A NEW METHOD FOR MEASURING SMALL LIGHT FLUXES USING PHOTOCELLS

N. S. Khlebnikov

For measuring small light fluxes, the most widely used devices are photocells with an external photoelectric effect, as practically the only ones suitable for use in direct-current amplification circuits and in electrometer circuits (and also as the most convenient when using alternating-current amplification). The sensitivity threshold of any such measuring circuit, i.e., the minimum value of the light flux that the circuit is still capable of registering, is determined by its noise level, which is set by the highest level of parasitic signal of some element of the circuit.^2 In principle, fluctuation noise caused by the shot effect of photoemission is unavoidable for a photocell with an external photoelectric effect. However, the attainment of the threshold sensitivity determined by the shot effect in electrometer circuits and with direct-current amplification is hindered by other phenomena. These are, above all: 1) the presence of leakage between the cathode and anode of the photocell along the glass of the bulb, 2) the presence of thermionic current from the photocathode at room temperatures.

These phenomena have the least effect in the case of antimony–cesium photocells, owing to the relatively high work function of the Sb–Cs cathode and the low vapor pressure of free cesium above the surface of such a cathode. The latter prevents the formation of a conducting film of alkali metal between the cathode and the anode, which, together with the high quantum yield, has earned these photocells the reputation of being the best measuring photocells.

However, the extent of the spectral response of these photocells, while allowing their use almost throughout the visible region and in the ultraviolet down to wavelengths of about 2000 Å, does not make it possible to use them in the red and near-infrared parts of the spectrum, for which the oxygen–silver–cesium photocell remains the only indicator.

It is precisely for photocells with such cathodes that the difficulties indicated are especially severe, since the work function of the Cs–O–Ag cathode is low (0.6–0.7 eV) and free cesium is only very weakly bound in it. To eliminate leakage, one increases the distance along the glass between the cathode and the anode by giving the neck ...

and the stem of the photocell a more or less complex shape, introducing highly insulating materials (quartz) into this part of the circuit and using guard rings. For suppressing the thermoelectron current there is, obviously, only one way—cooling the cathode, as was in fact used in connection with the application of electron multipliers for measurement purposes³. The impossibility of eliminating the thermoelectron current by any other method lies in the impossibility of separating thermoelectrons from photoelectrons, since both are emitted by the cathode and are collected at the anode of the photocell.

Fig. 1. Electron-optical image converter of Holst, de Boer, Teves and Venema.

It is possible, however, to implement a method of measurement in which, if not the thermoelectron and photoelectric currents themselves, then the effects produced by them are separated, which from the point of view of this type of application is entirely equivalent; at the same time the question of leaks loses its acuteness, since in the electrometric circuit itself an ordinary antimony-cesium photocell can be used. The basis of this method is the use of electron-optical conversion.

The electron-optical image converter was first described by Holst, de Boer, Teves and Venema⁴ (see Fig. 1) and was intended for converting an invisible optical image in infrared rays, projected onto a transparent (silver-oxygen-cesium photocathode) \(C\), sensitive to these rays, into a visible image on the fluorescent screen \(S\). This conversion is carried out as a result of the transformation at the cathode of the optical image into an electronic one and the reverse transformation on the fluorescent screen under bombardment by photoelectrons accelerated by the potential difference applied between \(C\) and \(S\). The optical image on the screen turns out to be somewhat blurred in comparison with that on the cathode, since the electron beams, in moving from the cathode to the screen, expand owing to the presence of tangential components of the velocities of the photoelectrons. Henneberg and Recknagel⁵ found that, for the converter of the Holst system et al., the radius \(\rho\) of the circle on the screen into which a point on the photocathode is transformed is expressed by the equality

\[ \rho = 4L \sqrt{\frac{\varepsilon}{V}}, \tag{1} \]

where \(L\) is the distance between the cathode and the screen, \(\varepsilon\) is the initial energy of the photoelectron, and \(V\) is the potential difference between the screen and the cathode. From this equality it is seen that the resolving power of such a converter (proportional to \(\frac{1}{\rho}\)) is the higher, the smaller

distance between \(C\) and \(S\), and the greater the accelerating potential difference. Since both reducing the first and increasing the second are feasible only within certain limits, a number of converter designs were proposed that make it possible to increase the resolving power by using more advanced electron optics\(^6\).

It is quite clear how, with the aid of an electron-optical converter, one can accomplish the tasks outlined above: to get rid of interference from the thermoelectronic current of the photocathode and from leakage in the measuring photocell. For this it is sufficient to assemble the circuit shown in Fig. 2, where \(A\) is the exit slit of the monochromator\(^*\), \(C\) and \(S\) are, respectively, the cathode and screen of the converter, \(D\) is a diaphragm located in the plane of the image of the screen \(S\) produced by lens \(L_2\), and \(P\) is the measuring photocell; lens \(L_1\) forms an image of slit \(A\) on the photocathode \(C\) of the converter.

Fig. 2. Schematic diagram of a measuring setup with an electronic converter.

With such an arrangement the measuring photocell can be chosen in accordance with the requirements imposed by the electrometric circuit, i.e., for example, it may be an antimony–cesium photocell, whereas the cathode of the converter may be selected solely on the basis of considerations of its desired spectral properties, i.e., for example, an oxygen–silver–cesium cathode may be used.

Such a separation of functions between the photocathodes that receive and measure the radiation, connected into a single whole through the radiation of the converter screen, further makes it possible, by means of diaphragm \(D\), to eliminate also the parasitic current produced by thermoelectronic emission. Indeed, in the image plane in which the diaphragm is located, there will exist a real image of a portion of the cathode \(C\) of the converter, containing also the image of the monochromator slit. The thermoelectronic current from the cathode will produce a more or less intense and more or less uniform (depending on the degree of uniformity of the surface and the value of the work function, as well as on the temperature of the cathode) glow of the screen, on which the image is—

\(^*\) It should be noted that this method of eliminating the influence of the thermoelectronic current has an advantage only provided that the measured light flux can be concentrated on a sufficiently small area of the converter cathode.

the slit image will appear as a narrow stroke brighter than the background. By choosing the size and position of the diaphragm so that it transmits radiation only from the portion of the cathode surface on which the slit image is located, we obviously obtain a reduction of the parasitic signal from the thermionic current in the measuring photocell by a factor equal to the ratio of the area of the converter cathode to the area occupied on it by the image of the monochromator slit. With the usual dimensions of both, this means a gain of two to three orders of magnitude and a corresponding lowering of the threshold sensitivity of the circuit.

Highly essential for the practical value of this method are the properties of the coupling element of this system—the screen of the converter. For convenient performance of measurements it is necessary, first of all, that there be a direct proportionality between the luminous flux incident on the converter cathode and the radiation of the screen. Special measurements show that such a relation exists over a wide range of cathode illuminations. To obtain reproducible measurement data, it is necessary that the voltage between the electrodes of the multiplier remain constant during at least one series of measurements, since the light output of the phosphor depends substantially on the energy of the bombarding electrons. As it turns out, this condition is satisfied with sufficient accuracy when the high-voltage rectifier supplying the converter is connected to the alternating-current mains through an ordinary ferroresonant voltage stabilizer.

Among the advantages of the method described should also be included the fact that the quantum yield of the system as a whole can be made greater than the quantum yield of the receiving photocathode. This is a consequence of the circumstance that, at the operating voltages of converters (kilovolts), one electron striking the screen gives rise to several tens of quanta. Therefore, despite light losses in the optics, the overall increase of the signal (so long as the device operates sufficiently far from the sensitivity threshold of the receiving photocathode, determined by the shot effect) may amount to several times. Taking this circumstance into account, the use of converters with antimony–cesium cathodes also appears expedient.

The use in the circuit of Fig. 2 of a Holst-type converter, i.e. an instrument with a transparent cathode and a small distance between the electrodes, is not convenient. First, transparent cathodes never have the maximum sensitivity attainable for the given type of cathode, and their sensitivity in the best case is approximately only half of this value; second, with such a converter the possibility is excluded of using thin-walled glass windows transparent to ultraviolet radiation, which greatly restricts the working wavelength interval. Both of these difficulties can be

eliminated provided that the distance between the cathode and the screen of the converter is considerably increased, but, as follows from relation (1), this will lead to a strong reduction in resolving power. Therefore, a rationally designed instrument must use one or another electron-optical system. In Fig. 3 a simple converter with purely electrostatic focusing is shown schematically, in which it is possible to use both opaque cathodes and glass windows transparent to ultraviolet that have proved so satisfactory. Here \(C\) and \(S\) still denote the cathode and the screen; \(l_C\) and \(l_S\) are the leads from them, \(W\) is the thin-walled window, and \(E\) is the focusing electrode. The electrostatic lens is formed by the field between it and the cathode. The optical axes of the system are shown by dashed lines.

Fig. 3. Electron converter with a thin-walled window and an opaque photocathode.

The applications of this method may evidently be very diverse. Initially it was intended for various work with a monochromator. Here, in addition to the improvements evident from the preceding discussion, it provides a fundamentally new possibility of visual observations in invisible regions of the spectrum. This is of interest, for example, when calibrating a monochromator by wavelengths or, conversely, when investigating unknown line spectra. For observations of this kind, it is advisable to replace lens \(L_2\) (Fig. 2) with a suitable eyepiece. Another broad field of application appears to us to be astronomy and astrophysics. In this case, by replacing the slit diaphragm \(D\) (Fig. 2) with a circular one, it is possible to reduce interference from the thermoelectronic current by approximately one further order of magnitude.

References

1. N. S. Khlebnikov and A. E. Melamid, Inf.-Tekhn. Byull. of Plant No. 632.
2. See, for example, F. Preisach, Wireless Eng., 16, 169, 1939.
3. J. Raichmann, Arch. Sci. Phys. et Nat. (5th series), 20, Sept., Oct., Nov., Dec., 1938.
4. G. Holst, J. H. de Boer, M. C. Teves and C. F. Veenemans, Physica, 1, 297, 1934.
5. W. Henneberg and A. Recknagel, Zschr. techn. Phys., 16, 230, 1935.
6. W. Heimann, Elektr. Nachr. Techn., 12, 68, 1935; W. Kluge, Z. Phys., 93, 789, 1935; M. v. Ardиinne, Electr. Nachr. Techn., 13, 230, 1936; V. K. Zworykin and G. A. Morton, J. O. S. A., 26, 181, 1936; V. K. Zworykin, Zschr. techn. Phys., 17, 170, 1936; W. Shaffernicht, Zschr. Phys., 93, 762, 1935; Jahrb. Forsch.-Inst. AEG, 4, 45, 1933–1935. See also V. Shaffernicht, UFN, 17, 491, 1937.