Full Text
Stable Isotopes of Light Elements
A. I. Brodsky, Dnepropetrovsk
In recent years great advances have occurred in the study of isotopy. From passive observation of isotopes it has become possible to proceed to experimentation on them in the sense in which this is usually understood in chemistry (separation, the study of differences in the properties of isotopic elements and their compounds, and their chemical behavior). So far this has been achieved for hydrogen and for the following elements: lithium, carbon, nitrogen, oxygen, and neon. For the remaining elements, the study of stable isotopes has not yet gone beyond the stage of mass-spectrographic and spectral observations. Thus a review of work on the isotopes of the elements listed is, at the same time, a review of the present state of isotope chemistry.
Hydrogen is not included in this review, since the literature on it is so extensive that it would require a separate article. Moreover, several surveys on hydrogen isotopes already exist, including ones published in this journal. The review also does not concern nuclear reactions and the recently discovered radioactive isotopes artificially obtained as a result of these reactions. I recently published a survey on oxygen isotopes \(^{8,19}\), covering the literature up to the end of 1936. Here only additions to it are given (chiefly works of 1937).
In arranging the elements I have departed from the usual order for reasons of convenience in describing isotope-separation methods. The review covers all the most essential material published up to the beginning of 1938.
Table 1 gives the atomic masses of the first ten elements of the periodic system and of their stable isotopes according to the latest summary by Livingston and Bethe \(^{1,3}\). These numbers are based on an analysis of the most reliable mass-spectrographic investigations by the doublet method (chiefly Bainbridge’s) and on the study of the energy balance of the corresponding nuclear reactions. In the same table the most probable isotope proportions are given. The justification for the numbers is given in the corresponding places in the text. The masses are given on the isotope scale \((\mathrm{O}^{16} = 16.0000)\). Table 1 also contains atomic weights calculated from the masses and the isotope proportions, divided by 1.00025 for conversion from the isotope scale to the chemical scale \((\mathrm{O} = 16.0000)\). This
TABLE 1
Isotopic composition of the elements
| No. in order | Nucleus | Mass | Proportion | Atomic weight, calculated | Atomic weight, according to the 1938 table |
|---|---|---|---|---|---|
| 1 | H¹ D² T³ |
1,00813 ± 0,00002 2,01473 ± 0,00002 3,01705 ± 0,00007 |
~1/6600 <1·10⁻⁷ |
1,0080₇ | 1,0081 |
| 2 | He⁴ | 4,00389 ± 0,00007 | 4,0028₄ | 4,003 | |
| 3 | Li⁶ Li⁷ |
6,01685 ± 0,00020 7,01818 ± 0,00013 |
7,9 92,1 |
6,937₂ | 6,940 |
| 4 | Be⁸(?) Be⁹ |
8,00792 ± 0,00028 9,01504 ± 0,00025 |
~0,05 ~99,95 |
9,00₉ | 9,02 |
| 5 | B¹⁰ B¹¹ |
10,01631 ± 0,00025 11,01292 ± 0,00017 |
~20 ~80 |
10,81₁ | 10,82 |
| 6 | C¹² C¹³ |
12,00398 ± 0,00010 13,00761 ± 0,00015 |
99,0 1,0 |
12,010₈ | 12,010 |
| 7 | N¹⁴ N¹⁵ |
14,00750 ± 0,00008 15,00189 ± 0,00020 |
99,62 0,38 |
14,007₆ | 14,008 |
| 8 | O¹⁶ O¹⁷ O¹⁸ |
(16,00000) 17,00451 ± 0,00007 18,00369 ± 0,00020 |
99,77 0,04 0,19 |
(16,0000) | (16,0000) |
| 9 | F¹⁹ | 19,00452 ± 0,00017 | 18,999₆ | 19,00 | |
| 10 | Ne²⁰ Ne²¹ Ne²² |
19,99981 ± 0,00011 20,99968 ± 0,00023 21,99864 ± 0,00035 |
90,0 0,27 9,73 |
20,192 | 20,183 |
the atomic weights are compared with those which are given in the table of the International Commission on Atomic Weights for 1938^A,7, for which direct determinations by ordinary physical and chemical methods were used. The agreement of the numbers in both last columns is very remarkable.
Most modern methods of isotope separation are based on the fact that the substance being separated is distributed between two volumes (1) and (2), which are in the same or in different states of aggregation. In such a separation partial fractionation of the isotopes occurs, repeated many times, which leads to their more or less complete separation. If the mixture co-
consists of only two isotopes1, then we shall define the separation coefficient as the ratio
\[ \alpha=\left(\frac{N}{1-N}\right)_1:\left(\frac{N}{1-N}\right)_2, \tag{1} \]
where \(N\) and \(1-N\) are the atomic fractions of the two isotopes in the mixture. The magnitude of it characterizes the effectiveness of the given method. If the fractionation process is repeated \(p\) times, then, as is easy to see, expression (1) remains valid with \(\alpha\) replaced by \(\alpha^p\), if for each subsequent operation the product of the preceding one is used. Passing to infinitesimal quantities of the mixture and integrating from its initial amount \(V_0\) to the final \(V\), we obtain from (1) the well-known Rayleigh formula
\[ \left(\frac{N}{N_0}\right)^{\alpha} \left(\frac{1-N_0}{1-N}\right) =\vartheta^{\alpha-1}, \tag{2} \]
where \(N_0\) and \(N\) are the initial and final atomic fractions of one of the two isotopes when the amount of their mixture is reduced as a result of fractionation by \(\vartheta=\frac{V_0}{V}\) times.
In the case of sufficiently small \(N_0\) and \(N\) in comparison with unity, relations (1) and (2) become
\[ \alpha=N_1/N_2 \tag{3} \]
and
\[ N/N_0=\vartheta^{\frac{\alpha-1}{\alpha}}. \tag{4} \]
This case applies to the isotopes of hydrogen, carbon, nitrogen, and oxygen, if the enrichment is not carried to too great a value.
The Rayleigh formula is valid for a separation process in which the fraction being separated off is not returned to the mixture being enriched (for example, simple distillation or a separate stage of electrolysis). The more general case of a continuous process (for example, distillation with a fractionating column) will be considered below.
The relations obtained follow formally from the definition of the separation coefficient and acquire concrete content only after the quantity \(\alpha\) has been determined through the physical or chemical properties of the isotopes being separated (vapor pressures in the case of distillation, molecular weights in the case of diffusion, equilibrium constants in the case of the use of exchange reactions for separation, etc.).
I. Lithium
Isotopic composition. Two stable isotopes of lithium have been reliably found: \(\mathrm{Li}^6\) and \(\mathrm{Li}^7\), discovered in 1921 simultaneously by Aston and Dempster,[^2] with the aid of a mass spectrograph. The existence of \(\mathrm{Li}^6\) was then confirmed by a number of authors on the hyperfine structure
of the atomic spectrum of lithium and on the bands Li₂ and LiH. Recently BrewerB,16, with the aid of a new mass spectrograph constructed by him, found the isotope Li⁵ in the amount of 1/20,000 of Li⁷, but this was not confirmed by subsequent investigations by Sampson and BleakneyS,1.
The proportion Li⁷ : Li⁶ = 92.1 : 7.9 = 11.66, given in the last table of the International Atomic Commission,A,6, is confirmed by a number of new mass-spectrographic investigationsB,15; B,16; B,17. We shall not dwell on earlier contradictory and, apparently, inaccurate mass-spectrographic and spectral determinations. In one of Brewer’s recent papersB,7 he points to a correction that must be introduced into this number because of isotope fractionation during the emission of positive ions from the anode of the mass spectrograph. An incandescent platinum disk impregnated with a small amount of lithium salt at first emits both isotopes in the proportion 11.60 : 1, but then this ratio increases, reaching 14 : 1 in the last portion. For potassium and rubidium an analogous effect is practically absent. Brewer explains this difference by the fact that a necessary prerequisite for fractionation of isotopes during emission is the formation of their solution in the anode metal. This is possible for lithium ions with a radius of 0.6 Å and impossible for the large potassium and rubidium ions with radii of 1.33 and 1.48 Å, which cannot be situated between the lattice nodes of the anode metal, but enter it as a structural element. The calculated change in the isotope ratio during the course of emission greatly exceeds the observed effect. For an approximate calculation it is sufficient to take into account the difference in the numbers of collisions of the two isotopic ions with the anode surface, proportional to the square roots of their masses. In this way Brewer obtains Li⁷ : Li⁶ = 11.60√(7/6) = 12.52.
In natural samples of lithium compounds no appreciable change in the proportion of the two isotopes has so far been found. In particular, mention should be made of Brewer and Baudis’sB,18 investigation of fossil algae Cryptozoa of the Novocambrian period from the vicinity of New York. Along with a considerable increase in the content of the heavy isotope of potassium K⁴¹, they found a normal proportion of 11.8 ± 0.1 for the lithium isotopes. An increased content of K⁴¹ was also found in living algae.
For the atomic weight of Li⁷, the most reliable are the mass-spectrographic measurements of Bainbridge and JordanB,4. The separation of the doublet Li⁷⁺ — N¹⁴⁺⁺ is equal to 0.01443 ± 0.00010 atomic-weight units, which gives Li⁷ = 7.01822 ± 0.00014. From the balance of nuclear reactions involving Li⁷, OliphantO,2 finds the close value 7.0180. The same nuclear reactions make it possible to find the mass of Li⁶ with great accuracy. The number in Table 1 is based on the balances of the reactions Li⁶ + D² (Cockcroft and WaltonC,3) and Li⁶ + D² = Li⁷ + H¹ (Oliphant, Kempton and RutherfordO,3). It is close to Oliphant’s number 6.0167O,2.
These atomic weights are considerably higher than those which were previously obtained by various authors with the aid of the mass spectrograph. They
lead to the value of the atomic weight of lithium on the chemical scale, coinciding with that which has been given unchanged in the tables of the International Commission on Atomic Weights ^A,7^ for a number of recent years.
Separation of isotopes in the mass spectrograph. The first positive results in the separation of lithium isotopes were obtained by separating both ion beams in a mass spectrograph. Aston had already long ago indicated this route; however, an approximate calculation made by him in 1922 ^A,2^ showed the hopelessness of this method for mass spectrographs of the construction then available. New instruments with more powerful emission gave quite satisfactory results in this direction. In 1934 Oliphant, Shire, and Crowther ^O,4^ obtained several micrograms of completely separated lithium isotopes, which were deposited in the mass spectrograph on metal disks cooled with liquid nitrogen and placed in the paths of both beams. The deposits were converted into lithium chloride by treatment with hydrogen chloride and were then used for certain nuclear reactions. This work will enter the history of science as the third case, after neon and hydrogen (1933), of complete isotope separation. Soon afterward similar results were achieved by Smythe, Rumbaugh, and West ^S,4^, and later Rumbaugh ^R,1^, having increased the anode emission in the mass spectrograph to 0.004 A, obtained appreciable yields. He collected 20 samples of separated lithium isotopes weighing from 5 to 1000 μg (the apparatus gives 0.1 mg of Li^7 in 1 hour). These quantities are sufficient not only for nuclear reactions, but also for certain spectral and chemical investigations.
It should be noted in passing that the same authors ^S,4; S,5^ achieved, by the method described, a considerable separation of the isotopes of potassium, and direct evidence was obtained that, of the three isotopes of potassium (K^39; K^40; K^41), the carrier of the natural radioactivity is K^41.
Electrolytic separation. The successful separation of hydrogen isotopes by electrolysis prompted a number of investigators to attempt to apply this method also to other elements. The comparatively large relative difference in the masses of the two lithium isotopes (Li^7 : Li^6 = 1.17) gave grounds for hoping that in this case the conditions for separation would be favorable. It should be noted, however, that the efficiency of electrolytic separation depends not only on the difference in masses alone. For example, for O^18 and O^16 the mass ratio is 1.12, i.e., a value close to the same ratio for the lithium isotopes; meanwhile, as we shall see below, separation of the oxygen isotopes by electrolysis is a hopeless task, whereas positive results were obtained for the lithium isotopes ^1^).
^1^) For O^16—O^18 the coefficient of electrolytic separation is equal to 1.008, whereas for Li^6—Li^7 it has a value close to 1.025. This corresponds to a separation approximately \(3^n\) times more effective after \(n\) stages of electrolysis.
There is as yet no reliable explanation of the mechanism of electrolytic separation of isotopes, and therefore it is difficult to make theoretical predictions as to the course of the fractionation of lithium isotopes by this method. The electrolytic separation of hydrogen isotopes has been the subject of many investigations. Although it has come into wide practical use, there is still no generally accepted theory of it. It is beyond doubt that the difference in the equilibrium potentials of the hydrogen and deuterium electrodes is not large enough for the separation to be explained by it. This difference amounts to only 0.003 V according to Abel, Bratu, and Redlich A, 1, or is equal to zero according to Drucker D, 2. It is also beyond doubt that the difference in the mobilities of the isotopic ions in solution plays no essential role. This follows already from the fact that the coefficient of electrolytic separation remains approximately constant in solutions of acids, bases, and neutral salts B, 20 ¹).
The simplest attempt to explain the separation of hydrogen isotopes during the electrolysis of water was made by A. and L. Farkas F, 1, who attributed the separation to a secondary reaction at the cathode between the hydrogen evolved and the water of the electrolyte:
\[ \mathrm{HD} + \mathrm{H_2O} \to \mathrm{H_2} + \mathrm{HDO}. \]
The equilibrium of this reaction is shifted in the direction indicated by the arrow, which leads to the accumulation of deuterium in the residual electrolyte. In the event that this equilibrium had time to become established and that it were the sole cause of the separation, the ratio \(\left(\frac{\mathrm{H}}{\mathrm{D}}\right)_{\mathrm{gas}} : \left(\frac{\mathrm{H}}{\mathrm{D}}\right)_{\mathrm{solution}}\), equal to the separation coefficient, would practically coincide with the equilibrium constant of the reaction mentioned, i.e. would be equal to 3.3 at room temperature. However, both of these assumptions are not fulfilled. Direct experiments have established that exchange between gaseous deuterium and water proceeds slowly, so that the mechanism described could have decisive significance only at extremely low current densities (see, for example, W, 6). Under ordinary conditions it can lead only to some decrease in the separation coefficient, the value of which is usually considerably higher than 3.3.
It is natural to ascribe electrolytic separation to one or another nonequilibrium cathodic process; in choosing among them we encounter the same difficulties as in the theory of overvoltage, with which the theory of electrolytic isotope separation is closely connected. According to the views now prevailing, hydrogen overvoltage depends on a delay in the neutralization of \(\mathrm{H^+}\) at the cathode. This point of view was developed in detail in the well-known works of Volmer, Frumkin, and Terny. On this basis, Topley and Eyring T, 3 and Bell B, 9 attributed the enrichment of the electrolyte with deuterium to the discharge of \(\mathrm{D^+}\) at the cathode proceeding more slowly than the discharge of \(\mathrm{H^+}\). The assumptions underlying the calculation of the first authors are not irreproachable, and the values of the separation coefficient calculated by them exceed the observed values by a factor of 2–3. Na-
¹) See also the theoretical study by Fowler F, 3.
STABLE ISOTOPES OF LIGHT ELEMENTS
on the contrary, Bell’s calculation, based on Gurney’s overvoltage theory, gives correct numbers. Bell’s final approximate formula has the form
\[ \ln \alpha=\frac{\Delta E_0}{\gamma kT}, \tag{5} \]
where \(\alpha\) is the coefficient of electrolytic separation, \(\Delta E_0\) is the difference (for two isotopes) of the zero-point hydration energies of unexcited ions, and \(\gamma(>1)\) is a quantity determined by the form of the energy barrier between the hydrated ion and the ion on the electrode surface. The empirical value of \(\gamma\) is close to two. For \(\Delta E_0\) one may take the difference of the heats of hydration of \(\mathrm{D}^+\) and \(\mathrm{H}^+\), close to \(1500\) cal. Putting \(\alpha=5\), we obtain \(\Delta E_0=2.1380\lg 5=1930\) cal, i.e. an acceptable value. Bell does not confine himself to the case of hydrogen, but also gives a calculation for lithium isotopes. For this he calculates \(\Delta E_0\) of hydrated lithium ions by means of the well-known relation
\[ \Delta E_0=\frac{1}{2}h\cdot\Delta\nu =\frac{h}{4\pi}\cdot\sqrt{f}\cdot\left(\Delta\frac{1}{\sqrt{\mu}}\right), \]
where \(\nu\) is the natural frequency of oscillation of the ion in the hydrate, \(\mu\) is its reduced mass in the same hydrate, and \(f\) is the constant of the quasi-elastic binding force, which Bell considers the same for both isotopes and proportional to the hydration energy of the \(q\)-ion. Hence, and from (5), after some computational simplifications it is easy to obtain
\[ \frac{\alpha_{\mathrm{Li}}}{\alpha_{\mathrm{H}}} = \exp\left[ \left(\frac{q_{\mathrm{Li}}}{q_{\mathrm{H}}}\right)^{1/2} \left( \frac{\sqrt{7}-\sqrt{6}}{\sqrt{2}-\sqrt{1}} \cdot \frac{\sqrt{1\cdot2}}{\sqrt{7\cdot6}} \right) \right]. \]
From the heats of hydration of solid halide salts Bell finds \(q_{\mathrm{H}}=200\) and \(q_{\mathrm{Li}}=15\) k² cal, whence, for \(\alpha_{\mathrm{H}}=5\), he calculates \(\alpha_{\mathrm{Li}}=1.04\). This value, as we shall see below, is of the same order as the experimental one. However, Elkin and Bratscher \(^{E,2}\) consider that the values adopted by Bell for \(q\) and \(\alpha_{\mathrm{H}}\) are incorrect. Taking \(\alpha_{\mathrm{H}}=15\) (too large a value), \(q_{\mathrm{H}}=270\), and \(q_{\mathrm{Li}}=150\) k² cal, they obtain, by means of the same relation, the implausibly large value \(\alpha_{\mathrm{Li}}=1.21\). Thus the question of the correctness of Bell’s calculation remains open, especially since the very foundations of his theory inspire doubts.
The theory of the separation of hydrogen isotopes was examined in detail by Urey and Teal \(^{0,2}\), who also took into account the possibility that isotope separation depends on the different rates of recombination of H and D atoms on the electrode. Proceeding from Bell’s premises, by a more exact calculation (taking the frequencies from T, \(^{3}\)) they obtain \(\alpha\) equal to \(4\)—\(4.5\). The second possibility is analyzed by Halpern and Gross \(^{H,1}\). Assuming that recombination obeys a second-order equation and that the activation energies for recombination in \(\mathrm{H}_2\), HD, and \(\mathrm{D}_2\) differ only by the values of the zero-point energies of these molecules, they find 11—13 for the upper
of the limit \(\alpha_{\mathrm H}\), which has the order of magnitude of an experimental quantity. The decisive role of the recombination rate is indirectly confirmed by the new work of A. Farkas \(^{F,2}\), who found that in the electrolysis of diluted heavy water with a palladium cathode, previously saturated with hydrogen, the ratio
\[ \frac{\mathrm H}{\mathrm D} \]
in the occluded gas is \(1/3\) greater than in the escaping gas.
Horiuti and Okamoto \(^{H,10}\) accurately measured the value of \(\alpha\) for the separation of hydrogen isotopes in acid solutions and found that the cathode metals can be divided into two groups: one with \(\alpha\) equal to 6—7.5 (Ni, Au, Ag, Cu, Pt), and another with \(\alpha\) about 3 (Sn, Hg). This was confirmed by recent measurements of Walton and Wolfenden \(^{W,6}\), who also discovered a different influence of temperature on \(\alpha\) in the two groups. Horiuti and Okamoto, in contrast to earlier authors, assume two mechanisms for the overvoltage and separation of hydrogen isotopes. The first, which they call catalytic, predominates in the first group of metals and has as its slow stage the recombination of two hydrogen atoms adsorbed at active sites on the metal. The second—the electrochemical mechanism—predominates in the second group of metals and has as its slow stage the recombination of one hydrogen atom adsorbed on the metal with a hydrogen atom supplied by the electrolyte. For separation on a nickel cathode, Okamoto, Horiuti, and Hirota \(^{O,6}\) calculated the separation coefficient statistically, using Eyring’s transition-state theory, and obtained remarkable agreement with experiment (the calculation gave, for 3% \(\mathrm{D_2O}\) at \(25^\circ\), \(\alpha = 7.1\) instead of the observed 6.9). However, the anomalies with temperature coefficients and the influence of inhibitors noted by Walton and Wolfenden cannot always be reconciled with the dualistic theory of Horiuti and Okamoto, nor with the other hypotheses listed above.
We have had to dwell in greater detail on the separation of hydrogen isotopes in order to show that the theoretical conclusions for this case cannot be directly transferred to lithium and to other elements, about the theory of whose separation by electrolysis nothing reliable can now be said.
Several unsuccessful attempts to separate lithium isotopes by electrolysis did not give much hope for the success of this method. In 1933 Kendall \(^{K,5}\) briefly reported the negative result of a series of experiments carried out in his laboratory. Even earlier Kendall and Crittenden \(^{K,4}\) tried to make use of the difference in the mobilities of the two isotopic ions. According to Lindemann’s calculations \(^{L,2}\), this difference exceeds 1% for free ions. Hydration, especially significant for lithium ions, smooths out this difference, and therefore the attempt apparently ended in failure. Eken and Brattler \(^{E,2}\) electrolyzed a concentrated aqueous solution of lithium sulfate with a cathode of flowing mercury, on which a considerable current density of \(2\ \mathrm{A/cm^2}\) was maintained. The amalgam was then decomposed with water, and the resulting hydrate, after dissolution
in sulfuric acid, was again subjected to electrolysis. Of 30 moles of salt, the first portion, amounting to \(1/8\) of this quantity, was taken for the second electrolysis, and for the remaining four electrolyses the first portions were \(1/4\) of the quantities taken for each preceding stage. Thus the final residue amounted to 0.015 mole. After each electrolysis, the atomic weight of the Li obtained was determined by titration of a weighed sample. The accuracy of the determination of the atomic weight is estimated by the authors as 0.01 unit. The difference in the atomic weights of the extreme fractions did not exceed 0.02 unit, and comparison of the individual analyses shows that this difference lies within the limits of analytical error. Champley and RenboC,2 subjected to electrolysis, under analogous conditions, a \(9\,N\) solution of lithium chloride. The electrolysis was carried out in 3 stages down to a residue of 5 g from the initial kilogram for each of the extreme fractions of the amalgam and solution. The difference in atomic weights was \(0\)—0.03 unit, with an accuracy of 0.02 unit. It is easy to understand the reason for the failure of both of the works mentioned. At \(\alpha = 1.02\) (see below), in the experiment of Eigen and Bradtler the enrichment, according to (4), does not exceed 17%, which corresponds to a change in atomic weight of 0.02 unit. This change is still smaller under the conditions of the experiments of Champley and Renbo, where the extreme fractions constituted \(1/200\) of the initial quantity. Thus, with the stated accuracy of determination of atomic weights, both works obviously could not give a positive result.
In contrast to these unsuccessful attempts, definite results were achieved by Urey and TaylorT,2 in a recent work. They decomposed, by a single electrolysis, \(800\ \mathrm{cm^3}\) of 10% lithium hydrate down to a residue of 1 g. The cathode likewise consisted of flowing mercury, which was vigorously stirred, and the anode of an inverted nickel cone. Water cooling maintained the temperature at \(25^\circ\). Twenty such electrolyses were carried out, each time with fresh portions of electrolyte. The residues were combined and again subjected to electrolysis down to a final residue equal to \(1/600\) of the initial quantity. A mass-spectrographic determination, carried out by Brewer, gave an increase in the ratio
\[ \frac{\mathrm{Li}^7}{\mathrm{Li}^6} \]
from the initial 12.51) to 14.2. From these data, by Rayleigh’s formula (2), a very small value is obtained for the separation coefficient, \(\alpha = 1.020\).
Considerably greater success in the separation of lithium isotopes was achieved in the recent work of Lewis and MacdonaldL,1, who made use of exchange between lithium dissolved in amalgam,
1) This ratio is higher than that usually found (11.7). The authors express the supposition that the increased content of \(\mathrm{Li}^7\) in their sample depends on the conditions of obtaining lithium salts: in certain technological processes lithium salts are obtained by exchange of ionic adsorption, which may lead to partial fractionation of the isotopes. These considerations prompted them to the successful attempt, described below, to separate by means of permutite.
and lithium ions in solution. Here the separation is probably also due to the different rates, for the two isotopes, of the electrode process: $\mathrm{Li}^+ + \theta \to \mathrm{Li}$. This work, exemplary in its execution, should be considered in somewhat greater detail.
The exchange was carried out in a glass column 18 m high and 4 mm in diameter. The column was filled with a solution of a lithium salt.
From above there entered a stream of fine droplets of amalgam which, after passing through the solution, collected at the bottom. The solution displaced by the amalgam left the column at the top and was replaced by fresh solution entering below. The latter was prepared by dissolving lithium from the amalgam that had passed through the column in a solution of HCl in the corresponding solvent. Thus new portions of lithium entered the apparatus only with the amalgam. Decisive for the success of the separation was the absence of mixing of the solution in the column while the amalgam was passing through it. For this purpose the latter was fed from above through a fine orifice, which atomized it into small droplets 0.1 mm in diameter. The drops fell in a zigzag path at such a rate that they passed through the column in 4–5 min. The rate of feed of the amalgam was 100 cm³ in 13–15 min. During this time the preceding 100 cm³ of amalgam had to be freed from lithium and the latter had to be transferred into 100 cm³ of LiCl solution. For this purpose the portion of amalgam poured out of the lower receiver was treated with a solution of HCl in the given solvent until exact neutralization. Excess acid, on contact with the amalgam, gives hydrogen, the bubbles of which mix the solution and disturb the regime. To avoid the evolution of hydrogen, the solvent used was not water, but ethyl alcohol or its mixture (2:1) with dioxane. Washing lithium out of the amalgam was carried out in two vessels. From the column the amalgam entered vessel I, where it was treated until complete neutralization with HCl solution previously used in vessel II. After this the amalgam entered vessel II, where the remaining lithium was washed out with fresh acid. In Fig. 1 the upper and lower parts of the column are shown. Fresh amalgam entered from $P$ through atomizer $M$, and the solution displaced by it left through tube $N$.
Fig. 1.
At the bottom of the column the amalgam collected in $F$, and fresh solution was introduced into $G$ through clamp $E$ with clamp $C$ open (communication with the air). When 100 cm³ of amalgam had collected in $F$, it was let into $G$ by opening clamps $A$ and $B$. At the same time, from $G$ a volume of solution equal to the volume entering was forced into the column.
ing into the amalgam in \(F\). The amalgam was passed through \(D\) into the neutralizer with cock \(C\) open, while vessel \(G\) was filled with the next portion of solution. The pressure of the liquid column at the bottom of the column, equal to 2 atm, doubled while the amalgam was being passed.
For analysis, lithium was extracted from the amalgam in the form of LiCl, then converted into carbonate, a weighed portion of which was titrated with hydrochloric acid. The accuracy was \(0.01\%\). The authors incidentally describe an interesting method for determining isotopic composition from the change in the density of a lithium salt. For this, 1 mg of salt is sufficient. Lithium was converted into LiF, which was fused into a bead. By flotation of fragments of the latter in a mixture of two organic liquids, it was possible to determine the density with an accuracy of up to \(0.1\%\).
The most successful results were obtained in the 3rd cycle, where 10 l of 0.6 N amalgam was passed in 24 hours through a solution of LiCl in anhydrous ethyl alcohol. In the last portions of the solution the ratio \(\frac{\mathrm{Li}^7}{\mathrm{Li}^6}\) fell from the initial 11.6 to 5.1, which corresponds to an increase in the content of \(\mathrm{Li}^6\) in the amalgam1 from 8 to \(16.3\%\), i.e., twofold. From the data obtained one calculates a separation coefficient \(\alpha = 1.025\), in good agreement with the results of the above-mentioned work of Taylor and Urey. If one roughly assumes that the difference in the equilibrium potentials of the electrodes of the two isotopes corresponds to the difference in the zero energies of the hydrated ions, then relation (5) gives 0.0006 V at \(\gamma = 1\), as Lewis and Macdonald suppose, or 0.0003 V at \(\gamma = 2\) (Bell’s formula). Comparison with the work of Taylor and Urey shows that this difference is the same in alcoholic and aqueous solution, i.e., that the ratio of the solubilities of the salts of the two isotopes is the same in both solvents. These approximate calculations are confirmed by the fact that Taylor and Urey found no fractionation of the isotopes in extracting an aqueous solution of lithium bromide with methylamyl alcohol, even when only \(1/1000\) of the initial amount of salt remained in the water. It also follows from this that the separation in the experiments of Lewis and Macdonald must be attributed not to different distribution in the two phases, but to the electrochemical process, as indicated above.
Turning to an assessment of the electrolytic method for separating lithium isotopes, it should be noted that the conditions for work here are not very favorable. However, as Taylor and Urey justly observe, the small separation coefficient found corresponds to electrolysis at low current densities, when the electrode processes are close to equilibrium. It is possible that at higher rates of separation the irreversibility of the electrode processes will increase the value of \(\alpha\). Here one may mention the electrolysis of water, where the equilibrium \(\alpha = 3.3\) for the separation of hydrogen isotopes rises to 6–7 and higher at ordinary current densities, when the rates of nonequilibrium electrode processes acquire decisive importance.
Other methods of separation. Positive results were obtained in attempts to fractionate lithium isotopes by exchange adsorption. Guided by the considerations indicated above, Taylor and Urey[^2] extracted portions of 300 g of permutite with 30 g portions of lithium chloride from a 20% solution. At each extraction the solution was shaken for 20 min. and filtered. During the exchange adsorption \(\mathrm{Li^+—Na^+}\), sodium chloride accumulated in the solution and was removed. The residue from the extraction of 70 g of lithium chloride was examined in a mass spectrograph. The ratio \(\frac{\mathrm{Li}^7}{\mathrm{Li}^6}\) in it increased from 11.6 to 12.7, which, according to Rayleigh’s formula, corresponds to a separation coefficient \(a = 1.022\), close to the electrolytic one. A second experiment consisted in passing a lithium chloride solution through a 10 m column packed with permutite. In this case the first portions of the solution that had passed through the column gave an increase in the ratio \(\frac{\mathrm{Li}^7}{\mathrm{Li}^6}\) to 13.3. In efficiency this method of separation approaches the electrolytic method and will probably prove simpler than the latter if the process is organized according to the principle of continuous fractionation.
II. Oxygen1
Isotopic composition in the atmosphere and in water
The proportion of oxygen isotopes in the atmosphere and in natural compounds still remains not precisely established. Murphy and Brandt[^4], comparing the atomic weights of helium determined from density \((\mathrm{O} = 16)\) and by the mass-spectrographic method \((\mathrm{O}^{16} = 16)\), find for the two atomic-weight scales the ratio 1.00054 instead of the generally accepted 1.00025. This new coefficient gives better agreement between ordinary and mass-spectrographic atomic weights than the old one for the elements fluorine, aluminum, and phosphorus, which have no stable isotopes. For oxygen, however, it leads to the proportion \(\mathrm{O}^{16} : \mathrm{O}^{17} : \mathrm{O}^{18} = 248 : 0.2 : 1\) (if \(\mathrm{O}^{18} : \mathrm{O}^{17} = 5\) is assumed), sharply diverging from the mass-spectrographic and spectral determinations of a number of authors.
The difference in the isotopic composition of oxygen in water and in the atmosphere obtained earlier by Dole, corresponding to a difference in the densities of water of \(6—7 \gamma\), was confirmed in two new works. Smith and Mettisen[^3] compared the densities of river water and of water from atmospheric oxygen and commercial hydrogen. To exclude the influence of different deuterium content, both water samples were normalized with respect to hydrogen by repeated saturation (up to 4 times) with ammonia. The difference in densities after this amounted in two experiments to \(8.5 \gamma\) and \(8.7 \gamma\)—a value somewhat greater than that obtained by Dole. This difference was confirmed by comparing the densities of water from atmospheric and electrolytic oxygen with the same
electrolytic hydrogen, which gave 7.4 γ and 7.9 γ. The authors consider the first method more reliable and take 8.6 γ as the difference in the isotopic composition of the oxygen of water and air.
Direct exchange between the oxygen of air and of water proceeds sufficiently rapidly only on heated catalysts. Morita and Titani^m,6 unsuccessfully attempted to carry out exchange with liquid water or with water vapor on platinum, copper oxides, iron, or nickel at temperatures below 500°. On spongy platinum above 500° the exchange proceeds sufficiently rapidly and to completion. Thus, a mixture of water vapor with an excess content of O^18, corresponding to 18 γ, with electrolytic oxygen from ordinary water in the proportion 1:2, when passed over incandescent spongy platinum at temperatures of 580° and above and at a rate of 1 cm^3/sec, gave an increase in density of 16.3–17.2 γ. After correction for the fractionation of the oxygen isotopes in the electrolyzer, this corresponds exactly to the excess O^18 taken.
Jones and Gold^j,2, by the following simple and elegant method, found the magnitude of the difference in the isotopic composition of the oxygen of air and water. They made use of the same exchange reaction
\[ 2\mathrm{H}_2\mathrm{O}^{16} + \mathrm{O}_2^{18} = 2\mathrm{H}_2\mathrm{O}^{18} + \mathrm{O}_2^{16} \]
between water vapor and air on incandescent platinum. The calculation by Urey and Greiff was extended by them to high temperatures and gave for the equilibrium constant a value practically equal to unity, beginning at 1500° K. Thus, after complete exchange with a large excess of air above this temperature, the oxygen of the water should have the same isotopic composition as the oxygen of the air. Indeed, when water vapor in admixture with air was passed over platinum wire heated to 1800° K for 3 days, an increase in the density of the water by 7 γ was obtained (average of 6 experiments). After correction for slight fractionation of the water in the apparatus, this number decreases to 6.1 γ, in full agreement with the value previously found by Dole. Exchange on incandescent platinum was repeated by Brodsky, Skarre, Dontsova, and Slutskaya^B,2,3, who, after passing 2000 l of air with 75 cm^3 of river water over incandescent platinum wire, obtained an increase in density of 12.6 γ. Of these, 10.7 γ are attributable to equilibration of the isotopic composition of oxygen. The excess over 6–7 γ in the increase in density due to O^18, and the increase of 2 γ due to deuterium, are explained by fractionation during evaporation from the apparatus, which was confirmed by the corresponding calculations.
To the earlier attempts to explain the different isotopic composition of atmospheric oxygen and the oxygen of water must be added the interesting hypothesis of Green and Voskuyl^g,1, in which the principal role is assigned to the vital activity of plants. The carbon dioxide of the air is in equilibrium with the water of the earth’s crust at temperatures on the order of 0°C. As a result of isotopic exchange this leads to a decrease of O^18 in water, equivalent to 10.2 γ, as shown by ther-
hydrodynamic calculations of Urey and Greiff and the experimental determinations of Webster, Wahl, and Urey. This difference in the densities of the particles is compensated by the fact that CO₂ is absorbed by plants, where its oxygen is diluted by the lighter oxygen of water from the soil. If this process too is considered an equilibrium one, then the density difference decreases to 6.8%, which agrees well with the observed value.
The isotopic composition of snow water was determined in the work of Brodsky, Skarre, Dontsova, and Slutskaya\(^{8,23}\). In three portions of snow collected in Dnepropetrovsk they found, relative to river water, a decrease in density equal to 1.9–2.9%, which is in good agreement with the earlier measurements of Riesenfeld and Chang and of Garad and Titani in various countries. Thus snow water has a rather stable isotopic composition. In contrast to previous authors, in this work not only the change in density was measured, but also the change in the proportions of the hydrogen and oxygen isotopes separately. For this purpose the densities and refractive indices were determined simultaneously. From both quantities it was possible to calculate the proportions of the isotopes of both elements. The unexpected result was obtained that, whereas the deuterium content is \(1/5\)—\(1/3\) less than in river water (equal to \(-3.5\)—\(6.2\%\)), the content of \(O^{18}\), on the contrary, is increased by 1–2%.
In the same work the isotopic composition of the high-mountain rivers of the Karachay was determined. Three rivers gave the normal composition of ordinary river water, while three others gave the composition of snow water. This difference is probably explained by the fact that the samples were taken in winter far from the glaciers feeding these rivers, so that their waters were in some cases strongly diluted by spring waters, and in others by melting snow. We have now begun a more detailed study of the isotopic composition of glacial and firn waters and of the waters of high-mountain lakes. The study of the isotopic composition of the waters and ices of the Arctic has also been begun.
A systematic study of the isotopic composition of various atmospheric precipitations is being carried out by Demidenko. For a number of samples of snow and hoarfrost he obtained results agreeing with those given above. The ice of the Dnieper River in winter has an isotopic composition almost coinciding with that of the water.
The verification described below of the influence of \(H_2O^{18}\) on the refractive index of water, carried out by Brodsky and Skarre, eliminated the earlier uncertainty in the dependence coefficients used for determining the isotopic composition of water from densities and refractive indices. The formulas given in\(^{8,23}\) must be somewhat changed, which, however, does not introduce substantial changes into the calculation. If the difference between the refractive indices of the water under investigation and ordinary water is equal to \(\Delta n\), and the difference between the densities is equal to \(\Delta d\), while the excess of deuterium and \(O^{18}\) over their content in ordinary water is equal to \(x\) and \(y\), then
\[ x = 1.224 \cdot \Delta d - 190.5 \Delta n, \]
\[ y = 8.082 \cdot \Delta d + 190.5 \Delta n. \]
Reduced deuterium content together with increased O¹⁸ content in snow water makes it necessary to reconsider earlier hypotheses put forward to explain the different densities of snow and river water, according to which one could expect only a simultaneous increase or decrease in the proportion of both heavy isotopes. This is all the more necessary because the isotopic composition of atmospheric moisture also does not conform to these hypotheses and does not explain the coincidence of the densities of rain and river water. Thus, DoleD,3 found that the water of atmospheric moisture is by 1.3–3.0 γ (on average by 1.5 γ) less dense than lake water, while Okabe and Titani0,5 found in various samples of moisture a decrease from 0.8 to 6.8 γ in comparison with the fresh water of Osaka.
Of considerable interest is the study of the isotopic composition of oxygen in natural compounds, especially organic ones. Up to now this has been done only in the single work of Morita and TitaniM,7, who found an increased content of O¹⁸ in sugar, equivalent to an increase in the density of water by 4 γ.
Concentration of heavy oxygen. In 1937, Geffcken and UreyH,9 published a detailed description of the fractionating column used by them, the results obtained with it, and its theory. Since these data are of significance extending far beyond the particular case of the separation of oxygen isotopes considered here, a detailed description of Urey’s work is given below, supplemented by results obtained in the laboratory of the author of this review.
It is easy to see that substantial concentration of O¹⁸ requires the use of fractionating columns with a very large number of theoretical plates, while the volume of the still and of the column itself must be small. This is possible only when the equivalent length of one theoretical plate in the column is small. The problem was brilliantly solved by Urey’s collaborators through the use of plates made of rotating cones. This design is described below. In the following table, the lengths corresponding to one theoretical plate are compared for different types of packing:
| Type of column/packing | Length corresponding to one theoretical plate |
|---|---|
| Column without packing | 50 cm |
| with beads | 15 cm |
| with glass spirals | 5–8 cm |
| Urey column | about 2.5 cm |
Thus, with a length of 10 m, Urey’s column corresponds to approximately 400 theoretical plates, which provides an enrichment of O¹⁸ by ~40 times in one operation, i.e., by 3.3 times at 100° (α = 1.003) and by 16 times at 60° (α = 1.007). In fact, at 60° only a 4.5–5-fold enrichment was achieved, which should probably be ascribed to too high a rate of evaporation from the still, since otherwise the attainment of a stationary state would have required too much time (many days).
In columns with so large a number of theoretical plates, the time required to achie—
of a stationary state. If, for example, in Urey’s column the boiler has a volume of \(0.5\) liters, and the column is filled with \(4\) liters of water in the form of reflux and vapor, then in order to enrich the boiler by a factor of \(4.5\) it is necessary to introduce into it \(3.5\) g of \(\mathrm{O}^{18}\) and about \(12\) g of it into the column, for a total of \(15.5\) g. Each liter of water fed from above constitutes in the column
\[ N_0 \cdot 1000\left(1-\frac{1}{\alpha}\right)\ \text{g of } \mathrm{O}^{18}. \]
For \(N_0 = 1/500\) and \(\alpha = 1.007\) this amounts to \(1100\) liters of water that must be fed into the column in order to obtain limiting enrichment in the boiler. Even at the high feed rate of \(50\ \mathrm{cm}^3/\mathrm{min}\), attainment of equilibrium requires a time of the order of 20 days. A considerably higher feed rate is hardly attainable without increasing the volume of the boiler, which in turn lengthens the time required to reach equilibrium.
Under these conditions, calculation of the kinetics of the column becomes of great importance; its foundations were given in the work of Gefman and Urey, cited in detail below\(^{18,19}\). Such a calculation not only makes it possible to establish the enrichment and yields reached by a given moment, but also makes it possible to predict the final result long before the stationary state has been attained. This considerably accelerates the preliminary experiments for establishing the most favorable regime. The calculation must be based on some experimentally justified assumption concerning the distribution of concentrations in the column before the stationary state has been reached. Gefman and Urey proceed from the assumption that, by some given moment, a certain part of the length of the column adjacent to the boiler (the case considered here is meant) has reached the equilibrium state, while in the remaining part the initial concentration is retained. The concentration distribution in the equilibrium part is determined by relation (3). As the column operates, the length of the equilibrium part grows, and when it reaches the entire length of the column, limiting enrichment sets in. Further operation of the column does not increase this enrichment, but can only lead to accumulation of the product, if it is withdrawn from the apparatus sufficiently slowly so as not to disturb the equilibrium that has been attained.
The following calculation, based on this assumption of Gefman and Urey, differs somewhat in form from that made by them\(^{1}\). Denote the amount of liquid in the boiler and in the column (including vapor) by \(V\) and \(H\). Both remain constant. The boiler evaporates, per unit time, a certain amount of liquid \(\Delta\), and the same amount returns to it in the form of reflux. In the same time, an amount \(v\) of fresh liquid enters the column from above, and the same amount leaves it in the form of vapor. Let the initial concentration of the heavy isotope be \(N_0\), and let its concentration at time \(t\) in the boiler and (on average) in the column be \(N_t\) and \(n_t\). At time \([t]\) limiting enrichment occurs, when \(N_t = N\) and \(n_t = n\). By the mo—
\(^{1}\) A detailed theory of the column has been considered in the work in press by Skarre and Brodsky.
at time \(t\) an amount \(vtN_0\) of the heavy isotope has entered the column, and an amount \(\dfrac{vtN_0}{\alpha}\) has left it, since on the upper plate the concentration of the escaping vapor, according to (3), is \(\dfrac{1}{\alpha}\) times smaller than the concentration of the liquid. Initially the apparatus contained \(N_0(V+H)\) of the heavy isotope, and at time \(t\) its amount had increased to \(VN_t+Hn_t\).
Writing the balance for the heavy isotope, we obtain\(^1\):
\[ t=\frac{V}{v}\cdot\frac{N_t-N_0}{N_0(\alpha-1)}+\frac{H}{v}\cdot\frac{n_t-N_0}{N_0(\alpha-1)}. \tag{6} \]
Both terms are of essential importance, since in large columns the ratios \(\dfrac{V}{v}\) and \(\dfrac{H}{v}\) cannot be made small. To calculate the mean concentration in the column (without the still) \(n_t\), we shall use the assumption made above. If by time \(t\) a fraction \(\lambda\) of the whole length \(L\) of the column has reached equilibrium, then this fraction corresponds to \(k=p\lambda\) theoretical plates, where \(p\) is their total number. At any height \(l\) (in fractions of \(L\)) the number of plates is equal to \(pl\), and the concentration is \(N_{tL}=N_0\alpha^{pl}\). The total amount of the heavy isotope in the equilibrium part is equal to
\[ \int_0^\lambda N_0\alpha^{pl}\,dH = N_0h\int_0^\lambda \alpha^{pl}\,dl = N_0\,\frac{\alpha^k-1}{p\ln\alpha}, \]
where, for the integration, \(H=hl\) has been put (\(h\) is the filling per unit length of the column). In the remaining part of the column, with volume \(H(1-\lambda)\), the amount of heavy isotope is equal to \(H(1-\lambda)N_0\). Dividing by \(H\) and replacing \(\alpha^k\) by \(\dfrac{N_t}{N_0}\), where \(N_t\) is the concentration in the still, we find
\[ n_t=\frac{N_t-N_0+N_0\ln\dfrac{N}{N_t}}{p\ln\alpha}, \tag{7} \]
since \(N=N_0\alpha^p\). Substituting (7) into (6), we find
\[ t=\frac{V}{v}\frac{N_t-N_0}{N_0(\alpha-1)} + \frac{H}{v}\frac{N_t-N_0-N_0\ln\dfrac{N_t}{N_0}}{N_0(\alpha-1)p\ln\alpha}. \tag{8} \]
The second term is identical with the expression of Teffman and Urey if in it one puts \(1-N=1\) and \(\ln(1-N)=-N\). The first term was added only in the later work of Taylor and Urey\(^{2,3}\), where its calculation was not given. Expression (8) makes it possible to find \(p\) and the limiting enrichment from the found \(N_t\) for some moment before the apparatus reaches the stationary state. For the latter, (8) gives, as is readily seen,
\[ [t]=\frac{V}{v}\frac{N-N_0}{N_0(\alpha-1)} + \frac{H}{v}\frac{N-N_0-N_0\ln\dfrac{N}{N_0}}{N_0(\alpha-1)p\ln\alpha}. \tag{9} \]
\(^1\) Factors \(\alpha\), close to unity, may be neglected.
For the numerical example given above, (9) gives \([t] = 17\) days, in agreement with the number obtained earlier.
If the column operates without the introduction of fresh liquid into it and without loss of vapor, then equilibrium is reached in a considerably shorter time; however, the limiting enrichment is smaller than in the first case. With this mode of operation, of course, \(n < N_0\), since enrichment in the boiler is achieved at the expense of part of the contents of the column.
After these calculations let us proceed to a description of the construction of the Geffcken and Jory column\(^{11,13}\), a preliminary communication on which was given in 1936.\(^{4}\) This column was successfully used for fractionating the isotopes of oxygen, nitrogen, and carbon. Its characteristic difference from other types is that the flat plates are replaced by cones with their apexes downward.
Fig. 2.
Fig. 3.
The stationary cones, fastened to the jacket of the column, alternate with cones rotating between them on a common shaft. The rotating cones have a diameter smaller than that of the column, so that they leave a peripheral gap. The stationary cones are provided with an opening at the apex, through which the shaft passes; vapors and reflux move through this opening, making a long path, as can be seen from Fig. 2. The liquid is thrown by the centrifugal force of rotation over the inner surface of the rotating cone and flows over its edge onto the one below, which is stationary. The vapor moves in the opposite direction. Owing to the long path and the close contact of the vapor with the reflux, it proved possible to accommodate more than 50 pairs of cones per 1 m of length. In its action each pair of cones approxi-
STABLE ISOTOPES OF LIGHT ELEMENTS
...approaches one theoretical plate. After a number of preliminary tests with columns of small height, Themen and Urey built a large column 10.7 m long with 621 pairs of cones. The column consists of 7 sections fastened by flanges and provided on the outside with a heating winding.
Fig. 3 gives a drawing of one middle and two end sections. The details of the construction are described in the original paper. Only the most essential points are given here. The jacket is made of steel tubes with an internal diameter of 183 mm and a wall thickness of 6.4 mm. The steel shaft is 25.4 mm thick. It is rotated by an asynchronous motor placed at the top of the column. The shaft also consists of sections fastened by hinges. Each of the sections is provided with a guide ball bearing fastened to the jacket by a conical support with 8 large cutouts for the passage of reflux and vapor. At the top the shaft is brought out of the column through a ball bearing and a hermetic stuffing box; at the bottom it rests on a thrust bearing placed inside the column. The end sections have 2 openings each for reflux and vapor. The fixed steel cones have slots 44 mm in the center and, at the periphery, flanges 1.2 mm high, clamped between peripheral rings on the jacket, separating them from one another. The rotating cones have a horizontal bottom 40 mm in diameter and are clamped on the shaft between washers, 15.8 mm high, fitted on it. These cones do not come within 40 mm of the jacket. The inclination of the cones is 40°. At the bottom of the column is a boiler, and at the top a condenser.
Preliminary tests were carried out with a section of 15 pairs of cones in which 5% heavy water was concentrated. It was found that the speed of rotation, if not less than 240 rev/min, did not affect the results. The yield, however, fell sharply with a decrease in the reflux ratio. If the latter was of the order of 20, then the column, with respect to the isotopes of hydrogen, was equivalent to 14 plates. In it an enrichment of O¹⁸ by a factor of 1.051 was achieved, which corresponds to 16.5 plates.
Subsequently a column 1.5 m high with 105 pairs of cones was tested, in which 0.3% heavy water was fractionated. It gave \(\alpha^p = 15.1\) for hydrogen and \(\alpha^p = 1.31\) for oxygen, corresponding to 110 and 90 plates. The rate of evaporation from the boiler was 60 cm³/min. Another section of the same length, with 87 pairs of cones, gave an enrichment of O¹⁸ by 25%, corresponding to 74 plates. Thus the number of theoretical plates approaches the number of pairs of cones. Considerably less favorable results were given by the large 10.5-meter column, in which 1.5–6 pairs of cones correspond to one theoretical plate.
Four fractionations were carried out with the large column, the samples being analyzed at different moments of its operation. In experiments I and II the water boiled under atmospheric pressure. Their purpose was to obtain a light fraction. For this purpose the column was provided with a large boiler and samples of condensate were taken at its top.
The stationary state was reached after 70 hours, when the content of \(O^{18}\) in the water fell from 0.00194 to 0.00110—0.00107 atomic fractions, i.e. almost by a factor of 2. The results of the intermediate analyses are given in Table 2. Verification of formula (8) may be
TABLE 2
Operation of the Urey column
| Experiment No. | I | II | III | IV |
|---|---|---|---|---|
| Evaporation rate \(v\), cm\(^3\)/min | 95 | 92 | 45 | 33 |
| Temperature in °C | 100 | 100 | 67 | 65 |
| Ratio \(\dfrac{H}{p}\) | 20,8 | 20,5 | 25,9 | 9,34 |
| \(H\) in cm\(^3\) | 4050 | 4000 | (4000) | (4000) |
| \(p\) | 195 | 195 | 154 | 428 |
| Content of \(O^{18}\) | I: \(t\), hours | I: \(N\) | II: \(t\), hours | II: \(N\) | III: \(t\), hours | III: \(N\) | IV: \(t\), hours | IV: \(N\) |
|---|---|---|---|---|---|---|---|---|
| Content of \(O^{18}\) | 0 | 0,00194 | 0 | 0,00194 | 0 | 0,00194 | 0 | 0,00194 |
| Content of \(O^{18}\) | 13 | 0,00156 | 6 | 0,00159 | 6,5 | 0,00285 | 29 | 0,00405 |
| Content of \(O^{18}\) | 24,6 | 0,00136 | 21 | 0,00134 | 29 | 0,0036 | 91,5 | 0,00587 |
| Content of \(O^{18}\) | 46,8 | 0,00113 | 27 | 0,00124 | 53 | 0,00396 | 141 | 0,00642 |
| Content of \(O^{18}\) | 60,8 | 0,00110 | 42 | 0,00116 | 77 | 0,00414 | 162 | 0,00723 |
| Content of \(O^{18}\) | 70,8 | 0,00109 | 54 | 0,00109 | 100 | 0,00453 | 188,5 | 0,00811 |
| Content of \(O^{18}\) | 62,25 | 0,00108 | 150 | 0,00468 | 218 | 0,00872 | ||
| Content of \(O^{18}\) | 77 | 0,00107 | 175 | 0,00492 | 240 | 0,00830 | ||
| Content of \(O^{18}\) | 195 | 0,00543 | 269 | 0,00828 | ||||
| Content of \(O^{18}\) | 293,5 | 0,00956 | ||||||
| Content of \(O^{18}\) | 307 | (0,00872) |
carried out in the following way. If one neglects the boiler, in which, owing to its large volume (50 gallons), the concentration practically remained constant and equal to the initial \(N_0\), then the quantity \(N_t - N_0 - N_0 \ln \dfrac{N_t}{N_0}\) is a linear function of the time \(t\) (for \(t < [t]\)), which was confirmed for all 4 experiments. From the slope of this straight line it was possible to calculate the ratio \(\dfrac{H}{p}\), since all the remaining quantities are given. On the basis of an analysis of the limiting portion at \(\alpha = 1{,}003\), \(p = 195\) theoretical plates is calculated, whence \(H = 4000\) cm\(^3\). Experiments III and IV were carried out with collection of the heavy fraction. For this purpose a small boiler was taken (\(V = 200\) cm\(^3\)), and samples were withdrawn from it. The distillation was conducted under reduced pressure at boiling temperatures of 67° (\(\alpha = 1{,}0055\)) and 60° (\(\alpha = 1{,}006\)). Experiment III continued for 195 hours,
and experiment IV—307 hours. During this time the limiting enrichment had still not been reached, since the reduced pressure used to increase the dilution coefficient led to a twofold decrease in the rate of evaporation. Assuming, on the basis of experiments I and II, \(H = 4000 \text{ cm}^3\), \(p = 154\) and 428 were calculated in the two experiments. The low effectiveness of experiments I–III depended on the instability of the regime. The limiting enrichment in experiment IV corresponds to an increase in the concentration of \(O^{18}\) almost fivefold, i.e. almost to 1%. After the establishment of equilibrium (about 13 days), 200 \(\text{cm}^3\) of this concentrate per day can be taken from the column.
The fractionation of oxygen isotopes was successfully repeated in the laboratory of the author of this review. In preliminary experiments by Brodskii, Skarre, and Aleksandrovich\(^{B,22}\), a glass column 180 cm high packed with glass spirals was used. An enrichment in the isotope \(O^{18}\) equivalent to 7 \(\gamma\) was obtained, with a total increase in the density of the water of 19 \(\gamma\), in 66 hours of operation (the limiting enrichment had not yet been reached in this time). The isotopic composition of the concentrate was determined by simultaneous densimetric and interferometric measurements and was checked by washing out the excess deuterium by passing CO. By the latter method, 7 \(\gamma\) due to \(O^{18}\) was also found. The small volume of the column, with the high rate of evaporation, made it very difficult to conduct the process with a sufficiently uniform regime, and the contents of the boiler were often diluted with fresh water. This explains the small concentration.
After this, on the basis of a brief preliminary communication by Urey, Huffman, and Pegram\(^{U,4}\), Skarre and Aleksandrovich designed and built a two-meter copper column 15 cm in diameter with 200 stationary and 200 rotating cones. The column differs from Urey’s column by the smaller distance between the cones and by certain constructional details that are not of fundamental importance. The boiler contained about 250 \(\text{cm}^3\) of water, and the contents of the column itself amounted to about 1000 \(\text{cm}^3\). The column worked very uniformly, especially under reduced pressure, and maintaining a constant regime did not present great difficulties. The product was analyzed by the method described above. After 76 hours of continuous operation under atmospheric pressure, an increase in the density of the water in the boiler of 75 \(\gamma\) was obtained, of which 46 \(\gamma\) were due to enrichment in \(O^{18}\). A considerably greater enrichment was obtained at a reduced pressure of 23–24 cm Hg. In one experiment, after 80 hours of operation, an increase in density of 195 \(\gamma\) was obtained, of which 88 were due to \(O^{18}\), corresponding to an enrichment in this isotope of more than 50%. Another experiment with 89 hours of operation gave an increase in density of 204 \(\gamma\), of which 104 \(\gamma\) were due to \(O^{18}\). In all three cases the limiting enrichment was reached. For the three experiments the following number of theoretical plates for heavy oxygen is calculated: 80, 65, and 74, i.e. a sufficiently constant value.
around one plate by 3 cm. Dependence (9) gives, for the time required to obtain the limiting enrichment, 83, 72, and 75 hours. After the third experiment the lower and upper halves of the phlegm were collected and analyzed. For the first, an enrichment by deuterium of 4.4 times and by heavy oxygen of 1.37 times was obtained, while calculation by formula (7) for the limiting case gives 4.0 and 1.42. An equally good agreement between theoretical calculation and experiment was given by analysis of the upper half of the phlegm.
The limiting enrichment obtained in the isotope \(O^{18}\), above 60%, is sufficient for a more or less accurate quantitative study of the equilibrium of the kinetics of the reactions of isotopic oxygen exchange, already begun by us, and also for the use of \(O^{18}\) as an isotopic indicator. Details of the construction and operation of the column are given in the work, now in press, by Skarre and Brodsky.
The enrichment obtained is sufficient for an approximate check of the correctness of isotopic analysis of water by simultaneous determination of densities and refractive indices. For such a check both these quantities were determined by us both in the initial concentrate and after its liberation from excess deuterium by burning its oxygen with hydrogen of normal isotopic composition on a palladium catalyst. The electrolyzer for obtaining oxygen from the water being analyzed is also described in the work mentioned. The two most concentrated solutions gave \(+0.00060\) and \(+0.00073\) for the difference in refractive indices of pure \(H_2O^{18}\) and ordinary water at \(20^\circ\) (for the yellow He line). These numbers are in good agreement with those calculated from a comparison of the density and refractive index of water freed from excess deuterium. The coefficient obtained is probably accurate to within a couple of tens of percent, which is quite sufficient for a reliable isotopic analysis of water that is not too highly enriched. It differs somewhat from the number \(+0.0008\) previously used by us, based on the old data of Lewis and Cornish, but this small difference does not introduce substantial corrections into the analyses of the isotopic composition of natural waters published by us earlier \(^{3,23}\).
The value obtained for the refractive index of \(H_2O^{18}\) makes it possible to find the refraction of this isotope. It is greater by 0.015 units than the refraction of \(H_2O^{16}\) and greater by 0.053 units than the refraction of \(D_2O^{16}\). Thus, of the three water molecules, the first is most capable of deformation.
After successful experiments with the two-meter column, we propose to build a large column 10 m high, with which we expect to obtain an enrichment of \(O^{18}\) by 8–9 times.
The second method of Hertz diffusion through a jet of mercury vapor described below was also applied to the concentration of \(O^{18}\). Barvikh \(^{8}\) describes work with an apparatus consisting of 48 mercury diffusion pumps. Fractionation in it of water vapor at a pres-
inlet of 2.5 mm in a receiver of volume 1200 cm³ should have given an enrichment to the ratio \(O^{18}:O^{16}=1:7.4\), which followed from a preliminary study of the operation of the apparatus on mixtures of helium and neon. In fact, an enrichment only to \(1:10\) was obtained, which the author explains by the presence, on the glass walls of the apparatus, of a film of water not participating in the fractionation. Scher and Bleakney\(^{5,6}\), by this same method, in an apparatus of 10 pumps, obtained an enrichment to \(O^{18}:O^{16}=1:330\). The receiver had a volume of 200 cm³. The stationary state was reached already after an hour and a half. Since the separation coefficient increases exponentially with the number of pumps, the results of Barvikh and of the latter authors agree: \(\alpha=1.025\) and 1.032 for one pump.
Exchange reactions in solution. The obtaining of water sufficiently enriched with heavy oxygen opened possibilities for studying exchange reactions between oxygen isotopes in solution. The few works carried out up to now have already yielded interesting results and show how important the further study of these reactions is. From elementary theoretical considerations\(^{10,1,B,21}\) it follows that, in all cases of exchange of oxygen isotopes between water and a dissolved substance, the deviation from the probable distribution (the same \(O^{18}/O^{16}\) ratio in the water and in the dissolved substance after equilibrium has been reached) must be very small and can be detected only by extremely precise measurements. Indeed, in none of the cases so far investigated have such deviations been reliably found. On the other hand, the kinetics of oxygen exchange reveals a number of remarkable features deserving detailed study.
Blumenthal and Herbert\(^{B,14}\) studied exchange in potassium phosphate \((K_3PO_4)\). Water with an excess density of \(330\gamma\) due to heavy oxygen was obtained by the Hertz—Barvikh diffusion method. Because of its small quantity, it had to be diluted to \(160—280\gamma\), and a micromethod had to be developed which made it possible to study the exchange quantitatively in saturated solutions of 10—30 mg of salt. In 3 hours at room temperature complete exchange of all four oxygen atoms in the phosphoric-acid ion occurred. Basing themselves on X-ray data on the structure of this ion (the phosphorus ion is surrounded by four closely packed oxygen ions), the authors propose the following mechanism of exchange. When a water molecule approaches the \(PO_4\)-ion, the latter captures a proton
\[ PO_4^{3-}+H_2O=(PO_3—OH)^{2-}+OH^-. \]
This attachment weakens the screening action of the oxygen ions, and the phosphorus ion attaches, by means of covalences, one more hydroxyl, forming an unstable intermediate compound,
\[ (HO^{18})^-+(PO_3—OH)^{2-}=(HO^{18}—PO_3—OH)^{3-}, \]
which then, splitting off hydroxyl and a proton, is again converted into a phosphoric acid ion with one oxygen replaced by its isotope
\[ (\mathrm{O}^{18}\mathrm{H}-\mathrm{PO}_3-\mathrm{OH})^{3-} = (\mathrm{O}^{18}\mathrm{H}-\mathrm{PO}_3)^{2-}+\mathrm{OH}^- = (\mathrm{PO}^{18}\mathrm{O}_3)^{3-}+\mathrm{H}^+ + \mathrm{OH}^- . \]
Since both isotopes have equal chances of substitution, in the final outcome their probable distribution between the water and the salt is established. The mechanism described must proceed in alkaline solution. Indeed, hydrolysis of the phosphoric-acid salt provides a sufficient concentration of free hydroxyl ions in the solution.
An analogous mechanism was proposed by Datta, Deem, and IngoldD, 4 to explain exchange in sodium sulfate, which proceeds at an appreciable rate (complete exchange in 18 h at 100°) only in alkaline solution
\[ \text{1) }\quad \mathrm{H}_2\mathrm{O}^{18}+\mathrm{OH}^- \rightleftarrows \mathrm{H}_2\mathrm{O}+\mathrm{O}^{18}\mathrm{H}^- \]
\[ \text{2) }\quad \begin{array}{c} \mathrm{O}\quad \mathrm{O}^{2} \\ \diagdown\quad \diagup \\ \mathrm{S} \\ \diagup\quad \diagdown \\ \mathrm{O}\quad \mathrm{O} \end{array} +\mathrm{O}^{18}\mathrm{H}^- \rightleftarrows \begin{array}{c} \mathrm{O}\quad \mathrm{O} \\ \diagdown\quad \diagup \\ \mathrm{S}-\mathrm{O}^{18}\mathrm{H} \\ \diagup\quad \diagdown \\ \mathrm{O}\quad \mathrm{O} \end{array} \rightleftarrows \begin{array}{c} \mathrm{O}\quad \mathrm{O}^{2} \\ \diagdown\quad \diagup \\ \mathrm{S} \\ \diagup\quad \diagdown \\ \mathrm{O}\quad \mathrm{O}^{18} \end{array} +\mathrm{OH}^- . \]
Herbert and LauderH, 11 studied exchange in acetaldehyde, upon dissolution of which in water, according to earlier data, ethylene glycol is formed. In 20 h at 20° complete exchange takes place, which, apparently, should be ascribed to the reversible reaction
\[ \mathrm{CH}_3\cdot\mathrm{C} \begin{array}{c} \mathrm{O}\\[-2pt] \diagup\\[-2pt] \mathrm{H} \end{array} +\mathrm{H}_2\mathrm{O}^{18} \rightleftarrows \mathrm{CH}_2\cdot\mathrm{CH} \begin{array}{c} \mathrm{OH}\\[-2pt] \diagup\\[-2pt] \mathrm{O}^{18}\mathrm{H} \end{array} \]
The rate of exchange is such that the half-period is 2 h. The authors draw attention to the fact that this rate is of the same order as the rate of oxidation of acetaldehyde to acetic acid on a nickel catalyst, in the presence of water and oxygen; for this reaction Wieland likewise assumed the formation of ethylene glycol as an intermediate stage.
In a brief preliminary communication, Cohn and UreyC, 5 described the following results of a study of the kinetics of oxygen-exchange reactions in aqueous solution. The oxygen of hydroxides and carboxylic acids (except chloroacetic acid) does not exchange under ordinary conditions. In the carbonyl group of acetone and acetaldehyde exchange does occur. For acetone the catalysts are hydrogen and hydroxyl ions, and also molecules of salicylic acid (but not its anion). Comparison of the rate of exchange in acetone with the rate of its enolization according to
shows that oxygen exchange, in contrast to hydrogen exchange, does not proceed by enolization.
RobertsR, 2 continued the investigations of Cohn and Urey. He found that at room temperature no exchange occurs between methyl alcohol or nitrobenzene and water, either in acid or in alkaline solutions. Likewise, no exchange was observed over two days in a solution of lead acetate. In the presence of HCl, acetic acid in 40 days at \(25^\circ\) exchanges two oxygen atoms. The same amount is exchanged by benzoic acid at \(100^\circ\) in 4 hours, both in the presence of HCl and without it.
MersM, 9 found exchange of both oxygen atoms in hydrochloric-acid glycine within twenty-four hours in solution with \(\mathrm{pH} \leq 1.9\). Glycine under the same conditions gave no exchange.
The rather considerable material on oxygen exchange, collected in recent months, gives inconclusive results, which are still insufficient for any generalizations; however, it is enough to warrant the expectation that further study of this exchange will yield important results, no less interesting than those obtained for the exchange of hydrogen for deuterium.
III. Neon
Isotopic composition. Neon was the first nonradioactive element in which isotopy was found. Passing a beam of canal rays from neon ions through parallel electric and magnetic fields (parabola method), Thomson discovered in 1912, along with \(\mathrm{Ne}^{20}\), also an admixture of its isotope \(\mathrm{Ne}^{22}\). This observation was confirmed in the very first work of Aston in 1919A, 2 with his mass spectrograph. Soon thereafter Aston reported a third isotope found by him, \(\mathrm{Ne}^{21}\), noting, however, the unreliability of this observation. The existence of \(\mathrm{Ne}^{21}\) was finally proved in 1928 in the work of Hogness and KvalnesH, 3 with the aid of Dempster’s mass spectrograph. The indications of Kalman and LazarevK, 1 of the existence of a new isotope \(\mathrm{Ne}^{23}\) were refuted by BainbridgeB, 1 and BleakneyB, 13; S, 1. The error should probably be ascribed to the ion \(\mathrm{Ne}^{22}\mathrm{H}\), which was taken for \(\mathrm{Ne}^{23}\).
The masses of the neon isotopes were accurately measured by the mass-spectrographic method. From the distance \(\mathrm{Ne}^{20} - \tfrac{1}{2} A^{40}\), equal to 0.01088 atomic-weight units, AstonA, 5 found \(\mathrm{Ne}^{20} = 19.9986 \pm 0.0006\). Later measurements by Bainbridge and JordanB, 4 give for this doublet the larger distance \(0.01130 \pm 0.00020\), whence \(\mathrm{Ne}^{20} = 19.99917 \pm 0.00019\), a value somewhat smaller than that found by them in an earlier work. For the remaining isotopes, measurements of the corresponding doublets gave \(21.00013 \pm 0.00029\) and \(21.99870 \pm 0.00040\). The results of the treatment of Bainbridge’s measurements, made by Livingston and BetheL, 3, are given in Table 1.
New studies give similar results for the proportion of the three isotopes in atmospheric neon. For the ratio \(\mathrm{Ne}^{20}:\mathrm{Ne}^{22}:\mathrm{Ne}^{21}\), Boan, Williams, and Tate\({}^{V,1}\) found \(90.0:9.73:0.27\); Bleakney\({}^{B,13}\) found \(91.75:7.98:0.27\), and Muravkin\({}^{M,2}\) found \(100:8.2:(0.28—0.30)\). Preference should probably be given to the first figures, which are included in the table of the Atomic Commission for 1937.\({}^{A,6}\) From this proportion and Bainbridge’s masses, the atomic weight of atmospheric neon on the chemical scale is calculated, very close to the value found from densities (Table 1).
The spectrum of atomic neon confirms the results of the mass-spectrographic study of the isotopes of neon.
Fractionation of neon isotopes by distillation. This method, in combination with fractional adsorption, was applied by Aston\({}^{A,2}\) as early as 1913. The apparatus consisted of four bulbs of charcoal, cooled by liquid air, in which the fractionation was carried out. The end fractions were withdrawn and subjected to further treatment. After 3000 distillations, 7 fractions were obtained, identical in density to the original gas.
Successful results were recently obtained by Keesom, van Dijk, and Gantess\({}^{K,2}\) in a fractionating column 2 m high, equivalent to 60 theoretical plates, cooled by liquid hydrogen. From 420 l of gaseous neon, 2 fractions of 5 l each were obtained, with atomic weights 0.092 unit less and 0.391 unit greater than the original. In a subsequent work\({}^{K,3}\), a column of 85 plates was used, each of which held \(1\ \mathrm{cm}^3\) of liquid. The distillation was conducted at the temperature and pressure of the triple point. Equilibrium was established in 4 days. A series of fractions was obtained with atomic weights from 20.043 to 20.785. For the different fractions the vapor pressure was determined, which proved to be a linear function of the atomic weight. The triple point of \(\mathrm{Ne}^{22}\) is \(0.134^\circ\) higher than the triple point of \(\mathrm{Ne}^{20}\), and the heat of vaporization is higher by \(1.18\ \mathrm{kcal/mol}\).
Separation by diffusion. After fractionation of neon isotopes by a series of successive distillations gave no results, Aston applied fractional diffusion through porous clay tubes. A portion of gaseous neon was repeatedly fractionated, as a result of which a difference of 0.13 unit was obtained for the atomic weights of the extreme fractions. The work was interrupted in 1914 with the outbreak of the war.
The method of separating gases by diffusion through clay tubes was brought to a high degree of perfection by Hertz\({}^{H,4;H,5}\). He applied a method resembling the operation of a fractionating column: successive fractionation in a multistage apparatus forming a closed system. This method gave a considerable separation of the isotopes of hydrogen, carbon, nitrogen, oxygen, and neon. We shall examine it in more detail. In the diffusion of a mixture of gases through po-
STABLE ISOTOPES OF LIGHT ELEMENTS
through a porous partition there occurs their partial separation owing to the different rate of diffusion, which is proportional to the square roots of the molecular weights. If, as before, we define the separation coefficient $\alpha$ as the ratio of the isotope proportions in the end fractions, then, as is easily seen, Rayleigh’s relation (2) remains valid with $\alpha$ determined by the ratio
\[ \sqrt{\frac{M_1}{M_2}}, \]
where $M_1$ and $M_2$ $(M_1 < M_2)$ are the molecular weights.
Hertz’s apparatus consists of a series of cells, each of which is formed from two clay tubes inserted into glass sleeves, and mercury vapor-jet pumps (of the Langmuir type), serving
Fig. 4.
to pump the gas from one cell to another. The end cells are connected with a reservoir $V_l$, in which the light fraction is collected, and a receiver $V_s$, in which the heavy fraction is collected. In Fig. 4 is given a diagram of the apparatus of Wooldridge and Smythe$^{W,5}$, not differing in principle from Hertz’s apparatus; one of its cells is shown schematically in the same figure below. There also is given a schematic representation of one of the pumps $P$. Let us consider the operation of one of the middle cells. The gas mixture enters from the right into the porous tube $B$. A part of it, enriched in the light component after diffusion through tube $B$, is pumped back into the preceding cell. The remaining part enters tube $A$, where the gas stream again branches. Part of it diffuses into the sleeve and has the composition of the gas entering $B$ on the right. It is pumped back to the inlet into tube $B$. The heavy fraction enters the next cell to the left, where the entire process is repeated. It is easy to see that the portions of gas being mixed have approximately the same composition and that, during operation of the apparatus, the light fraction is collected in the right-hand vessel $V_l$, and the heavy one in the left-hand $V_s$, if the pumps pump the gas from left to right. The degree of fractionation
tion increases and then reaches the limit corresponding to the stationary operating regime of the apparatus; it is larger, the greater the number of cells.
Hertz found that the enrichment in one cell obeys the relation
\[ q=\frac{f^\alpha}{1-(1-f)^\alpha}, \tag{10} \]
where \(\alpha=\sqrt{M_1/M_2}\) is the separation factor, \(q\) is the change in the ratio of the densities of the two components of the mixture in the heavy and light fractions, and
\[ f=\frac{l_2/l_1}{2+l_2/l_1} \]
(\(l_1\) and \(l_2\) are the lengths of tubes \(B\) and \(A\)).
If the apparatus consists of \(p\) cells, then instead of \(q\) we have \(q^p\). In the case of large \(V_l\) compared with \(V_s\), as is usually the case, and of a small content of the heavy component, the quantity \(q\) is close to the degree of enrichment of the heavy component in \(V_s\) as compared with the initial mixture. More precisely, this relation was established by Waldmann and Smyth.
In Hertz’s original apparatus \(l_1=l_2\), or \(f=\frac{1}{3}\). This case is not the most favorable one. Obviously, the greatest enrichment will be at \(l_1=0\), when \(f=1\); however, as \(f\) increases, the time required to reach the stationary state increases. Therefore it is not at all advantageous to eliminate the \(B\) tubes. This disadvantage is felt the more strongly the more enriched the mixture is, since in this case equilibrium is established more slowly. It is therefore expedient to equip the cells on the \(V_s\) side with longer \(B\) tubes than those adjoining \(V_l\), while keeping the \(A\) tubes of equal length. An apparatus with tubes of unequal length was built by Harmsen,¹² who thereby considerably improved the operation of Hertz’s apparatus. The instrument consisted of 24 cells, which according to (10) should have given in \(V_s\) an enrichment of the isotope \(\mathrm{Ne}^{22}\) by a factor of 8.4.
Diffusion was carried out at a pressure of 10 mm in \(V_l\) and 7.5 mm in \(V_s\). The stationary state was reached after several hours, after which the ratio \(\mathrm{Ne}^{20}:\mathrm{Ne}^{22}\) decreased from \(1:9\) to \(1:2.5\), which corresponds to an enrichment by a factor of 7, close to that predicted. The apparatus consisted of 19 cells with \(l_1=0\) and 5 cells with \(l_2/l_1=4\). In this apparatus Harmsen subjected Hertz’s heavy fraction with \(\mathrm{Ne}^{20}:\mathrm{Ne}^{22}=1:2.5\) to further fractionation and obtained \(2:1\). The heavy fraction was again fractionated twice, after which the ratio \(\mathrm{Ne}^{20}:\mathrm{Ne}^{22}=1:80\) was achieved, while the light fraction gave pure \(\mathrm{Ne}^{20}\) after \(\mathrm{Ne}^{21}\) had been separated from it in the proportion \(\mathrm{Ne}^{20}:\mathrm{Ne}^{21}=1:1\). In the following work by Harmsen, Hertz, and Schütze,³ spectrally pure \(\mathrm{Ne}^{22}\) was obtained. Subsequently an apparatus of 48 cells with more powerful pumps was built, in whi-
...by Hertz\(^{4,5}\), spectrally pure deuterium was separated from hydrogen. These works will go down in history as the first case of complete separation of the isotopes of nonradioactive elements.
A tenfold enrichment in the isotope Ne\(^{22}\) was achieved by Wooldridge and Smythe in an apparatus of 14 cells. In this way they studied the efficiency of the apparatus, which was then used for fractionating the isotopes of nitrogen and carbon (see below).
An important improvement is the second type of Hertz apparatus,\(^{6}\) in which diffusion through clay tubes is replaced by diffusion through a jet of mercury vapor in the pump itself, which transfers the gas mixture from one cell to another. The principle of operation is explained in Fig. 5. On the left is shown the nozzle of a vapor-jet mercury pump of the usual type. The jet of mercury vapor rises upward through the inner tube and, condensing at the top of the outer tube, entrains the gas entering the wide tube from below. At the same time part of the gas diffuses in a direction perpendicular to the jet, entering inside the vapor jet. In this process partial fractionation of the components of the mixture occurs, owing to the different rates of their diffusion. By narrowing the outer tube above the nozzle (Fig. 5, middle), the separation can be considerably increased. In such an arrangement, only that part of the gas mixture which has diffused through the vapor jet is pumped away into \(B\). The remaining part stays in \(A\). The final form of the nozzle, with dimensions, is shown in the right-hand part of Fig. 5. For better action the gas being separated is brought to the mercury jet from below through a gap formed by the inner and an additional intermediate tube. With this arrangement the jet of the outgoing gas is in close contact with the jet of mercury vapor.
Fig. 5.
Working at a pressure of 1–2 mm, Hertz obtained with an assembly of 6 pumps a threefold enrichment in Ne\(^{22}\). An apparatus of 12 pumps gave, after 45 min., a tenfold enrichment. During this time a concentration of Ne\(^{22}\) of 50% in the heavy fraction was attained. The degree of enrichment in the new apparatus is approximately the same as in the old one with an equal number of cells, but the stationary state is established more rapidly and the apparatus is considerably simpler in construction and in operation.
Barwich\(^{5,5}\) studied in detail the operating regime of the apparatus and developed its theory.\(^{8,8}\) His final apparatus consists of 48 cells, the manner of connecting them to one another being clear from Fig. 6. The terminal reservoir \(V_L\) had a volume of 10 l. The apparatus was filled with a mixture of equal volumes of neon and helium at a pressure of 2.2 mm. After 2.5 hours a stationary state was established, in which in \(V_s\) only the lines of Ne\(^{22}\) were observed in the spectrum, and in \(V_L\) only the lines of Ne\(^{20}\). Successful fractionation of the oxygen isotopes in this apparatus was reported above.
Sherr$^{5,7}$ constructed an apparatus of 29 cells, in which 100 cm$^3$ of neon gave, after 10 hours (stationary state), 1.5 cm$^3$ of a heavy fraction containing 80% Ne$^{22}$.
The apparatus described was used by Barwich and Shütze$^{3,6}$ for the separation of argon isotopes. After 60 hours the ratio $A^{40}:A^{36}:A^{38}=100:12:0.6$ was attained in the light fraction, instead of the initial $100:0.32:0.06$. The analysis was carried out with a mass spectrograph. Konfermann and Krüger$^{\mathrm{K},5}$ brought, after 300 hours, the ratio $A^{40}:A^{36}$ to $1:1$. They obtained 0.5 l of the light fraction.
Fig. 6.
IV. Carbon
Isotopic composition. Carbon has two stable isotopes, C$^{12}$ and C$^{13}$. The second was discovered in 1929 by King and Birge$^{\mathrm{K},6}$ in the band spectrum of C$_2$. The quantum of the 1.0 band in the Swings band (transition $^3\Pi \to {}^3\Pi$) near 4737.1 Å, belonging to the C$^{12}_2$ molecule, is accompanied by a weaker quantum at 4744.5 Å of the C$^{12}$C$^{13}$ molecule. The same authors$^{\mathrm{K},7;\mathrm{B},11}$ found a splitting of bands caused by isotopy in the far ultraviolet spectrum of CO (4th positive group, transition $^1\Pi \to {}^1\Sigma$) and in the violet bands of cyanogen. Carbon isotopy was also discovered in Raman spectra. Bhagavantam$^{\mathrm{B},10}$ found in the spectrum of benzene, cyclopropane, and ethane, near the frequencies 985, 1175, and 974 cm$^{-1}$, a weak satellite at 992, 1188, and 993 cm$^{-1}$, belonging to the molecules of the listed compounds in which one atom of C$^{12}$ is replaced by its isotope C$^{13}$. This observation was confirmed in a new work by Cheng, Hsu, and Ta-Yu Wu$^{\mathrm{C},4}$, who found the satellite of frequency 990 cm$^{-1}$, belonging to a molecule with one C$^{12}$ replaced by its isotope C$^{13}$, in the Raman spectra of benzene, chlorobenzene, bromobenzene, toluene, and cyclohexane. The separation of the satellite is 7–11 cm$^{-1}$, and its intensity is 6:100; calculation confirms the correctness of the proposed explanation of the nature of the satellite.
Initial rough determinations of the ratio \(C^{12}:C^{13}\), made by King and Birge, gave an erroneous value of 400. More accurate spectral determinations by Jenkins and Ornstein\(^{J,1}\) gave a value of the ratio equal to 106, in agreement with subsequent spectral and mass-spectrographic determinations. In a recent work by Broza and Harkins\(^{B,24}\), comparison of the intensities of the aforementioned 1.0-band of both isotopic \(C_2\) molecules is recommended as a fairly accurate and simple analytical method. These authors found that neither CO nor \(CO_2\) is suitable for accurate measurements because of the superposition of the 4835 Å band. Methane is quite suitable; it must first be purified of hydrogen impurity by diffusion through palladium, if this impurity is present in significant quantity. The background and the superposed zero branch are suppressed by the addition of helium. For good measurements it is necessary that the partial pressure of \(CH_4\) in the discharge tube be within the range from 0.1 to 0.5 mm, and the partial pressure of helium from 12 to 20 mm. These authors found, in normal methane, the ratio \(92.2 \pm 3.7\) for \(C^{12}:C^{13}\).
Mass-spectrographic measurements by Boan, Tate, and Williams\(^{V,1}\) gave \(91.6 \pm 2.2\), in contrast to an earlier work by the same authors\(^{T,1}\), where 100 was found. The number 140, obtained by Aston\(^{A,3}\), is hardly correct, and its inclusion in recent international tables\(^{A,6}\) is unjustified.
The most accurate mass-spectrographic measurements of the mass of the \(C^{12}\) atom are based on measurements of the doublet \(C^{12}H_4^{1}—O^{16}\), for which Aston\(^{A,5}\) found a separation of \(36.01 \cdot 10^{-3}\) atomic-weight units, and Bainbridge and Jordan\(^{B,4}\) \((36.49 \pm 0.08)\cdot 10^{-3}\). Hence the former author obtains \(12.00355 \pm 0.00015\), and the latter: 12.00428 or \(12.00402 \pm 0.00007\). From energy balances of nuclear reactions, Cartan\(^{C,1}\) found 12.0037. Aston’s original number 12.0048, retained in some tables, must be regarded as obsolete.
For \(C^{13}\) Bainbridge and Jordan find \(13.0079 \pm 0.0002\) from the distance of \((4.5 \pm 0.1)\cdot 10^{-3}\) units in the doublet \(C^{12}H^{1}—C^{13}\). Bainbridge’s data formed the basis for the numbers of Livingston and Bethe, given in Table 1.
The above figures give 12.011 for the atomic weight of natural carbon on the chemical scale. This number is in excellent agreement with the new careful direct determinations of Baxter and Hale\(^{B,7}\) and with the latest table of the Commission on Atomic Weights\(^{A,7}\), in which the former number 12.00 has been replaced by the more correct 12.010.
Variations in the proportions of carbon isotopes under natural conditions have not so far been observed.
Fractionation of carbon isotopes. Noticeable successes in the separation of carbon isotopes have so far been obtained only by the methods of diffusion and chemical exchange.
Urey and Jenkins\(^{W,3}\) constructed an apparatus of the type proposed by Hertz, with clay tubes, consisting of 34 cells. Diff-
25 liters of methane at a pressure of 8 mm were subjected to diffusion. After 12 hours a stationary state had been reached. The heavy fraction, with a volume of 200 cm³ at a pressure of 6 mm, contained 6.6% C¹³ instead of the initial 0.95%. The analysis was carried out from the spectrum of a mixture of CH₄ with argon in the ratio 1:20, at a total pressure in the discharge tube of 100 mm. In the photographs appended to the paper, the band head at 4744 Å of the Swan band of the molecule C¹²C¹³ (alongside the head at 4737 Å of the molecule C¹²C¹²), which is considerably more intense in the heavy fraction, is clearly visible. It should be noted that in the enriched sample there also appeared the head of the molecule C¹³C¹³, observed for the first time in emission spectra (Sanford⁵˒¹⁰ had observed it earlier in stellar spectra). Its position agrees with calculation. In the subsequent work of Wooldridge and Smythe⁴; ⁵, still greater enrichment was achieved: the heavy fraction contained about 16% C¹³. The vessels \(V_l\) and \(V_s\) (Fig. 4) had volumes of 22 liters and 0.3 liters. The pressure difference in them was about 4 mm at 10–11 mm in \(V_l\). The stationary state was reached after 24 hours, after which 300 cm³ of the heavy fraction (about 1.2 mg of methane) could be withdrawn daily. Below will be described the results of fractionating nitrogen isotopes in the same apparatus. The authors point out that enrichment with the heavy isotopes of carbon and nitrogen could be doubled by passing the diffusing mixture through discharge tubes in which the process
\[ 2\,\mathrm{C}^{12}\mathrm{C}^{13}=\mathrm{C}_2^{12}+\mathrm{C}_2^{13} \]
or
\[ 2\,\mathrm{N}^{14}\mathrm{N}^{15}=\mathrm{N}_2^{14}+\mathrm{N}_2^{15} \]
takes place, since, for example, between \(\mathrm{C}_2^{12}\) and \(\mathrm{C}_2^{13}\) the difference in molecular weights is twice as large as between \(\mathrm{C}_2^{12}\) and \(\mathrm{C}^{12}\mathrm{C}^{13}\). This method was successfully applied by Hertz in the separation of hydrogen isotopes.
Sherr⁵˒⁷, in an apparatus of 29 pumps (Hertz’s second method, with diffusion through mercury vapor), obtained 1.5 cm³ of 5–5.5% C¹³H₄, which corresponds to a separation factor \(Q=13\). The author hopes to bring the enrichment in his apparatus up to 12% C¹³.
Reactions of isotopic exchange open up prospects for obtaining considerable quantities of carbon compounds substantially enriched in the heavy isotope. Table 3 gives the equilibrium constants and separation coefficients for four such reactions. The values of \(\alpha\) differ appreciably from unity, but nevertheless they are so close to it that significant separation can be achieved only with exchange repeated many times. For this purpose a fractionating column may be used, in which, instead of water and vapor, two phases move countercurrent to one another, between which isotopic exchange occurs—for example, a solution of \(\mathrm{CO}_4^{2-}\) and gaseous \(\mathrm{CO}_2\). We have already encountered an example of an analogous use of a fractionating column in the fractionation of lithium isotopes. Another example, relating to the fractionation of nitrogen isotopes, will be considered in detail below. According to Table 3, the most favorable exchange is CO—CO₂, leading to concentration of the heavy isotope in CO₂. However, carrying out this process in the form of a closed cycle is associated with considerable technical difficulties.
STABLE ISOTOPES OF LIGHT ELEMENTS
TABLE 3
Equilibrium constants and separation coefficients
for isotopic exchange reactions
| Reaction | \(K_{273.1}\) | \(K_{298.1}\) | \(K_{600}\) | \(\alpha_{273.1}\) | \(\alpha_{298.1}\) | \(\alpha_{600}\) |
|---|---|---|---|---|---|---|
| \(\mathrm{CO_2^{16} + CO_2^{18} = 2CO^{16}O^{18}}\) | 3.9990 | 3.9993 | 4.0000 | 0.9997 | 0.9998 | 1.0000 |
| \(\mathrm{C^{13}O + C^{12}O_2 = C^{12}O + C^{13}O_2}\) | 1.098 | 1.086 | 1.029 | 1.098 | 1.086 | 1.029 |
| \(\mathrm{C^{13}O_2 + C^{12}O_3 = C^{12}O_2 + C^{13}O_3}\) | 1.015 | 1.012 | 0.997 | 1.015 | 1.012 | 0.997 |
| \(\mathrm{C^{12} + C^{13}O = C^{13} + C^{12}O}\) | 1.074 | 1.030 | 1.074 | 1.030 | ||
| \(\mathrm{C^{12} + C^{13}O_2 = C^{13} + C^{12}O_2}\) | 0.989 | 1.002 | 0.989 | 1.002 | ||
| \(\mathrm{Cl^{35} + 2HCl^{37} = Cl^{37} + 2HCl^{35}}\) | 1.007 | 1.006 | 1.003 | 1.004 | 1.003 | 1.0015 |
| \(\mathrm{Br^{79} + 2HBr^{81} = Br_2^{81} + 2HBr^{79}}\) | 1.001 | 1.0008 | 0.99994 | 1.0005 | 1.0004 | 0.99997 |
| \(\mathrm{Li^{7}H + Li^{6} = Li^{7}H + Li^{6}}\) | 1.028 | 1.025 | 1.008 | 1.028 | 1.025 | 1.008 |
| \(\mathrm{N_2^{14} + 2N^{15}O = N_2^{15} + 2N^{14}O}\) | 1.033 | 1.030 | 1.015 | 1.016 | 1.015 | 1.007 |
The exchange \( \mathrm{C—CO} \) requires work at high temperatures, when \(\alpha\) approaches unity. There remains exchange between gaseous \(\mathrm{CO_2}\) and a carbonate solution. One might have expected equally favorable results from exchange with a bicarbonate solution, which is simpler to work with,
\[ \mathrm{C^{13}O_2 + HC^{12}O_3^- = C^{12}O_2 + HC^{13}O_3^-}. \]
A calculation of the equilibrium constant of this reaction could not be made. In a short communication, Urey, Aten, and Keston\(^{5,6}\) describe preliminary results obtained by this route. The large column with rotating cones described above was used for separating oxygen isotopes. A 25% potassium bicarbonate solution was fed into it from above. At the bottom of the column, \(\mathrm{CO_2}\) was liberated from this solution by boiling with sulfuric acid. The gas moved from bottom to top, countercurrent to the solution. In doing so it gave up to the fresh bicarbonate part of its heavy isotope. After passing through the column, the gas was released from it to the outside. In the first experiment, which lasted 45 hours, the solution was fed at a rate of \(50\ \mathrm{cm^3/min}\), while in the second, which lasted 12 hours, the rate was twice as great. An enrichment in \(\mathrm{C^{13}}\) was achieved: in the first case by 30%, and in the second by 15%. This corresponds to the transfer of 0.78 and 0.39 \(\mathrm{C^{13}}\) through the column, since its charge amounted to 240 g of carbon in the form of \(\mathrm{KHCO_3 + CO_2}\). From these data it is easy to find the separation coefficient, if one recalls that the transfer is equal to \(vtN_0(\alpha - 1)\), where \(vt\) is the amount of carbon introduced into the column during the time \(t\), and \(N_0\) is the initial content of \(\mathrm{C^{13}}\) in it. With \(N_0 = 0.0106\), for the two experiments one obtains \(\alpha = 1.013\) and 1.014, which is close to the number 1.012 calculated for the reaction with carbonate at \(25^\circ\).
It is important to note that initially fractionation did not succeed. The authors attributed this to the excessively slow course of the reaction of carbonic-acid formation upon dissolution in water. They then added, as a catalyst, the enzyme carbonic anhydrase in the form of
chloroform solution of an extract of red blood corpuscles of the bull. After this the results mentioned above were obtained. Apparently, nevertheless, the reaction rate was insufficiently high, since the enrichment obtained corresponded only to approximately 10 theoretical plates, instead of 200–400 in the fractionation of water in the same column. This work limits the application of chemical methods for the separation of carbon isotopes.
V. Nitrogen
Isotopic composition. Isotopy in nitrogen was discovered by NaudéN,1 in 1929 in the spectrum of NO (the γ-bands of the transition \(^{2}\Sigma \to {}^{2}\Pi\) in the region 2052–2269 Å), where, along with the bands of \(N^{14}O\) with three oxygen isotopes, he also observed band heads attributed by him to the molecule \(N^{15}O^{16}\). The existence of the isotope \(N^{15}\) was finally proved by HerzbergH,7 on the \(^{3}\Pi \to {}^{3}\Pi\) bands of the spectrum of \(N_{2}\). The proportion of the two isotopes \(N^{14}:N^{15}=346\), found by Murphy and UreyM,3 from the band spectra of NO, is in good agreement with the new measurements on the \(N_{2}\) spectrum by Uldridge and SmythW,5, but disagrees with the more reliable mass-spectrographic measurements, which give a number close to 265 (Boan, Williams, and TateV,1 found \(265 \pm 8\); Wahl, Huffman, and HippleW,1—from 262 to 268, and Urey with co-workersU,6 \(263 \pm 1\)). This same number is given in the latest table of the Atomic CommissionA,6, \(N^{14}:N^{15}=99.62:0.38\). It gives the correct value of the atomic weight of nitrogen on the chemical scale (Table 1).
Mass-spectrographic determinations of the masses of both isotopes give, for different authors, numbers in good agreement. The most accurate measurements belong to Bainbridge and JordanB,2, who found a separation of \(0.01074 \pm 0.00020\) unit in the doublet \(N^{14}H^{1}—N^{15}\). It is well confirmed by the energy balance of the nuclear reaction \(N^{14}+D^{2}=N^{15}+H^{1}\), for the energy of which 8.57 MeV is calculated instead of the observed 8.53. Comparing this separation in the doublet with data obtained earlier, Bainbridge and JordanB,3 find for the masses of the atoms the values \(14.0076 \pm 0.0002\) and \(15.0050 \pm 0.0003\). The same numbers were found by MattauchM,1, comparing his new mass-spectrographic measurements with data of other authors. AstonA,4 obtained \(14.0073 \pm 0.0005\) for the light isotope. The value 15.0027 for \(N^{15}\), obtained earlier by BirgeB,12 from the above-mentioned spectral measurements of Herzberg, is too small.
Other stable isotopes, besides \(N^{14}\) and \(N^{15}\), apparently do not exist for nitrogen. No variations of the isotopic composition in natural nitrogen and its compounds have so far been detected. This question has not yet been studied.
Fractionation of the isotopes of nitrogen by physical methods. The different vapor pressure of liquid \(N^{14}H_{3}\) and \(N^{15}H_{3}\) may likewise be used for the fractionation of nitrogen isotopes, just as the different vapor pressure of \(H_{2}O^{18}\) and \(H_{2}O^{16}\) was used for the fractionation of oxygen isotopes.
The first attempt to determine \(\alpha=\dfrac{P_{\mathrm{N}^{14}\mathrm{H}_3}}{P_{\mathrm{N}^{15}\mathrm{H}_3}}\) was made by Wahl, Geffman, and HippleW,1 in 1935. This was the first work on the separation of nitrogen isotopes. A portion of liquid ammonia was evaporated to a residue amounting to \(1/32800\) of the initial quantity. In this residue the ratio \(\mathrm{N}^{14}:\mathrm{N}^{15}\) decreased from 268 to 254. Application of Rayleigh’s formula (2) gives \(\alpha=1.0052\) at the boiling temperature of ammonia under atmospheric pressure. The errors of the mass-spectrographic analysis introduce into this number an error of the order of \(+0.0013\). The authors consider the value they obtained for the separation coefficient \(\alpha\) to be too low because of splashing of the ammonia during boiling. Such a comparatively large value of \(\alpha\) made it possible to expect good results from fractionation of liquid ammonia. One may cite the example of water, where, at \(\alpha=1.003\), substantial enrichment in the isotope \(\mathrm{O}^{18}\) is obtained. The authors used a fractionating column with a vacuum jacket, 3 m high and of internal diameter 25 mm. Ammonia at the bottom of the column evaporated at a rate of 2 l/hour, and at the top it condensed and returned as reflux. After several hours of operation the condensate gave an increase in \(\mathrm{N}^{14}:\mathrm{N}^{15}\) only from 268 to \(274\pm4\). In the same work another unsuccessful attempt is described to detect a change in the isotopic composition of ammonia during its rectification under industrial conditions.
It was later found that the value of \(\alpha\) found by Wahl and co-workers was too large. Urey and AtenU,3 measured it more carefully. They subjected portions of liquid ammonia to isothermal distillation without boiling, at different temperatures, down to a residue of \(1/15\)—\(1/70\). The change in the ratio \(\mathrm{N}^{14}:\mathrm{N}^{15}\) was likewise measured in a mass spectrograph. The average of 6 experiments was: \(\alpha=1.0025\pm0.0003\). No influence of temperature was detected. This value of the separation coefficient is considerably smaller than that found by Wahl, Geffman, and Hipple. Taking into account the cost of liquid ammonia, it does not promise good prospects for this method of fractionation.
Probably more favorable results would be given by fractionation of liquid nitrogen or air. The corresponding patent of KrausK,9 was mentioned above.
A considerable separation of nitrogen isotopes was achieved in the above-mentioned apparatus of Wooldridge and SmytheW,5, constructed on the principle of Hertz’s apparatus (fractional diffusion through clay tubes). There were obtained 300 cm³ of a heavy fraction at a pressure of 6 mm, in which the content of \(\mathrm{N}^{15}\) increased 10-fold (to 3%). This enrichment corresponded to the calculated value. The operating regime was the same as in the fractionation of carbon isotopes in methane. Determination of the \(\mathrm{N}^{15}\) content was carried out spectrally, from the intensity of the band heads of \(\mathrm{N}^{15}\mathrm{O}\) in the region of 3000 Å. The enrichment achieved in this work considerably exceeds all that had been obtained by other authors, but the yield of enriched product is very small: less than 3 mg of nitrogen per day.
Of great fundamental interest is the application of exchange ionic adsorption, \(\mathrm{NH_4^+—Na^+}\), on permutite. It was indicated above that Taylor and Urey\({}^{T,2}\) obtained in this way a noticeable fractionation of the isotopes of lithium. For this purpose they passed a solution of lithium chloride through a 10-meter column packed with permutite. The experiment was repeated with a solution of ammonium chloride. A single passage of this solution through the column gave an increase in the content of \(\mathrm{N}^{15}\) in it by 10%. It is interesting to note that here the heavy isotope is adsorbed preferentially, in contrast to lithium, where the heavy isotope accumulates in the solution. The reason for this difference cannot at present be explained.
Application of isotope exchange. Urey and Greiff’s\({}^{U,1}\) calculation of the equilibrium constant for the reaction
\[ \mathrm{N_2^{14}} + 2\mathrm{N}^{15}\mathrm{O} = \mathrm{N_2^{15}} + 2\mathrm{N}^{14}\mathrm{O} \]
shows that for it \(\alpha = 1.016\) at \(0^\circ\) and 1.007 at \(327^\circ\) (Table 3). This exchange should therefore lead to a noticeable fractionation of the nitrogen isotopes; however, the reaction is practically unsuitable because of the impossibility of quantitatively converting \(\mathrm{N_2}\) into \(\mathrm{NO}\) and back at the low temperatures required for a continuous process. By analogy one might have expected good results from the easily effected exchange
\[ \mathrm{N}^{15}\mathrm{H}_3 + \mathrm{N}^{14}\mathrm{H}_4^+ = \mathrm{N}^{14}\mathrm{H}_3 + \mathrm{N}^{15}\mathrm{H}_4^+, \]
the equilibrium constant of which cannot yet be calculated. The first attempt by Wahl, Geffcken, and Gipple\({}^{W,1}\) to apply this reaction ended in failure. They used a column 7.5 m high with packing. Gaseous ammonia entered it from below, and water from above. After 6 hours of operation the ratio \(\mathrm{N}^{14}:\mathrm{N}^{15}\) in the ammonia leaving at the top increased by an amount barely exceeding the accuracy of the analysis (from 262 to 270). The absence of appreciable separation should be attributed to the unsuccessful arrangement of the experiment. Fresh ammonia entered the column continuously, so that it operated as an apparatus with only a single theoretical plate, and for too short a time. Below is described the work of Urey and co-workers, in which, with proper operation, considerable separation was obtained.
Before proceeding to the final experiments, Urey and Aten\({}^{U,3}\) approximately determined the distribution coefficient for reactions of the type under consideration. In the first experiment ammonia was washed out of an aqueous solution by a stream of air until only a small residue remained. The latter showed only a very small, unreliable enrichment in the heavy isotope. In the second experiment, small portions of a sodium hydroxide solution were added successively to a concentrated aqueous solution of ammonium chloride. After each addition the ammonia formed was displaced by a stream of air. The initial amount of \(\mathrm{NH_4Cl}\) was 0.61 mole, and the residue 0.052 mole.
In the latter, the ratio \(N^{14}:N^{15}\) decreased from 250.2 to 233.8, which by Rayleigh’s formula gives \(\alpha = 1.027\). Significantly better results using this same reaction were published by Ogawa,^\(0,1\) but Urey and Aten believe that this work contains substantial errors, since it implies an improbably large separation coefficient \(\alpha = 1.7\)—1.8. In the same work Urey and Aten briefly mention a considerable concentration of \(N^{15}\) in gaseous hydrogen cyanide as a result of the exchange reaction
\[ \mathrm{HCN}^{15} + \mathrm{CN}^{14-} = \mathrm{HCN}^{14} + \mathrm{CN}^{15-}. \]
In a recent paper by Urey, Huffman, Soday, and Fox,^\(U,6\) which we shall discuss in greater detail, a method is described for obtaining large quantities of ammonium salts considerably enriched in the heavy isotope \(N^{15}\). This method, the idea of which was proposed by Urey as early as 1935, is also of great importance for the separation of isotopes of other elements. Its principle consists in the use of a fractionating column for the exchange reaction. If, for example, the following reaction is to be used for concentrating \(N^{15}\),
\[ \mathrm{N}^{15}\mathrm{H}_{3} + \mathrm{N}^{14}\mathrm{H}_{4}\mathrm{Cl} = \mathrm{N}^{14}\mathrm{H}_{3} + \mathrm{N}^{15}\mathrm{H}_{4}\mathrm{Cl}, \]
then the liquid mixture in the column must be replaced by a solution of ammonium chloride, and its vapor by gaseous ammonia. The two must move in the column toward one another. On leaving the column, the ammonia, which has given part of its \(N^{15}\) to the solution, is discharged outside, while the ammonium chloride, slightly enriched in the isotope \(N^{15}\), is converted into ammonia, which is returned to the column, there coming into contact with fresh portions of the solution. Under such operation of the fractionating column, the theory set forth above is fully applicable to it, with the sole difference that the separation coefficient \(\alpha\) entering into the relations derived above is not the ratio of vapor pressures, but the ratio of the isotope proportions in the solution and in the vapor, calculated from the equilibrium constant.^\(B,19; U,1\)
For the reaction just mentioned, \(\alpha\) has a value of the order of 1.02. A simple calculation shows that \(\alpha\) of about \(1.02^{100}\) is needed, i.e., 100 plates, to enrich \(N^{15}\) by a factor of 8. Thus, significant isotope separation requires the use of powerful fractionating columns corresponding to a large number of theoretical plates.
In the work of Urey, Huffman, Soday, and Fox, the same column with rotating cones was used which had served Pegram and Urey for fractionating the isotopes of oxygen, and Urey and Aten for fractionating the isotopes of carbon. A solution of the ammonium salt was fed to the top of the column, and the ammonia obtained from it to the bottom. The column itself was described in detail above, and the arrangement of the auxiliary apparatus is schematically depic-
shown in Fig. 7. The ammonium salt solution is fed to the top of the column by means of piston pump \(A\). After passing through column \(B\), it is pumped by centrifugal pump \(C\) into mixer \(D\), into which an alkali solution is also fed, delivered by piston pump \(E\). The liberated ammonia, after passing through the condenser and the water separator \(F\), returns to the bottom of the column and, after passing through it, is absorbed by water in the small absorption column \(G\), from which it leaves the apparatus in dissolved form. The residues of ammonia not liberated in mixer \(D\) are liberated in boiler \(J\), heated by steam, into which the mixture enters. The boiler is provided with column \(H\), in which ammonia is separated from water vapor. The operation was carried out under reduced pressure, for which vacuum pump \(K\) was used. The ammonia solution and the spent salt solution were removed from the apparatus through valves \(L\) and \(M\). Column \(H\) had a height of \(2.13\ \mathrm{m}\) and a diameter of \(16\ \mathrm{mm}\), and column \(G\) a height of \(90\ \mathrm{cm}\) and a diameter of \(7.5\ \mathrm{cm}\).
Fig. 7.
The isotopic analysis was carried out in a mass spectrograph; for this purpose nitrogen was liberated from a sample of the ammonia solution by the action of sodium hypobromite. The accuracy of measurement of the ratio \(N^{15}:N^{14}\) was of the order of \(1\%\).
It was very important that the ammonia from the ammonium salt solution, which had passed through the column, be returned completely to it. Calculation shows that a loss of \(1/50000\) of the ammonia would reduce the concentration by half at the rate of the process that was used. It turned out that the solution leaving \(J\) contained no more than \(1 \cdot 10^{-6}\) part of ammonia. From relations (8) and (9) it is clear that increasing the charge (in the present case, the amount of nitrogen in the apparatus) proportionally increases the time required to reach the stationary state. Since pump \(C\) cannot operate except with a considerable volume, water was fed into it in order to reduce \(H\). Operation under reduced pressure is advantageous in that it also reduces \(H\) and, at the same time, facilitates the liberation of ammonia from the solution of its salt.
Preliminary experiments were carried out with a small column
of 15 pairs of cones. Equilibrium in it was attained after 3–5 hours. Aqueous solutions of ammonium sulfite (15–35%) and ammonium nitrate (60%), as well as 7% ammonia solution in water and in methyl alcohol, were tested. The value of \(\alpha\) was especially large in 31–34% \((\mathrm{NH}_4)_2\mathrm{SO}_4\). Taking \(\alpha = 1.021\) (see below), \(p = 8\) was obtained, i.e., about two pairs of cones per one theoretical plate. For ammonia solutions, on the basis of previous investigations, \(\alpha = 1.006\) was assumed. This gives \(p = 7\)–8 theoretical plates in water and \(p = 8\)–12 in alcohol.
In the large column, 6 preliminary experiments of 10–16 hours each were carried out. Application of formula (8) made it possible to find the values of \(N\) and \(p\) without waiting for final equilibrium to be reached. The first three experiments with aqueous ammonia and with ammonium nitrate were conducted at a high feed rate of 75–108 \(\mathrm{cm}^3/\mathrm{min}\), and the following ones—with aqueous and alcoholic ammonia solution and with ammonium sulfite at a feed rate of 10–12 \(\mathrm{cm}^3/\mathrm{min}\). Intermediate analyses confirmed formula (8). The expression
\[ N - N_0 - N \ln \frac{N}{N_0} \]
varied linearly with time\(^{1}\). From the slope of the corresponding curve it was possible to calculate the ratio \(\frac{H}{p}\) for a given \(\alpha\). The first term of (8) under the described experimental conditions is close to zero. The filling \(H\) was taken as equal to \(4480\ \mathrm{cm}^3\) of solution, on the basis of an analytical determination of ammonia in the apparatus.
In the first 3 experiments the feed rate was too high. Equilibrium between gas and solution on the cones did not have time to be established, and \(p\) was equal to only 25–29 theoretical plates. In the second series, with a much lower feed rate, \(p = 135\)–207. For the calculation the above-cited values of \(\alpha\) were assigned. The experiments were carried out at a pressure from 15 to 40 mm Hg.
The final experiments were carried out with 31% ammonium sulfite at a pressure of 8 mm and a solution feed rate of \(15\ \mathrm{cm}^3/\mathrm{min}\). To prevent corrosion, 1% ammonium chromate was added to the solution. Both experiments were conducted for 12 days. During this time a stationary state was reached, and enrichment in \(\mathrm{N}^{15}\) up to 2.5 and 2.3% was obtained, i.e., by factors of 6 and 6.5. In all, the two experiments yielded 61 g of \(\mathrm{NH}_4\mathrm{Cl}\) with 2.5% \(\mathrm{N}^{15}\), 244 g with \(> 2\%\), and 1087 g with 0.7–1.5%.
\(^{1}\) In Urey, owing to a somewhat different method of calculation, instead of the expression
\[ N - N_0 - N_0 \ln \frac{N}{N_0} \]
there is given
\[ (1 - N_0)\ln \frac{1 - N_0}{1 - N} + N_0 \ln \frac{N_0}{N}. \]
Both are identical for small \(N\) and \(N_0\).
The value of \(\alpha\) in both experiments proved to be 1.013 and 1.021, which was found from the difference between the amounts of incoming and outgoing \(N^{15}\). The number of theoretical plates was 85 and 91, which, with 621 pairs of cones, is considerably less than the expected value. The small value of \(p\), despite the comparatively slow operation, is attributed by the authors to the low rate of absorption of ammonia by the solution, which prevents full utilization of the column. In part this depends on an oily film on the surface of the liquid, introduced by impurities. It should be recalled that in the fractionation of water the same column gave \(p\) up to 430 plates at a water-feed rate of \(33\ \mathrm{cm^3/min}\).
In a recent paper by Sode, Gorham, and Urey \(^{T,5}\), new successes in the separation of nitrogen isotopes are reported. The same isotope-exchange reaction between a solution of ammonium nitrate and ammonia was used. The authors constructed a unit of three fractionating devices. The latter were designed in such a way that the transfer of the heavy isotope through the cross section of each of the three columns per unit time was the same. For this purpose the capacity of each succeeding member of the unit was set \(\beta\) times smaller than that of the preceding one, where \(\beta\) is the enrichment in one member. The value of \(\beta\) was set equal to 10, on the basis of which the dimensions of the columns were chosen from preliminary experiments with individual sections. In actuality, only \(\beta = 7\) was obtained. The three fractionating columns had the following dimensions: the first \(15.2\ \mathrm{m}\) long and \(76\ \mathrm{mm}\) in diameter, the second \(12.2\ \mathrm{m}\) long and \(25\ \mathrm{mm}\) in diameter, and the third \(7.6\ \mathrm{mm}\) long and \(9\ \mathrm{mm}\) in diameter. The columns were made of glass with continuous packing. The columns were divided into sections \(4.5\ \mathrm{m}\) high, so that the entire apparatus could be placed in an ordinary laboratory room. Pumping of the solution from one section to another was accomplished by squeezing rubber tubes. In the published communication only the last two members of the unit were used. After two weeks of operation they gave an enrichment of \(N^{15}\) by a factor of 46, i.e. to \(14.8\%\). The capacity corresponded to an increment of \(N^{15}\) of \(150\ \mathrm{mg}\) per day in the concentrate. The authors believe that after the first member is also put into operation, it will be possible to obtain about \(1.5\ \mathrm{g}\) of \(60\%\) \(N^{15}\) per day.
In the same apparatus an increase in the content of the sulfur isotope \(S^{34}\) by a factor of 3, i.e. to \(6.8\%\), was obtained by applying the exchange reaction between a solution of sodium bisulfite and sulfur dioxide. In this case only the third member of the unit was used.
One of the samples of \(N^{15}H_3\) concentrate, enriched 1.7-fold in the isotope \(N^{15}\), was used by Schoenheimer, Rittenberg, Fox, Keston, and Ratner \(^{S,2}\) as an indicator for studying the intermediate metabolism of nitrogen in the organism. Glycine and hippuric acid were obtained from this ammonia. Special experiments established the absence of \(N^{14}\)—\(N^{15}\) exchange between hippuric acid and ammonia or glycine, and also between glycine and ammonia or tyrosine. After injection of hippuric acid-
of the acid, enriched with the heavy isotope of nitrogen, in the hippuric acid of the urine there was found \( \frac{2}{3} \) of this enrichment. After injection of glycine, only \( \frac{2}{3} \) of the initial enrichment passed into the urine. These observations indicate that hippuric acid is absorbed in the gastric tract and that glycine may serve as material for its synthesis in the organism.
Table 4 compares the most essential and practically important results of the fractionation of the isotopes of the elements considered in this review. The concentration of the product is expressed as the percentage content of that isotope whose proportion in the starting material is smaller. The ratio of the content in the concentrate to the initial content corresponds to the degree of enrichment. The daily yield of the enriched product is given approximately.
TABLE 4
Achieved separation of isotopes
| Element | Method | Product | Concentration of the enriched isotope in % | Degree of enrichment | Yield per day of product |
|---|---|---|---|---|---|
| Lithium | Exchange of \(\mathrm{Li}^{+}\) in solution with amalgam | \(\mathrm{LiCl}\) | 17.3 | \(2:1\) | \(\sim 10\ \mathrm{g}^{\mathrm{L},1}\) |
| Lithium | Mass spectrograph | \(\mathrm{LiCl}\) | Complete separation | Complete separation | \(2—3\ \mathrm{mg}^{\mathrm{R},1}\) |
| Carbon | Diffusion | \(\mathrm{CH}_{4}\) | 6.6 | \(7:1\) | \(1\ \mathrm{mg}^{\mathrm{W},3}\) |
| Carbon | Exchange \(\mathrm{CO}_{2}—\mathrm{HCO}_{3}^{-}\) | \(\mathrm{KHCO}_{3}\) | 1.3 | \(1.3:1\) | \(25\ \mathrm{g}^{\mathrm{U},5}\) |
| Nitrogen | Diffusion | \(\mathrm{N}_{2}\) | 3 | \(10.5:1^{1)}\) | \(3\ \mathrm{mg}^{\mathrm{W},5}\) |
| Nitrogen | Exchange | 2.5 | \(6.5:1^{2)}\) | \(10\ \mathrm{g}^{\mathrm{U},6}\) | |
| Nitrogen | \(\mathrm{NH}_{3}—\mathrm{NH}_{4}^{+}\) | \(\mathrm{NH}_{4}\mathrm{Cl}\) | 14.8 | \(46:1\) | \(1\ \mathrm{g}^{\mathrm{T},5}\) |
| Oxygen | Fractional distillation of water | \(\mathrm{H}_{2}\mathrm{O}\) | 0.9 | 5.1 | \(200\ \mathrm{g}^{\mathrm{H},9}\) |
| Oxygen | Diffusion | \(\mathrm{H}_{2}\mathrm{O}\) | 9 | \(50:1\) | \(3\ \mathrm{mg}^{\mathrm{B},5}\) |
| Neon | Diffusion | \(\mathrm{Ne}\) | Complete separation | Complete separation | several \(\mathrm{mg}^{\mathrm{H},8}\) |
\(^{1)}\) Calculated according to the authors of work \(\mathrm{N}^{14}:\mathrm{N}^{15}=365\).
\(^{2)}\) Calculated with \(\mathrm{N}^{14}:\mathrm{N}^{15}=265\).
REFERENCES
A, 1. E. Abel, E. Bratu and O. Redlich, Z. physik. Chem., A 173, 353, 1935.
A, 2. F. W. Aston, Isotopes (Isotopes, 1923).
A, 3. F. W. Aston, Proc. Roy. Soc., A 149, 400, 1935.
A, 4. F. W. Aston, Nature, 137, 613, 1936.
A, 5. F. W. Aston, Nature, 137, 337; 138, 1094, 1936; 139, 922, 1937.
A, 6. 2nd Report of the Atomic Commission, Union Intern. de Chim. for 1937.
A, 7. Report of the Atomic Weights Commission for 1938, Union Intern. de Chim.
A, 8. F. W. Aston, Nature, 130, 21, 1932.
B, 1. K. T. Bainbridge, Phys. Rev., 43, 1060, 1933.
B, 2. K. T. Bainbridge and E. B. Jordan, Phys. Rev., 49, 883, 1936.
B, 3. K. T. Bainbridge and E. B. Jordan, Phys. Rev., 50, 98, 1936; 51, 1007, 1937.
B, 4. K. T. Bainbridge and E. B. Jordan, Phys. Rev., 51, 384, 1937.
B, 5. H. Barwich, Z Physik, 100, 166, 1936.
B, 6. H. Barwich and W. Schütze, Naturwiss., 24, 663, 1936; Z. Physik, 105, 395, 1937.
B, 7. G. P. Baxter and A. H. Hale, J. Am. Chem. Soc., 58, 510, 1936; 59, 506, 1937.
B, 8. J. W. Beams and F. B. Haynes, Phys. Rev., 50, 491, 1936; J. W. Beams and A. W. Masket, ibid., 51, 384, 1937.
B, 9. R. P. Bell, J. Chem. Phys., 2, 164, 1934.
B, 10. J. S. Bhagavantam, Proc. Ind. Acad. Sc., A 2, 86, 1935.
B, 11. R. T. Birdge, Nature, 124, 182, 1929.
B, 12. R. T. Birdge, Phys. Rev., 37, 841, 1931.
B, 13. W. Bleakney, Phys. Rev., 43, 1050, 1056, 1933.
B, 14. E. Blumenthal and J. B. Herbert, Trans. Farad. Soc., 33, 849, 1937.
B, 15. A. K. Brewer, Phys. Rev., 47, 571, 1935.
B, 16. A. K. Brewer, Phys. Rev., 49, 635, 889, 1936.
B, 17. A. K. Brewer, J. Chem. Phys., 4, 350, 1936.
B, 18. A. K. Brewer and O. Baudisch, J. Am Chem. Soc., 59, 1578, 1937.
B, 19. A. I. Brodskii, Advances in Chemistry, 6, 152, 1937.
B, 20. A. I. Brodskii and N. P. Skrynnikova, Journal of Physical Chemistry, 6, 1429, 1935.
B, 21. A. I. Brodskii, Journal of Physical Chemistry, 9, 755, 1937.
B, 22. A. I. Brodskii, O. K. Skarre and V. A. Aleksandrovich, Journal of Physical Chemistry, 10, 558, 1937.
B, 23. A. I. Brodskii, O. K. Skarre, E. I. Dontsova and M. M. Glukhaya, Journal of Physical Chemistry, 10, 731, 1937.
B, 24. A. R. Brosi, W. D. Harkins, Phys. Rev., 52, 472, 1937.
C, 1. L. Cartan, C. R., 201, 1363, 1935.
C, 2. G. Champetier and P. Regnault, Bull. Soc. Chim. (5), 4, 592, 1937.
C, 3. J. D. Cockroft and E. T. S. Walton, Proc. Roy. Soc. A 144, 704, 1934.
C, 4. H. C. Cheng, C. F. Hsueh and Ta-Jou-Wu, J. Chem. Phys., 6, 8, 1938.
C, 5. M. Cohn and H. C. Urey, J. Chem. Phys., 6, 176, 1938.
D, 1. M. Dole, J. Chem. Phys., 4, 778, 1936.
D, 2. C. Drucker, Trans. Farad. Soc., 33, 660, 1937.
D, 3. M. Dole, J. Chem. Phys., 4, 268, 1936.
D, 4. S. C. Datta, J. N. E. Day and C. K. Ingold, J. Chem. Soc. London, 1968, 1937.
E, 1. A. Eucken and K. Bratzler, Z. physik. Chem. A 174, 269, 1935.
E, 2. A. Eucken and K. Bratzler, Z. physik. Chem., A 174, 273, 1935.
F, 1. A. and L. Farkas, J. Chem. Phys., 2, 468, 1934.
F, 2. A. Farkas, Trans. Farad. Soc., 33, 552, 1937.
F, 3. R. H. Fowler, Proc. Roy. Soc., A 144, 452, 1937.
G, 1. S. N. Greene and R. J. Voskuyl, J. Am. Chem. Soc., 58, 693, 1936.
H, 1. O. Halpern and Ph. Gross, J. Chem. Phys., 3, 452, 1935.
H, 2. H. Harmsen, Z. Physik, 82, 589, 1933.
H, 3. H. Harmsen, G. Hertz and W. Schütze, Z. Physik, 90, 703, 1934.
H, 4. G. Hertz, Z. Physik, 79, 108, 700, 1932.
H, 5. G. Hertz, Naturwiss., 21, 884, 1933.
H, 6. G. Hertz, Z. Physik, 91, 810, 1934; 19, 35, 1923.
H, 7. G. Herzberg, Z. physik. Chem., B 9, 43, 1930.
H, 8. T. R. Hogness and H. M. Kvalnes, Nature, 122, 441, 1928.
H, 9. J. R. Huffman and H. C. Urey, Ind. Eng. Chem., 29, 531, 1937.
H, 10. J. Horiuti and G. Okamoto, Sc. Pap. Inst. Phys. Chem. Res. Tokyo, 28, 231, 1936.
H, 11. J. B. M. Herbert and J. Lauder, Trans. Farad. Soc., 34, 432, 1938.
J, 1. F. A. Jenkins and L. S. Ornstein, Proc. Amsterd. Acad., 35, 10, 1212, 1932.
J, 2. Th. O. Jones and N. F. Hall, J. Am. Chem. Soc., 59, 259, 1937.
K, 1. H. Kallman and W. Lasareff, Z. Physik, 80, 237, 1933.
K, 2. W. H. Keesom, H. van Dijk and J. Haantjes, Physica, 1, 1109, 1934.
K, 3. W. H. Keesom and J. Haantjes, Physica, 2, 981, 986, 1935.
K, 4. J. Kendall and Crittenden, Proc. Nat. Acad., 9, 75, 1923.
K, 5. J. Kendall, J. Am. Chem. Soc., 55, 2612, 1933.
K, 6. A. King and R. T. Birge, Nature, 124, 127, 1929; Phys. Rev., 34, 376, 1929.
K, 7. A. S. King and R. T. Birdge, Phys. Rev., 35, 133, 1930; Astrophys. J., 72, 19, 1930.
K, 8. H. Kopfermann and H. Krüger, Z. Physik, 105, 389, 1937.
K, 9. A. Krauss, DRP (J. G.) 632071.
L, 1. G. N. Lewis and R. T. Macdonald, J. Am. Chem. Soc., 58, 2519, 1936.
L, 2. Lindemann, Proc. Roy. Soc., 99, 1921.
L, 3. M. S. Livingston and H. A. Bethe, Rev. Mod. Phys., 9, 373, 1937.
M, 1. J. Mattauch, Phys. Rev., 50, 617, 1936.
M, 2. H. Murawkin, Ann. d. Phys., 8, 203, 353, 974, 1931.
M, 3. G. M. Murphy and H. C. Urey, Phys. Rev., 41, 141, 1932.
M, 4. G. M. Murphy and P. F. Brandt, J. Chem. Phys., 5, 274, 1937.
M, 5. L. V. Mysovskii, Physics of the Atomic Nucleus, 1936.
M, 6. N. Morita and T. Titani, Bull. Chem. Soc. Japan, 12, 104, 1937.
M, 7. N. Morita and T. Titani, Bull. Chem. Soc. Japan, 11, 695, 1936.
M, 8. D. Mac Gillavry, Trans. Farad. Soc., 33, 433, 1937.
M, 9. W. H. Mears, J. Chem. Phys., 6, 295, 1938.
N, 1. S. M. Naudé, Phys. Rev., 34, 1499, 1929; 36, 333, 1930.
O, 1. E. Ogawa, Nature, 138, 294, 1936; Bull. Chem. Soc. Japan, 11, 425, 428, 1936; Rev. Phys. Chem. Japan, 11, 131, 1937.
O, 2. M. L. Oliphant, Nature, 137, 396, 1937.
O, 3. M. L. Oliphant, Kempton and E. Rutherford, Proc. Roy. Soc., A 149, 406, 1935.
O, 4. M. L. Oliphant, E. S. Shire and B. M. Crowther, Nature, 133, 377, 1934; Proc. Roy. Soc., A 146, 922, 1934.
O, 5. K. Okabe and T. Titani, Bull. Chem. Soc. Japan, 12, 11, 1937.
O, 6. G. Okamoto, J. Horiutia, K. Hirota, Sci. Pap. Inst. Phys. and Chem. Res. Tokyo, 29, 223, 1936.
R, 1. L. H. Rumbaugh, Phys. Rev., 49, 882, 1936; 50, 681, 1936.
R, 2. J. Roberts, J. Chem. Phys., 6, 294, 1938.
S, 1. M. B. Sampson and W. Bleakney, Phys. Rev., 50, 456, 1936.
S, 2. R. Schoenheimer, D. Rittenberg, M. Fox, A. S. Keston and S. Ratner, J. Am. Chem. Soc., 59, 1768, 1937.
S, 3. E. R. Smith and H. Matheson, Bur. Stand. Sci. Rep. 17, 625, 1936.
S, 4. W. R. Smythe, L. H. Rumbaugh and S. S. West, Phys. Rev., 45, 220, 724, 1934.
S, 5. W. R. Smythe and A. Hemmendinger, Phys. Rev., 51, 146, 178, 1937.
S, 6. R. Sherr, and W. Bleakney, Phys. Rev., 59, 882, 1936.
S, 7. R. Sherr, Phys. Rev., 51, 1007, 1937.
S, 8. D. F. Stedman, Canad. J. Res., B 13, 114, 1935.
S, 9. O. Stern, Phys. Rev., 51, 1028, 1937.
S, 10. R. F. Sanford, Publ. Astronom. S. Pacific, 41, 271, 1929.
T, 1. J. T. Tate, P. T. Smith and A. L. Vaughan, Phys. Rev., 43, 1054, 1933.
T, 2. T. I. Taylor and H. C. Urey, J. Chem. Phys., 5, 597, 1937.
T, 3. B. Topley and H. Eyring, J. Chem. Phys., 2, 217, 1934.
T, 4. V. I. Chernyaev, Usp. Fiz. Nauk, 14, 685, 1934.
T, 5. H. G. Thode, J. E. Gorham and H. C. Urey, J. Chem. Phys., 6, 296, 1938.
U, 1. H. C. Urey and L. Greiff, J. Am. Chem. Soc., 57, 321, 1935.
U, 2. H. C. Urey and G. H. Teal, Rev. Mod. Phys., 7, 34, 1935.
U, 3. H. C. Urey and A. H. W. Aten, Jr., Phys. Rev., 50, 575, 1936.
U, 4. H. C. Urey, G. B. Pegram and J. R. Huffman, J. Chem. Phys., 4, 623, 1936; Phys. Rev., 49, 883, 1936.
U, 5. H. C. Urey, A. H. W. Aten and A. S. Keston, J. Chem. Phys., 4, 682, 1936.
U, 6. H. C. Urey, J. R. Huffman, H. G. Thode and M. Fox, J. Am. Chem. Soc., 59, 1407, 1937; J. Chem. Phys., 5, 856, 1937.
V, 1. A. L. Vaughan, J. H. Williams and J. T. Tate, Phys. Rev., 46, 327, 1934.
W, 1. M. H. Wahl, J. F. Huffman and J. A. Hipple, J. Chem. Phys., 3, 435, 1935.
W, 2. H. E. Wirth, Th. G. Thompson and C. L. Utterback, J. Am. Chem. Soc., 57, 400, 1935.
W, 3. D. E. Wooldridge and F. A. Jenkins, Phys. Rev., 49, 404, 1936.
W, 4. D. E. Wooldridge and F. A. Jenkins, Phys. Rev., 49, 704, 1936.
W, 5. D. E. Wooldridge and W. R. Smythe, Phys. Rev., 50, 233, 1936.
W, 6. H. F. Walton and J. H. Wolfenden, Trans. Farad. Soc., 34, 436, 1938.