tube, through which water was passed for cooling. A current of 250 A was used, later reduced in the final stages of electrolysis, when the volume of the electrolyte became small. To avoid too high a concentration of electrolyte at the end of the process, we used 1 l of 5M alkali from the large electrolytic cell and 9 l of water from the distillation of the same liquid.

After five or six days the volume of the electrolyte had been brought to 1 l. Then 90% of it was placed in a bath at 0°C, where CO₂ was passed through it until all the alkali had been converted into ordinary Na₂CO₃; indigocarmine served as the indicator. This solution was then distilled in a copper still, which could be heated sufficiently for all the water of crystallization to be given off. The 900 cm³ thus obtained, together with 100 cm³ of alkaline solution left from the preceding electrolysis, were placed in a smaller but similarly constructed cell and electrolyzed to a volume of approximately 100 cm³. This, in turn, by the same procedure, reduced the electrolyte to a volume of 10 cm³ in a third, still smaller cell. In a fourth, very small cell, placed for cooling in ice water (instead of using a cooling coil), the volume of the electrolyte was brought to slightly more than 0.5 cm³. The resulting liquid, after CO₂ had been passed through it, was distilled in vacuo into a small vessel. The specific gravity of the water thus obtained, measured with a small pycnometer, proved to be 1.035; in other words, 31.5% of all the hydrogen in this water is H².

In the next series of experiments, changes were made in the first and last electrolytic concentrators. In the first, the glass cylinder of 10 l capacity was replaced by a 20 l Monel-metal cylinder with external cooling, and the current was brought up to 400 A. In the last concentrator an apparatus was installed for trapping the gases evolved. These gases were then recombined on platinized asbestos into water containing several percent H².

In this second experiment, in concentrating from 20 l to 0.5 cm³, we obtained water with a specific gravity of 1.073. Accordingly, 65.7% of the hydrogen in this water is H². *

Although the preceding experiment convincingly showed that in the process of electrolysis no large change occurs in the ratio of the oxygen isotopes, and although spectral analysis showed that the hydrogen obtained from our sample of water contains more than 50% \(H^2\), two further things still have to be done before we can judge the composition with confidence from the density. First, this water must be subjected to the process described above, in order to free it from oxygen, in which the isotope ratio might have changed; and second, it must be proved that the density varies linearly with the composition. These two experiments have not yet been completed.

Throughout the whole process of concentration we tried to keep as closely as possible to the conditions of our large commercial cell, until we could determine—theoretically or experimentally—the best conditions for electrolytic separation. Therefore all our electrodes and leads were made of pure nickel. Nevertheless, during the experiments copper was transferred from the cooling coil as a result of parasitic currents, and we suspected that the deposition of copper might slightly reduce the efficiency of the electrolytic separation, but as yet we have no proof of this. During electrolysis we tried to keep the temperature below \(35^\circ\)C, chiefly to avoid loss of water by evaporation, but also partly because it seemed probable that the efficiency of separation would be higher at low temperatures. Here again, however, we made no systematic efforts to prove this.

We tried to determine the efficiency of electrolytic separation by measuring the density of the water at various stages of concentration. Although there were individual deviations, all the results agree well with the assumption that the value in equation (1) is 0.20. In other words, under equal conditions five times more \(H^1\) is evolved than \(H^2\).

According to this figure, if water containing 65.7% \(H^2\) is reduced by electrolysis to one quarter of its volume, it will contain 99% \(H^2\). However, we shall postpone this experiment until we have at our disposal a larger quantity of heavy water.

For the moment, a little more can be said about the theory of electrolytic fractionation. It is clear that under the conditions of our experiments there is a difference in the electrical polarization of the two isotopes at the cathode, amounting to four hundredths of a volt. Assuming that this difference represents, chiefly, a difference in other conditions rather than a difference in the true electrode potentials, it is difficult to guess what part of it is due to the difference in the mobility of the two hydrogen ions, the difference in the rates of diffusion of the two kinds of atoms through the nickel surface, and the difference in the rate of recombination

atoms. These questions are now being investigated. In addition, Prof. Heyrovský intends to make direct determinations of the difference in potential differences for the two isotopes.

In conclusion, let us estimate the amount of the heavy isotope of hydrogen contained in ordinary water, assuming that samples from different sources may differ greatly. There are two simple ways of making this estimate. By successive electrolysis, collecting and recombining the evolved gases, we can approach the preparation of pure $\mathrm{H^1H^1O}$ and find the difference in density between it and ordinary water. On the other hand, we can concentrate ordinary water in an electrolytic cell of known efficiency and determine the increase in density. Preliminary measurements by both methods show that in the city of Berkeley (California) water contains one part $\mathrm{H^2}$ to approximately 6500 parts $\mathrm{H^1}$.