DIELECTRIC LOSSES OF ZnS—Cu AND ZnS—Cu, Fe PHOSPHORS

F. I. Vertunas and O. V. Agashkin

In zinc sulfide phosphors, under the action of ultraviolet light, ionization of the emission centers occurs. Some of the electrons become localized at trapping levels. Their thermal release and subsequent recombination with ionized centers give rise to afterglow. The probability of thermal release of localized electrons \(p\) is equal to \(p_0 e^{-\varepsilon/kT}\), where \(\varepsilon\) is the depth of the localization levels, and \(p_0\) is the frequency of attempts at releasing electrons from them.

Studying the thermoluminescence curves of ZnS—Cu phosphors, Randall [1] found \(p_0 = 10^{8\pm1}\), and without special proof assumed that for all zinc sulfide phosphors, irrespective of the kind of localization levels, \(p_0\) has the same value.

V. V. Nuvar’eva [2] showed in our laboratory that in one and the same phosphor \(p_0\) has different values for different levels. Thus, for a ZnS—Cu,Fe phosphor, as a result of analysis of its decay curves, two values of \(p_0\) were found—\(1.2 \cdot 10^6\) and \(1.5 \cdot 10^4\). Such small values, moreover different for different levels, were found for the first time. It was therefore of interest to determine these frequencies by another independent method. As the latter, we considered it possible to use the frequency dependences of \(\operatorname{tg}\delta\) (the tangent of the dielectric loss angle) of an excited phosphor. In [3] it was shown that electrons localized at sticking levels cause losses of a relaxation character. This was indicated by the linear dependence of

\[ \frac{\varepsilon_0 - \varepsilon}{\varepsilon - \varepsilon_\infty} \]

on \(\omega^2\), observed by the authors. On this basis it could be expected that, for losses caused by localized electrons, the dependence of \(\operatorname{tg}\delta\) on \(f\) would have a maximum at a frequency \(f\) equal to \(p_0\), and that each kind of level, characterized by a definite value of \(p_0\), would correspond to its own maximum.

In the present work the frequency dependences of two phosphors were investigated—ZnS—Cu (Cu—\(10^4\) g/g, firing temperature in the preparation process \(1200^\circ\)C) and ZnS—Cu,Fe (Cu—\(10^{-4}\) g/g, Fe—\(10^{-5}\) g/g).

Measurement method and its accuracy

The frequency dependence of \(\operatorname{tg}\delta\) was recorded on a \(Q\)-meter. The capacitor with the phosphor had the following construction. Onto a disk 6 cm in diameter, covered with aluminum foil, a phosphor layer \(18\text{–}25\ \mathrm{mg/cm^2}\) thick was deposited from ethyl alcohol. The foil served as one of the electrodes; the second electrode was a wire mesh, of wire diameter 0.1 mm, containing 575 openings per \(1\ \mathrm{cm^2}\) and transmitting 40% of the light incident on it.

light. To reduce the inhomogeneity of the field produced by the grid, a layer of mica (first-grade muscovite) 0.05 mm thick was placed between it and the phosphor. The mica weakened the excited light by another 20%. The electrodes were inserted into quartz holders, which made it possible to press the grid and the mica tightly against the phosphor. The capacitor was placed in front of a PRK-4 lamp with a Uviol filter. The lamp burning regime was kept constant. Between the lamp and the capacitor there were a photographic shutter and a lens, ensuring uniform illumination of the phosphor surface. The small thickness of the sample, approximately 0.1 mm, made it possible to excite it uniformly throughout its volume.

Fig. 1. Frequency dependence of \(\operatorname{tg}\delta\) of dolomite glass.

Fig. 2. Frequency dependence of \(\operatorname{tg}\delta\) of the equivalent circuit.

Our apparatus and measurement method contained two sources of error. The first source was the \(Q\)-meter itself, the second—the layered structure of the insulator (phosphor, mica). The errors of the \(Q\)-meter were determined in two ways: frequency dependences of \(\operatorname{tg}\delta\) were taken on it for glass and for our capacitor without phosphor, with a noninductive resistance connected in series. The surfaces of the glass, as well as of the phosphor, were not silvered. The parameters of the equivalent circuit were chosen so that the losses in it were approximately equal to the losses in the excited phosphor. All measurements were carried out at a constant value of the capacitance \(C_1 = 150\) cm. The frequency was varied by changing coils. Figure 1 gives the dependence of \(\operatorname{tg}\delta\) on \(f\) for dolomite glass, for which, by the resonance method in the electrophysics laboratory, a weak monotonic decrease of \(\operatorname{tg}\delta\) with frequency had been found; in Fig. 2 the theoretical curve \(\operatorname{tg}\delta = RC\omega\) for the equivalent circuit is shown, and the experimental points are plotted on it. From analysis of these curves it follows that the \(Q\)-meter, in the frequency intervals 49–78 kHz and \(1.2 \cdot 10^5\)–\(2 \cdot 10^5\) Hz, gives underestimated values of \(\operatorname{tg}\delta\). The deviation from the theoretical dependence at 50 Hz is 24%, and at 148 Hz, 19%. In the interval \(2 \cdot 10^5\)–\(4 \cdot 10^6\) Hz the \(Q\)-meter gives no systematic errors; its maximum relative error in this interval is 13%.

Subsequently, when measuring losses in phosphors, we made corrections in accordance with the curve in Fig. 2. It should be noted that they never changed the character of the frequency dependences.

The capacitance of the glass and of the equivalent circuit had no frequency variation up to \(2 \cdot 10^6\) Hz.

The errors due to the layered nature of the insulator were determined as follows. We considered the mica and the phosphor as a two-layer insulator, since under the conditions of our experiment the sample was excited uniformly throughout the entire volume. In work [4], nonuniformly excited thick

the samples were considered as multilayer insulators. \(\operatorname{tg}\delta\) of a two-layer insulator is equal to:

\[ \operatorname{tg}\delta = \frac{C_c \operatorname{tg}\delta_\phi + C_\phi \operatorname{tg}\delta_c}{C_c + C_\phi} = \frac{\dfrac{C_c}{C_\phi}\operatorname{tg}\delta_\phi + \operatorname{tg}\delta_c}{\dfrac{C_c}{C_\phi}+1}, \tag{1} \]

\[ \operatorname{tg}\delta_c = 5 \cdot 10^{-5} \qquad \varepsilon_c = 7;\quad \varepsilon_\phi = 10;\quad d_c = 005 \]

\[ d_\phi = 0.1\ \text{mm};\quad \frac{C_c}{C_\phi}=1.4 \]

\[ \operatorname{tg}\delta = \operatorname{tg}\delta_\phi \cdot 0.58 + 2 \cdot 10^{-5} \]

\[ \operatorname{tg}\delta = 0.58 \operatorname{tg}\delta_\phi . \tag{2} \]

It follows from formula (2) that for our capacitor the \(\operatorname{tg}\delta\) of the system, at least as to order of magnitude, coincides with the \(\operatorname{tg}\delta\) of the phosphor. When the frequency is changed, the ratio of \(\operatorname{tg}\delta\) to \(\operatorname{tg}\delta_\phi\) changes owing to a change in \(C_\phi\). In the case of an excited ZnS—Cu phosphor, the capacitance of the capacitor in the interval \(5 \cdot 10^5\)—\(4 \cdot 10^6\) cycles changes by 3%. Calculation shows that \(C_\phi\) then changes by 6%. From the same formula (1) it follows that if \(C_\phi\) changed even by 30%, the ratio of \(\operatorname{tg}\delta\) to \(\operatorname{tg}\delta_\phi\) would change only by 11%. Thus, owing to the weak dependence of \(C_\phi\) on \(f\) in the case of our phosphors, the frequency dependences of \(\operatorname{tg}\delta\) of the system and \(\operatorname{tg}\delta\) of the phosphor practically coincide. Consequently, we determine the absolute values of \(\operatorname{tg}\delta\) of the phosphor with a large error, but the character of the frequency dependence rather accurately.

Results obtained

1. ZnS—Cu phosphor. The \(\operatorname{tg}\delta\) and the increment of the capacitance of the capacitor upon excitation of the phosphor were measured in the frequency interval \(5 \cdot 10^4\)—\(6 \cdot 10^6\) cycles. The losses of the capacitor without phosphor and with phosphor, but without illumination, in the indicated frequency interval were below the sensitivity limit of the \(Q\)-meter, i.e., less than \(10^{-4}\). The capacitance of the capacitor was equal to 133 cm and did not vary with frequency. Upon excitation of the sample, its capacitance at 50 cycles increased to 138 cm. The change in capacitance \(\Delta C\) decreased monotonically with increasing frequency—at \(f = 10^6\) cycles, \(C = 134\) cm. The losses upon excitation increased to \(10^{-3}\). The frequency dependence of \(\operatorname{tg}\delta\) is shown in Fig. 3. It is seen that the curve has 4 weakly expressed maxima at \(f = 8 \cdot 10^4\), \(f = 2 \cdot 10^5\), \(f = 3.5 \cdot 10^5\), and \(1.2 \cdot 10^6\) cycles. The heights of the maxima are not within the limits of experimental error. The positions of the maxima did not change either with a change in the thickness of the sample (18 and 25 mg/cm\(^2\)), or with a change in the voltage across it (dependences were taken for the phosphor with mica and without mica).

After the exciting light was switched off, the losses of the phosphor decreased, and more slowly than the afterglow brightness. Only after 8 hours of bleaching did the losses fall below the sensitivity limit of the \(Q\)-meter. Infrared light accelerated the decay both after illumination and of \(\operatorname{tg}\delta\).

1. ZnS—Cu, Fe phosphor. In this phosphor, in contrast to ZnS—Cu, the dark losses were appreciable. The frequency behavior of \(\operatorname{tg}\delta\) is shown in Fig. 4 (curve 1). The frequencies of all three maxima of ZnS—Cu, Fe practically coincided with the corresponding maxima of the ZnS—Cu phosphor. Upon excitation of ZnS—Cu, Fe, the losses in the region of these maxima decreased (curve 2), and a new maximum appeared at \(f = 1.4 \cdot 10^6\) cycles, corresponding

to a maximum of \(1.2 \cdot 10^6\) Hz for the ZnS—Cu phosphor. When the exciting light was switched off, the losses in the region of the first three maxima slowly increased, while in the region of the fourth they decreased. When the excitation was switched on, the opposite picture was observed. The positions of the maxima and the frequency corresponding to the intersection

Fig. 3. Frequency dependence of \(\operatorname{tg}\delta\) of an excited ZnS—Cu phosphor.

of curves 1 and 2 did not change either with a change in the thickness of the sample (18 and 25 kg/cm\(^2\)) or with a change in the voltage applied to it. (Dependences were recorded for a capacitor with mica of 0.04 and 0.08 mm, without mica, and with air gaps of different thickness.)

Fig. 4. Frequency dependence of a ZnS—Cu,Fe phosphor.
1—in the dark; 2—under illumination.

Discussion of the Results

The maxima in the frequency dependence of ZnS—Cu give us grounds to associate the losses not with conductivity, but with electrons localized at sticking levels. From this point of view, the four maxima for \(\operatorname{tg}\delta\) mean that in the ZnS—Cu phosphor there are four kinds of localization levels. From what follows it will be seen that the behavior of the first three and of the fourth maximum is different; therefore we have assumed that in the ZnS—Cu phosphor there are two kinds of localization levels. Levels \(A\), consisting of three components, correspond to the maxima \(8 \cdot 10^4\), \(2 \cdot 10^5\), \(3.5 \cdot 10^5\) hertz, and levels \(B\), to which the maximum \(1.2 \cdot 10^6\) Hz corresponds.

In the unexcited ZnS—Cu phosphor the levels \(A\) and \(B\) are empty; therefore its dark losses are practically equal to zero. When the ultraviolet is switched on, the levels are gradually filled and the losses increase; after it is switched off they are emptied and the losses correspondingly decrease.

In the ZnS—Cu, Fe phosphor, in contrast to ZnS—Cu, the maxima corresponding to the levels \(A\) are retained even in the absence of excitation; consequently, the levels \(A\) in this phosphor are filled with electrons even in the unexcited state. The levels \(B\), however, remain empty, since the corresponding maximum is absent in the unexcited state. Electrons from the levels \(A\) are not thermally released, at least at room temperature, but they can be released under the action of the exciting light. The bleaching action of exciting light for a number of phosphors was proved as a result of optical investigations in [5]. Thus, when excitation is switched on in the ZnS—Cu, Fe phosphor, not only does ionization of the emission centers occur, but electrons are also released from the levels \(A\). The conduction electrons thus obtained populate the localization levels \(B\). Therefore the maxima caused by the levels \(A\) decrease, and an additional maximum appears, caused by the levels \(B\). As a consequence of this, the frequency dependences of \(\operatorname{tg}\delta\) for the unexcited and excited phosphor intersect. If our explanation is correct, then with an increase in the intensity of the exciting light \(E\), the negative change of \(\operatorname{tg}\delta\) at low frequencies and the positive change at high frequencies should increase. Moreover, \(-\Delta \operatorname{tg}\delta\), caused by the bleaching action of the exciting light, should vary with \(E\) according to a linear law, while \(+\Delta \operatorname{tg}\delta\), with increasing \(E\), should tend toward saturation, since \(+\Delta \operatorname{tg}\delta\) is due to the light sum accumulated on the localization levels, which with increasing \(E\) tends toward saturation. In order to verify this proposition we measured the dependences of \(\operatorname{tg}\delta\) on \(E\) at frequencies \(2.3 \cdot 10^5\) and \(2.34 \cdot 10^6\) hertz. The results obtained are shown in Fig. 5. It is seen from Fig. 5 that, in accordance with expectations, \(-\Delta \operatorname{tg}\delta\) increases with \(E\) according to a linear law, while \(+\Delta \operatorname{tg}\delta\) tends toward saturation with increasing \(E\). Consequently, the decrease of \(\operatorname{tg}\delta\) under illumination can be explained by the bleaching action of the exciting light.

Fig. 5. Dependence on the intensity of the exciting light of \(+\) and \(-\Delta \operatorname{tg}\delta\) of a ZnS—Cu, Fe phosphor.

Thus, from our measurements it follows that in ZnS—Cu and ZnS—Cu,Fe phosphors there are two kinds of localization levels—a three-component level \(A\) and a one-component level \(B\). The levels \(B\) behave identically in both phosphors: they are empty in unexcited specimens and are populated with electrons upon excitation. The levels \(A\) in these same phosphors behave differently. In the absence of excitation in the ZnS—Cu phosphor they are empty, while in ZnS—Cu, Fe they are “populated” with electrons. When the exciting light is switched on in the ZnS—Cu phosphor these levels are populated with electrons, whereas in ZnS—Cu, Fe electrons are released from them.

The attempt frequencies \(p_0\), equal to the frequencies of the \(\operatorname{tg}\delta\) maxima, are different for different localization levels of one and the same phosphor; moreover \(p_0\).

may be considerably less than \(10^8\). The frequency \(p_0 = 1.2 \cdot 10^6\), for the ZnS—Cu phosphor, is in agreement with the value \(p_0 — 1.2 \cdot 10^6\), found for the ZnS—Cu phosphor in [3] from the slope of the straight line in the coordinates

\[ \frac{\varepsilon_0 - \varepsilon}{\varepsilon - \varepsilon_\infty}, \ \omega^2. \]

In conclusion, the authors express their gratitude to G. I. Skanavi, K. A. Vodop’yanov, and V. N. Kessenikh for a valuable discussion.

LITERATURE

1. L. T. Randall, Proc. Roy Soc. 184, 365, 1945.
2. V. V. Nuvar’yeva. Dissertation, 1955.
3. G. F. Garlick and A. F. Gibson Proc. Roy. Soc. 188, 485, 1947.
4. J. H. Gisolf, Physica 6, 918, 1939.
5. V. V. Antonov-Romanovskii, Izv. AN SSSR, ser. fizich. 13, 91, 1949.

Siberian Physico-Technical
Institute

Questions and Answers

V. I. Odelevskii.—What does the frequency dependence of \(\varepsilon\) look like?

F. I. Vergunas.—In the region of the maximum, \(\varepsilon\) decreases.

V. I. Odelevskii.—How do you picture the resonance mechanism of the losses?

F. I. Vergunas.—Such a mechanism is only assumed to be possible for radio frequencies. In actual fact, it has been observed only in the region of centimeter waves. In this case, \(\varepsilon\) in the absorption region undergoes normal and anomalous dispersion.

M. P. Tonkonogov.—How strongly did the capacitance change as a result of illumination?

F. I. Vergunas.—The maximum change in capacitance due to illumination was 3%.

P. F. Veselovskii.—How were measurements carried out at frequencies above 1 MHz?

F. I. Vergunas.—For this, the same Q-meter was used. The good reproducibility of the results makes it possible to consider that, at least up to 10 MHz, the measurement data can be obtained on the Q-meter with an accuracy of not less than 5%.

V. N. Kessenikh.—Was the frequency behavior checked under other conditions? Was the capacitance of the specimen and the intensity of the exciting light varied?

F. I. Vergunas.—No, only the temperature and the field strength were varied. We always tried to increase the capacitance, since the larger the surface of the specimen, the greater the effect produced by illumination.

V. N. Kessenikh.—The shape of the electrodes can have a substantial effect on the results, since the properties of the contact layer can depend strongly on the frequency.

F. I. Vergunas.—We were interested only in the effect caused by illumination; since the contact layer remained exactly the same both under illumination and without illumination, it could not have affected the difference.

A. A. Vorob’ev.—Is it possible that the skin effect in the powder of the specimen has an influence?

F. I. Vergunas.—In that case, for all powders we would obtain the same picture. In our case, however, the picture changes for different phosphors, while for one and the same phosphor it is well reproduced.

A. A. Vorob’ev.—At such high voltages, taking into account the properties of ultraviolet light, ionization could have arisen under illumination. Then the differences between the losses in the dark and in the light would be explained by ionization.

F. I. Vergunas.—For ionization it is necessary that the total potential difference be no less than the ionization potential. We did not have this.

V. N. Kessenikh.—At low frequencies, illumination by ultraviolet light, producing a region of increased conductivity, will lead to an apparent decrease in the loss angle and an increase in capacitance. In the region of high frequencies, the increased conductivity will appear in the depth of the phosphor, and the observed effect will be the reverse.

F. I. Vergunas.—From this point of view, all phosphors should, under illumination, give increased conductivity and, consequently, one and the same effect. In our case, however, the opposite was obtained.