NEW PHOTOCATHODES

N. S. Khlebnikov and N. S. Zaitsev, Moscow

1. In 1931, carried away by the photoelements with blocking layers invented by him, B. Lange¹ somewhat hastily announced that photoelements with the external photoeffect had reached the limit of their sensitivity and that there was no future for them. Fortunately for technology, he was wrong.

As early as 1930, Koller first described oxygen–silver–cesium photocathodes, photoelements with which have now come into the widest general use. These photocathodes possess an integral sensitivity of, on average, 10, exceeding the sensitivity of those whose properties served Lange as the basis for his predictions by a factor of 10; and their quantum yield (in the region of the selective maximum), instead of 1–2%, can reach 20–25%. A further step of enormous importance was finally established in 1934 by the work of Kubetskii and P. Farnsworth² on secondary emission. The electron multipliers invented by them make it possible to obtain sensitivities, for photoelements with the external photoeffect, \(10^5\)–\(10^7\) times greater than those observed by Lange—sensitivities of such magnitude as can never be obtained with any other kind of photoeffect.

One may, of course, consider—and this would be quite fair—that the high sensitivity of electron multipliers has as its basis an entirely new phenomenon, secondary emission³, and not the photoeffect itself. But at the same time one must not forget that secondary emission can be used only in combination with the external photoeffect and cannot be applied to photoelements based on other kinds of photoeffect.

A new major step forward—this time already in the direction of the photocathodes themselves—was made in 1936. These newest photocathodes, possessing a whole series of remarkable properties, are the subject of our article.

2. If one traces the history of this discovery, it turns out that its sources lie in one comparatively old work (1910) on the classical photoeffect—by Pohl and Pringsheim⁴. This work was devoted to the study of the photoeffect of alloys of alkali metals (K) with certain heavy metals (Pb, Bi, Tl). In this work the influence of the relative concentrations of both components of the alloys was studied, and a number of conclusions were drawn which are not of fundamental interest from the point of view of modern views on the photoeffect.

But since at that time the photoeffect did not have such great practical significance as it has now, and, moreover, from the practical point of view these alloys offered no advantages compared with the existing (sodium–potassium) photocathodes, while for theoretical studies the investigation of alloys was a complication, this work did not attract responses. Later, when technology raised the question of highly sensitive photoelements (for sound cinema, television, etc.), improved

NEW PHOTOCATHODES

...ment of photocathodes under the influence of analogies between the photoelectric and thermionic emissions, the latter having been studied earlier for the photoeffect (for example, thoriated cathodes), proceeded in the direction of creating “thin-film” photocathodes⁵. Another analogy, of a more external character (with the field of photography), which had been established by Ollin⁶, also proved fruitful; it expanded the class of “sensitizers” of alkali metals, among which hydrogen was long known and used still earlier by Elster and Geitel⁷ (treatment of potassium surfaces by a discharge in hydrogen—hydride-potassium photocathodes). The most important practical result to which the development of these two directions led was the creation of oxygen–silver–cesium cathodes⁸.

The first continuer of Pohl and Pringsheim, 26 years after Germer⁹, turned out to have chosen the right path because all the existing photocathodes, not excluding the oxygen–silver–cesium cathode, failed to satisfy a number of requirements imposed on them by the developing technology of various photoelectric devices.

1. Possessing high integral sensitivity under direct illumination, which is due to the peculiarities of the distribution of sensitivity over the spectrum, oxygen–silver–cesium photocathodes do not make it possible to obtain high sensitivity under reverse illumination¹). This is a consequence of two circumstances. First of all, in such cathodes under reverse illumination there is inevitably absorption of light by the base metal (silver), which is present not only in the form of the substrate, but is also included in large quantity²) in the intermediate layer, and a considerable part of the light is lost before it reaches the outer surface of the cathode, where the emission centers are located. Therefore, in order even to have the possibility of working under reverse illumination, it is necessary to have as thin an intermediate layer as possible. However, as the thickness of the intermediate layer decreases, the number of active sites¹⁰ on which cesium atoms can be adsorbed, turning into emission centers with low ionization work, decreases. As a result of this it turns out that the curve of the distribution of sensitivity over the spectrum for a thin-layer oxygen–silver–cesium photocathode has no significant maximum in the long-wavelength part of the spectrum, even under direct illumination. For reverse illumination it is entirely absent. All this can be seen in Fig. 1, where on curve a, corresponding to a thin-layer oxygen–silver–cesium cathode under direct illumination, one can still notice a small rise between 700 and 800 mμ; on curve b (the same cathode under reverse illumination) there is not the slightest hint of a maximum.

Fig. 1. Distribution of sensitivity over the spectrum for oxygen–silver–cesium photocathodes (according to Koller): a — transparent cathode under direct illumination; b — transparent cathode under reverse illumination; c — opaque cathode

¹) By direct illumination we mean the case when light falls directly on the emitting surface of the cathode; reverse illumination is illumination of the cathode from the side of the substrate.

²) In the manufacture of oxygen–silver–cesium cathodes the sensitive layer is obtained by decomposing silver oxide Ag₂O with cesium vapors¹⁰ according to the equation

$$ \mathrm{Ag_2O + 2Cs = Cs_2O + 2Ag}, $$

so that silver is contained in such a layer in high concentration: 2 atoms for each molecule of Cs₂O.

The integral sensitivity even for the best specimens of such photocathodes, made under laboratory conditions, does not exceed 17 μA/Lm, as against 60–70 μA/Lm for ordinary oxygen–silver–cesium photocathodes (also made under laboratory conditions).

Meanwhile, obtaining high sensitivity under back illumination and especially sensitivity in the long-wavelength part of the spectrum is of enormous interest in connection with the development of new photoelectric devices—electron image converters[^11], for the Farnsworth television system[^12], and in a number of other cases.

1. Görlich set himself the aim of creating photocathodes that would be transparent in the sense that they would not completely absorb either the incident light or the electrons liberated in the layer. It is obvious that satisfying the second of these conditions would lead to the emission centers of photoelectrons capable of taking part in photoemission being located not only on the surface of the cathode, but also inside the layer. This

Fig. 2. Distribution of sensitivity over the spectrum for an opaque Bi—Cs cathode (according to Görlich)

Fig. 3. Distribution of sensitivity over the spectrum for a transparent Bi—Cs cathode (according to Görlich)

would lead to an increase in the probability of useful absorption of light and, consequently, to an increase in sensitivity. In order to eliminate absorption of light that does not lead to the escape of electrons, it was necessary to get rid of the metal substrate. This led Görlich to turn to the alloys studied by Pohl and Pringsheim[^4], since the emitting surfaces of such a structure are obviously homogeneous throughout their thickness. Since Pohl and Pringsheim had established that, at the same atomic concentration and in one and the same alkali metal (Pohl and Pringsheim), the red selective maximum and the threshold lie the farther toward the short-wave side the more electropositive the heavy metal is, while, on the other hand, the use of heavier alkali metals always shifts the boundary and the maximum toward the long-wave side, it turned out that, in order to obtain high sensitivity in the long-wavelength part of the spectrum, the use of an alloy of bismuth with cesium was most advantageous; this alloy was investigated especially thoroughly by Görlich.

1. According to Görlich, the method of making cathodes from the Bi—Cs alloy consists in evaporating Bi in vacuum onto the wall of the photocell bulb and depositing cesium on this layer. Figure 2 shows the sensitivity-distribution curve over the spectrum for an opaque cathode obtained in this way. The maximum sensitivity occurs at a wavelength of about 430 mμ, and the threshold lies between 700 and 800 mμ. The sensitivity-distribution curve for a transparent Bi—Cs cathode is shown in Fig. 3. It differs from the preceding one by a somewhat flatter course near the threshold, which gives a greater (relative) sensitivity in the long-wavelength region.

The two preceding figures characterize the properties of Bi—Cs cathodes in the visible region of the spectrum. Extending the wavelength region toward the ultraviolet shows (Fig. 4) that, along with the maximum at ~430 mμ, there exists an even shorter-wavelength maximum at about 370 mμ.

Since Görlich was primarily interested in high sensitivity in the long-wavelength region of the spectrum, he attempted to improve in this respect the sensitivity distribution of Bi—Cs cathodes by treating the cathode with oxygen. As is known, such treatment of alkali-metal photosensitive compounds under suitable conditions always leads to a shift of the work-function threshold toward longer wavelengths, to a displacement of the selective maximum in the same direction, and sometimes to the appearance of a new long-wavelength maximum. These cathodes proved to be no exception to the general rule. Fig. 5 gives a series of curves showing the spectral sensitivity distribution of a Bi—Cs cathode at various stages of sensitization with oxygen. Here the displacement of the maximum to the right proceeds in parallel with an increase in the amount of oxygen. Görlich does not indicate the method of treatment, just as he did not indicate the method of depositing cesium. There is every reason to think that the treatment consisted in introducing small portions of oxygen into the photoelement and subsequently heating the photoelement (perhaps while introducing new portions of the alkali metal).

Fig. 4. Distribution of sensitivity over the spectrum for a transparent Bi—Cs cathode in the visible and ultraviolet region (after Görlich)

Fig. 5. Change in the distribution of sensitivity over the spectrum for Bi—Cs cathodes treated with oxygen. The maximum shifts toward long wavelengths as the amount of oxygen in the layer increases (after Görlich and Zauner⁵)

Completely analogous results are obtained when transparent Cs—Bi layers are treated with oxygen. In Fig. 6 two curves relating to this case are shown. Thus, transparent Bi—Cs cathodes, in their spectral properties, are not inferior to massive ones, which is a major achievement.

It should be noted here—we shall return to this further on—that the Bi—Cs—O layers are no longer as homogeneous as the initial Bi—Cs layers, since the action of oxygen causes changes primarily in the layers of the cathode nearest to the surface. This, generally speaking, significantly changes the properties of the cathode, which is reflected in the character of the sensitivity distribution over the spectrum for direct and reverse illumination. In Fig. 7 three different possible cases are given. The curves in Fig. 7A show coincidence of the positions of the maxima for direct and reverse illumination. The curves in Fig. 7B and 7C show that the maximum of the curve for direct illumination lies at shorter wavelengths than for reverse illumination. From Figs. 7B and 7C it is also evident that the integral sensitivity for reverse illumination may exceed the sensitivity for direct illumination. Unfortunately, Görlich does not give the absolute values of the integral sensitivity of his Bi—Cs cathodes.

The excess of sensitivity under reverse illumination over the sensitivity under direct illumination is of great fundamental interest.

The possibility of such a situation indicates that the release of electrons may occur most intensively not at the outer surface, but in the depth of the cathode. The practical significance of this fact is hardly great. This can be judged only with data on the absolute values of the integral sensitivity of such cathodes. We shall return to this question below.

Fig. 6. Curves of spectral sensitivity for transparent sensitized Bi—Cs cathodes (according to Görlich)

Fig. 7A. Curves of spectral sensitivity for transparent sensitized Bi—Cs cathodes; a—for direct illumination, b—for reverse illumination (according to Görlich); the positions of the selective maxima coincide

In addition to Bi—Cs layers, Görlich also investigated other cesium alloys, the second components of which were lead, thallium, and antimony. The most interesting of these alloys is the antimony–cesium alloy, since it has a very sharp maximum in the short-wavelength region of the visible spectrum. This property of the Sb—Cs layer is well illustrated by Fig. 8, where

7B—the maximum for reverse illumination is shifted toward longer wavelengths; the integral sensitivity is greater for direct illumination

7C—the same as in B, but the integral sensitivity is greater for reverse illumination

the scale along the ordinate axis is the same for all curves. Görlich gives almost no data on this alloy. Some of its properties were described by Lukirskii and Dushevaia[^13].

1. The method of preparing the Sb—Cs layer used by these latter ...

authors, differed somewhat from the method of Görlich and consisted in applying a layer of antimony to the wall of a glass bulb, then heating it in cesium vapor, after which antimony was newly deposited, followed by another heating in Cs vapor, etc. By this method a layer of considerable thickness can be obtained.

Of great interest are the changes in the optical properties of the antimony layer that occur under the action of cesium vapor. A freshly deposited layer of antimony is an extremely good mirror on the side of the bulb wall, and on the reverse side is better than silver layers obtained by evaporation in vacuum. If the antimony layer is sufficiently thin to be transparent, then when viewed in transmitted light it appears slightly lilac. Heating in cesium vapor leads to a very considerable increase in transparency; moreover, in transmitted light the layer appears to be colored yellow, orange, and red, depending on the thickness of the layer (the red color corresponds to the thickest layer). When viewed in reflected light such a layer appears green. The increase in transparency is so great that a completely opaque layer of pure antimony, after treatment with cesium, makes it possible to view through it sufficiently brightly illuminated objects, to say nothing of the filament of an incandescent lamp.

Lukirsky and Lusheva investigated the distribution of sensitivity over the spectrum for their antimony–cesium cathodes. In Fig. 9 two curves (1 and 2) are given, providing the spectral sensitivity of the layers obtained by them under direct illumination. For comparison, an excerpt of the spectral characteristic of an oxygen–silver–cesium photoelement (manufactured at VEI; curve 3) is also given there. As is easy to see from comparison of this figure with Fig. 8, there is a substantial discrepancy between Görlich’s data and those of the last-cited authors regarding the position of the maximum (approximately 450 and > 500 mμ, respectively). The reason for this discrepancy

Fig. 8. Curves of spectral sensitivity for various photocathodes:
a — hydride–potassium cathode; b — oxygen–silver–cesium (transparent); c — bismuth–cesium (transparent, sensitized); d — antimony–cesium (transparent, sensitized); the scale along the ordinate axis is the same for all curves (according to Görlich and Zaüer)

Fig. 9. Curves of spectral sensitivity for Sb—Cs cathodes (1, 2) and Cs—Cs₂O—Ag under direct illumination (according to Lukirsky and Lusheva)

is not clear. Differences in the degree of purity of the antimony, the influence of gases (see above on Bi—Cs cathodes), etc., are possible here.

The same authors investigated the spectral distribution of absorption for an Sb—Cs layer. The corresponding curve is shown in Fig. 10. Here it is very clearly seen that the position of the minimum of the transmission curve coincides with the position of the maximum of the photoelectric yield in the preceding figure. The absorption properties of this layer explain the fact that the curve of the spectral distribution of sensitivity under back illumination has significant differences from the curve for front illumination, as is readily seen from a comparison of Figs. 9 and 11. The differences consist in a shift of the maximum from approximately 500 mμ to ~550 mμ, and also in a flatter course of the whole curve. It is easy to see that both these circumstances are a consequence of the absorption of light in the cathode layers remote from the emitting surface. The selective character of this absorption leads to the fact that, under back illumination, light that is closer to the emitting surface of the layer reaches, predominantly, longer-wavelength radiation, so that the maximum of the product of the probability of photoelectric absorption of a quantum of a given frequency by the number of quanta proves to correspond to a larger value of the wavelength than under front illumination. The decrease in the steepness of the curve is also the result of absorption of light in deep layers of the cathode, since it leads to a decrease in the amount of light reaching the layers, absorption in which can lead to the emission of electrons from the surface.

Fig. 10. Transmission of an Sb—Cs layer as a function of the wavelength of the incident light (according to Lukirskii and Lusheva)

Fig. 11. Spectral distribution of sensitivity for an Sb—Cs cathode under back illumination (according to Lukirskii and Lusheva)

In accordance with this, the integral sensitivity of such cathodes under back illumination is smaller than under front illumination, and this difference becomes more pronounced the thicker the layer.

Speaking of the integral sensitivity of Sb—Cs cathodes, it is necessary to distinguish the sensitivity at large and at small anode voltages. The fact is that the current-voltage characteristics of vacuum photoelements with such cathodes very often do not exhibit saturation. These characteristics may be regarded as consisting of two parts: from 0 to 80–100 V and above 100 V. In the interval 0 ~ 100 V the characteristic has the usual form of the characteristic of a vacuum photoelement: a rapid rise at first, gradually slowing down as 80–100 V is approached and showing a tendency toward transition into the saturation region. However, in reality this saturation very often does not occur, and with a further increase in the anode voltage \(V_a\) the curve continues to rise, its course in this interval being well expressed by a linear dependence of \(i\) on \(V_a\). This rise

may continue up to 1500 V and, probably, further. In some cases, when the voltage is increased above a certain limit (of the order of 1000 V, and for one and the same photoelement this limit depends on the luminous flux), a stepwise decrease of the current is observed at unchanged illumination1.

The sensitivity at high voltages may exceed by several times that at \(V_a\) equal to 80–100 V, and reach hundreds of microamperes per lumen. However, since the behavior of the photoelement under these conditions is undoubtedly determined by certain secondary processes, it is more correct to take as the sensitivity of the cathode itself that which is obtained at 80–100 V, at the boundary of the first and second sections of the current-voltage characteristic. This sensitivity is also very large, averaging about \(60\,\mu\mathrm{A}/\mathrm{Lm}\) (i.e., twice as much as for the average oxygen–silver–cesium cathode) and sometimes reaching \(80\,\mu\mathrm{A}/\mathrm{Lm}\) and more. As is easy to see from Fig. 9, this high sensitivity is due to a very sharply expressed maximum in the short-wave region of the visible spectrum.

Thus, antimony–cesium cathodes are the best for operation with visible light, but are unsuitable for operation in the infrared region.

7. On the basis of the data they obtained, Lukirskii and Lusheva attempted to explain the unusual current-voltage characteristics of photoelements with antimony–cesium cathodes. As these authors indicate, the increase in sensitivity with voltage takes place only in the case of thin layers2. The resistance of such cathodes along the layer has a very considerable magnitude. Thus, for example, in a photoelement specially made for such measurements with two leads from the cathode, the resistance between the leads proved to be of the order of \(10^{9}\).

In the presence of such a large resistance, in an ordinary photoelement with one cathode lead, at a sufficient distance from the lead the surface potential (when it is illuminated and, consequently, when current is drawn from it) may differ from the potential of the lead by a considerable value—of the order of hundreds of volts3. In just the same way, with respect to any two points of the surface located at different distances from the lead and sufficiently far from one another. As a result of the large potential differences between individual regions of the cathode surface, it is quite possible for electrons ejected by light from some region to strike not the anode, but another region with a higher potential (especially in spherical photoelements with a central anode, with which Lukirskii and Luneva worked, since in this case the field gradient at the cathode is small, and the local fields may probably be considerably stronger). The arrival of electrons from one region to another may lead to secondary emission at the bombarded region. Under the condition that the coefficient of secondary emission \(\sigma\) is greater than unity, this gives an increase in the number of ejected electrons and, consequently, an increase in the current to the anode, which will be perceived as an increase in sensitivity.

In order to explain in this way the observed increase in sensitivity, Lukirskii and Lusheva were compelled to assume that \(\sigma\) for an antimony–cesium layer exceeds 5 at 100 V. In reality,

however, they did not succeed in finding so high a value of \(\sigma\). In measurements with a special three-electrode photoelement they obtained a value of \(\sigma\) equal to 2–3, and in general, as it turns out, for a cesium-antimony layer \(\sigma\) at \(V_p = 100\ \mathrm{V}\) never reaches 5. Thus there is here a substantial quantitative discrepancy, despite the plausibility of the qualitative picture. We therefore consider the attribution of the explanation, at least, not exhausted. It is all the more probable that a number of other features of these cathodes were not taken into account by the cited authors.

1. Before speaking in more detail about the possible mechanism of action of antimony-cesium cathodes, it seems advisable to us to describe their properties in somewhat greater detail. A number of these properties, of which we have so far mentioned only the extraordinarily high integral sensitivity in the visible region, indicate that these cathodes will have exceptionally great practical significance. The data given on integral sensitivity refer to illumination of photoelements by light from an incandescent lamp (temperature about \(2400^\circ\ \mathrm{K}\)), in whose radiation energy there are few wavelengths corresponding to the maximum sensitivity of Sb—Cs photocathodes. Thus, for light sources richer in short-wave radiation, the ratio of the sensitivities of antimony-cesium and oxygen-silver-cesium cathodes will be still more favorable for the former. In particular, this should be noted in view of the use of these cathodes when working with daylight, in stellar photometers1 and in some investigations of fluorescence.

Another property of Sb—Cs cathodes, no less important from the applied point of view, is the absence of so-called photoelectric fatigue, even at cathode illuminations of the order of a thousand lux. It should be noted that at high anodic voltages some of these cathodes sometimes show an increase in sensitivity under completely unchanged external conditions (illumination, voltage). This increase proceeds rapidly at first, then gradually slows down, and after several minutes the sensitivity assumes a certain stationary value. This phenomenon indicates the presence of some internal processes in the photocathode.

Further, it is necessary to note the extremely great temperature stability of the new cathodes. Heating the photoelement does not lead to a change in its sensitivity, while operation at a lowered temperature does not cause an increase in fatigue. Only at the temperature of liquid air does the sensitivity fall sharply. It is restored to its initial value immediately after the temperature of the cathode rises.

It is also extremely curious that Sb—Cs (as also Bi—Cs) cathodes possess an incomparably greater uniformity of sensitivity over the surface than oxygen-silver-cesium cathodes. If for the latter the fluctuations of sensitivity from point to point may reach 50% of the mean sensitivity, then for Sb—Cs cathodes the fluctuations rarely exceed 5–6%. Summing up what has been said, it may be considered that Sb—Cs cathodes are, in all their properties, the best of all photocathodes existing to date for the visible region. We consider it quite probable that they will soon displace, in a whole range of applications,

oxygen–silver–cesium cathodes. The basis for such a view is, besides that just set forth, also the fact that the process of preparing Sb—Cs cathodes (see above) is incomparably simpler than the process of preparing Ag—O—Cs cathodes. In particular, it should be noted here that there is no such complex and difficult-to-control operation as oxidation of the silver substrate by a discharge in oxygen, and also that for Sb—Cs cathodes there is no danger of spoiling the cathode by introducing even a considerable excess of the alkaline metal. Such an excess can easily be removed from the photoelement by heating at a not too high temperature, without the slightest damage to the sensitivity of the cathode.

1. Regarding the mechanism of emission of these new cathodes, Görlich made one very significant and, undoubtedly, correct observation, namely that the emission centers in such a cathode are present throughout the entire thickness of the layer. This is a natural conclusion from the fact that in these layers, in contrast to ordinary composite cathodes, we have a structure homogeneous throughout the thickness.

Despite this, Lukirskii and Lusheva expressed the view that photoemission here too is due to the absorption of light only by the surface layer of the cathode. Confirmation of this view was taken to be the fact that these authors did not succeed in detecting a change in the resistance of the layer under illumination. It seems to us that the reason for this was rather shortcomings in the experimental method than the actual absence of such changes. Confirmation of Görlich’s view on the presence of emission centers not only on the surface of the cathode is, in particular, the fact that the sensitivity is invariably higher for thick layers than for thin ones. There is therefore no basis for thinking that the liberation of electrons by light does not also take place in deeper layers, from which the electrons can already emerge to the outside. It is quite possible that, with an increase in the anode voltage, owing to an increase in the potential gradient of the external field in the surface layers of the cathode, the number of centers participating in the outward emission of electrons increases. This supposition is all the more probable because Lukirskii and Lusheva’s explanation of the absence of saturation in the volt-ampere characteristics of Sd—Cs cathodes encounters serious difficulties, since the same curves are observed not only when the entire surface of the cathode is illuminated, but also when individual points are illuminated (with a light probe), even in the immediate vicinity of the lead-in, where there is no voltage drop along the layer. Thus we are inclined to the opinion that all the peculiarities of photoelements with these cathodes are due to the peculiarities of their structure and to the processes proceeding in them during photoemission.

It seems to us that the Sb—Cs cathode (as also Bi—Cs) is not simply an alloy of two metals, but their compound, corresponding to a definite stoichiometric composition[^14]. There is every reason for the formation of a compound here, since the indicated elements of group V of the periodic system possess negative valences. This is also indicated by the strong increase in transparency and the sharp increase in the resistance of the layer that occur when antimony is treated with cesium vapor, which points to a strong decrease in the number of conduction electrons. The presence of a definite stoichiometric composition is confirmed, for example, by the fact that removal of excess cesium by heating does not lead to a deterioration in the sensitivity of the cathode. This indicates that the cesium which is superfluous (from the point of view of high sensitivity) is bound incomparably more weakly than the principal cesium, which is an element of the structure of the layer.

If such a point of view is adopted, then we must consider that the Sb—Cs photocathode is a molecular lattice of a compound of antimony and cesium, possibly also containing a certain amount of free cesium. In this way it is easy to explain the extraordinary sharpness of the selective maximum for Sb—Cs cathodes, which from this point of view represents the absorption curve of homogeneous molecules of the compound-

…of Sb and Cs, whose absorption of light leads to photoionization1. Thus we believe that what occurs here is photoionization of the molecule, and there is no reason to think that, if it can occur in the surface layers of the cathode, it cannot occur in the depth; there it should manifest itself as a change in the resistance of the layer (an internal photoeffect), which Lukirsky and Lusheva were unable to detect.

For stationary emission to be possible, it is necessary that the depleted electrons be immediately replenished by new ones. Thus the presence of conduction electrons in the layer is necessary. Their presence may be due to thermal ionization of part of the Sb—Cs molecules. Since the work-function threshold of this cathode lies near \(750\ \mathrm{m\mu}\), which corresponds to a work function of about \(1.70\ \mathrm{V}\), the work of liberating an electron without its escape beyond the cathode should be still lower. Therefore one may expect rather strong thermal ionization at ordinary temperatures. When the cathode is illuminated, in addition, the number of free electrons increases at the expense of electrons liberated in the depth of the cathode and unable to escape outward. The assumption of thermal ionization as a source of free electrons is confirmed by the strong (but completely reversible) drop in the sensitivity of Sb—Cs photocathodes at the temperature of liquid air, mentioned above.

Sensitization of such cathodes by oxygen, in our opinion, leads to a change in the structure of the layers close to the surface. This change consists in the fact that, for such cathodes, along with normal emission centers there also appear centers consisting of cesium atoms adsorbed on active sites of the \(Cs_2O\) crystals, i.e. emission centers characteristic of ordinary complex photocathodes. It is characteristic that for sensitized cathodes fatigue is observed. Here one should also recall the data of Gerlich that, for sensitized (Bi—Cs) cathodes, the sensitivity under reverse illumination may be greater than the sensitivity under direct illumination. This is yet another confirmation of the assumption that emission centers exist in the bulk of the cathode.

As can readily be seen, all the properties of the new cathodes can be easily (qualitatively) explained by the above-outlined idea of a homogeneous molecular structure of the layer. In particular, the absence of fatigue is due to the absence of inhomogeneity in the layer, which makes impossible the electrolytic processes (the dragging of alkali-metal ions from the surface into the depth of the layer under the action of an external field) that play such a large role in the fatigue of ordinary complex photocathodes. The high uniformity of the surface with respect to sensitivity, as well as the temperature stability of the cathodes, is explained just as easily. A quantitative verification of the scheme we have proposed has not yet been fully carried out, but there are many measurable effects available for its implementation.

We believe that these new photocathodes, in addition to the great practical importance they will have, will also be useful in that they will permit a more detailed analysis of the mechanism of emission of complex, in particular ordinary \(Ag—Cs_2O—Cs\), photocathodes.

LITERATURE

1. B. Lange, Naturwiss., 1931; Uspekhi fizich. nauk, 11, 747, 1931.
2. P. T. Farnsworth, J. Frankl. Inst., 218, 411, 1934.
3. N. S. Khlebnikov, Uspekhi fizich. nauk, 16, 467, 1939; N. D. Morgulis, Uspekhi fizich. nauk, 16, 730, 1936; R. Kollath, Physik. Z.,

1. R. Pohl and P. Pringsheim, Verh. Dtsh. Phys. Ges., 12, 1039, 1910.
2. V. K. Zworykin and E. D. Wilson, Phys. Rev., 33, 633, 1929; J. Opt. Soc. Am., 19, 81, 1929; H. E. Ives and A. R. Olpin, Phys. Rev., 34, 117, 1929; N. Campbell, Phil. Mag., 6, 633, 1928.
3. A. R. Olpin, Phys. Rev., 36, 251, 1930.
4. J. Elster and H. Geitel, Ann. Phys., 48, 1892.
5. L. R. Koller, Phys. Rev., 36, 1639, 1930.
6. P. Görlich, Z. Physik, 101, 335. 1936.
7. J. de Boer, Electron Emission and Adsorption Phenomena, ONTI, 1936, ch. XIII.
8. V. K. Zworykin, Advances in Physical Sciences, 16, 814, 1936.
9. S. Yu. Lukyanov and A. A. Ravdel, Advances in Physical Sciences, 15, 814, 1935.
10. P. I. Lukirsky and N. I. Lusheva, Journal of Technical Physics, 7, 1900, 1937.
11. Ya. G. Dorfman and I. K. Kikoin, Physics of Metals, ONTI, 1933.
12. J. de Boer, Electron Emission and Adsorption Phenomena, ONTI, 1936, ch. X.
13. P. Görlich and H. Sauer, Z. Instrumentenkunde, 10, 423, 1936.

1. As is known, analogous phenomena occur in the coloring of halide salts of alkali metals[^15]. ↩↩↩

2. We believe that this is not so; see below. ↩

3. This was also shown by Lukirskii and Lusheva experimentally. ↩