NEW INSTRUMENTS AND METHODS OF MEASUREMENT

ELECTRON PROJECTOR AS A METHOD OF PHYSICOCHEMICAL RESEARCH

B. M. Tsarev

INTRODUCTION

The rapid development of electron microscopy over the last 10–15 years has given experimenters a powerful means for investigating many physicochemical phenomena, one that has significantly extended the range of magnifications in comparison with the accessible methods of optical microscopy. Electron microscopy has acquired particular value in the study of various kinds of electron emitters,^1 in the investigation of catalysis phenomena,^2 and in the biological sciences.

In its usual form, the electron microscope, especially in the case of high magnifications, is a complex of extremely complicated apparatus, consisting of the microscope proper, a vacuum system for evacuating it, and its power-supply circuit. At high magnifications the construction of the microscope itself is very complex, and in terms of cost such an electron microscope is accessible only to the largest institutes and laboratories. However, alongside such designs, various simplified designs of electron microscopes, not requiring complex electron-optical systems, have been successfully used in a number of investigations. In what follows we shall call these designs “electron projectors,” since the term “electron microscope” essentially arose from the far-reaching analogy of these instruments with optical microscopes, whereas in the simplified designs use is made of the possibility of obtaining large magnifications of the object under study by means of the radial propagation of electrons emitted by its surface—that is, essentially by “projecting” the surface of the object by means of a stream of electrons onto the surface of a luminescent screen. As we shall see below, electron projectors, despite their exceptional simplicity in comparison with the electron microscope, make it easy to obtain much greater magnifications, reaching in individual cases 1–2 million times.

The disadvantages of electron projectors include the possibility of studying objects only with the aid of the electrons emitted by those same objects. Nevertheless, as we shall see below, electron projectors are successfully used for the study not only of electron emitters, but also of a number of other physicochemical phenomena, such as, for example, the crystalline structure and recrystallization of metals, adsorption phenomena, surface migration and evaporation of various substances, the kinetics of chemical reactions of solids with gases, vapors, etc. In addition, electron projectors are indispensable for studying cathodes of ordinary designs used in various modern electrovacuum devices in the form, for example, of wires, ribbons, cylinders, and also single crystals of various metals, the investigation of which with an electron microscope is extremely difficult because of the peculiarities of the electron optics of the latter.

I. DESIGNS OF ELECTRON PROJECTORS

The principal types of electron projectors—namely the cylindrical, spherical, and conical (the latter being only a variety of the spherical)—are shown schematically in Fig. 1.

Fig. 1. Principal types of electron projectors:
a—cylindrical; b—spherical; c—conical; E—fluorescent screen, K—collector;
O—object under study: wire or point,
P—point heater.

Depending on the method by which the electron flux is obtained from the object under study, one may distinguish thermionic projectors, in which the object is heated to a temperature at which it gives electron emission sufficient to obtain an image on the screen, and field-emission projectors,

in which the electron image of the object is obtained by means of electrons torn from its surface by a strong external electric field. However, in the majority of cases, in one and the same instrument the image of the object on the screen can be obtained both by heating the object directly with current or through thermal conduction from a special heater, and by autoelectronic emission under the action of an external field; therefore this additional classification of projectors is superfluous.

1. Cylindrical electron projector, first proposed by Johnson and Shockley³ (see Fig. 1,a), is most convenient for studying objects in the form of thin wires and has been widely used for investigating auto- and thermoelectronic emission both of pure metals and of film cathodes, that is, films of various contaminants on the surface of a metal.

Fig. 2. Autoelectronic emission of an untreated tungsten wire: a—before cleaning of contaminants, b—after cleaning.

This projector consists of a glass cylindrical bulb, the walls of which are coated on the inside with a fluorescent substance. Along the axis of the bulb, the wire under investigation is stretched by means of a spring and fastened to two leads, which allow the heating current to be supplied to the wire. The external electric field is applied between the wire (−) and the anode (+), made in the form of a wire spiral,

located along the walls of the cylinder and having a lead passing through the wall of the bulb.

Since the fluorescent screen, owing to secondary electron emission from it under bombardment by electrons, acquires a high positive potential close to the potential of the anode (the spiral), the electrons emitted by the wire as a result of heating or pulled out by the external electric field pass from the wire to the screen practically along radial trajectories and give on the screen an enlarged image of the emitting surface of the wire.

Fig. 3. Emission of tungsten after removal of contaminants: a — unpolished wire; the junctions of individual crystallites are visible through the drawing marks; b — polished wire, with a distinctly revealed crystalline structure; c — structure of tungsten revealed by cesium absorption.

Examples of such images are shown in Figs. 2 and 3 for a pure tungsten wire, in Fig. 4 for thoriated tungsten, and in Fig. 5 for an oxide cathode.

A disadvantage of the cylindrical cathode is that it gives magnification only in one dimension, perpendicular to the axis of the wire under investigation. In this case magnifications on the order of hundreds and thousands of times are easily achieved. Thus, for example, with a wire diameter of 0.1 mm and a cylinder diameter of 60 mm, we obtain a radial magnification equal to 600; and with the same diameters equal respectively to 20 microns and 100 mm, we obtain a magnification of 5000 times. In the direction of the wire axis, however, the magni-

the magnification is equal to 1, that is, the image of the object on the screen is obtained with strong distortion. However, in many cases, not only for qualitative studies but also for quantitative estimates and measurements, such images are quite sufficient. In addition, for the study of wires or other objects with cylindrical symmetry, this type of projector is the most convenient.

1. Spherical projector. The above-mentioned drawback of the cylindrical projector is completely eliminated in the so-called spherical projector (Fig. 1, b), which has found wide application in the study of single crystals of various metals^4,5,6. The spherical projector is still used mainly for the study of auto- and thermionic emission from the thin single-crystal tip of the metal under investigation.

Fig. 4. Thoriated tungsten: a — immediately after annealing to 2800° abs.; b — after 4 minutes of activation at 1850° abs.; c — after 40 minutes of activation at 1800° abs.

A spherical projector usually consists of a glass sphere, half of the inner surface of which is coated with a fluorescent layer. At the center of this hemispherical screen is placed the metal tip under investigation, usually of length on the order of 1 mm, welded to a tungsten loop, which allows the tip to be heated by thermal conduction. The anode-collector is most often either a transparent layer of platinum deposited on the glass beneath the fluorescent layer of the screen, or a metal ring, arranged—

Fig. 5. Oxide cathode; a, b, c, d — different stages of activation.

placed somewhat below the tip, as we see in the projectors shown in Figs. 6 and 7. Martin^5 used a spherical projector for the study of a single-crystal tungsten hemisphere 1 cm in diameter, and with a spherical-screen diameter of 22 cm therefore had a magnification of only 22 times in all (see Fig. 7). In the case of studying objects in the form of a tip, however, magnifications of the order of several tens of thousands of times are comparatively easily attained. Thus, for example, even with a sphere of diameter

![Figure 6 and Figure 7 diagrams]

Fig. 6. Spherical projector with annular collector; the designations are the same as in Fig. 1.

Fig. 7. Martin’s spherical projector:
1 — fluorescent screen; 2 — tungsten hemisphere 1 cm in diameter; 3 — tungsten holders; 4 — tungsten heater; 5 — thermocouple; tungsten–tantalum, 6 — tantalum shields; 7 — tantalum collector; 8 — side arm with getter; 9 — side arm with ionization manometer; 10 — place of sealing-off; 11 — trap with cesium, cooled by liquid air.

10 cm in diameter, a readily obtainable tip with a radius of curvature of 5 microns gives a magnification of 10,000 times. Specially prepared tips in spheres of diameter 10, 15, and at most 20 cm make it possible to obtain magnifications of 100, 200, and 500 thousand times and higher. Thus, for example, Benjamin and Jenkins^4 in their experiments had tips with diameters from 0.5 to 1 micron and, with a screen diameter of 10 cm, obtained magnifications from 100 to 200 thousand times; Müller^7 succeeded in obtaining tips with a radius of curvature of 1120 Å and magnifications up to 420,000 times. In studying the limiting resolving power of the spherical projector, he succeeded in obtaining tips of 180 Å and even 110 Å and in reaching magnifications of 1,200,000 and 2 million times.

3. Conical projector.

In spherical projectors, large magnifications are achieved, as we shall see later, chiefly by reducing the size of the object, i.e. of the tip, since an excessive increase in the diameter of the glass bulb does not attain this aim and is impractical, causing considerable difficulties both in the manufacture of the bulb itself and in its subsequent glass-blowing treatment and in applying to the inner surface the fluorescent screen or the platinum layer. Therefore it is often preferred to use a conical design of projector, shown in Fig. 1,c (see, for example, ⁷˒⁸), whose screen is, in essence, a part of a spherical surface.

Such a design makes it possible to use relatively large radii of the sphere of the screen with ease, and, with small radii (5–10 cm), gives a very compact instrument. The use of only a small part of the spherical surface is, in this case, quite possible and sufficient, since, in projecting the faces of a single crystal onto a hemisphere, the whole image usually consists of a series of symmetrically arranged identical or mutually similar images of individual faces or their groups, and it is quite sufficient to limit oneself to obtaining the image of one of such faces or groups.

Fig. 8. Diagram of an ion-electron projector (after Maly).

A disadvantage of the conical projector, however, is the necessity of manufacturing bulbs with a good, geometrically regular dome, onto which the screen is applied, which is very difficult to accomplish in manual glass-blowing work. The manufacture of a small spherical, and still more of a cylindrical, projector, however, is quite within the reach of any glass-blower of average qualification.

Finally, it is also necessary to note the simple design of an electron projector proposed by Maly⁹ for simultaneous observation of the electron and ion emission of filamentary or cylindrical emitters. This design is shown schematically in Fig. 8. Electron emission from one side of the wire gives directly an image on the screen deposited on the anode and kept at a high positive potential relative to the cathode. At the same time, ion emission from the opposite side of this same cathode falls on a plate \(P\), negative relative to the cathode, where it causes emission of secondary electrons directed toward

anode–screen and which there give an image of the reverse side of the wire (cathode) in “ionic” rays, that is, a picture of its ionic emission.

Fig. 9. Images of monocrystals of pure metals: a — tungsten, b — molybdenum, c — nickel.

Fig. 10. Barium film on tungsten: a — barium deposited on the left; a monatomic film is visible at the boundary of the deposited layer; b — slight traces of barium.

A series of images obtained with the aid of spherical or conical projectors is given in Figs. 9, 10, and 11.

II. AREAS OF APPLICATION OF THE ELECTRON PROJECTOR

The principal purpose of the electron projector is the study of thermionic emission from various kinds of surfaces, ranging from pure metals to complex cathodes. However, in most such works, besides electronic emission, many other phenomena were also studied incidentally. Thus, for example, Benjamin and Jenkins4, in the first part of their work (1940), devoted to the study of emission from pure metals, also investigated the mobility of metal atoms under the influence of temperature and an external electric field, and determined the onset of noticeable mobility of surface atoms. Thus, for tungsten, molybdenum, and nickel they obtained values of this temperature equal respectively to 1170, 770, and 370° abs., whereas for iron they were unable to obtain a stable image of a single crystal even at room temperature.

Fig. 11. Cesium film on tungsten.

The mobility of surface nickel atoms under the influence of electric fields already at temperatures of the order of 100°C is used by these authors to explain the breakdown phenomena often observed in powerful generator tubes during their normal operation under conditions of good vacuum (see 10).

All works devoted to the study of thermionic emission, of films of electropositive contaminants on metals4, 5, 6, 7, 8, 11, 12, as a rule also include studies of the phenomena of adsorption, surface migration, and evaporation of the metals forming the film. For example, the mechanism of formation of a thorium film in the case of tungsten with an addition of thorium oxide, and the phenomena of thorium migration over the surface of tungsten, were definitively established as early as 1936 by Ehern and Becker13 and then confirmed and refined by subsequent works3, 4, 5, 6, 8. In contrast to this, the experiments of Benjamin and Jenkins11 on the migration of barium over the surface of tungsten initially gave negative results. Only by improving the degassing process of the entire projector and more carefully—

of the experiments, Becker and Muru,^12 and later Benjamin and Jenkins,^4 were able to establish the fact of barium migration on tungsten and molybdenum and to clarify the role of the slightest traces of oxygen and other contaminants in stopping the migration of barium in the first experiments.

The works listed made it possible finally to establish the difference in the adsorption properties of different faces of the crystal of the base metal with respect to various contaminants.

The same picture was obtained for the process of evaporation of an impurity, especially in the case of its monomolecular films. The rate of surface migration of contaminant atoms also proved to be different, depending on the crystallographic direction of the given surface region.

For the study of oxide cathodes, a cylindrical electron projector was recently used by M. D. Yagudaev and G. N. Shuppe.^14 In this work they compared the behavior of cathodes coated by the cataphoresis method with cathodes obtained by pulverization, and followed successively the processes of repeated activation and deactivation of oxide filaments.

Martin’s successful studies^5 of pure tungsten and of various films on its single crystal, at only 22-fold magnification, suggest the possibility of studying thermionic emission and various physicochemical processes on the surface of indirectly heated oxide cathodes by means of cylindrical or spherical projectors.

Using the possibility, already noted by Johnson and Shockley,^3 of observing the process of recrystallization of metals, Robinson^15 investigated, by means of a cylindrical projector, the dependence of the rate of recrystallization or, more precisely, of the growth of tungsten crystals on temperature, obtaining the following empirical dependence:

$$ R = 7.1 \cdot 10^{-3} \exp\left(-\frac{55\,000}{T}\right), $$

where \(R\) is the rate of recrystallization (crystal growth), in \(\mathrm{cm/hour}\), and \(T\) is the temperature in degrees absolute. This example illustrates the use of the electron projector for quantitative determinations, and not only for obtaining a qualitative picture of the phenomenon under study.

In addition to the phenomena enumerated above that have been investigated with electron projectors, mention should also be made of the diffusion of impurities in metals, an example of which is the mechanism, already mentioned by us, of activation of thoriated tungsten cathodes, and of the purification of metals from impurities and surface contaminants, in particular by various electronegative elements, such as oxygen. In this case we are directly confronted with questions of the kinetics of chemical reactions of metals with gases and vapors of various substances. As an example one may point to

work of Yagudaev and Shuppe[^14], who also investigated the adsorption of iodine on tungsten and the kinetics of the reaction of iodine with pure and thoriated tungsten. In doing so they found a considerable acceleration of the reaction of iodine with thoriated tungsten in comparison with its practically imperceptible reaction with pure tungsten. Apparently, here we are dealing with a phenomenon analogous to the corrosion of metals and the formation of microgalvanic elements (tungsten–iodine–thorium), which accelerate the reaction of iodine with tungsten; this is clearly confirmed by the deep etching of thoriated tungsten under the action of iodine vapors. In addition, it was found that after treatment of thoriated tungsten with iodine vapors and the subsequent removal of the latter, the activation process of tungsten proceeds more rapidly, which indicates a noticeable change in the structure of the surface layers of the tungsten.

In conclusion, one should point out the possibility of combined investigations of various objects by means of the electron projector and, of the same objects, by means of other instruments that give a quantitative estimate of various quantities. Thus, for example, Nichols[^16], making use of the results of a preliminary investigation of the single-crystal tungsten filaments obtained by him with a cylindrical projector, then placed them in a special apparatus in which, with the aid of a number of collectors of special design rotating around the filament, he was able to study the thermionic emission from different faces of the tungsten crystal and to determine the thermionic constants for these faces by the method of constructing Richardson straight lines, that is, dependences of the form $\lg I_e - 2 \lg T = f(1/T)$, where $I_e$ is the emission measured from a given face of the crystal (or in a given direction).

Nor is the possibility excluded of combining, in the electron projector itself, collectors that make it possible to carry out various quantitative determinations as well. Finally, as the work of Fox and Bailey[^17] on the electron-microscopic comparison of various complex emitting surfaces has shown, such quantitative determinations are apparently also possible by photometry of individual portions of the image on the projector screen or of photographs obtained from them. For this purpose only a preliminary “calibration” is necessary of the scale of the dependence of the intensity of the screen glow, or of the blackening of the photographic plate under the action of this glow, on the magnitude of the electron current falling on the corresponding portions of the screen.

The examples enumerated are quite sufficient to illustrate the possibilities of using electron projectors of various design in the most diverse fields of scientific research. It should be especially emphasized here the possibility of investigating various phenomena that require experiments under very clean conditions, attainable only with the aid of high vacuum, while at the same time obtaining, at high magnification, a visual picture of the phenomenon under study.

III. TECHNIQUE FOR MANUFACTURING ELECTRON PROJECTORS AND WORKING WITH THEM

1. General Structural Design

The desire to conduct the entire experiment under the cleanest possible conditions compelled most experimenters to use electron projectors sealed off from the pump and to manufacture their bulbs from the most refractory types of glass, in order to be able to degas, at as high a temperature as possible, both the bulb and the internal parts that cannot be baked out by high frequency or by passing current through them. Thus, for example, many authors use “Nonex”-type glass with tungsten seals, while Martin⁵ made his spherical projector (Fig. 7) from “Pyrex” glass and heated it for 60–70 hours during pumping at a temperature of about 500–510°C. Benjamin and Jenkins⁴ point out, among other things, that when working with bulbs made of low-melting glass they were unable to obtain stable images on the projector screen.

The use of a sufficiently effective getter, usually placed in a separate branch tube in order to prevent it, during sputtering, from reaching the object under study, makes it possible to obtain an extremely high vacuum, often reaching in sealed projectors \(10^{-7}\)–\(10^{-8}\) mm Hg and even lower. To measure the vacuum, the projector is provided with a separate branch tube with an ionization manometer sealed onto it. To remove excess readily condensable vapors introduced into the projector during experiments with them, another branch tube is required, with a trap cooled by liquid air or solid carbon dioxide by means of a Dewar vessel fitted over it.

Sources of various metals or impurities, whose influence on the object under consideration is to be investigated, are also placed in appropriately arranged branch tubes.

Among the advantages of sealed projectors, in addition to the possibility of carrying out experiments under conditions of the best vacuum and guaranteed cleanliness of all projector parts, one should also include the possibility of long-term preservation of projectors with the objects under study and of their repeated demonstration. At the same time, such projectors do not occupy the pumping apparatus.

In this respect, the designs of sealed projectors could serve as excellent visual aids, making it possible during the reading of special courses to demonstrate easily various phenomena from the above-listed areas of application of electron projectors.

However, a sealed projector has the fundamental disadvantage that it does not permit investigations with noncondensing vapors or gases after it has been sealed off from the pump, which—

significantly limits the area of its application. Moreover, for each object under investigation it is necessary to make a separate projector, which makes the investigations extremely laborious or forces one to confine oneself to single objects, as is seen from Martin’s work.^5

At the same time, the modern technique of obtaining a high vacuum and good vacuum sealing of various kinds of connections of vacuum instruments and circuits—such as, for example, plane or conical ground joints of large diameters—fully permits the construction of demountable electron projectors on ground joints with continuous pumping. Thus, for example, Yagudaev and Shupte^14, for their investigations of oxide cathodes and reactions of tungsten with iodine, used a cylindrical projector with a ground joint, schematically shown in Fig. 12.

Fig. 12. Cylindrical projector with a ground joint.

Such designs completely eliminate the above-mentioned shortcomings of the sealed-off projector, making it possible, with the aid of one instrument, to investigate an extremely large number of objects, recording the most interesting moments photographically or even on motion-picture film in the case of relatively rapid changes of the image on the screen. Moreover, with the aid of a single projector it is possible to repeat the same experiments many times with one or different objects, studying such phenomena as the influence of various gases and vapors, both condensing and non-condensing, pumping them out in the latter case for making observations on the screen.

A number of possible designs of such projectors with continuous pumping are shown schematically in Fig. 13.

2. Some instructions on preparing objects for investigation

a) Preparation of wires. The principal forms of objects investigated with the aid of projectors are wire and a point or sphere. In the case of polycrystalline wires, their preparation is most often reduced to polishing the surface in order both to obtain a diameter and cross section uniform over the entire length of the wire and to remove traces of drawing, which give on the screen a characteristic moiré (see Fig. 3), through which the crystalline structure of the metal shows.

Grinding and then polishing of wires are possible either mechanically, with the aid of a special machine (see, for example,^18), or

by chemical or electrolytic methods. Mechanical treatment makes it possible to obtain a much greater uniformity of the wire in diameter and cross section and is used in the case of more precise investigations. However, it requires special equipment. Chemical etching most often leads to the manifestation of the crystalline microstructure (even in the case of “bright” etching), owing to the different rates of action of the reagents on different crystal faces, and therefore does not make it possible to obtain the necessary “smoothness” of the wire surface. Therefore, most often it is preferred to use electrolytic polishing, which gives excellent results in many

Fig. 13. Possible designs of dismountable projectors: Ш — ground joint; for the remaining designations see Fig. 1.

cases of its application. However, the rate of electrolytic polishing requires very careful control; otherwise it is not possible to achieve the necessary uniformity of the wire along its entire length. For details of the process of polishing various metals and alloys we refer the reader to the book by L. Ya. Popilov[^19] and to the extensive list of specialized literature cited therein.

A more complex process is the production of single-crystal wires. Thus, in the case, for example, of tungsten, they can be prepared from polycrystalline wire, using the method proposed by Nicolson[^15]. Robinson[^15] converted polycrystalline wires into single-crystal ones by heating them for a long time in vacuum at constant temperature. Usually in this case the wire consists of several crystals arranged along it and partially overlapping one another, and, consequently, in individual cross sections of the wire there may occur

per 2–3 crystals. The graph in Fig. 14 clearly shows the process of formation of individual crystallization centers and the growth of crystals with time; moreover, after 3 hours of heating at \(2100^\circ\) abs., 7 crystals were located along the filament.

After completion of the principal recrystallization process by any method, additional heating to higher temperatures is recommended (for example, for tungsten at \(2950^\circ\) abs. in an atmosphere of hydrogen or at \(2700^\circ\) abs. in vacuum), with the aim of absorbing small surface crystals by the single crystals.

b) Fastening and tensioning of wires. In the case of a cylindrical projector, especially when using autoelectron emission from thin wires, that is, with the application of relatively strong external electric fields, which exert a considerable mechanical action on the wire, the choice of springs tensioning the filament under investigation plays an extremely important role. In this case the spring must not only keep the filament in a taut state during its thermal expansion while being heated, but also prevent the possibility of vibration of the filament. The frequency of these vibrations proves to be entirely unrelated, for example, to the frequency of the alternating current used for heating the wire, and oscillations of the latter may occur in a plane continuously changing its position about the axis of the instrument, along which the wire is stretched. The intensity of the vibration of the wire depends mainly on the voltage on the collector \(U\) and on the tensioning force of the wire \(F\). Thus, for example, for tungsten Danielson\(^{20}\) gives the following expression for the condition for the onset of vibration of a wire:

\[ U > 3.4 \cdot 10^{4}\left(\frac{F^{1/2}}{l}\right)\cdot D \cdot \lg \frac{D}{d}, \]

where \(D\) and \(d\) are respectively the diameters of the collector and the wire in cm; \(F\) is the tensioning force of the wire in g, and \(l\) is the length of the wire in cm.

Fig. 14. Graph of the process of recrystallization of a tungsten wire, obtained with the aid of a cylindrical projector.

From this, for the tension force of the wire at a given voltage on the collector \(U\), we obtain the expression:

\[ F \geqslant 8.64 \cdot 10^{-10}\, \frac{U^{2/2}}{D^2 \left[\lg \frac{D}{d}\right]^2}. \]

For the manufacture of springs, tungsten or molybdenum is usually used in the form of wire or ribbon. The springs are most often made helical. The method of calculation and design of tungsten and molybdenum springs is given, respectively, in the papers of Blodgett and Langmuir\(^{21}\) and of Touney\(^{22}\). In the case of tensioning wires of other metals, it is sufficient to calculate the permissible tension force of the tungsten spring required for this purpose, multiplying the tension force required for a tungsten wire of the same diameter by the product of the ratio of the melting temperature of the given metal to the melting temperature of tungsten (3355°) and the ratio of their linear coefficients of thermal expansion (see\(^{20}\)). In this way, for tantalum wires we obtain a factor equal to 0.83, for molybdenum—0.77, and for nickel—1.06.

c) Preparation and mounting of a metallic point. In the case of spherical or conical projectors, objects in the form of a sphere or a point are most convenient for investigation.

The method of making a sphere from a single crystal of tungsten is given in Martin’s paper\(^{5}\). We shall not dwell on it here, since for obtaining large magnifications it is more interesting to use points. They are made with equal success both from single-crystal and from polycrystalline wire, and the repeated experience of many experimenters has shown that, in the case of very fine points, usually obtained by chemical or electrolytic etching, a single crystal almost always remains in the point.

Tungsten and molybdenum points are usually etched electrolytically in a solution of caustic soda, and then in ammonia. Nickel is etched (also by electrolysis) in a solution of potassium perchlorate in dilute hydrochloric acid. The etching process, accompanied by repeated observations under the microscope, is continued until the point ceases to be clearly distinguishable. For tungsten, such points can readily be obtained with a radius of curvature of the order of 0.2–0.5 micron. Daniel\(^{6}\) gives a recipe for preparing tungsten points by electrolytic etching in a solution of 30 g of caustic soda per liter of water, and recommends carrying out the etching with alternating current, which gives more homogeneous points than in the case of direct current; the best results for wires 75 microns in diameter he obtained at a voltage of 4.4 volts, applied to the bath through a resistance of 20 ohms.

Subsequent heating of the tungsten tip for a short time to \(2950^\circ\) abs., or for a long time at \(2600^\circ\) abs., changes neither its shape nor its dimensions. Prolonged heating (of the order of 10 hours) at \(2700^\circ\) abs. removes only irregularities and some asymmetry, without any noticeable change in the diameter of the tip. Such tips, “smoothed” by heat treatment, give a single crystal with faces smooth to within atomic distances, which no longer affects autoelectron emission, since these distances are small compared with the “thickness” of the surface potential barrier.

Higher temperatures of treatment of the tip lead to its melting, with the formation of a bead and a considerable increase in the diameter of the tip. Müller\(^7\) succeeded in obtaining tips of much smaller dimensions, down to diameters of 360 and 220 Å, by applying cathodic sputtering for this purpose, i.e. bombardment of the tip with ions of inert gases or mercury.

Fastening the tip to the heater loop may be carried out in two ways, schematically shown in Fig. 15. In the first case (a), a tip of any metal may be spot-welded to the tungsten loop; moreover, as approximate calculations by Daniel\(^6\), made with an obvious overestimate of the results, showed, the temperature difference between the tip itself and the point of fastening on the loop is respectively \(6^\circ\), \(100^\circ\), and \(500^\circ\) at loop temperatures equal to 1000, 2000, and \(3000^\circ\) abs.

Fig. 15. Methods of fastening the tip on the heater.

The actual values of the temperature drop along the tip are undoubtedly considerably lower, and at temperatures not exceeding \(2000^\circ\) abs. they may in most cases quite safely be neglected. Somewhat substantial corrections, apparently, are necessary only in the case of annealing the tip to higher temperatures, required in treating objects, and not when observing them on the screen. In the second case (b), the tip is made directly at the end of one of the branches of the heater loop, and this method is therefore applicable only in the case of refractory metals. With such fastening of the tip, the temperature difference between the loop and the tip should apparently be still smaller, owing to the additional local heating of the contact of the two branches of the loop at the expense of its resistance.

d) Methods for determining the radius of curvature of the tip. Since one often has to deal with tips whose radii of curvature lie beyond the resolving power of an ordinary microscope, and, for example, Heffer\(^3\) had to determine their dimensions by means of an electron microscope at magnifications from 3 to 23 thousand times, some kind of

method of indirect determination of the dimensions of the tip. Such a determination is possible directly in the spherical (or conical) projector itself. According to Müller\(^7\), the radius of curvature of the tip \(r\) can be determined from the measured value of the field-emission current \(I_e\) from the tip under investigation at a given collector voltage \(U\), by the formula:

\[ r=\frac{2U(\lg 2\pi A-\lg I_e+2\lg 2U)-2\lg 2{,}3(\lg R-\lg r)} {2{,}3B(\lg R-\lg r)}, \]

where \(A\) and \(B\) are constants of the equation of field emission

\[ I_e=AF^2\exp\left(-\frac{B}{F}\right). \]

Although this equation is somewhat cumbersome, it nevertheless can in practice be used for calculations by the method of successive approximation. For this it is sufficient first to substitute in the right-hand side of the equation any approximate value of \(r\), and in subsequent substitutions the values of \(r\) calculated from this equation converge very rapidly. Since, however, a tenfold change in \(R\) changes the value of \(r\) by only 2%, it is quite possible to use the simpler, semiempirical formula

\[ r=\frac{0{,}85}{\varphi^2}\,U^{3/4}\qquad (\text{for } I_e=10^{-5}a), \]

where \(\varphi\) (the work function of the tip) and \(U\) are measured in volts, and \(r\) in Å.

d) Construction of the collector. In the case of a cylindrical projector, as we have seen, either a wire spiral directly adjoining the fluorescent screen (see Fig. 1a), or a platinum layer deposited in the form of a spiral on the inner surface of the bulb (see Fig. 2b), may serve as the anode-collector. The latter method, however, is more complicated, and the relatively broad strip of the platinum layer obscures too large a part of the image on the screen. Finally, Yagudaev and Shuppe made in their demountable projector a collector in the form of a welded molybdenum mesh placed at a distance of 5 mm from the screen. Such a collector can easily be degassed by heating it with high-frequency currents or by electron bombardment. For the same reasons, in the case of spherical or conical projectors it is most convenient to use a collector in the form of a molybdenum or, better, tantalum ring (see Figs. 6 and 7). Sometimes in spherical projectors\(^5\) the collector is made in the form of a transparent platinum layer deposited beneath the fluorescent layer of the screen over the entire hemispherical inner surface of the bulb. In this case, however, both the method of depositing the layer and making contact with it for leading it outside are very complicated and laborious.

e) Manufacture of the fluorescent screen. The fluorescent screen is usually made of willemite (orthosili-

zinc silicate) or calcium tungstate, or mixtures of these. For applying the screen it is simplest to use one of the following methods:

1) After thorough cleaning of the inner surface of the bulb by washing, successively, with acidified and alkaline water, a solution of dichromate in sulfuric acid, running water, distilled water, and alcohol, and then drying, it is coated with a 10% solution of potassium silicate and, after removing the excess, the layer is dried for 2 minutes, after which a dry phosphor powder is applied with a spray gun.

The final layer must be sufficiently thick so that, in the case of thermoelectronic investigations, the object under study, heated to high temperatures, would not shine through it, interfering with observation and photography.

2) The second method, which does not use a silicate binder, consists in grinding (matting) the inner surface of the bulb with alumina powder passed through a 120-mesh sieve, using a rotating brass disk mounted on a long shaft. After this, the matte surface is rubbed dry with phosphor powder. This method gives a more uniform, but also too transparent, layer, suitable, consequently, only for work with autoelectronic emission.

3) To obtain large uniform screens, the following method is more convenient: a mixture of willemite and calcium tungstate is ground with acetone in a porcelain ball mill for 6–8 hours.

Immediately before applying the screen, several drops (per 50 cm³ of suspension) of a solution of nitrocellulose in diatol (diethyl carbonate) are added to the ground mixture, and the suspension is applied to the glass with a spray gun.

In this case the nitrocellulose promotes better adhesion of the layer to the glass. To prevent scattering of the layer under the influence of electron bombardment and the fall of its particles onto the object under study, and, in work with vapors of alkali metals and especially cesium, to avoid their destructive action on the screen, the latter must be protected by a protective layer of potassium silicate.

For this purpose the bulb with the applied phosphor layer is heated in air to 550°C, burning out the organic impurities; and, after cooling the bulb, a 5% solution of potassium silicate is carefully poured into it (so as not to damage the layer). After draining off the excess solution, the bulb is dried, continuously rotating it in order to obtain a more uniform coating.

After removing the layer from unwanted places by wiping, the bulb is again heated in air to 550°C. Then it is thoroughly washed with a large excess of distilled water to remove free particles, and the entire process of applying the protective layer is repeated once more. A screen treated in this way requires more prolonged degassing

during pumping, but at the same time the possibility of contamination of the object under investigation is completely eliminated.

g) Pumping of projectors. The process of pumping projectors, apparently, can be carried out with the aid of both mercury-vapor and oil-vapor diffusion pumps. In the first case, after the instrument has been heated, liquid air is poured into the trap and the further treatment of the projector is carried out.

In the case of an oil-vapor pump, liquid air is not required, and water cooling and the use of special oil-vapor traps along the path from the pump to the projector system are sufficient.

However, in some more precise investigations it is necessary to bear in mind the possibility both of reaction of the object being heated, or of other parts of the projector, with residual oil vapors entering the instrument during pumping, and of the formation of insulating films as a result of the polymerization of organic vapors under the influence of electron bombardment and the catalytic action of metals. Such films were observed, for example, by Stuart^24, who proved their nature and origin under conditions of a purely electronic discharge in a vacuum with traces of organic vapors remaining even when the trap was cooled with liquid air. Therefore, until special experiments have been carried out on the comparative pumping of sealed-off projectors by mercury-vapor and oil-vapor pumps, and until a process has been developed for pumping them with oil-vapor pumps that guarantees the absence of any influence of residual oil vapors, it is apparently possible to recommend that sealed-off projectors be pumped only by mercury-vapor pumps with continuous cooling of the trap by liquid air.

The process of pumping sealed-off projectors is usually reduced to prolonged (60–70-hour) heating at a temperature of about 500–510° C (in the case of bulbs made of Nonex or Pyrex glass), degassing by current of the wire under investigation or of the heater with a tip, heating of the collector by high-frequency currents, and degassing of the ionization manometer, getter, and other auxiliary parts of the instrument. In addition, the screen may be additionally heated by electron bombardment. After thorough degassing and the performance of all processes necessary for preparing the given object for investigation, it is checked whether the maximum attainable vacuum has been reached; the getter is then evaporated, and the instrument is sealed off from the pump. With sufficiently thorough treatment during pumping, it is possible to obtain in sealed-off projectors a vacuum of the order of \(10^{-8}\)—\(5 \cdot 10^{-9}\) mm Hg.

Approximately the same order of vacuum can also be obtained with continuous pumping of projectors on ground joints; in the case of a mercury-vapor pump the trap must be cooled with liquid air the whole time, while in the case of oil-vapor pumps it is necessary to use special traps that protect the instrument from the entry of oil vapors into it.

In addition, with continuous pumping, in certain cases it is also possible to use various kinds of getters. Finally, apparently, the possibility is not excluded of constructing sealed demountable projectors with sufficiently good sealing of the ground joints and with the introduction of a getter in an amount sufficient to absorb gases throughout the entire period during which experiments with the projector are carried out.

3) Getters. The best type of getter for projectors is metallic barium; moreover, in its purest form it is obtained from the so-called barium–beryllate getter, which is

Fig. 16. Getter designs: a, b, c — barium–beryllate getter;
d — tablet getter.

a narrow, thin tantalum strip shaped like a small boat (see Fig. 16, a and b) or a spiral made of tantalum wire (Fig. 16, c), coated with a layer of barium beryllate obtained by calcining equimolar amounts of beryllium oxide and barium oxide hydrate or barium carbonate at a temperature of about 1300° C. On first heating to 1200° K, such a getter gives off an extremely small quantity of gas occluded in it, and then, owing to the reaction of tantalum with barium beryllate at temperatures above 1470° absolute, it releases exceptionally pure barium; the compound of beryllium and tantalum oxides formed in the process (beryllium tantalate) is extremely stable at the temperatures at which the getter is evaporated.

Evaporation of the barium–beryllate getter is possible either by passing current directly through the tantalum boat or spiral (Fig. 16, a–c), for which one extra seal-in is required for the current lead (any seal-in present in the instrument may be used as the second lead), or by heating with high-frequency currents, for which the getter strip is welded to a nickel bow (Fig. 16, b).

It is also possible to use, as a getter, tablets of an alloy of 50–65% barium with aluminum, fastened in special capsules consisting (see Fig. 16, d) of a nickel holder (shelf) and a nickel cover, which make it possible to evaporate the getter by heating it with high-frequency currents. To reduce gas evolution from barium–aluminum tablets, it is sometimes recommended to use tantalum holders and covers for them. To prevent the entry—

tact with barium on the object under investigation, the getter is usually either placed in a separate branch sealed to the projector, or, making use of the directed evaporation of barium owing to the presence of a special screen (Fig. 16, v) or a special form of the getter (the boat in Fig. 16, a–b), the getter is placed in the projector itself in such a way that, during evaporation, it cannot under any circumstances reach the object under investigation. Evaporation of the getter must, of course, be carried out only after it has first been thoroughly degassed under conditions of a sufficiently high vacuum, checked before evaporation of the getter by an ionization manometer.

Besides gettering, barium, along with many other metals, is also used for depositing films on metals for the purpose of studying, for example, its absorption, migration, or evaporation.

Strontium and calcium can be obtained in pure form, analogously to barium, from beryllates. Of the alkali metals, cesium, rubidium, and potassium are conveniently introduced in glass ampoules with capillaries, which are opened mechanically at the required moment. Sodium and lithium can be obtained from carbonates by coating a tantalum ribbon with them (analogously to barium beryllate). On heating, the carbonate first decomposes, and then, at higher temperatures, the oxides formed react with tantalum, liberating the pure metal. In order to obtain an absorbed layer and to avoid the formation, on the surface of the base metal, of crystallites of the deposited metals, evaporation of the latter must be carried out at the lowest possible temperatures, namely not above 1200° abs. for barium, 400° abs. for potassium, and 300° abs. for cesium; that is, with the calculation that an atomic layer is obtained in approximately 20–30 minutes.

IV. SOME QUESTIONS CONNECTED WITH THE USE OF ELECTRON PROJECTORS

1. Determination of the temperature of the object under investigation

To determine the temperature of a wire or of an edge heater in the case of tungsten, it is quite sufficient to restrict oneself to calculating the temperature from the values of the heating current and the wire diameter, using the tables given in the book by A. P. Ivanov[^26]. For tantalum, similar tables are given in the article by Maltter and Langmuir[^27]. Determination of the temperatures of wires and points made of other metals is best carried out with the aid of an optical micropyrometer. For this purpose, in the screen deposited inside the bulb, a small hole is burned through opposite the central part of the wire or of the edge heater.

The accuracy of temperature determination in both cases is from ±20° to ±50°, depending on the magnitude of the temperature itself. The temperature of a point welded to a heater can quite well,

as we have already seen, be regarded as approximately equal to the temperature of the central part of the heater loop. In the case of long filaments (10–20 cm), the influence of the cold ends may in all practical cases be neglected.

2. Circuit for Supplying the Electron Projector

The supply of the electron projector itself consists mainly of the heating supply for the object under investigation and a high-voltage source for the collector; in addition, separate supply circuits are needed for the ionization manometer, as well as for the gettering circuits for various metals used in the investigations.

The choice of current for heating the object plays an extremely important role, since the course of the wire recrystallization process depends on the kind of heating current. At sufficiently low temperatures, heating by alternating current produces no changes; at higher temperatures, recrystallization of the wire (crystal growth) is observed, followed by a slight “etching” of the surface due to evaporation.

An entirely different picture is observed when wires are heated by direct current. Under the action of the voltage drop along the filament, migration of negative metal ions is observed in the direction of the positive end of the filament; moreover, this phenomenon is found not only in the case of surface contamination by foreign substances, but also in the case of pure metals. At the ends, stepwise recrystallization appears, which was observed for tungsten and a number of other metals by Johnson28, and for tantalum by Mrowka29. Therefore, in order to preserve a homogeneous surface, it is best to heat the wires under study by alternating current. Heating of the tip heater in a spherical heater is likewise more conveniently carried out by alternating current.

The power of the collector supply sources proves to be comparatively low, thanks to the extraction from the object under investigation of extremely small currents, not exceeding tens or hundreds of microamperes in the case of field emission, or a few milliamperes for heated cathodes. Even with a collector potential of the order of 10–15 kilovolts, the power consumed by the collector does not exceed a few tens of watts, especially since, as the emission from the cathode increases, the potential required to obtain a good image decreases considerably. This is especially noticeable in the case of fine tips, i.e. at large magnifications, and when field emission is used, since the intensity of the latter is determined not by the collector potential, but by the field gradient at the wire or tip. Table I gives the collector voltages \(U\) at which, from a tungsten tip of diameter \(d\), a current to a spherical collector 14 cm in diameter equal to \(10^{-5}\) A is obtained.

Table 1

Dependence of the collector potential \(U\) on the tip diameter \(d\)
at an emission current of \(10^{-5}\) A

Even when selecting thermionic currents of the order of 15–20 µA in Martin’s projector[^5], for the purpose of heating the screen, a voltage of the order of 6–8 kilovolts was required, which amounts to a power of the order of 90–160 W.

Fig. 17. Circuit of a high-frequency high-voltage source for the collector.

Consequently, in any case one may use, for feeding the collector, the circuit of the so-called high-frequency direct-current source, consisting of a high-frequency tube oscillator, a high-frequency step-up transformer, and a rectifier using special kenotrons with a small capacitance between cathode and anode and a directly heated cathode powered from the high-frequency circuit. An example of such a circuit is given in Fig. 17.

and Table II gives the principal data of such circuits, calculated for various voltages.

Table II

Principal data of high-frequency high-voltage direct-current sources for supplying electron-optical instruments

Owing to rectification at high frequency, the pulsations of the rectified current are easily filtered out by a simple capacitor filter with a capacitor of very small capacitance, which makes the circuit, at low powers, entirely safe, because only insignificant charges accumulate on the capacitors; only at powers of the order of tens and hundreds of watts is protective interlocking of the circuit necessary. Such circuits are very compact and, in addition, make it possible easily to stabilize the output direct-current voltage. A more detailed description, with calculation of the circuits, may be found in Reich’s book30 and in the articles by Shade31, Mautner and Shade32.

3. Resolving power of the electron projector and the quality of the image on the screen

a) Cylindrical projector.

The resolving power of a cylindrical projector may be estimated by considering the paths of two electrons: one leaving the cathode (wire) in the radial direction, and a second leaving the same point with initial velocity \(\left(\dfrac{2kT}{m}\right)^{1/2}\), normal to this radius. The angular distance \(\theta\) of the end points of the paths of both electrons on the screen is given by the formula [see3]

\[ \theta = 2 \left( \frac{300 kT}{eU} \right)^{1/2} \left( \ln \frac{D}{d} \right)^{1/2} \int_{0}^{(\ln D/d)^{1/2}} \exp(-x^2)\, dx, \]

which, for \(D = 50.8\ \mathrm{mm}\), \(d = 0.127\ \mathrm{mm}\), \(U = 10\ \mathrm{kV}\), and \(T = 2000^\circ\) abs, gives \(\theta = 1^\circ\), or, for \(D = 50\ \mathrm{mm}\), a blurring of the image of a point

is approximately equal to 0.44 mm. When recalculated to the object, on the basis of a 400-fold magnification \(\left(\dfrac{D}{d}=\dfrac{50.8}{0.127}\right)\), we obtain a resolving power equal to 1.1 microns. An analogous calculation for the initial velocity along the wire axis shows that points at a distance of the order of 0.2 mm can be distinguished. At higher magnifications the resolving power in the transverse section of the wire increases. The field near the surface of the wire, for a given collector potential \(U\), may be calculated from the formula

\[ E=\frac{dU}{dr}=\frac{U}{(r-x)\ln \dfrac{R}{r}}, \]

where \(r\) is the radius of the wire, \(R\) is the radius of the collector, and \(x\) is the distance from the wire at which the potential gradient is determined.

Owing to the fact that the magnification occurs only in the direction perpendicular to the axis of the cylinder, the emitting points on the cathode surface should appear on the screen in the form of very thin lines along the circumference of the screen. In reality, however, they are imaged as bright, rather broad spots elongated along the circumference of the screen (see Fig. 2, \(a\)).

The reason for this is the nonuniformity of the field near such points. In addition, these spots may be further blurred under the influence of contact potential differences between separate portions of the surface possessing different work functions. A very vivid picture of local nonuniformities on the cathode surface is given by the image of an oxide cathode shown in Fig. 5, with the formation of a “lunar landscape,” i.e., with a clear predominance of emission from the edges of the regions of the cracked oxide layer.

b) Spherical projector. The resolving power of a spherical projector with field emission from a point is determined mainly by two factors, namely: diffraction phenomena and the distribution of both components of the initial electron velocities tangential to the sphere of curvature of the point. According to Müller\(^7\), the diameter \(D_0\) of the circle of scattering of the image of a point on the screen can be calculated from the equation

\[ D_0=\frac{0.78R}{U^{1/2}}\sqrt{2.89e+15.5\,\frac{\varphi^2}{U^{3/4}}} \]

or, when recalculated to the object, the resolving power is determined by the quantity

\[ d_0=\frac{U^{3/4}}{\varphi^2}\sqrt{2.89e+15.5\,\frac{\varphi^2}{U^{3/4}}}, \]

where \(\varphi\) is the work function, \(e\) is the electron charge, \(R\) is the collector radius, and \(U\) is the collector potential. Here the first term under the radical takes into account the influence of the initial velocities, and the second, diffraction.

To this value one must also add the influence of contact potential differences between individual facets or spots on the surface of the object. Practically, in the case of the magnifications cited above, of 420, 1200, and 2000 thousand times, the value \(d_0\) was respectively 40, 15, and 12 Å. It should be noted here that, in the case of a tip 360 Å in diameter, the structure of tungsten was still clearly distinguishable on the screen, whereas a tip 220 Å in diameter no longer made it possible to distinguish this structure.

In conclusion, Tables III and IV give the basic data for cylindrical and spherical projectors used by various authors.

Table III

Basic data for cylindrical electron projectors

Table IV

Basic data on spherical electron projectors

CITED LITERATURE

1. Brüche and Scherzer, Geometrical Electron Optics, Lenizdat, 1943, ch. VI.

2. Collection Methods for the Study of Catalysts, Publishing House of the Academy of Sciences of the USSR, 1948, section I, pp. 58—74 (Schechter’s article).

3. R. P. Johnson and W. Shockley, Phys. Rev. 49, 436 (1936).

4. M. Benjamin and R. O. Jenkins, Nature No. 3623, 599 (1939); Proc. Roy. Soc. A176, No. 3, 965, 262 (1940), A180, No. 981, 225 (1942).

1. S. T. Martin, Phys. Rev. 56, 947 (1939).
2. J. H. Daniel, Phys. Rev. 61, 9/10, 657 (1942).
3. E. W. Müller, Zeits. f. Phys. 120, No. 5/6, 270 (1943).
4. E. W. Müller, Zeits. f. Phys. 106, 541 (1937); 108, 668 (1938).
5. H. Mahl, Zeits. f. Phys. 108, 771 (1938).
6. B. Gossling, J. I. E. E., 71, 461 (1932).
7. M. Benjamin a. R. O. Jenkins, Phil. Mag. 26, 1049 (1938).
8. J. A. Becker a. G. E. Moore, Phil. Mag. 29, 129 (1940).
9. A. J. Ahearn a. J. A. Becker, Phys. Rev. 54, 448 (1938).
10. M. D. Yagudaev and G. N. Shuppe, Bull. Acad. Sci. Uzbek SSR, No. 7, 7—11, (1947).
11. C. S. Robinson, J. Appl. Phys. 13, No. 10, 647 (1942).
12. M. H. Nochols, Phys. Rev. 57, 297 (1940).
13. G. W. Fox a. F. M. Bayley, Phys. Rev. 59, 174 (1941).
14. R. P. Johnson, A. B. White a. R. B. Nelson, Rev. Sc. Instr. 9, 253 (1938).
15. L. Ya. Ponilov, Electropolishing of Metals, Mashgiz (1947).
16. J. H. Daniel, J. Appl. Phys. 12, No. 8, 645 (1941).
17. K. B. Blodgett and I. Langmuir, Rev. Sc. Instr. 5, 321 (1934).
18. G. L. Tawney, Rev. Sc. Instr. 10, 152 (1939).
19. R. Haefer, Zeits. f. Phys. 116, No. 9/10, 604 (1940).
20. R. L. Stewart, Phys. Rev. 45, 488 (1934).
21. E. A. Lederer, RCA Review 4, No. 3, 117 (1937).
22. A. P. Ivanov, Electric Light Sources, Vol. 1, G. E. Izd., 1938, pp. 145 and 152.
23. L. Malter and D. B. Langmuir, Phys. Rev. 55, 743 (1939).
24. R. P. Johnson, Phys. Rev. 54, 459 (1938).
25. B. A. Mrowca, J. Appl. Phys. 14, 684 (1943).
26. J. G. Reich, Theory and Applications of Electronic Devices, G. E. Izd., 1948, Ch. 14.
27. O. H. Schade, Proc. I. R. E. 31, No. 4, 158 (1948).
28. R. S. Mautner and O. H. Schade, RCA Review 8, No. 1, 43 (1947).