On the Pre-Breakdown Glow of Alkali Halide Crystals

R. V. Gusarova

(Presented by the scientific seminar of the Department of Experimental Physics)

Electroluminescence is the name given to luminescence in which the luminous body draws its energy from an electric field. Good electroluminophores are compounds of the ZnS type, activated mainly by Cu and belonging to the class of semiconductors. These compounds have a comparatively small forbidden-band width, possess acceptor and donor levels, and in their conduction band there are always free electrons. To explain the phenomenon of electroluminescence, three mechanisms are invoked: the tunnel effect, carrier injection, and the mechanism of impact ionization, which in most cases is regarded as the most probable [1].

According to the theoretical concepts of Piper and Williams, in order for electroluminescence to be produced by inelastic collisions between fast electrons and the activator system, at least three conditions must be satisfied. A region of strong field must be created, in which charge carriers can be accelerated to large values of kinetic energy; electrons or holes must be introduced into this region of strong field; finally, the activator centers must be arranged so that they can absorb a considerable fraction of the energy acquired from the field by the charge carriers. Since the named mechanism of excitation of electroluminescence can operate at a field strength somewhat lower than the breakdown strength, such field configurations are of great importance in which the stability of the crystal with respect to breakdown is increased. For example, a narrow region of strong field connected in series with a wide region of weak field permits this mechanism to act over a larger interval of applied voltages without creating unstable breakdown conditions. These conditions are realized if the crystal contains a barrier region with high resistance. Such barrier layers for ZnS crystals may be depletion barriers of the Mott–Schottky type or a \(p\)—\(n\) junction operating in the blocking direction.

However, the impact-ionization mechanism, like other mechanisms of electroluminescence, has not yet been sufficiently developed, and there is still no unified and coherent theory of this phenomenon. It is known that alkali halide crystals are used very successfully as objects for studying breakdown processes and certain fundamental problems.

photo- and X-ray luminescence of crystalline phosphors. Therefore, it is of great interest to use this class of compounds also for the study of electroluminescence phenomena. However, for a number of reasons it is more difficult to excite electroluminescence in crystals with a large fraction of ionic bonds than in crystals with covalent bonding. Ionic crystals have a wider forbidden band. The mean free path of electrons and holes before scattering by thermal vibrations of the lattice in ionic crystals is several tens of times smaller than in covalent ones.

In crystals with covalent bonding, regions that concentrate electric fields are readily formed ($p$—$n$ junctions, depletion barriers, etc.); in crystals with ionic bonding, because of the small number of free current carriers, this, as a rule, does not occur [2]. All these difficulties have made it impossible to obtain electroluminescence in alkali-halide crystals. So far only one report is known, by Georgobiani and Golubeva, who obtained electroluminescence in a sublimed CsJ—Tl film 1 $\mu$ thick in an alternating field of strength

$$ E_{10} - 2 \cdot 10^{5}\ \frac{\mathrm{V}}{\mathrm{cm}} \ [3]. $$

Experimental data obtained at the Tomsk Polytechnic Institute in recent years make it possible to assert that electrical breakdown of alkali-halide crystals begins with impact ionization and that strengthening of thin dielectric layers is possible, since at small distances appreciable electron avalanches cannot develop [4, 5]. Thus, in the case of thin dielectric layers in fields close to breakdown fields, electron avalanches arise which do not yet lead the dielectric to breakdown, but which, along with the tunnel effect, may lead to electroluminescence of the crystal. Detection of these electron avalanches is of interest both from the point of view of the theory of breakdown of solid dielectrics and from the point of view of the theory of electroluminescence.

We undertook an attempt to detect the glow of thin layers of alkali-halide crystals in pre-breakdown fields.

For the investigations we used the method of counter-drilling [6], which makes it possible to obtain thin layers of a single crystal (100–10 $\mu$). After a whole series of preliminary experiments it was found that the glow is observed when an alternating voltage is applied to the crystal, which apparently can be explained as follows. It is known that luminescence in alkali-halide crystals is of a recombination character. The conditions necessary for recombination of electrons and luminescence centers can most probably be realized in an alternating field. In addition, an alternating field promotes the production of pre-breakdown avalanches (states) that do not bring the dielectric to breakdown.

It turned out that the most suitable object of investigation is KJ—Cu, which has the smallest forbidden-band width, the largest fraction of covalent bonding among alkali-halide crystals, and gives good photoluminescence. Single crystals of KBr—Cu and KCl—Cu were also investigated.

Fig. 1 presents a photograph of the glow of KJ—Cu, obtained with the aid of a microphoto attachment. The glow, in the form of separate spots, appears at the electrodes at an average field strength of $\sim 2 \cdot 10^{5}\ \frac{\mathrm{V}}{\mathrm{cm}}$. Between these spots a glow arises in the form of separate filaments, connecting-

these luminous spots, if the distance between them is sufficiently small. When the voltage is increased, or even at a given voltage but after some time has elapsed following the application of the voltage, breakdown occurs. It has been established that breakdown begins from these luminous spots. In some cases the glow is observed not at the electrodes, but in the volume (almost in the middle) of the crystal.

Fig. 1. Microphotograph of the glow of a KJ—Cu crystal

For KBr—Cu an analogous glow is observed, but it is less bright, and it begins at a higher field strength \((E \sim 3.4 \cdot 10^5\ \frac{V}{cm})\); in KCl—Cu only a very weak glow is observed, in the form of

Fig. 2. Brightness waves for the glow of a KJ—Cu crystal under excitation by an alternating electric field with \(v = 50\ \text{Hz}\)

spots, and breakdown quickly occurs after the appearance of the luminous spots.

The field strength at which the glow occurs is \(E_{\mathrm{cr}} \sim 6 \cdot 10^5\ \frac{V}{cm}\).

Figure 2 shows brightness waves for KJ—Cu. It is seen from the figure that one half-period corresponds to one peak of luminescence, and the luminescence arises some time after the application of the voltage half-period and leads the voltage in phase. There is no constant component. The luminescence exists only during one half-period, which indicates the sensitivity of the mechanism responsible for excitation of the luminescence to the polarity of the applied field. It turned out that the region of the crystal capable of being excited luminesces only when the corresponding electrode is charged negatively and, consequently, can emit electrons.

Thus, the observed luminescence is associated with pre-breakdown processes of ionization by electrons and apparently represents electroluminescence.

It should be assumed that the regions of local luminescence are sites of concentration of high fields caused by surface defects.

REFERENCES

1. W. W. Piper and Williams. Phys. Rev., 98, 1809—1813, 1955.
2. A. N. Georgobiani. Opt. i spektr., 22, no. 6, 1962.
3. A. N. Georgobiani and N. P. Golubeva. Opt. i spektr., 12, 802, 1962.
4. V. A. Kostrygin. FTT, 11, no. 8, 425, 1960.
5. A. A. Vorob’ev, G. A. Vorob’ev, V. A. Kostrygin. Izv. vuzov, Fizika, 6, 166, 1960.
6. V. A. Kostrygin. Physics of alkali-halide crystals. Proceedings of the Second All-Union Conference. 1961. Latvian State University. Riga, 1962, p. 365.