ELECTRON-OPTICAL STUDIES OF CATHODES

N. D. Morgulis, Kiev

I. Introduction

The development of electron optics and of the electron microscope based upon it, which began so fruitfully around 1932, has led to the acquisition of a whole series of new and very interesting data in the field of the physics of processes occurring during the operation of cathodes, and first of all of thermionic cathodes of various types.

The electron-optical method of investigation, in those cases where it can be applied, is of great importance, since it makes it possible to observe directly the course of processes taking place on the surface, in contrast to the integral method, ordinarily used up to now, of studying the general character of electron emission. The method of simultaneous electron-optical and electron-emission analysis that has been developed recently promises to give us the possibility of going much further in the investigation of emission processes at cathodes of various types. Unfortunately, however, the results of these investigations are for some reason completely passed over in silence in contemporary monographs on electron emission, such as, for example, those of Reimann¹ and de Boer², and are only briefly mentioned by Becker³, which is entirely unjustified in view of their considerable value. Only in the special monograph on electron optics by Brüche and Scherzer⁴, not translated into Russian, is there a review of these works, carried out chiefly in 1932–1933, whereas in the last two or three years (1934–1936) many more interesting works have appeared, produced with a more refined technique. Therefore the purpose of the present article is to give a review of the results obtained in works in this field carried out up to 1937, without at all touching upon the designs of electron microscopes or the methods of investigation, which in the present case would lead us too far afield. Those interested in this question should consult the monographs by Brüche and Scherzer⁴, the article by Brüche and Henneberg⁵, and also directly the original works. Finally, we shall dwell

briefly also on the results of electron-optical studies of photoelectron and secondary-electron cathodes, which, although they stand somewhat aside from the general subject of the present article, are at present of very great significance. In all this, on questions connected with the electron emission itself of various cathodes, we shall in what follows dwell only very briefly, referring those interested in it, for convenience of citation, almost exclusively to monographic literature.

II. Cathodes of Pure Metals

One of the first objects of investigation with the aid of the electron microscope was a cathode of pure tungsten; three successive images of a tungsten ribbon, obtained by Brüche and Johannson ^6 with increasing temperature, are presented in Fig. 1 at a magnification \(v = 75\). From these photographs it is clearly seen how, with such a gradual increase of the temperature of pure drawn tungsten, its recrystallization takes place, i.e. the formation of large crystals, which is especially clearly visible in Fig. 1,c, showing its state before burning out. For comparison, Fig. 2 presents two ordinary optical microphotographs of it, corresponding to the states 1,b and 1,c. It must nevertheless be admitted that the quality of the images obtained in this case (especially for non-equipotential cathodes ^7,8) with the aid of the electron microscope is not particularly good, especially in comparison with images of cathodes of other types (see below), and this was in part the reason for the transition to another method of studying the surface structure of pure cathodes, to which we now turn.

However, let us first dwell preliminarily on the following fact, which is very important for us. Despite the fact that, as it would seem, the emission properties of clean surfaces should have had the simplest homogeneous character, the matter, unfortunately, is not so. This is evident already at least from Fig. 1c cited above, representing the character of the electron emission of an ordinary recrystallized tungsten cathode, from which it is seen that, contrary to expectations, its emission proves to be by no means uniform over the surface, but has different values on its different crystalline surfaces. Thus we arrive at the conclusion that, owing to the crystalline structure of the surface of a pure cathode—which, moreover, changes during its thermal treatment—not only does the actual surface of the cathode prove to be larger than the geometric one, but the contact properties of differently oriented surfaces of the crystal also prove to be unequal. This last conclusion is in complete agreement with the data of experimental studies of the photoelectric effect from differently oriented surfaces of single crystals of zinc and copper ^9, which directly showed that the possible difference in the values of the work function may amount to several tenths of a volt. From this it also follows that, strictly speaking,

To the article by N. Morgulis

Fig. 1.

Fig. 2.

Fig. 3.

Fig. 4.

Fig. 5.

Fig. 6.

even for a cathode made of a pure, but polycrystalline, metal, the emission formula should already be written in a more complicated form, namely

\[ I=\frac{\sum I_n f_n}{\sum f_n}=T^2\sum A_n f_n e^{-\frac{e\varphi_n}{kT}}, \]

where the indices \(n\) refer to crystal surfaces differing in their emission properties. The application of this equation, instead of the usual simple one, introduces a whole series of additional complications, on which we shall not dwell here[^10]. To an equal extent, on the basis of this fact one can explain those deviations from Schottky’s theory which were obtained for cathodes with a nonuniform surface structure[^11].

Let us now return to our question concerning the electron-optical study of cathode structure. A number of inconveniences associated with the use of pure specimens for this purpose (indistinctness of the images, the necessity of exposure at high heating temperatures, etc.) led to a somewhat modified procedure in these investigations; namely, in the study by Brüche and Johannson of barium-azide cathodes on a nickel core (see below) it turned out[^12] that, as a result of a certain thermal treatment of the specimen under investigation, when the excess barium evaporates and only a thin film remains on the core, the crystalline structure of the nickel core appears especially clearly, as is seen from Fig. 3 (magnification \(v=30\)) and from comparison with an ordinary micrograph after etching (Fig. 4). It is found, moreover, that the thin barium film on the inhomogeneous crystalline surface of the core, by lowering the work function to unequal degrees because of its contact inhomogeneity, makes it possible to carry out the investigation under much more favorable conditions, greatly improving the quality of the images obtained. In this case the surface structure seems to show through this film, and its image proves to be very sharp and reproducible. Subsequently it turned out that, for a similar “development” of the surface structure of a metallic cathode, one can proceed in various ways:

1. Cover the core with a layer of oxide paste, subsequently subjecting it to the corresponding thermal treatment[^13].

2. Cover the surface of the core with barium from an atomic beam[^14], or, more simply still, by evaporation from an oxide cathode located nearby[^15].

3. Cover the surface with a layer of cesium[^17], formed by evaporation or deposition in equilibrium with its vapors.

4. Deposit on the surface a thin layer of thorium[^18] (for tungsten and molybdenum cores), etc.

Generalizing all this, one may say that, in order to obtain an image of the core, it must be covered with a thin, apparently of the order of monatomic, layer of an electropositive element, as this

as indicated above. Even a contamination film, usually present on an outgassed cathode and consisting partly of alkali metals, can be used here[^19]. In this case it turns out that, upon an initial increase in the temperature of the core (platinum–rhodium), the crystalline structure begins to appear and is already clearly visible at a relatively low temperature (890°C); if, however, the temperature is raised further, the image disappears and does not appear again even when the temperature is lowered to its former value. Of course, the quality of the image may be different with different film materials; thus, for example, with a cesium film the details of the structure are transmitted better than with a barium film. Also very interesting is the character of the image of a nickel core when coated with a cesium film, upon a gradual increase of its temperature, as shown in Fig. 5 for \(T = 400—630^\circ\mathrm{C}\). We see that at first the brightness and sharpness of the image increase, reaching a maximum at approximately 520°C, and then begin to decrease. This character of these images becomes quite understandable if one recalls that, as is known from the corresponding experimental investigations, at temperatures above approximately 500°C and at ordinary cesium vapor pressures the cesium film on the surface of the core becomes unstable and begins gradually to evaporate[^20].

The assumption adopted above in order to explain the images of the core structure obtained with thin electropositive films—that the presence of such a film leaves unchanged the contact inhomogeneity of the crystalline surface of the core—received its direct confirmation in the experimental work of Schenk[^21]. Owing to this inhomogeneity, the electron emission of the surfaces of different crystallites, which are visible in the image with unequal brightness, must also be unequal. To check this, a small hole of 1 mm was made in the screen, behind which a Faraday cylinder was placed, by means of which it was possible to measure the electron emission of the given small area. Moving the resulting image of the nickel core, covered with a barium film, along this hole, with the aid of an additional magnetic field it was possible to record the curve of the emission distribution while simultaneously photographing the image obtained. This is presented in Fig. 6, from which it is seen that at \(T = 1235^\circ\mathrm{K}\) the maximum inhomogeneity corresponds to an emission ratio of approximately \(1:20\). If one then makes use of Richardson’s formula

\[ I = AT^2 e^{-\frac{e\varphi}{kT}}, \]

assuming, approximately, the quantity \(A\) to be constant, then

\[ \ln \frac{I_1}{I_2} = 1.16 \cdot 10^4 \frac{\varphi_2 - \varphi_1}{T} \]

and, substituting here \(T=1235^\circ\mathrm{K}\) and \(\dfrac{I_1}{I_2}=20\), we obtain that

\[ (\varphi_2-\varphi_1)=0.22\ \mathrm{V}. \]

A similar value, obtained directly from experiments, for the difference in the work functions of differently oriented surfaces of crystalline nickel agrees quite well, in its order of magnitude, with the data indicated above, obtained by the direct application to this study of the photoeffect\(^9\). It is also interesting to note that the absolute value of the electron-current density obtained in this case (at \(1235^\circ\mathrm{K}\)) corresponded, approximately, to \(10^{-5}\ \mathrm{A}/\mathrm{cm}^2\), which is many times smaller than that obtained with a monatomic coating of the grain by barium. This indicates a different character of the given coating. One may think that in the present case we are dealing with an incomplete monatomic coating\(^ {22}\), and its degree may be different on different surfaces. Let us also note that Dosse and Knoll\(^ {23}\), using a method approximately analogous to Schenck’s method\(^ {21}\), measured the distribution of the electron current over the surface of the anticathode of an X-ray tube, which in this case proved to have a peculiar ring-like form.

From the equation written above one can draw one more interesting conclusion concerning the conditions for obtaining contrast images of the crystalline structure of the specimen under study. We shall assume that the brightness of the image at a given place of the fluorescent screen is proportional to the strength of the emission current arriving there, which has been directly proved for small current densities. In that case, for the ratio of the brightnesses of the image of two crystalline surfaces whose work-function difference is a constant quantity, we obtain

\[ \left(\frac{I_1}{I_2}\right)^T=\mathrm{const}. \]

Hence it is clear that the contrast of the image on the screen will increase with a lowering of the cathode temperature; for example, if in the above-mentioned example at \(T=1200^\circ\mathrm{K}\) we have \(\dfrac{I_1}{I_2}=20\), then at \(2400^\circ\mathrm{K}\) we would obtain 4.5, and at \(600^\circ\mathrm{K}\)—400. This consideration in particular justifies the transition, in studies of the structure of metallic cathodes, to the region of sufficiently low temperatures, for which it is necessary, in order to obtain greater emission, to cover them with a very thin film of an electropositive metal.

With this we shall end the exposition of the main factors connected with the study of the surface structure of metallic cathodes, leaving aside a number of other questions, such as, for example, the question of the so-called “fine structure” of the surface of individual crystallites\(^ {24}\), of phenomena at the boundary between individual crystallites\(^ {25}\), etc.

The above-described property of thin active films—namely, as it were, to reveal, when observed in an electron microscope, the crystalline structure of the metallic grain lying beneath them—has found an interesting application in the study of the surface structure of iron in its various phases and of the conditions of their mutual transformations. As is known, when iron passes from the $\alpha$-phase into the $\gamma$-phase, which occurs at approximately $898—906^{\circ}\mathrm{C}$, the structure of its crystal lattice changes from cubic with centered faces to centered-cubic, and this is accompanied simultaneously by an abrupt change in its other properties. Therefore the question was posed of the possibility of observing such recrystallization by means of the electron microscope, with sufficiently good visibility and reproducibility, not impaired as a result of possible instability of the revealing film. It turned out, however, that in this respect everything is quite satisfactory. For observation, the iron specimen under study was coated with a thin barium film, deposited by evaporation from an oxide cathode, and its crystalline structure was observed during heating up to about $1000^{\circ}\mathrm{C}$. The first successful experiments were carried out by Brüche and Knecht$^{26}$ with iron containing an admixture of carbon. In the photographs the transformations occurring in this process were very clearly visible. Subsequently, successful experiments were performed with pure electrolytic iron, and from the 12 photographs presented, taken after annealing it at different temperatures, it is evident that the transformation temperature lies at approximately $898^{\circ}\mathrm{C}$, which is in complete agreement with the known data. In view of the fact that the transformation of $\alpha$-iron into $\gamma$-iron occurs at $900^{\circ}\mathrm{C}$, when the emission of the specimen is very small, Brüche and Knecht did not succeed in photographing the actual dynamics of the transition, and they had to limit themselves only to visual observations. This, however, was accomplished by Burgers and Ploos van Amstel$^{27}$, who even recorded the dynamics of the transition on motion-picture film. In Fig. 7 are presented 6 separate photographs, in which the growth of an $\alpha$-crystal (beginning on the left) in the initial $\gamma$-phase is clearly visible. There is evidently no doubt that these interesting possibilities provided by electron optics for the study of crystalline structures will be successfully used in a whole series of other directions as well. We shall not dwell here on another method of studying the crystalline structure of metals with the aid of radiation from radioactive preparations$^{28}$.

In conclusion, let us dwell also on the result of investigations of autoelectronic emission by means of the electron microscope. As is known, when a strong electric field $E$ is created at the surface of a cathode, the surface potential barrier changes its form in such a way that even electrons situated at low energy levels can pass through this potential barrier and escape outward. The autoelectronic current obtained in this way can be determined from the following corre—

For the article by N. Moreulis

Fig. 7.

Fig. 9a.

Fig. 10.

Fig. 9c.

relation given by Fowler and Nordheim:

\[ I=A\frac{E^{2}}{\varphi}e^{-B\frac{\varphi^{3/2}}{E}f\left(\frac{\sqrt{eE}}{\varphi}\right)}. \]

Numerous experimental investigations \(^{29}\) confirm in general the character of this dependence, indicating, however, the presence of the following additional factors: 1. The actual electric-field intensity at the surface \(E_e\) is considerably greater than that determined from macroscopic calculations, i.e. \(E_e=\beta E\), where \(\beta \gg 1\). 2. The effective emission surface \(S'_e\) is considerably smaller than the geometrical one and is of the order of \(10^{-10}\)—\(10^{-9}\ \mathrm{cm}^2\). The meaning of both these factors becomes quite clear if one takes into account that, owing to the crystalline roughness of the surface, emission will occur chiefly from its active peaks. It must be admitted, however, that this quite natural conclusion is made only on the basis of indirect measurements of field-electron emission in the present case; therefore, in the work of Welter and Schilling \(^{30}\) an attempt was made to verify this with the aid of an electron microscope. The emitting surface of the cathode was a ground blade with an opening angle \(2\theta \simeq 10—15^\circ\), located at a distance of \(0.5\ \mathrm{mm}\) from the anode; Ta, Mo, W, Pt, Ni, Fe, Cu, Ag, and C were used as cathode materials. Assuming the cross-section of the cathode blade to be hyperbolic, it was possible to calculate the relation between the potential difference between the electrodes \(V_a\) and the field strength at the cathode, which corresponded to the condition \(E=620\,V_a\ \mathrm{V/cm}\). Then, in order to obtain satisfactorily reproducible images of the emission pattern on the screen, thorough degassing of the electrodes and maintenance of good vacuum conditions were, naturally, required. Despite all the measures taken, it was established that, after 10—30 min. of tube operation, the pattern, previously stable, began to change greatly, which the authors attributed to the onset of appreciable gas evolution from the cathode as a consequence of its strong heating in the emission centers. Then, in parallel with observations of the image of the emission pattern on the screen, measurements were made of the field-electron current.

Fig. 8.

Next, a graph is constructed from the Fowler—Nordheim formula, plotting on the abscissa the value

\[ \beta=\frac{E_e}{E}, \]

and on the ordinate \(S_e=\frac{i}{I}\), for the experimentally specified average

values of \(E\) and \(i\) and for two values of the work function \(\varphi\), equal to 3 and 5 V—this graph is shown in Fig. 8. In view of the fact that the actual values of \(\beta\) and \(s'_e\) will lie between these curves, to specify the problem an additional condition is still necessary, which Schilling attempts to introduce as a relation between the quantities \(E\), \(E_e\), \(s'\), and \(s'_e\), and which, for \(E = 3 \cdot 10^6\ \mathrm{V/cm}\) and \(s = 10^{-3}\ \mathrm{cm}^2\) (for \(2\Theta = 12^\circ\)), is shown in Fig. 8 as a dashed curve. The points of intersection of this curve with the two preceding ones give the values of interest to us, which are listed in Table 1. From consideration

TABLE 1

of this table we may conclude that these data of electron-optical analysis lead to such values of \(s'_e\) and \(\beta\) as previously had only been indirectly estimated in order of magnitude from measurements of field-electron emission, and which by their nature are quite natural.

III. Thoriated and carbide cathodes

One of the first, and in practice very valuable, types of active cathodes was the cathode made of thoriated tungsten. The phenomena occurring during the operation of such a cathode, which at first, on the basis of Langmuir’s first theory\(^ {31}\), seemed relatively simple, proved in subsequent investigations\(^ {32}\) to be much more complicated and to this day still far from clarified. Along with cathodes of thoriated tungsten, cathodes of thoriated molybdenum\(^ {33}\) and of carburized thoriated tungsten\(^ {34}\), partly related to it, also acquired considerable interest. This whole group of cathodes became the object of very interesting electron-optical investigations, to the exposition of which we now turn.

Already in one of the first works, Brüche and Johannson\(^6\) made an attempt to study the processes in thoriated cathodes by means of an electron microscope. Unfortunately, these experiments were technically insufficiently perfected, and therefore, without dwelling on them for the present, we shall pass directly to the most important works in this field, carried out by Brüche and Mahl\(^ {35}\).

In Fig. 9 are shown images of a strip of thoriated tungsten according to Brüche and Mahl at 25-fold magnification after its reduction at \(2900^\circ\ \mathrm{K}\) for \(\frac{1}{2}\) min. (a) and

during further activation at 2000°K for 60 min. (c). Consideration of these photographs, along with others that were obtained here, immediately leads us to the following important conclusions. The diffusion of thorium from within to the surface of the grain does not proceed uniformly over the entire surface through the crystal lattice, as is assumed in Langmuir’s old theory, but only at certain discrete places, whose location may change, say, as a result of annealing the cathode during the decomposition of ThO₂. Only after this does it spread over the entire surface owing to migration; however, on the nature of such migration, on the basis of these electron-optical investigations, we can as yet say nothing. It further turns out that the rate of activation of the cathode increases rapidly with increasing temperature, provided, of course, that we have not yet entered the region of temperature evaporation. It is a great pity that in this work, in parallel with the electron-optical investigation, no measurement was made of the electron emission of the cathode, as was done, for example, in the work of Heinze and Wagener^36 with an oxide cathode (see below), which could have given us a number of additional interesting data.

In Fig. 10 photographs are presented of the process of deactivation of the cathode during its annealing at 2600°K for 10 min., in which the crystalline structure of the tungsten grain is already clearly evident. This creates the impression that evaporation of Th from the surface also occurs nonuniformly, and that on some crystalline surfaces the thorium film is held especially firmly. Having obtained such a good image of the grain structure, Brüche and Mahl raise the question of whether the places where thorium emerges onto the surface are located precisely at the boundaries of these individual crystallites. From the experiments, however, it appears that the places where thorium emerges onto the surface, while quite discrete, are not confined to the crystal boundaries, but are also found within them. Apparently, however, one should for the time being refrain from such a conclusion, since our electron microscope, while showing the coarse-crystalline structure of the grain, may fail to show the presence in it of fine cracks through which diffusion of thorium to the surface is possible. Otherwise the presence of such discrete preferred places of thorium diffusion becomes completely incomprehensible; in any case, this question remains open for the time being and requires further investigation. We additionally note that evidence of the discreteness of the places where thorium emerges onto the surface of tungsten was also obtained in the works of Ahearn and Becker^37, as well as of Johnson and Shockley^38.

It is also interesting to note that, according to Brüche and Mahl, the number of centers from which thorium emerges onto the surface of tungsten is of the order of 1–100 per 1 mm², i.e., many times smaller than the number of thorium sources beneath the surface, which, apparently, can be explained by the necessity of the presence of pores for the diffusion of thorium outward. Nor can one fail to note that the image of the thoriated ...

of the cathode strongly resembles the “chessboard” structure that had earlier been introduced by a number of authors to explain the anomalous character of the Schottky effect[^11] obtained in this case.

In the work of Brüche and Johannson[^6], interesting photographs were obtained showing the character of the emission of a partially activated thoriated cathode as its temperature was gradually lowered. In this process the emitting film seems to break up into smaller, and then altogether contracts into small separate emitting regions; unfortunately, the work does not indicate whether this character of the emission is reversible, say, when the temperature is raised again. The authors also draw attention to the fact that this picture resembles certain well-known capillary phenomena with thin films, but they do not yet venture to say definitively what the nature of this analogy is.

Let us now turn to thoriated molybdenum[^35]. After cleaning the surface of the Th—Mo ribbon of all impurities by heating it at \(2200^\circ\mathrm{K}\), during which decomposition of thorium oxide \(\mathrm{ThO}_2\) also occurred, activation of the cathode was begun at \(1550^\circ\mathrm{K}\) for only 60 min. The character of activation of thoriated molybdenum coincides with that for thoriated tungsten, which was considered in fairly great detail above. A study of the deactivation of Th—Mo under bombardment by positive nitrogen ions likewise led, approximately, to the same results as in the thermal deactivation of Th—W. Finally, the rate of migration of thorium over the surface of a molybdenum core at the same temperature proves to be much greater than in the case of a tungsten core; for example, it is the same at \(2000^\circ\mathrm{K}\) on tungsten and at \(1540^\circ\mathrm{K}\) on molybdenum.

Let us turn to carburized thoriated tungsten[^35]. The Th—W strip used was carburized by heating in naphthalene vapors to such a degree that its resistance increased by 30%. It turned out that the presence of carbon makes possible the decomposition of \(\mathrm{ThO}_2\) even at \(2000^\circ\mathrm{K}\), i.e., at a much lower temperature than for an ordinary uncarburized cathode. If two identical strips of ThW are taken, of which only the second is carburized, and they are heated for a short time to \(2700^\circ\mathrm{K}\), the image thereby obtained shows that, in contrast to the ThW already known to us, carburized ThW gives a more blurred picture over the surface. This difference will be still more indicative if only half of a ThW ribbon is carburized and then the boundary between the two halves is observed with an electron microscope—such photographs are presented in Fig. 11. Fig. 11, \(a\) gives the picture after heating the strip at \(1500^\circ\mathrm{K}\) without a special regime for decomposition of \(\mathrm{ThO}_2\); in this case only the carburized half of the ribbon gives emission. After heating for 1 min at \(2650^\circ\mathrm{K}\), the following picture is obtained, in which points already appear that are associated with dif-

Fig. 11.

Fig. 12.

Fig. 13.

Fig. 14

thorium diffusion on the surface of pure tungsten. After heating at \(2800^\circ\mathrm{K}\), the carburized half is deactivated, whereas the non-carburized half gives an even more sharply expressed point structure. Finally, after heating in the activation regime—\(2200^\circ\mathrm{K}\) (Fig. 11, \(d\)), the non-carburized half is normally activated, while the carburized one, owing to the loss of all thorium during the preceding overheating regimes, remains dark.

From the figures one might form the impression that, in the case of carburized tungsten, thorium emerges uniformly onto the surface. However, the data of these authors concerning the emission pattern at a small degree of carburization (an increase in resistance by \(1\%\)) show that in this case the pattern corresponds in general to that obtained also for ordinary non-carburized thoriated tungsten, with the only essential difference that in this case we obtain a much larger number—depending on the degree of carburization—of emission centers facilitating the diffusion of metallic thorium from within to the surface of the grain. The authors relate this phenomenon to the loosening of the tungsten surface that occurs during its carburization, and confirm this by an electron-optical study of a Pinsch thoriated filament deposited on the surface, investigated by Maly’s method\(^8\).

In conclusion let us dwell on Maly’s work\(^39\), in which the question is investigated of the character of the poisoning action of oxygen on the electron emission of thoriated tungsten and thoriated molybdenum. As is known\(^40\), according to Langmuir, under appropriate conditions oxygen can form a monatomic film on the surface of tungsten, which will increase the work function as a result of the reverse orientation of the oxygen dipoles. If a strip of incompletely activated thoriated tungsten is heated at \(1600^\circ\mathrm{K}\) in an atmosphere of rarefied oxygen at a pressure of \(\sim 10^{-3}\) mm, then dark, ever-expanding spots appear in the image, indicating that oxygen is adsorbed over the entire surface nonuniformly, but chiefly at the boundaries of the emission spots, whence it then spreads further. If the oxygen is now pumped out and the temperature of ThW is raised to \(1700^\circ\mathrm{K}\), desorption of the oxygen begins. However, it evaporates again not uniformly, but starting from the edges of the spots covered by it. After complete desorption of all the oxygen, the emission pattern proves to be the same as before poisoning, i.e. the thorium layer that had been under the oxygen is practically unchanged (at temperatures of the order of \(1700^\circ\mathrm{K}\) diffusion of thorium from within to the surface is practically absent). It should be noted that at lower pressures (of the order of \(10^{-5}\) mm) adsorption of oxygen also in the middle of a thorium spot proved possible. It then turned out that, in the state of poisoning of thoriated tungsten by oxygen, those surfaces of the tungsten grain crystallites which, before poisoning,

were the brightest will now be the darkest, and in general the whole picture is now similar to that given by pure tungsten,—hence the author concludes that, with respect to electron emission, the dipole action of the oxygen film as it were compensates the dipole action of the thorium film. It further turns out that such an oxygen subfilm on the surface of a thorium film proves to be somewhat more stable than on the surface of pure tungsten, i.e., it permits a somewhat greater rise in temperature. As for thoriated molybdenum, the character of its poisoning by oxygen is approximately the same as in the case of thoriated tungsten.

IV. Barium cathode

By a barium cathode, in the direct sense of the word, one should understand a cathode made from some metallic core, covered on the outside with a thin film of active barium[^41], in contrast, say, to the thoriated cathode, where the active thorium layer diffuses to the surface of the core from within. However, barium cathodes are also the name given to cathodes made by other technologies, such as, for example, azide, thermite, etc.[^42], which in their final structure are rather close to the oxide cathodes considered below. In the second case, before the barium is applied the cathode core is preliminarily treated, and then, in order to obtain good emission, the cathode requires still additional activation, fairly similar in its character to that used in the case of oxide-paste cathodes. Following, however, the accepted differentiation, we shall now consider those works which are devoted to the electron-optical study of barium cathodes, i.e., cathodes made of a pure or preliminarily prepared core, onto which a layer of active barium is deposited by evaporation from some external source.

The first data concerning the character of the emission of azide-barium cathodes were obtained in the work of Knoll, Gautermans, and Schulze[^43]. The cathode used operated under the conditions of a gas discharge in argon at a pressure of 4 mm, at \(I = 0.75\) A and \(V_a = 30\) V for 10 min. As a result of intense bombardment by ions, which in particular led to sputtering, its emission pattern changed greatly, as is seen from Fig. 12. Fig. 12,a shows the appearance of the cathode before the discharge, and 12,b after it, at \(v = 15\); the contraction of the emission obtained here into a single active spot was studied at a greater magnification of 33 and 100 times (c and d). We see that this spot in fact has a highly inhomogeneous structure, which, during a prolonged discharge, burns out and gives emission only from a number of separate points.

Subsequently the azide-barium cathode was studied by Brüche and Johannson[^12]. The nickel core of this cathode had been preliminarily oxidized, and therefore, as a result of decomposition of the layer of barium azide \( \mathrm{BaN}_6 \) applied to its surface at \(890^\circ\mathrm{C}\), we shall be...

have the following structure: Ni—BaO—Ba (thick layer). The emission pattern corresponding to it is presented in Fig. 13, a, which after some time of heating changes into that shown in Fig. 13, b; this pattern apparently corresponds to a thick, relatively uniform layer of barium on the core. Raising the temperature to 970° C gave the pattern 13, c, after which, as a result of the action of the high temperature and the load, “burning out” of the barium began from within, with the formation of a dark, ever expanding spot (Fig. 13, d and f). Subsequently the structure of the core gradually begins to appear, which indicates that the barium layer present on the surface has become very thin as a result of the evaporation from it of all its excess. As for the physical conditions corresponding to this final state of it, with the manifested structure of the core, we have already become acquainted with them above, in the analysis of the conditions for obtaining an image of the crystalline structure of the core.

Finally, let us dwell also on another work of Brüche and Johannson¹⁴, devoted to a cathode obtained by depositing barium by evaporation from an external source onto a clean nickel core on which a grid had first been scratched. The first heating after the application of barium gave a pattern from which it was evident that, contrary to expectation, emission is given only by the scratched places and by a slight vein-like structure, while the rest of the surface remains dark. After some time of heating, the scratched places no longer stand out against the general background, giving approximately the same emission as the entire remaining core, while the veins in this case cease to give emission, becoming dark. In all this the barium layer on the core still remains rather thick, and only upon its further evaporation do we begin gradually to obtain an image of the structure of the core itself⁴⁴. Films of other active metals, such as, for example, cesium, etc., deposited on the surface of the core, will behave in approximately the same style, as was already indicated above.

Finally, mention should also be made of the work of Kemnitz, Knoll, and Walcher⁴⁵, in which the influence of artificially made scratches of various size and shape on the core of a cathode, upon the character of its electron-optical image, was investigated (the cathode used was a nickel core covered with an evaporated layer of barium).

Unfortunately, all the material available in this field of electron-optical investigations is rather limited, evidently in view of the fact that all investigations of this group proceeded in the direction of studying the structure of the core, as has already been discussed in detail above. As for the investigation of the emission processes themselves in thermionic-barium and azide-barium cathodes, which in particular have very great practical importance, in this field, apart from the material set forth above, we have nothing more for the time being. Although these cathodes are close to the ordinary oxide-paste cathodes, which have been the object of quite numerous investigations (see below), nevertheless alongside

At the same time they have their own noteworthy, specific features, about which, on the basis of the material cited above, apart from general considerations, nothing more can yet be said. On the other hand, it must be noted that, as is evident from the photographs obtained, contrary to expectations, the barium evaporated from an external source is not deposited at all uniformly on the exposed surface, but collects at certain discrete places, say, in the present case in the fissures of scratches, whence it then spreads—diffuses—over the rest of the surface. The latter interesting process of surface diffusion will be considered by us in detail below, in the analysis of the data from studies of oxide cathodes. As for the former—the nonuniformity of the metal film when it is deposited by evaporation onto the surface—it is of great independent interest, especially in connection with a number of data obtained by other methods^46 concerning the structure of such films, and therefore requires further development.

V. Oxide cathode

The oxide-paste cathode has been the object of the greatest number of investigations by means of the electron microscope, for quite understandable reasons: despite the fact that it was discovered more than 30 years ago, up to the present time it is still difficult for us to say anything definite about the character of the processes occurring here, which have always been studied by its electron emission^47; on the other hand, cathodes of this type also have very great practical importance. It should therefore be expected that in this case we must obtain a whole series of new and very important data. Let us therefore proceed to the exposition of these works.

Already in Broucke’s first work^48 it was shown that with the aid of the electron microscope one can obtain a good emission picture of an oxide cathode and then observe the character of certain processes occurring here. Subsequently this method was considerably improved in a very detailed work by Broucke and Johansson^49. In this work the processes on three oxide-paste indirectly heated cathodes were filmed cinematographically, with a total of 4497 frames obtained, at an average magnification \(v = 50\). In doing so the authors observed the following phenomena. If the surface of the core plate, after being coated with a layer of oxide, is scratched in the form of a grid, then at first the scratched grid appears dark against the general bright emission background; then a “negative” of the grid is obtained, when it is already luminous; then one can again obtain a “positive,” and finally once more a “negative.” Thus, regions of the cathode which at first did not emit can, as a result of the corresponding processes on the cathode, begin to emit, and conversely. Considering the possible causes of such reversal of the image, the authors note that the possibility is not excluded that this is caused by a surface displacement of barium oxide. Then the authors observe-

yielded a picture associated with the sputtering of an oxide cathode under bombardment by ions and under conditions of a not especially good vacuum in the tube, showing that the sputtering of the cathode, beginning accidentally at some point, then spreads more and more over the entire surface, ultimately leading to the deterioration of the whole cathode. Finally, it turns out that the picture obtained after reactivation has nothing in common with the picture that had been obtained with this same cathode at the very beginning of its operation, before its deterioration.

In the following paper by Knoll, Gutermans, and Schulze[^43], a number of new data concerning oxide cathodes are presented. First of all, these authors emphasize that the emission pattern of oxide-paste cathodes has a clearly pronounced nonuniform character: in it we see a number of islands, the shape and size of which depend on the method of applying and treating the oxide layer. These incandescent islands, the existence of which had been assumed here much earlier, according to the authors’ observations have a size on the order of \(10^{-2} — 5 \text{ mm}^2\). On the other hand, even the relatively large emission spots in thin-film oxide cathodes are essentially not homogeneous either, but have their own “fine structure.” Next, further interesting data are given concerning the nature of the process of activation of oxide cathodes under different methods of their treatment. The authors point out that, although it is difficult to give a definite conclusion about this process of activation of oxide cathodes, since for different cathodes the emission pattern may often differ considerably, it is nevertheless possible, as it were, to establish the following general rule. When activation is carried out only by strong overheating, without anode voltage, a relatively uniform picture of the distribution of emission is obtained, with a fine structure of the spots. If, however, activation is carried out under heating with the simultaneous application of an anode voltage (electrolytic activation), this leads to the formation of emission islands. Then, all other conditions being equal, a thin-film cathode gives a more uniform picture than a thick-film one. During activation under conditions of a gas discharge, thick-film cathodes continue to activate, giving as a result a peculiar structure with cracks. Thin-film cathodes also undergo their own characteristic changes in this case. During prolonged operation there occurs a gradual deactivation of the cathode, having approximately the same character as in the preceding work of Brüche and Johannson[^43]. The authors then further indicate that, when the emission of the cathode is calculated per actual surface of the emitting islands, very large densities are obtained, on the order of \(1 — 10 \text{ A}/\text{cm}^2\). During operation of the cathode, it is sometimes visible in the image that places which had been dark in the image begin to emit, and that at the same time a thin film of active barium is evidently creeping over them. The authors even estimate the rate of its spreading as \(0.01 — 0.1 \text{ mm}/\text{s}\). Finally, the work concludes

another attempt to observe the space charge of electrons at strong cathode incandescence.

As was already indicated above, in the work of Brüche and Johannson^49 an interesting fact was discovered: the “reversal” of the image, in which places scratched on the cathode, which had been dark at the beginning of operation, after a certain time of operation became light, i.e. began to emit, etc. To explain this fact the authors, in particular, advanced the hypothesis of the possibility of diffusion of barium oxide. However, this point of view was not confirmed later, for in studying the surface structure of the cathode with an ordinary microscope it was not possible to detect traces of barium oxide in these places, at first scratched but then having begun to emit.^50 Therefore one must turn to the assumption—perhaps even more natural—of the possibility of surface diffusion not of barium oxide, but of metallic barium obtained during activation of the oxide cathode. This group of phenomena was investigated in considerable detail by Richter,^51 to whose work we shall now turn.

First of all Richter shows that the observed phenomenon of image reversal is not connected with any defects of the experiment, and that, generally speaking, it has a much more general character. Then it is shown that such surface diffusion of the active substance can be observed under different experimental conditions; for example, in Fig. 14, a there is shown first the image of a pasted cathode partially (only in the center), then its electron-optical image at the beginning (Fig. 14, b) and at the end (Fig. 14, c) of operation—from this it is seen how the active substance spreads in all directions. A special experiment with a cleaned platinum strip laid on one half of the cathode showed that this phenomenon cannot be explained as a distortion connected with possible wetting of the oxide paste and the remaining surface of the cathode during its pasting. Finally, another special experiment showed that the reversal phenomenon is also obtained in the case of an oxide-paste grid, when the surface of the oxide and of the pure metal are in one plane. To explain this phenomenon the following point of view is put forward. The system of interest to us, consisting of a metallic surface only partially pasted with oxide, may be regarded as it is represented in Fig. 15. In all this it must be taken into account that the oxide layer has a considerably lower electrical and thermal conductivity than the metallic core. Therefore, if at the very beginning the system of Fig. 15 is heated and an anode voltage \(V_a\) is applied, then, owing to the absence of activation in the oxide, it will not yet give emission, and owing to its poor thermal conductivity it will be at an even lower temperature than the pure metal, which in this case is slightly emitting (Fig. 15, a). As the activation of the oxide spot increases, its emission current will increase considerably, liberating in it (owing to its poor electrical conductivity) a significant amount of Joule heat; consequently its temperature will become higher than that of the metallic core.

Owing to the onset of evaporation of barium from its surface in all directions (Fig. 15, b) and to the now lower temperature of the pure metal, the barium deposited on the latter will gradually activate it, while at the same time the oxide surface may possibly be deactivated, owing to its high temperature. If the anode potential is now switched off, we shall again return to the state represented in Fig. 15, a, when, owing to the absence of continuous replenishment, the adsorbed barium evaporates from the surface of the metal, thereby exposing it, and so on. Thus, by varying the cathode temperature, the anode potential, etc.,

Fig. 15.

one can obtain all those phenomena connected with the reversal of the image that were already discussed above. Here, however, it is important to emphasize the following circumstance. The formation of a barium film on the metal in fact occurs not by surface diffusion over it, as, for example, in the case of thoriated tungsten (see above), but by deposition from a peculiar atomic beam issuing from the oxide[^52]. This is evident if only from the fact that, when a depression lies in its path, barium is first found not in it but on the other side of it, while small elevations on the surface impede the spread of barium. A possible explanation of this is that the operating temperature in this case is considerably lower than in the case of a thoriated cathode, where, therefore, surface diffusion in its pure form becomes possible.

In Knecht’s work[^53], in which it was established that barium oxide itself does not take part in surface diffusion (see above), an interesting picture was also obtained of the change in the structure of an oxide-paste cathode under strong overheating. If it is heated to \(1150^\circ\text{C}\), in the absence of anode voltage, the oxide layer forms until, as a result of the evaporation of all its excess, the structure of the nickel core, usually obtained with a fairly thin coating of it by active barium (see above), begins to show through distinctly.

In the work of Mergulis, Nakhutin, and Nagorskii[^54], the processes of activation, deactivation, and reactivation of filament-type oxide cathodes were investigated, using a simplified-type electron microscope proposed by Johnson and Shockley[^38]. It was found that the process of activation of the oxide filament usually begins in its middle part, spreading from there gradually

on both sides toward its ends. The deactivation process proceeds in the reverse direction, i.e., the emitting zone of the cathode as it were contracts from the ends of the filament toward its middle, and, finally, the process of repeated reactivation again proceeds from the middle toward the ends. Such a cycle can be repeated many times on one and the same specimen. The reason for this peculiarity of all these processes lies in the specific nature of the temperature distribution along the filament, cooled at its ends by the leads supporting it; in this case the temperature will have its maximum value only in the middle, the least cooled part, and will gradually decrease in the direction of its ends. Since the rate of activation, and hence the degree of balancing of the oppositely directed deactivation process during cathodic sputtering of the active film, must increase strongly with increasing temperature[^55], this can explain in this way all the phenomena observed in this case. It should also be noted that the image of the cathode emission proved, of course, to be of an inhomogeneous character, with the usual fine structure only partly visible in the figure. We omit here a description of some other phenomena observed in this work.

From all that has been set forth above it is quite clear how great is the importance of the material that was obtained as a result of all the electron-optical investigations of oxide cathodes, especially in its further comparison with the material obtained by analysis of the electron-emission data. It must be said, however, that such a comparison, absolutely necessary for studying the very complex processes occurring here, is by no means always permissible, in particular owing to the considerable difficulty connected with creating sufficiently simple and reproducible experimental conditions. This circumstance, which presented great difficulties in investigations of electron emission, can be eliminated only if a method is created whereby it proves possible simultaneously to observe the electron-optical image and to make measurements of the electron emission of one and the same specimen. Such a very significant step forward in the methodology of cathode investigations, already partly realized in the works of Schenk[^21], and also of Venelt and Schilling[^30], found its further development in the new work of Gehnke and Wagener[^36]. In this work the activation processes of various oxide-paste cathodes were investigated, and the experimental method made it possible, together with electron-optical observations, to carry out simultaneously measurements of their electron emission. For example, in Figs. 16 and 17 there is presented the curve of the increase of emission of a heated oxide cathode with a nickel core under a purely thermal method of activation in the regime indicated on the graphs, as a result of which its emission increased approximately \(10^6\)-fold. In parallel with this, at the points marked by numbers, photographs were taken of the electron-optical image of the cathode. From a compa-

of all these data, obtained by both methods, one can at once draw one very important conclusion: the surface of our oxide cathode, already as a result of the first step of activation, when its emission still remains very small, is practically completely covered with emission centers, the character of whose distribution subsequently remains practically unchanged. In the electrolytic activation of the cathode the character of the growth of the emission current and of the change in the electron-optical image remained the same as in the preceding case of purely thermal activation. It then turned out that also in the case of poisoning

Fig. 16.

Fig. 17.

of the oxide cathode by oxygen, which leads to a strong decrease in its electron emission, the character of its electron-optical image also remains practically unchanged.

Therefore the authors come to the conclusion that the activation of oxide cathodes is very little connected with an increase in the effective emitting surface of the cathode, but depends chiefly on the decrease, in this process, of the value \(\Delta E\) in the Schottky emission formula\({}^{56}\) for oxide cathodes

\[ I = S \alpha V r T^{5/4} e^{-\frac{\varphi + \frac{1}{2}\Delta E}{kT}}, \]

where \(\varphi\) is the external work function of the electron, \(\Delta E\) is the work of transfer of an electron from the impurity level of metallic barium from the con-

tration \(n\) at the conduction level of barium oxide, \(S\) is the effective emission surface. It should, however, be borne in mind that even with an unchanged effective surface of the cathode such a considerable increase in the emission of the cathode upon its activation may be connected not only with a decrease in the quantity \(\Delta E\) or

\[ \left(\varphi + \frac{1}{2}\Delta E\right), \]

but at the same time also with an increase in the constant \(A\), associated with the growth of the concentration of the impurity of metallic barium in barium oxide \(n\), which leads to an increase in the concentration of free electrons in the conduction band of the semiconductor \(\mathrm{BaO}^{57}\). The conclusions obtained from this work clearly indicate that Espe’s representation\({}^{47}\), as well as de Boer’s thermionic theory of the oxide cathode\({}^{58}\), which attributed activation to a process occurring with only an increase in the effective surface at unchanged work function, must be recognized as incorrect, although doubts as to their correctness had arisen earlier as well. Finally, one should also note the complete possibility, shown in this work, of a purely thermal activation of an oxide-paste cathode with indirect heating, as well as the explanation of the absence of this in Espe’s older investigations.

These works for the time being exhaust all the electro-optical investigations of oxide cathodes available at present, which, despite their relatively small number, have nevertheless greatly advanced us toward an understanding of a number of processes occurring here.

§ VI. Photoelectron Cathode

After setting forth the data obtained by the electron-optical method concerning the character of the emission of thermionic cathodes, the natural question arises to what extent a similar investigation of cathodes of other types, and first of all photoelectron cathodes, is possible. It should be borne in mind that the photoelectron cathode has a number of features specific in this respect as compared with the thermionic cathode, namely: if we have a photocathode complex in its emission structure, the image of which we wish to observe with the aid of an electron microscope, then for this one may proceed in two ways. Either illuminate the photocathode with monochromatic radiation of such a frequency that, according to Einstein’s relation, only part of it could produce photoemission while another part would not. Or proceed by another, simpler path, illuminating the photocathode with such radiation under which in fact the whole surface would emit, and the structure of the surface would be visible because, under different conditions at different points of the surface, the absolute value of the photoemission would be different, although the ratio of photoemissions would thereby be smaller than that of thermionic emissions. In this respect lies one of the differences between the photocathode and the thermionic cathode: in the former case, as is known, the absolute value of the emission in dif-

particular points of the surface only partially and not to such a great extent depends on the electron work function as in the second. Therefore we shall encounter below cases in which the photoelectron and thermoelectron images of some specimen will be similar to one another, and also cases in which they will not be identical.

The first image of a photocathode, consisting of a zinc plate with drilled holes, when it was illuminated by a quartz arc, was obtained by Brüche ^59. These experiments were then continued by Pohl ^60, who carried them out with other materials; for example, Fig. 18 presents the electron-optical (18, a) and ordinary (18, b) images of an aluminum cathode (a = 5). In these experiments the well-known sensitivity of photocathodes to surface contamination, and in particular to grease, immediately became apparent; for example, the inscription on the image in Fig. 18 was made with grease. Subsequently Pohl ^61 investigated a photocathode with a gas film on the surface in its various states, with simultaneous observation of the crystalline structure of the photocathode. For example, a platinum strip was taken which was first heated for 20 hours in air at 1450°C, which led to its recrystallization. If one now examines the image of its photoemission, it turns out that at first it gives a fairly uniform photoemission; then, if its temperature is raised in vacuum to 900°C, its emission already becomes nonuniform; with a further increase in temperature to 1040°C, thermoelectron emission already predominates, showing in this case a typical crystalline structure; finally, if the cathode is cooled, then the photoemission obtained in this case will already be nonuniform, giving an image of the structure very similar to the thermoelectron one. Thus, we encounter a similarity between the photo- and thermoelectron images and the possibility of photoelectron observation of the cathode structure. In contrast to this, ^62 cases are observed in which the thermoelectron and photoelectron images differ considerably from one another.

Then, in the work of Mahl and Pohl ^63, images were obtained of a photocathode made of a copper core coated with a layer of potassium. First a thick layer of potassium was deposited on the copper plate; then this layer was treated by a prolonged discharge in hydrogen, as a result of which the photoemission was reduced to the emission of droplet-shaped potassium spots. Finally, with additional deposition of potassium, the photocathode retained its “droplet-shaped” structure, only with greater density. It is interesting to note in the figure the presence of shadows from the droplets, associated with lateral illumination of the surface and lateral deposition of potassium. The authors then also observed a photoelectric image of the process of melting tin. Finally, it proved possible also to observe photo-images of a number of minerals, such as, for example, galena, chalcopyrite, and others ^64. Thus, the indicated works demonstrate the possibility and fruitfulness of such an investigation of electron-optical

images of various photoelectron cathodes. Unfortunately, here one can only note the absence of published works on the electron-optical study of photocathodes of other types, and first of all cesium ones.

At this point one may also note works connected with the conversion of optical images into electronic ones, subsequently observed on a screen^65. With the use of appropriate photocathodes, even images in the invisible infrared region can be used, which, in particular, is of very great importance also in the methodology of laboratory investigations^67.

§ VII. Cold Cathode

Let us also dwell on the results of electron-optical investigations of cold cathodes, for the imaging of which one may use both their secondary electrons and primary ones—transmitted or reflected. In these investigations, however, we encounter one very substantial complication, which was absent for all the preceding types of cathodes. The point is that, as a result of bombarding a metallic surface with a monochromatic beam of primary electrons of velocity \(v_p\), we obtain the emission of an entire group of electrons with different initial velocities in the range \(0 < v < v_p\)^68: elastically reflected, inelastically reflected, and secondary electrons proper. The ratio between them depends on a whole series of factors, such as, for example, the magnitude of \(v_p\), the material and thickness of the specimen being bombarded, etc. The resulting velocity inhomogeneity of the electron beam emerging from the specimen leads to great difficulties in focusing it, in contrast to the preceding types of cathodes, which gave electrons with practically identical initial velocities almost equal to zero. Some relief in this case is provided only by the fact that, practically, in the velocity spectrum of electrons emitted from a cold cathode, we have a predominance mainly of two groups: 1) secondary electrons proper with a very small initial velocity, and 2) elastically reflected electrons with almost the full velocity \(v_p\). Despite this difficulty, the electron-optical study of cold cathodes has yielded a number of interesting results, to the exposition of which we now turn.

On the basis of what has been said above, all the investigations performed on cold cathodes can be divided into the following four groups:

I. Obtaining an image of the front side of the bombarded metal specimen using fast reflected electrons.

II. Obtaining an image of the rear side with the aid of fast electrons transmitted through it.

III. Obtaining an image of the front side with the aid of slow secondary electrons.

IV. Obtaining an image of the rear side with the aid of slow secondary electrons.

To the first group belongs the work of Ruska^69. The object investigated

To the article by N. Morgulis

Fig. 18.

a  b

Fig. 20.

a  b  c

Fig. 21.

a  b  c

Fig. 22.

“illuminated” by electrons with a velocity of 20–80 kV, and the image obtained thereby had \(v = 10\). In this case we are dealing with reflected, or so-called back-diffused, electrons, which retain almost completely their initial velocity; of course, in all this the thickness of the specimen under investigation must be greater than the so-called Lenard thickness of back diffusion \(^{70}\). In this way images were obtained for a gold-platinum plate of thickness \(0.4\) mm, at an angle of incidence of the electrons of \(60^\circ\). The surface structure of the plate is seen rather poorly in these images, although nevertheless differently than in an ordinary optical image. The reason for this lies in the fact that the reflected electrons can still penetrate rather deeply (approximately to hundreds of atomic diameters) into the plate. When the thickness of the specimen is decreased, the image becomes weaker, owing to the decrease in the number of back-diffused electrons, but, on the other hand, sharper owing to their greater monochromatization.

The second group includes the works of Borris and Ruska \(^{71}\) and of Ruska \(^{72}\). In this case the image given by the specimen under investigation—for example, a thin metallic plate, Lenard windows, etc.—was studied when a beam of fast primary electrons passed through the specimen. From these experiments one may attempt to draw certain conclusions about the nature of the absorption and scattering of electrons, the structure of the plate being shot through, etc. A number of photographs were also obtained at very high magnifications; for example, Fig. 19 shows an image of aluminum foil of \(0.8\mu\), at magnification \(v = 7800\), which, owing to strong heating by the bombarding electrons, was partially destroyed.

Fig. 19.

Subsequently this line of work was developed by Bersch \(^{73}\), who analyzed the meaning of the pictures of the “thickness structure” of the leaf under investigation obtained in this case and further developed this method of investigation, obtaining in particular a number of interesting photographs. Then he proposed, for investigating the structure of individual small crystallites of the leaf, first obtaining on the first screen a magnified image of the whole specimen (\(v = 10\)), after which, by diaphragming the required place, to investigate it by the diffraction method on the second screen. All this enabled him to carry out an analysis of the shape, posi-

of the orientation and structure of individual small crystals ($<0.1\,\mu\mathrm{m}$) of the sheet under investigation. Finally, let us also point to the work of Marton^74, who carried out an investigation of organic substances by observing the electron-optical image obtained when a rapid beam of electrons passed through them (after appropriate treatment).

The third group includes the works of Zworykin^75 and Knoll^76,^77. In Zworykin’s work the surface of carbonized nickel was bombarded with electrons at a velocity of 20 V, and the image produced in this case, in Zworykin’s opinion, by slow secondary electrons was photographed; evidently, reflected electrons also played some role here. On the surface of this emitter the letter A had been scratched, so as to remove the upper layer of carbon. Since the surface of pure nickel gives a greater secondary emission than the surface of carbon, this letter was visible on the screen quite clearly as a dark one on a lighter background. An attempt to obtain a similar image from a scratched thin layer of silver on mica did not lead to satisfactory results, owing to charging of the mica at the site of the scratches*.

The fourth group of works includes the works of Knoll and Lubszynski^78 and Béné^79. In the first of these, the image of a metal mesh, produced by its secondary electrons and visible on the screen as light against a general dark background, was investigated, in contrast to the dark image on a light background obtained with primary electrons. The authors point to the possibility of investigating in this way the secondary emission of various objects and its distribution over the surface, and cite as an example the case when the mesh was half covered with zirconium, which reduced its secondary emission there; therefore the image of this mesh turned out to be simultaneously half light and half dark.

Let us turn to the new work of Béné^79, on which we shall dwell in somewhat greater detail in view of the fact that it is perhaps the most interesting of this entire cycle of works. In this work an aluminum foil $0.6\,\mu$ thick was shot through by rapid electrons obtained on its reverse side. The author’s method made it possible in this case to focus only this image, separating it from the image produced by the primary rapid electrons. In Fig. 20 are shown images of this plate at three different velocities of the bombarding electrons: 6, 7, and 9 kV at $v = 5.5$; we see that the best image of the structure is obtained at 7 kV. The white spot visible in the photographs is the trace of primary electrons that passed through the hole in the foil, which was shown by the author by the action of an external magnetic field on the image.

The necessity of using electrons of a definite velocity in order to obtain a clear picture of the structure, as is seen from Fig. 20,

* In the same work of Knoll^76,^77, using a peculiar method with a synchronously tuned Braun tube, some data were obtained on the secondary emission of metals and an insulator ($\mathrm{Al_2O_3}$).

may be explained as follows. To obtain an optimal image, it is necessary to choose the velocity of the scanning electrons so that the thickness of the specimen under investigation corresponds to Lenard’s “limiting thickness” \(\delta^{70}\). Indeed, if it is greater than \(\delta\), then most of the fast electrons incident on the surface will be reflected back. In the other case, if it is less than \(\delta\), then most of the primary electrons will pass through without appreciable loss of velocity. Only in the case when it approximately corresponds to this value will a large number of slow secondary electrons arise in the boundary layer at the rear surface, and these will give the image of interest to us. Confirmation of this explanation is provided by the fact that direct experiments on scattering and emission in Al gave, for \(v_p=6\ \mathrm{kV}\), the value \(\delta \simeq 0.7\,\mu\). For gold foil of \(0.1\,\mu\), the image optimal in contrast is obtained at \(4.5\ \mathrm{kV}\), whereas according to direct experiments for this electron velocity \(\delta \sim 0.08\,\mu\). Then, using the formulas of Henneberg and Recknagel, Bene, in agreement with the data of Becker’s direct investigations, found that the initial velocity of its secondary electrons is of the order of \(2\ \mathrm{V}\).

Next, a comparison was made of the images obtained under similar electronic and ordinary optical transillumination of a gold leaf \(0.1\,\mu\) thick, shown in Fig. 21 \((v=65)\). We see that the optical image (a) is somewhat similar to the electronic image (b and c) with respect to the demonstration of the structure of the leaf, although it is inferior to the latter in sharpness. Finally, Bene also succeeded in observing an interesting electron-optical image of the process of gradual melting of the leaf, owing to the large energy developed in it when it is bombarded by primary electrons. This image is shown in Fig. 22 \((v=60)\), and from it one can see how, around the melted place, the metal of the leaf assumes a new structure in comparison with the structure of the undamaged part visible around it. Thus this work of Bene undoubtedly represents a considerable step forward in comparison with all preceding works and promises to yield much of interest, in particular for the study of the structure of Lenard windows for various cathode oscillographs, Coolidge electron guns, etc., directly under the different regimes of their operation.

§ VIII. Conclusion

From all the foregoing we see how much that is new the electron-optical method has given us in roughly five years for the investigation of processes occurring during the operation of cathodes of different types, and first of all thermionic ones. During this time the very methodology of electron-optical investigations has also been greatly improved.\(^{80}\)

It should be noted that in the methodology of these investigations there is at present apparently another considerable step forward in prospect. The point is that, as has already been indicated above, almost all

The investigations carried out up to the present time have in fact amounted to observing the electron-optical image of the cathode on a screen, and the results of these observations have been compared with the results of earlier investigations of these cathodes, carried out by the usual method of analyzing their integral electron emission. Recently, however, it has begun to become clear—and this is seen most distinctly in the example of the work of Heinze and Wagener described above, ^36 that such a comparison is not always and in every case so directly permissible. Therefore it should apparently be considered most correct and desirable to recognize a synthesis of the old electron-emission method and the new electron-optical method of investigation; in this direction one may apparently expect, in the near future, the appearance of a number of interesting works. In this case one may hope to obtain a whole series of new data connected primarily with processes in complex cathodes, such as, for example, the determination of the degree of coating corresponding to optimal electron emission, the determination of the actual emission constants and work functions of cathode regions differing in their structure, the Schottky effect, autoelectronic emission, space charges, etc., i.e., in those cases where the series of very valuable data obtained by us with the aid of the indirect electron-emission method, owing to the considerable complexity of the surface conditions, unfortunately still has a rather limited significance.

REFERENCES

1. A. Reimann, Thermionic emission, 1934.
2. J. de Boer, Electron emission and adsorption phenomena, 1935 (Russian translation, 1936).
3. J. Becker, Thermionic emission and adsorption (I), Rev. Mod. Phys., 7, 95, 1935.
4. E. Brüche u. O. Scherzer, Geometrische Elektronenoptik, 1934, S. 239—269.
5. E. Brüche u. W. Henneberg, Geometrische Elektronenoptik, Ergebnisse der exak. Naturwiss., 1936.
6. E. Brüche u. H. Johannson, Physik. Z., 33, 898, 1932.
7. H. Seeman, Z. Physik, 79, 742, 1932.
8. H. Mahl, Z. Physik, 98, 321, 1935.
9. A. Nitzsche, Ann. d. Phys., 14, 463, 1932; N. Underwood, Phys. Rev., 47, 502, 1935.
10. E. Brüche, Z. Physik, 98, 97, 1935; A. Recknagel, Z. Physik, 98, 355, 1935.
11. See, for example, J. Becker, Rev. Mod. Phys., 7, 118, 1935; E. Brüche, Z. Physik, 98, 99, 1935; Z. techn. Phys., 17, 591, 1936.
12. E. Brüche u. H. Johannson, Z. techn. Phys., 14, 487, 1933; E. Brüche u. O. Scherzer, l. c. p. 250.
13. W. Knecht, Ann. d. Phys., 20, 161, 1934; E. Brüche, Z. Physik, 98, 79, 1935.
14. E. Brüche u. H. Johannson, Z. Physik, 84, 56, 1933.
15. E. Brüche u. W. Knecht, Z. techn. Phys., 15, 461, 1934.
16. D. Schenk, Z. Physik, 98, 753, 1936.
17. D. Schenk, Z. Physik, 98, 753, 1936; R. Johnson a. W. Shockley, Phys. Rev., 49, 436, 1936.

1. E. Brüche and H. Mahl, Z. techn. Phys., 17, 262, 1936.
2. See E. Brüche and O. Scherzer, l. c., p. 244.
3. See, for example, A. Reimann, l. c., p. 135.
4. D. Schenk, Ann. d. Phys., 23, 240, 1935.
5. See E. Brüche, Z. Physik, 98, 87, 1935.
6. I. Dosse and M. Knoll, Arch. Elektrotechn., 29, 729, 1935.
7. H. Mahl and D. Schenk, Z. Physik, 101, 117, 1936; E. Brüche and W. Knecht, Z. Physik, 92, 452, 1934.
8. See E. Brüche, Z. Physik, 98, 102, 1935.
9. E. Brüche and W. Knecht, Z. techn. Phys., 15, 461, 1934; 16, 95, 1935.
10. W. Burgers and J. Ploos van Amstel, Nature, 136, 721, 1935.
11. G. Tamman and A. Löwis of Menar, Z. anorg. allg. Chem., 205, 145, 1932.
12. See, for example, S. Dushman, Rev. Mod. Phys., 2, 471, 1930; A. Ahearn, Phys. Rev., 44, 277, 1933.
13. A. Wehnelt and W. Schilling, Z. Physik, 98, 286, 1935; W. Schilling, Berl. Dissert., 1936.
14. I. Langmuir, Phys. Rev., 22, 357, 1923; S. Dushman, Rev. Mod. Phys., 2, 398, 1930.
15. See, for example, A. Reimann, l. c., p. 104; J. de Boer, l. c., 104; I. Langmuir, Journ. Frankl. Ins., 217, 543, 1934.
16. See, for example, A. Reimann, l. c., p. 158.
17. See, for example, H. Simon, Herstellung von Glühelektroden, Handb. Experimentalphysik, XIII/2, 1928, p. 301; L. Koller, The physics of electron tubes, 1934, p. 31.
18. E. Brüche and H. Mahl, Z. techn. Phys., 16, 623, 1935; 17, 81, 262, 1936.
19. W. Heinze and S. Wagener, Z. techn. Phys., 17, 645, 1936.
20. A. Ahearn and J. Becker, Phys. Rev., 49, 879, 1936.
21. R. Johnson and W. Shockley, Phys. Rev., 49, 436, 1936.
22. H. Mahl, Z. techn. Phys., 17, 653, 1936.
23. See, for example, S. Dushman, Rev. Mod. Phys., 2, 417, 1930.
24. See, for example, A. Reimann, l. c., p. 159.
25. See, for example, W. Espe and M. Knoll, Werkstoffkunde der Hochvakuumtechnik, 1936, p. 287.
26. M. Knoll, F. Houtermans and W. Schulze, Z. Physik, 78, 340, 1932.
27. W. Knecht, see E. Brüche, E. Physik, 98, 88, 1935.
28. G. Kempnitz, M. Knoll and W. Walcher, Z. Physik, 96, 612, 1935.
29. See, for example, E. Riehl, Collection of New Ideas in the Field of Catalysis, 1932, p. 61. Yu. Khariton and A. Shalnikov, The Mechanism of Condensation and Formation of Colloids, 1934, p. 18.
30. See, for example, A. Reimann, l. c., p. 188, W. Espe and M. Knoll, l. c., p. 270.
31. E. Brüche, Naturwiss., 20, 49, 1932; E. Brüche and H. Johannson, Naturwiss., 20, 357, 1932.
32. E. Brüche and H. Johannson, Ann. d. Phys., 15, 145, 1932.
33. W. Knecht, Ann. d. Phys., 20, 177, 1934; E. Brüche, Z. Physik, 98, 80, 1935.
34. E. Richter, Z. Physik, 86, 697, 1933; 87, 686, 1934.
35. See, for example, E. Brüche, Z. Physik, 98, 81, 1935.
36. W. Knecht, Ann. d. Phys., 20, 161, 1934.
37. N. Morgulis and I. Nakhutin, Journal of Experimental and Theoretical Physics, 6, 1108, 1936; N. Morgulis and A. Nagorskii, Journal of Experimental and Theoretical Physics.
38. See, for example, N. Morgulis and A. Patiokha, Journal of Technical Physics, 7, 483, 1937.
39. W. Schottky, Naturwiss., 23, 116, 1935.
40. W. Schottky, Z. techn. Phys., 17, 653, 1936; N. Morgulis and I. Nakhutin, Journal of Experimental and Theoretical Physics, 6, 1107, 1112, 1936.
41. J. de Boer, l. c., p. 339.
42. E. Brüche, Z. Physik, 86, 448, 1933.

1. J. Pohl, see E. Brüche and O. Scherzer, l. c., p. 263.
2. J. Pohl, Z. techn. Phys., 15, 579, 1934.
3. G. Seitz, see E. Brüche, Z. Physik, 98, 93, 1935.
4. H. Mahl and J. Pohl, Z. techn. Phys., 16, 219, 1935.
5. H. Mahl, Mineral. u. Petrogr. Mitteilungen, 46, 289, 1935.
6. See, for example, reviews: V. Zworykin and G. Morton, Journ. Opt. Soc. Am. 26, 181, 1936; W. Schaffernicht, Z. techn. Phys., 17, 596, 1936.
7. W. Schaffernicht, Z. Physik, 93, 782, 1935.
8. W. Schaffernicht, Z. techn. Phys., 17, 596, 1936.
9. See, for example, N. Morgulis, Advances in the Physical Sciences, 16, 730, 1936.
10. E. Ruska, Z. Physik, 83, 492, 1933.
11. See, for example, W. Bothe, Durchgang von Elektronen durch Materie, Handb. d. Physik, XXII 2, 1933, p. 25.
12. B. Borries and E. Ruska, Z. Physik, 83, 187, 1933.
13. E. Ruska, Z. Physik, 87, 580, 1934.
14. H. Boersch, Ann. d. Phys., 26, 631, 1936; 27, 75, 1936.
15. L. Marton, Nature, 133, 911, 1934; Phys. Rev., 46, 527, 1934; see also F. Krause, Z. Physik, 102, 417, 1936.
16. V. Zworykin, Journ. Frankl. Inst., 215, 535, 1933.
17. M. Knoll, Z. techn. Phys., 16, 467, 1935; see also Advances in the Physical Sciences, 16, 277, 1936.
18. M. Knoll, Naturwiss., 24, 345, 1936.
19. M. Knoll and C. Lubszynski, Physik. Z., 34, 671, 1933.
20. R. Behne, Ann. d. Phys. 26, 372, 385, 1936.
21. E. Brüche, Z. techn. Phys., 17, 622, 1936.