ARTIFICIAL RADIOACTIVITY PRODUCED BY NEUTRON BOMBARDMENT—II *

E. Amaldi, O. d’Agostino, E. Fermi, B. Pontecorvo,
F. Rasetti and E. Segrè

Introduction

This article describes some further results on artificial radioactivity induced by neutrons, obtained in the physics laboratory of the University of Rome after the publication of our first paper[^1] on the same subject. The principal results of the present work have already been published by some of us in a series of preliminary communications[^2].

The most interesting new results were obtained on the question of the influence of substances containing hydrogen on the activation of certain elements when they are irradiated with neutrons (§ 1). We give an explanation of this phenomenon, assuming that neutrons are slowed down in collisions with hydrogen nuclei (§ 2). Certain anomalously large effective cross sections for the absorption of slow neutrons are discussed in § 3. In many cases the anomalous absorption is connected with the emission of $\gamma$-rays (§ 4). Section 5 describes some experiments made to estimate the energy of neutrons. Experiments on the scattering of slow neutrons are described in § 6. Section 7 describes experiments on obtaining slow neutrons by means of substances that do not contain hydrogen. In § 8 the results obtained above are discussed from the standpoint of theory. Sections 9 and 10 describe chemical methods for the separation of radioactive isotopes, as well as improvements in the technique of measurement. The results of a systematic investigation of various elements are given in § 11. They are collected in a table placed at the end of the article.

§ 1. Influence of hydrogen-containing substances on activation

In our previous work we noted certain irregularities in the intensity of the activation of silver when it was irradiated with neutrons from a source consisting of radon (radium emanation) and beryllium ($\mathrm{Rn}+\mathrm{Be}$). These deviations undoubtedly depended on

* Proc. Roy. Soc. A, 149, 522, 1935, transl. by L. V. Groshev.

some not entirely clear geometrical factors. Further investigation showed that the activation strongly depends on the bodies surrounding the neutron source. In particular, it increased greatly when the neutron source and the substance being activated were surrounded by a large amount of water or paraffin. It immediately became clear that this effect is caused by the presence of hydrogen, since substances containing no hydrogen did not produce a similar effect (see § 7).

To determine what causes the strong activation—neutrons or γ-rays, which were emitted in large quantity by our source—we repeated the experiment using 100 mg of radium in the absence of beryllium. In this case the induced radioactivity was absent. It follows from this that the observed effect is caused by neutrons. This is also confirmed by the fact that we observed the very same proton effect with neutrons from a Po + Be source*; in this case the intensity of the effect was found to be in agreement with the number of neutrons emitted.

Not every substance that is activated under the action of neutrons shows an increase in activity when it is irradiated under water. Among the substances whose activity changes strongly are: Na (15 h.)**; Al (2.3 min.); V (3.75 min.); Ag (22 sec., 2.3 min.); Cu (5 min.); Rh (44 sec., 3.9 min.); I (25 min.). The activity of other elements or of individual decay periods (in the case where several radioactive elements arise) does not change when irradiation is carried out through water. These include: Si (2.3 min.); Al (10 min.); Mg (40 sec.); Zn (5 min.). We observed that in all cases where the active elements are isotopes of the irradiated element (about 20 cases), the activity increases in the presence of water.

In order to make it possible to measure approximately the sensitivity of different activations with respect to the action of water and paraffin, we express it on a conventional scale. A cylinder (about 2 cm in diameter and 5 cm high) of the substance under investigation is irradiated with neutrons from a source located at the center of the body being irradiated. The source and the cylinder, with the aid of a thin metallic support, are placed at some distance from other objects. After a suitable irradiation time, the activity obtained in the cylinder is measured. Then the cylinder is surrounded by a large cylindrical piece of paraffin (diameter 27 cm, height 20 cm) and is again irradiated with neutrons under the same geometrical conditions and for the same interval of time. The ratio of the activities in the presence and in the absence of paraffin is taken as the measure of the sensitivity of the given activation with respect to the hydrogen-containing substance. We shall denote this ratio by α. On this scale, α = 1 means that the given substance does not exhibit an increase of activity when irradiated in paraffin. Of course, this definition of the sensitivity coefficient is purely empirical,

* Po, unlike Rn with its decay products, emits only soft γ-rays, and these in small quantity (Translator’s note).

* The half-lives of the active elements are given in parentheses (Translator’s note*).

since to some extent it depends on the geometrical conditions. The observed sensitivity coefficients differing from unity varied from $\alpha = 1.6$ for U (13 min., 100 min.) to $\alpha = 40$ for V (3.75 min.). It should be noted, however, that the latter value is not the largest, since many substances are not activated at all when irradiated in air, but are activated under water, though even in this case rather weakly.

The increase in activity caused by the presence of paraffin or water is much greater if the source and the irradiated substance are placed at a distance of several centimeters from one another (see § 5).

§ 2. Interpretation with the aid of slow neutrons

The experiments described in the preceding paragraph can be explained on the basis of the hypothesis which asserts that the effect of water or, more precisely, of hydrogen is caused by the scattering and slowing down of primary neutrons when they collide with hydrogen nuclei.

It is easy to show that, when a neutron collides with a proton, the energy of the neutron decreases on the average by a factor of $e$. It follows that after 10 collisions it decreases approximately to $1/20\,000$ of its initial value. Taking the initial energy to be $4 \cdot 10^6$ eV, we obtain for the energy of the neutron after 10 collisions the value 200 eV. Fewer than 20 collisions are required for the energy of the neutrons to decrease to values corresponding to thermal equilibrium.

The phenomena described above can now be explained on the basis of the assumption that slow neutrons are captured by certain nuclei much more readily than fast ones. In this and the following paragraphs, in discussing the results of our experiments, we shall adhere to this hypothesis.

The increase in activity caused by the presence of hydrogen may be attributed both to neutron scattering, which increases the neutron flux through the substance being activated, and to the greater effectiveness of collisions of slow neutrons as compared with fast ones. In order to show that the chief importance belongs to the second factor, we performed the following experiment.

A silver cylinder was irradiated with neutrons from an Rn + Be source containing 350 millicuries of emanation. The distance between the source and the cylinder was 20 cm. If the irradiation was carried out in air, then after irradiation the cylinder showed no activity whatever. However, if the source was surrounded by a cylindrical vessel with water (diameter 14 cm, height 14 cm), without changing the distance from the source to the irradiated body, then the silver cylinder acquired a large activity (about 100 pulses per minute in our counters). This experiment shows that slow neutrons are more effective than fast ones. Indeed, when the source is surrounded by water, the number of neutrons striking the silver must

within 1 sec does not increase (possibly it decreases slightly as a result of absorption), whereas the activity of silver increases greatly. From this one may conclude that the activity, calculated per neutron incident on the substance, increases greatly when the neutrons are slowed down.

§ 3. Absorption of Slow Neutrons

The fact that the effective cross section for the activation of many elements is much larger for slow neutrons than for fast ones makes it necessary to clarify whether slow neutrons will be strongly absorbed in those substances which they activate most strongly. To this end we carried out a systematic investigation of the absorption of slow neutrons in various substances.

The main purpose of this investigation was to find an element with an anomalously large absorption coefficient for slow neutrons, and therefore we usually used the thinnest possible layers of the absorbing substance. The arrangement for measuring absorption is shown in Fig. 1. The neutron source, a tube with Rn + Be, was placed inside a paraffin cylinder \(P\) (diameter 24 cm, height 14 cm) at a distance of approximately 2 cm from its upper surface. A second paraffin cylinder \(P'\) was placed on the first. In its lower part a recess several centimeters in diameter and 2 or 3 cm deep was cut out. Inside this recess was placed the slow-neutron detector—a rhodium plate (sometimes a silver plate was used). After irradiation with neutrons for a definite time, the activity of this plate was measured once in the absence of the absorbing substance and a second time with it; in the latter case the plate was placed between two layers of the absorbing substance, as shown in Fig. 1 (\(A\)—two layers of the absorbing substance). The ratio of the activities obtained in the two cases gives the absorption of neutrons in the given substance. In the experiments under consideration, as well as in many others, we usually used a rhodium plate as the detector for slow neutrons, in view of its very great activity. This made it possible to carry out very accurate measurements with the aid of an ionization chamber. In addition, of the two periods of rhodium—44 sec and 3.9 min—only the first is of practical importance, which makes it easy to repeat the measurements.

Fig. 1.

The measurements showed that the thickness of the absorbing layer \(\delta\), necessary for reducing the number of neutrons by a factor of two*, varies

* More precisely, \(\delta\) denotes the thickness of the absorbing layer necessary to reduce the activity of the detector by a factor of two. The statement given in the text

for different elements over rather wide limits. For some elements this quantity is extremely small; for example, we found for boron \(\delta = 0.004\ \mathrm{g/cm^2}\), for yttrium \(\delta = 0.015\ \mathrm{g/cm^2}\), for cadmium \(\delta = 0.014\ \mathrm{g/cm^2}\). For other elements \(\delta\) is several thousand times larger; for example, several centimeters of lead absorb less than a thin layer of boron of a few milligrams per square centimeter.

Recalculating the absorption coefficients into an effective cross section for activation in the collision of a slow neutron with a nucleus, we found in a number of cases unexpectedly large values, for example, \(\sigma = 3000 \cdot 10^{-24}\ \mathrm{cm^2}\) for B, \(\sigma = 7000 \cdot 10^{-24}\) for Y, \(\delta = 10\,000 \cdot 10^{-24}\) for Cd. This last value is the largest of all those found so far*. It is remarkable that these effective cross sections are many times greater than the geometrical cross sections of the nucleus, whereas for fast neutrons the effective cross sections and the geometrical cross sections are of one and the same order¹. Indeed, we were able to show directly that for boron the absorption of fast neutrons is at least 1000 times smaller than for slow neutrons.

Fig. 2.

In the absorption measurements carried out, the neutron beam is not homogeneous. In fact, the absorption curve does not fall exponentially—the neutron absorption coefficient gradually decreases as the thickness of the absorbing layer increases, as can be seen, for example, from the absorption curve for cadmium shown in Fig. 2. It should be noted that the thickness of the layer necessary to absorb half the neutrons depends somewhat on the position of the paraffin, since this changes the mean velocity of the neutrons. For example, the absorption of neutrons inside a cavity cut in the paraffin is greater than on the outer side of the paraffin.

¹ The quantity \(\delta\) will coincide with the value of \(\delta\) given in the note only in the case when the absorbing layer does not change the velocity spectrum of the neutrons passing through it. For the sake of brevity, we shall henceforth conventionally call \(\delta\) the “half-absorbing layer” (Translator’s note).

* Still larger values have recently been found for \(\sigma\). For gadolinium \(\delta = 30000 \cdot 10^{-24}\ \mathrm{cm^2}\). See ¹⁷ (Translator’s note).

§ 4. Emission of γ-rays upon capture of a neutron by a nucleus

In connection with the anomalously large absorption of slow neutrons it is necessary to investigate in greater detail the process of absorption itself. The simplest assumption is that a neutron is captured by a nucleus with the formation of an isotope whose mass is greater by one unit than that of the original element. If this heavier isotope is unstable, then strong induced radioactivity may be expected for elements with large absorption. This occurs, for example, for indium and iridium, in which, as is known, radioactive isotopes are formed. A rough estimate of the activity of the activated substance and of the number of absorbed neutrons shows that approximately one active atom is formed for each act of neutron absorption. In other cases it was established that the anomalously strong absorption is not connected with the appearance of induced activity, at least not of considerable magnitude (B, Y, Cd). In these cases we might expect that neutron capture is accompanied by the formation of stable nuclei. Obviously, this can occur more readily for elements possessing many stable isotopes that differ in atomic weight by one unit (Cd, Hg). Assuming simple capture of the neutron, we might expect that in both cases the absorption process would be accompanied by the emission of γ-rays whose energy corresponds to the binding energy of the neutron in the nucleus. There are indications from Li^4 of the existence of such γ-radiation in the case of fast neutrons.

Fig. 3.

In the case of slow neutrons we succeeded in establishing, for certain elements, the presence of such rather strong γ-radiation by means of the following experiment.

A Po + Be neutron source \(S\) of 60 millicuries (Fig. 3) was placed in paraffin together with our standard Geiger–Müller counter \(C\). A lead screen \(L\), 10 cm thick, served to protect the counter from the γ-rays of the source. The counter was surrounded by a layer of lead 2 mm thick. Under these conditions the counter gave about 30 pulses per minute. A small cylindrical layer of various substances was placed around the counter, on the outside of the lead layer. We usually observed a noticeable increase in the number of pulses.

pulses for the case of strongly absorbing substances. This occurred, for example, for Co, Cd, Y, Cl, Ir, Au, Hg, where the number of pulses sometimes increased by more than a factor of two. Exceptions in this respect are boron and lithium, which give no γ-radiation, despite the strong absorption of slow neutrons. It was established that, for these elements, the absorption of slow neutrons is associated with the emission of heavy particles (see § 11). Such emission of heavy particles is theoretically possible only for very light elements, for which the potential barrier of the nucleus is sufficiently low.

The following experiments show that the γ-radiation described above is caused by slow neutrons. γ-radiation is absent if the paraffin is removed. In addition, the effect is greatly reduced when the investigated substance and the counter are surrounded by a boron screen.

§ 5. Energy of Slow Neutrons

It would be very important somehow to estimate the mean energy of the slow neutrons that produce activity in a substance. In this section we describe some attempts made by us in this direction.

We have already noted that the mean energy of neutrons decreases by a factor of \(e\) at each elastic collision with a proton, provided that the energy is large compared with the energy of thermal motion. If the energy of the neutrons had indeed been reduced to thermal energy, then one might expect the diffusion process to depend on the temperature. We tried to detect this effect by means of the following experiment.

A detector made of rhodium or silver was irradiated by a Rn—Be neutron source under the same geometrical conditions, once in paraffin at a temperature of \(200^\circ\)C, and another time at \(20^\circ\)C in a mixture of benzene and pentane, which at this temperature had the same density and the same composition as paraffin at \(200^\circ\)C. In two series carried out, no difference in the activation in the two cases was found within 2%. The mixture filled a cylindrical vessel 26 cm in diameter and 15 cm high. The detector was placed on the axis of the cylinder, at a distance of 1 cm below the surface. The source was also located on the axis, 2 cm lower. The experiment showed that temperature does not affect the activity, at least under these conditions. This could be interpreted as meaning that the energy of the activating neutrons is greater than the energy of thermal motion. However, this is not entirely convincing, since for slow neutrons the dependence on velocity is unknown both for the effective cross section for activation and for the mean free path.

A direct method of measuring, or at least establishing an upper limit for, the energies of slow neutrons would consist in measuring the mean ionization produced at each collision of a slow neutron with a nucleus. This can be done by measuring the total ionization in a chamber filled with hydro-

… or else by direct measurement of the ionization in a separate process by means of a linear amplifier. In both methods the action of slow and fast neutrons can be separated by placing a thin layer of boron in front of the chamber. We attempted to carry out such experiments, but as yet they have not given definite results.

We investigated whether the observed increase in activity produced by paraffin would also occur with other neutron sources. We did this with neutrons emitted by Be when it is irradiated by the γ-rays of radium⁵, and found a large effect with this source. This shows that the slow neutrons obtained in paraffin have lower velocities than the neutrons arising when Be is irradiated by γ-rays.

§ 6. Scattering and Diffusion of Slow Neutrons

In connection with the absorption measurements described in § 3, we also made a rough investigation of the scattering of neutrons by certain elements. The experimental arrangement is shown in Fig. 4. The neutron source \(S\) was placed in the paraffin cylinder \(P\), as described in § 3. The detector—a rhodium plate—was placed on the upper surface of the paraffin. The activity of the detector was measured once in the presence of the substance under investigation, which was placed as a layer above the rhodium plate, and once without it. In some cases an increase in activity was found in the presence of the substance under investigation. This indicates the action of neutrons scattered by the substance under investigation in the backward direction. For example, a layer of carbon several centimeters thick increases the activity of the detector by a factor of 5. Similar results were found for some light elements (Be, C, Si). Heavy elements usually give a much smaller effect. Boron, which has a large absorption coefficient, does not increase the activity of the detector at all. This shows that the anomalously large absorption of this element is due chiefly to true absorption, and not to scattering. We repeated the same experiment with water. The results obtained for the dependence of the intensity of the plate activity on the thickness of the scattering layer of water are shown in Fig. 5. Similar results are obtained if paraffin is used instead of water.

Fig. 4.

Other scattering experiments were made with the aid of the apparatus shown in Fig. 6. At the center of a paraffin cylinder 13 cm in diameter and 11 cm high is placed the neutron source \(S\). The detector—a silver or rhodium cylinder \(R\)—is loca-

is located at a distance of 30 cm from the source. Between the source and the detector a screen of the substance under investigation, \(D\), is placed. Its dimensions are such that all neutrons that could enter the detector from the paraffin cylinder fall upon it. The difference between the activities measured with the screen and without it gives a measure

Fig. 5.

Fig. 6.

of the effect of scattering plus absorption. We found with the aid of this arrangement that a paraffin layer of \(0.5\ \mathrm{g/cm^2}\) reduces the activity of silver to half its value. It was also established that the absorption does not follow an exponential law. Analogous results are obtained if water is used as the scatterer. From these experiments we can infer the order of magnitude of the mean free path of slow neutrons in water and paraffin—it is comparable with the observed thickness of the layer required to reduce the activity of the detector by a factor of two.

Fig. 7.

For carbon this “half-absorbing layer” is about \(5\ \mathrm{g/cm^2}\). Lead of thickness \(4\ \mathrm{g/cm^2}\) gives only a weak decrease in activity, which is due mainly to scattering.

Another series of experiments was carried out in order to establish how the intensity of activation of rhodium in water depends on the distance between the source and the detector. The source was immersed in a water tank measuring \(40 \times 40 \times 100\ \mathrm{cm^3}\), and the induced activity of rhodium was measured with the aid of an ionization chamber.

plate for different distances of it from the source. The results are shown in Fig. 7, where the dependence of the activity of the plate on distance is plotted. Similar experiments were carried out with a 2% solution of boric acid and with water. The curve obtained is similar to the one shown, differing from it only in the magnitude of the intensity, which is greatly reduced owing to the absorption of neutrons by boron.

On the basis of these experiments the following picture of neutron slowing down can be inferred. A large part of the diffusion process takes place when the primary neutrons still possess a large part of their initial energy, after having collided with hydrogen nuclei only a few times. When the velocity of the neutrons decreases to values at which absorption in boron becomes appreciable, the mean free path in hydrogen becomes small, and the diffusion process after this stage of slowing down has been reached is limited to only a small volume. This also explains the fact that the absorption coefficient of boron for neutrons in water is almost independent of the distance from the source.

§ 7. Effect on the Activation of Substances Not Containing Hydrogen

We tried to determine whether the effect of increasing the activity is also produced by substances that do not contain hydrogen. In view of the large amount of material required for these experiments, it was possible to investigate only a small number of substances: Pb, SiO₂, C, Fe. In all cases, with the exception of iron, an increase in the activity of rhodium was established. Under the geometrical conditions described below, the effect increased by a factor of 2 to 5, whereas in water this increase could have been several hundredfold. The observed increase in activity is caused chiefly by a decrease in the velocity of the neutrons, and not by their scattering. This is proved by the fact that the activation of silicon under these conditions does not increase; moreover, a thin layer of cadmium greatly reduces the activation of rhodium. These facts seem strange, especially for lead, since here it would be difficult to explain the slowing down of neutrons by elastic collisions.

The experiments were carried out as follows. From pieces of lead we assembled a cube with sides of 50 cm, at the center of which the neutron source was placed, and at a distance of 10 cm from it—a rhodium detector. The activity observed was approximately three times greater than under the same geometrical conditions in the absence of lead. Similar results are obtained with a silver detector instead of a rhodium one, whereas for a silicon detector no increase in activity at all is observed. This fact indicates that the velocity of the neutrons decreases in collisions with lead nuclei. The same is confirmed by the observation that the introduction of an absorbing cadmium layer of 1 g/cm² reduces the activity of rhodium in lead by a factor of two. Comparing this

absorption of neutrons by cadmium with the absorption of the same element for slow neutrons in water (“half-absorbing layer” equal to 0.014 g/cm²), one may conclude that the average energy of the neutrons in lead is not as small as in water.

Analogous experiments were carried out with a graphite cylinder of volume 3 l. The neutron source was placed at the center of the cylinder, and the rhodium detector at a distance of 5 cm from it. The increase in activity was approximately the same as in the case of lead. The absorption of neutrons by cadmium in this case was no less than in the preceding one. Similar results were obtained with silica, taken in the form of kieselguhr. The absence of such an effect for iron is probably explained by its relatively large absorption.

From these experiments we conclude that most substances possess the property of increasing activation. In a quantitative investigation of radioactivity induced by neutrons, it is always necessary to take into account the possible action of surrounding bodies and, possibly, even of the air. For these reasons the data given above on the influence of certain substances should be regarded only as preliminary indications.

§ 8. Theoretical Discussion of the Properties of Slow Neutrons.

In our first paper we left unresolved the question of what process accompanies the formation of a radioactive isotope of the initial element under neutron bombardment—the capture of the incident neutron or the emission of an additional neutron from the nucleus, i.e. what atomic weight the newly formed radioactive element has—\(A + 1\) or \(A - 1\) (\(A\) being the atomic weight of the initial element). We think that the accumulated data speak in favor of the first hypothesis. The main points of this evidence are the following:

a) Two new weak activities were discovered: one with a decay period of 15 h in sodium, the other with a period of 2.3 min in aluminium. In both cases it proved possible to identify the decay periods with those known previously—in the first case with the period of aluminium and magnesium, in the second with the period of silicon and phosphorus. Chemical data (see § 11) show that the carriers of these activities are respectively an isotope of Na and an isotope of Al. Since both of these elements have only one stable isotope each, namely \(^{23}\mathrm{Na}\) and \(^{27}\mathrm{Al}\), the choice for radiosodium can be only between \(^{24}\mathrm{Na}\) and \(^{22}\mathrm{Na}\), and for radioaluminium between \(^{28}\mathrm{Al}\) and \(^{26}\mathrm{Al}\). However, both light isotopes, \(^{22}\mathrm{Na}\) and \(^{26}\mathrm{Al}\), as is known from other reactions, have different periods and emit positrons instead of electrons. Thus one may with sufficient grounds consider that, at least in these cases, capture of a neutron by the nucleus takes place.

b) Whereas there is no theoretical difficulty

in the understanding of the capture by a nucleus of a neutron with insignificant kinetic energy, it seems implausible that such a neutron can knock another neutron out of a stable nucleus. It would be still more difficult to understand, in the second case, the emission of \(\gamma\)-rays.

In what follows, in discussing the experimental facts, we shall adhere to the point of view that neutrons, in particular slow ones, can readily be captured by many nuclei.

There are, however, certain theoretical difficulties in understanding this capture process, or at least in explaining the large effective cross sections that have been observed experimentally in a number of cases. Nevertheless, it is useful to indicate some general conclusions of the theory that must always be borne in mind in considering the present problem.

Fig. 8.

Let us suppose, as is usually assumed, that the interaction of the neutron with the nucleus takes place only at distances comparable with the radius of the nucleus itself. If this is so, then the de Broglie wave for fast neutrons is comparable with the radius of interaction, and consequently for slow neutrons it is much larger. The well-known theory of collisions, in which the nucleus is regarded as a potential well, assumes in this case a very simple form. Let \(\psi\) be the eigenfunction of the state \(S\), corresponding to zero energy. In Fig. 8 the dependence of \(r\psi\) on the radius vector \(r\) is shown; \(\rho\) denotes the radius of interaction. The curve has an irregular course for \(r<\rho\), whereas for \(r>\rho\) it is a straight line. For the normalization of \(\psi\) we put \(\psi(0)=1\). Then the equation of the straight line expressing the dependence of \(r\psi\) on large \(r\) will have the following form:

\[ r\psi \to \eta(a+r). \]

The geometrical meaning of \(\eta\) and \(a\) is clear from Fig. 8.

The values of these quantities can easily be calculated from collision theory, provided only that the form of the potential well representing the nucleus is known. The effective cross section for elastic collisions, in the limit of small velocities, takes the value:

\[ \sigma_{\mathrm{el}}=4\pi a^{2}, \tag{1} \]

whereas the probability density of finding the neutron at the center of the nucleus is:

\[ P=\frac{n}{\eta^{2}}, \tag{2} \]

where \(n\) is the density of neutrons outside the nucleus.

Whatever the mechanism of capture may be, it is natural to assume, at least in some approximation, that the probability of capture of a neutron by a nucleus per unit time is proportional to \(P\), i.e., that this probability will be given by the expression \(kn/\eta^{2}\), where \(k\) is a constant for each nucleus.

This probability can be expressed through the effective cross section \(\sigma_{\mathrm{ca}}\) for the capture process by the following relation:

\[ \sigma_{\mathrm{ca}}=k/\eta^{2}v, \tag{3} \]

where \(v\) is the velocity of the neutron. The limit of applicability of (3) is set by the circumstance that \(\sigma_{\mathrm{ca}}\) can evidently be at most of the order of magnitude of the square of the de Broglie wavelength. This makes it possible to establish an upper limit for the energy of slow neutrons by considering the large effective cross sections found experimentally (for Cd, \(\sigma_{\mathrm{ca}}=10^{-20}\)). The limit found proves to be of the order of several hundred volts. It must be borne in mind, however, that these conclusions are valid only under the assumptions stated above.

Formula (3), valid only for small velocities, shows that the effective capture cross section for a given nucleus varies inversely as the velocity of the neutron. This explains why the effective capture cross section is usually greater for slow neutrons than for fast ones. The same thing may be expressed by saying that the mean lifetime of slow neutrons in a substance does not depend on their velocity. Whereas the effective capture cross section is inversely proportional to the velocity, the effective cross section for elastic collisions, expressed by formula (1), does not depend on velocity. This means that the mean free path for this type of collision likewise does not depend on velocity.

Another characteristic feature of (3) is the circumstance that \(\sigma_{\mathrm{ca}}\) is inversely proportional to \(\eta^{2}\). The straight line of Fig. 8 may by chance be almost parallel to the axis of abscissae. In that case \(\eta\) is very small, and consequently the effective capture cross section is very large. Such behavior of the eigenfunction is probably also the cause of those anomalously large effective cross sections that have been observed for some nuclei.

In order to derive from (3) the absolute values of the effective cross sections, it would also be necessary to know \(k\), the magnitude of which depends on the physical mechanism of the capture itself. From the experimental data two different processes are known. In some light elements (\(\mathrm{Li}\), \(\mathrm{B}\)) neutron capture is accompanied by the emission of heavy particles, whereas in the case of heavier elements the process apparently consists in the capture of a neutron with the emission of a \(\gamma\)-quantum. When an additional neutron is bound in the nucleus, a certain amount of energy is liberated, whose average value is about \(7 \cdot 10^6\ \mathrm{eV}\). This excess energy could in some cases cause the emission of \(\alpha\)-particles, provided that the potential barrier surrounding the nucleus is sufficiently low for the rapid ejection of the \(\alpha\)-particle. Consequently, this process can be expected only in the case of light elements, whereas in activation by fast neutrons the emission of charged particles is also possible for elements with higher atomic weights, owing to the presence of the additional kinetic energy of the incident neutron, which is added to the binding energy\(^6\).

In the case of capture with emission of a \(\gamma\)-quantum, observed for elements beginning with a certain atomic weight up to the heaviest, the usually accepted mechanism of emission gives, for \(k\), undoubtedly too small values. \(k\) depends on two factors which are difficult to estimate: the matrix elements and the energy of the emitted \(\gamma\)-quantum. Since the probability of emission of a \(\gamma\)-quantum ceteris paribus is proportional to \(\nu^3\), it must be assumed that processes with a large neutron binding energy occur predominantly. This binding energy in some cases may appreciably exceed \(10 \cdot 10^6\ \mathrm{eV}\). Nevertheless, one probably has to adopt too large values for the matrix elements in order to obtain a plausible picture of the relative abundance of elements with anomalously large effective cross sections. From theory one might also expect that usually an anomalous effective cross section for the capture process is connected with an anomalous effective cross section for elastic collision. There are as yet no experimental data on this question.

In this paragraph we shall also briefly discuss the law of the velocity distribution for slow neutrons in hydrogen. From the above theoretical considerations [formulas (1) and (3)] it seems probable to assume that neutrons possessing velocities below a certain limit have one and the same mean free path \(\lambda\) for elastic collisions and a constant lifetime \(\tau\) before capture by the nuclei of the substance. It is easy to show that, for a substance containing hydrogen, the number of neutrons whose velocities lie between \(v\) and \(v + dv\) will be proportional to:

\[ \frac{v \cdot dv}{(v + \lambda/\tau)^3}, \tag{4} \]

for \(v\) less than the above-mentioned limit.

This law of distribution of slow neutrons by velocities can be applied to explain the fact that the absorption curves of slow neutrons are not exponential.

§ 9. Separation of Radioactive Isotopes

Szilard and Chalmers⁷ were the first to succeed in separating radioactive iodine from ordinary iodine by chemical methods. We extended their method to a number of other cases, using inorganic compounds instead of organic ones. The essence of the method is as follows. Suppose that, before irradiation with neutrons, an atom was part of a molecule or radical which, once decomposed, has a practically negligible probability of recombination. When struck by a neutron, the atom is usually knocked out of the molecule and tends to remain in the atomic or ionic state. It follows from this that the atoms knocked out of the molecules after irradiation are in a chemical state different from that of the entire remaining mass, and therefore can be separated by an appropriate reaction. The energy of a chemical bond is only a few volts, and therefore, even if one assumes that the neutron incident on the nucleus has negligible energy, the recoil of the atom upon emission of the γ-quantum appearing as a result of neutron capture by the nucleus would be sufficient to overcome the chemical bond.

Szilard and Chalmers separated radioactive iodine from irradiated ethyl iodide by adding traces of free iodine to it and precipitating I⁻ with AgNO₃. The very same method was applied by us to bromoform, chloroform, carbon tetrachloride, and certain other organic halogen compounds. We were almost always able to separate the radioactive halogen almost completely from the inactive substance.

We also separated radioactive chlorine (35 min.) from sodium chlorate. The chlorine atom is knocked out of the ion ClO₃⁻ by the impact of a neutron. By adding a small amount of Cl⁻ to capture the radioactive isotope and carefully precipitating with AgNO₃ (with the addition of HNO₃ to avoid precipitation of AgClO₃), we found the activity concentrated in the AgCl precipitate. Analogous results were obtained with sodium bromate and iodate. In these compounds the nitric acid was replaced by ammonia. From 70 to 90% of all the activity was concentrated in the precipitate. Cacodylic acid \((\mathrm{CH}_3)_2\mathrm{AsOOH}\) is a good starting material for concentrating radioactive arsenic. The activity can be concentrated by precipitating trisulfide of arsenic from irradiated \((\mathrm{CH}_3)_2\mathrm{AsOOH}\).

Potassium permanganate irradiated with neutrons and then filtered through an ordinary paper filter leaves on the filter a large activity in manganese dioxide formed as a result of oxidation of the paper. By adding manganous salt and precipitating manganese carbonate, it was possible to separate about 80% of all the activity. The ion \(\mathrm{MnO}_4^-\) is disrupted by a neutron; upon

in this case the manganese atom remains in a lower state of oxidation. These atoms are collected with manganese dioxide on the filter or are precipitated with manganese carbonate. We carried out this separation starting from a solid salt and from a solution; however, no substantial difference was found in these cases. Nor was any noticeable influence of the acidity or alkalinity of the irradiated solution observed.

The physical method for separating radioactive isotopes is analogous to the well-known method of obtaining a radioactive deposit from emanation. We applied this method to a gaseous iodine compound. In a glass cylinder of volume about 1 liter, a cylindrical aluminum electrode was placed close against the walls of the cylinder, and another (nickel) electrode was arranged along the axis. The vessel was filled with ethyl or methyl iodide. The temperature of the cylinder was chosen so that the vapor pressure was about atmospheric. The whole apparatus was placed in hot water. A potential difference of 3000 V was established between the electrodes. The neutron source \((\mathrm{Rn} + \mathrm{Be})\) was located outside the vessel. After irradiation, the nickel electrode was removed. It showed the presence of activity with the period of iodine. (The active deposit was collected on nickel because the latter is not activated under the action of neutrons.) The intensity of the deposit was rather small. Reversal of the polarity of the electrodes did not give consistent results.

§ 10. Methods of Measurement

The strong activation caused by slow neutrons opens the possibility of obtaining more intense sources of artificially radioactive elements than last year. This makes it possible to carry out a more accurate measurement of the decay constants of radioactive elements, using an ionization chamber instead of Geiger–Müller counters.

The ionization chamber was made of steel; the internal electrodes were a wire of pure metal and a brass rod. The electrons entered the chamber through a round aperture 6 cm in diameter, made in the lid and covered with aluminum foil 0.01 cm thick. The chamber was filled with \(\mathrm{CO}_2\) at a pressure of 3 atm. It was connected to a Perucca electrometer, whose sensitivity was 0.02 V per division. The capacitance of the entire system was about 20 cm. The zero effect and the sensitivity of the system, checked with the aid of a standard uranium preparation, remained constant. In Figs. 9 and 10 some decay curves obtained with this apparatus are presented.

The very same apparatus was used for measuring the absorption of \(\beta\)-rays. In this case the window was covered with aluminum foil, and the activity was measured for various thicknesses of aluminum. The absorption curves obtained are approximately exponential.

Sometimes substances, besides \(\beta\)-activity, exhibit strong \(\gamma\)-radiation. We conventionally attribute to \(\gamma\)-radiation that residual

ionization obtained in the chamber if it is closed with 2 cm of lead. In calculating the absorption coefficient of the β-rays, the presence of this γ-radiation was taken into account. As a check, the absorption coefficients of the β-rays of RaE and UX₂ were measured. The results obtained were in agreement with the usually accepted values.

Fig. 9.

§ 11. Systematic investigation of the elements

In this section we report all the new data obtained by us for each element, both with respect to induced radioactivity and with respect to the properties of the elements in relation to slow neutrons. Some of the data differ slightly from our previous data because of the increased accuracy of the measurements.

Fig. 10.

1 — Hydrogen. No activity could be detected either in water or in paraffin irradiated in a large vessel of water for several days by a source of Rn + Be in 500 millicuries of emanation.

3 — Lithium. Lithium oxide hydrate gave no activity after irradiation with slow neutrons (14 hours, 400 millicuries). Although lithium remains inactive, it strongly absorbs slow neutrons. The “half-absorbing layer” for lithium is equal to 0.05 g/cm². Absorption of neutrons is not accompanied by the emission of γ-rays. It was shown independently of one another, by us and by Chadwick and Gold-

gaber,^8 that, upon absorbing slow neutrons, lithium emits charged heavy particles. According to Chadwick and Goldhaber, the process can be represented by the following formula*:

\[ {}^{6}_{3}\mathrm{Li}+{}^{1}_{0}\mathrm{n}={}^{4}_{2}\mathrm{He}+{}^{3}_{1}\mathrm{H}. \]

4—Beryllium. Metallic beryllium (99% pure), strongly irradiated with slow neutrons, showed extremely small activity, caused, perhaps, by impurities. Impurities can easily distort the results because of the strong activity that some of them acquire when irradiated under water.

5—Boron. Metallic boron remained inactive after irradiation under water for 14 hours from a 500-millicurie source. It has the largest absorption coefficient so far found for slow neutrons. The “half-absorbing layer” \(\delta=0.004\ \mathrm{g}/\mathrm{cm}^{2}\), which corresponds to an effective cross section of about \(3\cdot10^{-21}\ \mathrm{cm}^{2}\). The absorption process is not accompanied by \(\gamma\)-radiation. In this case, as in the case of lithium, instead of \(\gamma\)-radiation \(\alpha\)-particles are emitted, as was shown by Chadwick and Goldhaber and by us. The emission of \(\alpha\)-particles can readily be detected from the strong discharge in an ionization chamber filled with \(\mathrm{BF}_{3}\), surrounded by paraffin and irradiated with neutrons from a \(\mathrm{Po}+\mathrm{Be}\) source. Shielding the ionization chamber with a thin layer of cadmium (for the absorption of slow neutrons) considerably reduces the ionization current. The same effect is observed in a chamber with air, on the bottom of which there is a small quantity of boron. The emission of \(\alpha\)-particles was also established with the aid of a small ionization chamber connected to a linear amplifier, both for the case of boron deposited on its walls and for the case of filling it with \(\mathrm{BF}_{3}\).

To explain the process we propose the following reaction:

\[ {}^{10}_{5}\mathrm{B}+{}^{1}_{0}\mathrm{n}={}^{7}_{3}\mathrm{Li}+{}^{4}_{2}\mathrm{He}. \]

Chadwick and Goldhaber proposed another reaction:

\[ {}^{10}_{5}\mathrm{B}+{}^{1}_{0}\mathrm{n}=2\,{}^{4}_{2}\mathrm{He}+{}^{3}_{1}\mathrm{H}. \]

At present there are still insufficient data for choosing between these two possibilities. We are now trying to carry out a more precise measurement of the number of ions formed in each process in an ionization chamber containing boron either in gaseous form (total effect) or in a state deposited on the walls (effect of one or two particles). We are also attempting to observe the disintegration of the nucleus in a Wilson chamber containing a gaseous boron compound.**

* Experiments by Kurchatov and Latyshev with a Wilson chamber showed that this reaction is indeed accompanied by the ejection of two heavy charged particles. (Translator’s note.)

* (Note added in proof. Taylor and Goldhaber, Nature 135*, 341, 1935, showed that the reaction occurs according to the first scheme.)

Kurchatov and Latyshev observed the splitting process in a Wilson chamber with a gaseous boron compound. They established that the reaction proceeds with the ejection of two heavy particles. (Translator’s note.)

6—Carbon. No activity is present (see hydrogen). For neutron scattering see § 6.

7—Nitrogen. Ammonium nitrate, irradiated under water for 12 hours from a 600 millicurie source, gives no activity.

8—Oxygen. No activity is present (see hydrogen).

9—Fluorine. Both activities of this element (periods 9 and 40 sec.) do not change in the presence of a substance containing hydrogen.

11—Sodium. This element has two activities, one of which (period 40 sec.) is insensitive to substances containing hydrogen.* A very weak activity with a long period was discovered by Bjerge and Westcott^9. Since this activity is greatly increased in the presence of water, we were able to determine its decay period fairly accurately. It proved to be 15 hours. In view of the theoretical importance of this activity (see § 8), we carefully compared its decay curve with that for aluminum (long period), in order to establish their identity. For the chemical investigation of the radioactive substance we irradiated pure sodium carbonate (Kahlbaum) with neutrons. We dissolved the irradiated substance in hydrochloric acid and added magnesium chloride and aluminum chloride to the solution. The oxide hydrates of the latter metals, precipitated by adding ammonia, proved to be inactive. Then we added a little sodium fluoride to the solution and precipitated the fluorine in the form of barium fluoride. This precipitate also proved inactive. The solution containing the original sodium was then evaporated, and the residue was carefully ignited in order to eliminate neon, in which an active isotope might have formed. Activity was found in the dried sodium chloride. Hence we conclude that the active element is the sodium isotope \(^{24}\mathrm{Na}\). This same isotope was obtained by us last year when aluminum and magnesium were irradiated with neutrons. \(^{24}\mathrm{Na}\) has also recently been obtained in noticeable quantities and studied in considerable detail by Lawrence^10. In this case \(^{24}\mathrm{Na}\) was produced by bombarding certain elements with artificially accelerated particles.

12—Magnesium. Pure magnesium oxide (Kahlbaum), specially examined by us for the absence of aluminum in it, was irradiated under water. The substance was placed at some distance from the source, so that activity insensitive to water would not arise. A new, very weak activity with a period of about 10 min was found. Since this period coincides with the 10-minute period of aluminum, which belongs to \(^{27}\mathrm{Mg}\) (see aluminum), it is quite probable that it too belongs to the same isotope, formed by the capture of a neutron by the \(^{26}\mathrm{Mg}\) nucleus, present in an amount of 11% in ordinary magnesium.

13—Aluminum. Aluminum irradiated under water shows a rather strong new activity with a period of 2.3 min

* That is, \(\alpha = 1\). (Translator’s note.)

(measured with an ionization chamber). When irradiated outside water this activity is extremely small. The period of the new activity coincides with the 2.3-minute period of silicon, belonging to \(^{28}\mathrm{Al}\); therefore we assume that the new activity belongs to the same isotope \(^{27}\mathrm{Al}\), formed when the nucleus captures a neutron.

The second period of aluminum, measured with an ionization chamber, proved to be equal to 10 min instead of 12. This activity is insensitive to water. A chemical separation of the carrier of this activity was carried out. The irradiated metallic aluminum was dissolved in a solution of caustic soda; then magnesium chloride was added. The precipitate of magnesium hydroxide possessed the 10-minute activity. We assume that the active isotope is \(^{27}\mathrm{Mg}\), formed by the following reaction:

\[ {}^{27}_{13}\mathrm{Al}+{}^{1}_{0}\mathrm{n}={}^{27}_{12}\mathrm{Mg}+{}^{1}_{1}\mathrm{H}. \]

14 — Silicon. We determined, with an ionization chamber, the short period of this element. It proved to be equal to 2.3 min. This activity does not change in the presence of water.

In addition to this activity, when irradiating silica under water we found a new activity with a period of several hours. This activity is very weak, but very sensitive to water. We suppose that its carrier is probably \(^{31}\mathrm{Si}\), which is also obtained upon irradiation of phosphorus and has a period of 2.4 hours. \(^{31}\mathrm{Si}\) could be formed by the capture of a neutron by a \(^{30}\mathrm{Si}\) nucleus, present in an amount of about 3%.

15 — Phosphorus. The activity of this element with the short period (2.3 min.) does not increase in the presence of water. Curie, Joliot, and Preiswerk \(^{11}\) attribute this period to \(^{28}\mathrm{Al}\). The following chemical experiment speaks in favor of this hypothesis. We irradiated phosphoric acid, neutralized the solution with sodium carbonate, and added aluminum chloride. The activity proved to be concentrated in the aluminum precipitate.

We obtained the decay curve of phosphorus with the long period by using an ionization chamber. The period proved to be equal to 2.4 hours instead of the 3 indicated earlier. We also measured, with the ionization chamber, the “half-absorbing layer”* for the corresponding \(\beta\)-rays and found it to be equal to \(0.15\ \mathrm{g/cm^2}\) Al.

16 — Sulfur. With the aid of an ionization chamber we determined the period of phosphorus extracted from irradiated sulfur. The period proved to be equal to 14 days. The half-absorbing layer for \(\beta\)-rays is equal to \(0.10\ \mathrm{g/cm^2}\) Al.

17 — Chlorine. Chlorine irradiated under water reveals a new period of 35 min., determined electrometrically. For the chemical identification of the carrier of this activity see § 9. Chlorine strongly

* By the “half-absorbing layer” for \(\beta\)-rays we mean the thickness of the absorbing layer (placed directly at the chamber window), upon passing through which the electrons produce in the ionization chamber a current two times smaller. This quantity characterizes the absorption of electrons in the given substance. (Translator’s note.)

absorbs slow neutrons. The “half-absorbing layer” is 0.3 g/cm². The absorption process is accompanied by the emission of γ-rays.

19—Potassium. We found in irradiated potassium an induced radioactivity strongly dependent on the presence of water and decaying with a period of 16 hours. A chemical investigation of the carrier of this activity, carried out by the same method as in the case of sodium, excluded the elements Cl, A, Ca. From this we conclude that the carrier of the activity is probably an isotope of potassium. According to Hevesy¹² this isotope should be identified with \(^{42}\mathrm{K}\), which he obtained by irradiating scandium with neutrons and which has the same decay period.

20—Calcium. No activity was detected upon 14-hour irradiation in water of calcium fluoride from a source of 600 millicuries.

23—Vanadium. The decay of the induced activity of vanadium was measured with an ionization chamber. The results were as follows: half-life 3.75 min. The “half-absorbing layer” for β-rays is 0.17 g/cm² Al. The β-rays are accompanied by γ-radiation. The activity of vanadium is very sensitive to substances containing hydrogen; according to the determination of § 1, \(\alpha = 40\).

24—Chromium. The activity of chromium is insensitive to water.

25—Manganese. The activity with the short period is insensitive to water (\(\alpha = 1\)). The activity with the long period (2.5 hours; determined with the aid of an ionization chamber), however, is greatly increased in the presence of water (\(\alpha = 23\)). The “half-absorbing layer” for β-rays, measured electrometrically, is 0.14 g/cm² Al. The decay is accompanied by γ-radiation. It is known that the element with a period of 2.5 hours is an isotope of manganese. In § 9 a method was described for concentrating this radioactive element. To obtain new proof that the active product is an isotope of manganese, we first concentrated the activity obtained in irradiated manganese (\(\mathrm{KMnO_4}\)) by precipitating manganese carbonate. The radioactive manganese carbonate was then dissolved in hydrochloric acid, and a large quantity of salts of chromium, vanadium, and iron was added to the solution. Then the manganese was again separated in the form of dioxide by means of nitric acid and sodium chlorate. The manganese precipitate possessed activity, whereas the fractions containing chromium, vanadium, and iron were found to be inactive.

26—Iron. The activity of this element (period 2.5 hours) is insensitive to water. The “half-absorbing layer” for the absorption of slow neutrons is 8 g/cm².

27—Cobalt. This element absorbs slow neutrons rather strongly. The “half-absorbing layer” is 0.7 g/cm². The absorption is accompanied by the emission of γ-radiation.

29—Nickel. Strongly irradiated nickel reveals only doubtful traces of activity.

29—Copper. Both induced activities of this element (periods 5 min and 10 hours) are greatly increased in the presence of water. For

of the first activity \(\alpha = 15\). Copper absorbs slow neutrons with a “half-absorbing layer” of about \(3\ \mathrm{g/cm^2}\). This absorption is accompanied by weak \(\gamma\)-radiation.

The irradiated metallic copper was dissolved in hydrochloric acid, and small quantities of salts of cobalt, nickel, and zinc were added to it. Copper sulfide was precipitated from the acidic solution. It proved to be active. The precipitates of zinc, cobalt, and nickel sulfides, obtained by neutralizing the solution and adding ammonium sulfide, proved to be inactive. Since the time required for these experiments was rather long, the results apply only to the long period. The carrier of this activity may be considered to be an isotope of copper, as was assumed by Bjerrum and Westcott (see above).

30—Zinc. The zinc activity with the short period does not increase in the presence of water. The long period of another activity was measured electrometrically and was found to be equal to 10 hours. To establish the carriers of this activity the following experiment was carried out. The irradiated metallic zinc was dissolved in hydrochloric acid, and small quantities of salts of copper, nickel, and cobalt were added to it. Copper was precipitated partly by reduction on insoluble metallic zinc and partly as copper sulfide in acidic solution. The copper obtained was strongly active. The other elements, precipitated by neutralizing the solution and adding ammonium sulfide, proved to be inactive. This confirms the results of Bjerrum and Westcott, namely that the long period of zinc belongs to an isotope of copper and, probably, to the same one that gives the long period of copper itself. There is some discrepancy between the data of these authors and ours only with respect to the magnitude of the period (according to Bjerrum and Westcott the period is equal to 6 hours).

31—Gallium. The activity with a 20-minute period (measured electrometrically) is not very sensitive to water (\(\alpha = 3\)). The “half-absorbing layer” for the corresponding \(\beta\)-rays is \(0.17\ \mathrm{g/cm^2}\) Al. The carrier of this activity is in all probability an isotope of gallium. To verify this assertion we irradiated gallium nitrate and then added traces of copper and zinc to the solution. The copper was separated in the form of a metallic precipitate on zinc powder, and the zinc was separated by adding mercuric rhodanide. Both elements proved to be inactive.

In addition to this 20-minute activity, upon irradiation under water we found still another new activity, accompanied by rather strong \(\gamma\)-radiation. Its decay period is equal to 23 hours (measured electrometrically).

33—Arsenic. The activity of this element depends strongly on the presence of water (\(\alpha = 6\)). We measured electrometrically its decay period (26 hours) and the “half-absorbing layer” for the \(\beta\)-rays (\(0.16\ \mathrm{g/cm^2}\) Al). For the concentration of this activity, see § 9.

34—Selenium. The activity of this element (period 35 min.) is sensitive to water (\(\alpha = 4\)). The irradiated selenous anhydride was dissolved in 30% hydrochloric acid, and arsenic was added to the solution—

ic anhydride. We precipitated metallic selenium by reduction with gaseous sulfurous anhydride. The precipitate proved to be inactive. We precipitated trisulfide arsenic from the solution and found it inactive. These tests also appear to exclude germanium. Hence we conclude that the activity belongs to an isotope of selenium.

35—Bromine. Both activities of this element are sensitive to water. The activity with the short period has \(\alpha = 10\). The periods of the activities were measured electrometrically and proved equal to 19 minutes and 4.2 hours. The “half-absorption layer” for \(\beta\)-rays for both activities is \(0.12\ \mathrm{g/cm^2\ Al}\). Both activities are accompanied by \(\gamma\)-radiation. For the concentration of the activity, see § 9.

38—Strontium. Activity was absent after strong and prolonged irradiation under water.

39—Yttrium. Strongly irradiated yttrium oxide gave only a very weak activity, which was possibly caused by impurities. Yttrium absorbs slow neutrons very strongly. The “half-absorption layer” is \(\delta = 0.015\ \mathrm{g/cm^2}\). This absorption is accompanied by \(\gamma\)-radiation.

40—Zirconium. Strongly irradiated zirconium nitrate gave only a very weak activity, which was possibly caused by impurities.

41—Niobium. The same as zirconium.

45—Rhodium. The activity with the short period is sensitive to water (\(\alpha = 15\)). The period and “half-absorption layer” for \(\beta\)-rays were determined electrometrically (44 sec, \(0.15\ \mathrm{g/cm^2\ Al}\)). We also made a more accurate measurement of the long period with an ionization chamber and found for it the value 3.9 min. The activity is accompanied by weak \(\gamma\)-radiation. Rhodium absorbs slow neutrons rather strongly. The “half-absorption layer” is \(0.3\ \mathrm{g/cm^2}\). The half-absorption probably corresponds to the formation of active isotopes.

46—Palladium. The activities of this element are also sensitive to water. We found two periods: one short, about 15 min, and another about 12 hours. MacLennan, Grimmett, and Read found a period of 14 hours, which agrees with ours within the limits of measurement accuracy.

47—Silver. The two periods were again determined with an ionization chamber. They are equal to 22 sec and 2.3 minutes. Both activities are very sensitive to water, having respectively \(\alpha = 30\) and 15. The strong activation of this element corresponds to considerable absorption for slow neutrons. The “half-absorption layer” is \(\delta = 1.2\ \mathrm{g/cm^2}\).

We added rhodium chloride and palladium nitrate to a solution of irradiated silver nitrate. By adding hydrochloric acid we precipitated silver, which was found to be active. From the filtered solution we precipitated palladium with dimethylglyoxime and, by reduction, rhodium. Both precipitates were inactive. This experiment, relating only to the long period because of the long...

...of the operations, shows that the carrier of this activity is in all probability an isotope of silver.

48 — Cadmium. Cadmium irradiated under various conditions gave several weak activities with different periods (not yet measured). Cadmium absorbs slow neutrons very strongly. The “half-absorbing layer” is \(\delta = 0.013\ \mathrm{g/cm^2}\). The corresponding effective cross section is the largest of all those so far found for slow neutrons (\(\sigma = 10^{-20}\ \mathrm{cm^2}\)). The absorption is accompanied by intense \(\gamma\)-radiation and probably corresponds to the transformation of one stable isotope of cadmium into another stable isotope of the same element.

49 — Indium. The activity induced in indium has three periods. The activity corresponding to the shortest of them (13 sec.) is sensitive to water (\(\chi = 12\)). The second activity (period 54 min., measured electrometrically) is also very sensitive to water. Experiments with a magnetic field show that the emitted electrons have a negative sign. The corresponding “half-absorbing layer” is \(0.045\ \mathrm{g/cm^2}\) Al. An even longer period of several hours is reported by Szilard and Chalmers \(^{13}\). This last activity is either quite insensitive to water or has moderate sensitivity.

Chemical experiments were carried out in order to establish the carriers of the last two activities. Silver was added to a solution with irradiated indium nitrate, and was then precipitated from the solution in the form of silver chloride. The precipitate proved inactive. Tin, antimony, and cadmium were then added to the solution; these were precipitated in the form of tin sulfide, antimony trisulfide, and cadmium sulfide with the aid of hydrogen sulfide. The acidity of the solution was adjusted in such a way that the indium remained in solution while the other metals precipitated. This precipitate also proved inactive. By neutralizing the solution, we precipitated indium sulfide, which proved active. It was established that, together with strong activation, this element exhibits appreciable absorption for slow neutrons. The “half-absorbing layer” is \(\delta = 0.3\ \mathrm{g/cm^2}\).

50 — Tin. Tin strongly irradiated under water shows no activity.

51 — Antimony. We found for this element one activity with a period of 2.5 days. The activity is sensitive to substances containing hydrogen. The “half-absorbing layer” for the emitted \(\beta\)-rays is \(0.09\ \mathrm{g/cm^2}\) Al. The following chemical experiment shows that the carrier of this activity is an isotope of antimony. We dissolved irradiated metallic antimony in aqua regia and added to the solution a certain amount of tin. After separation of tin sulfide (according to Clark), we found activity in the precipitate of antimony trisulfide. The antimony trisulfide was then dissolved again. Indium was added to the solution, and antimony trisulfide was precipitated from a solution of medium acidity. The solution was then neutralized and indium was precipitated from it; it was found inactive. Tellurium and iodine were added to a new solution of antimony.

At first tellurium was precipitated by reduction, then iodine was precipitated in the form of silver iodide. Both precipitates proved to be inactive.

52—Tellurium. Shows weak activity, sensitive to water. Its period is 45 min, instead of the 30 indicated in our first paper.

53—Iodine. The period and the “half-absorbing layer” for β-rays were determined electometrically (25 min, 0.11 g/cm² Al). The activity is of medium sensitivity with respect to water (α = 5). For concentration of the activity see § 9.

56—Barium. A new activity was found, with a period of 80 min, sensitive to water (α = 8). The following chemical experiment confirms the supposition that the carrier of this activity is an isotope of barium. We dissolved the irradiated barium hydroxide hydrate in hydrochloric acid, added a small amount of sodium chloride, and precipitated barium sulfate. Activity was detected in this precipitate. After evaporating the solution, we found the remaining sodium precipitate to be inactive.

57—Lanthanum. After strong irradiation no activity whatever was found.

58—Cerium. The same as lanthanum.

59—Praseodymium. The activity with a short period is insensitive to water. Upon irradiation under water we found a new activity, sensitive to water and decaying with a period of 19 hours. The “half-absorbing layer” for the corresponding β-rays is 0.12 g/cm² Al (both values were obtained electometrically).

64—Gadolinium. We irradiated under water a very pure preparation of gadolinium oxide, kindly provided to us, together with other rare earths, by Prof. Rolla. We found an activity decaying with a period of 8 hours.

73—Tantalum. After 12-hour irradiation only a doubtful activity was detected.

74—Tungsten. Metallic tungsten irradiated under water showed an activity (α = 15) decaying with a period of about one day[^14]. We irradiated tungstic oxide and dissolved it in a solution of caustic soda, then added and separated tantalum pentoxide, which proved to be inactive. To the solution with tungsten a solution containing rhenium was added, and by addition of hydrochloric acid tungstic oxide was precipitated. The precipitate was active, whereas the rhenium, precipitated from the filtrate in the form of rhenium sulfide, proved to be inactive. Since we had no hafnium, we performed the following experiment in order to exclude an isotope of this element as the carrier of the activity. From an irradiated solution of tungstic oxide in ammonia we precipitated zirconium hydroxide. The precipitate gave no activity. We conclude that the activity produced in tungsten probably belongs to an isotope of this element.

75—Rhenium. We irradiated pure metallic rhenium under water. The activity arising in it increases strongly

in the presence of water and decays with a period of 20 hours. The “half-absorption layer” for electrons is \(0.12\ \mathrm{g/cm^2}\) Al. The activity probably belongs to an isotope of rhenium. The irradiated rhenium was dissolved in nitric acid. Tantalum and tungsten were added to the solution and then precipitated in the form of tantalum pentoxide and tungsten trioxide. Both precipitates were inactive, whereas the rhenium retained its activity.

77—Iridium. The induced radioactivity of this element is very sensitive to water. The activity period and the “half-absorption layer” for \(\beta\)-rays were determined with the aid of an ionization chamber (19 hours, \(0.12\ \mathrm{g/cm^2}\) Al). The strong activation of iridium corresponds to a strong absorption of slow neutrons, the “half-absorption layer” for which is \(0.3\ \mathrm{g/cm^2}\). The absorption is accompanied by the emission of \(\gamma\)-rays.

78—Platinum. Very pure metallic platinum irradiated under water (purity standard 4 according to Heraeus) possessed an activity with a period of about 50 min. MacLennan, Grimmett, and Read give a period of 36 min.

79—Gold. The activity of this element is sensitive to water. Its period is 2.7 days (determined electrometrically). Deflection of the \(\beta\)-rays in a magnetic field showed that they consist of negative electrons. The \(\beta\)-rays have very low penetrating power: the “half-absorption layer” for them is \(0.04\ \mathrm{g/cm^2}\) Al. When gold is irradiated with slow neutrons, \(\gamma\)-rays are emitted.

80—Mercury. No activity was detected after strong irradiation. This element strongly absorbs slow neutrons. The “half-absorption layer” is \(0.2\ \mathrm{g/cm^2}\). Upon absorption of neutrons, \(\gamma\)-radiation is emitted.

81—Thallium. No activity was detected after strong irradiation.

82—Lead. The same as thallium.

83—Bismuth. The same as thallium.

90—Thorium. Activities with periods of 1 and 24 min (measured electrometrically) are almost insensitive to water.

92—Uranium. We also investigated the effect of substances containing hydrogen on the induced activity of this element (periods 15, 40 sec.; 13, 100 min.). It was established that the activities corresponding to the 1st, 3rd, and 4th periods are somewhat increased upon irradiation under water, whereas the activity with the period of 40 sec. is not changed thereby. We determined the increase in activity upon irradiation in water for the periods 15 sec., 13 min., and 100 min. For all of them we found \(\alpha \simeq 1.6\). For the activity with a period of 15 sec., the measurement was made with counters and is not very accurate because of the shortness of the period. For the other two activities the coefficient \(\alpha\) was determined with the aid of an ionization chamber; in doing so special attention was paid to establishing that \(\alpha\) has the same value for these two periods. For this purpose we compared three decay curves of activity obtained after 14-hour irradiation of one-

one and the same amount of uranium oxide, once in air, placing the uranium in a special tube around the neutron source; a second time under the same geometrical conditions, but with the tube surrounded by paraffin; and a third time with the neutron source placed at a distance of 5 cm from the tube with uranium oxide and with all of this immersed in a large mass of paraffin. In all these measurements we found that the decay curves were proportional, i.e., showed one and the same ratio of activities. We think that the identity of the sensitivity coefficients for these two activities has been established more accurately than their absolute magnitude and their identity with the sensitivity coefficient of the 15-second activity. It is clear that all active products arising as the result of one and the same primary process must have one and the same sensitivity coefficient. Hence we conclude that the activity with a period of 40 sec is caused by an independent primary process, whereas the three other activities are probably the result of one and the same primary process. This conclusion is limited by the possibility of a chance coincidence of the sensitivity coefficients within the sufficiently wide limits of accuracy of our measurements.

Assuming the identity of $\alpha$ for three periods, these three activities should belong to a family of elements (the initial product would be the activity with the short period), perhaps complicated by the presence of branching. Some evidence in favor of such an assumption, at least with respect to the 13- and 100-minute periods, is provided by the following experiment. We measured electrometrically the decay curves for a thick layer of irradiated uranium. Analysis of these curves showed that the ratio of the initial activities with periods of 13 and 100 min is about $100 : 45$. The “half-absorbing layer” for $\beta$-rays of the 13-minute activity is $0.14\ \mathrm{g/cm^2}$. For the long period this quantity cannot be measured with any accuracy, but it is definitely less than the preceding value and probably by a factor of two. These results agree with the assumption that the number of decays for the activities with periods of 13 and 100 min is the same.

In our first article we cited certain chemical facts which seemed to indicate that the carriers of the 13- and 100-minute activities were not isotopes of the known heaviest elements; rather, the activity belonged to elements situated beyond uranium. From that time our point of view was criticized by Gross and Agruss,$^{15}$ who, although they had never experimented with activated uranium, nevertheless, on the basis of our experiments, came to the opposite conclusion: that the carriers of these activities are isotopes of protactinium. We therefore carried out some new chemical experiments on the behavior of these activities.

The separation of the activity with sulfide compounds of certain metals (silver, copper, lead, mercury) was repeated. The acidity of the solution (hydrochloric acid) was about 20%; sometimes it

varied somewhat to facilitate the precipitation of the sulfide of the element being used. The co-precipitation of the activity with the precipitate was usually good—about 50%—and varied with changes in the precipitation conditions. Nitric acid very greatly reduces the amount of activity precipitated. In a sulfide precipitate a large amount of activity is precipitated in the presence of a hydrofluoric-acid solution of tantalum. We also carried out an experiment to see whether the induced activity would take part in the reaction which, according to Gross and Agruss, is most characteristic for protactinium.

We dissolved previously purified and irradiated uranium oxide in a 25% solution of hydrochloric acid and then added zirconium nitrate and phosphoric acid to the solution. The zirconium phosphate precipitate proved inactive. After separating the zirconium, we precipitated the sulfide from the solution. It contained activity in the usual amount. According to Gross and Agruss this reaction should be regarded as evidence that the carrier of the activity is not an isotope of protactinium.

Having carried out various chemical experiments, Meitner and Hahn\(^ {16}\) also came to the conclusion that the carriers of the 13- and 100-minute activities are in all probability elements lying beyond uranium. We repeated some of their experiments and found the same results.

The two activities, without doubt, behave identically in chemical respect. Some slight evidence of the possibility of separating them was obtained only in the following experiment. Thoroughly purified uranium oxide, irradiated with neutrons, was dissolved in hydrochloric acid. This solution was then poured into a solution of ammonium carbonate until the uranium precipitate was again completely dissolved. By adding lead nitrate or manganese, we collected the precipitate of the corresponding carbonate salt, which possessed the 13- and 100-minute activities. In the filtrate we precipitated copper sulfide and found both activities in it. It seems that the ratio of the activities in the two precipitates was somewhat different—the 13-minute activity was present in greater amount in the sulfide precipitate.

These experiments give further confirmation of our hypothesis that the carriers of the induced radioactivity of uranium, with periods of 13 and 100 min, are elements lying beyond uranium. The simplest interpretation, consistent with the known facts, consists in the assumption that the activities with periods of 15 sec, 13 min, and 100 min are products of one and the same chain, probably with atomic numbers respectively 92, 93, and 94 and atomic weight 239.

12. Tabular Results

The main results on the radioactivity induced by neutron bombardment are collected in the tables. The 1st column contains the atomic numbers and symbols of the elements investigated. Column 2 gives the isotopic composition of the given element. Isotopes indicated in boldface are ...

Table 1.

Continuation of Table 1.

Continuation of Table 1.

Continuation of Table 1.

isotopes constituting not less than 20% of the given element. Column 3 gives the half-life periods found, in the order of their increase. Column 4 gives the “half-absorbing layer” for β-rays in g/cm of aluminum. The mean energy of the β-rays in millions of volts can roughly be obtained by multiplying the data of columns 4 and 8. Column 5 indicates whether the β-emission is accompanied by γ-rays. Column 6 gives the activation sensitivity with respect to substances containing hydrogen. It is given either by indicating the numerical value of the sensitivity coefficient (for its definition see § 1; a sensitivity coefficient equal to unity means that activation is not increased in the presence of substances containing hydrogen), or by the letter a, which means that activation is increased by substances containing hydrogen, but the sensitivity coefficient has not been measured. Column 7 indicates the carrier of the activity. Column 8 gives the half-absorbing layer for slow neutrons.

In some cases, two active products obtained by neutron bombardment of different elements were assigned the same period (although only one of them may have been measured exactly), if chemical experiments established the identity of these products. The same applies to some half-absorbing layers for β-rays.

LITERATURE

1. Fermi, Amaldi, D’Agostino, Rasetti, Segrè, Proc. Roy Soc. A 146, 483, 1934 (see Uspekhi fizich. nauk 14, no. 8, 1934).

2. Fermi, Amaldi, Pontecorvo, Rasetti, Segrè, Ric. Scient. 2, 280, 1934; Fermi, Pontecorvo, Rasetti, Ric. Scient. 2, 380, 1934; Amaldi, D’Agostino, Segrè, Ric. Scient. 2, 381, 1934; Amaldi, D’Agostino, Fermi, Pontecorvo, Rasetti, Segrè, Ric. Scient. 2, 467, 1934; 1, 123, 1935; some of our experiments were repeated by Bjerge and Westcott (Bjerge a. Westcott, Proc. Camb. Phil. Soc., 31, 145, 1935), and analogous results were obtained.

3. Dunning, Phys. Rev. 45, 586, 1934.

4. Lea, Nature 133, 24, 1934. Proc. Roy. Soc. A, 150, 637, 1935.

5. Szilard a. Chalmers, Nature 134, 494, 1934.

6. Meitner, Naturwiss. 45, 789, 1934.

7. Szilard a. Chalmers, Nature 134, 462, 1934.

8. Chadwick a. Goldhaber, Nature 135, 65, 1935.

9. Bjerge a. Westcott, Nature 134, 286, 1934.

10. Lawrence, Phys. Rev. 47, 17, 1935.

11. Curie, Joliot a. Preiswerk, C. R. 198, 2089, 1934.

12. v. Hevesy, Nature 135, 96, 1935.

13. Szilard a. Chalmers, Nature, 135, 493, 1935.

14. Mc Lennan, Grimmett a. Read, Nature, 135, 147, 1935.

15. Grosse a. Agruss, Phys. Rev. 46, 241, 1934.

16. Hahn u. Meitner, Naturwiss. 23, 37, 1935.

17. Dunning, Begram, Sink a. Mitschell, Phys. Rev. 48, 265, 1935.