ELECTRON-OPTICAL SYSTEMS AND THEIR APPLICATIONS *

V. K. Zworykin

I. ELECTRON OPTICS OF IMAGE TRANSMISSION

The beginning of the development of electron optics may be dated directly to the moment of the discovery of the electrons themselves. The first investigators working in this field knew that a body placed in the path of a cathode ray casts a sharp shadow on a screen fluorescing under the action of this cathode ray, so that, at least in this respect, cathode rays proved to be similar to rays of light. Some time after the discovery of electrons, such investigators as Busch, Davisson, and Kolbick, later—Knoll and Ruska (High-Voltage Laboratory of the Technische Hochschule, Berlin), then Brüche, Johannsen, and Scherzer (AEG Research Institute), as well as Picht (Berlin) and Glaser (Prague), undertook the determination of electron trajectories, thereby laying the foundation for the study of electron optics. Since then these questions have been subjected to increasingly detailed investigation as the importance of electron optics has become more and more evident.

Of particular interest is the calculation of electron trajectories in fields possessing axial symmetry. It can be shown that in electrostatic or magnetic fields of this kind electron rays are refracted in exactly the same way as rays of light when passing through a lens. This possibility of establishing an analogy between an electron-optical system and a system of optical lenses in many cases simplifies quantitative and qualitative investigations of electron-lens systems.

The analogy between electron and light rays is evident, for example, from the behavior of electrons flying through an aperture with potential \(V\) from a point at which the field strength is \(E_1\) to a point with field strength \(E_2\). In this case the aperture causes the electrons to converge in exactly the same way as would a spherical lens with focal length \(f = \dfrac{4V}{E_2 - E_1}\). Just as a glass lens gives an image of a source of light, by means of the aperture one can obtain an electron image.

* Z. techn. Phys. No. 6, p. 169, 1936. Translated by D. V. Zernov.

It was further found that every axially symmetric electrostatic field produces an effect analogous to that of a positive or negative lens, provided only that the electron path passes not too far from the axis of symmetry.

It is not our intention to give a description of theoretical investigations in the field of electron optics, and we shall confine ourselves only to considering the known practical details relating to the production of electron images. Further, we shall limit ourselves mainly to the question of obtaining electron images by means of electrostatic fields, to the detriment of systems consisting of magnetic or combined magnetic and electrostatic lenses.

If electron optics were of merely academic interest, it would never have received such great attention as it does at the present time. The two most important and most thoroughly studied practical applications are the electron projector and the electron microscope. A characteristic feature of the former is that in it an extremely small electron source must be focused into an extremely small image or spot. In this case both the image and the object lie near the axis of symmetry, and the various distortions of the image do not play a major role. A typical example of an electron projector is shown in Fig. 1. The system consists of two lenses (cf. the equivalent lens system beneath the drawing of the electron projector), which do not give an image of the cathode itself, but instead project onto the screen a point situated in front of the cathode (the point of intersection of the electron rays). This type of electron optics plays an important role in cathode oscillographs in general and in the technique of cathode television in particular. The latter application has already been described previously[^1].

Fig. 1. Electron optics of the kinescope and its analogous system of optical lenses

Fig. 2. Image obtained by means of the kinescope
(343 lines)

The quality of image transmission depends first of all on the accuracy of focusing of the electron beam both in the transmitting and in the

in the receiving device. The achievements in this field may be seen from Fig. 2, which gives a photograph of the television image of a motion picture when scanned into 343 lines.

In the second practical application—the electron microscope—electron optics is used to obtain a highly magnified image of a small object. Fig. 3 shows an electron microscope. The drawing is taken from the excellent book by Brüche and Scherzer devoted to electron optics. There exist a large number of types of these microscopes, using combined electrostatic and magnetic fields. Discussion of these subjects, however, lies outside the scope of the present article.

Fig. 3. Electron microscope according to Brüche and Scherzer

In the course of our investigations in electron optics, we were faced with the task of obtaining an image of an object of relatively large dimensions. It then proved necessary, in order to achieve this aim, to make use exclusively of electrostatic means.^2 The optical analogue of this system is the lens system of a copying camera. A satisfactory solution of this problem requires:

1) that the image be free from distortions,
2) that the region of best definition be as large as possible,
3) that the effective aperture of the lens be large,
4) that the dimensions of the object and of the image be large.

Fig. 4. Electron-optical system consisting of an electron lens formed by two coaxial cylinders

The magnification of this system had to be approximately equal to 1, i.e. remain within the limits from \(1/2\) to 3.

As the fundamental basis of this system we shall consider the field between two coaxial cylinders of the same diameter. The complete apparatus is shown schematically in Fig. 4. The cathode \(K\) is a semitransparent photoelectric layer. The fluorescent screen \(L\), like the screen in cathode oscillographs, is applied by depositing a thin layer of tungstate

calcium onto the bottom of the tube. A potential difference is applied between the two cylinders. The cathode and the cylinder connected with it are charged negatively, while a positive potential is applied to the screen and to the long cylinder. If the optical image is now projected onto the photocathode, electrons will fly out from the cathode in a quantity proportional to the intensity of the light falling on the given area of the surface. These electrons are projected by an electrostatic lens (the field between the two cylinders) onto

Fig. 5. Distribution of potential and electron trajectories in an electronic image converter

the screen in the form of an inverted image. Since this system forms the basis of tubes used for image conversion, it is meaningful to consider it in greater detail from both the theoretical and the experimental points of view. The potential distribution can be determined by means of the integral presented in Fig. 5. Its solution gives the axial distribution, which, together with the second derivative of the potential, is shown in the figure as a curve. From this the trajectories of the electrons can be calculated by means of the known formula for beam propagation. The lower curves show the electron trajectories; one of them represents the trajectory of an electron emerging from the middle of the cathode at a certain angle to the direction of the axis and determines the position of the image, while the other shows the path of an electron emerging from the cathode with an initial velocity equal to zero, at some distance from the axis, and determines the magnification of the system. The position of the image is determined by the distance \(u\) between the cathode and the line separating the two cylinders, which we shall conventionally call the “lens.” The dependence of the distance between the lens and the image \(v\) on \(u\) was determined experimentally with the aid of a tube with a movable cathode and screen.

The results of both methods of measurement are presented as curves in Fig. 6, giving the dependence of \(u\) on \(v\) (expressed in fractions of the diameter of the surface emitting electrons). In the same figure

an increase \(m\) is presented as a function of \(\dfrac{v}{2u}\). From the drawing it is seen that, over a wide range, the magnification \(m\) is determined with sufficient accuracy by the relation \(m=\dfrac{v}{2u}\). It should be mentioned that the position of the image does not depend on the potential difference between the two cylinders. A practical inconvenience of this system is the almost complete impossibility of constructing and arranging the cylinders so that the image is absolutely sharp, since it is very difficult, if not impossible, to make the tube in such a way that after evacuation and sealing off it would be possible to change its

Fig. 6. Dependence of magnification on tube dimensions

geometrical dimensions. In order to achieve focusing, it is necessary to change the curvature of the equipotential surfaces near the lens, without changing the field near the cathode so much that the distortion of the image increases. This can best be achieved if the cylinder located between the cathode and the lens is made of a material with high resistance, so that a definite potential gradient is established along its length. In practice it has turned out that, when the cathode cylinder is divided into four or more rings with gradually increasing potential,\(^3\) the distortion of the image is no greater than with mechanical focusing of the tube. The magnification of this type of tube is determined from the same relation as in the preceding case, namely:

\[ m=\frac{v}{2u}. \]

Both types of tubes give a relatively satisfactory image if the part of the cathode surface used is small. If this is not taken into account, the image will have two shortcomings:

1) curvature of the contours,

2) a genuinely sharp image can be obtained on the screen only for a definite diameter.

In Fig. 7 is shown a photograph of the image obtained on a fluorescent screen if the entire cathode is illuminated through a rectangular grid. In this figure the image appears as though concave in the direction toward the center of the lens. We have two means of correcting the above-mentioned distortions without changing the existing lens system:

1) the formation of a radial gradient on the cathode,

2) imparting the corresponding shape to the surface of the cathode.

Fig. 7. Distortion of the image produced by an electrostatic lens

The first method at present is not of especially great importance, but is of theoretical interest. It consists in making the transparent cathode layer possess a relatively high ohmic resistance and allowing the current to pass in the radial direction from its middle. The resistance of the layer is usually not uniform; its lowest value corresponds to the middle of the cathode, increasing in proportion to the square of the radius in the direction of the edges. If the middle of the cathode is given a positive potential and the resistance is made exactly proportional to the radius, then both the curvature of the contours and the phenomena of unsharpness can be eliminated. Since, however, in this case an electric lead must be brought to the middle of the cathode, a spot inevitably appears at the center of the image.

It is interesting to note that such cathodes, for a certain illumination, can be made self-correcting if the layer is made of sufficiently high resistance. In this case the photoelectric current creates a potential drop across the cathode in the same way as the radial current in the above-mentioned case. The drawback of this solution of the problem is that the correction is complete only for one definite value of illumination and that the illumination of the optical image must be approximately uniform.

The second method—imparting a special shape to the surface of the cathode—is considerably more practical and gives an excellent image over the entire surface of the fluorescent screen. In this method of correction one may first of all assume that the cathode must be curved so that its convex side is turned toward the lens, so that the middle of the object is at a shorter distance from the lens than would be the case if the curvature of the contours and ...

distortion of the field. However, the principal cause of these distortions is a consequence of the change in the radii of curvature of the equipotential surfaces with increasing distance from the axis near the plane cathode. When a convex surface is introduced, this effect is only intensified. If, on the contrary, the cathode faces the lens with its concave side, the shape of the equipotential surfaces can be changed so that the image is obtained free of distortions. In practice, for a spherical cathode with a radius of curvature equal to the distance to the lens \((u)\), quite satisfactory results were obtained.

Figure 8 schematically shows the construction of a tube with a curved cathode and concentrating rings. In this case it is not necessary to make separate leads for each ring. Instead, a voltage divider maintaining definite potentials on the rings can be introduced inside the tube itself. A photograph of the image of the same grid as in Fig. 4 is shown in Fig. 9. It is evident from the figure that the image is equally sharp everywhere, and the curvature of the contours is negligibly small.

Fig. 8. Electron image converter with constant magnification

Fig. 9. Improvement of the quality of the electron image by giving the cathode a concave shape

Up to now we have spoken only of lens systems without an aperture diaphragm. Although the presence of an aperture diaphragm somewhat complicates the form of the electrostatic field forming the electron lens, the known general characteristics remain the same. The field, as before, remains axially symmetric; the lens system as a whole remains positive and gives a real image; and, finally, if an uncorrected tube is used, the same image distortions arise as in the tubes described above. Three ways of arranging aperture diaphragms in the tube described above are presented in Fig. 10. The first of these designs contains two symmetrical diaphragms. For this system, as in the preceding case, the magnification proves to be

equal to \(m=\dfrac{v}{2u}\). The focal length, however, for a given ratio \(V_1:V_2\) is made shorter, i.e., in practice the potential difference between the anode cylinder and the last concentrating ring can be made smaller. This circumstance eases the problem of insulation and reduces the danger of a discharge occurring between the electrodes.

If the aperture diaphragms are asymmetrical, then the magnification no longer satisfies the equality \(m=\dfrac{v}{2u}\). If the diaphragm of the anode

Fig. 10. Methods of arranging the diaphragm in an electronic image converter

Fig. 11. Electronic image converter with variable magnification

cylinder is smaller than the diaphragm of the last concentrating ring, then the magnification is greater than follows from the preceding equation, whereas in the case when the anode diaphragm is the larger one, the magnification is smaller. The use of asymmetrical aperture diaphragms allows for a greater variety in the operation of tubes. The latter case, which has been investigated, is shown in Fig. 11. In this lens system the aperture diaphragm is placed midway between the anode cylinder

and the last concentrating ring; moreover, its potential \(V_3\) may be varied at will. It should be noted that, in the case where the aperture diaphragm and the anode cylinder are at the same potential \(V_3\), the conditions will be analogous to the case described above, in which the anode diaphragm is smaller than the diaphragm of the focusing ring, and the magnification obtained is greater than \(\dfrac{v}{2u}\). In the same way, the magnification becomes minimal if the potential \(V_3\) becomes equal to the potential of the ring.

Fig. 13. Electronic image converter

Fig. 12. Construction of an electronic viewing tube

Fig. 14. Installation for infrared microscopy

For intermediate values of the potential \(V_3\), the magnification lies between these limiting values.

The distortion of the contours is greatest when \(V_3\) is equal to or less than the potential of the ring; however, as \(V_3\) increases, this effect rapidly weakens, tending toward a certain constant value. This distortion is almost completely corrected over the entire magnification range by means of a bent cathode.

The described design makes it possible to produce a tube that gives a quite satisfactory image, and the magnification can be varied at will. An interesting practical application of this type of tube is the electron telescope (Fig. 12).

Fig. 15. Image of plant fibers obtained in infrared light

Fig. 16. Electronic image of a film illuminated by infrared light

A large-aperture objective is arranged so that the picture at which the telescope is directed is projected onto the cathode of a tube sensitive to infrared light. The electronic image falling on the fluorescent screen makes the infrared image visible. Such a device may be used to study the permeability of fog and smoke to infrared rays and may be used for signaling and similar purposes.

Another field of applications is being created in connection with infrared microscopy. Fig. 14 shows a tube and microscope for studies in infrared light. The image of a microscopic specimen on the fluorescent screen is shown in Fig. 15 (plant fibers).

Fig. 17. Electronic image of a film illuminated by infrared light

The following two figures (Figs. 16 and 17) show photographs of the fluorescent screen when an infrared image is projected onto the cathode. As is evident from these figures, the image obtained is quite sharp.

V. K. Zworykin

II. Electron Optics of Secondary-Electron Converters (Multipliers)

We shall now consider another field of application of electron optics, one not connected with its use as a means of obtaining an image, but in which electron optics nevertheless plays an extremely important role. This field is that of secondary-electron converters. Here the greater significance should be attached to the general principle and mode of operation of the converters; however, the importance of electron optics in this problem is self-evident.

It became known many years ago that certain surfaces, when bombarded by cathode rays, emit electrons. This effect, known as secondary emission, had already been studied in considerable detail long ago by a number of investigators, such as Lenard, Hull, Bayard, and others.

The study of this phenomenon showed that the number of emitted secondary electrons is proportional to the number of incident (primary) electrons, the coefficient of proportionality varying from values less than unity up to 10. The magnitude of this ratio depends on the surface used and on the velocity of the bombarding electrons. Although this effect has been known for a long time, until now it had found no practical application, with the sole exception of the dynatron invented by Hull. In fact, secondary emission was regarded chiefly as a substantial difficulty in the design of electron tubes, and a whole series of investigations was undertaken with the aim of eliminating it or, at least, reducing it. During the last 15 years it has been discovered that secondary emission can be used to amplify very weak currents, and a number of investigators have taken up the development of this problem. Patents for methods realizing this idea were proposed as early as 1919 by Slepian and later by Jervis and Blair, Iams, Farnsworth, and others.

Fig. 18. Diagram of an electron multiplier

The basic principle of operation of secondary-electron converters is as follows: a stream of primary electrons is directed onto a surface specially activated for the purpose of obtaining high secondary emission. The secondary electrons emitted by this surface are directed onto the next surface, where they in turn produce secondary emission, and so on. This process may be repeated the desired number of times. An explanation of this process is given in Fig. 18. The electrodes \(A, B, C,\) etc., are flat surfaces with a high coefficient of secondary emission. Gradually increasing—

positive potentials. The amplified electron current is directed onto plate \(A\), on which secondary electrons are liberated, going to plate \(B\). Here, in turn, secondary electrons are liberated, going to plate \(C\), and so on. After this process has been repeated a sufficient number of times to obtain the required total amplification, the electrons emitted by the last plate are captured by the collecting electrode \(O\). If at each cascade the number of secondary electrons per one primary electron is \(\sigma\), and the total number of cascades is \(h\), then the applied current \(I_0\) gives at the output a current \(I\), equal to:

\[ I = I_0 \sigma^h . \]

Thus the total amplification is proportional to \(\sigma^h\). This indicates that, with an increase in the number of cascades, the total amplification grows very rapidly.

Another type of secondary-electron converter was described by Farnsworth. In these converters the electrons oscillate between a single pair of electrodes, between which a high-frequency electrostatic field is created. In our article, of these two types of converters only static multipliers will be described, using cathodes of secondary emission placed one after another, since these converters permit reliable regulation and possess very considerable stability in operation.

The problem of constructing a converter with a large coefficient of efficiency, however, is by no means as simple as it may seem at first glance. A converter constructed according to the diagram of Fig. 18 would give almost no effect, since in this case practically all the electrons leaving a plate would not fall onto the next plate, but would fly along the tube and be captured directly by the collecting electrode, without producing secondary electrons on their path. To construct a truly operating converter, it is necessary not only to make plates having a high coefficient of secondary emission, but also to find means that make it possible to concentrate the electrons on each plate and force the secondary electrons emerging from one plate to fall onto the next.

Before proceeding to the analysis of the methods of electron concentration specially used in converters, it is necessary to consider some basic principles of operation of static converters.

A. General considerations

1. Secondary emission

Since the successful operation of such converters depends on the high secondary emission of the plates, it is extremely important to find the surface most suitable for this purpose. In our search for good sources of secondary emission we found almost no support from theoretical physicists,

The theory of secondary emission from the point of view of quantum mechanics was published in 1932 by Fröhlich. In this work the probability is calculated of energy exchange between a primary electron penetrating into a metal and a conduction electron of the metal, moving in it in the periodic field of the lattice. This exchange takes place in such a way that the free electron acquires a quantity of motion sufficient to escape from the metal. From this it is concluded that the best sources of secondary emission are metals with a large lattice constant and a small work function. These conclusions are valid only for simple metallic surfaces.

It was found experimentally that the secondary-emission coefficient of simple metallic surfaces in all cases proved to be lower than for complex surfaces, just as in the case of photoelectric sensitivity. Since the theory of secondary emission does not extend to these complex surfaces, investigations have to be carried out in a more or less empirical manner, taking into account that, other conditions being equal, in the majority of cases a surface possessing a small work function is at the same time a good source of secondary emission.

Fig. 19. Dependence of the secondary-emission coefficient on the velocity of primary electrons for the surface Cs—Cs₂O—Ag

For this purpose a large number of surfaces with a small work function were investigated, in which the base metals were Ag, Be, Ta, Ni, Al, Zr, Ca, W, etc., while the surface layer consisted of Na, K, Rb, or Cs. The most satisfactory among them proved to be: oxidized Ag, Be, or Zr with a surface layer of cesium. The maximum secondary-emission coefficient for these surfaces reaches a value of 8–10 at primary-electron velocities of the order of 400–600 V.

In Fig. 19 a curve is given showing the dependence of the secondary-emission coefficient on the velocity of the primary electrons

for the surface Cs—Cs₂O—Ag. This curve is characteristic of surfaces often used in the converters described below. The method of preparing these surfaces is analogous to the method used for preparing the cathodes of vacuum cesium photocells.

2. Efficiency of converters

The efficiency of secondary-electron converters can be defined from two points of view. On the one hand, the efficiency of a secondary-emission cathode can be expressed in the same way as for any source of electrons,

Fig. 20. Efficiency of the Cs—Cs₂O—Ag layer in A/W

in amperes per watt of power consumed. On the other hand, one may consider the amplification achieved at a given total voltage as a function of the number of stages, or as a function of the voltage at each stage. Of these two definitions, the second is more important from the practical point of view. Nevertheless, both should be examined. From the curve given in Fig. 19, the curve of the dependence of the gain per volt, or of the electron output in mA, on the velocity (in volts) of the primary electrons can easily be calculated and plotted (Fig. 20). This curve shows that the maximum efficiency lies near 30 V and corresponds to 60 mA/W. The curve then falls to 45 mA/W at 100 V and to 17 mA/W at 500 V. For comparison it may be mentioned that a good thoriated tungsten cathode gives from 55 to 75 mA/W, whereas a good oxide cathode gives up to 100 mA/W.

The question of the total amplification becomes clearer from consideration of Fig. 21. The family of curves shown in it gives the amplification obtained with multistage converters as a function of voltage. These curves show that the best operating conditions

the converter correspond to voltages of 40–50 V at each stage. With a suitable number of stages, very large values of the efficiency factor can be attained. An example is a 10-stage converter which, at a

Fig. 21. Dependence of the gain factor of an electronic multiplier on the total voltage and the number of stages

voltage of 500 V, gives an amplification of the order of 20,000, whereas a 15-stage amplifier at 800 V amplifies the primary current by \(10^8\) times. It should be noted that the values of the total voltage given in Fig. 21 do not include the voltage between the collecting electrode and the last plate, since this latter depends on the circuit for which the tube is intended.

3. Noise level at the output of a secondary-electron converter

In approaching the discussion of the question of the noise level that sets a limit on the amplification of secondary-electron converters, it is necessary chiefly to bear in mind the statistical fluctuations of the electron current in the tube. From theoretical considerations, confirmed by experiment, it follows that the ratio of the signal to the noise level attained under amplification is practically determined by the shot effect in the primary photocurrent.

Further, two other phenomena also limit the sensitivity of these converters. These include: 1) thermionic emission of the cathodes of the secondary emission, and 2) noise caused by the emission of positive ions. Both of these noise-producing factors can be compensated if appropriate measures are taken. In general, it should be noted that the shot effect is the most substantial limitation on the sensitivity of secondary-electron converters. At low illuminations, the gain in the signal-to-noise ratio obtained is from 60

to 200-fold compared with ordinary amplifiers under the normal operating regime.

4. Dependence on frequency

The frequency characteristic of secondary-electron converters over a wide interval runs parallel to the frequency axis.

B. Magnetic secondary-electron converters

1. Theory of magnetic multipliers

We shall now consider secondary-electron converters in which crossed magnetic and electrostatic fields are used to focus secondary electrons on the cathodes of secondary emission. This configuration of fields and electrodes was first proposed by Slepian in 1919 and was intended for use as a cathode giving a large current strength.

The arrangement of a converter based on this principle is shown schematically in Fig. 22. It consists of two rows of electrodes, of which the lower row represents sources of secondary emission, while the upper serves only to create

Fig. 22. Magnetic electron multiplier

Fig. 23. Defocusing of an electron beam in a magnetic multiplier

a transverse field between the two rows of plates. Each plate of the lower row has a positive potential with respect to the preceding one, so that when electrons coming from the preceding plate strike it, it can emit secondary electrons. The magnetic field is directed perpendicular to the axis of the tube and to the direction of the electrostatic field between the two rows of plates. The electrons emerging from the first plate are bent by the combined fields in such a way that they reach the neighboring plate, where they knock out secondary electrons, which, in turn, are bent toward the next plate, and so on along the entire tube.

Under the action of both fields the secondary electrons are separated from the primary ones and are focused on the neighboring plate. This focusing may be compared with the optical system of cylindrical lenses. As in that system, concentration takes place predominantly—

respectively in one direction and has absolutely no component in the direction perpendicular to the first. This means that electrons emerging from any point of the cathode produce, on the first plate, a spot of elliptical shape. This spot, in turn, expands into an even larger ellipse on the second electrode and then continues to grow as the electrons move along the tube. This process is schematically represented in Fig. 23.

Finally, the spot becomes so large that some electrons cease to fall on the plates at all, thereby causing a lowering of the efficiency of the subsequent stages. The expansion of the spot is so great that, if no special measures were taken against such deconcentration, considerable losses would arise even with a comparatively small number of stages.

2. Principle of the design and manufacture of magnetic converters

A schematic drawing of a cascade photoelement constructed according to the principle described above is shown in Fig. 24. Photoelectrons emerging from cathode \(1a\) are concentrated on plate \(2a\). Secondary electrons emerging from this electrode are concentrated on plate \(3a\) and, in turn, liberate electrons there. This process is repeated as many times as the given converter has stages.

Fig. 24. Magnetic electron multiplier

The plates are fastened inside the tube in such a way that the upper and lower plates are located as close as possible to one another. This is done in order to increase the directionality of the field and to increase the current strength from the plate, which is limited by space charge. The minimum gap is half the distance between the midlines of two neighboring plates. With such a construction, the current at the output of the tube is limited exclusively by power losses in the final stages as a result of their heating under the action of the electrons falling on them,

To limit defocusing, the electrons are accelerated on vertical solid strips. The charges accumulating on these strips alter the field in such a way that the additional action caused by them limits the defocusing. In this device the limit on the number of stages used is set by axial concentration. The latter, however, is so small that it permits the use of a very large number of stages. The upper limit on the number of stages has not been found experimentally. Converters with more than 12 stages were made without any noticeable decrease in the gain at each stage.

If a converter providing high gain operates into an external circuit with a large apparent resistance, certain difficulties arise, associated with a tendency toward oscillation. This tendency can be eliminated if a shielding grid is placed before the collecting electrode (Fig. 24). Such a grid serves as an electrostatic shield, preventing the reverse action of fluctuations of the collector potential on the preceding stages. The characteristic of the converter is thereby transformed from the characteristic of a triode into the characteristic of a tetrode. During operation of the converter the upper electrodes must be maintained at a constant positive potential relative to the corresponding lower electrodes, and the potential drop between adjacent electrodes must be the same. To make the number of necessary leads as small as possible, it is advisable to connect all the upper electrodes to the buses. Satisfactory results can be obtained if each upper electrode is connected to the next lower plate.

Fig. 25. Construction of the electrodes of a magnetic electron multiplier.

Since in the first stages of the multiplier the current consumption is negligible, their potentials can be maintained by means of a voltage divider of comparatively small dimensions. In order to reduce still further the number of leads when there is a large number of stages, it is advisable to place the voltage divider supplied to the first stages inside the tube itself.

The converter can also operate with alternating voltage on the electrodes. In this case, naturally, the tube will operate only during a certain part of each period. The frequency of the applied alternating voltage must exceed the highest frequency that the converter is to reproduce.

A satisfactory means of producing the required magnetic field is the use of permanent magnets. These latter

more convenient than electromagnets, since they are smaller in size and do not require an external current source.

A photograph of the internal arrangement of a 12-stage converter, in which the voltage divider for the first 5 stages is placed inside the tube, is shown in Fig. 25.

C. Electrostatic Converters

Where the field of application of secondary-electron converters does not permit the use of a magnetic field, it is necessary to use converters in which the focusing of the electrons is accomplished exclusively by means of electrostatic fields.

The problems that arise in designing the focusing system of such a converter are analogous to the problems that arise in designing an electron microscope. It turns out that, in the general case, a radially symmetric electrostatic field in a region close to the axis of symmetry possesses the properties of a lens. The distance in the radial direction for which these conditions hold depends on the configuration of the field. Focusing

Fig. 26. Electron lens of an electrostatic multiplier

Fig. 27. Arrangement of electrodes in a type I multiplier.

in electrostatic converters is produced by a field between two coaxial cylinders. Since it is desirable to have as few lead-ins as possible on the tube, the electrodes are constructed so that one of the cylinders is part of one of the electron-emitting plates, while the other cylinder is connected to the next plate. The electrodes are arranged in such a way that the electrons leaving the central zone of the first plate are collected at the center of the next plate and that the magnification of the electron image proves to be equal to unity.

An electron-optical system analogous to the system already described in the present article is shown in Fig. 26. In this drawing the lens is formed by cylinders \(A\) and \(B\), with cylinder \(A\) grounded and cylinder \(B\) at potential \(E\). The electrons,

emitted by cathode \(A\) with very low velocities, are deflected by the lens between the two cylinders and are focused on the screen or electrode \(B\), which they reach with a velocity equal to \(E\) volts.

The use of this system leads to the so-called converter of type \(L\), the construction of which is shown in Fig. 27. As regards focusing, this converter is quite satisfactory. However, because the field accelerating the secondary electrons emerging from the individual plates is comparatively weak, at extremely low current strengths a space charge arises in the converter. Further, the emitting spot on the first electrode must be small, since its image has the same dimensions.

The action of an electron multiplier of another construction does not depend to so great an extent on the sharpness of the image. Moreover, in this construction large field gradients are obtained at the surface of the secondary-emission cathodes. This construction, representing the so-called type \(T\), is shown in Fig. 28.

Fig. 28. Arrangement of electrodes in a multiplier of type \(T\)

This converter is made in such a way that the cylindrical entrance apertures are as short as possible, and yet long enough that the electrons entering the base of the letter \(T\) are not deflected by the field of the next electrode so much as to fail altogether to strike the surface emitting secondary electrons. The emitting surfaces are made in such a way that the entire inner side of the cylindrical crossbar of the letter \(T\) is sensitized.

Fig. 29. External appearance of a multiplier of type \(L\).

In order to obtain currents of the order of 1 mA, a converter of this type, for the purpose of eliminating the space charge, must operate at very high potentials.

on each cascade, namely on the order of 200–400 V. In Figs. 29 and 30 cascade photocells of types \(L\) and \(T\) are shown.

D. Applications

The most widespread practical application of these converters is the amplification of photocurrents. This field of application is especially simple, since surfaces possessing large secondary emission are at the same time the most sensitive photocathodes. For this purpose a device is usually used that employs combined electrostatic and magnetic fields, since such a device gives the best concentration of electrons and a large current strength at the output.

Fig. 30. External view of a multiplier of type \(T\)

A tube of this type gives an amplification on the order of several millions and has a sensitivity of 10 A/Lm or more. Its dimensions only slightly exceed those of an ordinary amplifying tube. These tubes require an operating voltage on the order of 1500 V and, since they consume an insignificant current, they can be supplied with a socket of small dimensions.

Since the secondary-electron converter replaces not only the photocell but also the amplifier, with its aid a great simplification of the entire apparatus and a considerable saving of space can be achieved. In addition, these converters are distinguished by extremely great stability, independence from external influences, and a good frequency characteristic. Especially important is the circumstance that the interference at the output is determined exclusively by the noise created by the shot effect at the photocathode, as a result of which, under normal conditions, the ratio of the signal magnitude to the amplitude of the interference at low light intensities is increased by a factor of 60–100. All these qualities, taken together, make this type of converter the best means for converting light signals into electrical ones. For comparison of the dimensions of the tube with an ordinary amplifying tube, in Fig. 31 a 10-cascade converter and an amplifying tube of type RCA-59 are placed side by side.

Electron-Optical Systems

The field of application of multistage photocells is very broad and includes sound recording, image transmission, various control, signaling, and sorting automatic devices, etc.

Although at present the most important technical application of secondary-electron converters is the cascade photocell, there is nevertheless a whole series of other possible uses for them, which are acquiring ever greater significance. Generally speaking, these converters may be used in all devices in which the signal to be amplified is an electron current. This field of application includes so-called alternating-direction tubes used in high-speed switches, secret telephony, and modulations of frequency.

As a voltage amplifier this device cannot be used directly. In order to connect the input of a voltage amplifier with an external circuit, it is necessary to use an intermediate coupling resistance, which in the case of an ordinary tube amplifier limits the ratio of the signal magnitude to the amplitude of the disturbances. If, in connection with the converter, an ordinary heated cathode provided with a controlling grid is used, great difficulties arise, if one wishes thereby to obtain results better than with the aid of an ordinary vacuum electronic tube. There is a possibility of obtaining better control of the anode current if one deliberately refrains from using a large part of the emission current of the cathode, and then uses an electron multiplier in order subsequently again to increase considerably the mean current strength. These experiments, however, are still in the initial stage, and it is still too early to say what will come of them.

Fig. 31. Difference in dimensions between an electron multiplier and an amplifying tube

The secondary-electron converter is still too new an instrument for it to be possible to outline the entire field of its applications. However, it is already quite evident at the present time that it can become a serious competitor of electronic tubes in fields where, for a long time, the latter have occupied an exclusive position, and also conquer new fields of application.

References

1. V. K. Zworykin, J. Inst. Electr. Eng. 73, 442, October 1933. See also V. K. Zworykin, Uspekhi fiz. nauk 14, 788, 1934.
2. G. Holst, J. H. de Boer, M. C. Teves and C. F. Veenemans, Physica 1, 297, 1934.
3. M. Knoll, Arch. f. Elektrot. 28, 7, 1934.
4. H. Frohlich, Ann. d. Phys. (5) 13, 229, 1932.