Spectroscopy of Solutions of Rare-Earth Salts

A. Zaidel and Ya. Larionov, Leningrad

The purpose of this review is to introduce the reader to the state of work on the spectroscopy of rare-earth ions in solutions.

Questions concerning the spectroscopy of solutions have only in the last five years begun to be intensively developed; they are therefore being studied much later and far less than, for example, questions concerning the spectroscopy of metal atoms in the vapor state. This is especially true of the interpretation of the spectroscopic properties of solutions of rare-earth salts. The accumulation of empirical material on the absorption spectra of the rare earths began long ago—the middle of the nineteenth century—and there is a very extensive literature on it. The old literature is represented in this review only to a very small extent, since in most cases the long-known experimental data are insufficiently reliable and are no longer of great interest. On the contrary, the literature of the last five years, in which attempts are being made to resolve the questions that interest us, is given as fully as possible. However, because of lack of space, a number of works have been omitted. Thus, for example, the review does not include several currently available works on combination scattering in rare-earth solutions. We do not touch on this question at all, since its investigation is still in a very rudimentary state. Nor do we touch on the question of the magneto-optical phenomena in rare-earth solutions. The large number of works on crystals of rare-earth salts, although sometimes of considerable interest for understanding the processes in solutions, also could not be reflected here, with the exception of those which seemed absolutely necessary for the integrity of the exposition.1 It seemed expedient to us in the first short section to present some of the properties of the rare earths, knowledge of which is necessary for understanding what follows. The second and third sections are devoted to the absorption and luminescence spectra of rare-earth solutions. The works belonging here are, in most cases, of an unfinished character. This character of theirs has deliberately been emphasized everywhere, so that the circle of problems whose solution is only just being outlined and on which work is now being carried out may be clearer.

§ 1. Properties of the Rare Earths from the Point of View of Atomic Structure

The rare earths were discovered in 1794 by Gadolin, who found, in a mineral now called gadolinite, oxides of new elements, which received the name yttrium earths. At the beginning of the nineteenth century Berzelius discovered the existence of cerium earths, in which in 1839 Mosander discovered didymium. After this, a number of investigators, among whom

it is necessary to mention the names of Lecoq de Boisbaudran, Marignac, Soré, Demarçay, Auer von Welsbach and, finally, Urbain, who for almost a hundred years continued to discover and investigate new elements belonging to the same group of rare earths.

In the history of the discovery of these elements it happened several times that a newly discovered element in fact turned out to be a mixture of several elements, or, conversely, what had previously been considered a mixture later proved to be a simple element. Moreover, one and the same element was often discovered several times and appeared under several different names1.

This is easily explained by the special position occupied by the group of rare earths among the other elements. All rare earths are extraordinarily close in their chemical properties and, as a rule, occur in nature as a mixture of almost all the elements of this group, usually accompanied by a number of other elements as well. The closeness of their chemical properties naturally makes the preparation of the rare earths in pure form difficult, all the more so because until very recently there were no characteristic reactions for the elements of this group; even now such reactions are known only for very few of them.

To illustrate the difficulty of separating the rare-earth elements, let us point out that to separate didymium salts into their constituents—neodymium and praseodymium—one must carry out more than a thousand recrystallizations, or approximately the same number of precipitations2.

Therefore even now pure preparations of the rare earths, especially of some of the rarest and most difficult to separate, are of very great value. Among such preparations are, among others, the salts of terbium and europium3.

The principal distinction between the rare earths from one another lies in certain of their physical properties, above all their spectroscopic properties. For this reason, from the very first steps of its development, spectral analysis played a major role in the discovery of the rare earths and in establishing their number. The study of absorption spectra, emission spectra, and cathodoluminescence spectra has served at all times as a powerful auxiliary means in the search for new elements and in purifying them from impurities[^4].

Table 1 contains the elements of the rare-earth group and some of their properties (the table was compiled from data taken from G. Hevesy’s monograph[^8], corrected in accordance with the latest literature data).

This table includes 16 elements, but there are only 14 rare earths in the narrow sense of the word. Strictly speaking, yttrium and lanthanum do not belong to this group. However, the chemical properties of these elements are so similar to those of the other 14 that these two elements are often also included in the rare-earth group. A few further

TABLE 1

The rare-earth elements and some of their properties

¹) For Pr and Tb no other tetravalent compounds are known except their oxides.

It will be stated which elements should be called rare earths from the point of view of atomic structure.

The special position of the rare-earth group is explained by the fact that, beginning with lanthanum \((z = 57)\) and up to Lu inclusive \((z = 71)\), the filling of the outer electron shells ceases and the filling of the \(4f\) shell proceeds. The latter, according to the Pauli principle, must contain 14 electrons.

Thus, not counting La, which has not a single \(4f\)-electron, one may expect the existence of a total of 14 rare earths²).

In addition to the spectral data, which will be discussed in detail below, an indirect indication of such a structure of the atoms of the rare earths can be seen precisely in the similarity of their chemical and physicochemical properties, most of which are determined by the valence electrons. Chemical—

²) Up to the present time only 13 rare earths are known. The element \(z = 61\) is still unknown; its existence cannot yet be regarded as proved. Recently Hopkins reported that he had obtained a 1% concentrate of element 61⁴⁹a), and the firm A. Hilger in its catalogs for 1935 announced that florencium (element 61), isolated by Rolla, would soon go on sale.

It is necessary, however, to note that Ida Noddack⁴⁹ doubts the correctness of Hopkins’s work, as well as of all previous works in which the discovery of element 61 is reported. Having carried out an enormous amount of work on the fractionation of hundreds of kilograms of rare earths, she came to the conclusion that element 61, if it exists, must be more than \(10^7\) times rarer than samarium.

...similarity indicates the same arrangement and number of valence electrons, while the insignificant change in atomic weight upon passing from one element of this group to another shows that they cannot differ from one another, for example, by a filled inner shell, as is the case with at least the alkali metals, whose chemical properties, though also very close, especially at the end of the series (Rb, Cs), nevertheless have atomic weights that change very greatly (Rb — 85.5; Cs — 132.0).

Table 2 gives the structure of the electron shells of normal and triply ionized atoms of the rare earths.

This table was compiled on the basis of Albertson’s data¹ ², checked against the latest literature data. It is seen from it that the number of \(4f\) electrons in a rare-earth atom is not equal to \(z = 57\) (where \(z\) is the atomic

TABLE 2

Electron shells and principal terms of the rare earths

number), as was previously thought. In some places, as, for example, in the transition from europium to gadolinium, the number of \(4f\)-electrons does not change, and the structure of the atom proceeds by the addition of one \(5d\)-electron. However, for the ions of the rare earths (\(M\) IV), there evidently is a strict correspondence between the atomic number and the number of \(4f\)-electrons, as is seen from the same table. The fundamental terms given in Table 2 were calculated according to Hund’s scheme under the assumption of Russell–Saunders coupling \(^{40}\). To what extent this assumption is fulfilled for rare-earth ions found in solutions and crystals will be seen below.

It should be pointed out that Hund’s calculations serve as confirmation of such a structure of the electron shell of the rare-earth ion. Proceeding from the indicated distribution of electrons among the orbits, Hund calculated the coefficients of magnetic permeability for rare-earth ions (\(M\) IV) \(^{41}\). With the exception of europium and samarium, these coefficients agree satisfactorily with the experimental data of Cabrera \(^{16}\) and Meyer \(^{44}\) (Fig. 1). Van Vleck \(^{77}\) and Frank \(^{32}\) also explained the deviations observed for Eu and Sm.

Fig. 1. Paramagnetic susceptibility of rare-earth ions (according to Hund):
1 — theoretical calculation, 2 — Meyer’s data, 3 — Cabrera’s data. Along the ordinate axis is plotted the number of Weiss magnetons.

As was already said above, the similarities of the chemical properties of the rare earths are explained by the peculiarities in the structure of the electron shells. The spectral differences of the elements of this group from other elements owe their origin to these same peculiarities. Soon after the discovery of the rare earths it was observed that solutions of their salts have absorption spectra characteristic for each element. Most investigators engaged in the separation of rare earths used, among other methods, absorption spectra as well to monitor the course of purification of the preparations. Especially favorable for these investigations is the fact that the absorption bands essentially do not shift either when the concentration of the solution is changed or when one solvent is replaced by another. The spectrum of the salts also depends little on the anion entering into the compound \(^{24, 29, 54, 55, 72—74}\). All this made absorption spectra a very important tool of the analytical chemistry of the rare earths of the nineteenth ...

and the beginning of the twentieth century, despite the comparatively low (in most cases) sensitivity of absorption analysis in comparison with emission spectral and X-ray analyses.

If the chemistry of the rare earths has widely made use, and continues to make use, of spectroscopy in its work, then spectroscopy, in turn, finds in the rare earths an exceptionally interesting object of study. The interest that the rare earths present from the point of view of the spectroscopist is due to the fact that the special position of the \(4f\)-electrons, which are in an unfilled shell, leads to the presence in the rare earths of terms that are well protected from external influences. The energies of these terms are such that the transitions observed between them often lie in the infrared, visible, and near ultraviolet regions\({}^{43}\). The spectra of fluorescence and absorption of the rare earths, as will be seen below, in a number of cases consist of comparatively narrow bands, analogous to broadened lines emitted by strongly perturbed gas atoms (on this basis it seems advisable to call these spectra quasi-line spectra). This character of the spectra makes it possible to approach their interpretation using those ideas about the mechanism of absorption and emission of light that have been developed for gases. The good agreement of experimental results with the calculations carried out indicates the validity of transferring certain concepts borrowed from gas spectroscopy to this case. This circumstance greatly simplifies the study of the behavior of the ion in solution, making it possible here to use methods that have been well developed and tested for free atoms. Thus one may hope that further work in this direction will play an essential role in the spectroscopy of liquids.

Moreover, there is no doubt that the study of the absorption and fluorescence spectra of rare-earth solutions can shed light on the phenomenon of solvation, on the coordination of solvent molecules around the metal ion, and on complex formation in solution.

§ 2. ABSORPTION SPECTRA OF SOLUTIONS OF RARE-EARTH SALTS

The experimental material on the absorption spectra of solutions of the rare earths is quite rich. However, the older works, pursuing purely chemical-analytical aims, are not at present of great interest, chiefly because they often contain a large number of contradictions arising both from the low accuracy of the technique and from the absence of pure preparations.

One of the first works in which an attempt was made to establish regularities in the absorption spectra of rare-earth solutions belongs to Ikema\({}^{32}\). He tried to establish a relation between the position of the absorption bands in the visible region of the spectrum and the atomic number of the element. However, he did not succeed in discerning any regularities. In a later work, Hopkins investigated the absorption spectra of solutions containing, with his aim being the discovery of absorption bands belonging to element 61\({}^{40,40a}\). A number of bands observed in the absorption spectrum of various fractions were attributed to this element.

Somewhat later, Prandtl and Scheiner\({}^{51}\) investigated the absorption spectra of aqueous solutions of the rare earths in the visible, ultraviolet, and near infrared parts of the spectrum. Their graph, placed in Fig. 2, is at present the most complete summary of the absorption spectra of rare-earth solutions in the visible and ultraviolet regions.

Let us note here also that the absorption lines of terbium: \(\lambda = 5225\) and \(5728\ \text{Å}\), placed in this graph, as Prandtl himself indicates in his later work\({}^{52}\), in fact belong to neodymium present as an impurity. To the broad absorption bands lying

Fig. 2. Absorption spectra of solutions of the chlorides of the rare earths. From bottom to top the concentrations decrease in the following sequence: normal, \(N/2\), \(N/4\), etc. Thickness of the absorbing layer \(5\ \mathrm{cm}\) (after Prandtl and Scheiner).

in the region shorter than 3000 Å, we shall return below, and now proceed to investigations in the infrared region.

The investigation of the absorption spectra of solutions of rare earths in the infrared region was carried out by Freymann and Takvorian[^33]. The authors used a thallium photoelement as the indicator. A summary of their results, obtained with preparations very reliable as regards purity (Urbain’s laboratory), is given in Table 3.

TABLE 3

Infrared absorption spectra of solutions of rare-earth chlorides

Thickness of the column 1 cm, concentration \(N = 1\) mole/l, \(\lambda\) in Å

Recently, measurements of the absorption spectra of rare earths in the infrared region were made by Gobrecht[^38]. The measurements were made on crystals of octahydrate sulfates, and for the far infrared region—on solid solutions of rare earths in borax. The region up to \(2.5\,\mu\) was investigated. Up to \(1.2\,\mu\) the spectra were studied photographically, from \(1.2\) to \(2.5\,\mu\)—with a thermoelement with a galvanometer. Detailed tables are included in the cited work.

In the works cited above, the wavelengths of the absorption bands of solutions of all rare earths are given, with the exception of ytterbium in the ultraviolet, visible, and near-infrared parts of the spectrum. For Ce and Yb it should not be expected that narrow absorption bands1 will be present, because the \(4f\) shell of these elements contains, respectively, 1 and 13 electrons[^34,^35].

Absorption spectra of rare earths can also be detected when the rare earth is present as an impurity in some transparent mineral, for example fluorite. The absorption spectra of each rare-earth element retain in the crystal the same general appearance as in solution; however, the individual bands are usually split into narrow lines and may be somewhat shifted in different crystals ^5), ^6), ^2).

From the graph in Fig. 2 it is seen that the absorption spectra of rare-earth solutions consist of bands of two types: 1) narrow bands located in the visible, infrared, and near ultraviolet parts of the spectrum; 2) broad bands passing into a continuum, beginning for all rare earths near 3000 Å and extending toward shorter wavelengths. The question of the nature of the absorption bands of the first and second types cannot yet be considered resolved, but something has already been done in this direction. We shall first dwell on the question of the nature of the broad bands (continuum) in the ultraviolet region. The works devoted to this question concern chiefly cerium. This is readily explained both by the relative simplicity of the Ce IV ion (4f), and by the well-expressed structure of the continuum. Investigations were carried out both on crystals and on solutions of various cerium salts.

Here it is first necessary to mention the works of Freed and Spedding, and also of Datta. In 1931 Freed investigated the absorption of single crystals of cerium ethyl sulfate, acetate, and chloride ^34. The investigation was carried out at room and low temperatures. Crystals of pure cerium salts are opaque beginning from 2700 to 2000 Å (2000 is the limit of the investigation). To increase the transparency (to decrease the number of absorbing centers), the cerium salts were dissolved in lanthanum salts isomorphous with them, which are completely transparent down to 2000 Å. In such mixed crystals three broad bands and one very weak, comparatively narrow band were found, detected only in the chloride at room temperature. Table 4, taken from the cited work, contains the wavelengths of the centers of the absorption bands observed by Freed. It is of interest that the width of these bands

TABLE 4

Absorption bands of cerium crystals,
\(\lambda\) given in Å

^2) See also a series of works by Freed, Spedding, and their collaborators in Phys. Rev. 1931–1938 and J. Chem. Phys. 1937–1938.

substantially does not change on going from room temperature to the temperature of liquid hydrogen.

Fried comes to the conclusion that these bands are due to \(4f—5d\) transitions. He explains the great breadth of the bands by the great sensitivity of the excited \({}^{2}D\) level, belonging to the outer \(5d\)-shell, to the influence of the external electric field.

Bose and Datta\({}^{3}\) studied the absorption spectrum of aqueous solutions of \(\mathrm{CeCl}_{3}\) at room temperature. The bands found by them agree well with the bands of Table 4, except for the band \(3020\ \text{Å}\), instead of which a band \(2970\ \text{Å}\) was observed. In addition, they succeeded in detecting one more absorption band at \(2105\ \text{Å}\). The authors attribute the observed bands to \(4f—5f\) transitions, and also to transitions between the excited levels \(5\,{}^{2}D\) and \(6\,{}^{2}P\). Thus a discrepancy arose between Fried’s interpretation of the absorption bands, according to which all the absorption bands correspond to transitions from the ground level, and the interpretation of Bose and Datta, who assigned three of the five bands observed by them to transitions

\[ 5\,{}^{2}D_{\frac{3}{2},\,\frac{5}{2}} — 6\,{}^{2}P_{\frac{1}{2},\,\frac{3}{2}} . \]

To resolve this contradiction, Datta and Debb\({}^{17}\) investigated the absorption spectrum of \(\mathrm{CeCl}_{3}\) solutions with a monochromator. Under illumination with a continuous spectrum, ions in an excited state may be present in the solution; however, in investigation with monochromatic light, excitation of \((5\,{}^{2}D)\) ions in the solution will not occur (unless, of course, the frequency of the exciting light corresponds to the energy of the transitions \(4\,{}^{2}F — 5\,{}^{2}D\)). Therefore the absorption bands corresponding to transitions from the excited level should not be observed.

The investigation showed that the \(2400\ \text{Å}\) band is present in the solution under the conditions of Datta and Debb. In the authors’ opinion this refutes the view of Bose and Datta concerning this band as the result of the \(5\,{}^{2}D — 6\,{}^{2}P\) transition. In agreement with Fried’s opinion, Datta and Debb believe that all the observed absorption bands of cerium correspond to \(4f—5d\) transitions.

The question of the transitions with which the ultraviolet bands of \(\mathrm{Ce}^{\mathrm{IV}}\) are connected may be regarded as finally solved after the work of Roberts, Wallace, and Pierce\({}^{53}\).

These authors determined the oscillator strengths \(f\) for these bands. The number \(f\), as is known, is related to the probability of the corresponding transition \(A\), and hence to the intensity of the band (line), by the relation

\[ f=A_{rq}\frac{g_rmc^2}{g_q\,8\pi^2 e^2\nu_{rq}^2}. \tag{1} \]

Experimentally the quantity \(f\) can be determined in various ways. The authors used a somewhat simplified Kravets formula\({}^{42}\), relating the quantity \(f\) to the absorption coefficient within the absorption curve,

\[ f=\frac{4\nu_0 m c}{e^2 N}\int_{0}^{\infty} k_\nu\,d\nu, \tag{2} \]

where \(e\) and \(m\) are the charge and mass of the electron, \(N\) is the number of atoms in \(1\ \mathrm{cm}^3\), \(c\) is the velocity of light, \(d\) is the thickness of the absorbing column, and \(k_\nu\) is the absorption coefficient for frequency \(\nu\), determined from the relation

\[ I=I_0 e^{-\frac{4\pi d}{\lambda}\,k}. \tag{3} \]

For \(f\) the following values were found: for \(\lambda = 2540\ \text{\AA}\), \(f = 2\cdot 10^{-2}\); for \(\lambda = 2960\ \text{\AA}\), \(f = 8\cdot 10^{-3}\). A theoretical calculation, carried out by methods of wave mechanics, on the assumption that these bands correspond to the transitions

\[ 4^{2}F_{\frac{3}{2}} - 5^{2}D_{\frac{3}{2},\,\frac{5}{2}}, \]

gave \(f_{2960} = 1.4\cdot 10^{-3}\), \(f_{2540} = 2.3\cdot 10^{-2}\), which is in good agreement with experiment.

Thus it may be regarded as established that the absorption bands of Ce IV are connected with transitions of the \(4f\) electron to outer electronic levels.

An equally detailed investigation of the ultraviolet continuum for other rare earths has unfortunately not yet been carried out. The work on the absorption spectrum of Yb IV, whose ground state is the same as that of Ce IV—\({}^{2}F(4f^{13})\)—done by Fried and Mezirow,\(^{55}\) showed that the ultraviolet absorption of ytterbium is considerably weaker than that of cerium; this is understandable if one assumes that here too we are dealing with transitions to terms belonging to the \(5d\) shell. Indeed, the first excited level of Yb IV will be \({}^{4}K\); consequently, transitions from the ground level \({}^{2}F\) will be forbidden (\(\Delta L > 1\)), whereas for Ce IV, whose excited level is \({}^{2}D\), the corresponding transitions are allowed. Fried and Mezirow consider it probable, on the assumption that the absorption of ytterbium is associated with its photoionization and transition into the quadruply ionized ion Yb V.

Apparently, for all rare earths the continuum in the ultraviolet is connected with the transition of an electron from the inner \(4f\) shell to the peripheral shells of the ion.\(^{1}\) This origin of the continuum, besides the analysis carried out in the case of Ce, is also indicated by a number of indirect considerations: the great sensitivity of these bands to the influence of the surrounding medium (thus, for example, the wavelengths in crystals and in solutions differ greatly from one another); the fact that this continuum does not change its structure and is not resolved into separate bands when the temperature is lowered also indicates that here we are dealing with transitions to outer levels. The narrow bands observed in the absorption spectra of solutions of salts of all rare earths, with the exception of Ce, Yb, and Lu, behave quite differently.

First of all, attention should be drawn to the fact that in minerals, in pure salts of different composition, and, finally, in solutions the absorption spectra of the rare earths do not undergo substantial changes. This, of course, indicates that the carrier of the absorption bands is in all these cases one and the same. Such a carrier can only be the rare-earth ion. Indeed, in the most diverse compounds into which the rare earth enters with the same valence, the absorption spectra are almost identical, but when the valence changes the absorption spectrum changes sharply. Of extraordinary interest in this respect is the work of Butement and Terrey\(^{12}\) on the absorption spectrum of Sm III salts, which is isoelectronic with Eu IV; indeed, the absorption spectrum of Sm III proves to be similar to the spectrum of Eu IV, but shifted relative to it into the red part of the spectrum, as should be the case for spectra of isoelectronic series. However, although a line-like absorption spectrum is also characteristic of the rare-earth ion, the surrounding medium can nevertheless sometimes exert a substantial influence on it, which is manifested especially strongly in the spectra of crystals. This influence is noticeable—

\(^{1}\) It is necessary to point out that the absence in Prandtl’s table of ultraviolet absorption of praseodymium is explained by experimental error. The existence of an ultraviolet absorption band in a solution of praseodymium salts was definitively proved by the authors in a work carried out jointly with Novikova-Minaš.\(^{52a}\) The absorption band of praseodymium lies at \(2150\ \text{\AA}\). The praseodymium absorption bands found by Meckerel\(^{47}\) and ascribed by him in the region \(2800\text{--}2200\ \text{\AA}\), as Lange showed,\(^{42a}\) belong to cerium.

... Bunzen also observed this,^4 finding that the corresponding absorption bands of solutions of the sulfate, acetate, and chloride of didymium are shifted relative to one another. He also points to the difference between the absorption spectra of crystals of \(Di_2(SO_4)_3 \cdot 8H_2O\) and aqueous solutions of this salt^2).

Later Brauner^5 noted that the absorption bands of \(Di\) solutions shift somewhat if \(Sm\) is present in the same solution.

J. Becquerel^6 showed that the absorption spectra of solid salts of didymium change quite appreciably when the anion is replaced, but the absorption spectra of solutions of different salts are almost identical.

Comparatively recently, the question of the influence of the anion on the absorption spectra of rare earths was studied by Ephraim and collaborators.^24–29 In their works they established that the absorption bands of anhydrous halide salts of rare earths shift into the violet part of the spectrum as the atomic weight of the halide decreases. The authors explained this shift by a change in the degree of contraction of the rare-earth ion when one anion is replaced by another. In the study of hydrated salts, a violet shift was found that increases with increasing degree of hydration; moreover, on increasing hydration the spectra of different salts approach one another, becoming identical in dilute solutions within the limits of measurement error. The authors explain this by the fact that in very dilute solutions, where the degree of dissociation is high, the anion cannot exert any substantial influence on the degree of solvation of the metal ion. Thus the absorbing ions (solvates) of the metal in solutions of different salts are identical and, consequently, the absorption spectra are identical as well.

However, a certain influence of the surrounding medium on the narrow absorption bands in solutions can nevertheless be detected. Unfortunately, a systematic investigation of this question has not yet been carried out, and in the literature there are only very fragmentary indications of it. Thus, for example, Becquerel^6 noted that the position of the absorption bands of a solution of neodymium chloride in ethyl alcohol differs somewhat from that observed for solutions in methyl alcohol. The addition of nitric acid to aqueous solutions of rare-earth salts likewise causes small shifts of the bands.^13, ^14 Finally, the absorption spectra of solutions of \(Nd_2(SO_4)_3\) in \(D_2O\) differ somewhat from the spectra of ordinary aqueous solutions.^36

A detailed qualitative investigation of the absorption spectra of neodymium acetylacetonate in various organic and inorganic solvents was carried out by M. Radojčić.^54, ^55 In contrast to the results of Uzymasa, she came to the conclusion that it is impossible to relate the observed changes in the spectrum either to the dielectric constant of the solvent or to the dipole moment of its molecules. To illustrate the observed changes in the spectrum, part of the table given by Radojčić in her article^55 is reproduced here (Table 5). All these fragmentary data on the influence of the solvent indicate that a more detailed investigation of this question may prove very important for clarifying a whole series of questions connected with the structure of liquids. In fact, the absorption spectrum of a rare earth retains its individuality in all solutions; however, the specific properties of the solution prove capable of causing quite noticeable changes in this spectrum. These changes may concern the positions of the band maxima, the number of these maxima (“fine structure” of the band), and, apparently, the intensities of the bands. The absorption spectra can be observed at very low concentrations of rare earths, so that the macroscopic properties of the solvent will practically not be changed by the introduction of a foreign impurity in the form of a rare-earth salt. It is perfectly clear—

^2) \(Di\) — didymium. This was the earlier name for the mixture \(Nd + Pr\), which for a long time was considered a single element.

TABLE 5

Shift of the neodymium absorption bands in various solvents, \(\lambda\) in Å

\[ \mathrm{Nd(C_5H_7O_2)_3} \]

s—strong, sl—weak.

This is because these absorption spectra may serve as a powerful tool for the study of intermolecular interactions in solution. As an example of a problem that can be solved with the aid of such a method, we point to the following: Hütting and Spedding \(^{56}\) showed that the structure of the absorption bands of various crystals of one and the same rare earth is determined, basically, not by the chemical composition but by the crystallographic structure of the given salt. According to the available data, there is every reason to believe that the structure of the absorption bands in an aqueous solution of a given salt is similar to the structure of the absorption bands of crystals of this same salt (if one disregards the width of the bands, which in solution is usually somewhat greater than in the crystal). From this a very probable conclusion is drawn regarding the connection between the quasi-crystalline structure of the liquid and the absorption spectra of solutions of salts of rare earths \(^{1}\).

The influence of temperature on the absorption spectra of rare-earth solutions has been studied very little. Solid solutions have been subjected to study, since it is possible to work with them over a wide temperature interval. With a considerable increase in temperature the bands broaden strongly and merge into the red portion of the spectrum \(^{79}\). Upon cooling from room temperature to the temperature of liquid air, no noticeable changes occur in the width and position of the bands. However, at low temperatures in some cases, apparently, an increase in the intensity of absorption takes place \(^{7,19}\).

We now turn to the question of the spectroscopic interpretation of the narrow bands observed in solutions of rare earths.

\(^{1}\) Very recently Fried and his co-workers \(^{34a}\) have been working in this direction. They compared the spectra of aqueous dilute solutions of europium chloride, nitrate, and sulfate and of solid crystals of these salts. In the spectra of these there is a far-reaching analogy, which, according to the authors, indicates that the field around the europium ion in solution has approximately the same orientation and intensity as the field around the europium ion located in the crystal lattice. The authors believe that the results obtained by them prove the inapplicability of the Debye–Hückel theory to electrolytes containing trivalent ions in the concentration range studied by them (\(1.5\)—\(0.0007\) mole).

This question began to be developed only in very recent times. First of all, it is necessary to decide to which electron shells the terms belong between which transitions occur upon absorption. It seems most probable that here we are dealing with forbidden transitions between terms belonging to the filled \(4f\) shell. However, of course, such a point of view is not the only possible one. Thus, as early as 1930, Laporte put forward the hypothesis that the narrow absorption bands are connected with transitions to electron levels belonging to the outer shells of the ion. This point of view still has its adherents in Spedding and his collaborators\({}^{56}\). Van Vleck in 1937 for the first time posed clearly the question of the type of transitions leading to the appearance of the narrow absorption bands of the rare earths\({}^{77}\). Analyzing the possible types of transitions, Van Vleck points out that the transition probabilities must be substantially different depending on whether we adopt one point of view or the other. In the case of forbidden transitions \(4f—4f\), the magnitude of the oscillator strength \(f\) must be of the order \(10^{-5}—10^{-8}\). For allowed transitions to outer electron shells the magnitude \(f\) should be considerably larger \((\simeq 1—10^{-3})\). Until now no direct determination has been made of the transition probabilities for the narrow absorption lines of the rare earths in solutions. Therefore such a measurement was undertaken by the authors\({}^{57}\). As the object of study three absorption bands of praseodymium were chosen, lying in the blue part of the spectrum. These bands belong to the number of the brightest absorption bands of all the rare earths among those located in the visible and ultraviolet regions of the spectrum (excluding, of course, the ultraviolet continuum). Thus the value of \(f\) obtained for these bands gives an upper limit for the values for all the other absorption bands of the rare earths.

The measurements were carried out by the absorption method described above. For an exact measurement of the band contour, photographs were taken with a large diffraction grating. Calculation by Kravts’s formula gave for \(f\) the values \(f_{444}=2\cdot10^{-5}\); \(f_{469}=6\cdot10^{-6}\); \(f_{483}=4\cdot10^{-8}\) (the subscript denotes the wavelength of the corresponding band in millimicrons). At the same time we calculated the values of \(f\) for the bands necessary to cover it, cited in the article of McKeridge\({}^{46}\). The resulting numbers likewise do not exceed several units of the sixth decimal place. These measurements give every basis for asserting that in all these cases the narrow absorption bands are connected with transitions within the \(4f\)-shell.

However, the resolution of this question constitutes only the first and least difficult part of the problem. It is considerably more important and difficult to determine those terms between which transitions give rise to the narrow absorption bands.

The difficulty of this problem is explained by the fact that, as yet, neither the selection rules to which the transitions in these cases are subject are known, nor even all the possible terms. The latter were calculated by Gibbs, Wilber, and White\({}^{37}\) for equivalent \(f\)-electrons and are given in Table 6.

It is necessary to point out that the calculation was made on the assumption of the applicability of Russell–Saunders coupling between the moments \(L\) and \(S\), which, as we have seen, is quite well confirmed for rare earths possessing a discrete fluorescence spectrum (see below). However, the calculated terms cannot yet be considered fully reliable.

Nevertheless, such an assumption is at present the only guiding thread that can be used in attempts to explain the observed absorption spectra. In the works pertaining here, almost exclusively Pr IV and Nd IV were studied as the simplest objects (\(2\) and \(3\), \(4f\)-electrons).

In the interpretation of the absorption spectra of praseodymium, two works appeared almost simultaneously, by Goobrecht\({}^{38}\) and Merz\({}^{48}\). As is evident from Table 6, for praseodymium the following terms are possible: \({}^{1}(S, D, G, I)\),

TABLE 6

Possible terms for \(n\) equivalent
\(f\)-electrons

\({}^{3}(P,F,H)\); of these, the term \({}^{3}H_{4,5,6}\) is the ground term. On the basis of the works cited, it appears probable that the blue absorption bands of Pr IV, \(\lambda = 444, 469\), and \(482\,m\mu\), are connected with the transitions \({}^{3}H_{4}—{}^{5}P_{0,1,2}\). It has not yet been possible to determine unambiguously the upper terms for the bands \(\lambda = 589\) and \(597\,m\mu\). In exactly the same way, the infrared bands \(\lambda = 1.44\) and \(1.92\,\mu\) are interpreted by Gobrecht with insufficient reliability. They may be correlated not only with the transitions \({}^{3}H_{4}—{}^{3}F_{2,3,4}\), as Gobrecht does, but also, for example, with the transition \({}^{3}H_{4}—{}^{3}H_{6}\) (Fig. 3). The possibility of such transitions within one and the same term is by no means excluded a priori. Thus we see that even in the simplest case of two \(4f\)-electrons, only three bands out of the eight observed have been interpreted more or less reliably.

Fig. 3. Scheme of the terms of the Pr IV ion
(according to Gobrecht)

For the case of neodymium, the question of the systematics of the absorption bands has likewise been solved only partially. Meggers attempted to fit a series of observed absorption bands of neodymium into a term scheme. The scheme given by him is shown in Fig. 4. However, of course, this scheme can be regarded only as an attempt to approach the solution of the problem. The large number of lines that do not fit into it, as well as the low accuracy of the measurements due to the relatively large width of the bands, make the solution ambiguous to a considerable degree. It may also be noted that Gobrecht attempts to combine certain groups of infrared absorption bands into a single multiplet. Thus, for example, according to Gobrecht, a multiplet consists of six absorption bands of samarium\({}^{38}\): \(\lambda = 1.56;\ 1.50;\ 1.39;\ 1.25;\ 1.06\), and \(0.92\,\mu\).

TABLE 7

Fluorescence bands of rare-earth solutions
according to Scheibe (abridged table),
\(\lambda\) given in mµ; the intensities of the bands are indicated in parentheses

For these bands the interval rule is approximately fulfilled, so that, apparently, in this case we are in fact dealing with a multiplet; however, it is unknown whether these bands correspond to the transition \({}^6H — {}^6P\) or \({}^6H — {}^6F\).

Fig. 4. Absorption scheme of the Nd IV ion (after Meggers)

Summarizing this section, one may say that the narrow absorption bands of the rare earths are associated with transitions within the \(4f\)-shell. However, with isolated exceptions, the terms connected with these transi-

dams, have not been established. Meanwhile, the importance of knowing the terms is quite obvious, since only such knowledge will make it possible for us to explain the distribution of bands in the spectra of the various rare earths, to clarify the selection rules prevailing under these conditions, to clarify the influence of electric and magnetic fields—in short, to understand that mysterious mechanism with which the appearance of such quasi-line spectra in the liquid and solid states is connected. It must be noted, however, that intensive work in this direction has been under way comparatively recently, and the results obtained allow one to hope that these questions will soon be resolved to a considerable extent.

§ 3. LUMINESCENCE OF RARE-EARTH SOLUTIONS

Just as do the absorption spectra, the emission spectra of the rare earths possess very characteristic properties.

Luminescence under the action of cathode rays of various substances containing rare earths was discovered by Crookes, who studied the action of cathode rays on glass, precious stones (gems), minerals, etc. The spectra of the cathodoluminescence of the rare earths in the period 1880–1900 were investigated by a number of researchers, but most of all in this field was done by Crookes himself, Lecoq de Boisbaudran, and Urbain. These investigations, like most of the old spectroscopic investigations, pursued purely chemical-analytical aims and only from this point of view can they now be of interest. We shall not here concern ourselves at all with the extensive literature on phosphors activated by rare earths, and in accordance with the problem posed shall confine ourselves only to an analysis of the luminescence of solutions of rare-earth salts.

The method of exciting the cathodoluminescence of solutions was proposed by Lecoq de Boisbaudran8,9, who gave it the name “reversed spectra” (Spectres renversés). To observe these spectra, Lecoq de Boisbaudran passed a spark from a small inductor between a solution of a rare-earth chloride and a platinum electrode. The surface of the liquid around the spark glowed when the solution was made the positive pole. The spectrum of this luminescence in the case of solutions of Tb, Eu, Sm, and Dy salts consists of a small number of narrow bands.

Subsequently, experiments on cathodoluminescence were repeated by Demarçay20–23, and also by Urbain and Urbain and Perrin74–76. The latter authors modified the method of Lecoq de Boisbaudran and transferred the experiments into vacuum. They subjected solutions of rare earths in sulfuric acid to the action of cathode rays. Almost simultaneously with these works, Soret discovered the photoluminescence of solutions of rare-earth salts65. According to this author, solutions of rare-earth salts, when excited by the light of a cadmium spark, fluoresce: cerium and ytterbium—violet, didymium—blue, lanthanum—blue, samarium—green, terbium—greenish-yellow, and erbium, which contains holmium—with yellow color. Soret did not investigate the fluorescence spectrum, but attempted, using an eyepiece with a fluorescent screen, to determine the region of excitation; more precisely, he determined which lines of cadmium, aluminum, and other sparks best excite fluorescence.

After Soret, solutions of rare-earth salts were investigated by Stark and Steubing66, who, in a work devoted to the study of the fluorescence of organic compounds, give a table containing the wavelengths and intensities of the broad bands they found in the fluorescence spectra of aqueous and alcoholic solutions of various rare earths. The authors indicate that nothing is known to them about the purity of the preparations used (Table 7).

Attention is drawn to the very good agreement of the wavelengths of the bands observed by Stark and Steubing in the fluorescence spectra of various rare earths, which, of course, is explained precisely by...

SPECTROSCOPY OF SOLUTIONS OF SALTS OF THE RARE EARTHS

contamination of all preparations by extraneous rare earths. This, of course, greatly devalues the data obtained by them.

A systematic study of the phenomenon of fluorescence in solutions of salts of the rare earths was begun only two or three years ago. The fluorescence of aqueous solutions of terbium salts has been investigated in the greatest detail. Seidel, Larionov, and Filippov \(^{58,60}\) showed that the spectrum of this fluorescence consists of seven narrow bands situated in the visible and near infrared regions of the spectrum (Fig. 5). The wavelengths of these bands are: 488, 545, 585, 620, 648, 670, and 681 \(m\mu\). These bands could be correlated with definite transitions within the \(4f\) shell. The scheme of the transitions is given in Fig. 6. According to this scheme, emission occurs in transitions from one upper term to all the constituent terms of the main term \(\mathrm{Tb}^{IV}\)—\({}^{7}F_{6,5,4,3,2,1,0}\). The correctness of this scheme is indicated by a whole series of facts: 1. Coincidence of the short-wavelength fluorescence band with one of the absorption bands of terbium solutions (\(\lambda = 488\ m\mu\), Fig. 2). 2. Satisfactory observance of the interval rule for the differences of the frequencies

Fig. 5. Fluorescence spectrum of a solution of terbium nitrate. Comparison spectrum—neon. Band wavelengths are given in millimicrons (Seidel, Larionov, and Filippov)

of the fluorescence bands (Fig. 7). 3. The presence of a long afterglow of the solutions (\(\simeq 0.006\) sec), which indicates that the excited level from which the transition occurs upon emission belongs to the inner \(4f\) shell. This is also indicated by the fact that up to now it has not been possible to detect the quenching of this fluorescence, which is usually always observed in solutions.

It is extremely interesting to point out those features which sharply distinguish this fluorescence from previously known cases of fluorescence of liquids. The first thing that catches the eye is the presence of an extremely characteristic quasi-line spectrum. In contrast to this, in other cases of fluorescence of liquids, very broad bands are usually observed, often merging into a continuous spectrum. The second feature of the phenomenon is the long afterglow. The only previously known case of afterglow in liquids is the afterglow of uranyl solutions discovered by Vavilov \(^{89}\). However, in the latter case the duration of the afterglow is considerably less than the duration of the afterglow of terbium (\(10^{-4}\) sec for uranyl, \(10^{-3}\) sec for terbium). These features of the fluorescence are, of course, connected—as are the features of the absorption spectra—with the specific structure of the electron shells of the rare earths. Quasi-line fluorescence spectra are observed, besides terbium, also for solutions of salts of Sm, Eu, Gd, and Dy \(^{68,69,54}\). In all these cases the mechanism of emis-

radiation, apparently, is one and the same. Thus, for Gd IV, whose ground term is \({}^{8}S_{7/2}\), only one fluorescence band is observed, \(\lambda = 411\,m\mu\), coinciding with the absorption band of Gd IV, which agrees with the fact that all \(S\)-terms are singletons.

Fig. 6. Diagram of fluorescence of the Tb IV ion (Zaidel, Larionov, and Filippov)

Table 8 gives the wavelengths of all fluorescence bands located in the visible and ultraviolet regions, observed in solutions of Sm, Eu, Gd, Tb, and Dy. From comparison of these numbers with the data of Table 6 it is seen that the fluorescence and cathodoluminescence spectra coincide quite well with one another. In both these cases, apparently, completely identical electronic transitions take place. It is interesting that in those, and only those, rare earths, quasillinear fluorescence spectra are observed in which the phenomenon of cathodoluminescence of solutions is observed (Gd was not investigated for cathodoluminescence in the ultraviolet region, where all of its absorption bands lie);

Fig. 7. Observance of the interval rule for terbium fluorescence bands. Along the abscissa are plotted the quantities \(\dfrac{j(j+1)}{2}\), along the ordinate axis—wave numbers. The radii of the circles are equal to the half-width of the band (Zaidel)

TABLE 8

Fluorescence bands of aqueous solutions in the visible and ultraviolet regions, \(\lambda\) in \(m\mu\)

exactly the same, cathodoluminescence of Eu, Sm, Tb and Dy in the infrared region of the spectrum has not been studied by anyone).

Soon after the publication of our works on the fluorescence of terbium, there appeared a number of papers by German authors which arrived essentially at the same results. The data of Gobrecht^[38] (schemes of the fluorescence of Sm and Eu are shown in Figs. 8 and 9). In these two cases a somewhat more complicated picture is observed than in the case of terbium. In the case of Sm there is a transition to two multiplets: the fundamental \({}^6H\) and the excited \({}^6F\) or \({}^6P\). In the case of europium, two levels are observed from which transitions occur. However, the correctness of the latter scheme is doubtful. In a work from the same laboratory published somewhat later^[68], a scheme of the fluorescence of the Eu IV ion is given which contradicts Gobrecht’s scheme. In any case, it may be considered established that in the latter two cases the fluorescence bands located in the visible region are the result of transitions to the components of the ground term. It is understood, for the reasons set forth in the preceding paragraph, that the question of the unambiguous determination of the upper term from which transitions occur in emission remains open for the time being. The large number of terms in the \(4f\) shell (more than 100 for terbium) makes the solution of this question very difficult.

Fig. 8. Scheme of the terms of the Sm IV ion (after Gobrecht)

It is interesting to note the following fact: the fluorescence of europium can be excited either by absorption of light in a broad absorption band in the far ultraviolet, or by absorption in narrow absorption bands situated in the near ultraviolet and visible regions of the spectrum; it is remarkable that, irrespective of the method of excitation, no substantial changes are observed in the emission spectra. The same phenomenon is observed for terbium. Sm and Dy could not be excited in the region of continuous absorption, but upon excitation in any of the narrow absorption bands the emitted spectrum is always one and the same^[19]. However, the intensity of the glow and the optimum concentration, for which the brightness of the glow is greatest, depend on the method of excitation. Thus, for Eu the minimum concentration at which glow is observed when it is excited in the region of continuous absorption is \(<0.01\%\). Upon excitation in discrete absorption bands the minimum observed concentration rises to \(\sim 0.2\text{--}0.3\%\). However, at the optimum concentrations (\(0.2\%\) for excitation in the continuous band and \(3\text{--}4\%\) for excitation in narrow bands) the intensities are considerable if excitation is carried out in the discrete absorption bands. Apparently this is connected with the difference in the absorption coefficient for different bands^[64]. For Tb, upon excitation in the far ultraviolet, the minimum detectable concentration lies below \(10^{-6}\%\). At the same time, according to the data of Gobrecht and Tomaschek^[69], who excited Tb by absorption in discrete bands, the detectable concentration is only \(10^{-3}\%\).

For the time being it remains quite unclear why, irrespective of

whether the transition upon absorption occurs to a given level, emission always occurs from one and the same level. In other words, why, before returning to the normal state, does the ion always pass through one and the same excited state (of course, Stokes’ rule must be obeyed, and the absorption band in which excitation occurs must have a wavelength shorter than the short-wave emission band). This phenomenon can, of course, be explained by the fact that return to the normal state always occurs by stepwise transitions, and that only from one definite term do direct transitions to the ground term occur, accompanied by emission in the visible region (Fig. 10). Such an explanation, however, appears rather forced and unconvincing. All these questions still await their solution.

Fig. 9. Term scheme of the Eu IV ion (after Gobrecht)

Fig. 10. Hypothetical scheme illustrating the mechanism by which fluorescence of rare-earth solutions arises (in particular, of the Tb IV ion) when they are excited in the region of continuous absorption

In favor of the correctness of the general scheme of fluorescence of solutions of the indicated rare earths, i.e. of the fact that the transitions occur to the component of the ground term, there is also the work carried out by Blank[^7]. She succeeded in showing that one of the absorption lines of the Sm IV ion (\(\lambda = 422\ \mathrm{m}\mu\)) corresponds to a transition from the thermally excited level Sm IV \(\left({}^{6}H_{\frac{7}{2}}\right)\). The frequency difference between this line and the corresponding line originating from the ground term \({}^{6}H_{\frac{5}{2}}\) coincides, within the limits of experimental error, with the frequency difference of the two short-wave fluorescence bands of samarium, and also with the calculated splitting of the ground term of the samarium ion. An analogous observation was made by Selwood[^67] in 1930, who observed a samarium absorption band at 5960 Å. This band is extremely weak, and according to Selwood its intensity increases in warm solutions (80° C). If one takes into account that the wavelength of this band agrees well with the wavelength of the second fluorescence band of samarium solutions (\(\lambda = 595\ \mathrm{m}\mu\)), then it seems highly probable that the absorption band discovered by Selwood belongs to the thermally excited ion Sm IV \({}^{6}H_{\frac{7}{2}}\) and has as its final

with the term from which transitions occur in Sm fluorescence.

The question of the structure of the fluorescence spectra of rare earths as a function of the fluorescing compound and the solvent has been studied very incompletely. The general form of the spectra for any solutions, solid (glass, borax)7,19 and liquid (water, alcohols, benzene)68, is the same, but the structure and position of the bands are somewhat different. For solutions of terbium salts in water, upon excitation in the short-wave region, different numbers of maxima are found in the bands emitted by different salts. It has been found that acids that are bivalent and univalent in sequence give a structure of the spectra determined only by valence (Fig. 11). A more sharply expressed influence of chemical composition on fluorescence was observed by the authors in the case of europium59, excited in the short-wave region. Solutions of sulfuric-acid europium fluoresce with red light, whereas solutions of chlorous europium do not fluoresce; moreover, fluorescence appears

Fig. 11. Photographs of the spectra of aqueous solutions of terbium sulfate and chloride: a — chloride, b — sulfate. The dots mark the lines of scattered light (Zaidel, Kremenevskii, Larionov)

when traces of sulfuric acid are added to the solution of chlorous europium. These observations compel one to suppose the presence of complex formation even in strongly diluted solutions. In this case some complexes apparently prove capable of fluorescing, while others do not. Analogous images may also be explained by the differences in the spectra of fluorescence of solutions of chlorous and sulfuric-acid terbium.

In addition to rare earths that give line fluorescence in solutions, there are rare earths that give fluorescence consisting of broad bands lying in the ultraviolet and visible regions. The most characteristic is the fluorescence of cerium solutions, the band maxima of which extend from 314 to 407 mμ61,65,66. This fluorescence, of a deep violet color, excited best throughout the region 2400–2700 Å, was observed in aqueous and alcoholic solutions, and also in borax (Fig. 12).

Gobrecht39 gave an attempt at its interpretation; however, as follows from our observations61, this interpretation is erroneous and is explained by an incorrectly calculated splitting of the ground term Ce IV (²F), and also by an incorrectly set experiment. Apparently, in the case of Ce, as also in the cases of Nd and Pr, in whose luminescence spectra broad bands are also observed, we are dealing with a considerably more complex phenomenon than, in the case of the rare earths that give quasi-line fluorescence spectra. It is possible that in these cases fluorescence is caused not by purely electronic transitions within the ion, but by a more complex process, the carrier of which is some kind of

complex, similar to a chromophoric group, which accounts for the fluorescence of organic dyes. Therefore it is hardly possible, as Hobrecht tried to do, to approach the fluorescence of cerium from the standpoint of a simple atomic scheme.

Fluorescence whose spectrum consists of two broad bands lying in the ultraviolet region is given by solutions of praseodymium excited by the light of a nickel or cadmium spark. In addition to these two bands, one more diffuse band is observed in the green region of the spectrum.

Solutions of neodymium salts give in the fluorescence spectrum, apparently, only one band; although Tomaschek indicates the presence of two bands, we have not succeeded in detecting the second band. In general it must be said that the fluorescence spectra of the last two elements have so far been very little studied and have not been at all clarified.

Fig. 12. Fluorescence band of cerium. Microspectrophotogram (Zaidel, Larionov, and Filippov)

Quite recently a paper by Mekerdzhi^47 appeared in which an attempt is made to explain the mechanism of the origin of the diffuse fluorescence spectra of solutions of salts of cerium, praseodymium, and neodymium. We shall not dwell here on this interpretation, since all Mekerdzhi’s conclusions are based on a misunderstanding: using insufficiently pure rare-earth preparations, the author simply was unable to distinguish impurities from the principal substance. In solutions of praseodymium and neodymium salts Mekerdzhi found three diffuse bands; in reality one of these bands (the long-wavelength one) belongs to cerium, which was present in both preparations as an impurity, while the other two—to praseodym-

Fig. 13. Qualitative analysis of a europium preparation (Urban) for its content of cerium and gadolinium. Microspectrophotogram (Zaidel and Larionov)

mμ. The band discovered by Mekerdzhi in solutions of lanthanum salts in fact belongs to cerium. Thus, the very elegant theory proposed by the author, explaining the identity of the luminescence spectra of solutions of a series of rare earths, has to be abandoned.

The failure suffered by Mekerdzhi shows what tremendous sensitivity the phenomenon of luminescence of rare-earth solutions possesses, and what great benefit it can bring in solving analytical problems when skillfully applied. Figs. 13 and 14 show this. Fig. 13 depicts a microspectrophotogram of the ultraviolet luminescence spectrum of a solution of an extraordinarily pure preparation of europium prepared by Urbain. The narrow maximum belongs to gadolinium, and the broad one to cerium, present in this preparation as impurities. Fig. 14 gives a quantitative microspectrophotogram, permitting

Fig. 14. Analysis of praseodymium purified by Prof. I. N. Zaozerskii for cerium: a — intensity of the middle of the fluorescence band of Ce in a 0.2% solution of Pr; b, c, d, and e — the same for solutions of Ce: \(6\cdot 10^{-6}\%\), \(8\cdot 10^{-6}\%\), \(1\cdot 10^{-5}\%\), and \(2\cdot 10^{-5}\%\), respectively (Seidel, Larionov, and Novikova-Minash)

one to estimate the presence of cerium in the pure preparation of praseodymium obtained by Prof. I. N. Zaozerskii. a is the intensity of the middle of the fluorescence band of cerium in a 0.2% solution of praseodymium; b, c, d, and e are the same for standard solutions of cerium \(8\cdot 10^{-6}\), \(6\cdot 10^{-6}\), \(2\cdot 10^{-5}\), and \(1\cdot 10^{-5}\%\), respectively. It is easy to see that the cerium content in Prof. I. N. Zaozerskii’s preparation does not exceed \(5\cdot 10^{-5}\%\) relative to praseodymium. Let us note for comparison that Rowland’s praseodymium, sold by Hilger under the name H. S. (spectroscopically pure), contains approximately 10 times more cerium. All this allows one to think that fluorescent analysis will in the near future occupy in rare-earth chemistry the position that, until recently, had been held there by the comparatively insensitive absorption analysis.

Let us note in conclusion one extraordinarily surprising fact: five rare earths grouped around gadolinium possess quasi-line fluorescence spectra. These same five elements possess remarkable nuclear properties—the largest effective cross section for the capture of thermal neutrons. At the same time, gadolinium, the element most brightly fluorescing, has the largest capture cross section, \(\sigma\). For the other four elements, the brightness of luminescence decreases in parallel with the decrease of \(\sigma\).

Of course, it is difficult to say anything about a connection between these two facts. However, such a parallelism in the features of electron shells and nuclei, observed for five consecutively arranged elements, deserves to be noted.

LITERATURE

1. Albertson, Phys. Rev., 47, 370, 1935.
2. Albertson, Astrophys. J., 84, 26, 1936.
3. Bose and Datta, Z. Physik, 80, 376, 1933.
4. Bunsen, Pogg. Ann., 125, 100, 1866; Bethe and Spedding, Phys. Rev., 52, 454, 1937.
5. Brauner, J. Chem. Soc., 23, 278, 1879.
6. Berquerel, J. Cedenbock aangeboden aan H. Kammerling-Onnes, Leiden, 1922.
7. Blank, Acta Physicochimica URSS, 9, 103, 1938.
8. Boibaudran, C. R., 100, 14, 1885.
9. Boibaudran, C. R., 100, 37, 1885.
10. Boibaudran, C. R., 102, 899, 1886.
11. Boibaudran, C. R., 114, 575, 1892.
12. Bucement and Terrey, J. Chem. Soc., 1112, 1937.
13. Brunetti, Chim., 5, 391, 1928.
14. Brunetti, Chim., 6, 345, 1929.
15. Brunetti and Ollano, Z. Physik, 75, 415, 1932.
16. Cabrera, J. Phys., 6, 253, 1925.
17. Datta and Debb, Ind. J. Phys., 10, 111, 1936.
18. Datta and Debb, Ind. J. Phys., 10, 163, 1936.
19. Deutschbein, Z. Physik, 102, 772, 1936.
20. Demarçay, C. R., 122, 728, 1896.
21. Demarçay, C. R., 131, 995, 1900.
22. Demarçay, C. R., 132, 1484, 1901.
23. Demarçay, C. R., 130, 1471, 1900.
24. Ephraim and Bloch, Chem. Ber., 61, 65, 1928.
25. Ephraim and Bloch, Chem. Ber., 61, 72, 1928.
26. Ephraim, Chem. Ber., 61, 80, 1928.
27. Ephraim and Ray, Chem. Ber., 62, 1509, 1929.
28. Ephraim, Chem. Ber., 62, 1520, 1929.
29. Ephraim, Chem. Ber., 62, 1639, 1929.
30. Ellis, Phys. Rev., 49, 875, 1936.
31. Ellis, Phys. Rev., 50, 675, 1936.
32. Frank, Phys. Rev., 39, 119, 1932.
33. Freyman et Takvorian, C. R., 194, 963, 1932.
    34a. Freed and Jacobson, J. Chem. Phys., 6, 654, 1938; Freed, ibid., 6, 655, 1938; Freed and Weissman, ibid., 6, 297, 1938.
34. Freed, Phys. Rev., 38, 2122, 1931.
35. Freed and Mesirow, J. Chem. Phys., 5, 22, 1937.
36. Fialkovskaia and Katiushkov, ZhTF, 8, 288, 1938.
37. Gibbs, Wilber and White, Phys. Rev., 29, 790, 1927.
38. Gobrecht, Ann. Phys., (5), 28, 673, 1937.
39. Gobrecht, Ann. Phys., (5), 31, 181, 600, 755, 1938.
    40a. Hopkins, J. Chem. Education, 13, 363, 1936.
40. Harris and Hopkins, J. Am. Chem. Soc., 48, 1583, 1928.
41. Hund, Z. Physik, 33, 855, 1925.
42. Kravets, Absorption of light in solutions of colored substances, Dissertation, Moscow, 1912.
    42a. Lange, Ann. Phys., 31, 609, 1938.
43. Laporte, Phys. Rev., 35, 130, 1930.
44. Meyer, Physik. Z., 26, 51, 478, 1925.
45. Mukherji, Ind. J. Phys., 10, 319, 1936.

1. Mukherji, Ind. J. Phys., 11, 123, 1937.
2. Mukherji, Ind. J. Phys., 11, 399, 1937.
   47a. Mukherji, Z. Physik, 109, 573, 1938.
3. Merz, Ann. Phys., 28, 569, 1937.
4. Noddak, Uspekhi Khimii, 4, 11, 1935.
5. Nutting, and Spedding, J. Chem. Phys., 5, 33, 1937.
6. Prandtl, and Scheiner, Z. allg. anorg. Chem., 220, 107, 1936.
7. Prandtl, Z. anorg. allg. Chem., 232, 267, 1937.
8. Roberts, Wallace and Pierce, Phil. Mag., 17, 934, 1934.
9. Radoitchitch, C. R., 204, 1469, 1937.
10. Radoitchitch, C. R., 206, 335, 1938.
11. Spedding, Howe and Keller, J. Chem. Phys., 5, 416, 1937.
12. Zaidel and Larionov, DAN, 19, 683, 1938.
13. Zaidel, Larionov and Filippov, DAN, 1 (10), 247, 1937.
14. Zaidel and Larionov, DAN, 3, 116, 1936.
15. Seidel, Nature, 139, 248, 1937.
16. Zaidel, Larionov and Filippov, DAN, 20, 351, 1938.
17. Zaidel, Larionov and Kremenevskii, Izvestiya Imen. str. 207, 1937; Acta Physicochimica URSS, 6, 481, 1937.
    62a. Zaidel, Larionov and Novikova-Minash, DAN, 21, 330, 1938.
18. Zaidel, Larionov and Filippov, Journal of General Chemistry, 8, 943, 1938.
19. Zaidel and Larionov, DAN, 16, 449, 1937.
20. Soret, Arch. Sc. phys. et nat., 4, 290, 1880.
21. Stark u. Steubing, Physik. Z., 9, 661, 1908.
22. Selwood, Ann. Chem. Soc., 52, 1937, 1930.
23. Tomaschek u. Deutschbein, Ann. d. Phys., (5) 29, 320, 1937.
24. Tomaschek u. Gobrecht, Ann. d. Phys., (5) 29, 324, 1937.
25. Tomaschek u. Deutschbein, Physik. Z., 34, 374, 1933.
26. Tomaschek u. Mehnert, Ann. d. Phys., 29, 320, 1937.
27. Uzumasa, Bull. Chem. Soc., Japan, 6, 147, 1931.
28. Uzumasa, Bull. Chem. Soc., Japan, 7, 85, 1932.
29. Urbain, Introduction à l’étude de la spectrochimie, Paris, 1911.
30. Urbain, J. Chim. Phys., 4, 247, 1906.
31. Urbain, Ann. Chim. Phys., (8) 18, 231, 1909.
32. Van-Vleck, Electrical and magnetic Susceptibilities, Oxford, 1936.
33. Van-Vleck, J. Phys. Chem., 41, 67, 1937.
34. Wood, Phys. Rev., 37, 1687, 1931.
35. Wawilov u. Lewschin, Naturwiss., 15, 899, 1927.
36. Kheveshi, Rare Earths from the Standpoint of Atomic Structure, NKHTU VSNKh, Leningrad, 1929.
37. Yntema, J. Am. Chem. Soc., 45, 907, 1923.

1. If, of course, one accepts the structure of the electron shells of the ion \(M\) IV, which is given in Table 2. The absence of narrow absorption bands in solutions of cerium and ytterbium salts is serious confirmation of precisely such a structure of the ion. The bands observed by Freymann and Takvorian[^33] in a mixture of Yb + Lu in the region \(950\,m\mu\) (Table 3) possibly correspond to a transition between the components of the ground term Yb IV — \({}^{2}F_{\frac{5}{2},\,\frac{7}{2}}\). The frequencies of these bands are in satisfactory agreement with the splitting of this term calculated by Blanc[^7]. ↩↩↩↩

2. In this respect the rare-earth group somewhat resembles a mixture of isotopes of one element. However, the difference in chemical properties in the former case is considerably greater than in the latter, and therefore the separation of rare earths is still usually considerably easier than the separation of isotopes. ↩

3. It is interesting to note that the terbium oxide supplied by a firm specializing in the preparation of spectrally pure reagents (A. Hilger), under the mark V. P. (very pure), proved, as our spectral analysis showed, to contain mainly yttrium and only a very small amount of terbium. ↩