DISCUSSION ON HEAVY HYDROGEN*

(Held at the Royal Society of London)

Lord Rutherford. In the history of physics it very often happens that a new discovery, which at first appears to be of purely scientific interest, ultimately finds useful practical applications. This proposition can be well illustrated by the discovery in the atmosphere of the rare gases—neon and argon—which are now used in large quantities for industrial purposes. The fundamental discovery in 1919 of the isotopic composition of most of our elements, for which we are chiefly indebted to Aston, at first sight seemed to have purely scientific significance; however, from today’s discussion we shall see that this discovery may, after a time, also have diverse practical applications in many directions.

It is hardly necessary to review in detail the history of the discovery and separation of heavy hydrogen. The discovery that oxygen is not a simple element, but contains small quantities of isotopes with masses 17 and 18, indicated a small discrepancy, amounting to 2 parts in 10,000, between measurements of the ratio of the masses of hydrogen and oxygen made by Aston and the same ratio found by direct physical and chemical methods.

Birge and Menzel came to the conclusion that this discrepancy might arise from the presence in ordinary hydrogen of an isotope of mass 2. This suggestion prompted Urey, Brickwedde, and Murphy to try to establish the existence of H² by direct optical methods. Their experiments were crowned with success, showing faint traces of H², the amount of which relative to the isotope H¹ was initially estimated as 1:4000. It was found that the wavelength of the α-line of H² is 1.79 Å shorter than for H¹, which agrees very well with the theoretical value of this difference in wavelengths that was to be expected for the hydrogen isotope of mass 2. The mass of the new isotope was also measured directly by Bainbridge with the aid of a modified mass spectrograph and was found

* Proc. Roy. Soc. A, 144, No. 851, 1, 1934. Translated by V. I. Chernyaev.

equal to 2.0136, i.e. slightly less than the mass of the molecule of ordinary hydrogen, which on the scale \(O = 16\) is 2.0156.

We have no data on the exact structure of \(H^2\) and do not know whether it may be regarded as a simple particle or whether it is built up of two or more constituent parts. At first it was natural to suppose that the nucleus \(H^2\) might consist of two protons and a negative electron, but the subsequent discovery of the neutron indicated that this nucleus is more likely a close combination of a neutron and a proton. Taking Chadwick’s value for the mass of the neutron—1.0067—we obtain the sum of the masses of the proton and neutron equal to 2.0145, whereas the mass of the nucleus \(H^2\) is 2.0136, i.e. somewhat less, which indicates that the binding energy of the neutron–proton combination is less than a million volts. If this were so, one would expect that the nucleus \(H^2\) should split upon collision with fast \(\alpha\)-particles. Together with Kempton, in order to test this, I carried out experiments, but was not able to establish with any certainty the presence of neutrons when heavy water was bombarded with polonium \(\alpha\)-particles. If neutrons were nevertheless produced in this process, their number was probably less than \(1\%\) of the number of neutrons obtained from a beryllium plate under the same conditions. If splitting of \(H^2\) with emission of a neutron does occur in this case, it must occur very rarely in comparison with the number of possible collisions between \(\alpha\)-particles and \(H^2\) nuclei.

It is interesting to note here the conclusion drawn by Lawrence. In his experiments on bombarding various substances with high-velocity \(H^2\) ions, he found that from some elements a group of protons is thereby obtained with almost one and the same velocity for all of them. To explain this he suggested that the \(H^2\) nucleus is then broken up into a neutron and a proton either in the bombarded nucleus or in the strong field in its vicinity. From considerations of conservation of energy it is necessary to assume in this case that the mass of the neutron is considerably less than that found by Chadwick, and is 1.0006 instead of 1.0067. From this point of view the \(H^2\) nucleus contains a store of energy corresponding to approximately 5 million V, which is sometimes released in nuclear collisions. Further experiments are required to establish the validity of these considerations.

It was also of interest to me to establish whether the force fields near the nuclei \(H^1\) and \(H^2\) are the same. For this purpose I attempted to compare the velocity distribution of recoil atoms of \(H^1\) and \(H^2\), produced when \(\alpha\)-particles pass through ordinary and heavy hydrogen respectively. Whereas the recoil particles \(H^2\), as should be expected, traversed a path slightly longer than the particles \(H^1\), in the first approximation the number and distribution of the recoil atoms were in both cases almost the same. In view of the fact that, in a close collision, the \(\alpha\)-particle and the \(H^2\) nucleus approach one another to distances of the order of \(10^{-12}\) cm, these results indicate,

that the scattering fields for the nuclei of H¹ and H² are, with great accuracy, identical even down to these extremely small distances.

The first known successes in concentrating H² were achieved in the separation of liquid hydrogen. Washburn and Urey observed that in old electrolytic vessels H² is present in a higher concentration, and found that in the electrolytic residue there is a rapid increase in the content of H².

This method for obtaining concentrated portions of heavy water on a large scale was first used by Lewis and Macdonald of the University of California. In this way they were able to obtain quantities of heavy water on the order of several hundred cubic centimeters, practically in a pure state. The authors concluded that in ordinary hydrogen there is normally one atom of H² for every 6500 atoms of H¹. Lewis and his collaborators found that the density of this new water is approximately 11% higher than the density of ordinary water, its freezing point is 3.8° C, and its boiling point is 101.42° C. Heavy water has its maximum density, as was established, at 11.6° C instead of 4° C for normal water. It is also of interest to mention one more method of concentration, namely the method of pure diffusion. Hertz informed me that he had succeeded in obtaining, in pure form and in a small quantity, the new isotope by applying his improved diffusion method to ordinary hydrogen. He asserts that in this way he obtained such pure heavy hydrogen that he could not detect the α-line of ordinary hydrogen in its spectrum. I shall not speak further about the practical methods of separation, since I hope that Harteck will tell you about the ways of obtaining heavy water, by which a portion of the new water of approximately 25 cm³ was prepared in the Cavendish Laboratory for experiments on the transformation of matter.

It is obvious that this new discovery opens up a broad and important field of work, but I leave it to my chemist colleagues to deal with this question. Owing to the greater mass of H², one should expect that the rate of diffusion and the rates of chemical reactions will be different if H¹ is replaced by heavy hydrogen, and also that compounds formed with the new isotope should in some cases exhibit properties quite different from those of compounds of normal hydrogen. Likewise, this new discovery raises a number of interesting questions concerning the action of heavy water on the normal course of physical and chemical processes in the animal and plant world. Some information on this interesting aspect of the question has already been obtained.

For me, another question is of great interest, and it is to this that I wish to turn, namely the question of using the nucleus H² as a high-speed projectile for studying the transformation of elements. It was a fortunate coincidence that, while Lewis was preparing concentrated portions of H², Lawrence, from the same university, was working on his ingenious apparatus for obtaining ions of high velocity, corresponding to an energy of more than a million volts.

Lawrence found that ions of $\mathrm{H}^2$ of high velocity are often significantly more effective than protons of equal energy as a means for transforming certain elements. For example, if lithium is bombarded with $\mathrm{H}^2$ ions, $\alpha$-particles are emitted with a velocity considerably exceeding the velocity of the fastest $\alpha$-particles from radioactive substances. It is now clear to us that an $\mathrm{H}^2$ particle sometimes penetrates the lithium nucleus of mass 6, and the resulting nucleus then disintegrates into two $\alpha$-particles flying apart from one another in almost opposite directions. The correctness of this supposition is excellently confirmed by photographs of the tracks of $\alpha$-particles in a Wilson chamber obtained by Dee and Walton. The action of $\mathrm{H}^2$ on the isotope of mass 7 is still more complicated, since Oliphant and I observed that in this case $\alpha$-particles of a wide range of velocities are liberated. In this case, apparently, the capture by the lithium nucleus of mass 7 of an $\mathrm{H}^2$ nucleus leads to the disintegration of the whole system into two $\alpha$-particles and one neutron. According to our estimate, the maximum energy of the emitted neutron may reach 15 million V. This conclusion was confirmed by the fact, established by us, that neutrons can be detected in quantities corresponding to this type of transformation when using $\mathrm{H}^2$ particles with an energy of about 200 thousand V. Lauritsen found that abundant neutron radiation can be obtained by bombarding beryllium with $\mathrm{H}^2$ particles, and Lawrence obtained a large number of them from lithium, although he is inclined to think that the greater part of the neutrons observed in his experiments arose from the disintegration of the $\mathrm{H}^2$ nucleus into a neutron and a proton.

As has already been mentioned, Lawrence observed that $\mathrm{H}^2$ bombardment leads to the emission of one or a greater number of groups of fast protons from certain elements. These observations were confirmed by Cockcroft and Walton for certain light elements: Li, C, and Fe, using $\mathrm{H}^2$ particles with energies of the order of 500 thousand V; but they did not succeed in observing groups of protons from Cu and Au. In general, apparently, the $\mathrm{H}^2$ particle is exceptionally effective in causing the transformation of many elements and leads, in a large number of substances, to the liberation of both $\alpha$-particles and protons and neutrons. There is no doubt that this new projectile, like the proton, will be of great service in the study of the processes that take place in the transformation of elements, which will give us further important information about the structure of the nucleus.

It is evident that this new isotope, which can so easily be obtained in sufficient quantities in the pure state, is of such great importance for science that it is desirable to give it a definite name. Urey proposed for the new isotope the name “deuterium” (deuterium). It is important, moreover, to give a suitable name also to the nucleus $\mathrm{H}^2$, not only as a projectile serving for the bombardment of atoms, but also as a possible constituent part of atomic nuclei. Lewis proposed for this nucleus the name “deuton” (deuton) or “deuteron” (deuteron). Although we all recognize,

that the person who has discovered a new substance has the greatest right to give this substance a definite name, but the question of a convenient nomenclature is in this case so important for scientists that it deserves careful discussion. Although the name “deuton” is in some respects suitable, nevertheless in conversation it can be confused with the neutron, and this difficulty is further deepened by the recent discovery that neutrons are often liberated when elements are bombarded with deutons. In consultation with some of my colleagues, both physicists and chemists, somewhat earlier than these names were published, the names “diplogen” (διπλοῦς—double) for heavy hydrogen and “diplon” for its nucleus received a certain recognition. Whatever view one may take on this question, it is important that the new isotope should have a definite symbol assigned to it, and the symbol D seems convenient for this purpose.

In my brief introduction I have not touched upon a large number of interesting questions which, I hope, will attract the attention of the subsequent speakers.

Chairman. I think it would be well if each speaker expressed his opinion on the question of names; it would be useful to have the collective opinion of all those taking the floor.

N. V. Sidgwick (N. V. Sidgwick). I should like, in agreement with the wishes of the chairman, to express very strong support for the terminology proposed by Lord Rutherford. I would not say that the objection he raises against the name “deutron,” because of its similarity to “neutron,” was so very insignificant; from persons who had been at the meeting of the American Chemical Society in Chicago this summer, I heard that they really experienced great difficulties in determining whether a speaker had said “neutron” or “deutron.” It must be clearly borne in mind that this case is not like the discovery of some rare earth. For any of us it is unimportant what No. 61 is to be called, since we rarely have occasion to deal with such substances; however, the isotope of hydrogen is becoming one of the most important substances in the future development of chemistry.

The names proposed by Lord Rutherford give us precisely what we need. First, they give us the relation between the nuclei of the new and the old isotopes; indeed, it is considerably more reasonable to call this nucleus “double” than “second,” since what is interesting with respect to it is not that it is the second particle in order of lightness—for this is a matter of convention, and one may, for example, say that the first particle is the proton and the second is the particle with atomic number 2. What is of interest to us is that this particle is double in relation to the first, and this consideration is the reason for the name proposed by Lord Rutherford. Then you have the relation between the particle and the atom as a whole, which is called “diplogen,” meaning that from which the diplon is obtained; and the only way of obtaining this particle is to obtain it from the atom.

isotope; moreover, “diplogen” expresses an obvious analogy with hydrogen, which is most remarkable. “Deutium,” “deuterium,” or “deuteronium” do not indicate a relation to hydrogen. Moreover, if one takes compounds such as, for example, CH$_3$D and C$_6$H$_4$D$_2$, then it is quite simple to call them diplomethane or diplobenzene; but if one calls them deuteromethane or dideuterobenzene, then one must remember that the prefix deutero is an independent prefix in scientific terminology and means not “containing deuterium,” but simply “second.”

I have been asked to say something about the chemical side of the matter, and I shall do this as briefly as possible. I want to speak about what properties should be expected of substances formed from this new element, in comparison with the properties of substances obtained from hydrogen, and in what directions one should expect that investigations of the new substance will lead to the acquisition of new knowledge.

It should be expected that the atomic volume of the new element is very close to the atomic volume of hydrogen, just as, so far as we know, the isotopes of lead have the same atomic volume. Moreover, since the atomic volumes are practically identical and the atomic number is one and the same, the usual physical properties of a substance prepared from this element should differ only slightly from the properties of a substance prepared from ordinary hydrogen. This is, first of all.

Secondly, as was shown by Urey and Rittenberg on the basis of theoretical physics, the equilibrium constant for reactions in which the new isotope takes part is not exactly the same as for reactions in which old hydrogen takes part. The difference is hardly more than 50%; thus, for the reaction H$_2$ + J$_2$ = 2HJ they found a difference of 20%.

This applies to reactions taking place at ordinary temperatures; if you carry them out at absolute zero, then this difference should increase to infinity, but at ordinary temperatures you may expect only a fairly small difference. To this belongs the experimental fact already established by Lewis, that the density maximum of D$_2$O lies at 11.6°C, whereas for ordinary water it lies at 4°C. We all imagine, of course, that water in some temperature interval near the freezing point consists of different polymers of one kind or another, and that the anomalous change in density near the freezing point is due to this fact. It is now obvious that, since D$_2$O has a temperature of maximum density different from that of ordinary water, the equilibrium between these polymers, or between the various states of water, is different for these two substances. Therefore—and this, I think, is not always noticed—the observed differences in the physical properties of heavy and light water, i.e. the difference of 3.8° in the freezing point and 1.4° in the boiling point, about 1/2% in molecular volume and about 1% in molecular refraction,

are not normal differences that one should expect, but greater than normal, owing to differences in the equilibria of these substances. These differences are greater than you might expect, for example, for dissimilar substances such as benzene and hexadiphenylbenzene.

The next thing we should expect is that the reaction rates of the new hydrogen should be smaller than the reaction rates of the old, and by an amount that can easily lead to a ratio of \(5:1\) in the rate constants of reactions; and in exceptional cases these differences may be considerably greater. This is obviously very important, since there are many reactions, particularly in organic chemistry, in which the point at issue is whether a hydrogen atom takes part in them. If you can replace one of the hydrogen atoms by an atom of diplogen and measure the rate, this should give you an indication either as to whether a hydrogen atom participates in this reaction, or as to whether the reaction in which the hydrogen atom participates is sufficiently slow to affect the rate of the overall reaction.

On the other hand, the discovery of this new isotope enables us to mark hydrogen atoms in a compound in a manner very similar to that in which Hevesy and Paneth marked atoms with the aid of radioactive isotopes mixed with other non-radioactive isotopes of the same elements. Here, however, we have the enormous advantage that, whereas they of necessity had to confine themselves to isotopes of the heaviest elements, which form part of only a very small number of molecules, we can now do the same with hydrogen, which is a constituent part of a larger number of chemical compounds than any other element.

From this follow the most varied applications. One of them consists in determining how readily hydrogen enters into reaction in ordinary compounds. Bonhoeffer has already shown that if sugar is dissolved in heavy water and allowed to crystallize, redistribution is achieved—namely, an almost equal partition of the atoms of the heavy isotope between the water and the hydroxyl groups of the sugar—but that under these conditions no substitution of hydrogen in the \( \mathrm{C—H} \) groups of the sugar occurs. Sugar consists mainly of \( \mathrm{H—O—C—H} \) groups; in this case one part can exchange hydrogen with water, while the other cannot at all. In addition, Polanyi established that \( \mathrm{DH} \) gas reacts with water in the presence of a catalyst—platinum black. It is clear that it is desirable to continue investigations of this kind, since they should shed much light on the nature of the processes occurring here. In particular, it is important to establish how far the readiness of hydrogen to undergo such substitution is influenced, first, by the nature of the atoms with which it is combined, and secondly—what is also extremely important—by the nature of the other groups in the very same molecule. This should provide a great deal of information about the nature of chemical reactions.

As one more example, I may mention the question concerning the electrolytic increase of concentration. To anyone who has followed the literature on this question it must be evident that, if we learn the actual differences in the conditions of the process, this knowledge will shed much light on the intimate nature of these electrolytic processes themselves.

I have only tried to indicate some of the directions in which we may expect the new reagent to be of value, but it will render its most important services to science in fields of which we at present have no suspicion.

F. W. Aston. I wish to dwell on only two points. First, I should like to emphasize how small a discrepancy in the numbers made possible the discovery of this element. Measurements of the atomic weight of hydrogen by chemical and physical methods, as you know, give, in comparison with \(O = 16\), the number 1.00777, and when I compared the masses of hydrogen 1 and oxygen 16, I obtained 1.00778, this measurement being made through helium, and therefore it could include an accumulated error. I used here the method of changing the electric field. Comparing \(O^{8*}\) with \(He^4\), I could use a small change of field, since I had the possibility of using the carbon atom as a point of reference between them, and changed the field by only 33%; but for \(He^4\) and \(H^2\) I had to change the field by 100%, and if any systematic error was present here, then for these two measurements together it might have been very serious. I estimated the possible error of the entire measurement as 1.5 parts in 10,000, but thought that it was considerably smaller.

There are three entirely different measurements: the chemical atomic weight, the isotopic weight, and the abundance of the rare isotopes of oxygen, established by optical spectroscopic observations. The accumulated error of these three experiments, in one of which it may be taken as equal to 1.5 in 10,000, must be less than 2 in 10,000 in order to demonstrate the existence of heavy hydrogen. I think that Urey may be congratulated as a courageous experimenter, who undertook the task in the presence of such exceptionally small differences from which it was necessary to proceed.

The second point is the question of nomenclature, which I consider extremely important. In June of last year I was at a meeting in Chicago, at which Urey and Lawrence were present. At that time I had a conversation with Professor Bohr, and he at first, of course, insisted that the new isotope be called hydrogen; he said that this was not a new element; this element has atomic number 1, and therefore is hydrogen. I agree that

* \(O^8\) denotes doubly positively ionized oxygen \(O^{++}\), which, as is known, gives in the mass spectrograph a line corresponding to an atom of mass half that of oxygen and with unit charge. F. U.

this is so, but it is completely impossible to develop the chemistry of this substance without a new name for he:o, and at this meeting I remarked to Professor Urey that, although tradition allows precisely the discoverer to give a name to his discovery, this, of course, does not guarantee that this name will be used by the general public. We all know, for example, the history of the “corpuscule,” and for quite recent years such a change in the meaning of the word “isotope” is characteristic that its author could hardly have foreseen it. My own word “mass spectrograph” has also changed. I think that we must have a word as useful as possible, and therefore I fully support the names proposed by Lord Rutherford, “diplogen” and “diplon.”

P. Harteck. First of all I should like to show you two experiments illustrating the difference in the vapor pressure of ordinary and heavy water and the different melting points of normal and heavy ice. According to Lewis and Macdonald, the vapor pressure of heavy water is less than that of normal water: at \(25^\circ\text{C}\) it is \(12\%\) lower, and at \(60^\circ\text{C}\) approximately \(8\%\) lower. Normal water at room temperature has a vapor pressure of about \(24\) mm, and heavy water about \(21\) mm—a difference of \(3\) mm. At \(60^\circ\) the vapor pressure of heavy water is \(138\) mm as against \(150\) mm for normal water. I can show you this difference in vapor pressure by means of a differential manometer. I have here a U-shaped tube containing mercury; in one limb of it there is a small quantity of heavy water, in the other—a little normal water; on one side the vapor is produced by the heavy water, and on the other by ordinary water, and you observe, at room temperature, a difference of levels of \(3\) mm. With an increase in temperature the difference will increase. I have another apparatus of the same type, which I immerse in hot water, and you can see how great the difference of levels became as a result.

The second experiment illustrates the difference in the melting points. Normal water has a freezing point of \(0^\circ\text{C}\), while heavy water, according to Lewis and Macdonald, has \(-3.8^\circ\text{C}\). I have two tubes, in one of which there is a white bead, and in the other a blue one. Both are placed in a vessel whose temperature was approximately \(1^\circ\text{C}\), and you see that the heavy water still has the form of ice, whereas the normal water has melted.

I now wish to make some remarks concerning the preparation of heavy water.

Lewis and Macdonald* described the preparation of heavy hydrogen by means of electrolysis of an alkaline solution with nickel electrodes. They did not give details of the experimental methods. However, certain precautions must be observed in order that the enrichment of heavy hydrogen (\(\mathrm{H}^2\)) proceed satisfactorily. This article gives a description of a convenient method of work. Great assistance—

* J. Chem. Phys., 1, 341, 1933.

in the development of the method was the possibility of investigating slightly enriched portions of water in the apparatus of Oliphant and Rutherford*, in which lithium is bombarded by hydrogen ions. The observed long-range $\alpha$-particles were liberated in collisions with $\mathrm{H}^2$ according to the equation:

\[ \mathrm{H}^2 + \mathrm{Li}^6 = 2\mathrm{He}^4 \quad \text{(range 13 cm).} \]

The amount of heavy hydrogen in ordinary water gives a measurable number of disintegrating particles, which constitutes a rapid method for determining the amount of $\mathrm{H}^2$ at low concentrations.

In this way about 60 portions of enriched hydrogen were investigated, prepared by electrolysis of ordinary water under various experimental conditions. When the method had been developed, the concentration of $\mathrm{H}^2$ was determined by density measurements.

The electrolytic cells were made of nickel manufactured by Mond Nickel Co. The cells were carefully soldered on the outside, so that the soldered places practically did not come into contact with the electrolyte.

Electrolysis was begun with $\frac{1}{2}\%$ NaOH. With a current of 150 to 200 A and water cooling, at about $10^\circ$ C the temperature of the electrolyte did not rise above $25^\circ$ C; dirt and organic compounds had to be carefully removed. The moist surface was cleaned with glass paper and steel wool, and electrolysis was begun immediately.

More than 25 electrolyses were carried out in this manner in four different cells without any failures. Since the initial alkali concentration was very small, it was possible to electrolyze the water until $\frac{1}{20}$ or $\frac{1}{30}$ of the initial volume was reached. The solution was then neutralized with $\mathrm{CO}_2$, and the water was distilled by the method described by Lewis and Macdonald.

The efficiency of electrolysis apparently does not depend on the alkali concentration over wide limits. No definite increase in efficiency was observed with an increase in current density, provided only that this density was greater than $0.6\ \mathrm{A/cm^2}$. The dependence of electrolysis on temperature was not investigated, since it was always necessary to work at low temperatures so that evaporation would be small.

If the electrodes are thoroughly cleaned, the efficiency assumes its full value from the very beginning.

When the concentration of $\mathrm{H}^2$ reached 12%, the mixture of oxygen and hydrogen leaving the cells was burned in water in order to avoid losses.

* Oliphant and Rutherford, Proc. Roy. Soc., A, 141, 359, 1933.
Oliphant, Kinsey and Rutherford, Proc. Roy. Soc. A, 141, 722, 1933.

In the course of these experiments we obtained enriched portions of heavy water containing from 30 to 90% heavy hydrogen, which in total corresponds to 25 cm\(^3\) of pure heavy water.

F. Soddy. In view of the confusion which since 1927 has arisen between the chemical and physical (positive-ray) units of atomic masses, respectively, I think that the disputed application to heavy hydrogen and to other substances, detectable only by means of band spectra, of the now well-understood term “isotope” is appropriate. The concept of chemically inseparable modifications of one and the same element, which appeared in 1907–1910, has its basis in the absolute chemical identity of certain radioactive elements of different masses and different origins. This identity followed from the law of displacement in radioactive transformations, which appeared in its final form as early as 1913, the critical evaluation of which was based chiefly on the work of Alexander Fleck in my laboratory in Glasgow. The double and single displacement of position in opposite directions in the periodic table, occurring when an \(\alpha\)- or \(\beta\)-ray is emitted from a disintegrating atom, respectively, served as the reason for introducing the word “isotope.” For me this word was, and still is, a brief designation of individual members of a group of two or more chemically identical elements, existing in a constant natural proportion and separable only by means of certain physical methods which depend directly on mass and, consequently, on molecular velocities.

At least in the country where it arose, this independent experimental concept of chemistry was always obscured by the more modern conception of the nuclear atom of Rutherford and Bohr and by van den Broek’s concept of atomic number, after the latter had been experimentally determined from X-ray spectra by Moseley, ultimately for all elements with the exception of the elements of the first short period. Thomson’s positive-ray method, Aston’s application of the concept of isotopy to non-radioactive elements, and his discovery of the whole-number rule are now the best confirmations of the correctness of the very concept that appeared in the narrow field of the chemistry of radioactive elements, and in particular confirm the constancy of the natural ratio. The only exception, in some respects, to the second cri-

…criterion, the constancy of the natural ratio of isotopes, is boron, investigated by Briscoe et al., which, as Aston indicated, can more easily be separated by such physical processes as diffusion.*

Before 1929, chemical and physical methods for determining atomic weights led to the most satisfactory agreement for four elements, including oxygen, which, as Aston established with the aid of the mass spectrograph, were “pure” and for which it is possible to determine the atomic weight with very great accuracy. Thus, with respect to oxygen 16, the chemical and physical (mass-spectral) values were as follows:

\[ \begin{array}{lcccc} \text{nitrogen} & \ldots\ldots\ldots & 14.008 & \text{and} & 14.008\\ \text{carbon} & \ldots\ldots & 12.0025 & \text{and} & 12.0036\\ \text{hydrogen} & \ldots\ldots\ldots & 1.00777 & \text{and} & 1.00777 \end{array} \]

Then, in 1929, from consideration of band spectra it was concluded that small quantities of the isotopes \(O^{18}\) and \(O^{17}\) exist. If these are truly isotopes in the original sense of the word, then their presence requires the existence, in almost exactly equal proportions, of the corresponding isotopes in the other three elements; and, as if it could be said with sufficient certainty, the band spectra indicated the existence also of \(N^{15}\), \(C^{13}\), and \(H^2\).

First of all—and this has not been noted today—heavy hydrogen apparently does not exist in a quantity sufficient for the agreement between the chemical and physical numbers to be restored again. It seems to me that Aston ascribes to the relative amounts of the two isotopes a ratio of \(35\,000:1\), while Lord Rutherford today gave a ratio equal to half of this.**

Lord Rutherford. I said 6,500.

F. Soddy. In any case, a ratio of 6,000 is required in order to obtain the correct values of the atomic weights again; and how such proportions can be determined is very difficult for me to imagine. I think that these isotopes exist only in the imagination of investigators, unconsciously, of course, influencing the result of attempts to find the cause of the apparent discrepancy.

Furthermore, it was forgotten that this remarkable difference in the chemical properties of heavy hydrogen, which was discovered, completely destroys the basis of the prediction. For every chemist familiar with exceptionally careful work on atomic weights,

* It is quite natural that boron isotopes differ from one another more than heavier isotopes do, since their relative mass difference is greater. The same, to a much greater degree, should apply also to hydrogen isotopes, which for some reason Soddy does not want to understand. V. N.

* Earlier, for this ratio, the numbers \(30\,000:1\), \(35\,000:1\), and even \(80\,000:1\) were given; but now, apparently, a ratio of \(5\,000:1\) or \(6\,000:1\) has been established. V. Ch.*

in which the ratio of hydrogen to oxygen was found, it is beyond doubt that a substance so easily separable must act more in a chemical than in a physical respect. There is no need at all to go into the question of whether heavy hydrogen is absorbed by palladium or not, or to speak of the ordinary methods by which chemists purify their gases. Either hydrogen and heavy hydrogen are completely separated, or they are not separated at all. Otherwise the difference in purification must inevitably entail differences in the determination of atomic weights. The fact that the values of atomic weights have always agreed for chemists, although both substances readily separate, seems to me a conclusive argument against the applicability of the original prediction. For me this is a play of errors; it is not at all a question of prediction.

As for separability, the essence of the matter seems to me much more like the separation of a pair of homologues, such as, for example, zirconium and hafnium, rather than the separation of isotopes. I do not know what confirms here the constancy of the natural ratio; however, I have seen no one who would rule out the possibility that we are dealing here with substances analogous to oxygen and ozone; i.e., that heavy hydrogen is by no means present in a constant ratio in hydrogen, but is formed in an electric discharge or in the electrolysis of alkaline solutions at high pressure and current density. Aston, in his recently published book Mass Spectra and Isotopes, refers to Sandford’s indication that the band originating from \(C^{12} C^{13}\) is abnormally intense in the spectra of certain types of stars, and since cosmic data are valuable for the chemist, this shows that these elements are not isotopes in the original sense of the word.

The relative ease with which light elements are transformed into one another must obviously lead to a cautious application of properties derived from the behavior of heavier elements to the elements of the first short period, and especially to hydrogen. I have never assigned a place to this element in the periodic system, and therefore I regard the expression “isotope of hydrogen” as a doubly incorrect name. It may even happen that it will be established that synthesizing heavy hydrogen artificially is easier than destroying the light elements.

I do not in the least wish to diminish the extraordinary interest and importance of this discovery. I believe that it is of primary importance, and the Americans should be congratulated on it. I recall that two Americans, Mac Coy and Ross, who encountered a pair of isotopes in 1907, had the courage to call things by their names and to declare that radiothorium and thorium are chemically inseparable in all processes. It seems to me that heavy hydrogen is one of the greatest discoveries of the century. I hope that it will be important not only for the destruction of certain atoms, but, more than that, for the destruction of certain physical theories of the present day, almost all of these theories.

M. Polanyi. I would like to dwell a little on Dr. Horiuti’s experiments with hydrogen containing a small percentage of diplothene, which were carried out in order to measure the rate of ionization of diplothene in connection with the problem of overvoltage and catalysis. If hydrogen containing \(1\%\) diplothene is placed in a vessel of volume \(100\ \mathrm{cm}^3\), \(10\ \mathrm{cm}^3\) of water is added, an additional half gram of platinum black is introduced, and all this is shaken for 10 min. at ordinary temperature, then the \(1\%\) of diplothene is reduced to \(1/2\%\). This proves the existence of catalyzed mutual exchange between diplothene and water.

As is known, platinum is a good hydrogen electrode, and the role of the hydrogen electrode consists in creating an equilibrium between hydrogen and the hydrogen ions of the solution in which it is immersed. In this process there arises a current passing into the solution through the electrode, carrying hydrogen atoms in the form of hydrogen ions, and at the same time and in the very same amount hydrogen ions pass out of the solution and form hydrogen on the electrode. If part of the hydrogen on the electrode is replaced by diplothene, then these two currents lead to mutual exchange of hydrogen and diplothene. In the presence of half a gram of platinum black a current arises approximately equal to \(0.2\ \mathrm{A}\) and flowing in both directions.

We performed several experiments in order to make sure that we were indeed measuring the rate of ionization, and not some accidental quantity, such as, for example, diffusion. We changed the solutions, taking now acidic, now alkaline ones, and moreover taking different acids as well as alcoholic solutions, and established that in doing so the rate of ionization changed very considerably; in particular, the rate of ionization in an alkaline solution proved to be many times smaller than in pure water. If the rate of ionization in ordinary water is taken as unity, then in \(n/1\) hydrochloric acid it is equal to \(0.7\), in \(n/1\) sulfuric acid to \(0.2\), and in \(n/4\) NaOH to \(0.4\); in alcohol with \(2\%\) water the rate is equal to \(0.4\), and, upon adding a small amount of alkali to the solution, ionization disappeared, so that we could not observe it over the interval of time usually used in these experiments.

This shows that we were indeed measuring not diffusion, on which, of course, such variable conditions could not have had such an effect, but the process of ionization itself. I think that, in addition, these experiments shed some light on the question of the nature of the inertia associated with the ionization of hydrogen. It is well known that when hydrogen is formed on an electrode, in order to force the hydrogen to leave the solution a certain overvoltage is necessary, and from this overvoltage it may be concluded that there exists here a certain chemical or physical resistance that must be overcome. Evidently this resistance is identical with the inertia associated with the ionization of hydrogen, which we measured in our experiments. We found that it depends on the nature of

solution into which the electrode is immersed. This convincingly shows that, of the two theories which existed concerning the nature of this resistance, at least for platinum only one is suitable. One of these theories assumed that the cause of the overvoltage is the dissociation of hydrogen molecules into atoms, accompanied by inertia. The other theory supposed that hydrogen instantly decomposes into atoms, and that the overvoltage arises from the next stage—the ionization of hydrogen, i.e., its transition from the atomic to the ionic state. It is obvious that the nature of the solution can hardly affect the first stage, which is a reaction between hydrogen and platinum, whereas this nature must, of course, have a strong influence on the second stage. Thus, at least for platinum, the cause of the overvoltage must lie in the inertia associated with the transition of the hydrogen ion from the solution into the atomic state in the form adsorbed on platinum.

Apparently, it is generally accepted that reactions with diplon always proceed more slowly than with hydrogen. Since I may be partly responsible for this assertion, I should like to point out that it is not always true. The slower reactions with diplon, as compared with hydrogen reactions, may arise from two causes: 1) from the existence of zero-point energy and 2) from quantum-mechanical transitions of particles through energy barriers. The probability of passage through a barrier is always greater for hydrogen than for diplon, but the effect of zero-point energy may in some cases act in the opposite direction. I shall confine myself to one particular case, since the general solution will soon be published by Bawn and Ogden. Let us compare the reaction of free hydrogen atoms and diplon atoms. In the initial state the atoms possess no zero-point energy, and therefore their energies are equal to one another. But at the top of the barrier there will already exist zero-point energy*, and it will be greater for the complex reacting with a hydrogen atom than for the one reacting with a diplon atom. Consequently, the action of zero-point energy at the top of the barrier consists in increasing the activation energy of hydrogen atoms by an amount greater than for diplon atoms.

E. K. Rideal. At present I can give you only a very brief résumé of the work of Drs. L. and A. Farkas, which has continued over the last several months. While in Germany, they worked in the laboratories of Haber in Berlin and of Bonhoeffer in Frankfurt on experiments on the conversion of ortho-hydrogen into para-hydrogen, and conversely, and it seemed possible to apply the methods of thermal conductivity used for the study of the ortho–para hydrogen system to the hydrogen–diplon system. Until now we have had only one method for determining the amount of diplon in hydrogen, namely

* Cremer und Polanyi, Z. phys. Chem. B, 19, 443, 1932; see also Eyring. Proc. Nat. Ac. Sci, Wasch. 19, 78, 1933.
* Eyring and Polanyi, Z. phys. Chem. (B), 12*, 279, 1931.

by the method of converting it into water and determining the density of this water. This process is cumbersome and requires at least several milligrams, whereas when the thermal-conductivity method is used only 0.002 mg is needed.

The experimental apparatus was a somewhat modified installation for precise measurements of the composition of ortho-para hydrogen, described in a recent communication by A. Farkas. This method is based on the different slopes of the curves of specific heats for the two isotopes of hydrogen: \(H_2\) at \(80^\circ\) K has a specific heat equal to \(3\ \mathrm{cal/mol}\), which at room temperature rises to \(5\ \mathrm{cal/mol}\), whereas the heat capacity of \(D_2\) (as also of \(HD\)) changes hardly at all in this temperature interval

Fig. 1.

and amounts to about \(5\ \mathrm{cal/mol}\). The concentration of diplon in hydrogen was determined by measuring the temperature of a wire (heated by a current) stretched in an atmosphere of hydrogen at a pressure of 0.04 mm. The difference in the wire temperatures obtained when placing pure \(H_2\) and pure \(D_2\) in the cell was approximately \(20^\circ\) C, which corresponded to a change in the resistance of the wire by \(6\ \Omega\). The instrument was calibrated by means of portions of water with different contents of D. We express our gratitude to Dr. Harteck for carrying out the very painstaking work of bringing the concentration of our heavy water to the final stages.

The next point was the study of the equilibrium

\[ H_2 + D_2 \rightleftarrows 2HD. \]

If a mixture of hydrogen and diplon is heated on a catalyst, which in our case was a nickel wire, then equilibrium is rapidly established. The formation of HD from H₂ and D₂ led to an increase, by approximately 1 Ω, in the resistance of the wire for a 50% mixture of H₂ and D₂, and this 1 Ω corresponds to the establishment of such an equilibrium in which 50% HD is formed, while 25% H₂ and 25% D₂ disappear.

By preparing suitable mixtures of hydrogen and diplon and bringing them to equilibrium, one can determine the equilibrium constant. In Fig. 1 the experimental curves are shown, agreeing well with the theoretical values. From these curves it is seen that the equilibrium constant lies between the values 3 and 4. First, here in agreement with theory it was established that the equilibrium constant

Fig. 2.

does not change greatly at temperatures above room temperature and up to 600°, and, secondly, that the equilibrium constant is very close to 4, which the theory also predicted.

It was then necessary to consider how this equilibrium is established. We saw that on a catalyst it can be established very readily at quite low temperatures; how does it become established in a homogeneous gaseous phase at high temperatures? It was found that on quartz below 600° C no reaction occurs. However, at higher temperatures the reaction takes place, and portions can be removed and analyzed in order to judge the order of the reaction.

Fig. 2 gives the reciprocal half-exchange times as a function of pressure at 725° C. The order of the reaction according to this curve lies between \(3/2\) and 2.

There are two possible mechanisms for this reaction. First, one may imagine a molecular reaction:

\[ \mathrm{H_2 + D_2 \rightleftarrows 2HD}, \]

and secondly, an atomic reaction

\[ \mathrm{H + D_2 = HD + D,\quad D + H_2 \rightleftarrows HD + H}, \]

where \(\mathrm{H}\) and \(\mathrm{D}\) are formed by the thermal dissociation of \(\mathrm{H_2}\) and \(\mathrm{D_2}\). The rate of the molecular reaction should be proportional to the square of the pressure \((p^2)\), while the rate of the atomic reaction is proportional to \(p^{3/2}\). Knowing the reaction rates at various temperatures, one can split the curve into two, corresponding to each of the reactions written above, and obtain the variation of the collision efficiency with temperature, and from this obtain what is called the activation energy. It was found that for the \(p^{3/2}\) reaction, i.e. the reaction in which atoms take part, the activation energy is equal to \(50\,000 + 7\,000\) cal, i.e. \(57\,000\) cal, where \(50\,000\) cal are expended on the thermal formation of \(\mathrm{H}\) and \(\mathrm{D}\) atoms. This is in good agreement with the result obtained for the thermal para-hydrogen reaction, which proceeds according to the scheme

\[ \mathrm{H + H_2 \to H_2 + H}. \]

The molecular exchange reaction has an activation energy of the same order, i.e. lying between \(50\,000\) and \(60\,000\) cal.

A. and L. Farkas also investigated the reaction with water.

TABLE 1

Reaction: \(\mathrm{D_2 + H_2O}\). \(1\) mm \(\mathrm{D_2} + 17\) mm \(\mathrm{H_2O}\).

From the data of Table 1 it can be shown that the activation energy of this reaction is also of the order of \(60\,000\) cal.

One may from time to time analyze the gas mixture and determine how these separate reactions proceed. The following two reactions are possible:

\[ \mathrm{H_2O + D_2 = D_2O + H_2\quad and\quad H_2O + D_2 = DHO + HD}. \]

It was established that the rates of these two reactions are related to one another approximately as \(1:3\).

It is also interesting to note that the efficiency of collisions, extrapolated to room temperature, is approximately \(10^{-43}\), which shows very clearly that the so-called Oliphant reaction does not proceed directly, but in all probability with the aid of catalysis, as Polanyi has shown.

Quite interesting phenomena were also discovered in studying the flow of a mixture of hydrogen and diplogen through a narrow capillary; thus, for example, such flow can lead to a very large change in the composition of the gas.

Table 2 shows the changes in the composition of the residual gas when the pressure \(p_0\) (several hundredths of a millimeter) is reduced to the pressure \(p\).

TABLE 2

The observed values are in very good agreement with the values calculated on the basis of the theoretical equation

\[ \frac{\left(\dfrac{H_0}{H}\right)}{\left(\dfrac{D_0}{D}\right)^s} = \left(\frac{p_0}{p}\right)^{s-1}, \]

where \(s\), the separation factor, is the ratio of molecular velocities, i.e. is equal to \(\sqrt{2}\).

The diffusion of both isotopes through palladium was also studied; this is one of the best methods for purifying hydrogen.

If one begins with a mixture of hydrogen and diplogen, equilibrium among all three kinds of molecules is rapidly established. The rate of diffusion is an exponential function of temperature and is proportional to the expression \(e^{\frac{16000}{RT}}\), where \(16000\) cal is the activation energy for diffusion. If the diffused portion of a hydrogen–diplogen mixture is analyzed and compared with the hydrogen–diplogen ratio for the initial mixture,

\[ \frac{(H/D)_{\text{diff}}}{(H/D)_{\text{init}}}, \]

a very interesting result is obtained. At low temperatures this ratio is relatively large, amounting, for example, to 1.5 at \(160^\circ\text{C}\), but as the temperature increases this ratio approaches unity more and more closely; in other words, the diplogen, if present, remains predominantly in the undiffused portion.

... whereas light hydrogen passes more easily. The expression for the ratio as a function of temperature looks as follows: \(e^{\frac{830}{RT}}\), where 830 cal represents the difference in activation energies for a certain process in diffusion. It is remarkable that the observed 830 cal is very close to the difference of the zero-point energies for these two substances.

This shows that hydrogen and diploren can be separated in those chemical reactions whose rates, at any values of the activation energy, are sensitive to changes on the order of 1,000 cal in activation energy. For example, if metals are taken and dissolved in water or dilute acids, it turns out that the activation energy for such a process is extremely small. One may therefore expect that the gas liberated in this process is a mixture of hydrogen and diploren, but the ratio of hydrogen to diploren in it should be many times greater than for the water itself. This in fact is the case, and I can cite some figures confirming this, obtained in the analysis of the gas liberated from water containing 30% hydrogen and 70% diploren. If, for example, zinc is taken with a small amount of sulfuric acid and dissolved in this water, then the ratio of hydrogen to diploren in the evolved gas is 3.5 : 1; in other words, almost the same separation is obtained as in electrolysis. Other metals are not so effective. For calcium this ratio is 1.5 : 1, for sodium 1.2 : 1, and for aluminum 2 : 1. In addition, there are many other possible methods of such separation by chemical means; I have indicated one of them because it is extremely simple and, possibly, will lead to technical applications.

As I said at the beginning, this is only a very brief résumé of the work which the two Farkases have carried out since they came from Germany.

R. G. Fowler. I personally was interested in the attempt to understand how it comes about that electrolysis separates hydrogen and diploren in such a way that, over a very wide range of concentrations of hydrogen–diploren in the water which you subject to electrolysis, you always find that the rate of liberation of hydrogen and diploren is, by a certain factor \(Q\), greater than the ratio of their concentrations in the water. I have tried to satisfy myself, for example, by the general proofs that the ideas expressed by Dr. Polanyi, which are probably correct, constitute the only possible way of describing the separation. In doing so I tried, as far as possible, to avoid all references to the mechanism and only to find what is required in order to obtain such a constant course of gas liberation, independent of the ratio of concentrations in the solution. At the same time one must realize that what you have is an established process of electrolysis and that nowhere, except for the gases liberated from the solution in general, do you have an accumulation of hydrogen or of any other substance.

Discussion on Heavy Hydrogen

I want to draw your attention to the following. If one considers the entire electrode system in which the process takes place, hydrogen enters this system by means of various processes—for example, by means of the process that maintains the current; hydrogen leaves the solution and is liberated in the form of gas bubbles, and these two phenomena must proceed at the same rate. With the help of such a simple consideration, it seems to me, one can, in a known way, check how far the correctness of the theories extends. I do not propose to do this now; I shall indicate only an example of the use of reasoning of this kind.

If, for example, diffusion in the liquid can be neglected, and the only process governing the transport of hydrogen is the actual current carried by hydrogen ions, then, whatever the electrode process may be, the ratio of the rates of liberation of hydrogen and diploterium will be exactly \(Q\) times greater than the ratio of the concentrations, where \(Q\) is merely the ratio of mobilities. However, this is incorrect, and in reality diffusion of atoms in the vicinity of the electrode prevents the establishment of what is required by this result; in fact, owing to diffusion of all the liquid in the vessel into its most enriched part, the ratio \(H/D\) is limited, despite the transport of H and D to the electrode.

If one goes further into the study of the process and imagines the exchange of hydrogen between the electrode itself and the boundary layer of the liquid, then in this case hydrogen from the boundary layer is deposited on the electrode. This deposition is a process that carries the current and may be controlled by overvoltages, which may be different for H and D, and which in itself is capable of giving a ratio of the rates of liberation of the two gases of the order of the observed values, provided only that this deposition is the sole effective process in the exchange of hydrogen and diploterium between the solution and the electrode. Then the factor \(Q\) is exactly equal to the ratio of the rates of deposition at the given concentration and given conditions of electrolysis. This would be true regardless of the manner in which the atoms, after deposition, react in order to turn into gaseous molecules; however, Dr. Polanyi in his interesting experiments has shown that not only an exchange process takes place. The process that produces the current is not only a process of exchange of hydrogen between the liquid and the electrode; new conditions also appear here, which must be taken into account. If these conditions are taken into account, then, of course, it can very easily happen—and in fact does happen at low current densities—that the process establishing the current is a trivial process, and equilibrium continues to be maintained between adsorbed hydrogen and the hydrogen of water; what is of greater importance is what occurs after this. The process governing the liberation will then depend on the rate at which the atoms of adsorbed hydrogen react in pairs, forming the gas that is liberated.

If one traces this whole process, it turns out that a certain relation must be satisfied between the effectiveness of the interaction of HD, HH, and DD, at which a ratio of the rates of evolution of the form \(\theta \mathrm{H}/\mathrm{D}\) is attained; however, the multiplier \(\theta\) may depend here not only on the overvoltage factors, but also on the factor of the effectiveness of collisions between adsorbed hydrogen atoms. Therefore, it seems to me that the question still remains open as to what part of the observed \(Q\) depends on one or another of these processes. This is not the place to go more deeply into such a special question; I wish only to point out that such questions arise, and they will require further interesting discussion after better acquaintance with the phenomena; such a discussion will inevitably lead to a better understanding of the whole process.

R. P. Bell. I wish to give a very brief report on some experiments carried out by Wolfenden together with me, on the effect of various factors on the electrolytic separation of water. We determined the concentration of diplogen by measuring specific gravity; in this, independent determinations by different observers agreed to an accuracy of up to \(1 : 100\,000\). Water with a known content of diplogen was subjected to electrolysis under various conditions, its volume was brought down to a known fraction of the initial volume, and the concentration of diplogen in the final portion was determined by measuring the density. We expressed the effectiveness of separation by the multiplier \(\alpha\) in the equation:

\[ d \log \mathrm{D} = \alpha\, d \log \mathrm{H}, \]

i.e. our \(\alpha\) is the same as Professor Fowler’s \(Q\).

Our first experiments had the aim of establishing the effect on the effectiveness of the process of various metals used as the cathode. The results are presented in Table 3.

TABLE 3

Influence of the cathode metal
\(\mathrm{D}/\mathrm{H} = 0.1\% \longrightarrow 0.3\%\)

The possible error in the value of \(\alpha\) is estimated at 0.05; however, it applies to the absolute values, and therefore differences between the values of a series of observations should be valid with an accuracy of approximately 0.02. Within the limits of experimental error, the effectiveness

...was obtained exactly the same, whether we used a nickel, platinum, or copper cathode. In one of the experiments the initial concentration of caustic soda was \(8\%\), whence the final concentration was approximately \(40\%\). It was published that separation becomes considerably less effective in a concentrated alkali solution. This is not confirmed by our results: the efficiency is exactly the same as for dilute solutions.

Table 4 gives the results of experiments determining the influence of temperature and current density.

TABLE 4

Influence of temperature and current density

I. \(D/H = 0.25\% \longrightarrow 0.5\%\)

II. \(D/H = 0.05\% \longrightarrow 0.15\%\)

From the table it is evident that, in practice, the efficiencies at \(10^\circ\) and \(100^\circ\) are identical. Some difference in the values of \(\alpha\) seems to be obtained in the experiments on current density, but it is doubtful that this difference exceeds the error of observation. It is necessary to repeat these experiments with all possible accuracy.

These are the experimental results that I wished to report, and I think that the most interesting thing in them is that the separation is considerably less sensitive to changes in conditions than might have been expected. The suggestion had been made that the separation is somehow connected with overvoltage phenomena. If this were so, one might have expected some change in efficiency when replacing one metallic cathode by another; in fact, however, it turned out that nickel, platinum, and copper are equally effective. Also unexpected is the absence of any influence of temperature, although perhaps the processes taking place in the vicinity of the cathode are not strongly affected by the temperature of the entire mass of electrolyte.

The average value of the efficiency coefficient \(\alpha\) found by us was approximately 0.2, which agrees with Lewis’s results. Dr. Harteck’s results seem to give exactly the same efficiency.

J. D. Bernal. Thanks to the kindness of Lord Rutherford, I was able to obtain a little 9% heavy water. It was placed in a small capillary; from it a single crystal of ice was obtained, which was subjected to X-ray analysis*. The measurements show a very small, but, I think, real difference from the values for ordinary ice obtained under similar conditions and with the very same camera. The \(a\)-axis of heavy ice was 4.50 Å as against 4.52 Å for ordinary ice, and the \(c\)-axis 7.36 and 7.39 Å, respectively. The difference is small and affects the specific volume by only a little more than 1%, but it is very interesting to note that, if this result is correct, then the difference in the volumes for ice has the opposite direction from that for water. The probable molecular volume of heavy water is slightly greater than that of ordinary water, while the probable molecular volume of heavy ice is somewhat smaller than that of ordinary ice.

It seems to me that I can point to theoretical reasons for such facts and, moreover, show that there are two entirely different reasons why deuterium compounds should differ from ordinary compounds. One of these reasons has already been mentioned: it is the difference in zero-point energies. The other reason is the difference caused by the double mass of deuterium in water, which affects the nature of all water in a very simple way. The essence of the difference between the so-called polymers of water is determined by the orientation, the regular or irregular arrangement of the molecules**. The moment of inertia of heavy water is approximately twice the moment of inertia of ordinary water, and therefore in general its molecule rotates more slowly, so that at the same temperature heavy water must be more like ice, i.e. its coordination number is smaller, which gives it a larger volume. This is precisely what takes place for liquid water: heavy water has a larger volume. On the other hand, for ice, where rotation is entirely destroyed, except for a small number of librations giving a large dielectric constant, and where, consequently, all atoms are in regular positions, we have a small volume.

All the properties of heavy water are explained by this hypothesis. The most interesting is the relation between surface tension and viscosity. The viscosity of heavy water is 30 to 50% greater than the viscosity of ordinary water, while its surface tension is only 0.8 of the surface tension of ordinary water. Here two phenomena are at work, both connected with polymerization and na—

* The photographs were taken and analyzed by Miss H. Megaw.

** Bernal and Fowler, “J. Chem. Phys.” vol. I, p. 515 (1933) (see also this issue of UFN, p. 586).

directed in opposite directions. The abnormally large viscosity of heavy water is not a viscosity of translation, but a viscosity of rotation. The ease with which a certain object can move in water depends on the spontaneous rotation of the molecules. It is small for heavy water, and consequently heavy water has a high viscosity. On the other hand, surface tension depends on the force of interaction of the molecules, and rotation does not directly affect it; therefore one should expect it to be the same or somewhat smaller for heavy water. A theory of these phenomena can be constructed, but I cannot go into it here.

The differences arising from this effect of mass in certain compounds, where the effects of other causes are not so noticeable, will manifest themselves very strongly. In practice this should occur in phenomena connected with the mobility of ions, especially in acids. It seems to me that the general theory of the mobility of the hydrogen ion, in any case in the form developed by Prof. R. H. Fowler and myself, must depend very strongly on the masses of the ions; and the difference in mobility obtained for the hydrogen ion is approximately 50% greater than that which could arise from the ordinary effect of viscosity.

V. Jevons. In today’s discussion no spectroscopic investigations connected with the isotope of hydrogen have been considered, except that, at the time of the discovery, Lord Rutherford mentioned the initial observations of Urey, Brickwedde, and Murphy, their measurements of weak Balmer lines D (i.e. H²), accompanying the considerably more intense lines H (i.e. H¹), and also the exceptional success of Hertz, who obtained D of such a high degree of purity that the line spectrum gave no traces of the Balmer lines H, while the molecular spectrum consisted of bands D₂ without any traces of bands H₂ and DH. In addition to what has been said, I wish to draw your attention to the spectroscopic results achieved so far.

Measurements of the Balmer lines D, and also the subsequent measurements of six lines of the Lyman ultraviolet series (Ballard and White), represent the only examples of observation of an electronic isotopic effect given directly by Bohr’s well-known expression for the Rydberg constant of one-electron atoms: H, He, Li++, Be+++. For atoms with two or three electrons, such as, for example, Li⁺ and Li, the theory of the observed isotope effects is, of course, less simple; this theory was given by Hughes and Eckart.

If we turn to the band spectra of diatomic molecules, then up to now we have observations of the isotope effect in three well-known systems that give band spectra: HCl, H₂, and OH.

1. It is now well known that in the infrared absorption band of HCl near $\lambda$ 3.46 $\mu$, each strong line of the molecule

HCl^35 has, on the side of low frequencies, the accompanying line of the molecule HCl^37, i.e., there exist two bands, arising from the isotopy of chlorine and superposed one upon the other. Hardy, Barker, and Dennison observed, near \(\lambda 4.8\,\mu\), a corresponding pair of bands due to DCl, with a similar doublet structure arising from the two isotopes of chlorine. The authors measured no fewer than 19 lines of the molecule DCl^35 and 17 lines of the rarer molecule DCl^37. The rotational displacement proved to be very large \((\mathrm{DCl}-\mathrm{HCl},\ \delta\nu=-805\ \mathrm{cm}^{-1})\); the coefficient \(\rho-1\) was considerably larger than for any previously observed isotopic displacements. The mass ratio derived from these observations is

\[ \mathrm{H}:\mathrm{D} = 1.00778:2.0137, \]

in excellent agreement with Bainbridge’s earlier determination by means of the mass spectrograph, which gave the number

\[ 1.00778:2.01363 \]

(on the chemical scale \(O=16\)). The estimate of the abundance ratio \(\mathrm{H}:\mathrm{D}=35000:1\) was refuted by the result of Bleakney and Gould, who gave \(5000:1\). The ratio \(6000:1\), mentioned by Lord Rutherford, is probably a later and more accurate estimate.

1. Using a gas mixture containing about \(25\%\ \mathrm{H}_2\), \(50\%\ \mathrm{DH}\), and \(25\%\ \mathrm{D}_2\), Miss M. Ashley and Prof. G. N. Lewis obtained spectrograms with a 21-foot grating, identified and measured several hundred new lines. Many of these lines were assigned to the \(P\), \(Q\), and \(R\) branches of the bands of DH and \(\mathrm{D}_2\), which correspond to the \(\alpha\)-bands \((3\,p\,^3\Pi - 2\,s\,^3\Sigma)\) of the \(\mathrm{H}_2\) spectrum. The observed isotopic displacements are in good agreement with those calculated from the data for \(\mathrm{H}_2\), and include not only the usual rotational and vibrational displacements, but also very considerable electronic displacements, which for \(\mathrm{DH}-\mathrm{H}_2\) reached \(2.4\ \mathrm{cm}^{-1}\). An alternation of line intensities was observed in the bands of the homonuclear molecule \(\mathrm{D}_2\), but, of course, not in the bands of the heteronuclear molecule DH. The alternation in the \(\mathrm{D}_2\) bands is such that, apparently, the spin (mechanical moment) of the D nucleus (diplon) is equal to unity, which is twice the spin of the H nucleus (proton).

2. The isotopy effect in the 0,0 band of OH, \({}^2\Sigma \to {}^2\Pi\) at \(\lambda 3064\), was found, though not with complete certainty, by Miss K. Chamberlain and G. B. Ketter, who observed and measured several new lines at the positions calculated for the lines of the branch \(^{s}R_{21}\) of the corresponding band of the isotopic molecule OD. Their results were confirmed and extended in the detailed investigation of Johnston and Dawson, who identified and analyzed the 1,0, 0,1, and 2,0 bands over a large part of their extent, as well as the above-mentioned branch of the 0,0 band. The observed displacements include clearly expressed isotopic effects of line doubling owing to the presence of mechanical moment in the \({}^2\Pi\) state, together with large vibrational and rotational displacements. It may apparently be expected that these investigations will lead to a very accurate determination of the mass ratio \(\mathrm{H}:\mathrm{D}\), and will also yield as yet unobserved isotopic effects of \(\Lambda\)-type splitting in the \({}^2\Pi\) state, as well as spin splitting in the \({}^2\Sigma\) state.

Lord Rutherford. I am sure that you will all agree that the discussion has been extremely interesting, and I regret that the lack of time does not allow many more people to speak on the question under discussion. At the present time only a few scientists in our country have even one cubic centimeter of heavy water for experimental purposes. I hope that this difficulty will soon be overcome and that industry will help us obtain supplies of new water at a low price. The cost of preparing new water in places such as Cambridge, where energy is expensive, almost does not permit its production.

I am sure that you would like me, before the close of the conference, to answer something regarding the points put forward by Prof. Soddy in his communication. As you all know, Prof. Soddy discovered the existence of isotopes among the radioactive elements and gave them this name because, as it turned out, the radioactive elements occupy one and the same place in the periodic table and are inseparable by chemical methods. Much water has flowed since he made this discovery, and now we speak of the isotopes of any element not as inseparable substances, but as consisting of atoms with the same nuclear charge but different masses. It seems to me that no one will doubt that heavy hydrogen has the same nuclear charge as ordinary hydrogen and has one electron revolving around the nucleus, and from this point of view it must, of course, be regarded as an isotope of hydrogen. I, of course, fully understand the reason for Prof. Soddy’s difficulties. The name “isotope” was first applied to atoms of heavy radioactive elements, differing very little in mass, chemically indistinguishable, and separable with great difficulty, even by diffusion. Now this name has been given to atoms of heavy hydrogen, which can easily be separated from ordinary hydrogen and possess certain very distinct properties. This contrast arises from the fact that in hydrogen the masses of the isotopes differ in the ratio \(1:2\), and therefore these isotopes are easily separable by diffusion or by processes similar to electrolysis, in which mass plays a large role. Even radioactive isotopes can be partially separated by diffusion methods, but because of the small difference in their masses the process will be slow and laborious.

I hope that I have been able to convince Prof. Soddy that, in using the word isotope for heavy hydrogen, we do not contradict the basic ideas embodied in this word when it was first used.