Recent Advances in the Field of Photovoltaic Cells*

V. Klug, Berlin

Introduction

This article gives a survey of photovoltaic cells in their present stage of development, with especially great attention being devoted to cells based on the external photoelectric effect. In the first section a comparative juxtaposition is given of three different kinds of photovoltaic cells, namely: photoresistive cells, cells with a barrier layer, and cells with the external photoelectric effect. In the second section, in view of the broad field of application of photovoltaic cells of the latter type, our present ideas concerning the relation between spectral sensitivity and the structure of the surface layer are briefly set forth. Although it would still be premature at present to establish general laws, nevertheless certain considerations may be expressed that can serve as a guiding principle.

The designs of photovoltaic cells are determined first of all by the conditions of the corresponding field of application. It should be noted that here nearly all possibilities have long since been exhausted. There remain only certain new or improved designs. The discussion concerns, chiefly, devices in which separated surfaces and the method of electronic imaging are used. Among them should be mentioned: the light meter, Zworykin’s iconoscope, and Farnsworth’s dissector.

1. Comparative Juxtaposition of Three Kinds of Photovoltaic Cells

The oldest photoelectric device that permits the conversion of oscillations of light energy into oscillations of current is the selenium photoresistive cell. Cells of this type at present are not of any very great technical interest. The explanation of the complicated relationships that occur in the photoelectric conductivity of selenium

* Z. techn. Phys., 16, 181, 1935. Translated by D. V. Zernov.

in recent times has practically not advanced further.^1 Another type of photoelement-resistance—the thallofide photoelement—because of its special properties has found for itself a particular, comparatively narrow field of application. Based on the internal photoeffect, the photoelement-resistance, in contrast to photoelements with alkali metals, which will be discussed below, is a light-sensitive resistance independent of the direction of current. Photoelements with a blocking layer^2 are photoelements in the full sense of the word, since under the action of light they generate an e.m.f. At present two types of photoelements with a blocking layer are encountered in technology: cuprous photoelements—front-wall and back-wall (Lange and Schottky)—and the front-wall selenium photoelement (Falkenthal, Lange, Bergmann).

The physical processes underlying the action of photoelements with the external photoeffect are different from those for photoelements-resistances and photoelements with a blocking layer. This explains the differences in their behavior from the behavior of photoelements-resistances and photoelements with a blocking layer, which have something in common in their properties. Each type of photoelement has found its own fields of application, determined by its properties. These fields cannot, of course, in all cases be precisely delimited. In most practical cases, however, such delimitation is possible without particular difficulty. In any event, for this it is necessary to compare all the properties of the photoelements with one another. Thus, for example, often only one property, namely its sensitivity, is indicated for characterizing and evaluating a photoelement. This, however, undoubtedly leads to incorrect conclusions. In choosing a photoelement for any particular purpose, it is necessary to keep in mind all the data set out in the table below. These data should be considered in somewhat greater detail.

In three columns are brought together the properties of the caesium photoelement, the thallofide photoelement-resistance, and the selenium photoelement with a blocking layer. The comparison is made for: 1) the long-wavelength limit, 2) quantum yield, 3) current sensitivity referred to a definite light source, 4) voltage sensitivity, 5) linearity of the dependence of the photocurrent on illumination, 6) inertia. In the first line is given the “relative” red limit of sensitivity for the above-mentioned types of photoelements. An exact indication of the limiting wavelength is impossible. It is obtained differently, depending on the arrangement of the measuring setup. Comparing the data given in the table with the spectral characteristics (Fig. 2), one can see that of all three types the thallofide photoelement is the most sensitive to red light. Following it comes the caesium photoelement, and in third place is the selenium photoelement with a blocking layer. Thus, when searching for the photoelement most sensitive to red light, one should choose the thallofide photoelement, if within the limits of the given case there is no need to take other properties into account.

Comparison of Three Types of Photocells

In the second row are given the percentage values of the observed quantum yield. These quantities refer not to the absorbed light, but to the incident light, since only this case is of interest for technical purposes. The value given for photocells with an external photoeffect (2%) refers to the selective maximum of a cesium photocell at the red end of the spectrum. This value can at present easily be achieved in technical photocells. In the literature, cases of still higher values are cited.^3 For absorbed light, the quantum yield may be taken as equal to 100%. Fleischer^4 found, for transparent gold foil with adsorbed potassium, a value of 37.9%. For photoresistor cells, an indication of the quantum yield has no meaning, since the primary photoelectric process gives rise to secondary phenomena. Calculations, depending on the magnitude of the applied voltage, have given values greater than 100%. For absorbed light, a quantum yield equal to 100% was established in the case of the primary photoelectric current of diamond.^5 It should be expected that this applies to all primary processes in the internal photoeffect. The value given for the photocell with a blocking layer is taken from the work of Kippenteiger.^6 It refers to the maximum of the spectral curve.

In connection with the values of the quantum yield is the current sensitivity given in the third row. In engineering practice it is customary to specify current sensitivity in microamperes per lumen when illuminated by an incandescent lamp. By contrast, even if the source

* \(T_w = 2600^\circ \mathrm{K}\).

one cannot object that the light was chosen quite arbitrarily, since the spectral distribution of the energy emitted by the light source is determined quite precisely by specifying the temperature. An inaccurate specification of the temperature entails large deviations in the value of the sensitivity, especially for photocells sensitive to red light. Detailed investigations of this question were undertaken by Koller.^7 In the table given above are listed the values of current sensitivity that can at present be obtained with a high degree of reliability. Higher values, which are sometimes obtained but are accidental in character and difficult to reproduce, are not given here. The current sensitivity for photocells with the external photoelectric effect increases if a gas filling is used. The amplification factor thus achieved depends on the field distribution, on the voltage, and on the intensity of illumination. Low illumination intensities permit higher voltages, which leads to higher amplification factors. These relations can best be established for each type of photocell by taking a family of characteristics. The highest value of the sensitivity of a gas-filled photocell is attained where the Townsend discharge, free of space charges, passes into a glow discharge accompanied by the formation of space charges.^8 The same considerations may be expressed with regard to the voltage sensitivity given in the next line. The magnitude of the voltage sensitivity determines the possibility of amplifying photocurrents and is therefore important for most technical applications. This quantity is given in volts per millilumen arising at the photocell when a resistance is connected in series with it. The choice of this unit is based on the magnitude of the values actually observed. In first place here stands the photocell–resistance, in second place—the photocell with the external photoelectric effect. If the latter is filled with gas, still higher values can be obtained. The voltage sensitivity of a photocell depends on the magnitude of the resistance connected in series with it. The optimum voltage sensitivity is obtained at a definite external resistance. This latter is determined from the internal resistance of the photocell at the given illumination.^9 The sensitivity of a selenium photocell with a blocking layer is almost \(10^3\) times lower than the sensitivity of a photocell with the external photoelectric effect. As a consequence, they are practically used only when connected directly in series with a low-resistance galvanometer. The next line of the table gives a comparison of the proportionality of the photocurrent to the intensity of the incident light. In first place stand photocells with the external photoelectric effect, which is what makes possible their broad technical use. Linearity is preserved here over wide limits.

The small deviations found by Ives and others^10 for photocells with certain photocathodes are of no significance

for technical purposes. If necessary, they can be determined by preliminary calibration. In any case, in order to obtain strict linearity, certain conditions must be fulfilled in the construction of the photovoltaic cell.^11 The use of gas-filled photovoltaic cells introduces known limitations in this respect.^12 For a selenium photovoltaic cell with a blocking layer there is also sufficient linearity. Deviations appear only over a wide range of changes in illumination. The requirement of linearity is least satisfied by the thalofide photovoltaic cell, although it has been considerably improved in comparison with its predecessor (the selenium resistance-type photovoltaic cell).^13

From the last line it is clear that the external photoeffect* and the photoeffect of the blocking layer have no inertia, i.e., the electrical action follows the light action without delay. This is not the case for the resistance-type photovoltaic cell, since the inertia-free primary effect itself is directly associated with an inertial secondary effect. For this reason the resistance-type photovoltaic cell is not suitable for fast-acting installations.

Fig. 1. Selenium photovoltaic cell with a blocking layer (after Lange)

The effect underlying the operation of a photovoltaic cell with a blocking layer is in itself inertia-free. Here, however, a noticeable inertia arises, caused by the inevitable relatively large self-capacitance. The order of magnitude of the latter is about \(1 \ \mu F\). Thus, in practice, only the photovoltaic cell with an external photoeffect remains an inertia-free photoelectric device.

Summarizing the data presented in the table, it may be said that for most practical purposes it is advisable to choose photovoltaic cells of this type. In its favor are its lack of inertia, the proportionality between photocurrent and illumination, and its high voltage sensitivity. In addition, among these properties is the possibility of varying the spectral sensitivity over wide limits by the appropriate choice of photocathode. These advantages outweigh the single shortcoming of the photovoltaic cell with an external photoeffect—its low current sensitivity. The next place in the number of practical applications is occupied by the selenium photovoltaic cell with a blocking layer; here one may point to illumination meters for photographic purposes and technical photometry. In Fig. 1 a special design of a selenium photovoltaic cell with a blocking layer is shown. The resistance-type photovoltaic cell in na-

* On the inertia of gas-filled photovoltaic cells compare Leo and Möller.^14

…at present finds almost no practical application. The reason for this is its inertia and low stability. It is used as a thallofide photoelement when it is necessary to have high sensitivity to red and infrared light, and when its shortcomings do not interfere too much with the matter at hand. In concluding the review of the numerical values given in the table, it should once again be noted that they serve exclusively for orientation as to the order of magnitude. Discrepancies in the values for a large number of photoelements are inevitable. Figure 2 presents curves of spectral sensitivity. They are characteristic curves, the course of which can be obtained quite reliably. The curves extend into the ultraviolet region as far as this can now be regarded as established. The maximum closest to the red boundary is taken as 100%.*

Fig. 2. Spectral characteristics of three types of photoelements

Labels in the figure: relative sensitivity, %; wavelength; selenium photoelement with blocking layer; cesium photoelement; thallofide photoelement.

The course of the curve for the thallofide photoelement in the visible region is associated with inaccuracy; strictly speaking, these photoelements should not be subjected to the action of visible light, since in this case photochemical processes take place in the semiconductor.

2. Photoelements with external photoeffect

The fact that, of the three types named here, photoelements with external photoeffect have the widest field of application, and also that in recent times they have been substantially improved, compels us to consider them here in somewhat greater detail. A whole series of earlier investigations showed that pure alkali metals usually give the so-called normal photoeffect1, i.e., the number of electrons released per unit of incident light—

light energy, continuously increases with the transition to shorter waves. According to more recent investigations, however, this principle proves to be violated for cesium in the spectral region accessible at the present time.^16 Cesium, even in its purest form, gives several maxima of spectral sensitivity in the ultraviolet region. The purity of the metal was to a high degree guaranteed by the fact that the so-called long-wavelength selective maximum located in the visible region was absent or was very weakly expressed. This criterion of purity was established for potassium^17 and applied to cesium.

In the manufacture of technically applicable photocathodes, use is made not of a pure alkali metal in the form of a thick layer, as was done earlier, but of so-called complex photocathodes formed in the photocell. The old methods of sensitizing the layer, for example by treating the alkali metal with a glow discharge in hydrogen,^18 have now been abandoned. The vapor pressure of such surface layers is too high and leads, at least over a long interval of time, to insufficient stability of the photocell. Modern photocathodes in most cases have the same structure: metal substrate—intermediate layer—alkali metal. An intermediate oxide layer is deposited on the metal substrate (in most cases silver), and a thin layer of alkali metal is adsorbed on it. The vapor pressure for such surface layers is considerably lower, which is a prerequisite for good stability and a long lifetime of the photocell. From the sometimes used abbreviated notation T—M₂O—M one can discern only the scheme. Nothing can thereby be said as to what quantities of the substrate metal or alkali metal are incorporated in the intermediate layer. The most preferred combination is the cathode Ag—Cs₂O—Cs. The red limit* for this layer, with the method of preparation employed, extends to 14000 Å. In the study of such photocathodes it proved that the characteristic position of the selective maximum and of the red limit is determined by the metal substrate and the selected intermediate layer. In Fig. 3 a comparison is given of the spectral-sensitivity curves of the cathodes Ag—Cs₂O—Cs, Cu—Cs₂O—Cs, and Ni—Cs₂O—Cs. Without touching on the short-wavelength maximum in the ultraviolet, we shall discuss only the so-called long-wavelength selective maximum in the visible region of the spectrum. It is precisely this maximum which, for technical purposes, must be sharply expressed and located in the long-wavelength region. The largest “selective” maximum and the red limit most shifted into the infrared region are possessed by the cathode Ag—Cs₂O—Cs. For the combination Cu—Cs₂O—Cs the red limit and the maximum are shifted toward shorter wavelengths. In addition,

* By the “relative” red limit is meant everywhere the limiting wavelength determined by the measuring setup (defined by the last actually recorded point of the curve).

selective maximum here is considerably lower. The combination Ni—Cs₂O—Cs gives no selective maximum, and the red boundary is shifted even further into the short-wavelength region. These curves show with complete obviousness the great influence of the metallic substrate on the absolute output, the selective effect, and the red boundary. The values of sensitivity to red light obtained for cesium (on silver) photocathodes are presented in Fig. 4. From this comparison it follows that the spectral position and height of the maximum in the red region vary from one photocell to another. This discrepancy becomes smaller the more homogeneous the manufacture of the photocells is. An uneven cathode layer, especially a thick intermediate layer, as a rule leads to a steep maximum. This is understandable, since the number of centers selectively absorbing light becomes especially large on such a surface. The properties of cathodes with a thick intermediate layer were studied in detail by Teves and de Boer. With their aid a large output can be obtained. Of course, one must take into account the accompanying undesirable phenomenon of fatigue. This latter appears the more sharply, the shorter the wavelength of the incident light.

Fig. 3. Influence of the metallic substrate on the spectral distribution of sensitivity

Fig. 4. Course of the spectral-sensitivity curves in the infrared region for cesium photocells fabricated in various ways

Complex photocathodes are characterized by the fact that, along with the long-wavelength maximum, they also possess another spectral maximum in the ultraviolet region¹⁹ (compare Fig. 3).

At the present time attempts are being made to relate the course of the spectral sensitivity to the structure of the cathode. In particular, much attention is being paid

is an explanation of the selective spectral photoeffect. Circumstances have lately become still more complicated, since, as was already mentioned above, complex cathodes exhibit multiple selectivity. Unfortunately, we do not yet have at our disposal parallel measurements of optical absorption in such layers that could facilitate our explanation of these results.

First of all, let us present the explanation of the selective photoeffect proposed by Gudden and Pohl.²⁰ It is based on Fleishman’s experiments.²¹ Gudden and Pohl say: the matter consists only in the fact that, when the spectral curve is taken, the optical absorption spectrum of the absorbing atoms or molecules is measured electrically. Fleishman measured the absorption of light by thin layers of alkali metals deposited on quartz, in polarized light, and established selective absorption of light for \(E_{\parallel}\), the maximum of which falls in the region of the spectral photoelectric sensitivity curves of alkali photocells. In Fig. 5 a similar curve is given for potassium. This explanation is suitable for the long-wave maximum of complex cathodes. If, by special measures,²² the surface layer of the alkali metal is removed, the long-wave maximum disappears. The short-wave maxima become weaker, but remain in their former positions.

Fig. 5. Absorption of light by a thin layer of potassium in the region of the selective surface photoeffect (after Fleishman)

Campbell and Fowler²³, in their theory, assume that the probability of escape is especially large for electrons with definite velocities. For this to occur, a definite distribution of potential must prevail at the surface. The required distribution is realized, for example, with an electronegative substance taken as the intermediate layer. This theory has so far been unable to find any satisfactory experimental confirmation. It is capable only of explaining, from the qualitative side, the existence of multiple selectivity.

De Boer and Teves²⁴ somewhat modified the theory of Campbell and Fowler. They believe that the atoms of the alkali metal adsorbed on the surface are ionized by light. When replenished with electrons from the substrate, its selective electronic conductivity produces a maximum in the sensitivity curve. Zuurman²⁵, in contrast to this, expresses the assumption that the selective photoeffect is due to the selective absorption of light by adsorbed-

surface atoms. As a result of this, electrons are emitted by the metallic underlayer, receiving energy through collisions of the second kind with optically excited atoms. This theory is closely related to Wolf’s old hypothesis.^26

The theory of Ives and Briggs^27 is comparatively simple and clear. It attempts to explain the entire characteristic course of the spectral response curve. The authors can, for example, explain very well the course of the response curve for thin invisible layers of alkali metals on a metallic mirror. In doing so they start from the simplified assumption that 1) all photoelectrons emerge only from the metallic film and 2) the effective intensity of light in the metallic film is equal to the intensity prevailing at the boundary surface

Fig. 6. Photoeffect and electric-field intensity on a mirror surface (after Fink)

vacuum—metallic mirror.

To test this theory, experiments were carried out on a silver mirror with a film of alkali metal deposited on it. These experiments provide an explanation of the effect in the sense that, for it, the intensity of the electric vector directly at the silver mirror is essential. Fig. 6 gives a diagram of the intensity of the electric vector on the surface of the mirror and, in parallel with it, the photoeffect as a function of wavelength. Verification of this theory is impossible on other surfaces, apart from mirror ones.

3. Designs of photocells with an external photoeffect

Photocells of this type are given the most varied form,^28 determined primarily by their purpose. From the many designs, two main groups may be distinguished. The first group comprises photocells that must operate as photometers in the full sense of the word. Here the action of the photocell as a black body is required, strict proportionality between the photocurrent and the illumination, and good insulation between anode and cathode. Thus there arises a more or less spherical form of photocell, the entire inner surface of which, except for the window through which light enters, is the cathode. Proportionality is achieved especially well if the cathode is located at the center, since then the arrangement...

the effect of space charges begins. Good insulation is achieved by increasing the glass gap between the anode and the cathode (a corrugated stem). Such photocells are shown in Figs. 7 and 12. The insulation resistance for them is \(10^{13}\)—\(10^{14}\ \Omega\). The second group includes all other photocells, for which observance of the three above-mentioned requirements is not unconditionally necessary. Here a large part of the surface of the walls may remain free. This entails, as a rule, insufficiently strict linearity, at least over a large range of changes in the intensity

Fig. 7. Spherical technical photocell with increased insulation

Fig. 8. Cylindrical photocell

of the incident light. Here it is necessary to take into account the disturbing action of charges on the walls. Such photocells cannot be used as photometers. They are more suitable for purely technical purposes, where what is required first of all is high voltage sensitivity. The insulation between the anode and the cathode need not be especially high, particularly if such photocells are intended for light of variable intensity. Thus, Fig. 8 shows a cylindrical photocell used in light relays, and Fig. 9 shows the so-called front photocell, used when illumination comes directly from above. These latter make it possible to arrange a large number of photocells close to one another in a single plane. In some cases the use of photocell-

tubes with a large surface. Then a correspondingly enlarged design is chosen, shown in Fig. 8. In conclusion, one more design of a photocell may be indicated, shown in Fig. 10. This photocell has no special window for the entrance of light, but the photocathode is made transparent and covers the entire inner surface of the bulb of the photocell. These photocells are very suitable for photometry of the sky, for volumetric photometry, and for special light relays. Transparent cathodes for photocells had already been used by Fleischer.^29

As has already been mentioned above, at present cesium is used predominantly for photocathodes. However, for special purposes it may prove necessary to use some other alkali metal, namely when it is necessary to obtain a selective maximum of \(r\) in some other part of the spectrum besides the infrared. This, for example, is required in measuring coloration and in measuring luminous intensity. For this it is necessary to have a sensitivity curve of the photocell that coincides as fully as possible with the sensitivity curve of the human eye. According to Zhurman’s proposal,^30 for this one may use the potassium photocell described by him together with Schott filters GG11 (1 mm) and GG12 (20 mm), which gives a sensitivity curve almost coinciding with the sensitivity curve of the human eye. The results that are obtained at present with the aid of a potassium photocell with a cathode of the type Ag—K\(_2\)O—K are seen from Fig. 11. Complete coincidence has not yet been achieved. The use of this combination is necessary only when amplification cannot be dispensed with.

Fig. 9. Front photocell

Fig. 10. Spherical photocell with transparent cathode

photocurrent, for example when transmitting measured values over a distance. In other cases one may use the tried-and-tested Dresler arrangement,^21 which employs a selenium photocell with a blocking layer. Fig. 12 shows yet another type of photocell belonging to the first group mentioned above. Here the photocell is made of hard glass with a fused-in quartz window, making it possible to carry out measurements from the short-wave ultraviolet \((1800\,\text{Å})\) to infrared rays with a wavelength of \(13000\,\text{Å}\). This photocell covers the widest spectral region of all photocells known up to now. It will probably find application in ultraviolet dosimetry. Until now, cadmium photocells have been used here almost exclusively. The cesium photocell, in comparison with the cadmium photocell, has a considerably greater absolute sensitivity.

Fig. 11. Coincidence of the sensitivity curve of a potassium photocell \((\mathrm{Ag—K_2O—K})\), equipped with Schott filters GG 11 and GG 12, with the sensitivity curve of the human eye

4. The Newest Photoelectric Devices

Here one should also mention devices that have appeared recently and that make use of the external photoelectric effect. Among them are: the light counter, Zworykin’s iconoscope, Farnsworth’s dissector, and the photoelectric image converter.

The light counter is a special type of gas-filled photocell. With the aid of a gas-filled photocell of ordinary design and an amplifier tube connected in an electrometer circuit, a sensitivity of about \(1.7 \cdot 10^{-15}\ \mathrm{cal.\ sec^{-1}\ cm^{-2}}\) can be achieved. Elster and Geitel^32, with the aid of a gas-filled photocell under special conditions, obtained a sensitivity as high as \(7.2 \cdot 10^{-17}\ \mathrm{cal.\ sec^{-1}\ cm^{-2}}\). They used a principle similar to that of the light counter.

on the photocell they gave values very close to the ignition potential. Then, under weak illumination, the electrometer is charged not continuously, but by pulses. Each pulse is proportional to a definite number of electrons emitted from the photocathode and, at the same time, to a definite number of light quanta incident upon it. In practice, a light counter is a Geiger–Müller counter whose wall is made in the form of a photocathode for visible light. In this case each

Fig. 12. Photocell made of refractory glass with a sealed-in quartz window (for photometry in the region from 1800 to 1300 Å)

photoelectron emerging from the wall of the tube is registered. At the present time a sensitivity of about \(2 \cdot 10^{-19}\ \mathrm{cal.\ sec.}^{-1}\ \mathrm{cm}^{-2}\) has been attained. The sensitivity of the human eye, whose threshold in the blue-green part of the spectrum is approximately \(2 \cdot 10^{-16}\ \mathrm{cal.\ sec.}^{-1}\ \mathrm{cm}^{-2}\), is thus exceeded by \(10^4\) times. The light counter is the most sensitive photometer that we possess at present.

Specially for the purposes of television, Zworykin developed the so-called iconoscope \(^{34}\). Its considerable advantage lies in the fact that it does not require mechanical scanning of the image. The image is projected onto a specially prepared flat electrode possessing photoelectric sensitivity. This electrode is a capacitor,

one plate of which consists of a continuous layer of silver, the spacer (dielectric) of a thin mica plate, and the second plate of a photocathode divided into small elements. When this surface is illuminated, the individual elements of the photocathode are charged in proportion to the intensity of the light incident upon them. An electron beam scans the surface of the photocathode, discharging the individual elements, which have been charged to a positive potential corresponding to the intensity of the light incident upon them. An amplifier is connected to the silver plate of the capacitor; it amplifies the pulses of discharge current from the individual elements and transmits them to the subsequent apparatus. The sensitivity of this device is greater by 6 orders of magnitude than that of those used hitherto.

Farnsworth35 proposed another device related to this one, since here too the mechanical parts are replaced by a moving electron beam. He projects the image to be transmitted onto a transparent photocathode. The different parts of the photocathode then emit different numbers of electrons, according to the intensity of the light incident on the given portion of the cathode. Next this “electronic image,” by means of electron optics, is thrown onto a receiving surface that has, in the middle, a small square window and, behind this window, a collecting electrode connected to an amplifier. If now, with the aid of an alternating magnetic or electric field, the electronic image is made to oscillate from side to side, then, in accordance with the elements of the image falling on the window, oscillations of current arise in the amplifier, reproducing the distribution of the light intensity on the photocathode. With the development of electron optics it also became possible to study photoelectron emission from surfaces. This was done by Pohl for the surface of platinum.36 Under uniform illumination of the surface, the local distribution of photoelectric emission can be observed on a fluorescent screen. In a similar way, when photocathodes of uniform sensitivity are used, especially alkali cathodes, optical images can be transformed into electron-optical ones.37 The photoelectric devices mentioned here make it possible to broaden the use of photography by extending it to red and infrared radiation.38

LITERATURE

1. B. Gudden, Lichtelektr. Erscheinungen, p. 195, J. Springer, Berlin, 1926.
2. R. Falkenthal (Berlin) proposed introducing for these photocells the designation “compound photocell.”
3. M. C. Teves and J. H. de Boer, Z. Physik, 83, 521, 1933.
4. R. Fleischer, Physik. Z., 32, 217, 1931.
5. B. Gudden and R. Pohl, Z. Physik, 17, 331, 1923.
6. K. O. Kiepenheuer, Die Sterne, No. 4–5, p. 81, 1934.
7. R. Koller, Rev. Sci. Inst. Dec. 1932, 760.
8. W. Kluge, Z. techn. Phys., 12, 591, 1934.
9. W. Kluge and H. Briebrecher, Z. techn. Phys., 12, 533, 1933.

1. G. Kortüm, Physik. Z. 32, 417, 1931.
2. R. Suhrmann, Z. wiss. Phot., 29, 156, 1930.
3. B. Gudden, Lichtelektr. Ersch., p. 127.
4. R. Seewig, Techn. Wiss. Abhandl. aus dem Osram-Konzern, 1931.
5. C. Müller, Physik. Z. 36, 111, 1935.
6. G. Wiedemann and W. Hallwachs, Verh. d. Deutsch. Phys. Ges. 16, 107, 1914; R. Fleischer and H. Dember, Z. techn. Phys. 7, 133, 1926.
7. W. Kluge, Z. Physik, 93, 636, 1935.
8. H. Simon and R. Suhrmann, Lichtel. Zellen, pp. 23 and 32, Berlin, J. Springer, 1932;
9. J. Elster and H. Geitel, Ann. d. Phys., 48, 625, 1892.
10. W. Kluge, Physik. Z. 34, 833, 1933.
11. B. Gudden, Lichtel. Erscheinungen, Springer 1928, pp. 109—110. R. W. Pohl, Naturwiss. 14, 217, 1926.
12. R. Fleischmann, Nachr. v. d. Ges. d. Wiss. z. Göttingen, Fachgruppe II, 1931, p. 252.
13. W. Kluge, Z. Physik (in press).
14. R. H. Campell, Photoelectric Cells and their Applications, Discussion of the Physical and Optical Societies of London, p. 10, 1930; R. H. Fowler, Proc. Roy. Soc. 128, 123, 1930.
15. J. H. de Boer and M. C. Teves, Z. Physik, 73, 192, 1932; 83, 521, 1933.
16. R. Suhrmann, Erg. d. exakt. Naturw. XIII, 191, 1934.
17. A. Th. Wolf, Z. Physik, 52, 158, 1929.
18. H. E. Ives and H. B. Briggs, Phys. Rev. 38, 1477, 1931.
19. R. Fleischer and H. Teichmann, Die Lichtelektrische Zelle und ihre Herstellung, Dresden 1932; H. Geffcken and R. Richter, Die lichtempfindliche Zelle als technisches Steuerorgan, Berlin, Tempelhof, 1933.
20. R. Fleischer, Z. techn. Phys., 2, 92, 1932.
21. H. Simon and R. Suhrmann, p. 251.
22. A. Dresler, Das Licht, 3rd year of publication, No. 2, 41—43, 1933.
23. J. Elster and H. Geitel, Physik. Z. 17, 268, 1916.
24. B. Rajewsky, Physik Z., 32, 121, 1931.
25. V. K. Zworykin, Hochfrequenztechnik u. Elektroakustik, 43, 112, 1934.
26. A. H. Brolly, Elektr. Eng., 53, 1153, 1934.
27. J. Pohl, Z. techn. Phys. 15, 579, 1934.
28. G. Holst, J. H. de Boer, M. C. Teves and C. F. Veenemans, Physica 1, 297, 1934; W. Schaffernicht, Z. Physik, 93, 762, 1935; W. Heimann, ENT, p. 291, 1935; W. Kluge, Z. Physik, 91, 789, 1935.
29. J. Eggert, Die technisch wichtigsten lichtempfindlichen Systeme, Z. techn. Phys. 13, 441, 1934.

1. The curve for the photoelement with a blocking layer was supplied to the author by Falkenhagen; for the thallofide photoelement, by Flechsig, for which gratitude is expressed to them. ↩