PROCEEDINGS
OF THE TOMSK ORDER OF THE RED BANNER OF LABOR POLYTECHNIC
INSTITUTE named after S. M. KIROV

Volume 184 1970

RADICAL-RECOMBINATION LUMINESCENCE OF SOLIDS. II. FEATURES OF RADICAL-RECOMBINATION LUMINESCENCE OF CERTAIN SULFIDES AND OXIDES

V. V. Stygov, V. A. Sokolov
(Submitted by Prof. V. A. Sokolov)

The known classification of luminescence according to types of excitation is not purely formal. Each type of luminescence, united by a common mode of energy input, reveals certain features of emission that indicate a difference in mechanism. In the present work, certain features of RRL of zinc oxide and zinc sulfide are reported, manifested in the luminescence spectra, in the temperature dependence of the luminescence intensity, and in the yield. ZnO and ZnS are good candoluminophors [1].

Spectra

RRL in our experiments was excited by hydrogen and oxygen atoms at low pressure in a vacuum apparatus described in detail earlier [2]. The concentration of atoms near the sample, measured by the method of [3], was \(10^{13} \div 10^{14}\ \mathrm{cm}^{-3}\). The RRL spectra were recorded automatically on an EPP 09. Usually the sample was trained, respectively, in an atmosphere of hydrogen or oxygen at a pressure of the order of \(10^{-2}\) torr for 30 minutes at a temperature of \(437^\circ\mathrm{K}\). After such treatment, when the discharge was switched on, bright luminescence gradually flared up at a pressure of about \(10^{-1}\) torr. The following samples were studied: ZnO, self-activated, brightly photoluminescent with a long afterglow; activator-free ZnS, prepared by calcining zinc sulfide of the “for luminophors” grade in an evacuated quartz ampoule at a temperature of \(1273^\circ\mathrm{K}\); samples ZnS—Sm \((10^{-3}\ \mathrm{g/g})\), ZnS—Tu \((10^{-2}\ \mathrm{g/g})\), and ZnS—Eu \((10^{-5}\ \mathrm{g/g})\), the first two prepared by calcination in a nitrogen atmosphere at \(1423^\circ\mathrm{K}\), and the last in an \(\mathrm{H_2S}\) atmosphere at the same temperature.

The main result of measuring the spectra of the above-listed phosphors is as follows: the RRL and photoluminescence spectra for the phosphors studied are not identical. In RRL spectra excited by hydrogen atoms, new bands may appear that are absent from the photoluminescence spectrum. Thus, for zinc oxide, Fig. 1, along with a band near \(510\ \mathrm{nm}\), usually attributed to excess-stoichiometric zinc and also observed in photoluminescence, we observed an additional band near \(470\ \mathrm{nm}\), of somewhat lower intensity, absent in the photoluminescence of the same sample. In the case of zinc sulfide the additional band lies

about 480 nm. Figure 2 shows the RRL spectrum under the action of hydrogen atoms for a ZnS sample prepared without an activator. The band at about 520 nm, common with the photoluminescence spectrum, is evidently due to traces of copper. We observed the same feature of the RRL spectra in zinc sulfide with activators.

Fig. 1. RRL spectra (solid curve) and photoluminescence (dashed curve) of zinc oxide

Figure 3 gives the RRL spectra, excited by hydrogen atoms at room temperature, of ZnS—Tu, ZnS—Eu, and ZnS—Sm samples in the region 400–600 nm. For comparison, for each sample the spectrum of its photoluminescence is shown by a dashed line. It is seen from Fig. 3 that in all cases a band appears in the RRL spectrum in the region 480–490 nm, which is absent in photoluminescence. It is completely resolved for the ZnS—Sm sample, since Sm itself gives emission in the orange-red region. Along with this band, each spectrum also contains the usual band, almost coinciding with the photoluminescence band (there is a small short-wavelength shift), belonging to the rare-earth ion [4]. These bands are rather narrow in the case of Sm and Tu, and relatively broad in the case of Eu.

It should be noted that the appearance and intensity of the RRL spectra are affected by adsorption on the phosphor surface of foreign gases and vapors. Oxygen adsorption, even at room temperature, strongly quenches luminescence excited by the action of hydrogen atoms. The effect of oxygen quenching, although to a lesser degree, is also well known for photoluminescence [5]. Heating the samples in a hydrogen atmosphere, on the contrary, promotes the ignition of bright RRL.

Fig. 2. RRL spectra (solid curve) and photoluminescence (dashed curve) of zinc sulfide

Adsorption of water at room temperature on a ZnS—Sm sample causes weakening and broadening of the band near 480 nm, while having relatively little effect on the samarium band. On a ZnS—Tu sample, after water adsorption, we observed the appearance in the RRL spectrum of yet another new band near 440 nm. Interestingly, in this region some authors observed the appearance of photoluminescence after wetting with water solids that previously had not luminesced [6]. The luminescence was ascribed to adsorbed water molecules or, possibly, to OH groups.

The 480 nm band, as was found, is more sensitive to surface conditions than the activator band. To elucidate the nature of this band, additional investigations are required. However, it can already be said that the luminescence centers responsible for its appearance are apparently located on the surface of the phosphor and are associated with the host substance. This is indicated, on the one hand, by the great sensitivity of this band to surface treatment and, on the other hand, by its absence from the photoluminescence spectrum when mainly the bulk of the substance is excited. In this respect, excitation of luminescence as a result of surface recombination of atoms, being a surface excitation, may be an effective method for detecting various kinds of surface centers. RRL, in this connection, may serve as a method for checking the correctness of certain hypotheses that attribute particular photoluminescence bands to surface centers.

Fig. 3. RRL spectra (solid curves) and photoluminescence spectra (dashed curves) of samples:
a) ZnS—Tu; b) ZnS—Eu; c) ZnS—Sm

Fig. 4. RRL and photoluminescence spectra of a ZnSCdS—Cu sample:
1 — upon excitation by hydrogen atoms, 2 — upon excitation by oxygen atoms, 3 — upon excitation by light

Upon excitation of the same phosphors by oxygen atoms, relatively weaker luminescence arises. In this case the maximum near 480 nm, characteristic of RRL under the action of hydrogen atoms, does not appear in ZnS-based phosphors. Figure 4 shows, for comparison, the spectra of RRL excited by oxygen and hydrogen atoms and the photoluminescence spectrum of a standard ZnSCdS—Cu luminophor. It is seen from the figure that the RRL spectrum under the action of oxygen atoms practically coincides for this sample with the photoluminescence spectrum, whereas in the RRL spectrum under the action of hydrogen atoms there is a weak maximum in the region of 480 nm. On the ZnS—Sm sample we observa-

... an interesting transformation of the RRL spectrum upon going from excitation by hydrogen atoms to excitation by oxygen atoms, shown in Fig. 5. Curve 1 represents the RRL spectrum under the action of oxygen atoms immediately after cessation of the action of hydrogen atoms and their pumping-out at room temperature. On this curve the Sm band near 570 nm coincides with that observed under excitation by hydrogen atoms, but is of much lower intensity, while the other band lies near 500 nm, i.e., it is shifted relative to the 480 nm band under hydrogen excitation (cf. Fig. 3). Half an hour after the beginning of excitation by oxygen atoms, the 570 nm band completely disappears and another band develops in the RRL spectrum, with a maximum near 530 nm. The 500 nm band is retained, increasing in intensity. At the same time, the photoluminescence spectrum remains unchanged and consists only of the Sm band near 570 nm (the apparatus made it possible to excite the photoluminescence of the sample in situ, through a special window, using a PRK-2 lamp and a UFS-2 filter). These results once again indicate the purely surface character of RRL. Since the crystal surface is its most defective region, it is unlikely that the recombination energy migrates appreciably into the bulk of the crystal. Thus the luminescence center, apparently, is located in the immediate vicinity of, or coincides with, the recombination center of the adsorbed atom with an atom arriving from the gas phase. Therefore, the light will most probably be emitted by those luminescent centers that are located near, or themselves are, centers of adsorption and recombination of atoms. In this case, various centers may be excited that are characteristic only of the surface of the solid and absent in the bulk. The existence of such specific surface centers has recently been found in studies of exoelectron emission from solids [7]. Thus the appearance in the RRL spectrum of new bands in comparison with photoluminescence becomes understandable. In this connection, from the disappearance of the Sm band in a ZnS—Sm sample upon going from excitation by hydrogen atoms to excitation by oxygen atoms, it must be concluded that oxygen atoms are not adsorbed on samarium centers. These results may be of interest for catalysis and for establishing the relation between the catalytic and luminescent properties of solids.

Fig. 5. Transformation of the RRL spectrum upon going from excitation by hydrogen atoms to excitation by oxygen atoms:
1 — RRL immediately after excitation by H atoms,
2 — RRL half an hour after excitation by H atoms.

Temperature Dependence of RRL

The surface character of RRL excitation is also manifested in its temperature dependence. Fig. 6 shows the temperature dependence of RRL of the 510 nm band of zinc oxide under excitation by hydrogen atoms.

It is seen from the graph that, as the temperature is raised, the intensity passes through a maximum; in this case three temperature regions can be distinguished. The intensity maximum occurs at approximately \(393^\circ\mathrm{K}\). For comparison, we recorded the curve of the temperature dependence of the photoluminescence of our specimen, shown in the same figure. It represents the usual curve of temperature quenching. The observed dependence of the RRL intensity on temperature can be explained as the result of the action of the following factors: 1) a change with temperature in the number of particles chemisorbed in the state of “strong bonding,” recombination with which, according to [8], is responsible for the luminescence; 2) the temperature dependence of the recombination coefficient \(\gamma\) of atoms on the surface of the phosphor; 3) the dependence on temperature of the probability of radiationless transitions. Factors 1 and 2 are not independent. On the basis of the course of the temperature quenching of the photoluminescence of this specimen (Fig. 6) and our data on the function \(\gamma(T)\), it follows that

Fig. 6. Temperature dependence of the RRL intensity for ZnO

the first factor is responsible for the appearance of the maximum in the observed temperature dependence of RRL. Indeed, in the low-temperature region the adsorbed atoms are for the most part in a state of “weak” bonding with the lattice and do not produce luminescence upon recombination. As the temperature rises, the population of the “strong” bonding state increases, owing to which an enhancement of the luminescence is observed. With a further rise in temperature, the population of the “strong” bonding state decreases because of the competing process of desorption; as a result of this, and also because of the increasing role of radiationless transitions, the RRL intensity passes through a maximum. From Fig. 6, the activation energy of the “strong” adsorption state is \(2\ \mathrm{kcal/mol}\), while the binding energy of the adsorbed atom with the surface (for this state), estimated by the semiempirical rule [9], is \(35\ \mathrm{kcal/mol}\). The value found for the activation energy is in agree-

... with the result of Myasnikov, obtained from electrical-conductivity data [10]. The binding energy of the “weak” form of chemisorption, found from Fig. 6, is about 2 kcal/mole.

Similar results for the temperature dependence of RRL were obtained upon excitation of a ZnSCdS—Cu specimen by oxygen atoms. The graph corresponding to this case is presented in Fig. 7. The maximum RRL intensity occurs at a temperature of about 360°K. An estimate of the binding energy of an O atom in the state of a “strong” bond with the phosphor lattice gives a relatively large value, approximately 100 kcal/mole. The low-temperature branch of the graph for this specimen lies outside the investigated temperature range. The data obtained agree with the predictions of the theory of RRL [8], based on the concepts of the electronic theory of chemisorption and the band scheme.

Fig. 7. Temperature dependence of the RRL intensity for ZnSCdS—Cu

In work [8], the following expression was obtained for the RRL intensity:

\[ I = I_0 \left[ 1 + B \exp \left( - \frac{\varepsilon - v}{kT} \right) \right]^{-1}, \]

where \(\varepsilon\) is the Fermi level at the surface, measured from the bottom of the conduction band, and \(v\) is the level of the chemisorbed atom in the forbidden band.

For \(\varepsilon - v > kT\), the graph of the function \(\lg I \left( \frac{1}{T} \right)\) should be expressed by a straight line, which agrees with experiment. On the basis of a comparison of the experimental dependence \(\lg I \left( \frac{1}{T} \right)\) with formula (1), one can estimate the depth of the level of the chemisorbed atom below the bottom of the conduction band. In the case of a hydrogen atom on ZnO, from Fig. 6 we have \(\varepsilon - v \approx 0.09\) eV.

For ZnO in the bulk of the semiconductor, from work [7] \(\varepsilon \approx 0.1\) eV. Taking into account the surface band bending on ZnO, which according to Morrison’s data is about 0.5 eV, we obtain for the ZnO surface \(\varepsilon \approx 0.60\) eV, which corresponds to \(v \approx 0.51\) eV. Unfortunately, we cannot make such an estimate for oxygen atoms on ZnSCdS—Cu, since the position of the Fermi level at the surface of this specimen is unknown.

The results presented here show that temperature quenching of RRL occurs at higher temperatures than quenching of photoluminescence. This can be explained by the large contribution to RRL of short-lived emission in comparison with phosphorescence. Phosphorescen-

...cence, as is known, is quenched at lower temperatures than fluorescence, owing to the rapid depletion with temperature of shallow traps. And indeed, a ZnO sample which, under excitation by light, possessed an appreciable afterglow, when excited by hydrogen atoms practically ceased to glow simultaneously with the switching off of the discharge. The data obtained also show that, by measuring the intensity of RRL as a function of temperature, one can follow adsorption processes on the surface of a solid and judge the type of bond between chemisorbed particles and the surface. The study of RRL is therefore of considerable interest for the theory of catalysis and the structure of solids.

Quantum Yield of RRL

The surface character of RRL indicated above is also expressed in the magnitude of the quantum yield of luminescence. By the quantum yield \(\eta\) in the present case is meant the ratio of the number of acts of light emission to the number of acts of recombination of atoms on the surface. A strong deviation from an ordered structure on the surface of a crystal leads to a relatively large contribution of nonradiative transitions in the process of relaxation of the recombination energy, and to a small yield of luminescence. In the case of excitation of a crystal as a result of surface recombination of atoms, the small value of the yield is apparently also connected with the low probability of transfer of recombination energy to electronic excitation of the crystal as compared with phonon excitation [11].

There are only a few works on measuring the quantum yield of RRL [12, 13, 14]. We have developed a method which, unlike existing methods, makes it possible to avoid both the measurement of small temperature changes and the necessity of knowing the coefficient of recombination of atoms on the surface. To measure the heat of recombination, we constructed a sensitive differential vacuum microcalorimeter of the Calvet type [15], responding to heat flow. The sensitivity of the instrument with an M-195 output galvanometer is \(10^{-4}\) cal/sec per division. The calorimeter responds directly to the heat of recombination of atoms on the phosphor surface. The light flux of the luminescence was measured, after calibration in absolute units, with an FEU-19M photomultiplier. The results of measurements of the quantum yield of RRL will be reported elsewhere. Here we shall only indicate that \(\eta\) of RRL for phosphors based on ZnO and ZnS, under excitation by atomic hydrogen, has, according to our data, a value of the order of \(10^{-4}\)—\(10^{-7}\). It was observed in this connection that \(\eta\) depends on the treatment of the surface and on its temperature. The reasons for such a low quantum yield of RRL, associated with the surface character of the excitation, were discussed at the beginning of the paragraph.

In summary, it should be said that the purely surface character of the course of the RRL excitation process gives rise to a number of features of the luminescence, manifested in its spectra, temperature properties, in the magnitude of the yield, etc. In its properties, low-voltage cathodoluminescence apparently stands closest to RRL. Certain common features with RRL should also be found in photoluminescence under excitation by short-wavelength ultraviolet light, absorbed mainly in the near-surface region of the crystal. Owing to the especially close connection of RRL with the state of the surface, one may hope to develop it into a method of a distinctive luminescent probing of the surface of a solid, which is of considerable interest for problems of catalysis and the theory of solids.

References

1. E. L. Nichols, H. L. Howes and D. T. Wilber. “Cathodoluminescence and luminescence of incandescent solids.” Washington, 1928.

2. Present collection.

3. E. E. Tollefson and D. J. Le Roy. J. Chem. Phys. 16, 1057 (1948).

4. V. L. Levshin, S. A. Fridman, V. A. Chikhacheva, V. V. Shchenko. Izv. AN SSSR, ser. fiz., 29, 500 (1965); 30, 1494 (1966).

5. K. V. Tagantsev, A. N. Terenin. Optics and Spectroscopy, 2, No. 3, 355 (1957).

6. Surface Chemical Compounds and Their Role in Adsorption Phenomena. Moscow State University Press, 1957, p. 243.

K. Przibram. Nature, 179, No. 4554, 319 (1957).

1. P. Mёnold. In: Exoelectron Emission. Moscow, Foreign Literature Publishing House, 1962, p. 235.

2. F. F. Volkenshtein, A. N. Gorban, V. A. Sokolov. In: Surface Properties of Semiconductors. USSR Academy of Sciences, Moscow, 1962.

3. I. D. Hirschfelder. J. Chem. Phys. 9, 645 (1941).

4. I. A. Myasnikov. DAN SSSR, 120, No. 6, 1298 (1958).

5. I. E. Lennard-Jones, C. Strachan. Proc. Roy. Soc. A150, 442, 1935.

6. K. Sommermeyer. Z. f. Phys. Chem. B 41, 433 (1938).

7. G. N. Rokhlin, Z. N. Kobina, E. B. Volkova. Svetotekhnika, No. 1, 8, 1965.

8. K. M. Sancier, W. I. Fredericks, H. Wise. J. Chem. Phys. 17, 857 (1962).

9. E. Calvet, A. Prat. Microcalorimetry. Moscow, Foreign Literature Publishing House, 1963.