ABSORPTION SPECTRA OF ALKALINE-EARTH METAL FLUORIDES

L. A. LISITSYNA

(Presented by the scientific seminar of the Faculty of Physics and Technology)

The fluorides of the alkaline-earth metals have lattices of the fluorite type. This lattice is an interpenetrating cubic lattice consisting of \(F^-\) ions and a face-centered lattice of \(Ca^{2+}\) ions (Fig. 1). The lattice parameter for \(CaF_2\) is \(5.45 \text{ Å}\), for \(SrF_2\)—\(5.78 \text{ Å}\), and for \(BaF_2\)—\(6.19 \text{ Å}\).

At room temperature, the principal type of defects in these crystals are anti-Frenkel defects \([1, 2, 3]\), consisting of a halide vacancy and a halide ion in an interstitial position. The number of anion vacancies and interstitial ions can be changed by introducing an impurity into the crystal. When monovalent impurity ions \(Me^+\) are introduced into the crystal lattice, substitution of \(Me^{2+}\) ions by \(Me^+\) ions is observed; charge compensation then occurs through the formation of an additional number of anion vacancies \([3, 2]\). When trivalent impurity ions \(Me^{3+}\) are introduced into the crystal, owing to the closeness of the ionic radii of \(Me^{2+}\) and \(Me^{3+}\), the formation of mixed crystals \(Me^{2+}—Me^{3+}\) proceeds by the type of isomorphous substitution of \(Me^{2+}\) ions by \(Me^{3+}\) ions, with the formation of \(F^-\) ions in interstitial positions. The latter was demonstrated by measurements of the lattice constant and the density of the crystal \([4, 5]\), and also by measurements of the electrical conductivity of impurity crystals \([2]\). Evidently, the closeness of the ionic radii of oxygen and fluorine (\(1.36 \text{ Å}\) and \(1.33 \text{ Å}\), respectively) can explain the high chemical activity of alkaline-earth metal fluorides in interaction with oxygen and water vapor when the crystals are heated in air. Oxygen, replacing fluorine ions, can enter either in the form of the hydroxyl group \(OH^-\), or in the form of \(O^{2-}\) ions, thereby forming in the crystal an additional number of anion vacancies. The presence of oxygen is the cause of the appearance of red fluorescence of the crystal \([6]\), several additional bands in the infrared region of the spectrum \([7]\), and an additional band in the ultraviolet in the wavelength interval \(205 \div 220\ \text{m}\mu\), according to data of various authors.

Fig. 1. Crystal lattice of the fluorite type.

Interaction with oxygen is also observed in crystals of SrF₂ and BaF₂. Thus, Karas [8] notes that upon heating in air (1000°C, 8 min), clouding of the surface of an SrF₂ crystal is observed, and an absorption band with a maximum at 205 mμ appears in the spectrum of the unirradiated crystal. Slight contamination with oxygen also occurs during additive coloration of these crystals [9].

In later studies of the optical properties of alkaline-earth fluorides, greater attention began to be paid to the prior history of the crystals, taking into account their interaction with oxygen, since it was found that differences in coloration (both additive and subtractive) from crystal to crystal are caused by the presence of oxygen, which may enter the crystal during growth, during the coloration process [7, 10], upon heating of the crystal not only in air but also in an inert atmosphere [1, 2, 6, 11], and during prolonged storage of crystals in air [1].

Additive coloration of CaF₂ crystals

Additive coloration of CaF₂ crystals in metal vapor has a number of distinctive features.

According to Smakula [12], the spectrum of an additively colored crystal does not depend on the experimental conditions and consists (Fig. 2) of four bands with maxima at 225, 335, 400, and 580 mμ.

According to Lüty [13], a spectrum consisting of four bands can be obtained only at a low coloration temperature (500–600°C). When coloration is carried out at a higher temperature followed by rapid cooling, the crystal spectrum consists of the so-called α and β bands (Fig. 3), lying at 370 and 520 mμ, respectively [14]. A number of authors, without specifying the experimental conditions, note that the spectrum of a CaF₂ crystal colored in metal vapor has bands at 320, 370, and 570 mμ [15], at 380 and 550 mμ, and strong absorption near 200 mμ [16].

Fig. 2. Appearance of the absorption spectrum of CaF₂ crystals under additive coloration [10, 12, 13] and under photochemical coloration—[20–22].

Fig. 3. Absorption spectrum of CaF₂ crystals under strong additive coloration.

Fong and Jacko [10] found that coloration at high temperature (850°C) followed by slow cooling causes the appearance of a spectrum consisting of four bands with maxima at 225, 335, 400, and 580 mμ, whereas coloration at this temperature followed by rapid cooling is accompanied by the appearance of bands with maxima at 380 and 555 mμ. Measurement of this spectrum at 77°K made it possible to reveal additional bands with maxima at 733, 670, and 520 mμ.

Lüty proposes the following explanation of the results obtained: the spectrum consisting of four bands, appearing under weak additive coloration, is due to the presence in the crystal of some impurity—

When crystals are colored at high temperature followed by slow cooling, the spectrum also consists of four bands; however, in this case the background, caused by the appearance of colloid in the crystal, increases strongly. The spectrum consisting of the α- and β-bands is attributed to calcium atomically distributed in the crystal. Lüty assumes that this spectrum overlaps a spectrum consisting of four bands. However, measurements at −130°C did not make it possible to resolve the two types of spectra; only a shift of the α- and β-bands toward shorter wavelengths and a decrease in their widths were found. In [17], when especially pure crystals were colored in metal vapors, the authors did not observe a spectrum of four bands, which is good confirmation of Lüty’s assumption.

Fong and Jaccard [10] believe that rapid cooling of additively colored CaF₂ crystals freezes the electrons captured during the coloring process in a certain equilibrium state for the given coloring temperature (850°C). The appearance of four bands in the spectrum when crystals are colored with subsequent slow cooling is explained by the establishment of a new stable state for a lower temperature. Thus, the authors of [10], in contrast to Lüty’s opinion, consider both types of spectra to be caused not by an impurity but by localization of electrons at intrinsic defects of the crystal lattice.

Bontinck [14], in order to clarify the nature of the α- and β-bands, bleached additively colored crystals with polarized and unpolarized light in the β-band. Strong bleaching of the β-band was found, along with a slight decrease of the α-band and the appearance of additional bands at 675, 750, and 825 mµ. Bleaching with light in the α-band causes a slight decrease of the α- and β-bands and the appearance of the same additional bands. On the basis of these experiments, a conclusion is drawn concerning the interrelation of the α- and β-centers and the isotropy of the properties of the centers responsible for the β-band. The following models of the centers are proposed:

α-center — a pair of antifrenkel defects that has captured an electron;

β-center — an electron localized at an anion vacancy.

However, this interpretation of the α-band is refuted by the works of Arends and Carras [9, 18], in which a model of the α-center is proposed as an electron captured at an anion vacancy. In [18] the EPR spectrum of an additively colored CaF₂ crystal is given; it is shown that the relative change in the intensity of the hyperfine-interaction lines in the EPR spectrum agrees well with the theoretical calculation for the interaction of an electron in an anion vacancy with the ions of the nearest environment in the CaF₂ structure. From experiments on the effect of bleaching in the α- and β-bands on the optical-absorption spectrum and the EPR signal of the crystal, it was found that bleaching in the α-band causes a decrease of the α-band after 1.5 hours, complete disappearance of the EPR signal after 4.5 hours, a slight decrease of the β-band, and growth of bands at 670 and 600 mµ. Bleaching in the β-band does not affect the EPR signal; the change in the absorption spectrum in this case is not given in the work. However, the incomplete correlation of the destruction curves of the α-band and of the EPR signal apparently casts doubt on the author’s conclusion that the EPR signal is caused by α-centers.

Hessick [19], on the basis of calculations, proposes a model of the α-center as a center consisting of an anion vacancy that has captured two electrons (F′-center).

Another method of producing color centers in CaF₂ crystals is electrolytic coloring from a cathode tip. According to the data of Smakula [12] and Beryl [20], in this case a spectrum consisting of four ...

absorption bands with maxima at 225, 335, 400, and 580 mμ, while according to Mollwo [1] the spectrum consists of α- and β-bands. In Arend’s experiments [18], when colored in air at 750°C, the crystal spectrum consisted of α- and β-bands and additional bands at 750, 670, and 600 mμ, whose considerable magnitudes, in the author’s opinion, are due to oxygen entering the crystal during coloring in air.

The reason for the appearance of the two types of spectra has not been studied.

Photochemical Coloring of CaF₂ Crystals

Irradiation of CaF₂ crystals with X-rays, electrons, γ-rays, and protons causes the appearance in the absorption spectrum of additional bands (Fig. 2) with maxima at 225, 335, 400, and 580 mμ [1, 17, 20—23].

As the irradiation dose is increased in X-rayed crystals, saturation of all bands is observed [14, 20, 22, 24], and it is noted that the ratio among all bands remains constant with irradiation dose [20]. The identical behavior of all bands under irradiation and the identical rate of destruction of all bands during thermal annealing [20] allowed Beryl to suppose that either all four bands are due to one type of center having different energy levels, or that there exist several types of interrelated centers.

In studying the growth curves of absorption bands as a function of proton-irradiation dose, the author of the present article found that, at doses considerably exceeding those used in [14, 20], anomalous behavior of the band with a maximum at 580 mμ is observed [25]. Whereas all the other bands become saturated, the growth curve of the 580 mμ band has three stages: an initial rapid growth, then saturation, and then a further growth of the band with irradiation dose. A conclusion is drawn as to the different nature of the defects responsible for the band with a maximum at 580 mμ and for the other three bands.

Much information on the nature of the bands can be obtained by studying, at low temperature, the spectra of crystals colored at different temperatures. However, there are still very few works in this direction. Measurement of the absorption spectrum at 77°K of crystals colored with electrons at room temperature resulted in a shift toward shorter wavelengths of the bands with maxima at 225, 335, and 400 mμ, and toward longer wavelengths of the band with a maximum at 573 mμ [3, 18]. Measurement of the spectrum at 4°K [1] made it possible to achieve greater resolution of the bands. Two new bands were found in the deep ultraviolet at 186 and 194 mμ, and two very weak bands at 265 and 285 mμ; no shift of the band with a maximum at 400 mμ was observed; in all other respects, the spectrum at 4°K is the same as in [3, 12].

Coloring at 77°K causes the appearance of a spectrum consisting of bands with maxima at 270, 320, and 550 mμ. It is assumed that the last two bands correspond to the bands at 335 and 580 mμ in the spectra of crystals colored at room temperature.

The absence, in the spectrum of crystals irradiated with electrons at 77°K, of bands corresponding to the bands at 400 and 225 mμ apparently indicates a different structure of the defects responsible for these bands as compared with the bands at 335 and 580 mμ.

The identity of the spectra of photochemically and additively colored crystals unambiguously determines the electronic nature of all four absorption bands with maxima at 225, 335, 400, and 580 mμ.

Undoubtedly, in the process of photochemical coloring, hole centers are also formed simultaneously with electron color centers.

Bands of a hole nature apparently lie in the far-ultraviolet region of the spectrum, which is inaccessible to measurement with ordinary quartz instruments. Some hole bands lying in the near-ultraviolet region of the spectrum, as has been found, are not stable at room temperature. Hayes and Twidell [26], in x-irradiated CaF₂ crystals at 77°K, found a band at 350 mµ. When the crystal is heated, the band is destroyed at 138°K.

The authors assign to this band an \(F_2^-\)-type center oriented along the [001] direction, the so-called \(V\)-centers.

V. A. Arkhangel’skaya and L. A. Alekseeva [27], upon irradiation with \(\gamma\)-rays at 77°K of CaF₂ crystals activated with the rare-earth ions Nd, Du, Ho, Er, Tu, found absorption at 315 mµ, which is destroyed at 300°K. The authors suppose that this band is due to intrinsic structural defects of CaF₂, since the position of the band depends neither on the type of activator nor on its concentration. Unfortunately, the authors did not carry out experiments on pure CaF₂ crystals; however, the possibility is not excluded of interpreting this band as belonging to self-trapped \(F_2^-\) holes. Cerro [28], in studying the EPR spectrum of natural CaF₂ crystals x-irradiated at 77°K, detected the signal of the \(V_k\)-center of Hayes and Twidell. At 138°K, simultaneously with the destruction of the \(V_k\)-centers, a new band appeared, which the author assigned to \(V_F\)-center complexes consisting of an \(F_2^-\) hole and a cation vacancy lying in the {110} plane. The absence of \(V_F\)-centers in synthetic CaF₂ crystals is explained by the author in the following way. In any CaF₂ crystal, because of the proximity of the radii, impurities of rare-earth ions are always present. However, compensation of the charge of a trivalent ion depends on the conditions of crystal growth. When grown from the melt, compensation of the excess charge of a trivalent ion is accomplished by interstitial \(F^-\) ions; when crystals are grown from solution (natural crystals), compensation is accomplished by cation vacancies, which also participate in the formation of \(V_F\)-centers. However, the author’s suggestion concerning different types of defects in natural and synthetic CaF₂ crystals apparently will not be confirmed, since the identical behavior of these crystals in the course of photochemical and additive coloration and in the course of thermal annealing does not support the proposed assumption.

Thus, it has not yet been possible to detect hole bands at room temperature. As indicated above, all four bands in the spectrum of colored CaF₂ crystals are of electronic nature. In order to clarify the models of the centers responsible for these bands, a number of authors [3, 11, 24, 29, 30] have used the following method: a definitely known type of defect is introduced into the crystal, and its influence on the character of the absorption spectrum is determined. It is known that when rare-earth ions are introduced into CaF₂, compensation of the excess positive charge is accomplished by \(F^-\) ions in interstitial sites, the concentration of which increases relative to the equilibrium one. According to Smakula [30], when a yttrium impurity is introduced, no additional bands arise in the spectrum of the irradiated crystal; only an increase of all four spectral bands is observed, and at high activator concentrations, in comparison with all the others, the band with a maximum at 400 mµ increases strongly. The supposition is made that fluorine atoms in interstitial sites are responsible for this band.

When a Na impurity is introduced into the crystals, an additional number of anion vacancies is introduced. The absorption spectrum of CaF₂—Na crystals irradiated with electrons consists of bands at 330 mµ, 385 mµ

and 605 mµ. It is assumed [11, 30] that these bands correspond to bands with maxima at 335, 400, and 580 mµ in pure CaF₂ crystals. The strong increase, upon introduction of Na, of the band at 605 mµ served as the basis for Avram [30] to attribute to this band and to the band at 580 mµ in the pure crystal a center of the type: an electron localized at an anion vacancy.

However, researchers’ opinions differ concerning the nature of the bands. A number of authors [24, 29, 31], on the basis of the facts cited below, conclude that all four bands in the spectrum of colored CaF₂ crystals are due to yttrium impurity: 1) pure CaF₂ crystals synthesized from especially pure salts do not become colored under the action of ionizing radiation; 2) introduction of yttrium impurity into an especially pure crystal causes the appearance of the spectrum of an irradiated crystal, consisting of four bands with maxima at 225, 335, 400, and 580 mµ; 3) in natural and synthesized CaF₂ crystals, as spectral analysis has shown, yttrium impurity is always present; 4) saturation of all four absorption bands with irradiation dose in x-rayed CaF₂ crystals indicates a finite concentration in the crystal of the defects responsible for these bands. The following mechanism for coloring CaF₂ crystals is proposed [29]. During irradiation, electrons and holes are formed. The localization centers for the electrons are trivalent yttrium ions. The four absorption bands are due, as shown by EPR, to four transitions of the electron on the divalent yttrium ion. The centers of localization of holes are not discussed in [29].

Attributing all four bands to a center of one nature agrees with Berrill’s data on the identical rate of destruction of all the bands during thermal destruction, but contradicts the following results: 1) the strong increase, upon introduction of yttrium, of only one band with a maximum at 400 mµ in [30]; 2) the absence of bands at 400 and 225 mµ in CaF₂ crystals irradiated at 77°K in [1]; 3) the anomalous behavior of the band at 580 mµ with dose of proton irradiation in comparison with the other three bands in [25].

Photochemical coloring of CaF₂ crystals prior to irradiation, containing a band in the ultraviolet region of the spectrum

In the preceding paragraph, the discussion concerned absorption spectra arising in the process of photochemical coloring in natural and synthetic CaF₂ crystals that, prior to irradiation, are transparent in the ultraviolet region of the spectrum. The interpretation proposed by various authors of the centers responsible for these bands was presented.

An entirely different picture is presented by the spectrum of a crystal which, prior to irradiation, has a band in the ultraviolet region of the spectrum, the maximum of which, according to various authors, ranges within 205–220 mµ. One of the presumed centers responsible for this band is the O⁻ ion, associated with an anion vacancy.

The spectrum of a photochemically colored crystal containing oxygen has the form shown in Fig. 4 and consists of bands with maxima at 320, 370, and 570 mµ [15], the position of the long-wavelength band being found at 540 mµ in [11] and at 550 mµ in [32]. In a number of works the weak band at 320 mµ was not observed. As is seen from comparison of Figs. 3 and 4, the spectra of additively colored crystals with subsequent rapid cooling and the spectrum of a photochemically colored crystal that, prior to irradiation, is not transparent in the ultraviolet are similar. From this similarity follows the electronic nature of at least the bands at 370 and 540 mµ.

It is of considerable interest to clarify the reason for the appearance of two types of spectra under photochemical coloration of CaF₂ crystals. As already mentioned above, many authors [1, 11, 16] see the reason for the appearance of the spectrum consisting of bands at 320, 370, and 540 mμ in the presence in the crystal of oxygen ions and the defects accompanying them.

The appearance of a similar spectrum under additive coloration was also observed in crystals that contained oxygen before coloration [16]. Bands with maxima at 225, 335, 400, and 580 mμ appear only upon coloration (photochemical and additive) of especially pure crystals [12], which before coloration do not contain oxygen [1, 11, 32]. However, there is also a completely opposite point of view.

Fig. 4. Absorption spectrum under irradiation of CaF₂ crystals that, before coloration, contained a band in the ultraviolet region of the spectrum [15]

A number of authors [13, 29, 24] attributed the four bands in the spectra of photochemically and additively colored crystals to the presence of an impurity in the crystals, whereas the spectrum consisting of the α- and β-bands is due to intrinsic defects of the crystal lattice.

There is also a third group of investigators [6, 19, 32], who believe that the appearance of both the spectrum consisting of four bands and the spectrum of α- and β-bands is due not to impurities but to defects introduced into the lattice during the coloration process. Moreover, the shift of the bands is explained differently by different authors. For example, Gessik [19] proposes to explain the shift of the α-band to 400 mμ in photochemically colored crystals by the influence of atomic calcium formed during irradiation [33].

Przibram, in his review [6], notes that under very prolonged X-ray and radium irradiation of natural CaF₂ crystals, a shift of the bands with maxima at 400 and 580 mμ to 380 and 540 mμ is observed. In pressed CaF₂ tablets, the long-wavelength maximum also lies at 540 mμ. Przibram sees the cause of such a shift in a large disturbance of the crystal lattice.

Thus, the question of the reasons for the appearance of two types of spectra during coloration of CaF₂ crystals and of the models of the centers responsible for the bands in spectra of both types still remains open.

Absorption spectra of synthesized CaF₂, SrF₂, BaF₂ crystals

The study of especially pure CaF₂, SrF₂, and BaF₂ crystals is of considerable interest, since it becomes possible to answer the question: is an impurity or are intrinsic defects of the crystal lattice responsible for the absorption bands in the spectra of these crystals? As for SrF₂ and BaF₂ crystals, they are not found in nature as natural minerals, and the only method for obtaining them is synthesis from the corresponding salts. However, the large amount of contradictory experimental material currently available in the literature does not make it possible to draw definite conclusions either about changes in the spectra depending on the method and conditions of coloration, or, still less, about the nature of the defects introduced into the crystals.

Table 1

Positions of absorption bands in the spectra of synthesized crystals of CaF₂, SrF₂, and BaF₂, colored by various methods (spectra measured at room temperature)

Note: * irradiation was carried out at room temperature; measurements at −225°C.

during coloration. Table 1 presents the positions of bands in the spectra of colored synthesized crystals of CaF₂, SrF₂, and BaF₂ according to data from several authors.

1. CaF₂. From the results given in Table 1 it follows that CaF₂ crystals have different absorption spectra depending on the type of irradiation and the method of coloration, according to different authors. Common to these spectra is the presence of a band with a maximum at 380 mµ (with the exception of the spectrum of a crystal irradiated with deuterons). The appearance of this band in photochemically and additively colored crystals unambiguously determines the electronic nature of the centers responsible for this band. The reasons for the appearance of the different types of spectra and the models of the centers responsible for the other bands of the spectra have not been clarified.

2. SrF₂. After growth, SrF₂ crystals are in some cases obtained transparent, with a band at 206 mµ, which indicates the presence of oxygen in the crystal [8, 36]; they may be green or yellow in color. The cause of the coloration has not been clarified. Depending on the initial sample, the spectra of colored crystals differ. In additively colored crystals the strongest bands in the spectrum are the bands at 631 mµ and 446.4 mµ. As Karras notes [9], these bands are similar to the F band of alkali-halide crystals: as the temperature is lowered the bands shift into the short-wavelength region of the spectrum, and irradiation of the crystals with light in these bands leads to their decrease.

3. BaF₂. Many works note the extremely weak sensitivity of BaF₂ crystals to photochemical coloration.

Additive coloration produces significant changes in the spectrum. Study of the EPR spectrum of additively colored crystals allowed Arends [18] to suggest that the EPR signal is due to electrons localized in anion vacancies. A band at 660 mµ is attributed to these centers; however, no comparison of the EPR spectra and optical spectra was made.

REFERENCES

1. D. Messner, A. Smakula. Phys. Rev., 120, 1162 (1960).
2. R. W. Ure. J. Chem. Phys., 26, 1363 (1957).
3. A. D. Franklin. J. Phys. Chem. Sol., 26, 933 (1965).
4. E. Zintl, A. Udgard. Z. anord u allgem. Chem., 240, 150 (1939).
5. I. Short, R. Roy. J. Phys. Chem., 67, 1860 (1963).
6. K. Przibram. Z. Physik, 154, 111 (1959).
7. W. Bontinck. Physica, 24, 650 (1958).
8. H. Karras. Phys. Stat. Sol., Б. 1. H. 1 (1961).
9. H. Karras. Phys. Stat. Sol., 1, 160 (1961).
10. F. K. Fong, R. N. Yocom. J. Chem. Phys., 41, 1383 (1964).
11. I. H. Schulman, R. J. Ginther, R. D. Kirk. J. Chem. Phys., 20, 1966 (1952).
12. A. Smakula. Z. Physik, 138, 276 (1954).
13. F. Liity. Z. Physik, 134, 596 (1953).
14. W. Bontinck. Physica, 24, 639 (1958).
15. L. P. Feofilov. Dokl. Akad. Nauk SSSR, KhSP, No. 3 (1963).
16. L. V. Krotova, V. V. Osiko, V. T. Udovenchik. FTT, 7, 238 (1965).
17. R. Görlich, H. Karras, K. Lehmann. Phys. Stat. Sol., 3, 98 (1963).
18. I. Arends. Phys. Stat. Sol., 7, 805 (1964).
19. B. R. Görlick. Physica, 26, 378 (1956).
20. S. Barile. J. Chem. Phys., 20, 297 (1952).
21. A. Smakula. Phys. Rev., 91, 1970 (1953).
22. A. Smakula. Phys. Rev., 77, 408 (1950).
23. E. K. Zavadovskaya, N. M. Timoshenko, V. A. Chernyshev, L. A. Litsyna. Izvestiya TPI, 19, 225 (1965).
24. P. Görlich, H. Karras, W. Lüdke, H. Mothes, R. Reimann. Phys. Stat. Sol., 3, 478 (1963).

1. L. A. Lisitsyna. Optics and Spectroscopy, 19, 295 (1965).
2. W. Hayes, I. W. Twidell. Proc. Phys. Soc., 79, 1295 (1962).
3. V. A. Arkhangel’skaya, A. A. Alekseeva. Report at the XIV Conference on Luminescence.
4. I. Sierro. Phys. Rev., 138, No. 2A, A648 (1965).
5. I. R. O’Connor, I. H. Chen. Phys. Rev., 130, 1790 (1963).
6. W. J. Scouler, A. Smakula. Phys. Rev., 120, 1154 (1960).
7. I. R. O’Connor, I. H. Chen. J. Phys. Chem. Soc., 24, 1382 (1963).
8. P. Feltham, J. Andrews. Phys. Stat. Sol., 10, 203 (1965).
9. W. Bontinck, W. D. Dekeyser. Physica, 22, 595 (1956).
10. P. Görlich, H. Karras. Proceedings of the International Conference on Semiconductors. Prague. (1960).
11. E. G. Chernevskaya. Optics and Spectroscopy, 11, No. 4 (1961).
12. H. Bruch, P. Görlich, H. Karras, R. Lehmann. Phys. Stat. Sol., 4, 685 (1964).