SEPARATION OF NONRADIOACTIVE ISOTOPES

V. I. Chernyaev, Leningrad

It is known that isotopes of one and the same element differ from one another in that the masses of their atomic nuclei differ by one or several units, whereas the nuclear charge, as well as the number and arrangement of the outer electrons, are the same. In view of the fact that the values of the spectroscopic terms or energy levels of an atom are determined chiefly by the nuclear charge or by the structure of the electron shell, the energy levels for all isotopes of one element are, to a first approximation, the same. The agreement of the terms is the better, the smaller the ratio of the difference of the isotope masses to the mass of one of them. The close agreement of the values of the spectral terms also determines the almost identical behavior of isotopes in chemical respects.

The principal difference in the behavior of isotopes may be expected in those phenomena whose course is substantially affected by the difference in the masses of atoms or molecules. First of all this includes the velocity of particles $v$, corresponding to a definite kinetic energy of them, $\frac{1}{2}mv^2$. Indeed, from the value of the mean kinetic energy of molecules of a gas or liquid, $\frac{3}{2}kT$, for the velocity we obtain $v=\sqrt{\frac{3kT}{m}}$. The latter, at a definite temperature $T$, is different for different isotopes, in view of the different masses of the isotopes.

Until quite recently, almost all attempts to separate isotopes, or at least to obtain a portion of an element enriched in one or another of its isotopes in comparison with the concentration of the usual mixture of isotopes, rested on this property.

Let us consider the principal methods of isotope separation, some of which have led to positive results.

1. Separation by recrystallization

In 1925 Robinson and Broisco attempted to separate from one another two isotopic compounds of bromine (bromine has isotopes with atomic weights 79 and 81), namely the compounds $\mathrm{NH_4Br^{79}}$ and $\mathrm{NH_4Br^{81}}$, by means of repeated fractional recrystallization. This method makes it possible to separate from one another substances that are slightly

different in their properties, but possessing not quite the same solubility. However, despite the fact that the authors carried out 2700 recrystallizations, the average atomic weights in the extreme portions proved identical; consequently, they did not achieve separation of the bromine isotopes.

2. Separation by the force of gravity and by centrifugal force

According to the hypsometric formula, when rising to a great height the atmosphere should become enriched in its heavier constituent parts. Therefore, for example, for neon, which consists of isotopes with atomic weights 20, 21, and 22 (perhaps also 23), portions of this gas obtained from the upper layers of the atmosphere should have an average atomic weight smaller than portions obtained from the lower layers. Similarly, chlorine obtained from water in the deep parts of the oceans should have a greater average atomic weight than that obtained from the upper layers. In both of the indicated cases, mixing of the layers as a result of various currents should somewhat diminish the separation of isotopes with height. There have as yet been no experimental investigations of changes in isotope concentration with height.

It is more practical to make use of the force field of a large centrifuge. If a tube is filled with gas and rapidly rotated about an axis passing through its middle, then, owing to centrifugal force, the heavier molecules of the gas mixture will collect predominantly at the periphery. An elementary calculation shows that if \(k_0\) is the ratio of the amount of the heavy isotope to the light one near the middle of the tube, then the ratio at its ends will be:

\[ k_1 = k_0 e^{-\frac{v^2}{2RT}(M_1-M_2)}, \]

where \(v\) is the velocity of the end of the tube, \(R\) is the gas constant, \(T\) is the absolute temperature, and \(M_1\) and \(M_2\) are the masses of the isotopes, with \(M_2 > M_1\). At a velocity \(v = 10^5\) cm/sec and room temperature, for neon \((M_1 = 20;\ M_2 = 22)\) there should be obtained a change in the average atomic weight from the center to the ends of 0.07 unit. For a liquid mixture of isotopes, the movement of particles from the center to the ends during centrifugation will be slow, but the final result will qualitatively be the same as for a gas. Experimental verification of this method is difficult, since during rotation it is hard to collect the gas or liquid from the ends of the tube.

In 1920, Joly and Poole attempted in this way to separate the isotopes of lead, centrifuging it in the molten state. They obtained no positive results, however. After them, Mulliken (1922) tried to apply the same method to the separation of the isotopes of mercury, but the mercury collected both from the ends and from the middle of the rotating tube proved to be of identical density. Neg-

…result the author explains by the presence of vibrations during rotation, which arose because it was impossible to balance the tubes with complete precision.

3. Separation by thermal diffusion

Enskog (1911) and Chapman (1916) showed theoretically that, if a mixture of two gases is placed in a long horizontal tube whose ends are at different temperatures, internal diffusion will occur, as a result of which at the cold end there should be a slight excess of the heavier component of the mixture, and at the warm end of the lighter one. An experimental verification of the theory was provided by experiments on the separation of a mixture of CO₂ and H₂ by the method of thermal diffusion, which were carried out by Chapman and Dootson in 1917. In principle, separation of isotopes is also possible by this method, but no experiments have been performed.

4. Separation by means of a mass spectrograph

As is known, in J. J. Thomson’s apparatus for investigating the masses of molecules, or in Aston’s “mass spectrograph,” by applying electric and magnetic fields, a narrow beam of positive gas ions is made to deviate from the rectilinear direction and to decompose into parts depending on the masses and charges of the particles of which it is composed. For given field strengths and a given construction of the apparatus, the line of motion of any particle is determined exclusively by its mass, charge, and initial velocity. In Thomson’s apparatus the directions of the electric and magnetic fields either coincide or are exactly opposite. As a simple calculation shows, in this case particles reaching a screen placed in their path, if they all have identical masses and charges, are arranged on one and the same parabola, different points of which correspond to different initial velocities of the ions. The screen used is a photographic plate, on which the different ions of the beam form a series of parabolas.

In Aston’s mass spectrograph the two fields are crossed, and moreover in such a way that the deflections of the particles caused by them are directly opposite in direction. This makes it possible to “focus” all particles of constant mass and of one charge into a small elongated spot, even if these particles possess a wide range of velocities. On the photographic plate there is obtained a characteristic “mass spectrum,” outwardly similar to an optical spectrum, the position of the “lines” of the mass spectrum being determined with great accuracy by the mass of the particles.

Since different isotopes of one and the same element possess different masses, in Thomson’s apparatus they will fall on different parabolas, and in Aston’s mass spectrograph they will give different…

lines of the “mass spectrum.” Historically, Thomson–Aston’s method led also to the discovery of isotopy of nonradioactive elements. From the relative intensity of the lines (or parabolas) of the isotopes of some element one can judge their relative amounts in the normal mixture, and the mean atomic weight calculated from these relative amounts agrees well with the chemical atomic weight.

If we turn to the question of the possibility of separating isotopes by the positive-ray method, it must be said that in principle complete separation is attainable by this method, because here it is only necessary, leaving in the screen a slit at the place where particles of a definite mass fall, to make them, after passing through the slit, enter a collecting vessel or to bind them in some other way, for example by a chemical method or by absorption. However, quite apart from the purely experimental difficulties of such collection of particles, even with their complete capture this method can give only microscopic quantities of pure isotopes. Aston calculates, for example, that with complete capture of the particles and with an ionic current reaching 5 mA (which is already difficult to obtain), in 100 sec. of operation of the apparatus one can obtain 0.1 mm³ of Ne²⁰ and 0.01 mm³ of Ne²², at pressures of the order of the usual pressures in discharge tubes, i.e. vanishingly small quantities.

However, in 1930 Yakovlev (Moscow)¹ published the results of his attempt to separate the isotopes of neon with the aid of the method described.¹ In an apparatus constructed essentially like Aston’s mass spectrograph, beams of positive rays, resolved according to mass, were made to fall on a fluorescent screen with one aperture. Any ion beam (consisting of ions of identical masses and charges) could, by changing the magnetic field, be brought to this aperture. Behind the aperture there was placed a rotating platform with ten glass ampoules, which could in turn be brought under the action of the positive rays. An individual ampoule was a glass bulb provided with capillaries, one of which was open, while the other ended in a funnel sealed at the top with a glass film 0.5 μ thick. The ampoules were placed on the rotating platform in the apparatus, the air was then pumped out of the latter, and the open capillary was sealed by means of a focused beam of cathode rays, and the evacuated ampoules were brought under the action of the positive rays through the thin film on the funnel. At a sufficiently high discharge potential the film had to transmit a sufficient quantity of positive rays. By bringing to the aperture in the screen a beam of ions \((\mathrm{Ne}^{20})^{+}\) or \((\mathrm{Ne}^{22})^{+}\), one could expect an accumulation of different isotopes in different ampoules.

After exposure, without admitting air into the apparatus, the capillaries leading to the funnels were sealed off from the ampoules, so that there remained a glass bulb filled with one or another isotope of neon in an amount of approximately 0.03 mm³. The contents of the ampoules were ...

filled Geissler tubes, and the spectrum of their glow was investigated with the aid of a diffraction grating giving, in the second order, an accuracy up to \(0.0006 \ \text{Å}\). The spectra of the presumed \(\mathrm{Ne}^{20}\) and \(\mathrm{Ne}^{22}\) proved to be completely identical, but the author sees the reason for this in the fact that the expected displacement of the lines owing to the influence of the mass of the nucleus for neon should not exceed \(0.0005 \ \text{Å}\). An attempt to analyze the contents of the ampoules with the aid of a mass spectrograph was unsuccessful because of the excessively small quantities of gas. Yakovlev leaves open the question of the results of his experiment. It must be noted, however, that as early as 1927 Hansen \(^{2}\) established that the neon lines have a hyperfine structure arising from isotopy, and in 1932 Hertz, in separating the isotopes of neon, used observations of the change in the intensity of the lines of the hyperfine structure of neon to determine the change in the concentration of the isotopes. Therefore it would not have been difficult for Yakovlev, using a spectral instrument of high resolving power, to establish whether or not he had actually achieved the separation of neon. Up to the present he has not done this, and therefore his separation of the isotopes of neon must be regarded as doubtful.

Only in 1934 did Oliphant and co-workers \(^{34}\) succeed in achieving separation of the isotopes of lithium with the aid of a mass spectrograph. So far only a brief communication on the results has appeared; the details of the experiment have not yet been described, and therefore we must confine ourselves merely to pointing out that the authors obtained in a very pure form the isotopes \(\mathrm{Li}^{6}\) and \(\mathrm{Li}^{7}\) on two separate metallic disks, onto which were directed ion beams separated in electric and magnetic fields. The ion current reached several milliamperes. In order to bind the lithium ions to the surface of the disks on which they fell, the disks were cooled with liquid nitrogen. Each of the types of lithium isotopes was obtained in an amount of the order of one microgram.

5. Separation by the photochemical method

As has already been noted above, the lines of the atomic spectra of different isotopes of one and the same element, since the difference in the masses of the isotopes may be neglected in comparison with their mass, are almost identical. In any case their difference can be established only with the aid of instruments of high resolving power; ordinary spectral instruments are not in a position to resolve the “hyperfine structure” of atomic lines arising from isotopy. But for molecular spectra, whose structure depends substantially on the masses of the nuclei of the atoms constituting the molecule, it is already possible in some cases, even with ordinary instruments, to establish the presence of isotopy of elements entering into the substance under consideration.

Molecular spectra have the form of bands grouped into systems; the bands in turn are divided into separate lines, of which there are very many, and they densely fill the given region.

Such a complication of the spectrum, as compared with atomic spectra, which consist of a relatively small number of separate lines, is due to the fact that a molecule, in addition to electronic energy levels, also has vibrational and rotational ones. The origin of a definite line of a molecular spectrum is due both to a change in the electronic state and to a change in the vibrational and rotational motions of the molecule. Since, for rotational and vibrational energy levels, the masses of the atomic nuclei are essential quantities, for molecules that are chemically identical but contain different isotopes these energy levels are different. Therefore the frequencies of the lines emitted or absorbed by these molecules will also be unequal. Instead of one band spectrum, two, three, or more (depending on the number of possible isotopic combinations) almost identical spectra are obtained, shifted somewhat with respect to one another. Such a spectrum may be regarded as the result of splitting of the individual lines of a band spectrum into a doublet, triplet, and in general into a multiplet.

For a diatomic molecule consisting of one simple element of atomic weight \(\mu\), and an element having isotopes of atomic weights \(\mu_1\) and \(\mu_1+\Delta\), theory gives the difference of frequencies arising from different rotational energies:

\[ \Delta \nu_r = \frac{m^2 h}{8\pi^2 r^2}\left(\frac{1}{\mu_1}-\frac{1}{\mu_1+\Delta}\right), \tag{1} \]

where \(m\) is an integer running through the series of values \(0, 1, 2,\ldots\) and characterizing the rotational energy level (rotational quantum number), \(h\) is Planck’s constant, and \(r\) is the distance between the nuclei of the atoms composing the molecule.

For the difference of frequencies arising from different vibrational energies, we obtain:

\[ \Delta \nu_v = \frac{1}{2}\frac{\Delta}{M}\frac{\mu}{\mu_1} \frac{1}{2\pi\sqrt{\mu_1}} \left(\sqrt{k}\,n-\sqrt{k'}\,n'\right), \tag{2} \]

where \(k\) and \(k'\) are constants characterizing the elastic forces in the vibrating molecule in the ground and excited states (the prime refers to the excited state), \(n\) and \(n'\) are the vibrational quantum numbers of the corresponding states, and \(M=\mu+\mu_1+\Delta\) is the mass of the heavier molecule of the two possible ones.

Both frequency differences (1) and (2) are superposed, and the total change in frequency is equal to:

\[ \Delta \nu = \Delta \nu_r + \Delta \nu_v. \tag{3} \]

Thus, instead of one line of the band spectrum, for the spectrum of diatomic molecules we obtain double lines, the distances between whose components on the frequency scale are equal to \(\Delta \nu\). Such splitting of lines into a doublet will, of course, occur only for diatomic molecules which, as was assumed, consist of one simple element and another having only two isotopes.

Relative intensities of the components of the doublets obtained correspond to the relative amounts of the isotopes in the mixture.

In more complicated cases, when both elements entering into a molecule have isotopes, or when the molecule consists of more than two atoms, of which some may have several isotopic modifications, the theory becomes considerably more complicated. Nevertheless, in this case as well we must obtain a multiplicity of bands, and the splitting of the components of multiplets must increase with increasing difference of the quantum numbers \((n — n')\).

Can one attempt to separate isotopes by a photochemical method?

It is known that in chemical reactions rearrangements of electron orbits and changes in the rotational and vibrational energies of the molecule take place. The possibility of chemical reactions is determined by valence, which depends on the structure of the outermost electron shell itself. If an atom or molecule is acted upon by light, then, if the light is absorbed, a rearrangement of the outer electrons occurs, which may be accompanied by a change in the states of vibration and rotation of the molecule. In other words, an absorbed quantum of light can change the valence of a molecule, which under certain conditions may lead to some reaction.

In addition, if, upon absorption of light, dissociation of a chemically inert molecule into chemically active atoms occurs, then in this case as well light can also lead to one or another chemical reaction. Thus, for example, the combination of hydrogen with chlorine to form hydrogen chloride begins only in light, in which there are frequencies capable of activating chlorine (exciting the \(Cl_2\) molecule or bringing about its decomposition).

Since chlorine molecules composed of different isotopes (35 and 37) must differ somewhat in the positions of their absorption bands, it may be thought that, if a mixture \(H_2 + Cl_2\) is irradiated with a very narrow spectral region, then molecules of one kind should enter into reaction more strongly than those of another. If the reaction thereby leads to the separation of the reacted molecules from the common mixture, then it may be expected that in this way it will be possible to separate isotopes.

The first experiment of this kind was the experiment of Merton and Hartley (1920), who attempted to isolate \(Cl^{37}\) from a mixture of ordinary chlorine and hydrogen by a photochemical method. However, their attempt was not crowned with success. They illuminated a mixture of hydrogen and chlorine with the light of a half-watt lamp, passing the light beforehand through a filter of ordinary chlorine. Since in ordinary chlorine the molecules \(Cl^{35}Cl^{35}\) predominate (the ratio of the amounts \(Cl^{35} : Cl^{37}\) is approximately \(3 : 1\)), it was expected that the light passing through it, in the second vessel, where the mixture \(H_2 + Cl_2\) was located, would be absorbed mainly by \(Cl^{37}Cl^{37}\) molecules, because in the filter the frequencies acting on \(Cl^{35}Cl^{35}\) are preferentially absorbed. Heavier mole-

where \(\mathrm{Cl}^{37}\mathrm{Cl}^{37}\) molecules will, consequently, enter into the reaction in greater quantity, and therefore the resulting hydrogen chloride should have a larger molecular weight.

However, it is now known that, in the formation of hydrogen chloride in a mixture of chlorine and hydrogen, a chain reaction occurs, in which up to \(10^6\) molecules of HCl may be formed. From this chain one cannot require preference of one isotope over another, since even if in the primary reaction \(\mathrm{Cl}^{37}\) combines with hydrogen, then in the further development of the chain such a selective reaction is destroyed. Moreover, as the German investigators Kuhn and Martin\(^3\) indicate, it is doubtful that the place of greatest transparency of ordinary chlorine coincides precisely with the absorption band of \(\mathrm{Cl}^{37}\mathrm{Cl}^{37}\), in view of the fact that the absorption spectrum of chlorine is very complex. Apparently, for the separation of isotopes by a photochemical method it is necessary to use other photochemical reactions.

The only attempt so far at photochemical separation that has yielded positive results is the attempt by Kuhn and Martin\(^3\) to separate chlorine isotopes in the decomposition of phosgene \((\mathrm{COCl}_2)\) under the action of light. A preliminary communication on this investigation appeared in October 1932, and a detailed one at the beginning of 1933. In accordance with the existence of two chlorine isotopes, three phosgene molecules are possible: \(\mathrm{COCl}^{35}\mathrm{Cl}^{35}\), \(\mathrm{COCl}^{35}\mathrm{Cl}^{37}\), and \(\mathrm{COCl}^{37}\mathrm{Cl}^{37}\), the relative amounts of which are proportional to the numbers \(11.2 : 6.7 : 1\), which follows from the fact that ordinary chlorine consists of \(77\%\) of the isotope \(\mathrm{Cl}^{35}\) and \(23\%\) of \(\mathrm{Cl}^{37}\).

Corresponding to these three kinds of molecules, each vibrational band of the phosgene spectrum is split into a triplet, the intensities of whose components are likewise in the ratio \(11.2 : 6.7 : 1\); this confirms the isotopic nature of this splitting.

It had previously been established that in phosgene vapors, at room temperature and pressures up to \(760\ \mathrm{mm}\), there is continuous absorption beginning at \(3050\ \text{Å}\), increasing in intensity toward the short-wavelength side of the spectrum; upon further movement in the same direction, vibrational and rotational structure is revealed. Around \(2750\ \text{Å}\), at first for individual bands, and at \(2720\ \text{Å}\) for all bands, the fine structure disappears, i.e. its separate lines merge and the band becomes continuous. This occurs as a result of so-called predissociation, in which the lifetime of the molecule becomes very short. As a result of such shortening of life, the molecule, before it has time to undergo changes in the state of its rotation, decomposes; and since the rotational structure arises from changes in the rotation of the molecule, the discrete spectrum thereby becomes diffuse.

As experimental studies show, the appearance of the predissociation spectrum always coincides with the beginning of the photochemical reaction. Therefore, if we were able to find such a region of the absorption spectrum

absorption of phosgene where predissociation would already have begun for phosgene molecules of one kind, and if this region were free from the lines of molecules of another isotopic composition, and if phosgene were irradiated by a narrow spectral interval corresponding to this region, then one might think that only phosgene of the given isotopic composition would be decomposed by the light. If we were able to collect the decomposition products, we would obtain a definite isotope of chlorine. However, at the boundary of predissociation the spectrum is so dense that it does not seem possible to isolate such a spectral interval; the predissociation region of molecules of one isotopic composition is overlapped by the convergence of the fine structure of molecules of another composition. Therefore it is necessary to seek a place in the absorption spectrum where the lines are not so dense and where photochemical decomposition is still possible.

These conditions are satisfied by the triple absorption maximum of phosgene at 2817.54 Å, 2818.25 Å, 2818.96 Å, arising from the presence of phosgene molecules containing, respectively, \(\mathrm{Cl}^{35}\mathrm{Cl}^{35}\), \(\mathrm{Cl}^{35}\mathrm{Cl}^{37}\), and \(\mathrm{Cl}^{37}\mathrm{Cl}^{37}\). Comparing the positions of these triplet components with the positions of the atomic lines of various elements, the authors find a very intense and isolated line of the spark spectrum of aluminum (2816.179 Å), close to the first of the indicated components.

If, therefore, phosgene is subjected to the action of this line, then it will be absorbed preferentially by \(\mathrm{COCl}^{35}\mathrm{Cl}^{35}\) molecules. Absorption exclusively by these molecules cannot be expected, since, in addition to the discontinuous absorption spectrum in this region, there is also an intense continuous background, for which the absorption coefficient amounts to approximately 76% of the total absorption of this aluminum line within the band under consideration. Therefore it must be assumed that 76% of the molecules absorb continuously and, in the event of decomposition, give ordinary chlorine (the normal mixture of isotopes), while the remaining 24% absorb discontinuously and, in the event of dissociation, give \(\mathrm{Cl}^{35}\). If it is assumed that all molecules which have absorbed the 2816.179 Å line actually dissociate, then, under the stated assumption concerning the number of molecules absorbing continuously, the calculation gives a change in the atomic weight of the chlorine thus separated from phosgene of \(\Delta A = 0.12\) atomic-weight unit. The accuracy of determining the atomic weight of chlorine permits this difference to be taken into account.

However, in reality one must expect worse results. Indeed, first, some of the molecules which have absorbed the 2816.179 Å line may not dissociate at all, but may give up the energy received either by radiation (fluorescence), or by means of an impact of the second kind, i.e., such a collision between an excited and an unexcited molecule in which the first molecule transfers the energy of its excitation to the second, as a result of which the latter may become excited (or acquire additional kinetic energy

motion), and the first passes into the normal state. Since collisions of the second kind are possible with all sorts of molecules, their presence will lead to a deterioration of the selective decomposition of phosgene.

The process of splitting the phosgene molecule is as follows: \(\mathrm{COCl_2}\to \mathrm{CO}+\mathrm{Cl_2}\), with a large amount of energy being absorbed. Owing to the fact that the reaction is strongly endothermic, there is scarcely any reason to fear a chain reaction. However, during decomposition secondary reactions are possible, after which, together with phosgene, we obtain free \(\mathrm{Cl}\), \(\mathrm{Cl_2}\), and the compounds \(\mathrm{CO}\) and \(\mathrm{COCl}\). These substances can again combine and give phosgene. This undesirable reaction can be hindered if the lifetime of the \(\mathrm{Cl}\) atom and of the \(\mathrm{COCl}\) radical is reduced and the concentration of \(\mathrm{CO}\) and \(\mathrm{Cl_2}\) is decreased. A decrease in concentration can be achieved by forcing phosgene, at a pressure of about 1000 mm, to flow through the irradiated space at such a velocity that the gas mixture emerging from there contains no more than 0.1 mm of chlorine and carbon monoxide. Atomic chlorine with undecomposed phosgene can also give \(\mathrm{COCl_3}\), and this compound can give further reactions that may lead to the replacement of the initially attached chlorine atom by another atom liberated from the compound. In order rapidly to remove chlorine atoms and the \(\mathrm{COCl}\) radical, the phosgene, before entering the illuminated vessel, was saturated with iodine vapor at room temperature and a partial pressure of about 0.13 mm. Then the chlorine atom or the \(\mathrm{COCl}\) radical entered, with the \(\mathrm{J}\) molecule, into the compound \(\mathrm{ClJ}\) or \(\mathrm{ClJ}+\mathrm{CO}\), with liberation of an iodine atom, possibly in the excited state. The latter does not combine with phosgene, and the energy of its excitation is insufficient to break phosgene into \(\mathrm{CO}+\mathrm{Cl_2}\), so that the free iodine atom is not dangerous. Thus a large part of the chlorine atoms is bound in \(\mathrm{ClJ}\), and these latter, reacting with free chlorine atoms, give \(\mathrm{Cl_2}\), liberating the \(\mathrm{J}\) atom. The chlorine obtained, or else \(\mathrm{ClJ}\), must be enriched in the isotope \(\mathrm{Cl^{35}}\).

To separate \(\mathrm{Cl_2}\) and \(\mathrm{ClJ}\) from the rest of the mixture, the gas was passed through solid mercuric iodide. Then chlorine and iodine chloride, reacting with \(\mathrm{HgJ_2}\), gave solid mercuric chloride and iodine: \(\mathrm{HgJ_2+Cl_2=HgCl_2+J_2}\) and \(\mathrm{HgJ_2+2ClJ=HgCl_2+2J_2}\), whereas pure phosgene did not enter into the reaction with \(\mathrm{HgJ_2}\). The chlorine bound in the solid compound could be preserved for a long time, and the iodine liberated was passed into the circulation process. \(\mathrm{CO}\) did not enter into the reaction and from time to time was removed from the apparatus.

In Fig. 1 a diagram of the experiment is given. After evacuation of the apparatus to \(10^{-4}\) mm, 200 g of phosgene was admitted into the system of tubes, vessel \(A_1\) was cooled with liquid air, and the phosgene condensed in it. Then valve \(v\) was closed by admitting mercury from \(K\), and the cooling of the vessel \(A_1\), filled with liquid phosgene, was stopped. For this purpose the Dewar vessel with liquid air, in which \(A_1\) had been immersed, was replaced by a Dewar vessel filled with water at \(14^\circ\mathrm{C}\). At the same time an identical vessel with melting ice was brought under \(A_2\). Owing to the temperature difference between \(A_1\) and \(A_2\), there was established

a pressure difference of about 400 mm was created, and the phosgene from vessel \(A_1\) began to pass into \(A_2\). Capillary \(k\) served to slow the rate of distillation of phosgene from vessel \(A_1\) into \(A_2\), and was chosen of such a cross-section that, under these conditions, 200 g of phosgene would be distilled into \(A_2\) over three days. Along the way the phosgene vapors in the U-shaped tube \(U\) were saturated with iodine vapors and passed through \(a\) and the irradiated vessel \(B\) into tube \(b\) with mercuric iodide. Tube \(a\) was filled with mercuric iodide only so that, during the reverse distillation of phosgene, a small amount of chlorine from the residues coming from \(B\) would be bound. Tubes \(l\), \(L\), and \(M\) serve to separate from the phosgene a small amount of CO; we shall not go into the details of this washing. After purification of the phosgene from carbon monoxide, \(A_1\) and \(A_2\) were brought to the same temperature and valve \(v\) was opened by raising the mercury in \(K\). In this way \(A_1\) and \(A_2\) were connected with one another, after which \(A_1\) was cooled with ice and, as a result, phosgene was distilled into it over the course of 1–2 hours; upon completion of the distillation, valve \(v\) was closed and the process began anew.

Fig. 1.

The aluminum spark inside vessel \(B\) was surrounded by a triple system of filters transmitting only the line 2816.179 Å. The liberated chlorine displaced iodine in tube \(b\) with \(\mathrm{HgJ}_2\). Tube \(c\) was installed only as a precaution; in reality all the chlorine remained in \(b\), and only traces of \(\mathrm{HgCl}_2\) were found in \(c\).

After lengthy repetition of the described processes (the experiments continued for half a year), the contents of tube \(b\) were used for measurement of the atomic weight of the chlorine bound in \(\mathrm{HgCl}_2\),

The results of the investigation are as follows:

1. Atomic weight of ordinary chlorine — 35.455
2. Atomic weight of chlorine obtained photochemically — 35.430
   and 35.431

Consequently, chlorine has been obtained with an atomic weight 0.0245 units less than that of ordinary chlorine. The reasons for the considerably smaller change in atomic weight than had been expected are apparently those indicated above.

This result, of course, is small in absolute magnitude, but it is of interest as the first successful attempt at the photochemical separation of isotopes. Here it has been shown experimentally that between isotopes, in addition to a difference in masses, there is also a certain chemical difference, which, however, can be detected only under especially pure reaction conditions. In the case considered, this chemical difference between isotopes is manifested in the fact that, under completely homogeneous conditions—namely, when placed in a radiation field in which the line 2816.179 Å is exceptionally intense—one isotopic compound proves to be more stable than the other.

6. Separation by Diffusion and Evaporation

In 1896 Lord Rayleigh showed theoretically that, when a mixture of gases diffuses through a porous material, the gas that has diffused through the partition becomes enriched in the lighter constituents of the mixture, while the residue becomes enriched in the heavier ones; moreover, the enrichment of the residue can be expressed by the formula:

\[ r=\sqrt[k]{\frac{V}{v}}, \tag{4} \]

where \(V\) is the initial and \(v\) the final volume of the gas, \(k=\frac{m_2+m_1}{m_2-m_1}\), with \(m_2\) being the mass of the molecule of the heavy constituent of the mixture, and \(m_1\) that of the light one.

The enrichment \(r\) is defined as the ratio

\[ r=\frac{y}{x}:\frac{Y}{X}, \]

where \(y\) is the number of molecules of one kind (here, the heavier ones), \(x\) the number of molecules of the other kind (the lighter ones) at the end of the process under consideration; \(Y\) and \(X\) are the corresponding quantities at the beginning of the diffusion process.

We see that the enrichment \(r\) is the greater, the larger the ratio of the volume of gas passed through the porous material to the volume of its residue before the partition, and the smaller, the larger the root exponent \(k\).

Since isotopes of one and the same element differ above all in mass, it is understandable that immediately after the discovery of isotopes

...element compounds, attempts arise to separate isotopes by the method of diffusion.

In 1913, at the meeting of the British Association in Birmingham, Aston reported on the results of his work on enriching the normal mixture of neon with the heavy and light isotopes. By means of repeated diffusion he succeeded in obtaining two portions of neon whose atomic weights proved to be 20.15 and 20.28, whereas the atomic weight of ordinary neon is 20.2. Aston’s experiments are the first in the separation of isotopes.

Comparing the possibility of separating isotopes of neon and of hydrogen chloride by diffusion, we find the root exponent in Rayleigh’s formula to be \(k = 21\) for Ne, and \(k = 37\) for HCl. Consequently, at equal \(V\) and \(v\) it is easier to enrich neon in some isotope. However, large quantities of neon are difficult to obtain and store, whereas hydrogen chloride can easily be obtained in any quantities; therefore, with respect to the ratio \(\frac{V}{v}\) we are in much better conditions when working with hydrogen chloride. In addition, purification of neon presents great difficulties, which means that a certain degree of uncertainty will always exist in the correctness of the determination of the atomic weight of neon, whereas hydrogen chloride is easily purified.

As a result of these considerations, the American investigators Harkins and his coworkers undertook a series of experiments on the separation of chlorine isotopes by diffusion of gaseous hydrogen chloride through the walls of porous tubes. These experiments began in 1916, but then, interrupted by the war, ceased, and only in 1920 were the first results obtained. In that year a communication appeared by Harkins and Broeker, who obtained about 10 g of chlorine of atomic weight 35.515, i.e., by 0.055 atomic-weight unit greater than the atomic weight of ordinary chlorine. For this they had to pass through the apparatus about 19,000 l of hydrogen chloride at atmospheric pressure.

In 1921 Harkins and Hayes[^4] published a description of their apparatus, which remained essentially the same in all subsequent experiments by Harkins and his collaborators.

The apparatus consisted (Fig. 2) of five identical glass cells equipped with porous tubes, through which carefully dried gaseous HCl passed successively. In this process, through the porous walls of the tubes there preferentially passed the lighter molecules \( \mathrm{HCl}^{35} \), which were carried by a stream of dry air into an absorber. The latter consisted of a series of glass tubes filled with pure water, the surface of which remained free; along it flowed HCl and was rapidly absorbed in the water. The heavier molecules \( \mathrm{HCl}^{37} \) diffused less well through the tube walls and predominantly passed through the entire system of porous tubes. At the outlet of the system the heavy fraction was absorbed in the same kind of absorber as the light one.

From the NaCl formed in the hydrochloric-acid absorbers, containing first the light and heavy fractions of chlorine, NaCl was obtained; the NaCl obtained in this way, in turn, served for the generation of gaseous HCl, which was passed into a repeated process, separately for each fraction. In this way, increasingly separated portions of chlorine were obtained.

After the third diffusion, Harkins and Hayes obtained 100 g of chlorine of atomic weight 35.494, i.e., having an atomic weight

Fig. 2.

0.034 atomic-weight units greater than ordinary chlorine. The fourth diffusion gave 9 g of chlorine of atomic weight 35.498 \((\Delta M = +0.038)\).

In addition, the authors obtained several kilograms of chlorine whose atomic weight was noticeably greater or less than the normal value, which was taken as 35.460.

In this work porous tubes with fairly large pores were used; therefore the efficiency of the process was somewhat lower than in the experiments of Harkins and Broeker.

All the work on the separation of chlorine isotopes by diffusion of HCl was carried out at atmospheric pressure. However, there is no doubt that vacuum diffusion is more effective, in which the gas diffuses through the wall into a vacuum. The pressure on the side with high

SEPARATION OF NONRADIOACTIVE ISOTOPES

the pressure (i.e., where the heavy fraction is located) must be maintained at such a value (for example, 10 mm) that the mean free path of the gas molecules is of the same order as the pore diameter. Low pressure assists the passage of molecules through the pores, as well as the mixing of the gas. But in diffusion in a vacuum the volume of gas per unit mass increases considerably, and certain experimental difficulties arise. Therefore Harkins and his collaborators carried out their experiments at atmospheric pressure.

In 1924 Harkins and Liggett ^5 obtained an increase in the atomic weight of the heavy fraction by 0.043 units.

Later, in 1926, Harkins and Jenkins ^6 took up the light fraction, using essentially the same method as before, only somewhat modifying the form of the individual cells. In addition, the authors applied a method of mixing certain fractions, proceeding from their approximately identical composition, when dividing the initial amount of gas into a large number of portions during the course of the process.

They obtained 28 g of chlorine of atomic weight 35.418, i.e., they decreased its atomic weight by 0.039 units. If this result is compared with the result of Harkins and Broeker, who obtained a heavy fraction with an atomic weight 0.055 units greater than the normal one, then for these two extreme portions we obtain a difference in atomic weights \(\Delta M = 0.094\) units. This is, as it were, the maximum separation achieved for chlorine. Of course, one cannot properly speak here of the separation of isotopes; here we have only enrichment—and that a very slight one—by one or another isotope. Actual separation by means of diffusion was achieved only in 1932 by the German investigator Hertz, by his own method. In 1933 Hertz’s collaborator, Harmsen, after somewhat improving the method, achieved still more significant successes. We shall return later to a consideration of their experiments.

Let us combine in one table the results of attempts by various authors to enrich chlorine with one or another of its isotopes.

Author	Year	Method	\(\Delta M\)	Obtained, g
Harkins and Broeker	1920	Diffusion	\(+0.055\)	10
Bronsted and Hevesy	1921	Evaporation	\(+0.02\)	See below
Harkins and Hayes	1921	Diffusion	\(+0.038\)	9
Harkins and Liggett	1924	”	\(+0.043\)
Harkins and Jenkins	1926	”	\(-0.039\)	28
Kuhn and Martin	1933	Photochemical	\(-0.0245\)

Let us now consider attempts to separate the isotopes of mercury, or, more precisely, to change the specific weight of its extreme fractions. Mercury has very many isotopes—at any rate seven \((196, 198, 199, 200,\)

201, 202, 204), so that distinguishing them from one another is in fact a very difficult task.

H. Wilde, as early as 1874, noted that the density of mercury depends somewhat on the method of its distillation; the same phenomenon was observed in 1883 by Marek. This is explained by the fact that, when a liquid evaporates, its surface acts as a plate with pores of molecular dimensions, and this is especially pronounced at low pressure. The lighter particles fly out of the liquid with greater velocities (since the velocities are inversely proportional to the square roots of the masses), but have a greater probability of passing back from the vapor phase into the liquid phase. Therefore a process is desirable in which the molecules that have flown out would immediately condense somewhere else, so that they would not be able to return. Then the concentration of heavy molecules in the liquid residue will increase, and in the condensate will decrease. This method was first applied by Brønsted and Hevesy\(^7\) for the separation of mercury isotopes (1920).

It was necessary to ensure that the displacement of mercury molecules due to molecular motion should hinder the accumulation of the heavy isotope in the surface layer, which must occur because of the preferential escape of light particles. The mean displacement of mercury molecules is equal to \(5\cdot 10^{-3}\ \mathrm{cm/sec}\). If no more than \(5\cdot 10^{-3}\ \mathrm{cm}^3\) evaporates from a square centimeter in one second, then the indicated accumulation in the surface layer will not occur. Therefore Brønsted and Hevesy used low temperatures for the evaporation of mercury—from 40 to 60°.

The evaporation was carried out in a high vacuum. At a distance of 1–2 cm from the evaporation surface there was a condensing surface cooled by liquid air. On it the emitted mercury atoms froze, and, because of the small length of their flight, the possibility of collisions between them was practically excluded. The change in the concentration of the isotopes was determined from the change in density of the condensate and of the residue. If the density of ordinary mercury is taken as unity, the authors’ first results were as follows:

Condensed mercury	0.999981
Mercury residue	1.000031

Later, using as starting material \(2700\ \mathrm{cm}^3\) of mercury, obtaining from it small extreme portions of \(0.2\ \mathrm{cm}^3\) each, and applying systematic fractionation, the same investigators obtained the following results:

Light fraction	0.99974
Heavy fraction	1.00023

which corresponds to a decrease in the mean atomic weight for the light fraction by \(\Delta M=-0.052\) units and an increase in the atomic weight of the heavy fraction by \(\Delta M=+0.046\) units. The same method Brønsted and

and Hewesch applied to the separation of chlorine isotopes, evaporating at \(-50^\circ\) a solution of HCl in water, obtaining a change in the atomic weight of chlorine of slightly more than 0.02 units.

By the same principle Egerton and Lee \(^{8}\) carried out in 1922 the separation of zinc isotopes (isotopes 64, 65, 66, 67, 68, 69, 70) and obtained the density of the zinc remaining after evaporation equal to 1.00026 (\(\Delta M=+0.017\)), and of the distilled zinc 0.99971 (\(\Delta M=-0.018\)), if the density of ordinary zinc is taken as unity.

In the same year 1922, by means of the method of Brønsted and Hewesch, Mulliken and Harkins \(^{9}\) separated mercury isotopes and obtained a difference in atomic weight between the extreme fractions equal to 0.027 units.

In 1923 Mulliken \(^{10}\) applied to the separation of mercury isotopes a combined method, combining, in order to obtain greater efficiency of the process, evaporation of mercury with diffusion of its vapors through porous partitions. In each of the systematically repeated operations, mercury vapors at low pressure (about 5 mm), produced by evaporation under such conditions that partial separation was obtained, diffused through porous partitions, which led to further separation of the isotopes.

The whole apparatus consisted of six identical units, which operated independently of one another, but in such a way that the products of one cell were used as the starting material for the subsequent cells.

Fig. 3.

A separate cell (Fig. 3) operated as follows. In vessel \(H\), which could be filled and emptied by means of tube \(F\) reaching to its bottom, evaporation of mercury was carried out. Mercury vapors passed through tube \(N\) with tube \(M\) of Whatman paper attached by means of liquid glass; part of them diffused through the walls of the latter (the light fraction), while part passed upward and condensed in the enlargement \(O\) of tube \(N\) and in the spiral \(P\) above it, cooled by water. The mercury condensing in these places flows through the rising vapors into vessel \(H\), evaporates again, part of the vapors again passes through the walls of tube \(M\), and part condenses and flows back, etc. The mercury vapors that have diffused through the walls of the paper tube condense on the inner surface of tube \(J\), cooled from outside, and flow downward. From the apparatus this mercury can be discharged through tube \(A\), fitted below with a stopcock. Obtained in this way

the mercury represents the light fraction, and the residue in vessel \(H\)—the heavy one.

The course of the whole process is as follows. When an individual cell was filled, a large quantity of the light fraction, produced earlier, was introduced into vessel \(H\). In the course of diffusion the density of the residue in vessel \(H\) increased. The initially produced fractions of ever greater density were gradually introduced into vessel \(H\), and the mercury that had diffused through was removed in successive portions, the volume of which was approximately 30% of the volume of the mercury contained in the vessel. The process was continued until the last of the former enriched fractions had been introduced into the system and the residue in vessel \(H\) had become too small.

Then the mercury remaining in vessel \(H\) was removed through a tube by means of air pressure admitted into \(H\); this mercury constituted the extremely heavy portion. When such a series of operations was repeated, the fractions of greatest and least density became more and more separated.

In several weeks of irregular work Mulliken obtained two extreme portions of mercury, each of which contained \(22\ \mathrm{cm}^3\) (300 g), the increase in atomic weight of the heavy fraction being \(\Delta M = +0.0504\) unit, and the decrease in atomic weight of the light one \(\Delta M = -0.0512\). As we see, he obtained a separation of the same order as that obtained by Brönsted and Hevesy; however, the quantities of enriched mercury obtained by Mulliken are approximately 100 times greater than those of the authors named, and this must be taken into account in evaluating the results, since, if these \(22\ \mathrm{cm}^3\) are reduced to \(0.2\ \mathrm{cm}^3\), a further strong increase in enrichment can be achieved. Undoubtedly, with Mulliken’s apparatus a considerably higher separation can be obtained.

Harkins and Madorsky[^11] developed an apparatus made of steel, operating on the principle of Brönsted and Hevesy, but more rapidly.

Finally, by Mulliken’s method (a combination of diffusion with evaporation), with almost exactly the same apparatus, except for improvements to accelerate the changing of the paper through which diffusion proceeds (which must be done from time to time), Harkins and Mortimer in 1928 carried out the separation of mercury.[^12] They obtained \(7.5\ \mathrm{cm}^3\) of mercury with atomic weight 0.0962 unit higher than ordinary mercury, and \(8\ \mathrm{cm}^3\) of mercury with atomic weight 0.0931 unit lower than ordinary mercury. Thus the total separation between the extreme portions reached 0.189 atomic-weight unit. But the heaviest and the lightest fractions contained approximately 100 g of mercury each, and according to the authors’ statement, with the aid of a small apparatus it was possible in less than one day to increase the total separation to 0.23 unit, and, when additional portions were used, to 0.25 unit. This requires only reducing the volume in the final portions to \(\sim 0.2\ \mathrm{cm}^3\), i.e. to the volume of the extreme fractions obtained by Brönsted and Hevesy. Thus, Harkins and Mortimer obtained \(7.5\ \mathrm{cm}^3\) of mercury of atomic weight 200.706 \((\Delta M = +0.0962)\)

SEPARATION OF NON-RADIOACTIVE ISOTOPES

and \(8\ \mathrm{cm}^3\) of mercury of atomic weight \(200.517\) \((\Delta M=-0.0931)\). Until now no greater enrichment of mercury isotopes has been achieved; it should also be pointed out that only in August 1932 was a better separation of the isotopes of another element (neon) achieved.

The authors consider that, for the individual isotopes of mercury, the results of their experiments appear as follows:

Isotope	198	199	200	201	202	204
Initial content in percent	12,90	16,13	22,58	9,68	32,26	6,45
Percent content in the light fraction	14,09	17,01	23,00	9,52	30,65	5,73
Percent content in the heavy fraction	11,51	15,03	21,96	9,83	34,23	7,45

Let us compare with one another the principal results of the enrichment of mercury. \(+\Delta M\) denotes an increase in the atomic weight of the heavy fraction, \(-\Delta M\) a decrease in the atomic weight of the light fraction; \(v_+\) and \(v_-\) are the volumes of the heavy and light fractions, respectively.

Author	\(+\Delta M\)	\(-\Delta M\)	\(v_+\) in \(\mathrm{cm}^3\)	\(v_-\) in \(\mathrm{cm}^3\)
Brenshtein and Hevesy	0,046	0,052	0,2	0,2
Harkins and Madorsky	0,052	0,014	0,28	0,32
Mulliken	0,0504	0,0512	22,10	22,00
Harkins and Mortimer	0,0962	0,0931	7,50	8,00

If one takes into account the quantity of mercury and the degree of enrichment, it is evident that we have listed the results in the order of improvement of the separation results.

We shall now turn to the works mentioned above on the development of a new, more perfect method for separating isotopes by means of diffusion, and to its application to the separation of neon isotopes.

The idea of this extremely ingenious and simple method in essence belongs to Hertz (Berlin). It was published in 1932 [13] together with the results of applying this fruitful method to the separation of neon isotopes. The essence of the method consists in forcing the process to proceed automatically until a stationary state is established in the sense of

cessation of further separation, whose degree is determined by the design of the instruments.

The entire apparatus consists of identical cells, of the type shown in Fig. 4, connected in series. \(R\) and \(S\) are two porous tubes separating the gas entering each of them into two parts of different composition with respect to the concentration of molecules of different mass that make up the gas stream. The gas enters the cell through \(A\), passes through the first porous tube \(R\), and the part of the gas enriched with light molecules passes through the walls of tube \(R\) and through \(B\), by means of a mercury pump, is pumped to the inlet of the neighboring right-hand cell. The remaining gas mixture, enriched with the heavy component, passes through the second porous tube \(S\), and, after undergoing secondary enrichment, exits through \(D\) to the neighboring left-hand cell. The gas that has diffused through porous tube \(S\) is pumped out by the mercury pump \(P\) and, being approximately of the same composition as the fresh portions entering at \(A\), mixes with them and again enters the same process. Such cells

Fig. 4.

may be connected in series in any number in such a way that each cell receives gas from both of its neighbors, and both streams into which this gas is decomposed in the cell again go to two neighboring cells. In addition, the cell separates off a third stream, which enters it again. With such a connection of cells, the concentration of the heavy constituent of the gas mixture will increase in one direction, and that of the light constituent in the other.

If, through \(B\), the pump pumps to the neighboring right-hand cell per unit time a fraction \(f\) of the total stream entering at the same time into \(A\), then through \(C\), in the same time, a fraction \(1-2f\) is pumped out, and through \(D\) to the neighboring left-hand cell there will again go a fraction \(f\), enriched with the heavy constituent. If the lengths of tubes \(S\) and \(R\) are denoted by \(l_s\) and \(l_r\), then \(f\) is determined from the relation:

\[ \frac{1}{f}=\frac{l_s}{l_r}+2. \tag{5} \]

In Hertz’s case both tubes \(S\) and \(R\) were equal to each other, so that \(f=\frac{1}{3}\), i.e. the stream was divided into three equal parts. If, for simplicity, it is assumed that the gas mixture consists only of two isotopes, with the heavy isotope present only in a small quantity, then, as Hertz shows, the ratio of the concentrations of the heavy

isotope in both streams leaving the cell is determined by the formula:

\[ q=\frac{f^\mu}{1-(1-f)^\mu}, \tag{6} \]

where

\[ \mu=\sqrt{\frac{m_e}{m_s}}, \]

with \(m_e\) being the mass of the light isotope, and \(m_s\) the mass of the heavy isotope.

The process asymptotically approaches the final stationary state, when each cell receives gas of the same composition as the gas it gives off. In this case the ratio of concentrations in a given cell differs from the ratio of concentrations at the corresponding point of the neighboring cell by the factor \(q\)

Fig. 5.

(the separation factor). Consequently, with \(m\) cells connected in series, we obtain the total separation factor \(Q=q^m\).

Hertz’s apparatus consisted of 24 cells of type \(f=\frac{1}{3}\). For neon isotopes \(\mu=0.953\); in this case \(Q=(1.092)^{24}=8.4\), i.e., upon reaching the stationary state the ratio of the concentration of the heavy isotope to the light one in the last (left-hand) cell should exceed the ratio of the same concentrations in the first cell by 8.4 times. The experiment gave approximately the same \(Q\).

Figure 5 shows a diagram of four cells. Each cell is separated from the neighboring one by a dotted line. At the beginning of the process, vessel \(V_l\) contains the ordinary mixture of gases; during the process the gas flows along the entire system of porous tubes, all the time giving off the light portions backward, and by the time it reaches vessel \(V_s\) it is already a portion enriched in the heavy isotope. At the end of the experiment, upon establishment of a stationary co-

of the state, vessel \(V_s\) contains the heavy fraction, and \(V_l\) the light one. It is clear that, in the case where vessel \(V_s\) is very small compared with vessel \(V_l\), upon establishment of the stationary state the concentration of the heavy isotope in vessel \(V_l\) will not change appreciably. If it is important for us to enrich the gas specifically in the heavy isotope (and, owing to its assumed small concentration, this is precisely what interests us), then in the case \(V_l \gg V_s\) the heavy portion can, by means of a branch tube with a stopcock, be removed from vessel \(V_s\), the stationary state re-established, another heavy portion taken, and so on, until a sufficient quantity of gas enriched in the heavy isotope has been obtained. Then, having pumped out the remaining gas from the apparatus and using the obtained heavy portion as the starting material, one can isolate from it a portion still more enriched in the heavy isotope. On the basis of these considerations, vessel \(V_l\) was taken to have a volume of \(3\) l, and \(v_s\) of \(0.4\) l.

After four hours of operation of the pumps, the stationary state was already established. In order also to obtain a fraction enriched in the light isotope, the two vessels were interchanged. The gas pressure in vessel \(V_l\) was about \(10\) mm, while in vessel \(v_s\), owing to the pressure drop in the tubes through which the gas passes, it was \(2.5\) mm lower.

Hertz examined the fractions obtained, first, by means of Thomson’s parabolas method and, second, measured the intensities of the components of the isotopic fine structure of the spectral lines of neon. It turned out that, whereas for ordinary neon the concentration ratio \(Ne^{20} : Ne^{22} = 9 : 1\), the intensity ratio of the components of the two isotopes for the first heavy fraction is approximately \(1 : 1\), and for the still heavier one \(Ne^{20} : Ne^{22} = 1 : 2.5\). If it is assumed that neon consists of only two isotopes, this result corresponds to a change in atomic weight by approximately an entire unit, namely from \(20.20\) (ordinary neon) to \(21.15\) (neon enriched by Hertz). One could hardly have expected such enrichment by the old methods.

In addition, another new result was obtained. Namely, on the mass spectrogram of the heavy fraction, with sufficient exposure, besides isotope 21 (known previously and present in an amount of approximately \(2\%\) of the total gas mass), there appears another parabola, corresponding to mass 23. It is possible that this is a new isotope of neon. Kallmann and Lazarev also indicated the existence of this isotope, although the Americans Bleakney and Bainbridge believe that it does not exist.

Hertz considers the possibility of improving the apparatus by changing the shape of the cells in order to increase the separation factor \(q\). From the formula for \(q\) it is seen that, by decreasing the quantity \(f\), i.e. by shortening tube \(R\) and lengthening \(S\) (Figs. 4, 5), one can achieve a substantial increase in the separation factor. However, it turned out that when \(f\) is decreased, the velocity decreases—

…rate of separation, i.e., the time required to establish the stationary state increases. Therefore it is necessary to seek some compromise solution of the problem, ensuring the most advantageous operating conditions of the apparatus by an appropriate choice of cells.

This question was studied in great detail by Harmsen, under Hertz’s direction.^14 He carried out many experiments to determine the separation factor of an individual cell, the rate of diffusion through the walls of porous tubes, and the rate of establishment of the stationary state. For this purpose he took a mixture of neon with helium and, introducing it into cells of various shapes, determined the change in concentration with time at a given point of an apparatus composed of different cells, and the change in concentration along the system of cells in the stationary state. For such a determination of the composition of the gas he sometimes used discharge tubes inserted at various points of the chain and examined the composition spectroscopically; sometimes, with the aid of an electrodeless discharge, he excited to luminescence all the gas contained in the apparatus and observed, in the stationary state, the transition from the pure luminescence of helium to the distinct luminescence of neon.

Fig. 6.

By these experiments Hertz’s theory of isotope separation with the aid of apparatus of this kind was fully confirmed. It turned out that, indeed, with a decrease in the ratio of the lengths of the tubes constituting the cell (a decrease in the value of \(f\)), the separation factor increases, but the rate of establishment of the stationary state decreases.

Of particular interest is the case \(f=0\), i.e., when the length of tube \(R\) is equal to zero \((l_r=0)\). In Fig. 6 a system of two such cells is shown. It turns out that the factor \(q\) can be increased if the diffusion of gases along the tube (which, for a cell with \(f=0\), cannot be neglected) is impeded at the place of the greatest concentration gradient, i.e., in the left part of the cell. This is achieved by narrowing the tube at the left end, as is shown in the figure. A formal theory of tubes with \(f=0\) becomes possible if the part \(DX\) of the right-hand cell is regarded as the right-hand tube of the neighboring left-hand cell.

Experimental investigations showed that the separation factor for cells of type \(f=0\) is approximately 1.5 times greater than for cells of type \(f=\frac{1}{6}\); however, the rate of establishment of the stationary state for these cells is very small. But this rate is located

in direct dependence on the speed with which some component of the gas mixture is transported through the system. The latter is proportional to the concentration of this component of the mixture. Therefore the course of the process in time throughout the apparatus is determined by the place of the smallest concentration, i.e. by the last cell. Thus the last cells must be chosen so that, with an average separation factor, they would have the greatest possible separation rate, while the remaining ones should be chosen so that the separation factor is increased at the expense of speed.

On the basis of his study, Harmsen constructed an apparatus of 24 cells, most of which were cells of the type \(f=0\). Only at the end, where the light component was enriched, were there five cells of the type \(f=\frac{1}{6}\), for the purpose of increasing the separation rate. Tests showed that with such an apparatus the stationary state was reached after 20 hours, and the total separation factor of the whole apparatus reached 35–40. Thus a considerably larger separation factor was obtained here, although at the expense of speed.

Harmsen also applied his apparatus to the separation of neon isotopes. At first the neon was enriched in the heavy isotope in the old apparatus of Hertz, with the volume ratio \((V_l=30\ \mathrm{l},\ V_s=5\ \mathrm{l})\) and with separation factor \(Q=8.4\); after establishment of the stationary state, the concentration ratio was \(\mathrm{Ne}^{20}:\mathrm{Ne}^{22}=2.5:1\). The enriched gas was removed and the process proceeded with fresh neon. In this way nine portions of gas enriched in the heavy isotope were obtained. This quantity of gas was again put into the process, and from it, by means of threefold removal of the portion enriched in the heavy isotope, a mixture was obtained with a concentration ratio of approximately \(\mathrm{Ne}^{20}:\mathrm{Ne}^{22}=1:2\). The heavy fraction obtained in this way then went for further enrichment into the new apparatus.

The residue in the apparatus was used for enrichment in isotope 21 in the following way. The light fraction was separated from it until the new residue gave a concentration ratio \(\mathrm{Ne}^{20}:\mathrm{Ne}^{21}=1:1\). Then the heavy fraction obtained earlier was again separated until the residue in the apparatus also approached the ratio \(1:1\) in concentration. In this case it must have turned out that each time a larger part of \(\mathrm{Ne}^{21}\) remained in the apparatus. The photograph (Fig. 7), by the parabola method, gives the intensity of parabola 21 for this residue as substantially greater than for the other portions of gas. Lines 20 and 22 have almost identical intensity. Parabola 23 is faintly visible. Thus, in principle, enrichment of an isotope whose mass lies between the masses of two others is possible.

The gas obtained from both first separations, with a content of \(\mathrm{Ne}^{22}\) of more than 50%, entered the new apparatus. At first the volume \(V_l\) was taken equal to 10 l, and \(V_s=1\) l. The heavy frac-

the resulting fraction was again fed into the apparatus with \(V_l = 3\) l and \(V_s = 1\) l. After a stationary state had been established in this second process, the light isotope was no longer observed spectroscopically in vessel \(v_s\). Taking into account the separation factor of the apparatus and the concentration in \(V_l\), the author establishes that the ratio of the concentrations in \(V_s\) corresponds approximately to
\(\mathrm{Ne}^{20} : \mathrm{Ne}^{22} = 1 : 80\).

The mass spectrogram (Fig. 8) of this fraction gives, among the neon parabolas, only parabola 22. The remaining parabolas were obtained from water and hydrocarbons introduced in order to serve as a mass scale. Parabola 11 comes from doubly ionized \(\mathrm{Ne}^{22}\). Fig. 9 presents the spectrogram of very pure \(\mathrm{Ne}^{20}\) (parabolas 20 and 10), obtained from the old apparatus.

Fig. 7.

Consequently, by means of Hertz’s method an almost complete separation of the isotopes of neon 20 and 22 was achieved. Such results could not even have been contemplated with the old technique.

7. Separation of the Isotopes of Hydrogen

The isotope of hydrogen with a mass approximately equal to two, discovered in 1932, is of such exceptional interest that the methods for separating it from ordinary hydrogen deserve separate treatment.

Owing to the extremely large relative difference in the masses of H and D, they should be considerably more easily separable than the isotopes of any other element. This should, co-

of course, strongly affects the separability of the hydrogen isotopes by physical methods (diffusion), and also leads to a difference

Fig. 8.

in the chemical behavior of H and D, which makes possible their separation also by means of chemical methods.

Fig. 9.

As early as 1919, Stern and Volmer^15^ attempted to establish whether ordinary hydrogen is a simple element or consists of a mixture of isotopes. To this end they tried to separate the heavy isotopes of hydrogen from the general mixture by the method of diffusion through the walls

porous tube. The authors obtained a negative result, which they explained by the absence of heavy isotopes in hydrogen, but which, obviously, should be explained by the fact that, with their imperfect instruments and methods of determining an increase in concentration (from the change in the density of water obtained from gases taken from the apparatus), it was impossible to notice an increase in concentration. Indeed, with the extremely small initial concentration of D (the natural ratio \(D : H = 1 : 5000\), and in the hydrogen evolved during electrolysis, which the authors used, still considerably smaller), only a sufficiently long process could lead to a noticeable increase in the density of the water obtained.

For the first time the enrichment of hydrogen with the heavy isotope was accomplished in 1932 by Urey, Brickwedde, and Murphy\(^{16}\), (to whom belongs also the credit for the discovery of D) by means of the evaporation of liquid hydrogen at the triple point, i.e. at a temperature close to the solidification temperature of hydrogen. For this purpose the authors used a pressure only a few millimeters above the pressure of the triple point. The vapor pressures arising at the triple point were calculated for the three possible hydrogen molecules \(H^1H^1\), \(H^1H^2\), \(H^1H^3\) (the probability of the molecules \(H^2H^2\), \(H^2H^3\), and \(H^3H^3\) in ordinary hydrogen is negligible), if it is assumed that besides \(H^3\) (D) there is also the isotope \(H^3\).

These pressures depend on the molecular weight of the hydrogen molecules and are in the ratio \(p_{11} : p_{12} : p_{13} = 1 : 0.37 : 0.29\), where \(p_{11}\) is the vapor pressure over solid hydrogen consisting of \(H^1H^1\) molecules, etc. Thus the residue from evaporation should be enriched with the less mobile isotopes. A small residue after evaporation of a 4-liter portion of liquid hydrogen was led into a discharge tube, whose emission spectrum was examined with the aid of a large 21-foot diffraction grating. The authors also photographed the lines of the Balmer series of ordinary hydrogen and, with a colossal overexposure, observed on the plates faint traces of the D lines at the positions calculated for them from Bohr’s theory, i.e. next to the H lines on the short-wavelength side, at distances of the order of \(1—2\ \text{Å}\). In photographs of the spectrum of the enriched portion these lines gained in intensity by approximately a factor of 5. From a comparison of the times required for the appearance on the plates of the H and D lines, the relative concentration of the isotopes was determined. It turned out that for ordinary hydrogen \(D : H = 1 : 4000\), and for the enriched portion \(D : H = 1 : 800\).

In Fig. 10 is shown a photograph of the line of the Balmer series H for ordinary (upper photograph) and enriched (lower) hydrogen. To the left of the central, most intense line \((H_\beta)\), an extremely faint trace is visible in the upper photograph, which in the lower appears considerably more clearly. This is the \(D_\beta\) line, situated, as follows from theory, at a distance of approximately \(1.3\ \text{Å}\) on the short-wavelength side from \(H_\beta\). Two symmetrical intense li-

are “spirits” imparted by the lattice on both sides of Hβ. No traces of the H³ isotope lines were found, from which the authors conclude that, if it is present, it is in a very insignificant amount; this conclusion is also confirmed by later investigations of portions of hydrogen considerably more enriched with the heavy isotope17*. Urey’s attempt, together with his collaborators, to enrich hydrogen with the heavy isotope by evaporating it not at the triple point, but at the normal boiling point, i.e. at atmospheric pressure, was not successful.

Fig. 10.

As has already been said, the concentration of D relative to H in ordinary hydrogen was estimated by Urey and his collaborators at 1 : 4000. However, the investigations of other authors that followed soon afterward give the figures 1 : 30,000; 1 : 40,000 and even 1 : 80,000. These discrepancies were explained by Washburn and Urey18 by the fact that, owing to the ease of separating H and D, it should be expected that in portions of hydrogen obtained by different methods the concentration of heavy hydrogen is different.

Thus, for example, during electrolysis the liberation of the different isotopes of hydrogen on the surface of the cathode must proceed at different rates, as a result of which the gas evolved should become enriched in one isotope, and the hydrogen remaining in solution in the other. As Washburn and Urey note, such separation can hardly be due exclusively to the difference in the mobilities of the ions of the hydrogen isotopes, whose ratio should be approximately \(1:\sqrt{2}\), and practically the whole process must take place on the surface or near the surface of the cathode. Later investigations confirm this assumption of the authors mentioned and lead to the conclusion that in the bulk of the solution itself separation does not occur, in all probability because, at small current densities, diffusion hinders the accumulation of ions of one isotope near the surface of the cathode, while at high current densities mixing processes must always arise, incomparably more effective than diffusion (vigorous formation of gas bubbles). Further, two processes must be distinguished. First, the passage of hydrogen ions from the boundary layer of the solution to the surface of the cathode, and here, at small current densities, the reverse process must also be taken into account—the passage of hydrogen atoms from the cathode into the boundary layer of the liquid; and, secondly, the liberation of hydrogen gas at the cathode, i.e. the formation of molecules from neutral

* Note added in proof.
However, in 1934, Tibbov and collaborators (Phys. Rev. 45, 840, 1934) succeeded in establishing that H³ is present in ordinary hydrogen in an amount of approximately \(1:10^9\) and is enriched during electrolysis (see below).

atoms of hydrogen. Concerning the mechanism of neutralization of hydrogen ions at the cathode surface, there are two principal points of view. According to one of them (Polanyi[^19]), hydrated hydrogen ions approach the cathode, then shed their shell of polarized solvent molecules and reach the electrode surface. In doing so the ion must pass through a certain energy barrier, which depends to a high degree on the mass of the particles; namely, this barrier is considerably greater for the deuton (the nucleus of heavy hydrogen) than for the proton. Therefore the light isotope should be preferentially discharged at the cathode, while the solution should gradually become enriched in the heavy one. In this potential threshold Polanyi sees the cause of the overvoltage.

Another point of view is that of Gurney[^20], who considers that hydrated ions are neutralized by electrons of the cathode passing from its surface to the ion, after which the hydrogen atom is deposited on the electrode. In his opinion, the energy threshold responsible for the overvoltage is created by these electrons issuing from the metal.

Bell[^21] applied Gurney’s theory to the process of separation of the hydrogen isotopes. He came to the conclusion that the separation must be due to the fact that, for the two isotopes, owing to the difference in mass, the energy of neutralization by an electron of the hydrated ion in its lowest vibrational-rotational state is different; this energy enters into Gurney’s formula.

Fowler[^22], sharing Polanyi’s point of view, considers, however, it possible that in some cases Gurney’s theory should also lead to correct results.

It should be noted that experimentally neither the one nor the other theory has yet been sufficiently confirmed; the experimental data concerning their verification are often contradictory, so that it is still difficult to prefer one to the other. Thus, for example, some authors have found that the separation efficiency does not depend on the substance of the cathode, which agrees with Gurney’s theory; others have found that it does depend on the cathode metal; some have found that the current density affects the efficiency, others that it does not, and so on. In the combination of neutralized ions into the hydrogen molecule there also exists the possibility of isotope separation with preferential liberation of the light isotope H.

Urey and co-workers spectroscopically investigated hydrogen obtained from the water of old electrolysis vessels and established a definite increase in the concentration of the heavy isotope as compared with its concentration in ordinary tap water.

It is well known how widely used such electrolytic enrichment of water with the heavy isotope of hydrogen has become, i.e. the production of “heavy” water. The first significant results in this direction were obtained by Lewis and MacDonald[^23], who, by electrolyzing in a vessel with nickel electrodes an alkaline solution of water taken from old electrolysis vessels, obtained

heavy water of density 1.073, i.e., water in which heavy hydrogen constituted 65.7% of all the hydrogen. For this purpose they had to decompose 20 l of solution down to a residue of 0.5 cm³, from time to time neutralizing the solution and distilling it in order to avoid too high a concentration of alkali.

Figure 11 shows a photograph of the first member of the Balmer series of hydrogen from this portion of heavy water, obtained by Lewis and Spedding^17. This line was photographed with a plane grating and a 10-meter astronomical objective in its third order.

Fig. 11.

The two close lines on the right are the line of heavy hydrogen \((D_\alpha)\), whose doublet structure is excellently resolved. The line on the left—the line of ordinary hydrogen \((H_\alpha)\)—is not so well resolved, because for the lighter atoms the Doppler broadening is greater as a result of the higher velocities.

In the cited work of Lewis and Macdonald it was established that the concentration of D in ordinary water should be approximately \(1 : 6500\) relative to H. This figure was subsequently corrected by Bleakney to \(1 : 5000\), which apparently is the most reliable value.

It is understandable why the earlier data on the concentration of D in ordinary hydrogen differed so greatly. Indeed, most often hy-

deuterium is obtained by the electrolytic method, while the concentration of the liberated gases will be different depending on the time during which a given portion of the solution has been in operation.

If the relative concentrations in water of H and D are denoted by \(A_1\) and \(A_2\), respectively, then the observed rates of evolution \(E_1\) and \(E_2\) for both hydrogen isotopes at the cathode are in the ratio:

\[ \frac{E_1}{E_2}=q\frac{A_1}{A_2}. \]

The multiplier \(q\) does not depend on \(\frac{A_1}{A_2}\) and in Lewis’s experiments reached a value of 5. There are indications that the most favorable conditions for separation are the following: electrolysis should be carried out in a solution of NaOH (or KOH) in a vessel with nickel electrodes at a current density of the order of \(1\ \mathrm{A}/\mathrm{cm}^2\). In this case the multiplier \(q\) may reach a value of 6 or 7. However, as already indicated, some authors believe that the efficiency of separation, i.e. \(q\), does not depend on the cathode material or on the current density.

There is also no agreement among the various authors on the question of the dependence of the efficiency \(q\) on temperature. Only one thing is clear: conducting the process at high temperature is extremely disadvantageous simply because of losses of water by evaporation. Therefore, in all installations for the electrolytic production of heavy water, the working vessels are cooled by means of running water.

Subsequently, by prolonged electrolysis, Lewis achieved the result that the percentage content of heavy hydrogen in water reached almost 100%.

At present, in many laboratories of the world, the production of heavy water has been established by the method described. In principle the installations do not differ from one another in any way. The difference lies chiefly in the fact that some authors (Lewis among them) use a very large current strength to accelerate the process (reaching up to 400 A), while others use a comparatively small one, but make the current pass through several vessels connected in series, whereby an increase in the rate of separation is again achieved.

In the USSR, for the first time, as far as the author knows, a small quantity of water enriched with heavy hydrogen for spectroscopic purposes was obtained by Prof. S. E. Frisch and V. I. Chernyaev32 at the State Optical Institute in Leningrad. Spectroscopically, the D content in this portion was estimated at 3%. At present, a portion of water with a much higher concentration of D is being obtained for further experiments. In addition, heavy water is now being produced in the laboratory of Academician Vernadsky at the Academy of Sciences, at the Physical Institute of Leningrad State University, and in several other laboratories.

Among the chemical methods for separating D and H one may point to

the following. If zinc is dissolved in water acidified with sulfuric acid, then in the hydrogen liberated the concentration of H increases in comparison with its concentration in water, and the residue of water, consequently, is again enriched in heavy hydrogen. In this process the efficiency of separation is also very high, being fairly close to the efficiency in electrolysis. In exactly the same way, when hydrogen is liberated from water with the aid of other metals, a certain separation is achieved, although the process is less efficient than in the case of zinc. A. Farkas and L. Farkas^24 dissolved zinc in sulfuric acid containing 25% D. The hydrogen liberated in this process contained only 8% D, i.e. the rates of liberation of H and D in this case were related approximately as 4 : 1. For Al, Ca and Na the ratios of the rates of liberation are 2; 1.5 and 1.2, respectively. The difference in the rates of liberation probably occurs because, in reactions proceeding at low temperatures (such as those considered above), the differences in zero-point energies for D and H are comparable with the magnitude of the activation energy.

The same authors^33, in addition, obtained some separation of the isotopes of hydrogen by a photochemical method; namely, it turned out that D reacts with Cl in light three times more slowly than H. In this case the difference in the rates of the reactions is in good agreement with the theoretical value of the difference in zero-point energies for the molecules H₂, HD and D₂.

Taylor and his co-workers^25, on the basis of Eyring’s theoretical considerations, carried out the following method of separating D and H. They removed hydrogen from activated charcoal that had previously absorbed the gas, maintained at the temperature of liquid air, and investigated with a mass spectrograph the last portions removed. Of the initially absorbed volume of 5025 cm³, the last portion of 35 cm³ was examined, and it proved to be enriched in HD molecules threefold. Then a portion of 5715 cm³ was reduced to a residue of 440 cm³. This latter was again absorbed, and the final evolved portion, with a volume of 90 cm³, was taken for investigation. Here the enrichment in HD molecules was already fivefold. Experiments at the temperature of liquid hydrogen should be considerably more effective.

In electrolysis, as was established by Lewis and Macdonald^23, and also by Selwood and Frost^26, the residue of water is enriched only in the heavy isotope of hydrogen, while the concentration of oxygen isotopes does not change appreciably. However, experiments can be carried out in which, simultaneously with the enrichment of water in the isotope H², it is also enriched in the heavy isotopes of oxygen (O¹⁸ and O¹⁷).

Washburn and Smith^27, and also Lewis and Cornish^28, evaporated water in a distillation flask and obtained an increase in the density of the residue at the bottom of the vessel and a decrease in the density of the first portion of evaporated water. Lewis^29, having obtained in the distillation flask a residue of density

1.000182, by means of an extremely ingenious method, into the description of which we shall not enter, found that in any case the increase in density, equal to 0.000097, was due to the presence of a residue of heavy water, while the concentration of the heavy isotopes of oxygen caused an increase in density of 0.000073. Thus the origin of part 0.000170 of the total value 0.000182 of the initial excess density was established.

Heavy water can be separated from light water by using the dependence of the adsorption capacity of porous materials and powders with respect to water on the isotopic composition of the water. Washburn and Smith[^27] immersed 300 g of activated charcoal for three weeks in 500 g of water of density 1.000053. The unabsorbed residue of the water and the last portion, separated from the charcoal by heating in vacuum to 500°C, were compared by density. It turned out that the specific weights of these portions differed by 12.4 parts per million. At the same time the density of the first decreased by 6.5 parts, while that of the second increased by 6.7 parts.

Extremely valuable results on the separation of the isotopes of hydrogen were obtained by Hertz[^30] with the aid of his apparatus for separating isotopes by diffusion. For hydrogen the ratio of the masses of the isotopes is very large, so that the Hertz separation factor \(q\) is large; however, because of the very small initial concentration \(D\), the accumulation process must proceed very slowly. Hertz therefore increased the number of cells in his apparatus to 48, so that the total separation factor for the whole apparatus \((Q = q^m)\) as applied to hydrogen reached \(10^{12}\). Water from an electrolytic apparatus, which contained about 0.1% heavy hydrogen, was used as the starting material. By the action of magnesium vapor on this water, hydrogen was liberated. Since this hydrogen contained chiefly the molecules \(\mathrm{H_2}\) and \(\mathrm{HD}\) and only a very small amount of \(\mathrm{D_2}\), a discharge tube was placed at a certain point in the apparatus, the discharge of which led to dissociation of the hydrogen molecules. Since at this point the concentration of \(\mathrm{HD}\) molecules had already increased in comparison with the concentration at the beginning of the apparatus, upon recombination a greater probability was created for the formation of the molecule \(\mathrm{D_2}\). Thus the \(\mathrm{HD}\) molecules partly passed into \(\mathrm{D_2}\), and partly into \(\mathrm{H_2}\), which increased the rate of separation. The gas from the vessel in which the heavy fraction was collected was used to fill discharge tubes with carefully calcined electrodes. In the emission spectrum of these discharge tubes no traces of the H lines were noticeable; all the lines belonged to the spectrum of D. We reproduce a photograph of the lines \(\mathrm{H_\alpha}\) and \(\mathrm{D_\alpha}\) in the second order of a small concave diffraction grating, obtained on one plate from two different discharge tubes, one of which contained pure D, and the other ordinary hydrogen (Fig. 12). When photographing the \(\mathrm{H_\alpha}\) line the upper third of the plate was covered, and when photographing \(\mathrm{D_\alpha}\)—the lower.

A. Farkas and L. Farkas[^31] pointed out two more methods of sepa-

of hydrogen isotopes by means of diffusion. These methods lead to small enrichments, the presence of which, however, is quite possible to establish. First, the authors mentioned pumped out, from a certain vessel, the mixture of light and heavy hydrogen contained there through a narrow capillary, the diameter of which was considerably smaller than the mean free path of the gas molecules enclosed in the vessel. Under such conditions the magnitude of the separation can be calculated from Rayleigh’s formula, which has the following form:

\[ \left(\frac{\left(\frac{H_0}{H}\right)}{\left(\frac{D_0}{D}\right)}\right)^S = \left(\frac{p}{p_0}\right)^{S-1}, \tag{7} \]

where \((H_0)\) and \((D_0)\) are the initial, and \((H)\) and \((D)\) the final concentrations of light and heavy hydrogen in the vessel; \(S\) is the ratio of the molecular velocities, i.e. in this case \(\sqrt{2}\); \(p_0\) is the initial pressure, and \(p\) the final pressure, which for the authors was equal to \(0.04\) mm. In the vessel there was indeed observed an increase in the concentration of D, and numerically in excellent agreement with the values calculated according to Rayleigh’s theory.

The authors give the following table:

\(p_0/p\)	% D observed	% D calculated
1	47.5	—
1.5	50.7	51.7
2	53.0	53.0
3.25	56.0	57.8

Fig. 12.

Some discrepancy in the figures of the last line is explained, probably, by the fact that the pressure used here (about \(0.13\) mm Hg) was too high, and hence the mean free path was small in comparison with the diameter of the capillary used.

It is interesting to note that A. Farkas and L. Farkas, having carried out an experiment on the adsorption of a mixture of both isotopes by charcoal, did not notice preferential adsorption of heavy hydrogen. At \(78^\circ\) K the gas remaining above the surface of the charcoal had the same concentration as the initial gas. Only during pumping was an enrichment of the residue in the heavy isotope observed, which apparently should be explained by the different diffusion rates through the charcoal of the three kinds of molecules present in it (\(\mathrm{H_2}\), HD, and \(\mathrm{D_2}\)). Therefore it is very probable that Taylor’s experiments with collaborators\(^{25}\) gave positive results not because the two kinds of hydrogen possess different adsorptive capacity, but precisely as a consequence of the reason indicated above.

In addition, both Farkas caused a mixture of H and D, contained in a certain vessel, to diffuse through a hot palladium tube heated by an electric furnace. The initial concentration of D in these experiments was from 40 to 50% at a pressure of about 5 mm. It was found that the gas which had diffused through the palladium always had a lower D content than the original portion, while the gas residue was, of course, enriched in the heavy isotope. However, the difference in D content between the initial portion and the portion that had diffused through the palladium decreases with increasing temperature of the palladium tube; at temperatures above 300° practically no separation occurs. At 150° the light isotope passes through palladium in an amount approximately 1.5 times greater than the heavy one, but the amount of gas passing through Pd at such low temperatures is insignificant.

The authors explain the different diffusion rates of the light and heavy isotopes through palladium by different values of the zero-point energy for these isotopes. Some enrichment of the heavy isotope in the gas remaining after diffusion through palladium was also detected spectroscopically by Prof. S. E. Frisch and V. I. Chernyaev^32 at the State Optical Institute.

Because the two isotopes of hydrogen are easily separable from each other by a wide variety of methods, the idea arose that conditions for their separation may be created directly in nature. At present, investigations are being carried out in America on the D content in the water of various natural bodies of water; however, no definite results have yet been obtained.

LITERATURE

1. Jakowjew, Z. Physik. 64, 378, 1930.
2. Hansen, Naturwiss. 15, 163, 1927.
3. Kuhn u. Martin, Naturwiss. 20, 772, 1932; Z. Physik. Chem. 21, 93, 1933.
4. Harkins a. Hayes, J. Am. Chem. Soc. 43, 1803, 1921.
5. Harkins a. Ligett, J. Physik. Chem. 28, 74, 1924.
6. Harkins a. Jenkins, J. Am. Chem. Soc. 48, 58, 1926.
7. Brönsted a. Hevesy, Nature 106, 144, 1920; Phil. Mag. 43, 31, 1922; Nature, July 14, 1921.
8. Egerton a. Lee, Proc. Roy. Soc. (A) 53, 499, 1923.
9. Mulliken a. Harkins, J. Am. Chem. Soc. 44, 37, 1922.
10. Mulliken, J. Am. Chem. Soc. 45, 1592, 1923.
11. Harkins a. Madorsky, J. Am. Chem. Soc. 45, 591, 1923.
12. Harkins a. Mortimer, Phil. Mag. 6, 601, 1928.
13. Hertz, Naturwiss., H. 26, 493, 1932; Z. Physik. 79, 108, 1932.
14. Harmsen, Z. Physik. 82, 589, 1933.
15. Stern u. Volmer, Ann. d. Physik. 59, 225, 1919.

1. Urey, Brickwedde and Murphy, Phys. Rev. 39, 164, 536, 1932; 40, 1, 1932.
2. Lewis and Spedding, Phys. Rev. 43, 964, 1933.
3. Washburn and Urey, Proc. Nat. Acad. Sci. Amer. 18, 496, 1932.
4. Polanyi, Naturwiss. 21, 316, 1933.
5. Gurney, Proc. Roy. Soc. (A) 134, 137, 1932.
6. Bell, J. Chem. Phys. 2, 164, 1934.
7. Fowler, Proc. Roy. Soc. (A) 144, 452, 1934.
8. Lewis, J. Am. Chem. Soc. 55, 1297, 1933; Lewis and Macdonald, J. Chem. Phys. 1, 341, 1933.
9. A. Farkas and L. Farkas, Nature 133, 139, 1934.
10. Taylor, Gould and Bleakney, Phys. Rev. 43, 496, 1933.
11. Selwood and Frost, J. Am. Chem. Soc. 55, 4335, 1933.
12. Washburn and Smith, J. Chem. Phys. 1, 426, 1933.
13. Lewis and Cornish, J. Am. Chem. Soc. 55, 2616, 1933.
14. Lewis, J. Am. Chem. Soc. 55, 3502, 1933.
15. Hertz, Naturwiss. 21, 884, 1933.
16. A. Farkas and L. Farkas, Proc. Roy. Soc. 144, 467, 1934.
17. S. E. Frisch and V. I. Chernyaev, Reports of the Academy of Sciences of the USSR, III, 245, 1934.
18. L. Farkas and A. Farkas, Naturwiss. 22, 218, 1934.
19. Oliphant, Shire and Crowther, Nature 133, 377, 1934.