From Current Literature

Dependence of X-ray Spectra on the Chemical Compound of Elements

N. D. Morgulis.

As is known, X-ray spectra, in contrast to optical spectra, were found to be completely independent of the physical or chemical state in which the element giving rise to this spectrum is found. The position of the lines in them depends exclusively on the atomic number of the element and changes gradually with a change in this atomic number. This fact is explained by the circumstance that X-ray spectra are determined by purely intra-atomic processes, which cannot be affected by all kinds of changes in the outer electron shells, by which, chiefly, the various chemical compounds of the elements are determined.

As it later turned out, this view was erroneous: the structure of X-ray spectra, both emission and absorption, depends on the chemical compound of the substance. This circumstance was first discovered for the absorption spectra of X-rays, and then also for the emission spectra. We shall begin with the absorption spectra.

X-ray absorption spectra in outward appearance resemble band spectra; on a photographic plate there is observed a band having a sharp boundary on the side of greater wavelengths. In the case of \(K\)-absorption spectra this sharp boundary of blackening approximately coincides with the hardest line of the \(K\)-series, the \(K_{\beta_2}\)-line. As it turns out, for the \(L\)-series there are 3 absorption edges, for the \(M\)-series—5 absorption edges, etc., the number of absorption edges corresponding exactly to the number of excitation thresholds of the given series. The results of investigations of X-ray absorption for the \(K\), \(L\), and \(M\) series, carried out by Fricke [1], Hertz [1], and Stenström [1], may be summarized as follows: the X-ray absorption edges by no means have such a simple character; instead of a sharp change of blackening on the photographic plate, regions with stronger and weaker blackening are observed, i.e. we observe something like a “fine structure” of the absorption edge. Moreover, on the photographic plate a white line is often observed precisely at the place where the absorption edge should be located; on both sides of this white line there is observed almost identical blackening of the photographic plate. This line reveals an external analogy with ordinary optical absorption lines.

Bergengren [2], in studying the \(K\)-absorption edges of different modifications of phosphorus, found that the position of this edge depends on the modification of phosphorus; thus he was the first to establish the erroneousness of the view stated above concerning the character of X-ray spectra. This phenomenon was later investigated in detail, for chlorine, sulfur, and phosphorus, by Lind [3], who established that \(K\)-spectra

N. D. MORGULIS

the absorption of X-rays in these three elements depends on the valence which the element has in the absorbing compound. For chlorine the following was found: the \(K\)-absorption edge of this element has the character of a fine structure; here there are observed, as it were, two absorption edges: the principal one—\(K_1\), and the subsidiary one—\(K_2\).

Table I.

Investigations showed that the absorption spectra corresponding to chlorine compounds with the same valence are approximately identical, and their absorption edges, within the limits of the accuracy of measurement, also coincide (the only exception is HCl); the results of the investigation of chlorine are presented in Table I. In the case of sulfur, which was taken both in different modifications and in various kinds of bivalent, tetravalent, and hexavalent inorganic and organic compounds, the results obtained were the same as for chlorine, i.e., here too the effect of the change of the absorption edge with valence was established. In this case the exception was 18 investigated sulfides, for which it turned out that the absorption edge depends on the metallic ion; its position lay in the range from 5005.3 X (Zn S) to 5011.7 X (Cr\(_2\) S\(_3\)).

In the case of tetravalent sulfur compounds the absorption edges, with the exception of SO\(_2\), agree well; the same holds for hexavalent compounds. All the results obtained for sulfur are presented in Table II. For phosphorus, Lindh’s \([3]\) results confirmed the results obtained by Bergenren \([2]\) for the dependence of the absorption edges on the modification (see Table III); in the case of sulfur a similar circumstance could not be established.

Table II.

Table III.

The general results obtained by Lindh \([3]\) for chlorine, sulfur, and phosphorus (Tables I, II, and III) may be summarized as follows. The position of the \(K\)-absorption edge depends on the valence of the absorbing element; it is the harder, the higher the valence. Generally speaking, the position of this absorption edge does not at all depend unambiguously on valence; for example, for tetravalent and hexavalent sulfur compounds the \(K\)-absorption edge of organic compounds is shifted relative to the \(K\)-absorption edge of inorganic compounds by approximately 6 X toward longer wavelengths.

Then the investigation of the absorption edges of a number of elements was continued by Coster \([4]\). He measured the \(K\)- and \(L_3\)-absorption edges and the positions of the white lines \(K\alpha\) and \(L_3\alpha\), which were quite often observed in the fine structure of the absorption edges. These anomalous white lines were usually situated on the soft side of the absorption edges, and it is precisely they that are of chief interest. These white lines do not change either with an increase in the thickness of the absorbing layer or with an increase in the exposure time, whereas the fine structure of the absorption edges in this case changes noticeably. A white line appears only in compounds with high valence; in the same case, if it is observed for some element in a compo-

...compounds with different valences, its position does not change as a result. In all the elements investigated by Koster (Ti, Va, Cr, Mn, Sn, Sb, Te, J), only the $K$- or $L_3$-absorption edges were measured; the $L_2$-absorption edge appeared only in certain cases and was much weaker than the $L_3$-edge; moreover, the impression was obtained that no white line exists here; the $L_1$-edge was altogether weak. From the results obtained by Koster it is seen that here, too, no regularity can be established between the position of the absorption edges and the valence; this is seen, for example, with sufficient clarity from the fact that the $K$-edge in the trivalent chromium compounds $\mathrm{Cr_2O_3}$, $[\mathrm{Cr(NH_3)_6}](\mathrm{NO_3})_3$, and $\mathrm{CrCl_3}$ has $\lambda = 2060.0;\ 2063.8$ and $2063.0$; although, generally speaking, its hardness, in contrast to the white line, increases with valence. Koster explains the phenomenon of the change of absorption edges with valence by a distortion of the electron orbits, which is especially strong in the case of compounds with high valence; he, incidentally, is the first to arrive at the conclusion that, in determining energy levels from absorption spectra, it is necessary to use, for the absorbing layer, the pure element or, at least, one present in a compound of low valence. Soon afterward, for potassium and calcium as well, Lind \[5\] established an effect of the dependence of the absorption edge on the chemical compound. In his other, more recent, work Lind \[6\] investigated the $K$-absorption edges of the elements Si, Ti, V, Cr, Mn, and Fe. Without going into the details of the experiments, which in some cases (e.g., Si) were very interesting, we shall try to summarize briefly the results obtained by him. For Si the pure element and its four valent compounds were used; in the latter case a displacement of the $K$-absorption edge toward harder rays by 23.5 X.E. was observed. In the case of Ti, likewise, the pure element and $\mathrm{TiO_2}$ were used; in the latter case a white line was observed. Measuring its edge from the side of longer wavelengths, Lind found that its wavelength agrees well with the absorption edge of the pure element; it must be supposed that this white line owes its occurrence to pure Ti, which was reduced from the compound ($\mathrm{TiO_2}$) during its illumination by x-rays. The correctness of this supposition was confirmed in the investigation of the absorption edges of V, Cr, and Mn, where the white line also coincides with the absorption edge of the pure element. In addition, here it is also seen that the change of absorption edges with valence is by no means as simple as is seen from the example of the trivalent chromium compounds, where for $\mathrm{Cr(OH)_3}$ and $\mathrm{Cr_2(SO_4)_3}$ the displacement has a magnitude of 4.3 X.E., while for $\mathrm{Cr_2S_3}$ only 2.4 X.E. Possibly this is explained by the circumstance that in the first case the Cr atom is bound to O atoms, and in the second to S atoms.

From Lind’s work it is unquestionably evident that the white line must be ascribed to the pure element, reduced from the compound under the action of x-rays; he also found that, in the case of $\mathrm{HJO_4}$ used by Koster, this compound under the action of x-rays becomes colored dark brown, the color being the more intense the more intense the incident x-radiation; and the $L_3\alpha$ line obtained in this case by Koster, according to Lind’s measurements, coincides with the $L_3$-absorption edge of iodine.

With regard to the $L_3\alpha$ white line of Sb and Sn, Chamberlain \[17\] had already earlier established the same thing as Lind did for the $K\alpha$-line. In addition, she showed that some of the compounds for which Koster indicated the existence of white lines are slowly reduced under the action of x-rays, and on the photographic plate the action of this reduced part will also be observed in the form of a white line. In her other work Chamberlain \[9\] gives the following table (IV), confirming Lind’s view on the nature of the anomalous white lines. From this table it is seen that the $K$-absorption edges of pure Ti, V, Cr, Mn and the $L_3$-edges of Sn, Sb, Te, and J are in good agreement with the corresponding anomalous lines of Koster, which, as a consequence of this, must be ascribed to the elements reduced under the action of x-rays. From a consideration of all these works it is clear that the effect of the dependence of absorption edges on the chemical compound of the element is fully established: for

N. D. Morgulis

For the \(K\)-absorption edges it was observed in 11 elements (P, S, Cl, K, Ca, Si, Ti, V, Cr, Mn, Fe), and for the \(L_3\)-edges in three elements (Sb, Fe, J). No definite regularity can yet be established, but in any case the absorption edge is almost always made harder with increasing valence of the absorbing element.

Table IV.

Let us now turn to the X-ray emission spectra and see whether a similar kind of effect can also be established here. All work in this direction may be subdivided into two categories: first, the structure of \(K\beta\)-lines for different elements in different compounds was investigated, and, second, the structure of \(K\alpha_{12}\)-doublets was investigated. To the first category belong the works of Lind and Lundquist \([9]\), who studied the structure of the \(K\beta_1\)-line of sulfur, chlorine, and phosphorus. As is known, the sulfur \(K\beta_1\)-line has a satellite on the short-wavelength side, \(\beta_x\); the authors investigated the influence of the chemical compound on the distance between these doublets \(\beta_1\) and \(\beta_x\). It was found that in some sulfides a separation of the doublets was obtained, while in others it was not. The following circumstance is interesting: for CuS sulfide and for pure S on a Cu anticathode, separation of the doublets was obtained; on the contrary, for ZnS, FeS, and for S on a Zn or Fe anticathode no separation was obtained; from this one may conclude that, when sulfur is bombarded by cathode rays on an anticathode consisting of Cu, Zn, or Fe, it combines with the material of the anticathode, transforming into CuS, ZnS, or FeS. Moreover, as it turned out, the position of both components is not always the same: for example, in the case of Ag\(_2\)S on an Al anticathode both components were shifted toward longer wavelengths, whereas in the case of Ag\(_2\)SO\(_4\) a shift toward shorter wavelengths occurred. Then the \(K\beta_1\)-line of chlorine and its compounds was studied. For a series of Cl compounds of different valence the \(K\beta_1\)-line and its satellite were obtained, with wavelengths 4394,2 and 4390,6 X. U. For CuCl\(_2\), deviations were obtained, namely \(\lambda_{\beta_1}=4394,75\) and \(\lambda_{\beta''}=4388,4\). For other compounds, a broadened line with \(\lambda=4393,3\) was obtained. In the case of phosphorus the authors obtained the following: red phosphorus and Cu phosphide gave the \(K\beta_1\) line and its satellite \(\beta_x\); the remaining compounds gave only one line \(\beta_{1x}\), which in the case of Zn phosphide was noticeably broadened. Then the structure of the \(K\beta\)-lines of phosphorus and potassium was studied in detail by Lundquist himself \([10]\). For phosphorus he obtained the following results: separation of the lines \(\beta_1\) and \(\beta_x\) was possible when pure P was used on a Cu anticathode, in Cu phosphides, and in all compounds with Na and K. The distance \(\Delta\lambda\) between \(K\beta_1\) and \(\beta_x\), with the exception of Cu phosphides, was always constant and equal to 12,0 X. U.; for Cu phosphide \(\Delta\lambda=13,9\) X. U. We observed an analogous phenomenon earlier: for CuCl\(_2\), where \(\Delta\lambda\) between \(\beta_1\) and \(\beta''\) was 3,0 X. U. greater than for other compounds. In the investigation of phosphorus compounds with a number of other compounds, instead of a pair of lines \(\beta_1\) and \(\beta_x\), a single line \(\beta_{1x}\) was obtained, which, when P was combined with Zn, had a width equal to the distance between the lines \(\beta_1\) and \(\beta_x\). The \(K\beta_3\) line was obtained only occasionally, chiefly when the \(\beta_1\) and \(\beta_x\) lines merged into one \(\beta_{1x}\); on the contrary, when the pair of lines \(\beta_1\) and \(\beta_x\) existed, the \(\beta_3\) line was often absent. Thus the structure of the \(K\beta\) lines of sulfur, chlorine, and phosphorus is influenced by the chemical compound; however, it is still impossible to establish any regularity here, as before. Highly interesting results were obtained for potassium. As is known, the \(K\beta\)-group of potassium lines consists of the principal line \(\beta_1\) and two weak lines \(\beta''\) and \(\beta_2\). The wavelengths of these lines are \(\lambda\beta_1=3446,80\);

\(\lambda \beta'' = 3442.70\) and \(\lambda_{\beta_2}=3434\), (in X. U.). These lines can be obtained when using KCl, but with the use of this same compound one can obtain the chlorine \(K\beta''\), which, as is known, depends on its chemical compound; therefore it was of interest to investigate potassium as well. In the investigation of potassium, 12 different chemical compounds of it were used, and for all of them both the \(\beta_1\)- and \(\beta'\)-lines were obtained. It turned out that no influence of the chemical compound of potassium (on a Cu anticathode) could be established either on the structure of the \(K\beta\)-lines or on their position. As a mean for the 12 different investigated potassium compounds, the result obtained was \(\lambda_{\beta_1}=3446.99\) and \(\lambda_{\beta''}=3443.09\) X. U.

Let us now turn to investigations of \(K_{\alpha_{12}}\) doublets. As is known, the difference in the wavelengths of doublets is a constant quantity, i.e. \(\Delta\lambda=\lambda_{\alpha_2}-\lambda_{\alpha_1}=\mathrm{const}\); its value fluctuates between 3.30 and 4.50 X. U. E. Bäcklin \([12]\) investigated whether the distance between \(K_{\alpha_{12}}\) doublets depends on the chemical compound; it turned out that such a dependence exists. Some of the results obtained by him are given in Table V. As is seen from this table, \(\Delta\lambda\), i.e. the distance between the \(K_{\alpha_{12}}\) doublets, decreases with increasing valence of the element. Moreover, it turned out that the position of both lines also depends on the valence of the element. This displacement of the doublets was investigated by Bäcklin \([13]\) for the light elements Al, Si, P, S and their compounds. For all these 4 elements the effect of displacement of the doublets under the influence of the chemical compound was established quite clearly and definitely. From all that has been said it is clear that X-ray emission spectra also depend on the chemical compound of the substance; this is expressed in a change in the structure of the \(K\beta\)-lines, in a change in the distance between the \(K_{\alpha_{12}}\) doublets, and, finally, in a displacement of the doublets themselves; to establish, however, any sort of regular dependence of the change in the X-ray emission spectrum on the change in valence, with some exceptions, is almost impossible.

Table V.

References

[1] M. Siegbahn. Spektroskopie der Röntgenstrahlen. 1924. S. 129.

[2] J. Bergengren. Zeitschr. f. Physik. 3; 247; 1920.

[3] A. Lindh. Diss. Lund. 1923.
M. Siegbahn. Spektroskopie der Röntgenstrahlen. 1924. S. 137.

[4] D. Coster. Zeitschr. f. Physik. 25; 83; 1924.

[5] A. Lindh. Ark. f. Mat. Astron. och Fys. 18; No. 14; 1924.

[7] A. Lindh. Zeitschr. f. Physik. 31; 210; 1925.

[6] K. Chamberlain. Nature. 114; 500; 1924.

[8] K. Chamberlain. Physic. Review. 26; 525; 1925.

[9] A. Lindh and O. Lundquist:
a. Ark. f. Mat. Astron. och Fys. 18; No. 14; 1924.
b. Ark. f. Mat. Astron. och Fys. 18; No. 34; 1924.
c. Ark. f. Mat. Astron. och Fys. 18; No. 35; 1924.

[10] O. Lundquist. Zeitschr. f. Physik. 33; 901; 1925.

[11] B Ray. Philos. Mag. 49, 168; 1925.

[12] E. Bäcklin. Zeitschr. f. Physik. 33; 547; 1925.