Full Text
ANTIMONY-CESIUM CATHODES AND PHOTOCELLS
N. S. Khlebnikov
I. INTRODUCTION
Since the time when Uspekhi fizicheskikh nauk first reported1 on complex photocathodes of a new type, one of this group of cathodes—the antimony-cesium cathode—has been and continues to remain the most popular object of photoelectric, secondary-emission, and other investigations. Moreover, during the indicated interval this photocathode has been used both in ordinary photocells and in other photoelectric devices, in particular in iconoscopes, and it may now be stated with complete confidence and justification that this cathode has become firmly established in practice and that its appearance constituted an epoch in the technology of photoelectric devices and related fields.
Antimony-cesium cathodes were first obtained abroad2. However, the credit for their investigation and introduction into production belongs entirely to Soviet scientists and engineers. A considerable number of people took part in this work—both workers of research laboratories and workers of industry, approaching the question from the most diverse sides. This circumstance found expression, in particular, in the programs of the last two conferences on the photoeffect and secondary emission[^3],[^7], where the question of antimony-cesium cathodes was invariably singled out into a special and extensive section.
The result of this varied and broadly organized work has been that serious achievement which antimony-cesium cathodes should be considered to represent. It is now time to sum up the work that has been done and to indicate its practically important results. In addition to its general interest, such a review is also significant for the development of further work in this and in fields close to it.
II. PHYSICAL PROPERTIES OF ANTIMONY-CESIUM PHOTOCATHODES
1. The nature of Sb-Cs cathodes.
As is known, in Görlich’s first publication1 on the new type of photocathodes, to which, in particular, the antimony-cesium cathode belongs, only certain photoelectric properties were reported in more or less detail—the distribution of sensitivity over the spectrum—and only in general terms was mention made of other physical features, in particular-
N. S. KHLEBNIKOV
...with considerable transparency and sensitivity when the cathode was illuminated from the side opposite to its emitting surface.
This was a consequence of the fact that Gerlich considered the antimony–cesium layer to be an alloy of two metals—cesium and antimony—and the true nature of this new material, and along with it the unusual properties of the new cathodes, remained unclear to him. In exactly the same way, in the first of the works published in our press² on antimony–cesium cathodes, where this material was regarded a priori as an alloy, considerations as to the reasons for the unusual behavior of these cathodes were reduced merely to the point that the properties of two metals, finely dispersed one in the other, might have nothing in common with those of each component separately.
It was sufficient, however, to abandon³ the preconceived idea (which in no way followed from the simplest observations on antimony–cesium cathodes) that the combination of two metals is necessarily and exclusively an alloy (that this is not obligatory had already been well known earlier⁴), for it to become possible to explain the peculiarities of antimony–cesium cathodes and to predict a number of their properties.
The further development of work in this field showed the complete correctness of the fundamental propositions built upon this new conception of the antimony–cesium cathode. Let us briefly recall those simple, but at the same time decisive, facts which served as the starting point for the development of modern views on the antimony–cesium photocathode.
The process of forming an antimony–cesium layer usually consists in depositing on the wall of an evacuated vessel a layer of antimony (by evaporation in vacuum) and in treating this layer with cesium vapor at a temperature of \(\sim 200^\circ\), these operations being capable of being alternated several times. As a result of such treatment, the original antimony layer, which is a pure metallic mirror and (if it is not too thick) has, when viewed in transmitted light, a violet coloration, completely changes its properties. It loses its metallic luster and greatly increases its transparency for long-wavelength radiation, as a result of which in transmitted light it now appears yellow, orange, or ruby- or cherry-red, depending on the thickness of the layer. Simultaneously with the changes in optical properties there occurs a strong increase in its resistance, which grows by approximately \(10^5\) times. Finally, as it turns out, the antimony–cesium layer, unlike metals, has a large negative temperature coefficient of resistance.
All these facts can be explained only by assuming³ that, first, the antimony–cesium layer is not an alloy but a compound, and, second, that this compound is a semiconductor. In favor of the fact that the antimony–cesium layer is a chemical compound, there is also a series of other data. First, it is known that antimony possesses negative valences and forms with hydrogen the compounds \(\mathrm{SbH_2}\) and \(\mathrm{SbH_3}\); compounds of antimony with such metals as Ni, Cu, Ag are also known, where Sb plays the role of a metalloid. Further, it was established that this cathode, in the course of treatment with cesium, behaves quite differently from ordinary complex photocathodes of the oxygen–silver–cesium type. This difference is explained by Fig. 1, where curve 1 schematically represents the course of the change
photosensitivity1) of an oxygen–silver–cesium cathode as a function of the time of arrival of cesium on the surface of oxidized silver, while curve 2 is the same dependence for an antimony–cesium cathode. In this case, at the usual temperatures of formation of the layer (~200°), the thermoelectron current is almost imperceptible because of the relatively high work function of these cathodes. A comparison of these curves shows that, whereas for the oxygen–silver–cesium cathode the cesium content in it (which determines the photosensitivity) is arbitrary in the sense that it depends on external factors (the time of treatment with cesium vapor), for the antimony–cesium cathode this dependence exists only up to a definite moment—until the cathode has acquired maximum sensitivity. Subsequently, the newly arriving cesium remains in a certain dynamic equilibrium with the layer and does not affect its photosensitivity. (It should be stipulated here that this is the case under the necessary condition that the possibility of cesium condensation on the cathode surface is excluded; if this is not so, then the sensitivity of the cathode falls, approaching the sensitivity of massive cesium; this decrease in sensitivity, however, in contrast to that for an oxygen–silver–cesium cathode, is reversible to a considerable degree.) Thus, the antimony layer can absorb only a certain definite amount of cesium—obviously, that which corresponds to the stoichiometric composition of the compound of these two metals.
Fig. 1. Curves of change in the sensitivity of oxygen–silver–cesium (1) and antimony–cesium (2) photocells as a function of the time of treatment with cesium vapor.
A direct experiment6, consisting in determining the amounts of Sb and Cs forming the photocathode, showed that this is indeed so. The amount of Sb was determined beforehand (a definite weighed portion of this metal was evaporated), while the amount of cesium was found by titration of the alkaline solution (CsOH) obtained after opening the bulb of the photocell and washing it with water. A number of determinations by this method gave, for the ratio of the numbers of Cs and Sb atoms, figures lying between 3.1 and 3.2. This showed that for the antimony–cesium layer the formula SbCs₃ must be adopted. The excess cesium, apart from errors in the determinations, must be attributed to the presence in such a cathode, in addition to chemically bound cesium, also of free cesium existing in the form of a surface-adsorbed film, and also within the bulk of the cathode (recently the view7 has become widespread that this cesium is distributed inside the cathode in the form of internally adsorbed atoms). Later determinations8, made by other, more complicated methods, gave ratios of the numbers of cesium and antimony atoms of 5.5 (first method) and 4.5 (second method). This forces the authors of the cited investigation to believe that in the Sb–Cs cathode
1) As is known, in general the thermoelectron current exhibits the same course.
there is no definite stoichiometric ratio between the quantities of the two metals. However, since five lines below they state that the SbCs layer is the compound SbCs₃, one may think that, in their opinion as well, the matter here lies in errors made in the chemical analysis.
In any case, at present there is no doubt that the antimony–cesium photocathode is a semiconductor. One of the most convincing proofs in favor of this view is the strong increase of its conductivity with temperature, discovered long ago³. The temperature dependence of the resistance of these layers has recently been studied quantitatively, and although the numerical data of different authors differ rather considerably, qualitatively they confirm one another. Thus, it has been established that the change of conductivity with temperature is well represented by the law
\[ \sigma=\sigma_0 e^{-\frac{U}{2kT}}, \]
where \(\sigma\) is the conductivity and \(U\) is the ionization work. The first determinations of \(U\) were made by N. D. Morgulis⁹, who found that \(U=0.3\text{ eV}\). A later determination by Luk’yanov and Mazover⁸ gave for \(U\) the value \(0.67\text{ eV}\)*). At present there are no grounds for giving preference to any one of these figures, since they describe the results of measurements on individual specimens of cathodes. Likewise, the value of the specific resistance of Sb–Cs layers has not been established with any precision. According to Morgulis, its value is about \(1\Omega\cdot\text{cm}\); according to Luk’yanov, \(\sim 10\Omega\cdot\text{cm}\).
The antimony–cesium layer is an electronic semiconductor, since the predominant type of conductivity in it is electronic conductivity. This was recently shown directly by P. G. Borzyak¹¹ by determining the contact polarity between an Sb–Cs layer and a metal when it is heated and illuminated. The same proposition had also been expressed earlier³ on the basis of the behavior of antimony–cesium cathodes with respect to photoelectric fatigue.
2. Internal photoeffect of antimony–cesium cathodes
In the first of the works in which an attempt was made to interpret the mechanism of action of antimony–cesium cathodes², the opinion was expressed that their considerable integral sensitivity and high quantum yield are due to very strong absorption of light in the very surface layer of the cathode. This opinion also followed in a rather forced way from the conception of the antimony–cesium cathode as an alloy or a finely dispersed mixture of two metals. However, a number of very simple observations³ showed that this is not so.
First of all, it was invariably found that the sensitivity of the antimony–cesium cathode (under direct illumination) proves to be higher in
*) In the latest work of N. D. Morgulis and B. I. Dyatlovitskaya¹⁰, the authors give, for the temperature interval from 90 to 290° K, \(U\) equal to \(0.28\text{ eV}\); for the interval from 290 to 400° K, \(U \leq 0.8\text{ eV}\).
ANTIMONY–CESIUM CATHODES AND THE PHOTOEFFECT
the case of thicker layers. From this circumstance followed the only possible conclusion, namely that in thin layers saturation of the active light is observed. This is in full agreement with the character of the variation of the yield of Sb–Cs layers as a function of thickness, and with the form of their spectral characteristics. Further, the curves of the dependence of sensitivity on the layer thickness for direct and reverse illumination, obtained on an antimony–cesium wedge, spoke of the same thing.\(^{12}\) These curves showed directly that the photosensitive region of the photocathode has a thickness much greater than a monatomic one. Finally, it was difficult to imagine that the quantum yield of cathodes, which, according to Görlich’s data,\(^{1}\) exceeded by several times that of all other cathodes, could be the consequence of photoelectric absorption of light only in the very surface layer of the cathode.
Thus, in the case of these cathodes the liberation of photoelectrons had to take place in a layer of considerable thickness, and since a large part of these electrons might not leave the cathode, this meant that Sb–Cs layers should exhibit a change in conductivity under the influence of illumination, i.e. an internal photoeffect. Such changes in conductivity were indeed found already in the first investigations,\(^{2,3}\) but some authors\(^{2}\) attributed them to the so-called “sliding photocurrent.” The reason for this error was\(^{4}\) defects in the experimental method, which were later clarified.\(^{3}\)
Changes in conductivity under the influence of illumination in an Sb–Cs cathode under ordinary conditions (room temperature) are small, owing to the fact that these layers possess a considerable intrinsic conductivity, masking the changes arising under the action of light. In consequence of this, the internal photoeffect of Sb–Cs layers is better observed at low temperatures. The most direct proof that photoconductivity in Sb–Cs cathodes exists as such, and is not a consequence of the presence of an external photoeffect (“sliding photocurrent”), consisted in observing the change in the resistance of the cathode and the magnitude of the external photoeffect when illuminated through a red light filter.\(^{3}\) These observations showed that, whereas such illumination gives an increase of the external photocurrent by 1–2%, the changes in the resistance of the layer (under the same conditions of illumination and temperature) reach 20–30%.
Fig. 2. Diagram of the apparatus with which P. G. Borzyak investigated the photoconductivity of Sb–Cs layers
A more detailed investigation of the photoconductivity of Sb–Cs layers was recently carried out by P. G. Borzyak\(^{11}\) at the temperature of liquid air. The arrangement of the apparatus with which these experiments were performed is shown in Fig. 2. The apparatus consisted of a glass tube about 20 mm in diameter, on a portion of the inner surface of which two electrodes \(E_1\) and \(E_2\) were deposited (separated from one another by strips of uncovered glass surface). From a tungsten spiral \(S\), by evaporation, a layer of antimony was deposited, overlapping one of the gaps between the electrodes. This antimony was treated in the usual manner with cesium vapor. Measurements of the photoconductivity were made with the apparatus immersed in liquid
when the antimony–cesium layer has negligible conductivity and, consequently, its masking effect is absent.
Under intense illumination with the unfiltered light of an incandescent lamp, Borzyak observed, in the gap between the electrodes, an increase in the conductivity between electrodes \(E_1\) and \(E_2\) by \(10^5\) times. The spectral distribution of the sensitivity was investigated only very approximately (with the aid of light filters), and it turned out that sensitivity exists both in the short-wavelength and in the long-wavelength regions of the visible spectrum.
The conductivity of the layer acquires its final value not immediately after the illumination is switched on, but only after some time has elapsed (in any case, with the very intense illumination that was used). In exactly the same way, after the illumination is switched off, the conductivity does not fall to its initial value at once, but only gradually decreases over a fairly considerable time. One might suppose that this circumstance is a consequence of the fact that illumination of the layer causes it to heat up, and this heating persists for a long time. However, since the cooling time \(t\) may be expressed by the formula
\[ t \simeq \frac{1}{15}\ln \frac{T_0}{T}, \]
where \(T_0\) is the initial temperature of the layer and \(T\) is the temperature at the moment \(t\), the temperature should already decrease by a factor of \(e\) in \(1/15\) sec. Such a rapid decrease in temperature cannot explain the considerably slower decrease in conductivity. In order to return the film quickly to its initial state, it is necessary to remove it from liquid air (heat it) and then cool it again without exposing it to illumination. It is also very interesting that, even when the film is illuminated without applying a potential difference, residual conductivity exists.
All this can apparently be explained, as Borzyak did, by the small probability of reverse thermal transitions of electrons from the conduction band (owing to the excessively large activation energy) at the temperature of liquid air, although the long persistence of conductivity, according to P. S. Tartakovskii, may also have another cause—space charges.
Whatever the mechanism of photoconductivity in Sb–Cs cathodes may be, at present there are still too few data to judge it with certainty; the experiments described by P. G. Borzyak show with complete certainty and clarity that photoconductivity does exist in Sb–Cs cathodes. With the aid of the same apparatus, Borzyak showed, in the manner mentioned above, that the conductivity of Sb–Cs layers is electronic in character.
3. External photoeffect of Sb–Cs cathodes
As has already been noted above, in the first works on antimony–cesium cathodes, with one exception, it was assumed that the photoeffect of these cathodes is purely a surface phenomenon. This point of view found its fullest expression in the extremely interesting and valuable work of S. S. Pri-
Plezhayev^13, who studied in detail the influence of the state of the surface of an antimony–cesium cathode on its sensitivity, the distribution of sensitivity over the spectrum, and also the distribution of photoelectrons by energy.
The principal results of this work consist in establishing the fact that the (integral) sensitivity of this cathode is determined (in the temperature range up to \(200^\circ\)) by the magnitude of the work function, which in turn depends on the degree of coverage of the surface by adsorbed cesium atoms, i.e., ultimately, on the surface temperature and the pressure of cesium vapor above it, since these factors determine the state of dynamic equilibrium between the adsorbed film and the vapor. For the dependence of the photocurrent \(i_s\) on the work function \(\varphi\), Prilezhayev gives the following empirical formula:
\[ \lg i_s = B_1 - B_2\varphi, \]
where \(B_1\) and \(B_2\) are constants depending on the spectral composition of the radiation. Despite the great interest and high practical value of this work for the creation of a rational technology of antimony–cesium cathodes, it cannot be considered to exhaust the question of the mechanism of their action, and one cannot agree with some of the author’s conclusions.
Thus, for example, it seems to us beyond doubt that the antimony–cesium cathode cannot be regarded, as Prilezhayev does, as belonging to the group of \([\mathrm{W}]\)-Cs, \([\mathrm{W}]\)-Ba, \([\mathrm{W}]\)-Th and similar cathodes, and one cannot accept for it, in de Boer’s notation^14, the formula
\[ [\mathrm{Sb},\mathrm{Cs}]\text{-}\mathrm{Cs}. \]
In fact, a cathode denoted in this way is assigned a quite definite emission mechanism: the extraction of conduction electrons mainly from the field of the surface potential barrier of the substrate (the base metal), while the role of the external adsorbed film consists almost exclusively in lowering the work function. As is known^15, such cathodes are characterized by small values of the quantum yield (of the order of fractions of a percent), whereas for the Sb–Cs cathode this quantity amounts (see below) to \(25\text{--}30\%\). Here we have no more than a limited analogy, relating probably only to the structure of the outer surface of the cathodes, and it is incorrect to elevate this analogy to the rank of an identity, all the more so since it is fundamentally wrong to try to force a new phenomenon into the framework of already known things.
In view of the fact that the semiconducting nature of the Sb–Cs cathode is at present already a firmly established fact, and also that the magnitude of the quantum yield can be understood only if the external photoeffect in this case is regarded as a volume one, the following formula for the Sb–Cs cathode seems to us correct:
\[ [\mathrm{SbCs}_3,\mathrm{Cs}]\text{—}\mathrm{SbCs}_3,\ \mathrm{Cs}\text{-}\mathrm{Cs}, \]
which is also in agreement with the facts of the existence of a definite stoichiometric composition and an excess amount of cesium.
In exactly the same way, taking into account the semiconducting nature of the Sb–Cs layer and the volume character of the photoeffect, the explanation by S. S. Prilezhayev of the differences in the distribution of electrons does not seem sufficient to us.
for Sb-Cs cathodes and metals is that in an Sb-Cs cathode the distribution of electrons differs from the Fermi distribution. First, since this material is not a metal, it must already be non-Fermi-like; second, because of the volume character of the effect, there can be no doubt that the processes of scattering and absorption of electrons on their way to the surface must inevitably affect the form of the curves of the distribution of electrons over energies. One of the most interesting questions concerning Sb-Cs cathodes seems to us to be the question of the emission centers in them. At present the most widespread opinion[^16] is that such centers are cesium atoms internally adsorbed in the lattice of \(SbCs_3\). In our opinion, however, this view has over the other[^17] view (that the emission centers are the molecules of the compound \(SbCs_3\) itself) only the advantage that this is the way matters stand in other known cathodes.
The most convincing argument in favor of the volume character of the external photoeffect of Sb-Cs cathodes has undoubtedly been, and remains, the exceptionally large value of the quantum yield in the region of the selective maximum of the spectral curve, which cannot be understood at all, in particular if, as regards the mechanism of action of these cathodes, one confines oneself to the circle of ideas contained in the cited work of S. S. Prilezhaev.
This circumstance served as a stimulus for determining the magnitude of the quantum yield by various authors using different methods. Already a comparison of the curves of spectral sensitivity of different cathodes, carried out in Gerlich’s first work,1 showed that an Sb-Cs cathode has a quantum yield several times greater than that for a hydride-potassium cathode, for which the value of this quantity was the largest of those known (\(2.3\%\)). However, these data of Gerlich referred to cathodes which apparently still had a comparatively low integral sensitivity (\(\sim 20\)—\(30\ \mu A/lm\)). Determinations made by the indirect method by Luk’yanov[^18] and Gurevich,[^19] as well as direct measurements carried out by Prilezhaev,[^20] gave for the quantum yield at the maximum of the spectral-sensitivity curve values from 25 (Gurevich, Prilezhaev) to \(30\%\) (Luk’yanov). These figures refer to cathodes with a sensitivity of the order of \(100\ \mu A/lm\) (at a color temperature of the source of \(2\,800\)—\(2\,850^\circ K\)).
In connection with these measurements, it seems advisable to dwell on the indirect method for determining the quantum yield, used by Luk’yanov and Gurevich. This method differs favorably from direct determination of the quantum yield by the relative simplicity of the necessary measurements (one need know only the spectral sensitivity distribution curve, expressed in relative units, and the integral sensitivity of the photoelement at a known temperature of the source). Nevertheless, it has not become sufficiently widespread among us, despite the fact that it was proposed comparatively long ago and that the question of the magnitude of the quantum yield is very often of great importance. In brief this method consists in the following1.
The integral sensitivity of the photocathode \(I\), expressed, as usual, in units of current strength (amperes) referred to a unit luminous flux (in lumens), can be represented as the ratio of two integrals
\[ I=\frac{\int\limits_{0}^{\lambda_0} i_0(\lambda)E(\lambda,T)\,d\lambda} {N\int\limits_{\lambda_1}^{\lambda_2} v(\lambda)E(\lambda,T)\,d\lambda}\ \mathrm{A/lm}, \tag{1} \]
where \(i_0(\lambda)\) is the absolute spectral sensitivity of the photocathode (in amperes per watt of incident energy); \(E(\lambda,T)\) is the spectral density of the radiation incident on the cathode (in arbitrary units); \(N\) is the maximum visibility coefficient (equal to \(621\ \mathrm{lm/W}\)) and \(v(\lambda)\) is the coefficient of relative visibility of radiation; \(\lambda_0\) is the wavelength corresponding to the threshold of emission; \(\lambda_1\) and \(\lambda_2\) are the limits of the visible spectrum. Since in ordinary measurements of spectral sensitivity it is expressed in relative units, in order to use formula (1) it is necessary to replace \(i_0(\lambda)\) by a function \(i(\lambda)\), expressing the spectral sensitivity in relative units and differing from \(i_0(\lambda)\), evidently, only by a constant factor
\[ i_0(\lambda)=k_0 i(\lambda). \tag{2} \]
Substituting \(i_0(\lambda)\) from (2) into (1), we can find \(k_0\):
\[ k_0=NI \frac{\int\limits_{\lambda_1}^{\lambda_2} v(\lambda)E(\lambda,T)\,d\lambda} {\int\limits_{0}^{\lambda_0} i(\lambda)E(\lambda,T)\,d\lambda} \ \mathrm{A/W}. \tag{3} \]
As is easily seen from (2), the quantity \(k_0\) is numerically equal to the photoelectric yield expressed in amperes per watt for the value \(\lambda=\lambda_{\max}\) (when \(i(\lambda)\) becomes unity). To determine the yield expressed in electrons per quantum (quantum yield), it is sufficient to pass to other units of charge and energy. If, following Lukyanov, the electron charge \(e\) is taken equal to \(4.80\cdot10^{-10}\) CGSE and Planck’s constant \(h\) equal to \(6.61\cdot10^{-27}\ \text{erg-sec}\), then we obtain:
\[ 1\mathrm{A/W}=1\ \text{coulomb/joule}=1.14\cdot10^{-4}\lambda^{-1}\ \text{electrons/quantum}. \]
Determining \(k_0\) from (3), expressing \(\lambda\) in angstroms and \(I\) in microamperes per lumen, we find for the magnitude of the quantum yield at the maximum:
\[ k=7.73\lambda^{-1}B\ \text{electrons/quantum}, \]
where \(B\) denotes the ratio of the integrals in formula (3). This quantity can be found graphically with the aid of the empirical curve of spectral sensitivity, the tabulated visibility curve, and the curve of the spectral density of radiation, which can be constructed according to Planck’s formula if the temperature of the radiation source is known. Thus, all necessary measurements svo-
is reduced to recording the spectral curve and measuring the integral sensitivity at a precisely known source temperature.
Figure 3 gives the spectral curves of antimony–cesium (1) and oxygen–silver–cesium (2) cathodes with sensitivities of about 80 and 60 μA/lm, respectively. These are curves reduced to unit energy, but not reduced to a unit ordinate at the maximum. Since they were recorded on one and the same monochromator under identical conditions, they have identical scales along the ordinate axis and make it possible to see the enormous difference in the spectral properties of these and other cathodes, in particular the differences in the magnitude of the quantum yield. Curve 3 of the same figure shows the distribution of energy in the spectrum of an incandescent lamp, measured directly behind the exit slit of the monochromator (i.e., without corrections for dispersion). This group of curves makes it possible to establish that the causes of the high sensitivity of these and other cathodes are quite different. For the oxygen–silver–cesium cathode, the high sensitivity is a consequence of the great extent of the curve along the wavelength axis, especially in that region of the spectrum where the radiation energy is large. In antimony–cesium cathodes the high sensitivity is due exclusively to a high quantum yield in a comparatively narrow (and disadvantageous in terms of the amount of energy) region of the spectrum. This property of the cathodes is one of the most important circumstances opening up new possibilities for their use, possibilities not realizable with photoelements having oxygen–silver–cesium cathodes, as will be explained below.
Fig. 3. Distribution of spectral sensitivity for Sb-Cs (1) and Cs-O-Ag (2) photoelements on one scale for photocurrent and (3) energy distribution in the spectrum of an incandescent lamp
At the present time, as can be seen from the preceding discussion, there is still no quantitative theory of the antimony–cesium photocathode. The first attempt in this direction is the work of S. Yu. Lukyanov, reported by him at the Kiev Conference on Cathode Phenomena¹⁶, where he tried to apply Fowler’s method to the case of a semiconductor cathode. The starting point is Wilson’s model²¹ of an electronic (see above) semiconductor containing impurities. Making use of the usual assumptions, namely, that the probability of absorption of a quantum of frequency \(\nu\) (in a narrow frequency interval) by an electron with energy \(\varepsilon\) is a constant quantity, and that the transmission coefficient of the surface for electrons with energy greater than the energy of the potential barrier is equal to unity and is equal to zero in the opposite case, and also using Maxwellian statistics for the electrons inside the semiconductor, Lukyanov arrives at the following formula for the photocurrent \(i\) as a function of the light frequency \(\nu\) and
ANTIMONY-CESIUM CATHODES AND PHOTOCELLS
surface temperature \(T\):
\[ i = a\sqrt{n_0}\, T^{\frac{5}{4}} e^{\frac{h\nu}{kT}-\frac{\frac{V}{2}+b}{kT}} \]
or, in a form convenient for experimental verification,
\[ \ln \frac{i}{T^{\frac{5}{4}}} = \ln a\sqrt{n_0} + \frac{h\nu}{kT} - \frac{\frac{V}{2}+b}{kT}, \]
where \(a\) is a constant, \(n_0\) is the number of impurity atoms in \(1\ \mathrm{cm}^3\), \(T\) is the absolute temperature of the surface, \(b\) is the work function, and \(V\) is the width of the forbidden band between the band of occupied levels and the conduction band.
Experimental verification has shown the applicability of the formula given above in the case when \(\nu \simeq \nu_0\), i.e., in the region of the spectral curve nearest to the red boundary. At shorter wavelengths strong discrepancies appear. This permits the conclusion that, near the threshold, the photoeffect of antimony-cesium cathodes is due to conduction electrons, whereas the emission mechanism for the region of the selective maximum has a different and as yet unclear character. Undoubtedly this is selective emission from some homogeneous emission centers, but at present there is no possibility of judging what these centers are.
III. PHOTOCELLS WITH ANTIMONY-CESIUM CATHODES
In this section we shall speak almost exclusively about vacuum antimony-cesium photocells, since one of the most important results of the development of antimony-cesium photocathodes was the possibility of replacing the gas-filled photocells used up to now by vacuum ones, owing to their high integral sensitivity, which presents very substantial advantages.
1. Integral sensitivity. The sensitivity of the first specimens of photocells with Sb-Cs cathodes was not great and usually amounted to from \(10\)—\(15\) to \(30\ \mu\mathrm{A}/\mathrm{lm}\). However, very soon, as a result of selecting the conditions for treating the antimony layer with cesium vapor, it proved possible to raise the upper limit of sensitivity considerably, to \(50\)—\(60\) and even \(80\ \mu\mathrm{A}/\mathrm{lm}\). A further increase in sensitivity was achieved by sensitizing the cathode with oxygen at very low pressures \((\sim 10^{-4}\)—\(10^{-3}\ \mathrm{mm\ Hg})\), which, in the light of the investigation by S. S. Prilezhaev cited above, should be regarded as the creation of an optimum surface coating as a result of oxidation of the excess free cesium adsorbed on the surface.
The sensitivity of modern antimony-cesium cathodes, even under conditions of serial manufacture (laboratory of the Moscow Electrolamp Plant), is \(100\ \mu\mathrm{A}/\mathrm{lm}\), while record specimens have a sensitivity of up to \(150\)—\(175\ \mu\mathrm{A}/\mathrm{lm}\). This is the value of the sensitivity at a radiation-source temperature of \(2848^\circ\ \mathrm{K}\) and under conditions excluding secondary factors that might give an apparent increase in the sensitivity of the photocathode.
2. Volt-ampere and light characteristics
Photoelements with antimony-cesium cathodes exist in the form of two basic types, differing in the construction of the cathode. These two types of photoelements behave identically under conditions of low cathode illumination and exhibit very substantial differences at high illuminations (of the order of thousands of lux and above).
Fig. 4. Diagram of the structure of Sb-Cs cathodes:
a—cathode without a backing, b—cathode on a backing
In Fig. 4, a the construction of a cathode of the first type is shown (it is the first chronologically); Fig. 4, b refers to the second type. In both drawings \(K\) denotes the antimony-cesium layer, \(P\)—the cathode lead-in (usually a platinum wire), and \(A\)—the anode. The difference between the two cases consists in the presence, in a cathode of the second type, of a metallic backing \(M\) beneath the layer \(K\).
As is clear from what has been set forth above, the longitudinal resistance of the cathode, owing to the comparatively high specific resistance of the Sb-Cs layer and its small thickness, may be very considerable. In fact, according to data of various authors\(^{2,3}\), the resistance between two platinum leads located at a distance of 2–3 cm from one another may reach \(10^7 \Omega\). This means that, at sufficiently high illuminations of regions distant from the cathode lead, a significant potential drop may be established along the layer, as a result of which two effects may appear: first, the absence of saturation at insufficiently high (for the given value of illumination) anode voltages and, second, the bombardment of regions of the cathode more distant from the lead by electrons liberated from regions close to it. The first effect should manifest itself as an increase in the saturation potential of the photocurrent and a decrease in the output of the photoelement at low voltages on the anode; the second—as an absence of saturation even at high anode voltages, owing to the occurrence of secondary emission increasing with the velocity of the bombarding electrons (as is known\(^{22,10}\), antimony-cesium layers are effective emitters with a secondary-emission coefficient sometimes reaching\(^{23}\) 14, at a primary-beam velocity of about 1000 V).
Since the presence of a metallic sublayer in cathodes of the second type eliminates the potential drop along the cathode, such cathodes should be free of the above-mentioned peculiarities even in the case of high illuminations.
What has been said is illustrated by Figs. 5, a and 5, b, showing families of volt-ampere characteristics of both types of photoelements at different values of illumination; moreover, Fig. 5, a refers to a photoelement with a cathode without a backing. In these figures the numbers at the right of the curves denote a fraction—the relative scale along the ordinate axis (its change from one
...curve to another was necessary in order to place all the curves in one drawing and the whole number—the magnitude of the cathode illumination. As can be seen, at low illuminations the course of the volt-ampere characteristics for both types of photoelements is the same, but, beginning with an illumination of approximately 1,000 lx, the photoelement with a cathode without a backing shows those deviations of which we spoke above. For photoelements of the second type a certain increase in the saturation potential was also observed, but it is small. Saturation nevertheless undoubtedly exists. The observed increase is a consequence only of an increase in the space charges at the surface of the cathode.
Fig. 5. Group of volt-ampere characteristics of vacuum Sb–Cs photoelements.
a—with a cathode without a backing; b—with a cathode on a backing
A noticeable decrease, in some curves for the photoelement with a cathode on a backing, of the photocurrent with increasing anode voltage (the curves for high illuminations) has nothing in common with the properties of volt-ampere characteristics as such and is a consequence of fatigue of the cathode, of which we shall speak below.
Some authors believed that the anomalous course of the volt-ampere characteristics of antimony–cesium photoelements is a consequence of the suction action of the anode field3 or of the field of charges forming on the surface of the cathode24. These ideas, however, are completely refuted by more recent experiments25, in which all necessary precautions were observed. The role of the longitudinal resistance of a cathode without a backing can also be demonstrated by changes in the volt-ampere curves under the action of a change in the cathode temperature. Fig. 6 gives a group of such curves, from which it is seen that an increase in temperature entails an improvement in saturation.
The differences between cathodes of the two types also appear in the course of the light characteristics. At insufficiently high anode voltages (in relation to the magnitude of the illumination or of the luminous flux), these characteristics show a bend toward the axis of luminous fluxes (Fig. 7, curve 1). Conversely, for photoelements with cathodes on a backing, the light characteristic is rectilinear up to the highest luminous fluxes hitherto used (illumination of 40,000–50,000 lx), provided that cathode fatigue has been eliminated (Fig. 7, curve 2).
Fig. 6. Change in the form of the volt-ampere curves of a photoelement without a backing as a function of temperature (the photocurrent in relative units is plotted along the ordinate axis)
Fig. 7. Light characteristics of Sb-Cs photoelements: 1 — cathode without a backing, 2 — cathode on a backing
The scale along the ordinate axis in this figure is not the same for the two curves. Curve 1 shows greater curvature at the beginning, when the secondary emission is small owing to the insufficiently large potential drop along the layer, which is due to the relative smallness of the luminous flux. Further on, the slope decreases because an increasingly important role begins to be played by the addition to the anode current due to secondary emission. Let us recall that photoelements with cathodes without a backing can, at high anode voltages (\(\sim 1500\) V), have an (apparent) sensitivity of 400–500 \(\mu\)A/lm\(^3\), with a true cathode sensitivity of 50–60 \(\mu\)A/lm. The light characteristics of photoelements without a backing, at fluxes of fractions of a lumen (approximately up to 0.5 lm), invariably show strict linearity.
Thus, the differences between the two types of photoelements are due only to the significant resistance of the cathode along the layer and have, as is usually said, not a physical but an electrotechnical character. These differences are substantial only when working with large luminous fluxes, when preference should be given to photoelements with cathodes on a backing. In the region of small illuminations—this includes the most widespread applications: sound cinema, phototelegraphy, ordinary automation—photoelements of both types may be used with equal success. The differences between them in rela-
With respect to changes in the spectral properties, fatigue, the magnitude of the integral sensitivity, etc., none is observed.
The most essential features that advantageously distinguish antimony–cesium photoelements from the oxygen–silver–cesium ones that have been almost exclusively widespread up to the present time are, on the one hand, their high integral sensitivity and the course of the spectral characteristic—properties of which we have already spoken above—and, on the other hand, comparatively very weak fatigue.
Speaking of the fatigue of antimony–cesium cathodes, it is necessary to distinguish two cases: fatigue at low illuminations of the cathode and fatigue at high illuminations. The greatest practical interest is presented by the first case, since in most applications (sound cinema, phototelegraphy, etc.) one has to deal with small luminous fluxes—of the order of hundredths and even thousandths of a lumen. At luminous fluxes usual, for example, for sound cinema, fatigue in vacuum antimony–cesium photoelements manifests itself extremely weakly²⁶. In Fig. 8 there is shown a fatigue curve for vacuum antimony–cesium photoelements (curve 1, upper scale along the time axis), obtained as an average for a large number of photoelements and thus representing a characteristic of this type of photoelement.
Fig. 8. Course of fatigue at low illuminations (~20 lux): 1—for an Sb–Cs photoelement (upper scale of the time axis); 2—for a Cs–O–Ag photoelement (lower scale of the time axis)
This curve is remarkable in that it can approximately be divided into two parts: an initial part, where a relatively rapid fall of sensitivity with time is observed, and its continuation, where the decrease in sensitivity proceeds much more slowly. Similar behavior is also exhibited by the curves for photoelements of other types. In the same Fig. 8, curve 2 depicts (also as an average) the fatigue curve for a gas-filled photoelement with an oxygen–silver–cesium cathode (photoelements of the TsG-4 type) under the same illumination conditions. Here too there is a rapidly descending branch of the curve, passing into an almost horizontal section. Thus, qualitatively the course of fatigue proves to be the same, but quantitatively the difference proves enormous. In fact, if for a vacuum Sb–Cs photoelement the sensitivity decreases to ~50% in 4,000 hours, then for an oxygen–silver–cesium photoelement with gas filling, already in 40 hours the sensitivity falls to 30% of the initial value (the time scale for curve 2 is shown by the figures of the lower row). If, for convenience of quantitative comparison of the fatigue of different photoelements, we assume that the lowering of sensitivity (the initial branch of the curve) occurs according to a linear law (dashed lines in Fig. 8), and take as the quantity characterizing the fatigue of photoelements the “average fatigue,” defined as the decrease in sensitivity, expressed as a percentage of the initial value per unit time (hour),
then from the data of Fig. 8 it is easy to find that the average fatigue of vacuum antimony–cesium photocells is about \(0.0125\%/\)hour, whereas for gas-filled oxygen–silver–cesium cells it is approximately equal to \(1.75\%/\)hour. In other words, the former photocells fatigue 140 times more slowly.
The significance of this circumstance is very great, both in that it means a considerable saving when oxygen–cesium gas-filled photocells are replaced by vacuum antimony–cesium ones, and because it increases many times over the reliability of operation and the service life of photoelectric devices. Against such a replacement the argument is sometimes advanced that the sensitivity of vacuum Sb–Cs photocells (which on average may be taken as equal to from 90 to \(110\,\mu\mathrm{A}/\mathrm{lm}\)) is considerably lower than the sensitivity of gas-filled oxygen–silver–cesium photocells (\(150\)—\(200\,\mu\mathrm{A}/\mathrm{lm}\)). Since, however, only the values of the initial sensitivity are being compared here, it is clear that this argument does not withstand criticism. Indeed, from the curves of Fig. 8 it may be seen that after only 40 hours, of the high initial sensitivity of a gas-filled photocell only \(30\%\) remains, i.e. from 50 to \(70\,\mu\mathrm{A}/\mathrm{lm}\), whereas the Sb–Cs photocell during this time completely retains its initial sensitivity.
Fig. 9. Fatigue of Sb–Cs photocells at high illuminations (\(\sim 4\cdot 10^{4}\,\mathrm{lx}\)). The lower scale of the time axis is for curve 1. Along the ordinate axis is plotted the photocurrent at constant illumination.
In Fig. 8 it may also be seen that the sensitivity of vacuum antimony–cesium photocells does not begin to fall immediately after switching on, but at first shows a certain increase. Such a course of change in sensitivity is characteristic of the overwhelming majority of photocells, and is observed both at small and at large illuminations.
Fatigue at large luminous fluxes proceeds, in general, in the same way as at small ones, but considerably faster. Fig. 9 shows the course of fatigue at an illumination of the cathode of about \(4\cdot 10^{4}\,\mathrm{lx}\) (curve 2). Here, too, an initial increase in sensitivity is visible (curve 1, time scale on the lower scale), and the establishment of a final value of the sensitivity at a level of about \(50\%\) of the initial one.
The character of the processes responsible for the phenomena of fatigue of antimony–cesium cathodes is at present still completely unclear. In any case, it should be thought that it differs radically from that which, according to de Boer, occurs in complex cathodes of the ordinary type (for example, the oxygen–silver–cesium cathode), where the primary process is the excitation of crystals of the substance of the intermediate layer. On the basis of the data of the work of Prilezhaev and of those changes in the form of the spectral characteristics which occur as a result of fatigue\(^{26}\), we believe that the primary cause of fatigue in the present case consists
in the irreversible removal from the cathode surface into its bulk of an externally adsorbed film of cesium atoms, whose thickness determines the work function of the surface.
IV. APPLICATIONS OF ANTIMONY–CESIUM PHOTOCELLS
The grounds for the broad application of antimony–cesium vacuum photocells are three circumstances:
- The exceptionally high average integral sensitivity of Sb–Cs cathodes, three times exceeding the sensitivity of ordinary oxygen–silver cathodes, which until now had been considered the most sensitive.
- Exceptionally slight fatigue.
- The character of the distribution of sensitivity over the spectrum.
Here we shall not dwell on a whole series of advantages of Sb–Cs cathodes with respect to the technology of their manufacture, since from the standpoint of applications this question is secondary.
Above we have already shown how the combination of high sensitivity with slight fatigue makes the displacement of gas-filled photocells by vacuum ones unquestionable. We shall now dwell on those advantages which are afforded by the spectral characteristic of the antimony–cesium cathode in a number of cases.
As is seen from Fig. 3, the main features of the spectral characteristic of the Sb–Cs cathode are the localization of sensitivity in a comparatively narrow region of the spectrum and high values of the quantum yield. The small extent over the spectrum determines the high spectral selectivity of these cathodes at high sensitivity, owing to the large quantum yield. This circumstance (together with high sensitivity and constancy) proved to be exceptionally important for the use of the new photocells in phototelegraphy, since they made it possible for the first time to transmit colored objects (text written in colored inks, underlined documents, colored drawings, diagrams, geographical maps, etc.) in two-color (black-and-white) reproduction. As was reported by the initiator of the introduction of antimony–cesium photocells into phototelegraphy, A. M. Gurvich, in his report at the above-mentioned conference in Kiev[^27], the impossibility of correctly transmitting colored objects with oxygen–silver–cesium photocells is due to the fact that both white and colored paper reflect very strongly in the near infrared region, where, first, the sensitivity of cathodes of this type is comparatively large and where, second, the maximum of the energy in the emission spectrum of the incandescent lamp lies (see curves 2 and 3 in Fig. 3). This leads to the result that all colored fields, with the exception of black and blue (pencil), are transmitted as white.
The introduction of vacuum antimony–cesium photocells into sound cinema, now being carried out, promises a large saving of funds, since the new photocells can operate without replacement for at least 5–10 times longer than the gas-filled ones now in use. Taking into account that sound cinema is the most mass-produced[^27]
consumer of photoelements (tens of thousands of units per year), it is clear that the appearance of antimony–cesium cathodes is indeed a step forward and has great national-economic significance. However, here too the matter is not exhausted by this. As it turns out, the projected prospects for the development of cinema require the replacement of oxygen–silver–cesium photoelements by antimony–cesium ones because of differences in the spectral sensitivity of the former and the latter. The point is that one of the important tasks facing the film industry is the replacement of ordinary celluloid film (positive), with a bromo-silver emulsion, by new grades that are cheaper and safer from the standpoint of fire. Such films have been developed by NIKFI. In investigations1 of their spectral transmission it turned out, however, that the darkened areas of these films possess very high transparency in the region from 6000–7000 Å and farther toward long waves (see Fig. 10, curves 1–2), i.e., again in that region of the spectrum where the sensitivity of the oxygen–silver–cesium cathode is great and where the light source used (an incandescent lamp) gives a maximum of energy (Fig. 10, curves 3 and 4). This makes sound reproduction by means of oxygen–silver–cesium photoelements extremely difficult, if not altogether impossible. Conversely, in the case of antimony–cesium photoelements (Fig. 10, curve 5) the dark areas of the film prove to be opaque2.
Fig. 10.
1—spectral transmission (with allowance for ordinary viscose film); 2—the same, for exposed film; 3—spectral distribution of the sensitivity of a Cs–O–Ag photoelement; 4—energy distribution of an incandescent lamp; 5—spectral sensitivity of an Sb–Cs cathode.
(On the ordinate axis all three quantities are plotted in relative units)
CONCLUSION
In addition to the applications indicated above, antimony–cesium photoelements are successfully used in “talking paper” apparatus, in various automatic devices (where the constancy of their sensitivity is especially valuable). Antimony–cesium cathodes are also used in iconoscopes with image transfer[^28], where they make it possible to obtain considerably greater sensitivity and to achieve correct color rendition with greater success than is possible, for example, with transparent
ANTIMONY–CESIUM CATHODES AND PHOTOCELLS
[[unclear: beginning of sentence obscured]] silver–cesium photocells, which, according to [[unclear]] data,
[[unclear: one short line]] 6 lumens.
The production of antimony–cesium photocells was mastered by the [[unclear]] plant, which at the present time [[unclear]]
[[unclear]] such photocells under the brands TsG[[unclear]]
[[unclear]] for phototelegraphy and cinema projection, and TsV[[unclear]]
[[unclear]] sound-cinema installations.
Addition during proofreading
From the time when this article was submitted to press, about [[unclear]] years have passed, during which work on antimony–cesium photocells was considerably reduced for reasons of the circumstances of wartime. Nevertheless, considerable successes have been achieved in this field.
One of the essential new data on Sb–Cs cathodes was obtained as a result of an investigation of their spectral sensitivity in the ultraviolet region, carried out first by the author and subsequently in collaboration with A. E. Melamud. These investigations showed that an Sb–Cs cathode possesses high sensitivity to radiation with wavelengths at least down to 2500 Å, and very weak fatigue under the action of ultraviolet radiation as well. This made it possible to construct very simple, sensitive, and stable photoelectric indicators of ultraviolet radiation for the indicated wavelength region.
A second important result of this work was the author’s detailed investigation of the fatigue of Sb–Cs photocells at large light fluxes, which led to the creation of practically non-fatiguing photocells for light fluxes of the order of tens of lumens.
LITERATURE
- P. Görlich, Z. Physik, 101, 335, 1936; see also N. S. Zaitsev, Uspekhi fiz. nauk, 19, 278, 1938.
- N. N. Lusheva, Zhurnal tekhn. fiziki, 7, 1900, 1937.
- N. S. Khlebnikov and N. S. Zaitsev, Zhurnal tekhn. fiziki, 9, 44, [[unclear: year]].
- See, for example, Ya. G. Dorfman and I. K. Kikoin, Fizika metallov, GTTI, 1933.
- See, for example, N. R. Campbell, Phil. Mag., 12, 174, 1931.
- N. S. Zaitsev, Zhurnal tekhn. fiziki, 9, 661, 1939.
- See, for example, report by S. Yu. Lukyanov at the Kiev Conference on Cathode Phenomena, 1940, Fizicheskie zapiski AN USSR, 9, 1941.
- S. Yu. Lukyanov and I. S. Mazover, Zhurnal tekhn. fiziki, 9, 14[[unclear]], 1939.
- See: Materials of the Conference on the Photoelectric Effect and Secondary Emission at the Leningrad Industrial Institute, January 20–22, 1939.
- A. D. Morgulis and B. I. Dyatlovitskaya, Zhurnal tekhn. fiziki, 10, [[unclear: page]], 1940.
- P. G. Borzyak, report at the Kiev Conference on Cathode Phenomena, 1940, Fizicheskie zapiski AN USSR, 9, 173, 1941.
- See Fig. 14.
- N. S. Zaitsev, Zhurnal tekhn. fiziki, 9, 1441, 1939.
- S. G. Ley-Bur, [[unclear: remaining reference obscured]].
S. G. [[name as printed: Svomya]] and R. Zurman, Photoelements and Their [[title continues]].
S. Yu. Lukyanov, Proceedings of the Kiev Conference of the Academy of Sciences of the Ukrainian SSR, p. 168, 1941.
[[fragment as printed: “from” the 1st row]]
Khlebnikov and N. S. Zaitsev, Uspekhi Fizicheskikh Nauk, 19, 838.
Journal of Technical Physics, 9, 1175, 1939.
Journal of Technical Physics, 10, 943, 1940.
Uilezhaev, Proceedings of the Kiev Conference of 1940, Physical Notes of the Academy of Sciences of the Ukrainian SSR, 9, 163, 1941.
Proc. Roy. Soc., 133, 458, 1931; 138, 277, 1932.
Khlebnikov, Journal of Technical Physics, 9, 367, 1939.
Binin, Proceedings of the Kiev Conference of 1940, Physical Notes of the Academy of Sciences of the Ukrainian SSR, 8, 163, 1941.
Ioffe and Yu. I. Lunkova, Journal of Technical Physics, 10.
Khlebnikov, Journal of Technical Physics, 10, 1908, 1940.
Khlebnikov and P. A. Sinitsyn, Journal of Technical Physics.
Gurevich, Proceedings of the Kiev Conference of 1940, Physical Notes of the Academy of Sciences of the Ukrainian SSR, 9, 223, 1941.
For example, Uspekhi Fizicheskikh Nauk, 24, 294, 1940.