Candoluminescence

V. A. Sokolov

Introduction

There exist many different kinds of luminescence, of which such types as photoluminescence, chemiluminescence, cathodoluminescence, radioluminescence, and others are now common knowledge. This cannot be said of candoluminescence, despite the fact that it has a long and still far from completed history.

The present article is devoted to a review of work on candoluminescence. The review does not claim to be exhaustive; its purpose is to acquaint the reader with the phenomenon under consideration, to illuminate briefly the main stages in the development of the problem touched upon, and, finally, to dwell on the very latest work on candoluminescence, associated mainly with the name of V. M. Kudryavtseva.

1. From the History of the Problem. Definition of Candoluminescence

As early as the beginning of the nineteenth century it was not only known, but also found practical application, that certain oxides possess the remarkable property of emitting a very bright light when incandescent at high temperature, especially under the action of high-temperature flames.

Thus, for example, the unusually bright glow of lime (CaO) when heated in an oxyhydrogen flame was widely used for a long time, under the name “Drummond light,” in projection lanterns, at lighthouses, and so forth, until this distinctive source of bright light was displaced by electricity. Also well known is the use of Auer incandescent mantles, coated with a layer of a mixture of thorium oxide and cerium oxide, which, when placed in the flame of a gas burner, kerosene lamp, and so on, considerably increase the brightness of the glow.

Despite the fairly wide use of the luminescence properties of such oxides, over a long period of time (1826–1891) this phenomenon was not subjected to detailed study and had no theoretical explanation.

In 1891 Nichols drew attention to the fact that certain oxides, in particular zinc oxide, have the ability to change their color upon heating[^1], and in 1892 he published his studies of the features of the temperature luminescence of zinc oxide, manifested at temperatures of 750–900° C[^2]. From that time on, abroad there arose an entire school, headed by Nichols, whose numerous works were devoted to the study of the luminescence features exhibited by certain oxides and compounds under purely thermal heating and especially under incandescence in high-temperature flames. Moreover, the latter was singled out especially in view of the opinion, widespread in Nichols’ school, that a flame, by virtue of its chemical activity, creates certain additional features to those luminescence features that are observed in these oxides and compounds under purely thermal heating (isolated from the flame). Common and fundamental for both cases in Nichols’ school is the assertion that these substances possess the ability for a special kind of luminescence, the so-called “temperature luminescence” or “luminescence of incandescent bodies,” known in foreign physics under the abbreviated name “candoluminescence”*). Despite the fact that for certain substances (Auer’s mixture, etc.) it had been proved that their special radiative properties in flames were mistakenly associated with luminescence and were due only to the features of their temperature radiation, the ability for candoluminescence is still ascribed to a whole series of different oxides and compounds.

The possibility of luminescence in the process of chemical transformations of a substance as a result of heating, as well as during chemical reactions in flames, is not excluded in principle, since a fairly wide range of phenomena associated with chemiluminescence is known; for this it is enough to recall the well-known works of V. N. Kondrat’ev and his collaborators on the chemiluminescence of flames[^3], although these works have no direct bearing on studies of candoluminescence.

Therefore, as E. I. Adirovich rightly points out[^4], the question of the ability for candoluminescence must be decided in each individual case and for each substance only by direct experiment, and the only possible criterion in this case is the Vavilov–Wiedemann criterion, which

) cando is the root base in the word incandescent*—incandescent.

CANDOLUMINESCENCE

clearly separates luminescent radiation from all other processes of radiation, regarding luminescence as an excess over the thermal radiation of a body in the case where this excess radiation has a finite duration, significantly exceeding the period of light oscillations\(^{5,4}\).

This definition first of all delimits luminescence from the thermal equilibrium radiation of bodies, which obeys Kirchhoff’s law:

\[ E_{\lambda T}=A_{\lambda T}\varepsilon_{\lambda T} =A_{\lambda T}\frac{C_1}{\lambda^5}\frac{1}{e^{\frac{C_2}{\lambda T}}-1}, \]

whereby the principal feature of luminescence is emphasized—the nonequilibrium character of the radiation process. But since a considerable number of cases of nonequilibrium radiation are also known which nevertheless do not belong to luminescence (for example, reflection and scattering of light, bremsstrahlung radiation, etc.), the second part of the Vavilov–Wiedemann criterion distinguishes luminescence among other types of nonequilibrium radiation.

In a whole series of cases, including those connected with the possibility of candoluminescence, in deciding the question of the character of the radiation the first part of the Vavilov–Wiedemann definition alone may prove to be a quite sufficient criterion, since if it is possible to establish that a given radiation process is an equilibrium one, then luminescence is excluded; it is clear that, in analyzing certainly nonequilibrium radiations, it is necessary to use the second part of the criterion considered.

Thus, the question of candoluminescence cannot be divorced from the general doctrine of luminescence, and the definition of candoluminescence can be given only on the basis of the Vavilov–Wiedemann criterion. In accordance with this criterion, by candoluminescence we shall understand an excess over thermal radiation, of long duration*) and arising at high temperatures in certain oxides and compounds, especially as a result of heating them to incandescence in flames.

2. THERMAL EXCITATION OF CANDOLUMINESCENCE

As already indicated, the idea of the ability of oxides to luminesce when heated was first developed by Nichols using zinc oxide as an example.

Zinc oxide is one of the most characteristic representatives of oxides exhibiting the features of temperature

*) In the sense of the Vavilov–Wiedemann criterion, i.e., significantly exceeding in duration the period of light oscillations (more than one ten-billionth of a second).

luminescence; its bright green-blue glow at high temperatures will seem an unusual and astonishing case of temperature radiation to anyone who observes it. It is therefore not surprising that the first object of investigation was precisely zinc oxide. These investigations were first carried out by Nichols and Snow in 1892 and published in an article entitled “On the character of the light emitted by incandescent zinc oxide”². In the named work, a comparison was made of the emissive power of zinc oxide with the emissive power of metallic platinum at the same temperatures for various wavelengths in the region of the spectrum under investigation. In these experiments the film of zinc oxide was deposited, by “smoking,” onto a strip of

Fig. 1.

platinum foil, which was then heated to the desired temperature by passing an electric current through it. The temperature spectrum of the oxide’s glow and the temperature spectrum of pure platinum were compared under the same conditions with a comparison standard, which was a specially selected incandescent lamp. As a result of the experiments the authors obtained a series of isochromatic and isothermal curves. The latter are shown in Fig. 1. On the ordinate axis in these graphs is plotted the emissive power in relative units, and on the abscissa axis—the wavelengths in microns.

The first (a) of the graphs shown gives the curves typical of the case of thermal radiation, both for platinum and for zinc oxide. Such a distribution occurs at a temperature below 700° C, and in this case the radiation from the oxide over the entire interval of the spectrum considered is lower than the radiation from platinum. The next graph (b) contains the corresponding curves for a temperature of 848° C. At this temperature, a maximum of radiation in the blue-green-

region of the spectrum, where the emissive power of the oxide already exceeds the emissive power of platinum at the same temperature. This effect is expressed more markedly at a temperature of 1013° C (curve 8), and in this case the maximum of the emission is shifted toward the long-wavelength part of the spectrum.

The general conclusion reached by the authors as a result of these experiments was the assertion that zinc oxide, when heated, possesses the capacity for a special kind of temperature luminescence. “We think,” they conclude, “that from these experiments it is clear that, in the case of zinc oxide, we are dealing with a substance which, in addition to ordinary incandescence due to temperature, becomes extraordinarily luminescent at temperatures above 880° C”².

“Because of the presence of ordinary incandescence at the same temperatures,” they go on to say, “this property of zinc oxide had until now escaped investigators, although it was, of course, known to everyone familiar with the reaction on a blowpipe tube that zinc oxide emits a light which differs very much from the light of coal or of most substances with whose incandescence we are familiar”².

Thus, an unusual feature of the temperature glow of zinc oxide, already distinguishable by eye, was regarded by Nichols and Snow as qualitative evidence of the presence of luminescence here. Their own experiments, which did indeed show the presence in zinc oxide of a spectrum of temperature glow that differs quite sharply from the ordinary spectrum of incandescence in the case, for example, of platinum, and the excess, in comparison with the latter, of the emissive power of zinc oxide at high temperature in the blue-green region of the spectrum, were already regarded by Nichols and Snow as quantitative proof of the existence in zinc oxide of a special kind of temperature luminescence*).

Already in this article the authors expressed the supposition that other metallic oxides too should show a peculiarity of emission upon heating, similar to that revealed in zinc oxide. The subsequent works of Nichols and his collaborators were devoted mainly to the investigation of the temperature glow of various oxides. Moreover, especially great attention was given to cases of excitation of their glow by means of a flame.

*) In the case of heating by thermal conduction, in the absence of chemical transformations in the substance, the possibility of non-temperature radiation is excluded, since the reverse would contradict the second law of thermodynamics; therefore, to explain luminescence in the case described, hypotheses connected with assumptions about chemical reactions in oxides arising under the influence of high temperature are invoked (see Section 4).

3. CANDOLUMINESCENCE OF OXIDES IN FLAMES

The Nichols school assigns an absolutely exceptional role to excitation of the temperature glow of oxides in flames. To illustrate this it is enough to say that, according to Nichols, zinc oxide, for example, heated to 800° by a flame, glows 115 times more brightly than when heated to the same temperature outside the flame6.

Coefficients of the same order of magnitude are also given for other oxides. The conviction that the flame has a special action of this kind was the reason why the overwhelming majority of investigations by the Nichols school were devoted precisely to the study of the peculiarities of the glow of oxides when excited by flames. At the same time, the decisive conclusion reached by Nichols and his collaborators is that the peculiarities of the glow of oxides in flames are explained by their capacity for candoluminescence.

The glow of lime (CaO) in a hydrogen–oxygen flame, mentioned above, was also assigned to this same kind of glow. Thus, in the introductory part of a report devoted to candoluminescence and read at a conference on luminescence in 1938 in Oxford, the speaker (Minchin) stated directly that “since 1921 there remains no further doubt that the glow of lime is an example of candoluminescence”7.

The principal characteristic features of candoluminescence in flames are considered to be the following6, 7:

a) the light output is many times greater than that of a black body, and the ratio may reach very large values, up to several hundreds;

b) this feature is clearly manifested in flames containing hydrogen and oxygen;

c) the most effective part of the flame is the boundary between the reducing and oxidizing parts of the flame;

d) a very short afterglow;

e) luminescence exists only between definite temperature limits, within which maxima appear that are characteristic of the given substance;

f) the spectrum of the emitted light gives bands, sometimes with regular frequency intervals. However, in a number of cases these bands are also observed under simple incandescence.

In the case of simple incandescence, features d) and e) are also clearly manifested.

Characteristic of many candoluminescent samples is the “fatigue” effect, consisting in the fact that in subsequent stages of excitation the intensity of the glow falls in comparison with the initial ones6.

Candoluminescence

Some authors speak of a substantial influence of activators on candoluminescence, while others deny this.^{7,12}

The substances in which candoluminescence can be observed in flames are chiefly oxides of metals of the 2nd group and of certain series of the 3rd, 4th, and 5th groups of the Mendeleev periodic system. They may be systematized as follows:

a) group 2: oxides of Be, Mg, Ca, Zn and, to a slight degree, Sr and Ba^{6,7,8};

b) group 3: oxides of B, Al, Ga and, probably, of all rare-earth elements^{6,7};

c) group 4: oxides of Si, Ti, Th and Zr^{7,9};

d) group 5: niobium oxide (Nb)^{7,10};

e) some sulfides, among which ZnS^{8} stands out especially.

The color of candoluminescence in a whole series of substances (ZnO, ZnS, TiO₂, etc.) is blue, or more precisely greenish-blue, but in some substances candoluminescence is characterized in general only by a very strong increase in brightness over the entire visible region when the oxide is introduced into a high-temperature flame.

Observation of the candoluminescence of oxides in flames was, in most cases^{6}, carried out by comparing, with the aid of an optical pyrometer, the emissive power of the oxide under investigation with the emissive power of some black oxide approaching an absolutely black body in absorption coefficient (uranium oxide, etc.). Thus, in the experiments of Nichols, Howes, and Wilber^{6}, such an oxide was applied to the surface of an alundum ring with an outside diameter of 1 cm and a width of 2 mm. The white*) oxide under investigation was used to coat the inner disk of this ring, likewise made of alundum. An assembly of this kind was then placed directly in the high-temperature flame in which the investigation was carried out, so that the surfaces of the disk and the ring were at the same level.

When observed through an optical pyrometer, the picture of candoluminescence in this case, according to the authors, appears especially distinct and vivid. Thus, for example, if the phenomenon is examined through a red filter (\(\lambda = 0.65\mu\)), the surface of the ring coated with the black oxide will appear bright, while the inner disk with the oxide under investigation will appear as a dark spot, since there is no candoluminescence in the red part of the spectrum and the radiation of the black body is more intense than the radiation of the oxide under investigation. If, however, the filament of the pyrometer lamp is brought to such a degree of incandescence that it will appear dark

*) As a rule, oxides possessing the described feature of temperature glow have a white color under ordinary conditions.

against the background of a light ring and light against the background of a dark disk with the oxide under investigation, then in the field of the pyrometer we shall have the picture shown in Fig. 2, a. When viewed through a blue filter \((\lambda = 0.45\,\mu)\), however, the effect is reversed (Fig. 2, b), which is interpreted as a consequence of the presence of candoluminescence in the blue region in the oxide under investigation, where its radiation exceeds in intensity the radiation of a black body under the conditions of one and the same flame.

By the method of the optical pyrometer, the authors carried out a comparison of the emissive power of a whole series of oxides placed in a high-temperature (usually hydrogen–oxygen) flame with the emissive power of a black body (uranium oxide or other black oxides) under the same conditions, in the temperature interval from 600 to 1600° C.

Fig. 2.

As a result of such measurements it was found that the excess of the intensity of luminescence of the oxides studied in the blue part of the spectrum over the temperature radiation of a black body under the same conditions reaches, in individual cases, extremely large values; this may be judged from Table I, where, for illustration, data are given for some oxides. The table contains the ratios of the radiation intensity of the oxides studied to the radiation intensity of a black body, obtained by Nichols and Howes\(^6\) for the blue region at a temperature of 665° C. In the table \(I_{\mathrm{b}}\) denotes the radiation intensity of the black body, and \(I_0\) the radiation intensity of the oxide studied.

Table I

4. EXPLANATIONS OF CANDOLUMINESCENCE

In Nichols’ school it was suggested that candoluminescence is the result of a molecular rearrangement occurring in oxides under the action of high temperatures\(^\text{11}\). The basis for such an assumption was the fact that, for a number of oxides, the temperatures optimal for candoluminescence correspond to the temperatures at which a change in the electrical conductivity of these oxides takes place\(^ {6,12}\). But this hypothesis was not accepted, since it cannot explain the duration of the process (luminescence is observed all the time while the ...

(heating), and, moreover, the proposition concerning molecular rearrangement has not been confirmed by X-ray investigations.

The hypothesis was also put forward that the energy of certain excited molecules or atoms in the flame is transferred to the crystal lattice of the solid body[^6]. The supposition has likewise been expressed more than once that the phenomenon of candoluminescence is connected with fluorescence excited by the ultraviolet radiation of the flame[^6].

But the most widespread and almost universally accepted hypothesis in Nichols’ school for explaining candoluminescence is the reduction–oxidation hypothesis[^6,^7,^13]. According to this hypothesis, advanced mainly for the case of excitation in a flame, candoluminescence is the consequence of a dual process in which reduction and oxidation rapidly follow one another in a series of unstable states arising under the action of the reducing and oxidizing zones of the flame.

This hypothesis can also be extended to the case of candoluminescence outside the flame, on the assumption that reduction of the metal from the oxide, i.e. decomposition of the oxide, can occur not only under the action of chemical reagents, the presence of which is characteristic of a flame, but also under the action simply of a high temperature (oxidation occurs both at the expense of the very oxygen liberated from the oxide and at the expense of the oxygen of the air); the chemical activity of the flame is the reason that greatly increases the intensity of the reduction and oxidation reactions, which explains the considerable excess of the brightness of the oxide glow in the flame as compared with purely thermal excitation of candoluminescence.

The arguments that have been advanced in favor of the reduction–oxidation hypothesis are as follows[^6,^7]:

a) the most effective region of the flame is (as already mentioned) the boundary between the reducing and oxidizing parts of the flame;

b) flames containing hydrogen and oxygen have the greatest effect on the excitation of candoluminescence;

c) most oxides capable of candoluminescence form suboxides;

d) when candoluminescence is observed during heating of an oxide in a vacuum vessel, evacuation of the vessel to remove air leads to a noticeable decrease in the glow.

These main arguments are also accepted as proofs of the validity of the reduction–oxidation hypothesis.

The most essential process in the hypothesis under consideration, as a result of which the visible directly arises,

V. A. SOKOLOV

luminescence, it is necessary to regard the oxidation process as a recombination process, which is usually the basis of the continuous spectrum of chemiluminescence. In explaining candoluminescence, the reduction process is assigned the role of a similar process that supplies material for oxidation; the following is considered to confirm this.

Although the most effective part of the flame is the boundary between the reducing and oxidizing regions of the flame, candoluminescence is generally observed in both parts, but in very different ways. In this connection, Nichols and Howes even introduced special terms—“phase R” and “phase O” (from “Reduction”—reduction, and “Oxidation”—oxidation)⁶. The principal difference between “phase O” (excess oxygen) and “phase R” (excess reducing agent—hydrogen) is the stronger and more distinct manifestation in the former of the characteristic features of candoluminescence, in particular the intensity of the greenish-blue glow.

As Nichols and Howes note, the very strong influence on candoluminescence of precisely the oxidizing part of the flame, and the very distinct difference between phases “O” and “R,” appear especially clearly in the example of niobium oxide. If a piece of this oxide is gradually introduced from the outside into the outer, i.e. oxidizing, part of a hydrogen flame, a pale greenish-blue glow immediately appears. When this piece is moved farther into the flame, i.e. into the reducing part, its color suddenly changes from blue to red. If, however, one part of the same piece of oxide is in the reducing region and the other in the oxidizing region of the flame, then a boundary can be observed between the red and greenish-blue glow, and the maximum of the glow, as in the case of other oxides, is situated at the boundary on the side of the outer, oxidizing, part of the flame.

The enhancement of candoluminescence observed in many cases at the boundary between the oxidizing and reducing regions of the flame is explained, according to the hypothesis under consideration, by the fact that here, on the one hand, reduction of the metal by hydrogen proceeds intensively, i.e. a “supply” of material for oxidation occurs, and, on the other hand, there is also the possibility of rapid oxidation by oxygen; farther on, in the very outer part of the flame, despite the presence of oxygen, the glow decreases because of the decrease in the concentration of reduced metal. In the reducing region of the flame, however, despite active reduction, intense candoluminescence cannot take place because of the lack of oxygen.

Thus, in summary, it may be said that the reduction–oxidation hypothesis of candoluminescence is a chemical hypothesis that regards candoluminescence as chemiluminescence arising as a result of an oxidation process.

5. DENIAL OF CANDOLUMINESCENCE

Another group of hypotheses put forward to explain candoluminescence in fact amounts to a denial of it as such, and to explaining the above-described features of the glow of oxides solely by the peculiarities of their thermal radiation.

Doubts concerning the existence of candoluminescence were expressed at various times by Lenard, Schmidt, Giesel, and others[^7]; moreover, these doubts gained real ground especially after a group of investigators headed by Rubens proved irrefutably that the exceptional radiative properties of the Auer–Welsbach mixture (99% thorium oxide and 1% cerium oxide), previously regarded as something entirely foreign to thermal radiation[^7], are explained solely by the peculiarities of the latter, as manifested in the components of the mixture when it is placed in a burner flame.

In brief, these peculiarities are as follows[^7][^14]. Pure cerium oxide \((\mathrm{Ce_2O_3})\), placed in the same flame in which thorium oxide \((\mathrm{ThO_2})\) is found, owing to its great absorptive and, consequently, emissive capacity in the infrared region of the spectrum, assumes a lower temperature than thorium oxide (at \(T = 1800^\circ\mathrm{K}\) for \(\mathrm{ThO_2}\), \(\mathrm{Ce_2O_3}\) has \(T = 1350^\circ\mathrm{K}\)). Moreover, cerium oxide, having a minimum of radiation only in the region adjacent to \(6\mu\), possesses a very considerable radiative capacity throughout the whole spectrum, including the visible part. In thorium oxide, however, the radiative capacity is negligible in the visible region and in a considerable part of the infrared spectrum. The addition of cerium oxide to thorium oxide considerably increases the emissive capacity in the visible spectrum, but at the same time the emission in the infrared region also increases. It turns out, however, that the addition of a small amount (of the order of 1%) of cerium oxide does not noticeably affect the infrared radiation and, consequently, the temperature of the mixture, and increases only the emission in the visible region without any noticeable lowering of temperature; as a result, the luminous efficiency of the mixture is considerably increased. An Auer mantle coated with this mixture, placed in a flame beside a black body, will in comparison with the latter have a higher temperature, because it expends much less energy on infrared radiation than the black body. And since in the visible region the absorptive capacity of the Auer mixture is close to unity, the brightness of the mantle, owing to its higher temperature, noticeably exceeds the brightness of the black body under the conditions of one and the same flame.

Smith[^15] experimentally showed that the selectivity of radiation is the cause of the increase in brightness and of other radiative—

bodies excited by a flame. It is obvious that such a phenomenon may also have occurred in many cases described by Nichols, Howes, Wilber, and others, especially if in these cases only the method of the optical pyrometer was used and no direct measurements were made of the temperatures of the oxide under investigation and of the black body, which under flame conditions, it must be said, are not easily carried out and are not entirely reliable.

Thus, the doubts expressed concerning the existence of candoluminescence are, in many cases, quite substantial. At the same time, the candoluminescence of certain solid bodies at temperatures below red heat apparently represents true luminescence. Thus, according to Smith \({}^{15}\), such luminescence in a low-temperature flame can be obtained from boron nitride and calcium oxide activated by small amounts of bismuth, manganese, praseodymium, etc. Pure substances, according to the author, do not produce this effect, since the presence of an activator is essential.

As for candoluminescence at high temperatures, the existence of luminescence in these cases cannot be considered proven, since the overwhelming part of the radiation then appears to be thermal in nature. The only correct way to reveal the nature of such radiation in each concrete case, as already indicated, is the application of the Vavilov–Wiedemann criterion. This was precisely the path chosen by V. M. Kudryavtseva in her studies of candoluminescence; in order to clarify the possibility of applying Kirchhoff’s law to the cases of luminescence being studied, she first undertook an investigation of the absorptive capacity of selected objects (ZnO, ZnS, etc.) under various temperature regimes (see section 6).

In concluding this section it is necessary to note that many purely qualitative arguments advanced in favor of candoluminescence may also be turned against it. Thus, the greater effectiveness of the hydrogen–oxygen flame is explained rather not by its chemical activity, but by its high temperature, since it is generally known that the oxyhydrogen flame is the hottest among many other flames.

The different action of the reducing and oxidizing regions of the flame on the oxide is also apparently a purely thermal effect, since it is well known that the outer, i.e. oxidizing, part of the flame has a higher temperature than its inner, reducing, region.

The decrease in the intensity of the oxide’s luminescence during prolonged heating in vacuum may be explained by an increase in the concentration of the metal due to decomposition of the oxide; the metal, not possessing those properties with respect to absorption, etc., which

characteristic of the oxide, and no longer exhibits the features of blue luminescence, as a result of which the intensity of this component decreases when the oxide is heated in vacuum.

The gradually increasing concentration of the metal can apparently also explain the “fatigue” effect observed in “candoluminescence.”

6. FEATURES IN THE ABSORPTION OF LIGHT BY CERTAIN OXIDES AND THEIR TEMPERATURE RADIATION

As early as Nichols[^1] drew attention to the change in the color of zinc oxide upon heating (with increasing temperature ZnO changes its white color to yellow), which indicated a change in its absorbing power with temperature. Nichols explained this by a change in the scattering of the substance in the short-wavelength part of the spectrum and, in his subsequent work[^3] on the study of the features of ZnO luminescence at temperatures above 700°C, attempted to connect this with the features of temperature luminescence; however, on the basis of the experiments described by us in Section 2, he came to the decisive conclusion that candoluminescence is present in ZnO.

Trow and Gislolf[^7] suggested that the region of continuous absorption in ZnO, which at room temperature lies in the ultraviolet, shifts with increasing temperature into the visible part of the spectrum; this accounts both for the change in color and for the features of the temperature luminescence of ZnO. At the same time Gislolf, having investigated the absorption spectrum of zinc oxide at temperatures from room temperature up to 500°C, did in fact discover the presence of such a shift. According to Gislolf’s data, the absorption limit of ZnO at room temperature lies near 3850 Å, and at 500°C near 4600 Å. Gudiv[^16] indicates that, according to his data, the beginning of the absorption band of ZnO at room temperature is near 4000 Å, and complete absorption begins at 3850 Å. F. I. Vergunas and F. F. Gavrilov[^17] investigated the transmission absorption of thin films of zinc oxide by a photographic method at the temperatures of liquid air, room temperature, and 100°C. It was established that at the temperature of liquid air the absorption band of zinc oxide begins at 3800 Å and complete absorption at 3700 Å; upon heating to +20°C the boundary of intrinsic absorption shifts toward longer wavelengths to 3860 Å, and at +100°C to 3880 Å.

V. M. Kudryavtseva and co-workers carried out a series of studies on the displacement of the edge of the absorption band of zinc oxide with changes in temperature up to high temperatures, since the latter is of particular interest in connection with the candoluminescence of ZnO, which appears at temperatures above 700°C.

The first experiments were carried out in 1941 by studying the spectra of diffusely scattered light from an electric lamp by pressed ZnO powder. These experiments gave the beginning of the absorption band at room temperature at 4100 Å and the boundary of continuous absorption at 3800 Å. At a temperature of 900°C the beginning of the absorption band shifted into the visible region beyond 4700 Å^18,19. Exploratory experiments were also carried out with zinc sulfide, for which a shift of the absorption band with increasing temperature into the visible region was likewise found.

Subsequently, P. S. Litvinova and N. L. Gasting^18 investigated, by the method of diffuse scattering, the shift of the edge of the absorption band of zinc oxide powder rolled onto glass, at temperatures from −45° to 500°C, and established a uniform shift of the absorption band with increasing temperature toward longer wavelengths.

Fig. 3.

Then V. M. Kudryavtseva and G. I. Sinyalkina^18,20, using a Pulfrich photometer (also by the scattering method), carried out a study of the absorption of zinc oxide (as well as zinc sulfide) in the visible region of the spectrum as the temperature was varied from room temperature to 950°C.

As a result of all these investigations, the nature of the absorption of zinc oxide and its variation with temperature up to the values characteristic of candoluminescence were definitively established.

Remaining complete in the region of higher frequencies, the absorption in ZnO falls sharply within a comparatively narrow frequency interval to very small values, characteristic of white powders. As the temperature is raised, the absorption band of zinc oxide shifts into the visible region of the spectrum, in such a way that the change in absorption at the boundary of the absorption band has one and the same character at all temperatures in the interval from −180° to +950°C.

Figure 3 gives the absorption curve of ZnO for a temperature of 950°; the dashed curve shows the absorption at

room temperature. Within these limits the curve shifts uniformly, with increasing temperature, toward longer wavelengths.

Investigations of ZnS were less detailed, but it was established[^18] that here too there is a systematic displacement of the edge of the absorption band with increasing temperature into the visible region, as is shown in Fig. 4.

In a subsequent paper by V. M. Kudryavtseva and G. I. Sinyakina[^31], the established features of the absorption of zinc oxide were related to the “candoluminescence” of ZnO under its thermal excitation, and the subordination of this kind of luminescence to Kirchhoff’s law was shown quantitatively,

\[ E_{\lambda T}=A_{\lambda T}\cdot e_{\lambda T}. \]

Fig. 4.

As follows from this law, a change in the absorptive capacity of a body \(A_{\lambda T}\) in the case of thermal radiation correspondingly leads to a change in its emissive capacity \(E_{\lambda T}\). Consequently, in accordance with the change in absorptive capacity \(A_{\lambda T}\) which is characteristic of zinc oxide, its temperature radiation must change from the radiation of a nonblack body to radiation close in intensity to that of a black body, according to the scheme shown in Fig. 5. In this scheme, \(E\) denotes the branch of the emissive-capacity curve as a function of \(\lambda\) for a nonblack body with some absorption coefficient \(A\) at temperature \(T\). \(E_0\) is the same for another body, whose absorption coefficient \(A_0\) at the same temperature is considerably larger than that of the first, approaching in value 1 (a black body). The solid curve gives the radiation at this same temperature for substances in which the absorption changes sharply from the values \(A\) to the values \(A_0\), as is shown in the lower part of the scheme, i.e. similarly to the change in absorption in zinc oxide (Fig. 3).

Consequently, all substances possessing a continuous absorption band with a sharply terminating edge must, in accordance with Kirchhoff’s law, have on the temperature-radiation curve an auxiliary maximum corresponding to the sharp change in absorptive

ability \(A\), which is also illustrated by Fig. 5. Since the boundary of continuous absorption in ZnO shifts with increasing temperature from the ultraviolet into the long-wave part of the spectrum, reaching at temperatures of \(800—1000^\circ\mathrm{C}\) the green-blue region of the spectrum, then, according to Fig. 5, a secondary, “green-blue,” maximum should also appear on the curve of the temperature radiation of ZnO.

V. M. Kudryavtseva and G. I. Sinyakina showed that the features of the temperature luminescence of ZnO fit entirely into this scheme. The position of the maximum usually corresponds to the position of the edge of the ZnO absorption band for the given temperature. Fig. 6 shows (from the data of V. M. Kudryavtseva and G. I. Sinyakina) the energy of the temperature radiation of zinc oxide \(E_\nu\) at \(900^\circ\mathrm{C}\) as a function of frequency in the green-blue region of luminescence.

Fig. 5.

Fig. 6.

The maximum here corresponds to the wavelength \(\lambda = 4650\,\text{\AA}\), at which, also for a temperature of \(900^\circ\mathrm{C}\), as Fig. 3 shows, complete absorption is attained.

The authors showed that \(E_\nu\) varies with frequency \(\left(\dfrac{1}{\lambda}\right)\) in proportion to the quantity

\[ \lambda^{-5}\cdot e^{-\frac{C_2}{\lambda T^*}}. \]

In Fig. 6 the straight line on the right depicts the course of \(\lg E\) at \(900^\circ\mathrm{C}\) for the branch of the curve falling toward the ultraviolet part of the spectrum (circles on the straight line); the dots give the values of

\[ \lg\left(\lambda^{-5}\cdot e^{-\frac{C_2}{\lambda T^*}}\right). \]

*) Kirchhoff’s law was applied in the form

\[ E_{\lambda T}=A_{\lambda T}\cdot \varepsilon_{\lambda T} =A_{\lambda T}C_1\times \lambda^{-5}e^{-\frac{C_2}{\lambda T}}, \]

where the function \(\varepsilon_{\lambda T}\) is taken in Wien’s form, which fully justifies itself in the visible region of the spectrum at temperatures not exceeding \(3000^\circ\mathrm{K}\).

The spectrogram shown in Fig. 6 was obtained photographically; as the authors note, the low sensitivity of the film in the long-wave region of the spectrum did not make it possible to obtain the second branch, i.e., the new rise of intensity toward the long-wave side of the spectrum. V. M. Kudryavtseva and G. I. Sinyakina also carried out visual photometric studies, in which this rise is detected quite distinctly. Having taken the logarithms of the energies of the temperature radiation of ZnO from the data of visual photometry and plotted them graphically as a function of frequency, they obtained two parallel straight lines, displaced relative to one another and representing, on a logarithmic scale, the branches \(E_0\) and \(E\) (Fig. 5), which quantitatively confirmed the subordination of the studied type of ZnO luminescence to Kirchhoff’s law.

Since the boundary of the absorption band of ZnO shifts, with increasing temperature, into the long-wave part of the spectrum, the radiation maximum corresponding to the sharp change in absorption must also, with increasing temperature, move in the same direction. The studies of V. M. Kudryavtseva and G. I. Sinyakina show that this does indeed occur.^21 In addition, the displacement of the subsidiary maximum on the curve of the temperature radiation of ZnO was also clearly revealed in the studies of Nichols and Snow, as is readily seen from a comparison of the radiation curves of zinc oxide at \(848^\circ\mathrm{C}\) and \(1018^\circ\mathrm{C}\) obtained by them and shown in Fig. 1.

Summarizing the foregoing, it may be said that the thermal nature of the features of the luminescence of zinc oxide in the case of purely thermal excitation (in the experiments of V. M. Kudryavtseva and G. I. Sinyakina the oxide was heated by means of an electric furnace) is definitively established, which proves the untenability of Nichols’s ideas concerning the temperature luminescence of ZnO.

A similar series of experiments was carried out by V. M. Kudryavtseva and G. I. Sinyakina^22, ^23 with respect to titanium dioxide (\(\mathrm{TiO}_2\)), which is included among actively candoluminescent substances.^6, ^7

The study of absorption in \(\mathrm{TiO}_2\) showed that it has the same character as in ZnO and ZnS, namely, a continuous band with an abruptly terminating edge, to which a subsidiary maximum on the curve of temperature radiation corresponds, as is readily seen from Fig. 7, where there are presented (according to the data of V. M. Kudryavtseva and G. I. Sinyakina) the dependence of absorption (upper curve) and of temperature radiation (lower curve) of \(\mathrm{TiO}_2\) on wavelength at a temperature of \(800^\circ\mathrm{C}\); moreover, as in the case of ZnO, there is also a displacement of the subsidiary maximum of the temperature radiation with increasing temperature toward longer wavelengths.

waves, which is clearly seen from Fig. 8, where the radiation curves of TiO\(_2\) are given for temperatures of 600 and 800° C. The quantitative subordination of TiO\(_2\) radiation to Kirchhoff’s law, i.e. its equilibrium character, is especially convincingly illustrated by Fig. 9, which represents the function \(f\left(\dfrac{1}{\lambda}\right)=\lg E+5\lg\lambda\) for the temperature radiation of TiO\(_2\) at 800° C. The values of \(f\left(\dfrac{1}{\lambda}\right)\), calculated on the basis of the experimental data corresponding to the two branches \(E_0\) and \(E_c\) of the curve of the temperature radiation of TiO\(_2\) (Fig. 7), fit well on this graph onto two parallel straight lines\(^*\), displaced relative to one another owing to the abrupt change in the absorptive capacity of the oxide.

Fig. 7.

Fig. 8.

G. I. Sinyapkina also studied the temperature glow of ZnS, for which exactly the same results were obtained\(^{23}\).

Thus, the investigations carried out with respect to TiO\(_2\) and ZnS showed that the features of the glow manifested in them under thermal excitation also have a purely thermal nature and are in no way connected with luminescent radiation.

\(^*\) Taking the logarithm of the formula

\[ E_{\lambda T}=A_{\lambda T} C_1 \lambda^{-5}\cdot e^{-\frac{C_2}{\lambda T}} \]

leads to the expression

\[ \lg E+5\lg\lambda=\lg(AC_1)-\frac{C_2}{T}\cdot\frac{1}{\lambda}, \]

which is the equation of a straight line of the form

\[ f\left(\frac{1}{\lambda}\right)=a+b\frac{1}{\lambda}. \]

The studies described confirm not only qualitatively, but also quantitatively, that the features of the temperature luminescence of ZnO, ZnS, TiO₂ (and other substances similar to them with respect to absorption)

Fig. 9.

are connected with the features of absorption and are entirely subject to the laws of equilibrium radiation.

7. THE THERMAL CHARACTER OF THE LUMINESCENCE DURING THE OXIDATION OF ZINC AND DURING EXCITATION OF ITS OXIDE IN FLAMES

The experiments of V. M. Kudryavtseva and G. I. Sinyakina, which showed the untenability of the notions of the presence of candoluminescence in ZnO, ZnS, and TiO₂ in the case of thermal incandescence, cast doubt on the existence of candoluminescence of these substances also when they are excited in flames.

V. A. Sokolov²⁴ investigated zinc oxide in detail in this direction; moreover, since the most widespread hypotheses explaining candoluminescence assign the decisive role to the chemical activity of the flame and especially to reduction–oxidation processes, in which oxidation is regarded as the cause of the appearance of the candoluminescence spectrum (see Section 4), the primary aim was to clarify (on the basis of applying the Vavilov–Wiedemann criterion) the nature of the luminescence during the oxidation of zinc.

The experiments were carried out on an apparatus that made it possible to burn zinc in an oxygen atmosphere and to photograph, through a spectrograph, the accompanying oxi-

... a bright glow of the flame. The spectrum obtained is shown in Fig. 10.

It is interesting that in the blue-green part of this spectrum there is a maximum whose appearance and position are characteristic of the spectrum of the temperature radiation of ZnO.

The repetition in the flame spectrum accompanying the oxidation of zinc of the spectrum of the temperature radiation of its oxide indicates that the luminescence during the oxidation of zinc owes its origin to the temperature radiation of particles of the oxide formed as a result of the reaction, heated to a high temperature by the heat of reaction. If this is so, then in the given region of the spectrum

Fig. 10.

there is only temperature radiation of the oxide particles. In this case, as in the cases considered earlier, the branches of the radiation curve \(E_1\) and \(E_2\), according to Kirchhoff’s law, must vary with frequency \(\left(\dfrac{1}{\lambda}\right)\) in proportion to the quantity \(\lambda^{-5} e^{-\frac{c_2}{\lambda T}}\). Indeed, as Fig. 11 shows, the experimental points corresponding to the values \((\lg E + 5 \lg \lambda)\) lie well on theoretical straight lines giving the values of \((\lg E + 5 \lg \lambda)\) for these branches at the temperature of zinc combustion (\(\sim 1000^\circ\)C). In addition, the calculation carried out by the author according to Kirchhoff’s formula showed that the appearance and position of the edge of the ZnO absorption band at this temperature correspond not only qualitatively, but also quantitatively to the position and shape of the secondary maximum on the radiation curve in the zinc-oxidation spectrum. In Fig. 12, at the top, the corresponding

CANDOLUMINESCENCE

in this case the emission curve; below is the absorption curve of ZnO according to the data of V. M. Kudryavtseva and others. The black dots mark the experimental data on the absorption of ZnO, while the white dots are the results of calculation by Kirchhoff’s formula on the basis of the experimental points of the emission curve given above.

All this indicates complete obedience to Kirchhoff’s law of the zinc-oxide spectrum in the spectral region studied. Since nonequilibrium radiation is not observed in the oxidation processes of zinc occurring in their pure form, still less can it be observed in those oxidation processes that might take place as a result of partial decomposition of ZnO under the action of high temperature or under the influence of chemically active flame reagents; this speaks to the untenability of the reduction–oxidation hypothesis of candoluminescence with respect to ZnO.

Fig. 11.

Fig. 12.

The author also carried out a series of experiments to study the luminescence spectra of ZnO upon excitation in flames. For this purpose, flames of a gasoline burner, illuminating gas and water gas, as well as mixtures of these gases with hydrogen, were used. The chemical composition of the flames was different, the most chemically active in the sense of reducing properties being water gas with a hydrogen content (according to the data of N. N. Norkinа[^25]) of more than 50% and carbon monoxide (CO) up to 30%. Despite

regardless of the differing chemical activity of the flames, the luminescence spectrum of ZnO when excited in them always had one and the same character, namely, a repetition of the spectrum of the temperature luminescence of zinc oxide.

Figure 13 shows the luminescence spectrum of ZnO when excited by a gasoline flame. It is easy to see that this is indeed the ordinary spectrum of the temperature luminescence of zinc oxide, with the characteristic secondary maximum for it in the blue-green region of the spectrum.

Fig. 13.          Fig. 14.

A quantitative confirmation of the purely thermal nature of this spectrum is also the good agreement of the experimental points corresponding to the values \((\lg E + 5 \lg \lambda)\) with the theoretical straight lines (Fig. 14); moreover, as the calculations showed, the form and position of the secondary maximum, within the scatter of the points, also agree quantitatively (according to Kirchhoff’s law) with the form and position of the absorption band of ZnO at the same temperature. Spectra of the same type and the same quantitative results were obtained also in the case of excitation of ZnO by the other flames mentioned above.

The results of the experiments described made it possible to answer unambiguously the question of the nature of the features of ZnO luminescence when excited in flames: this luminescence is not appreciably connected with nonequilibrium radiation processes, and the sole cause of its occurrence is thermal excitation due to the high temperature of the flame.

V. A. Sokolov also carried out exploratory experiments with ZnS, CaO, and MgO^26, which made it possible to suppose that, in the case of excitation of these substances in flames, the equilibrium component of the glow also plays the decisive role.

CONCLUSIONS

The brief survey of works on candoluminescence presented here shows that the history of this question is long and rather confused. Undoubted clarity is brought to this question by a series of works carried out under the general direction and with the direct participation of V. M. Kudryavtseva. The success of these works is due above all to the fact that the only correct path of investigation was chosen—namely, the application of the Vavilov–Wiedemann criterion, which must always lead to an unambiguous answer wherever the possibility of luminescent radiation is at issue.

True, these works do not yet make it possible to say definitively that candoluminescence does not exist at all, since so far they have investigated only a small number of substances among those to which the capacity for candoluminescence is ascribed; nevertheless, the clear results of these investigations compel one to express a fully legitimate doubt concerning the decisive significance of nonequilibrium radiation in all other cases connected with ideas about candoluminescence.

It should be supposed that in the future the word “candoluminescence” will find no application at all in scientific terminology; and if it remains, it will be used to designate the described peculiarities of the temperature glow of bodies with full awareness that this term entered science as the result of scientific errors and misconceptions and in no way reflects the essence of the phenomenon.

CITED LITERATURE

1. E. L. Nichols and B. N. Snow, Phil. Mag. 32, 401 (1891).
2. E. L. Nichols and B. N. Snow, Phil. Mag. 33, 19 (1892).
3. V. N. Kondrat'ev, Uspekhi khimii 12, 4 (1913).
4. E. I. Adirovich, Some Questions in the Theory of Luminescence of Crystals. Gostekhizdat, Moscow–Leningrad, 1951.
5. S. I. Vavilov, Izv. AN SSSR, ser. fizich. 9, 278 (1945).
6. E. L. Nichols, H. L. Howes and D. T. Wilber, Cathodo-luminescence and the luminescence of incandescent solids, 1928.
7. L. T. Minchin, Trans. Farad. Soc. 35, 163 (1939).
8. E. L. Nichols and H. L. Howes, Phys. Rev. 19, 314 (1922).
9. E. L. Nichols, Phys. Rev. 22, 420 (1923).
10. E. L. Nichols and Stanford, J. Opt. Soc. Amer. 26, 91 (1936).
11. E. L. Nichols and H. L. Howes, J. Opt. Soc. Amer. 22, 170 (1932).

1. E. L. Nichols and D. T. Wilber, Phys. Rev. (2), 17, 453 (1921).

2. E. L. Nichols and Boardman, J. Opt. Soc. Amer. 20, 115 (1930).

3. Ribo, Optical Pyrometry, GTTI, 1934.

4. A. Gaydon, Spectroscopy and the Theory of Combustion, Foreign Literature Publishing House, Moscow, 1950.

5. C. T. Goodeve, Trans. Farad. Soc. 33, 340 (1937).

6. F. I. Vergunas and F. F. Gavrilov, Transactions of SFTI 24, 181 (1947).

7. V. M. Kudryavtseva, G. I. Sinyakina, P. S. Litvinova, and N. L. Gastin, Scientific Notes of Tomsk University 8, 33 (1948).

8. V. M. Kudryavtseva, F. I. Vergunas, and P. S. Litvinova, Izv. AN SSSR, Physical Series 9, 404 (1945).

9. V. M. Kudryavtseva and G. I. Sinyakina, Transactions of SFTI 24, 214 (1947).

10. V. M. Kudryavtseva and G. I. Sinyakina, DAN 59, 1411 (1948).

11. V. M. Kudryavtseva, Izv. AN SSSR, Physical Series 13, 115 (1949).

12. G. I. Sinyakina, ZhETF 19, 581 (1949).

13. V. A. Sokolov, DAN 66, 45 (1949).

14. N. N. Norkin and G. D. Spetstsi, Izv. Tomsk Polytechnic Institute 66, 3 (1948).

15. V. A. Sokolov, Transactions of SFTI 27, 59 (1948); 28, 393 (1949).