MODERN THERMIONIC CATHODES

N. D. Morgulis

The study of thermionic emission and the further improvement of thermionic cathodes have acquired very great importance in recent years and are being carried out very widely. There are two reasons for this. First, the rapid development of electronic engineering, which we are witnessing at the present time, requires every possible improvement of old electronic devices and the creation of a large number of new ones. In solving these tasks, the thermionic cathode is often the narrowest and most difficult obstacle to overcome; the entire operation of a given electronic device—and hence of the whole apparatus as a whole—depends on its reliability and efficiency. Second, the physical processes that determine the emission activity of thermionic cathodes, and above all of semiconductor-oxide cathodes (see below), constitute a very complex and interesting set of problems, whose solution has required bringing into play an entire arsenal of diverse and new research methods. As a result of all this, substantial successes have now already been achieved, and a large number of studies have been published on this important problem of modern physical cathode electronics.

Several reviews¹ on the question of thermionic cathodes, and primarily oxide cathodes, have also been published; however, in connection with the vigorous development of research, these reviews have already become somewhat outdated and therefore cannot give a sufficiently clear picture of the present state of the question as a whole. This circumstance, as well as the importance of the problem, was the reason for writing the present review, which covers the literature up to the end of 1953.

Turning directly to the consideration of the material of the review, it is necessary to note that the assortment of thermionic cathodes known to us at the present time is very diverse in terms of ...

of their nature. Therefore, for the subsequent discussion of thermocathodes it will be useful to divide them into the following related groups: I—metallic cathodes, consisting of: 1) pure metals and 2) metals coated with monoatomic films, and II—nonmetallic cathodes, consisting of: 1) oxides of alkaline-earth metals, 2) various chemical compounds of metals, and 3) mixed metal–ceramic systems. Of course, such a classification is somewhat conventional; however, for our exposition it will be very useful. In conclusion we note also that, in comparing various cathodes for vacuum devices with one another, we shall characterize their quality by the density of thermionic emission at the operating temperature and by the rate at which the cathode material evaporates under these conditions; in the latter case one should also bear in mind the qualitative distinction between different materials.

I. METALLIC CATHODES

1. Pure metals

The use of pure metals as thermocathodes is naturally limited to the region in which the latter have to operate under such unfavorable and forced conditions that the normal functioning of various activated-type cathodes becomes practically impossible. In choosing the appropriate metals as thermocathodes, one should be guided by their emissive capacity, economy, and evaporation rate, which determine their service life; a summary of these data for the five most refractory, i.e., potentially usable, metals is given in Table I².

Table I

Metal	\(T_{\text{melt}}\ ^\circ\mathrm{K}\)	\(A\ \mathrm{a}/\mathrm{cm}^2\!\cdot\!\mathrm{degree}^2\)	\(\varphi\ \mathrm{eV}\)	\(T_0\ ^\circ\mathrm{K}\) for \(J = 3\ \mathrm{a}/\mathrm{cm}^2\)	Evaporation rate at \(T_0\ ^\circ\mathrm{K}\), \(\mathrm{mg}/\mathrm{cm}^2\!\cdot\!\mathrm{sec}\)
Re	3710	720	4.74	2590	\<0.043
W	3655	75	4.55	2780	0.043
Ta	3300	55	4.19	2585	0.043
Mo	2850	51	4.17	2580	14.0
Nb	2770	29	4.01	2560	0.42

It should be borne in mind that the values of the work function \(\varphi\) and the constant \(A\) given in this table are averaged values, obtained for polycrystalline specimens; the values of these quantities for different surfaces of a single crystal may differ noticeably, as is seen from Table II, where data for tungsten are presented.

Table II

Face index	111	112	116	001	110
\(\varphi\), eV	4.39	4.69	4.39	4.56	4.68
\(A\), A/cm\(^2\)·deg\(^2\)	35	125	53	117	15

In Fig. 1 a metallic tungsten cathode is schematically compared, in the range of operating temperatures and thermionic-emission densities, with various cathodes of other types, whose properties will be considered below.

Fig. 1.

As follows from Table I, the choice of pure metals as thermocathodes is at present limited only to W and Ta, the first being used in the straight-filament form, and the second in the heated form.\(^2\) The question of the possibility of using rhenium, as the most refractory metal, is still quite unclear, since on this point we still possess very limited data,\(^3\) obtained by a peculiar method using rhenium layers deposited on a tungsten core. In this case we are attracted by its relatively high emission capability and, at the same time, its very low evaporation rate, lower even than that of tungsten. The emission is calculated with the aid of the well-known Richardson formula:

\[ J = AT^2 \exp\left(-\frac{\varphi}{kT}\right), \tag{1} \]

where

\[ A = A_0 \overline{D} e^{-\frac{\alpha}{k}}. \]

Here \(\overline{D}\) is the mean transparency coefficient of the potential barrier, and \(\alpha\) is the temperature coefficient of the work function.

In Table I, attention is drawn to the anomalously large \((720\ \mathrm{a}/\mathrm{cm}^{2}\cdot \mathrm{deg}^{2})\) value of the coefficient \(A\) of the emission formula in the case of Re, as compared with the theoretical value \(A_0 = 120\ \mathrm{a}/\mathrm{cm}^{2}\cdot \mathrm{deg}^{2}\) for pure metals; similar anomalies were obtained previously also for Ni, Pt, and Zn. In connection with an attempt to explain such a phenomenon by overlap of the electron energy bands \(^{4}\), attention was drawn to the fact that in the most recent careful measurements this anomaly for Ni, Pt, and Zn proved already to have been eliminated by very careful cleaning of the surface of these metals, for which the following values of the emission constants may now be accepted \(^{5}\) (see Table III). As for the case of Re, since these are the first measurements in question, naturally, great accuracy still cannot be expected of them.

Table III

Metal	\(A\ \mathrm{a}/\mathrm{cm}^{2}\cdot \mathrm{deg}^{2}\)	\(\varphi\ \mathrm{ev}\)
Ni	30	4.61
Ni	120	4.50
Pt	32	5.32
Zn	120	3.78

In concluding this section, we note the latest measurements of the emission constants \(A\) and \(\varphi\) for a whole series of refractory metals (Ti, V, Cr, Mn, \(\gamma\)Fe, Co, Ni, Cu, Ag, Au, and graphite), carried out by means of the interesting method of applying the Clausius–Clapeyron formula to an electron gas formed inside a closed heated cavity with a small aperture \(^{6}\). Incidentally, in this case the value of the emission constant \(A\) never exceeded the above theoretical value. A summary of values of the work function for various elements is given in \(^{7}\).

2. Metals coated with monoatomic films

It is known that the first economical thermionic cathodes were metals coated with monoatomic films of an active substance. The first, best-known, and rather widely used cathode of this type was thoriated tungsten Th-W, which at the same time became the object of what are now classical studies of metals with monoatomic coatings \(^{8}\). However, thoriated tungsten lost its practical significance as a cathode for electronic devices rather quickly, owing to the extreme instability of the active monoatomic thorium film under conditions of accidental overheating or ion bombardment. The systems Ba-W and Cs-W \(^{8}\), which were also investigated in the past, although in principle even more effective with respect to their electronic

emission; than Th-W, but, owing to the impossibility of creating practically acceptable designs satisfying the severe operating conditions in electronic devices, they have never been used in practice.

Does it follow from all that has been said that all systems of this type are completely hopeless for practical use? In order to answer this question, let us examine it in somewhat greater detail.

The possibility of obtaining stable effective emission for a long time from such monolayer cathodes is connected with the favorable character of the balance of active atoms on the cathode surface,^9 namely:

\[ \frac{dN}{dt}=A-B, \tag{2} \]

where \(N\) is the surface concentration of active atoms, \(A\) is the rate at which they arrive at the surface, and \(B\) is the rate at which they are removed from the surface by evaporation, poisoning, ion bombardment (cathode sputtering), etc. Under practical operating conditions of cathodes the quantity \(B\) may reach substantial values. For example, under conditions of electron emission with density \(J_e=1\ \text{a}/\text{cm}^2\) (i.e., at \(v_a \simeq 1\ \text{kv}\)) and with the pressure of the inert part of the residual gases in the tube \(\simeq 10^{-6}\ \text{mm Hg}\), the density of the ion current bombarding the cathode reaches \(10^{-5}\ \text{a}/\text{cm}^2\); i.e., the surface monoatomic film will be sputtered in approximately a few seconds, not to mention the action of other factors. Therefore, under such conditions a cathode can operate only when, in the course of operation, its continuous and sufficiently intense (so that \(dN/dt \gg 0\)) reactivation takes place at the expense of certain reserves of active material. This method of creating practically effective monolayer-type cathodes is at present indeed being successfully realized, and we shall show it by examples of the following three cathodes, whose properties are brought together in parallel in Table IV.

Table IV

System	\(A\ \text{a}/\text{cm}^2\cdot\text{deg}^2\)	\(\varphi\), eV	\(T_{\text{oper}}\ ^\circ\mathrm{K}\)	\(J_{T_p}\), \(\text{a}/\text{cm}^2\)	\(N(T_p)\), \(\mu\text{kg}/\text{cm}^2\cdot\text{sec}\)
Th-W	3.0	2.6	2000	3.0	—
Ba-W	2.0	1.6	1320	3.0	\(4\cdot10^{-4}\)
Ba-W	2.0	1.6	1540	50	\(4\cdot10^{-2}\)
Cs-W	3.2	1.4	1400	79	—

1. As a first example let us consider the old case of a cathode made of thoriated tungsten, described in an extensive literature8 and still used at the present time in high-power generator tubes10. The idea of this application is connected with the additional use of the long-known technique of preliminary carburization of the cathode, i.e. the corresponding treatment of the cathode at high temperature in an atmosphere of hydrocarbon vapors. As a result, the surface of the tungsten is covered with a crust of tungsten carbide, the thickness of which depends on the conditions of its treatment. This crust of tungsten carbide has the following features11: a) the reduction of thorium oxide, as the initial impurity \((\sim 2\%)\) to tungsten, proceeds here continuously, practically at the operating temperature, almost without requiring a special activation operation; b) owing to the loose structure of this crust, diffusion through it of active thorium atoms to the surface proceeds comparatively easily and uniformly, etc.

However, along with this the tungsten-carbide crust has substantial disadvantages, among which are, first of all, its extreme brittleness and then the gradual reduction of tungsten carbide during operation of the cathode, which limits its lifetime. From all that has been said it follows, therefore, that in the optimum case one should create a tungsten-carbide crust of such thickness that, first, the time of its gradual reduction would appreciably exceed in advance the prescribed service life of the cathode and, secondly, that its thickness would be much smaller than the cathode radius, i.e. so that the purely tungsten “core” could retain sufficient mechanical strength for the cathode. All this is possible only when sufficiently thick tungsten cathodes are used, suitable, naturally, for use only in powerful electronic devices (as we in fact now observe), where their economy many times exceeds the economy of purely tungsten cathodes.

1. As a second example let us consider the case of a cathode consisting, for example, of tungsten covered with a monoatomic film of barium. The electronic properties of such a system were investigated very long ago8; however, attempts at practical use of such cathodes, first made by our industry as early as 1932 (a cathode for an arrester of the RB type), nevertheless had a limited character12, and then only in application to gas-discharge, not vacuum, devices. Only quite recently cathodes of this type, but in an entirely new structural-technological form, known under the name of metal-tube or porous \(L\)-cathodes, have begun to receive rather wide distribution and application in various electronic devices13. Therefore we shall dwell on them in greater detail. In their modern-

...In its final form these cathodes are a tungsten sponge, inside which there is a cavity filled with an active substance used in oxide-cathode technology—a mixture of \(\mathrm{BaCO_3}+\mathrm{SrCO_3}\) or simply \(\mathrm{BaCO_3}\) (Fig. 2). By carrying out an activation operation, somewhat similar to the activation of an oxide cathode (see below), these compounds are reduced, as one may suppose, to metallic barium, which then easily passes through the pores in the sponge to the cathode surface. Thus, the term \(A\) in formula (2) is here very large, and this means the possibility of forcing the cathode regime, i.e., of sufficiently prolonged cathode operation under heating to a high temperature, when it is capable of giving a high thermionic-emission density even in the presence of intense evaporation and ion bombardment. At the same time, by introducing in advance into the cathode the appropriate amount of active filler, it is possible, in principle, to regulate and ensure an acceptably long service life. Thus, a cathode of this type acquires many attractive features as a practical source of thermoelectrons with a high current density for various electronic devices, especially since, judging from the literature, it is comparatively little affected by poisoning, ion bombardment, and high electric fields.

Fig. 2.

Fig. 3.

An idea of the electron-current density \(J_e\) obtainable with a cathode of this type can be formed from consideration of the current-voltage characteristics (Fig. 3) for a cathode with a surface area of \(7\ \mathrm{mm^2}\). Here, for example, at \(1265^\circ\mathrm{C}\) it is possible easily to reach a density \(J_e \simeq 50\ \mathrm{a/cm^2}\). In this figure, attention is drawn to...

a peculiar course of the curves, characterized by a “fan,” i.e., an early departure from the three-halves law. This may be connected with various circumstances, such as, for example: a) nonuniformity of the surface emission activity, i.e., the density of the surface active film; b) the anomalous Schottky effect, sharply expressed owing to structural nonuniformity at the surface, etc. The actual presence of the influence of at least the first

Fig. 4.

of these factors was demonstrated both by observation of the emission pattern of such a cathode with the aid of an electron projector and by the discovery of its additional activation when the cathode temperature was lowered from the working value (\(1255^\circ\)C) to a reduced one (\(990^\circ\)C).

Of great importance for characterizing metal-sponge cathodes is the process of evaporation of their active filler, for this process, as well as the initial reserve of the active filler, primarily determines the practical service life of these cathodes. Thus, at \(T = 1270^\circ\)C, when \(J \approx 50 \ \dfrac{a}{\text{cm}^2}\), with an initial reserve of BaO equal to \(2\) mg, the service life of a cathode with a surf-

with an emission density of \(7\ \mathrm{A/cm^2}\) is approximately 200 hours. To investigate the evaporation process, the method proposed long ago by Becker (as applied to oxide cathodes) was used here: deposition of the evaporating active substance onto a test tungsten filament with monitoring of the time course of its thermionic emission. In this way the results shown in Fig. 4 were obtained; they give, incidentally, an unusual value for the heat of evaporation of the active substance from the \(L\)-cathode, equal to \(2.61\ \mathrm{eV}\). Preliminary consideration leads to the conclusion that what is involved here is mainly the evaporation of barium atoms, in view of the fact that: (a) measurements of the work function of this cathode, although still very approximate, give values closer to the value \(1.6\ \mathrm{eV}\), characteristic of the Ba–W system, than to the value \(1.0\ \mathrm{eV}\), characteristic of the BaO–W system; and (b) in the process of diffusion of barium oxide through the heated tungsten tube there must occur the chemical reaction of reduction of this oxide by tungsten\(^{14}\), which is currently used to accelerate and facilitate activation of oxide cathodes (see below).

In conclusion, we point to the fabrication of cathodes of a similar type both with a molybdenum sponge and with various other active fillers; the characteristics of some systems are given in Table V.

Table V

System (sponge–filler)	Operating temperature, °C	Emission density, \(\mathrm{A/cm^2}\)	\(\varphi\), eV	\(A\), \(\mathrm{A/cm^2\cdot deg^2}\)
W–Th	1500	3	2.8	100
W–Ba(Sr)CO\(_3\)	1000	3	1.6–2.0	1–1.5
W–BaCO\(_3\)+Si	900	4	—	—
W–BaBe	900	5	1.5	2.3
Mo–BaBe	850	5	—	—

Finally, systems have also been developed at the present time consisting of a pressed and annealed mixture of metal powder (W and Ni) and an active substance (BaO, ThO\(_2\), etc.); however, we shall return to their consideration somewhat later, after examining cathodes of the oxide type.

1. As a third example, let us consider the case of a cathode consisting of tungsten coated with a cesium film. The electronic and adsorption properties of such a system were investigated in great detail long ago\(^{8}\), since for the purposes of such investigations it offers incomparable conveniences. Despite the fact that a system of this type, i.e. a metal with a monolayer of cesium, has the greatest emission activity, the question of its practical use has never even arisen, since

as a cesium film on the cathode is very little stable. However, recently such systems have revealed an entirely different and new aspect of their application[^15]. Stabilization of a cesium monofilm at a somewhat elevated temperature, necessary for obtaining a sufficiently high density of thermionic emission, is possible, naturally, by increasing the pressure of cesium vapor in the lamp; however, in doing so the vapor pressure may reach such a magnitude that switching the lamp into a circuit will produce an arc discharge in it. This discharge should be distinguished by the feature that the potential drop in it will be very small, owing to the very easy ionizability of cesium atoms. Thus, here we arrive at cesium arc lamps, where the role of the cesium-vapor atmosphere at somewhat elevated pressure is reduced to two functions: the stable maintenance of an adsorbed film on the metallic cathode at its elevated temperature, i.e. the production of highly effective thermionic emission, and then the creation of a gas-discharge atmosphere with the minimally possible potential drop. As an illustration of all that has been said, one may cite from the literature[^15] an example of obtaining, in continuous operation, thermionic emission with a density of about \(70\,a/cm^2\) at a voltage drop in the arc of \(3.4\,v\) and a lamp bulb temperature of the order of \(250^\circ C\); such lamps can be used as highly economical rectifiers of alternating current.

Turning to the question of the nature of phenomena connected with the thermionic emission of film metallic cathodes, it is interesting to draw attention to two groups of important works carried out in this field in recent times.

The first group of these works concerns clarification of the rather old question of the influence of adsorbed atoms and molecules on the work function of metals. The matter was that until recently it had been customary to consider that, in the gradual build-up of an electropositive adsorbed film, the minimum electron work function \(\varphi_{\min}\) obtained and the corresponding maximum shift of the contact potential \(v_{k,\max}\) are connected with the fact that this film has reached an optimal, almost monoatomic coverage (usually the optimal degree of coverage \(\vartheta_{\mathrm{opt}} \approx 0.67\)). However, recently performed careful experiments[^16] have shown that, in the case of deposition of molecules of the oxides BaO and SrO on a metallic (W, Mo, Ta, Zr) core, the situation is entirely different. This is seen, for example, from Fig. 5, where the work function \(\varphi\) and the Richardson constant \(A\) are presented for the surface of tungsten as a function of the number of monomolecular BaO layers adsorbed on it. We see here a paradoxically large initial rate of decrease of the quantity \(\varphi\), reaching \(40\,ev/\)monolayer. This means that the main decrease of the quantity \(\varphi\) from \(4.5\,ev\) to \(1.5\,ev\) and of the quantity \(A\) from \(100\,a/cm^2\cdot deg^2\) to \(0.04\,a/cm^2\cdot deg^2\) occurs-

leads, upon adsorption of BaO, to the value \(\theta_{\mathrm{opt}}\lesssim 0.1\); further deposition of BaO up to \(\theta \simeq 25\) is already of no substantial significance. Similar experiments for the SrO–Mo system led to \(\theta_{\mathrm{opt}}\simeq 0.2\), etc. The authors relate their results to the presence in the BaO and SrO molecules of a large permanent dipole moment \(p=ed\), producing a strong change in the work function \(\Delta\varphi=4\pi ed\). Let us also note here

Fig. 5.

Fig. 6.

an attempt to refine these data by applying a mass selector for the dosed deposition of strictly definite particles \({}^{17}\).

Closely related to this work is another work \({}^{18}\), in which the question is posed of studying the influence on the work function of strictly dosed amounts of Ba atoms and polar BaO molecules, deposited in the form of a wedge on a tungsten strip surface about \(\simeq 200\) mm long under the best possible vacuum conditions. The measurements were made by the contact-potential method, i.e., without heating the surface under investigation and without the disturbances caused by this; calibration by the number of adsorbed particles was carried out with the aid of a very precise polarographic method. The results of the first part of these investigations are presented in Fig. 6, from which it is seen that, upon adsorption of Ba atoms, the work function \(\varphi\) decreases to \(\varphi_{\min}\simeq 1.6\) eV, corresponding to a surface concentration \(n(\mathrm{Ba})\) equal to \(4\cdot 10^{14}\ \mathrm{cm}^{-2}\) (without taking account of reflection during condensation). The initial dipole moment of Ba atoms adsorbed on tungsten is equal to \(7.6\cdot 10^{-18}\) CGSE, i.e., corresponds to an initial rate of about \(11\) eV/monolayer. The work also determined the change in the average surface-charge density with change in the density of adsorbed barium atoms. Similar investigations are now being extended to a number of other interesting systems.

Finally, this same question was also considered in a study of films evaporated from a thorium oxide source and adsorbed

on the surface of molybdenum.^19 Films were likewise deposited in the form of a wedge of variable thickness, and their properties were investigated by determining, from thermionic emission, the work function \(\varphi\) and the Richardson constant \(A\); the number of adsorbed atoms was determined in the thick part of the wedge by the contact-photography method, from the number of tracks of \(\alpha\)-particles produced by the natural radioactivity of thorium. The results of determining the dependence of \(\varphi\) on the degree of coverage \(\theta\) are given in Fig. 7. They are interpreted by the authors in the following way. In the thin part

Fig. 7.

of this wedge, thorium oxide is reduced by heating to metallic thorium, and the values of \(\varphi\) and \(A\) obtained here for \(\theta = 0.7\) practically coincide with the known values for thoriated molybdenum at \(\theta = \theta_{\mathrm{opt}} = 0.7\), while the values for \(\theta = 2\) coincide with those known for metallic thorium. In the thick part of this wedge, at \(\theta > 4\), the reduction of thorium oxide is only partial, and here we are dealing with an activated oxide-thorium cathode (see below), or else the values of \(\varphi\) and \(A\) obtained here coincide with those known for this cathode. Thus, all three of these works are, naturally, of substantial significance for the problem of film metallic cathodes.

The second group of works concerns questions of adsorption and evaporation of alkali atoms from a metal surface—questions that, although old, are now being studied by far more refined experimental means. In the first of these works^20 the adsorption and subsequent evaporation of potassium and sodium from a molecular beam when they struck tungsten were investigated. The molecular beam was switched on for a time \(\Delta t_1\) and switched off for \(\Delta t_2\) by means of a shutter vibrating with a frequency of \(100\) cycles. With the aid of a cathode oscillograph, direct observation was made of the decay, in time \(t\), during the period \(\Delta t_2\), of the ion current of these metals \(J_p\), obtained on tungsten as a result of the well-known phenomenon of surface ionization. In a certain temperature interval the dependence \(J_p = f(t)\) had a purely exponential character, from which it was possible to determine the mean residence time of an atom in the adsorbed state-

... of adsorption ($\tau$). By determining this quantity at different temperatures, one can, with the aid of a simple exponential dependence for potassium and a more complicated one for sodium, determine the heat of adsorption of their ions, $Q_p$; here the following was obtained: $Q_p(\mathrm{K}) = 2.55\ \text{eV}$ and $Q_p(\mathrm{Na}) = 3.3\ \text{eV}$. In another work,^21 where practically the same method and the same substrate metal (W) were used, the following was obtained: $Q_p(\mathrm{K}) = 2.9\ \text{eV}$ and $Q_p(\mathrm{Cs}) = 3.6\ \text{eV}$. The last result is somewhat strange, since for Cs one would have expected a smaller value of $Q_p$ than for K, at least from the well-known qualitative relation

\[ Q_p = \frac{e^2}{4R}, \]

where $R$ is the radius of the corresponding ion.

In conclusion, let us note the great interest also presented by the problem of the systematic investigation and creation of practically useful metallic cathodes coated with electronegative films that increase the work function $\varphi$. Unfortunately, in this direction one can note only a single recent work,^22 devoted to tungsten coated with a platinum film; in this case it is possible to increase the value of $\varphi$ from $4.53\ \text{eV}$ to $\approx 5.45\ \text{eV}$. The application of this system to the study of the surface ionization of sodium atoms unfortunately led to rather complicated results that are still poorly understood. This same problem also includes the very important question of creating anti-emission coatings that are also stable with respect to changes in contact potential difference, i.e., such substrates which, when an active mixture of Ba + BaO is deposited on them, for example from an oxide cathode, would not produce a sharp decrease in $\varphi$ and a corresponding increase in emission activity; this is very important for the grids of radio tubes.^23 Experience shows, in particular, the suitability for this purpose of thin gold coatings,^24 the nature of whose action is still unclear.

II. NONMETALLIC CATHODES

Already long ago, in the course of research and development of effective thermocathodes, the attention of investigators was drawn to cathodes of the nonmetallic type. Among these cathodes belongs first of all the oxide cathode, discovered as early as 50 years ago and still retaining the greatest importance. In addition, the continuously increasing requirements now imposed on thermocathodes have also led to intensive studies of a whole series of other diverse compounds, among which some have proved to be very promising. In view of the fact that all these cathodes must naturally be of the self-activating type, the discussion must concern chemical compounds of electropositive atoms of groups I, II, III

and IV groups of the periodic system of Mendeleev. Here it is necessary to choose a stable compound that combines within itself, to a reasonable degree, acceptable values of, for example, such factors of opposite character as the emission capability and the rate of evaporation of the cathode material. Cathodes of this type are a complex, usually semiconducting system with an entire gamut of interesting physical phenomena, which will be vividly illustrated below by the example of one of the most important representatives of cathodes of this type—the oxide cathode.

1. Oxides of alkaline-earth metals

Discovered as early as 1904, the thermionic oxide-barium or, more simply, oxide cathode (o. c.) continues to remain up to the present time the principal and most important among the other cathodes of various modern electronic devices, especially of the mass type. On the other hand, a cathode of this type is physically a very complex and interesting system, for

Labels in the figure: Vacuum; o. c. surface; surface layer; volume of the o. c. layer; near-contact layer; core.

Fig. 8.

the investigation and improvement of which, in connection with ever new requirements imposed on it, a numerous arsenal of new methods has been used in recent years (see below). The fruitfulness of these investigations of recent years is connected, above all, with the fact that only now have all investigators begun to rely on the conception of this cathode as a semiconductor film on a metallic substrate\(^2\) and have begun widely to develop semiconductor theories and methods of investigation of these cathodes. The results of these investigations have already been repeatedly covered in the past in various reviews and monographs\(^1\).

The structure of the oxide cathode may be represented schematically in the following form (Fig. 8). On a metallic (usually nickel) substrate there lies the main oxide layer of thickness \(\simeq 50\,\mu\), usually consisting of a mixture of barium and strontium oxides with the corresponding admixture of atomic Ba and Sr. In the contact between the substrate and the oxide there may be a thin intermediate near-contact layer of increased resistance; at the outer surface of the oxide layer there may be

MODERN THERMOELECTRONIC CATHODES

there is also a thin near-surface layer of increased resistance; finally, on the outer surface of the cathode there is an active surface film. In its operating regime this cathode is heated to a temperature of \(\approx 1100^\circ\text{K}\) and is placed under conditions of high vacuum; all this substantially complicates its investigation. Nevertheless, in recent years great and comprehensive progress has been achieved in its study, stimulated also by its great practical importance, in connection with which it is not without interest to give a list of those comprehensive methods that have been used in this work:

1. A pulsed method for measuring the emission of an oxide cathode over very short time intervals. Under these conditions, as is known, its enhanced emission capability was first discovered \(^{26}\).

2. A probe method for measuring the distribution of potential inside the layer of an oxide cathode in the process of emission, with separate determination of its bulk resistance and its contact resistance at the boundary with the substrate \(^{27}\), as well as radio-engineering methods for measuring the latter \(^{28}\).

3. The use of an electrostatic analyzer for determining the total resistance of the entire layer of an oxide cathode and isolating its near-surface part \(^{29}\).

4. Measurement of thermoelectromotive forces \(^{30}\) and the Hall effect \(^{31}\).

5. Optical methods of investigation, namely: spectral absorption and photoconductivity \(^{32}\), the external photoelectric effect \(^{33}\), and luminescence \(^{34}\).

6. X-ray and electron-diffraction studies of the structure both of the layer of the oxide cathode itself and of the contact layer formed at the boundary with the substrate \(^{35}\).

7. A method for growing both pure and impurity-containing single crystals of barium oxide and their use for investigating the primary properties of the cathode material \(^{36}\).

8. The use of an emission electron microscope to investigate the stationary emission of the cathode and of a spherical projector to study its pulsed emission \(^{37}\).

9. Spectrochemical \(^{38}\) and mass-spectrometric \(^{39}\) methods for investigating emissions from the cathode, as well as the use of a mass selector for the dosed deposition of Ba on an oxide cathode.

10. Thermochemical calculations and experimental investigations of possible thermochemical reactions in an oxide cathode \(^{40}\).

11. The use of a technique with labeled radioactive atoms \(^{41}\) for investigating diffusion inside the cathode, evaporation rates from it, etc.

12. Investigations of the secondary electron emission of oxide thermocathodes \(^{42}\), which, however, are of independent interest, etc.

Although all these studies are at present in full swing, many important and interesting results have already been obtained, to a brief consideration of which we shall now turn.

In modern, albeit elementary, notions of the nature of thermionic emission from an oxide cathode, this cathode is regarded as a completely structurally homogeneous layer (of thickness \(d\)) of an electronic semiconductor—barium oxide with an admixture of donor atoms of metallic barium; emission from it has the character of pure evaporation. In this case the following criteria hold for the dimensionless current \(\lambda\) and the screening depth \(x_0\):

Fig. 9.

\[ \left. \begin{aligned} \lambda &= \frac{eJx_0}{\sigma kT} \ll 1,\\ x_0 &= \left(\frac{\varepsilon kT}{8\pi e^2 n_e}\right)^{1/2} \ll d, \end{aligned} \right\} \tag{3} \]

where \(J\) is the emission-current density, \(\sigma\) is the conductivity of the cathode layer, \(\varepsilon\) is the dielectric constant, and \(n_e\) is the concentration of free electrons in the cathode, in accordance with the general expression

\[ n_e=\frac{2}{h^3}(2\pi m_e^* kT)^{3/2}\exp\left(-\frac{\zeta}{kT}\right) = B_e \exp\left(-\frac{\zeta}{kT}\right), \tag{4} \]

\(\zeta\) is the internal work function (Fig. 9).

In this case the emission-current density obeys the general formula

\[ J=D e n_e \sqrt{\frac{kT}{2\pi m_e^*}}\exp\left(-\frac{\chi}{kT}\right) = D A_0 T^2 \exp\left(-\frac{\zeta+\chi}{kT}\right), \tag{5} \]

where \(\chi\) is the external work function, \(A_0=\dfrac{4\pi m_e^* e k^2}{h^3}=120\ \dfrac{\mathrm{a}}{\mathrm{cm}^2\cdot \mathrm{deg}^2}\) is the well-known universal constant of the emission formula, and \(\varphi=(\zeta+\chi)\) is the total thermoelectronic work function. The latter (Fig. 9) is characterized by the energy gap between the level of the chemical potential \(\mu\) and the external space (vacuum); by the difference of precisely these quantities \(\varphi\) the contact potential difference is also determined. Anticipating somewhat, one may say that typical for an activated oxide cathode of continuous operation under working conditions (\(T \simeq 1100^\circ\mathrm{K}\)) are

will have the following values:

\[ J \simeq 0.1-1.0\ a/\mathrm{cm}^{2}, \quad n_e \simeq 10^{14}-10^{15}\ \mathrm{cm}^{-3}, \]

i.e. (for the effective electron mass \(m_e^* = m\)) on the basis of (4)

\[ \zeta = 1.0-1.2\ \mathrm{eV}. \]

The simplest case of emission from a semiconductor cathode considered here also corresponds to the new theory of the Schottky effect\({}^{43}\), which sees the action of the external electric field \(E\) not only in a lowering of the external work function \(\chi\), but also of the internal one \(\zeta\), as a result of the partial penetration of this field into the semiconductor. Hence

\[ \Delta\varphi=\Delta\chi+\Delta\zeta=(aeE)^{1/2}+\frac{2kT}{e}\operatorname{arcsh}\left[\frac{1}{2\sqrt{2}}\frac{e x_0}{\varepsilon kT}\left(E \pm 4\pi\sigma\right)\right], \tag{6} \]

where \(\sigma\) is the density of surface charges. In the case \(E=0\) we obtain a change in the magnitude \(\zeta\) due to the presence, on the surface levels of the semiconductor, of charges with density \(\sigma\), i.e.

\[ \Delta\zeta=\frac{2kT}{e}\operatorname{arcsh}\left[\frac{1}{2\sqrt{2}}\frac{e}{kT}\frac{4\pi\sigma x_0}{\varepsilon}\right]\simeq \frac{4\pi\sigma x_0}{\varepsilon}. \tag{7} \]

We see that here there is formed a double layer of macrodipoles with an arm \(\dfrac{x_0}{\varepsilon}\) that is very large in comparison with atomic microdipoles; for an oxide cathode under operating conditions it is equal to \(\simeq 10^{-6}\ \mathrm{cm}\). This theory now requires its further development for the cases \(\sigma=f(E)\) and \(\sigma=f(T)\), which may occur under real conditions.

The experiments on which we shall dwell below show, however, that in an oxide cathode, besides donor-type impurities, there may also be acceptor-type impurities; the latter manifest themselves mainly when the cathode is in the inactive state. In accordance with this (see Fig. 9), let us denote by \(N\) and \(q\) the concentration and excitation energy of the atoms of the donor impurity, by \(N'\) and \(q'\) the same quantities for the acceptor impurity, and by \(Q\) the width of the forbidden band in the substance of the principal semiconductor. Using now the neutrality equation, one can determine from it the value of \(\zeta\), namely\({}^{44}\):

\[ n_e=B_e\exp\left(-\frac{\zeta}{kT}\right)=B_p\exp\left(-\frac{Q-\zeta}{kT}\right)+ \]

\[ +\frac{N}{\left[\exp\left(\frac{q-\zeta}{kT}\right)+1\right]}- \frac{N'}{\left[\exp\left(\frac{q'+\zeta-Q}{kT}\right)+1\right]}. \tag{8} \]

Therefore the quantity \(\zeta\) is a complex function:

\[ \zeta=f(N,\ N',\ q,\ q',\ Q;\ T); \]

We pay special attention to the dependence of the internal work function \(\zeta\) on temperature. From this point of view, a careless use of the general formula (5) for semiconductor—for example oxide—cathodes, in order to determine the total and, consequently, the internal work function from the temperature dependence of thermionic emission, may sometimes lead to serious errors. This fact is, unfortunately, little known, although in the literature \(^{44}\) it is illustrated by a very good and convincing example. To this are added the complications, well known in the field of active film metallic cathodes, \(^{45}\) which are caused by the surface inhomogeneity present here of the external work function of the electron, \(\chi = f(x, y)\). In addition, the local values of the latter quantity are probably also temperature-dependent, owing to the dynamic conditions in the active film on the surface. In conclusion we note that the emission formula sometimes applied to an oxide cathode \(^{46}\)

\[ J=\overline{D}\,B\sqrt{N}\,T^{5/4}\exp\left(-\frac{\frac{1}{2}q+\chi}{kT}\right) \tag{9} \]

is a special case of the general formula (5), realized only when \(N'=0\), \(kT \ll (Q-\zeta)\), and \(kT \ll (q-\zeta)\), i.e., when thermionic emission takes place only at the expense of a donor impurity of one definite type, characterized by one definite value of the quantity \(q\), of which, for the real case, we have no certainty. Thus, in conclusion, we may say that, from the standpoint of these ideas (see (5)), obtaining a high emission activity from an oxide-type cathode reduces to obtaining small values of the internal \(\zeta\) and external \(\chi\) work functions of the electron. The latter can be accomplished by introducing into the interior and onto the surface of the semiconducting oxide layer a suitable quantity of impurity atoms of an electropositive metal; this operation is called activation of the cathode.

Unfortunately, the conception of the emission of an oxide cathode set forth here is very primitive, primarily because the structure of this cathode is far more complex even in the case where it is a two-component system \(\mathrm{BaO}+\mathrm{SrO}\). In fact, it is well known that the substances \(\mathrm{BaO}\) and \(\mathrm{Ba}\) are much more volatile than \(\mathrm{SrO}\) and \(\mathrm{Sr}\); as a result, the concentration ratio \(\mathrm{BaO}/\mathrm{SrO}\) is different at different depths of the cathode, changing, moreover, with time. A typical example of the distribution of the relative concentration of \(\mathrm{SrO}\) at different depths at different times is presented in Fig. 10 \(^{47}\).

It follows from this that, for example, after 400 hours of operation at \(875^\circ\mathrm{C}\), the near-surface layer of the cathode consists practically entirely of ...

only from SrO. At the same time, it may be thought that the emission capacity of the oxide cathode is determined mainly by its barium component, which, for example, can be seen from Table VI[^45], where the comparative emission at 750° K of various components of an oxide cathode is presented separately, i.e., of oxides of different alkaline-earth metals with ordinary admixtures of different metals of the same group.

Table VI

Emitter . . .	BaO + Ba	SrO + Sr	CaO + Ca	SrO + Ba	CaO + Ba
$J_e$ (amperes) . .	$1.2 \cdot 10^{-3}$	$1.0 \cdot 10^{-6}$	$1.0 \cdot 10^{-8}$	$2.0 \cdot 10^{-3}$	$1.8 \cdot 10^{-4}$

Finally, the oxide cathode does not at all have a homogeneous structure; on the contrary, it is a highly porous system consisting of very small, on the order of microns, partially sintered crystallites, which must affect its properties[^48].

Another very important structural feature of the oxide cathode was discovered as a result of the application of the probe method[^27], developed and used by us as early as 1941, which in the postwar years became very widely used. The essence of this method, borrowed from investigations of semiconductors and dielectrics, consists in introducing into the layer of a real oxide cathode two or three probes, which are thin ($\simeq 5\ \mu$) platinum or nickel ribbons located at different distances $x$ from the cathode core. By passing an emission current of different magnitude through the cathode, one can determine, by the compensation method, the potential of each of the probes, i.e., the distribution of potential in the layer

Fig. 10.                Fig. 11.

\(v=f(x)\). As a typical example, Fig. 11 gives a graph borrowed from our cited work of 1941.\(^{27}\) By extrapolating to \(x=0\), we obtain the values \(v=v_k\) and \(R_k=\dfrac{v_k}{J}\), which revealed the presence, at the boundary between the cathode layer and the core, of a particularly thin contact blocking layer with a potential jump \(v_k\) and resistance \(R_k\). On the other hand, the quantity \(R=\dfrac{v_2-v_1}{J}\) characterizes the resistance of the bulk part of the cathode layer of thickness \((x_2-x_1)=\Delta x\). The values \(v_{10}\) and \(v_{20}\) at \(J=0\) characterize the values of the thermoelectromotive force in the layer (see below). Thus, in this work the existence of a contact blocking layer in the oxide cathode was first discovered, and separate measurements were made of the resistances, at different temperatures, both of this layer and of the cathode bulk.

Fig. 12.

In subsequent studies a method was developed for measuring the total voltage drop across the cathode layer,\(^{29}\) which made it possible, in combination with the preceding one, to continue the curve of the potential distribution \(v=f(x)\) to the external surface of the cathode. One of the examples we obtained is shown in Fig. 12, where, unexpectedly

the appearance of still another barrier layer is observed—a near-surface one—with a potential jump \(v_n\). This layer must exert the most active influence on the electron emission of the cathode as proceeding directly from the region lying beneath the surface (at a depth of several free paths).

As for the nature of the barrier layers, the following may be said on this point: the contact barrier layer is, as a rule, chemical in nature; it is formed as a result of the chemical reduction reaction of barium oxide by various activating impurities to the core \(^{40}\) (see below). For example, in the typical case of a core made of silicon nickel, this layer consists of \(\mathrm{Ba_2SiO_4}\), the thickness of which may reach many microns \(^{53}\). As for the structure of the near-surface barrier layer, at present we can only make more or less plausible conjectures. The substance \(\mathrm{Ba_2SiO_4}\) has, under cathode operating conditions, a specific resistivity \(\rho_{z.c}\sim 10^{-6}\ \Omega\cdot\mathrm{cm}\) and an internal work function \(\zeta\sim 3\ \mathrm{eV}\), whereas for the bulk of the cathode a typical value is \(\rho\sim 10^{-2}\ \Omega\cdot\mathrm{cm}\). The total resistance of the entire cathode layer in good condition under operating conditions (\(\sim 800^\circ\mathrm{C}\)) is approximately several \(\Omega/\mathrm{cm}^2\). Finally, we note that the resistances \(\rho_{z.c}\) and \(\rho\) are ohmic at small emission-current strengths and sharply non-ohmic at large ones, obtained in the pulsed regime (see below).

Although the real oxide cathode (see Fig. 9) differs substantially from the ideal scheme set forth above, nevertheless one of its principal features—the activation process—is characterized here quite correctly. As a result of this process, in the cathode layer \(\mathrm{BaO}\) an impurity of atomic \(\mathrm{Ba}\) is liberated; its concentration, determined, for example, by Berdennikova’s well-known method, reaches \(^{49}\) the order of \(N \simeq 10^{17}\text{--}10^{18}\ \mathrm{cm}^{-3}\), and, in parallel with this, the internal work function \(\zeta\) is substantially decreased. Although during cathode operation there is continuous evaporation of its components, and primarily of the volatile, barium ones, nevertheless the value \(N\) must be maintained constant; otherwise the cathode will gradually lose its emission activity. The creation and subsequent maintenance of the concentration of the atomic barium impurity in the cathode at the required level are possible through processes of electrolysis and chemical reduction of the main semiconductor—\(\mathrm{BaO}\).

Perhaps one of the most significant achievements in the physics and technology of the oxide cathode in recent years has been the elucidation of the features of the process of chemical activation of the cathode and of the influence on this process of the material of the metallic substrate. The history of this question begins in 1935, when decisive experiments were performed and their correct interpretation was given \(^{50}\).

These experiments demonstrated the favorable role of Al and Ti impurities, as well as the unfavorable role of Mn. Then, already in 1946, in the development of oxide cathodes suitable for use in a pulsed mode with extraction of large emission-current densities, the favorable role of an electrolytically pure nickel core was shown[^51]. In connection with the very great practical importance of accelerating and facilitating activation of the cathode by introducing into its core small \((<1\%)\) amounts of impurities of certain metals (Al, Mg, etc.), a large number of works has recently been devoted to this question[^40],[^52]. As a result it became clear that the matter here consists in the thermochemical reduction of BaO by reaction with these impurities, i.e. in the liberation of metallic barium, which is then distributed throughout the entire volume and over the surface of the cathode, thereby lowering both the internal and the external work function according to (5). The degree of such activation can be estimated by calculating the equilibrium constant \(k\) of this reaction from the equation

\[ \Delta \Phi = -RT \ln k \]

(where \(\Delta \Phi\) is the change in free energy in the standard state) and then determining from this the vapor pressure \(p\) of the metallic barium obtained here. All that has been said may be illustrated by Table VII[^40], which gives the values of \(p\) at the activation temperature \(1000^\circ\mathrm{C}\).

Table VII

Type of reaction	\(p\), mm
\(2\mathrm{BaO} \rightleftarrows 2\mathrm{Ba}+\mathrm{O}_2\)	\(2\cdot 10^{-11}\)
\(\mathrm{BaO}+\mathrm{CH}_4\) (at \(10^{-5}\) atm) \(\rightleftarrows \mathrm{Ba}+\mathrm{CO}+2\mathrm{H}_2\)	\(0.1\)
\(\mathrm{BaO}+\mathrm{Ni}\rightleftarrows \mathrm{Ba}+\mathrm{NiO}\)	\(5\cdot 10^{-13}\)
\(3\mathrm{BaO}+\mathrm{Si}\rightleftarrows 2\mathrm{Ba}+\mathrm{BaSiO}_3\)	\(1\cdot 10^{-3}\)
\(\mathrm{BaO}+\mathrm{Mg}\) (liquid) \(\rightleftarrows \mathrm{Ba}+\mathrm{MgO}\)	\(40\)

From this point of view it becomes entirely clear, for example, the operating principle of our domestic cathodes with a core made of bronze-coated tungsten[^53]. In fact, the use of Cu-Al bronze amounts: a) to facilitated activation of the oxide layer through its thermochemical reduction by aluminum, and b) to the formation of an intermediate interlayer with the participation of Cu, possessing low radiating capacity, i.e. to a reduction in the power consumed by the cathode for heating.

It should be borne in mind, however, that in the process of thermochemical reduction of the oxide layer adjacent to the core there occurs the release of a by-product of this reaction: for example, \(Ba_2SiO_4\) in the case of the activating impurity Si, \(Ba_3WO_6\) in the case of the impurity W, etc. It is interesting to note the influence of a small (\(\sim 4\%\)) admixture of W to the Ni core, which has the most favorable effect on the emission of an oxide cathode\({}^{54}\), apparently by virtue of the reaction \(6BaO+W \rightleftarrows Ba_3WO_6+3Ba\). If there is also another small impurity in the core, for example Si, this reaction will proceed further according to the type \(2Ba_3WO_6+Si \rightleftarrows 3Ba_2SiO_4+W\), etc., as a result of which the compound \(Ba_2SiO_4\) will also be found in the contact blocking layer.

The layer of this new compound, whose thickness may reach several microns, has increased resistance and other features typical of chemical blocking layers. Barium orthosilicate \(Ba_2SiO_4\)\({}^{55}\), which we have already mentioned above, has been subjected to very detailed investigation.

The resistance of these blocking layers can be measured by various methods. Alongside the most direct and reliable method indicated above (see Fig. 11), which, however, requires the use of a complicated probe technique applicable only to laboratory specimens, several variants of radio-measurement methods\({}^{28}\) have recently been developed. All of them ultimately reduce to a known replacement of the blocking layer by an electrical cell consisting of a capacitance \(C_3\) with a resistance \(R_3\) connected in parallel, whose time constant is \(C_3R_3 \sim 1\) microsec. The value of these admittedly indirect and approximate methods lies in the possibility of applying them to the study of cathodes of ordinary electron tubes. In these studies a very interesting fact was established: the increase in the resistance of the blocking layer \(R_3\) during very prolonged operation of the cathode depends substantially on whether or not electron current was drawn. In the second case the increase in the value of \(R_3\) occurs much more strongly than in the first; it is possible that electrolytic processes in the cathode play some role here. At present it is accepted that the presence of these blocking layers plays a substantially negative role not only in the obvious case of a pulse cathode operating under conditions of a large current load, but also in the case of continuously operating cathodes with a small (\(\leq 0.1\, a/cm^2\)) density of emission current. This appears somewhat incomprehensible, since in this latter case criterion (3) applies \((\lambda x_0 \ll d)\), i.e., the emission should be determined by the near-surface layer of the cathode, with an extent of several mean free paths of the electron, for the study of whose properties ...

can be applied the combination, developed by us, of the probe method and the electrostatic analyzer (see above, Fig. 12). It is quite natural that the barrier layer should be absent if very pure electrolytic Ni is used as the core, which is of special importance in the case of a continuously operating oxide cathode with a high emission density \(J_e\).

At present there are indications of the creation of a cathode with \(J=19\ a/\mathrm{cm}^2\) in continuous operation\({}^{55}\). If this cathode had a barrier layer with a resistance \(\sim 3\ \Omega/\mathrm{cm}^2\), then in its volume a Joule power \(\simeq 1\ \mathrm{kW}/\mathrm{cm}^2\) would be released, which exceeds by a factor of 200 that required for its normal heating.

It is interesting to note that in the case of this cathode we obtain a high energy efficiency of thermionic emission, of the order of 5%. Finally, from the energy balance in the operation of a vacuum thermionic cathode it follows that the true limiting value of the efficiency of the cathode, taking into account the power \(J_e^2R\) expended at the expense of the anode source, is equal to

\[ H_{\max}=\frac{1}{(\varphi+v_0)}, \]

where \(\varphi\) is the total work function and \(v_0=\frac{2kT}{e}\) is the mean energy of the thermoelectron\({}^{36}\).

A large and interesting group of works was carried out with the oxide cathode in the study of its bulk properties by various electrical and optical methods. Among the first we note the investigation of the temperature dependences of the conductivity, separated from the contact and near-surface conductivity by the probe method\({}^{27}\), then the investigation of thermopower\({}^{30}\) and the Hall effect\({}^{31}\). By the thermopower method it was discovered that the cathode layer has conductivity of a mixed character, with predominance of hole conductivity in the unactivated state and electron conductivity in the activated states. It was then shown how, in the process of activation of a cathode with a silicon-nickel core, the zone of reversal of the sign of conductivity from hole to electron shifts from the core to the outer surface, thereby showing that the source of the formation of the metallic barium is located at the core, where the thermochemical reduction of BaO takes place.

There is an attempt to consider the features of the thermopower, taking into account the fact that the oxide-cathode layer has a porous structure\({}^{48}\). The latter is not only self-evident on the basis of technological factors, but is also suggested by some features of the temperature dependence of the cathode conductivity\({}^{48}\).

suitable for use in the future; however, it is still too early to speak of this, since the elements of group III of the periodic system Dy—Yt cited here have as yet been very little investigated. Thorium oxide, ThO$_2$, has been studied rather extensively and is often used in practice; we shall now turn to a brief description of it.

An oxide-thorium cathode consists of a layer of ThO$_2$ deposited in an appropriate manner on a high-temperature core of W, Ta, or Mo$^{66}$. Activation of this cathode, as always in these cases, amounts to the precipitation within the layer of Th metal atoms to the usual concentrations $\sim 10^{18}\ \mathrm{cm}^{-3}$; it is carried out by simple heating to $\simeq 2000^\circ$ K. The operating temperature has a comparatively high value, 1900—2000° K, and just as high (of the order of $\sim 40\ \mathrm{W}/\mathrm{cm}^2$, i.e., an order of magnitude greater than for an oxide cathode) is the power dissipated by it in radiation, without taking into account the power going directly to emission and equal to $J\left(\varphi+\dfrac{2kT}{e}\right)$. The steady and pulsed emissions of this cathode differ little from one another and have a value $\simeq 3—10\ \mathrm{A}/\mathrm{cm}^2$ at 2000° K. The thorium oxide layer is an ordinary semiconductor with a very low specific resistance in the activated state, $\sim 1\ \Omega\cdot\mathrm{cm}$, which does not form chemical compounds of the barrier type with the core material, and this is its great advantage. Thus, the oxide-thorium cathode is an ordinary semiconductor cathode, to which the above-described theory of a “thick” layer is unconditionally applicable.

2. Borides. Among various, as yet still laboratory, samples of new thermocathodes, compounds of various alkaline-earth or rare-earth metals with boron of the type MB$_6^{67}$ are of great interest. Six different borides have been investigated: CaB$_6$, SrB$_6$, LaB$_6$, CeB$_6$, ThB$_6$, and the boride of a mischmetal (a getter alloy of lanthanum with cerium). Among all these compounds, lanthanum boride LaB$_6$ is of particular interest. It is a solid with a high melting point (2210° C), with metallic conductivity in absolute magnitude and temperature coefficient, with a thermoelectric power of hole sign $\varepsilon_m \simeq 8\ \mu\mathrm{V}/\mathrm{degree}$, a current-carrier concentration $n_p \simeq 8.2\cdot 10^{21}\ \mathrm{cm}^{-3}$, and their mobility

\[ \mu \simeq 13.5\ \frac{\mathrm{cm}^2}{\mathrm{sec}\cdot\mathrm{volt}}. \]

The substance of a boride cathode in the heated state interacts vigorously with the metals W, Mo, Pt, Cb, and Ta, as a result of which inactive boride compounds of these metals are formed, while the active alkaline-earth or rare-earth metal evaporates; this fact still creates great difficulties for the investigation and use of these cathodes. To some extent, this difficulty can be...

2. Various chemical compounds of metals

1. Oxides. After the highly favorable role of barium and strontium oxides in producing an efficient thermocathode had become clear, it was natural to turn also to the oxides of other metallic elements. At the present time oxides of many metals of groups I—IV have been investigated,\(^2\) yielding a number of valuable results. For example, if for the oxides of metals of group I the value of the work function \(\varphi\) is, in general, close to 1 eV, then for group II these values lie in the region \(1—2\) eV, and for groups III—IV in the region \(2—3\) eV, i.e., they increase progressively. However, in the same direction the stability of the oxide also increases, i.e., the operating temperature; therefore the optimum may lie somewhere in the middle region. Thus, cesium oxide, despite the very small value \(\varphi \simeq 0.75\) eV, is entirely unusable because of its enormous volatility; conversely, thorium oxide, although it has \(\varphi = 2.6\) eV, being comparatively little volatile, permits heating up to 2000°K. Thus, in the final analysis the question will be decided by the emission capability at the practically highest permissible evaporation rate, from the standpoint both of evaporation of the cathode and of activation of the surrounding electrodes—a rate of evaporation which, incidentally, is known to be very unsatisfactory for many of these substances.

Table VIII

Oxides	\(\varphi\) eV	\(A\) A/cm\(^2\)·deg\(^2\)	\(T_p\) °K	\(J\) A/cm\(^2\)
Oxide Cs . . . . .	0.75	0.1	400	\(10^{-7}\)
» Ba . . . . .	1.5	0.1	1000	0.5
» Ba + Sr . . .	1.0—1.2	0.1—1.0	1000	3.5
» Dy . . . . .	2.28	1.6	1850	3.2
» Ga . . . . .	2.18	0.5	1850	1.6
» Sa . . . . .	2.38	2.0	1850	1.5
» Nd . . . . .	2.0	0.2	1850	2.6
» Pr . . . . .	1.9	0.1	1850	1.8
» Ce . . . . .	2.3	1.0	1800	1.2
» Yt . . . . .	2.4	1.0	1870	1.2
» Th . . . . .	2.6	5	1900	2.0

Table VIII gives the values of the total work function \(\varphi = (\zeta + \chi)\), the constant \(A\), the operating temperature \(T_p\), and the corresponding thermionic-emission density for several oxides. From this table it is evident that we already have at our disposal a fairly large stock of oxides which, possibly, will prove suitable for ...

Substituting for \(n\) the corresponding expression, and for \(\theta\) its value from the preceding one, the author arrives at the following:

\[ J = J_e \exp \left( \frac{\chi}{kT} \frac{\beta J_a^2}{1+\beta J_a^2} \right), \tag{14} \]

where \(J_e\) is the ordinary “thermal” emission of the cathode. Figure 16 serves to explain what has been said; from it one can see that, in the general case, the emission characteristic consists of three parts: \(I\)—the space-charge regime, \(II\)—the saturation regime, and \(III\)—the regime of the so-called “surface field.” In the case where the emission capacity \(J_e\) increases above a certain limit, the second case in Fig. 16 may occur, when throughout the entire region there is only the surface-field regime, i.e., the current is limited by the three-halves law. This new regime is characterized by the fact that the effective emission capacity \(J\) depends not only on the properties of the cathode, but also on the actually flowing anode current

\[ J_a = \gamma v_a^{3/2}, \]

which causes the appearance of an activating emission under the surface field

\[ E_n = J_a \rho_n. \]

Fig. 16.

Although the author attempts to use the data of this theory for a qualitative explanation of the behavior of an oxide cathode in pulsed operation, nevertheless, in view of the actual nonobservance of all the criteria of the theory cited above, this should be treated with great caution, although the general character of this theory is quite plausible. It is therefore desirable that this interesting theory be improved as quickly as possible and extended to the case corresponding to the operating conditions of an oxide-type cathode when high thermionic-emission current densities are drawn.

In conclusion of this section we note that a whole series of interesting results has been obtained on the question of secondary electron emission of heated oxide cathodes\({}^{42}\), which is of great importance as applied to the regime of their operation in magnetrons; however, since this question lies somewhat apart from our principal topic and is of independent interest, we shall not dwell on it.

22.3 A/cm², at an anode voltage equal to \(v_a = 867\) V, the emitted electrons had energies equal to 858, 730, 675 V; thus the energy deficit, i.e. the quantity \(v_k\), reaches in this case 200 V, having, most importantly, a discrete character. One may hope that further development and refinement of such experiments will probably lead to interesting conclusions about the nature of the pulsed thermionic emission of oxide cathodes.

Under conditions when an oxide cathode gives thermionic emission of high density, for example in a pulsed regime, very large electric fields can arise in the cathode layer; the latter, being partially concentrated in the emitting electron near-surface region of the cathode, as is seen, for example, from Fig. 12, can exert a decisive influence on the emission[^56]. For example, if this region has a resistance of the order of \(1\ \Omega/\text{cm}^2\), then with an ohmic extrapolation to the case \(J_e = 100\ \text{A}/\text{cm}^2\) and the assumption that the thickness of this near-surface layer is \(d_n \simeq 1—10\,\mu\), we obtain very large near-surface fields \(E_n \simeq 10^5—10^6\ \text{V}/\text{cm}\). Moreover, in this near-surface layer the processes may differ substantially from the conditions inside the cathode; in particular, here, at high thermionic emission, the thermal equilibrium of the electrons may be disturbed. For example, at \(T = 1100^\circ K\) and a concentration of free electrons

\[ n_e = 10^{13}\ \text{cm}^{-3} \]

the disordered current density

\[ j = en\sqrt{\frac{kT}{2\pi m}} \simeq 800\ \text{A}/\text{cm}^2, \]

whereas we are capable of extracting outward thermionic emission with \(J = 100\ \text{A}/\text{cm}^2\). All this indicates that, in the region of high emission densities of semiconductor cathodes, new regularities should occur, including the influence of the electric field, undoubtedly important in this case, in the near-surface layer of the cathode[^56].

In connection with all that has been said, much interest attaches to the new theory of thermionic emission of semiconductor cathodes in the region of high thermionic-emission densities[^65], based on the considerations set forth above. For a polar semiconductor, for example, such as an oxide cathode, in an electric field \(E\) there occurs a change in the concentration \(n_e\) of free electrons; the latter, at comparatively small \(E\) (\(\ll 10^3\ \text{V}/\text{cm}\)) and \(T\) \((\varepsilon \ll \hbar\omega)\), and in the range of applicability of Ohm’s law, preserve a Maxwellian distribution in velocities, but with a different effective temperature \(\theta\):

\[ \theta = T(1+\alpha E^2) = T(1+\beta J_a^2), \tag{12} \]

where \(J_a\) is the actual anode current. Hence the emission current is equal to

\[ J = Dn(T,E)\sqrt{\frac{k\theta}{2\pi m}}\exp\left(-\frac{\chi}{k\theta}\right). \tag{13} \]

At the same time, indications appeared\(^{54,61}\) that the best cores not only for continuous but also for pulsed emission of an oxide cathode are those made of Ni + 4% W.

If the duration of the pulse of the emission current being drawn is gradually increased, then its gradual decrease with time, usually called self-poisoning,\(^{62}\) begins to manifest itself to an ever greater extent; therefore, in order to eliminate this phenomenon, the pulse duration should not be made greater than approximately 1 microsecond. A large number of different assumptions were made in due course regarding the nature of the phenomenon of self-poisoning of the oxide cathode, among which, in particular, an undoubted one is that which connects it with the care taken in manufacturing the entire tube as a whole. The matter consists in the poisoning of the cathode by various impurities, such as, for example, gases released from the electrodes of the tube and from the films covering them, gas ions formed in the volume when current passes through the tube under investigation, electrolytic components of the current through the cathode layer, etc. In general, elimination of gas evolution in the tube may also prove important in the case of an oxide cathode of continuous operation with a comparatively small current density, if only increased requirements are imposed on its service life.\(^{63}\)

It may be thought that within certain limits, at not too high emission currents, so long as the phenomenon of thermionic emission has the character of pure evaporation described above, there should be no fundamental difference between pulsed and stationary emission, which has also been confirmed experimentally.\(^{61}\) In the case when the magnitude of the pulsed thermionic emission reaches very large values and conditions exist inside the cathode favorable for the formation in it of layers with increased resistance, especially in the near-surface layer (see above, Fig. 12), the situation may change radically. In this case a very high electric field \((>10^{5}\ \text{V/cm})\) can be established inside the interlayer, which may have the most active influence on the emission. Interesting indications in favor of the latter conclusion can, for example, be obtained from an experimental attempt to determine the energy distribution function of the emitted thermoelectrons in the case of high-density pulsed emission.\(^{64}\) If in the case of small stationary emission we obtain a practically Maxwellian distribution, then in this case, when an electrostatic analyzer of the oscillographic-tube type was used, it was found that the energy spectrum has, for some unknown reason, a discrete character, with a very large deficit associated with a large potential drop in the cathode layer \(v_k\). For example, in one case of an oxide cathode which at \(925^\circ\text{C}\) gave pulsed emission with a density

First of all, one should note the especially important significance, in this case, of the material of the cathode core, which, for example, is seen from Fig. 14[^51]. In the course of prolonged tests, a cathode with a core of electrolytically pure nickel proves to be much better than a core of Ni + 0.2% Si. A more detailed study of the role of the core material shows, as is seen from Fig. 15[^55], that the application of Si (in this case 2%) reduces this role to the appearance of the following two harmful factors: 1) the current-voltage characteristics from the very beginning already begin to depart from the theoretical ones (three-halves law), evidently as a consequence of the appearance of an appreciable potential drop in the contact barrier layer, and 2) the limiting emission capability obtained here proves to be very

Fig. 14.

Fig. 15.

lowered, owing to the early appearance of sparking of the cathode (marked in Fig. 15 by a circle with a cross). The nature of the latter phenomenon, which acts destructively on the cathode, is still not entirely clear; it is associated with local arc formation, breakdown of blocking layers in the cathode due to the formation here of very large electric fields, etc.[^60] A more detailed physical investigation of this important question would be of great significance for the problem of semiconductor cathodes with high emission capability. Finally, in the very last-

of the expression following from (6)

\[ \zeta=\frac{1}{2}q+kT\ln\sqrt{\frac{B_e}{N_a}} \]

will now be equal to

\[ \zeta=\left(\frac{1}{2}W+H\right)+kT\ln\sqrt{\frac{B_p}{N_a}} \]

and so on. Thus, this theory not only refines certain expressions important for us, which are evidently also applicable to the oxide cathode, but also makes it possible to foresee certain parameters. For example, for the system of an impurity single crystal BaO + Ba, on the basis of (10) and (11) we obtain \(H=0.37\) eV and \(W=1.59\) eV, i.e. \(\zeta=1.8\) eV; from the latter one gets the impression that the magnitude of the external work function \(\chi\) can sometimes be quite small.

In our preceding basic version of the theory of thermionic emission of a semiconductor cathode, the conditions (3) were assumed, conditions wholly applicable to an ordinary oxide cathode of continuous operation, since here approximately \(\lambda \simeq 0.01\) and \(x_0 \simeq 10^{-5}\) cm. However, in principle, cases are possible of such thin-film semiconductor thermocathodes—for example, of the type of the old “barium,” oxide-cesium, etc.—that here only the first inequality \(\lambda \ll 1\) is satisfied, while the second \(x_0 \ll d\) is not satisfied. In this case it is possible to obtain unusual regularities of thermionic emission, owing to the fact that the electric field of the internal contact semiconductor—metal core will extend to the external emitting surface of the cathode\({}^{59}\). We consider it necessary to note this fact; however, in view of the fact that at present there are as yet no concrete cathodes of this type in use, we shall not dwell on it in greater detail.

In recent years the oxide thermocathode has come to be widely used in a new regime of short-time (of the order of a microsecond) pulses with a long pause between them. In this case the possibility was found of obtaining very large thermionic-emission current densities, reaching in one case as much as \(140\ \mathrm{a/cm^2}\) \({}^{51}\). In this case certain special conditions may be created, differing from the conditions of the steady-state regime, although even in the latter case, with a core of pure Ni and very thorough evacuation and degassing of the investigated tube and all its electrodes, it is possible to obtain very large densities, up to \(19\ \mathrm{a/cm^2}\) \({}^{53}\). In connection with the large and important applications of the pulsed emission of oxide cathodes, a number of experimental works were devoted to the study of its features, the main content of which consists briefly in the following.

for example, the peculiarities of evaporation from the cathode in technical radio tubes, where the amount of evaporated substance may reach several percent of its total amount, then during the evaporation of individual components of the tricarbonate \((\mathrm{BaO}+\mathrm{SrO}+\mathrm{CaO})\) coating, etc. It should be noted that the struggle against thermionic emission of the grids of radio tubes, which appears as a result of deposition on the grid of evaporation products from the oxide cathode, is a very difficult matter \(^{23,24}\) and to this day is still practically unresolved.

In concluding our consideration of the oxide cathode of continuous operation, let us turn to certain features of the theory of thermionic emission that go beyond those elementary notions of which we have already spoken above.

The oxide cathode in the activated state is a mixture of \(\mathrm{BaO}\) and \(\mathrm{SrO}\) crystals, which are polar compounds, with a volumetric admixture of Ba and Sr atoms. The basis of this cathode—the \(\mathrm{BaO}+\mathrm{Ba}\) crystal—turns out to be not an electronic but a polaron semiconductor, which has its own specific features, already described in the literature \(^{57}\). In this connection certain features will also be possessed by the theory of thermionic emission of such substances, for example in the activated state, consisting in the following \(^{58}\). The emission formula has the former form (5), but the energy of dissociation of an electron from the impurity center is \(q=(H+W)\), where \(H\) is the energy of thermal dissociation of a polaron with the formation of a band electron, \(W\) is the dissociation energy of an \(F'\)-center (an impurity divalent barium atom) into an \(F\)-center and a polaron. Both these quantities \(H\) and \(W\), and consequently also the quantity \(q\), can be precalculated on the basis of the following parameters: \(n\), the optical refractive index; \(\varepsilon\), the static dielectric constant; \(\omega\), the limiting frequency of optical vibrations; and \(\mu\), the effective mass of the band electron. The latter can be determined by comparing the experimentally determined position of the maximum of the impurity absorption band with the theoretical one; for the \(\mathrm{BaO}\) crystal it was found that

\[ \frac{\mu}{m} \approx 5.8. \]

Thus,

\[ H = 0.054\,\frac{me^4}{\hbar^2}\,\frac{\mu}{m}\,c^2 + \frac{3}{2}\hbar\omega,\quad \text{where } c=\left(\frac{1}{n^2}-\frac{1}{\varepsilon}\right), \tag{10} \]

\[ W=\frac{me^{-4}}{\hbar^2}\,\frac{1}{9\varepsilon}\left(\frac{4}{n^2}+\frac{3}{\varepsilon}\right)\frac{\mu}{m}-\frac{3}{2}\hbar\omega, \tag{11} \]

and the internal work function, for example, in the well-activated state at low temperatures \([kT \ll (q-\zeta)]\) instead of

a number of interesting regularities. For a general characterization of the latter phenomenon we shall give in Fig. 13 the spectral characteristic of the quantum yield as a function of photon energy (in ev) for an activated cathode of BaO with the following values of the thermionic constants: \(\varphi = 1.44\) ev, \(A = 0.026\ \mathrm{a/cm^2\cdot deg^2}\); here a maximum is observed in the region \(3.6—4.0\) ev.

In another work, a second sharp increase in the quantum yield of the BaO photoeffect was obtained upon irradiation with light with \(h\nu > 5\) ev, which the authors attribute to the onset of the proper external photoeffect. Study of the current-voltage characteristics of the photoeffect of this cathode in the retarding electric field of a spherical condenser confirmed the above value of the threshold energy of the proper photoeffect of the BaO cathode, \(\Phi_{\phi}\simeq 5.0\) ev, and also indicated that the thermionic work function here is \(\varphi\simeq 1.6\) ev.

Fig. 13.

The oxide cathode and barium oxide also exhibit appreciable cathodoluminescence\(^{34}\) with a band spectrum in the region \(250—860\) mμ. Moreover, the character of this spectrum depends on the state of activation of the cathode, in the course of development of which a new band appears in the region of 410 mμ and another band in the region of 460 mμ is greatly weakened. All these data indicate the great advisability of further development of optical methods for investigating physical phenomena in the oxide cathode.

On the basis of these investigations\(^{32,33}\), attempts were made to construct an energy-band diagram of the oxide cathode, one of the variants of which we presented in Fig. 9.

Of substantial importance in investigations of the oxide cathode are questions of evaporation from it of its active components, their deposition on all nearby electrodes, and the influence which they exert on the contact potential and thermionic activity of these electrodes. The importance of this problem is connected with the stabilization of the operation of the control grids of modern radio tubes with oxide cathodes. In these investigations, in recent years, the techniques of Berdennikova,\(^{49}\) the mass spectrometer\(^{39}\), and especially labeled radioactive atoms Ba\(^{140}\) and Sr\(^{89}\) have been successfully applied. Using the latter, it has been found, for example,

By means of the Hall effect, a whole series of important features of this cathode was also clarified. First of all, the presence in the oxide cathode of two signs of conductivity—electronic and hole conductivity—and the possibility of its inversion when both the cathode temperature and the pressure of the oxygen introduced into the tube are varied were confirmed here. Numerical data were then obtained concerning the concentration of electrons and holes in the cathode and their mobilities; unfortunately, these data are not of an entirely definite character. In general, it may be assumed approximately that at \(1100^\circ\) K the concentration of free electrons in the oxide cathode is

\(n_e \approx 10^{14}\ \text{cm}^{-3}\),

and their mobility

\(\mu \approx 10^3\ \text{cm}^2/\text{sec}\cdot\text{volt}\).

At present, considerable attention is being attracted by various optical investigations; the objects of these investigations are both ordinary oxide cathodes and, especially, pure and impurity single crystals of BaO, for which a growth technique has already been developed\({}^{36}\) and which serve, as it were, as a model of the oxide cathode. The value of these investigations, especially those carried out with BaO single crystals, lies in the possibility of determining the band scheme of this crystal in the purest form, not obscured by extraneous factors, and thereby of coming substantially closer to an understanding of the physical phenomena in the oxide cathode. In these spectral investigations, questions of spectral absorption, internal and external photoeffect, and cathodoluminescence were touched upon\({}^{32,33,34}\). Experiments show that BaO has a broad band of optical absorption\({}^{32}\), beginning at the wavelength \(\lambda_0 \approx 330\ \text{m}\mu\) (\(h\nu \approx 3.8\ \text{eV}\)) and rapidly reaching a maximum in the region \(\lambda_m \approx 200\ \text{m}\mu\) (\(h\nu \approx 6.2\ \text{eV}\)). At this maximum the value of the absorption index reaches \(k \approx 0.56\), i.e. the absorption has an almost metallic character, thereby indicating that we are dealing here with intrinsic absorption associated with interband transitions; the width of the forbidden band for BaO is therefore \(Q = 3.8\ \text{eV}\) (according to other data, \(4.8\ \text{eV}\)). In the case of a BaO single crystal activated by an impurity of metallic Ba, an additional absorption band appears with a maximum in the region \(\lambda \approx 510\ \text{m}\mu\) (\(h\nu' = q = 2.4\ \text{eV}\)), typical in its character of impurity \(F'\)-absorption; from the absolute magnitude of the absorption coefficient in this region, an estimate was made of the concentration of this Ba impurity, leading to the value

\[ N_a \approx 4 \cdot 10^{18}\ \text{cm}^{-3}, \]

which agrees with data obtained by entirely different methods.

Ba oxide reveals the presence both of an internal photoeffect\({}^{32}\), whose threshold coincides with the threshold of optical absorption (\(330\ \text{m}\mu\)), and of an external photoeffect\({}^{33}\), revealing ...

are fabricated by carburizing the surfaces of metals that are in contact with boron carbide, or by using the graphite crucibles of these cathodes for the investigation. The emission constant of lanthanum and mish-metal borides, as well as their emission at \(1900^\circ\) K, are presented in Table IX.

Table IX

Borides	\(\varphi\) eV	\(A\) A/cm\(^2\)·deg\(^2\)	\(J\) A/cm\(^2\)
LaB\(_6\) . . . . . .	2.66	29	10.5
Mish-metal boride . . . . . .	2.64	14	5.0

In comparison with other cathodes (see Fig. 1), attention is drawn to the fact that LaB\(_6\), in its emission activity, is inferior only to the oxide and metal-sponge \(L\)-cathode, but it considerably surpasses not only pure metals, but also film cathodes with thorium and oxide-thorium coatings.

It may be supposed that, in its activated state, which is easily attained by brief heating at \(1500\)—\(1600^\circ\) C, this LaB\(_6\)-cathode has on its surface an active monatomic film of metallic lanthanum. This is confirmed by the fact that, after prolonged operation, the inner surface of the bulb becomes covered with a deposit consisting, as it turns out, practically of lanthanum. An advantage of the LaB\(_6\)-cathode is the very easy formation of the surface active film, i.e., the restoration of the cathode emission after poisoning or cathodic sputtering; the threshold for the latter for mercury ions is approximately 20 eV. At a temperature below \(2200^\circ\) K, i.e. in the operating region, the diffusion rate \(N_{\text{д}}\) of La atoms to the surface exceeds the rate of its evaporation \(N_{\text{и}}\), i.e., according to (2), there are created

Fig. 17.

conditions favorable to the possibility of the existence of a surface active film, and consequently to the efficient operation of the cathode. This is seen from Fig. 17, where the temperature dependences of both quantities \(N_{\text{d}}\) and \(N_{\text{i}}\) are presented in the usual semilogarithmic scale; hence, incidentally, the heats of diffusion and evaporation for the LaB\(_6\) cathode are \(Q_{\text{d}} = 4.0\ \text{eV}\) and \(Q_{\text{i}} = 7.3\ \text{eV}\), respectively. Thus, this cathode is of undoubted interest; however, its practical application must still be preceded by considerable work, primarily of a technological nature.

3. Various compounds. Purely qualitative tests were carried out on carbides and nitrides of several metals\(^2\).

Table X

Carbides	\(\varphi\), eV	\(A\), A/cm\(^2\)·deg\(^2\)	\(J_e\), A/cm\(^2\) at 1800° K
ZrC	2.18	0.31	0.9
ThC\(_2\)	3.5	550 (!)	0.35

Of these, only ZrC and ThC\(_2\) are so far considered promising; they are characterized by the following data (Table X). The compound ThC\(_2\) operated stably at 1900° K for 900 hours with a current draw of 1 A/cm\(^2\). In general, however, it is still premature to say anything definite about these cathodes.

Of somewhat greater interest are the results of an investigation of the properties of barium sulfide—BaS\(^ {68}\)—in the form of a thin layer on a metallic substrate. This compound is an electronic semiconductor which, in the activated state, has an internal work function \(\zeta \gtrsim 1.2\ \text{eV}\) and a thermoelectric power \(\simeq 1\text{–}2\ \text{mV/deg}\). The emission capability of the best specimens reaches a value corresponding to a BaO layer, i.e. \(\simeq 1\text{–}10\ \text{mA/cm}^2\) at 1000° K, with a total work function \(\varphi \simeq 2.0\ \text{eV}\). In an equiconcentration mixture BaS—BaO, the vapor pressure of BaS is approximately two orders of magnitude lower than that of BaO.

3. Mixed ceramic systems

Along with the cathodes of the nonmetallic type described above, which are continuous layers of the corresponding substance, a whole series of tests was also carried out with systems consisting of mixtures of powders of the corresponding active substance and a metal, the grains having dimensions of the order of microns. Usually this mixture is pressed and then sintered in hydrogen, as a result of which a cathode suitable for practical use is obtained. It was assumed that in this case the role of the metallic component would consist both in facilitating

restoration, i.e., activation of the cathode, and in reducing its transverse resistance, i.e., increasing its resistance to sparking, etc. Unfortunately, at present we do not possess serious studies of the physical structure of these systems. In the main, only the emission properties of these cathodes have been studied; we shall now turn to their description.

1. Cathodes on an oxide–barium base (oxide sinter cathode)\(^{69}\). In the initial stage, cathodes of this type used layers of powder paste consisting of 45% (by weight) double oxide and 55% nickel. Subsequently this system was replaced by a matrix applied to a core, sintered from Ni powder, into which an oxide paste was pressed.

Such a cathode proved quite acceptable for operation in pulsed magnetrons owing to a strong reduction in the probability of sparking and, as a consequence, an increase in the service life of the cathode. A new variant of this cathode has recently been described, consisting of a pressed and annealed pellet composed of a mixture of powders of 30% double carbonate and 70% nickel.

The emission capability of a cathode of this type is represented by a curve in summary Figure 1. For a short time this cathode gave a stationary emission with a density of \(\simeq 10\ \mathrm{a/cm^2}\) at \(950\text{–}1000^\circ\mathrm{C}\); however, its chief feature is the possibility of prolonged operation at high anode voltage, for example, 5000 hours at \(J \sim 0.5\ \mathrm{a/cm^2}\), \(T = 850^\circ\mathrm{C}\), and \(v_a = 500\text{–}1200\ \mathrm{v}\). Another important feature is its much greater resistance, compared with ordinary oxide cathodes, to the poisoning and sputtering action of gases.

Another variant of this cathode, close in character to the L-cathode, is a porous tungsten block filled with 5–10% (by weight) barium oxide. In this case it is necessary to start from such barium compounds which, on reacting with W, would lead to the liberation of metallic barium activating the system of this cathode; this, for example, is not the case for \(\mathrm{BaCO_3}\), since the reaction here proceeds as follows:

\[ 3\mathrm{BaCO_3} + \mathrm{W} \rightarrow \mathrm{Ba_3WO_6} + 3\mathrm{CO}, \]

whereas in the reaction with \(\mathrm{BaO}\) we obtain

\[ 6\mathrm{BaO} + \mathrm{W} \rightarrow \mathrm{Ba_3WO_6} + 3\mathrm{Ba}^{40}. \]

The emission constants of such a cathode are \(\varphi = 1.56\ \mathrm{ev}\) and \(A = 0.6\ \mathrm{a/cm^2\cdot deg^2}\). Tests of these cathodes showed the possibility of their operation for 650 hours at \(1000^\circ\mathrm{C}\) with emission \(> 8\ \mathrm{a/cm^2}\); in general, from these cathodes it was possible to obtain emission up to \(100\ \mathrm{a/cm^2}\), and no difference between stationary and pulsed values was observed here.

1. Cathodes on an oxide–thorium base (ceramic cathode). A cathode of this type\(^{70}\) is, in contrast to the ordinary oxide–thorium cathode, a solid body of oxide-tho-

...ceramic. To reduce the resistivity, such a cathode is made from mixtures of the basic ThO$_2$ powder with powders of the metals W, Mo, Re, or the nitrides ZrN and TiN; this mixture is pressed and then sintered in hydrogen. As a result, a substance is obtained that permits the manufacture of directly heated cathodes with a low specific resistivity $\simeq 3 \cdot 10^{-2}\ \Omega \cdot \text{cm}$ at $1800^\circ$K. These cathodes are readily activated and their emission is fairly stable; however, their mechanical properties are still very unsatisfactory.

The emission constants of the cathode are equal to $\varphi = 2.97$ eV and $A = 89\ \text{a}/\text{cm}^2 \cdot \text{deg}^2$, i.e., their emissive capacity is somewhat lower than that of an ordinary oxide-thorium cathode. Tests of cathodes at $1700^\circ$C with an emission of $3.5\ \text{a}/\text{cm}^2$ gave a service life of about 1000 hours; at a temperature of $1920^\circ$C (a very forced regime), under pulsed conditions it was possible to obtain up to $60\ \text{a}/\text{cm}^2$.

The experiments carried out indicate that the increased activity of this cathode is associated with the formation on its surface of an active film of metallic Th; the latter is formed inside by dissociation of ThO$_2$, followed by diffusion outward, as in the case of an ordinary thoriated cathode. For a system consisting of 67% ThO$_2$ + 33% W, these diffusion and evaporation processes were investigated; the heats activating these processes proved to be considerably lower than in the case of thoriated tungsten, namely: the heat of evaporation $Q_a \simeq 2.0$ eV, and the heat of diffusion from the inside outward $Q_g \simeq 1.4$ eV (instead of values $\simeq 7.8$ eV and 4.1 eV, respectively, for Th—W); the fact that $Q_a > Q_g$ is, for the cathode, a favorable factor, making possible the formation on its surface of a stable active film of metallic thorium. With improvement, these cathodes may find application under severe operating conditions, where the economy of emission, which is comparatively low here, is of secondary importance.

In concluding by summarizing everything set forth in the present review, we may note the very great successes achieved recently in the study of the physical properties and in the development of new types of thermionic cathodes. However, these successes still lag behind the rapidly growing demands of practice, which are connected with the use of these cathodes in the most diverse electronic devices that form the basis for the development of a large number of branches of new technology. It is therefore necessary to accelerate the further development of a broad front of physical investigations and technological developments in this important field, and to strengthen in every way the creative collaboration of workers in science and industry.

Cited Literature

1. See, for example, H. Friedenstein, S. Martin and G. Munday, Rep. on Progr. in Phys. 11, 298 (1946—1947); S. Ptitsyn, ZhTF 17, 965 (1947); O. Weinreich, Rev. gen. d’élect. 56, 75 (1947); A. Eisenstein, in the collection Advances in electronics 1 (1948); H. Germann and S. Vainer (with additions by B. Tsarev), Oxide Cathode, Gostekhizdat, 1949; S. Ptitsyn, Physical Phenomena in the Oxide Cathode, Gostekhizdat, 1949; K. Herring and M. Nichols, Thermionic Emission, IL, 1950; collection: Problems of Radiolocation Engineering No. 6, 81 (1951), IL; L. Dobretsov, Electron and Ion Emission, Gostekhizdat, 1952; D. Wright, Proc. I. E. E. 100, 125 (1953); Collections of Abstracts: Electron and Ion Emission, issue 1 (1953); issue 3 (1954), IL.

2. Collection: Problems of Radiolocation Engineering No. 6, 81 (1951); D. Wright, Proc. I. E. E. 100, 125 (1953).

3. R. Levi and G. Espersen, Phys. Rev. 78, 231 (1950).

4. E. Wohlfarth, Proc. Phys. Soc. 60, 360 (1948); H. Jvey, Phys. Rev. 74, 983 (1948); E. Wohlfarth, Phys. Rev. 74, 984 (1948).

5. L. Whitney, Phys. Rev. 50, 1154 (1936); H. Wahlin, Phys. Rev. 61, 509 (1942); A. Wahl, Phys. Rev. 82, 574 (1951).

6. S. Jain and K. Krishnan, Proc. Roy. Soc. A 213, 143 (1952); 215, 431 (1952); 217, 451 (1953).

7. H. Michaelson, Journ. appl. phys. 21, 536 (1950).

8. See, for example, J. de Boer, Electron Emission and Adsorption Phenomena, Gostekhizdat (1936); A. Reiman, Thermionic Emission, Gostekhizdat, 1940; K. Herring and M. Nichols, Thermionic Emission, IL, 1950; L. Dobretsov, Electron and Ion Emission, Gostekhizdat, 1952.

9. N. Morgulis, Sputtering of a Metallic Surface under Impacts of Positive Ions, Acad. Sci. Ukr. SSR (1936).

10. H. Dailey, Electronics 21, No. 1, 107 (1948); Z. Atlee, Electr. Eng. 68, 863 (1949); R. Avery, Proc. I. R. E. 40, 591 (1952).

11. M. Andrews, Phys. Rev. 33, 454 (1929); L. Koller, The physics of electron tubes, 1934, p. 31; E. Brüsche und H. Mahl, Zeits. f. techn. Phys. 17, 81 (1936); C. Hörsting, Journ. Appl. Phys. 18, 95 (1947).

12. Yu. Boldyrev, Tekhnika zavoda “Svetlana” No. 9 (1932); collection: Gas-Discharge Devices, Glavvakuumprom, 1939, p. 106; A. Vorob’ev, Fiz. zap. AN USSR 9, 235 (1941); A. Hull, Phys. Rev. 56, 86 (1939); J. Martin, Phys. Rev. 86, 584 (1952).

13. H. Lemmens, M. Jansen and R. Loosjes, Philips Techn. Rev. 11, 341 (1950); J. Babakian, Phys. Rev. 82, 573 (1951); F. du Pré and E. Rittner, Phys. Rev. 82, 573 (1951); H. Katz and K. Rau, Die Frequenz 5, 192 (1951); H. Gudde, Funk u. Ton. 5, 103 (1951); D. Shaefer and G. White, Journ. Appl. Phys. 23, 669 (1952); E. Rittner and R. Ahlert, Phys. Rev. 85, 390 (1952); G. Espersen, Proc. I. R. E. 40, 284 (1952); H. Katz, Journ. Appl. Phys. 24, 597 (1953).

14. R. Hughes, P. Coppola and H. Evans, Journ. Appl. Phys. 23, 635 (1952).

15. M. Gurtovoi and G. Kovalenko, Fiz. zap. AN USSR 9, 240 (1941); N. Morgulis and P. Marchuk, ZhTF 19, 135 (1949); A. Hull, E. Burger and R. Turrentine, Phys. Rev. 73, 1228 (1948); Gen. El. Rev. 54, No. 8, 16 (1951); R. Steinberg, Journ. Appl. Phys. 21, 1028 (1950); A. Günterschulze, E. T. Z. 72, 575 (1951); M. Coquerel, Le vide No. 46—47, 1384 (1953).

16. G. Moore and H. Allison, Phys. Rev. 77, 246 (1950).

1. L. Aldrich and C. Smith, Phys. Rev. 77, 744 (1950).
2. V. Gavrilyuk, Proceedings of the Institute of Physics, Academy of Sciences of the Ukrainian SSR, issue 5, p. 87 (1953).
3. A. Shulman and A. Rumyantsev, DAN SSSR 93, 455 (1953).
4. S. Starodubtsev, ZhETF 19, 215 (1949).
5. F. Knauer, Zeits. f. Phys. 125, 278 (1949).
6. M. Kruglova and I. Sokolskaya, ZhTF 19, 1292 (1949); Kim Hen Bon and I. Sokolskaya, Vestnik LGU No. 12, 67 (1952).
7. B. Gorelik, ZhTF 15, 549 (1945); H. Sorg and G. Becker, Electronics 18, No. 7, 104 (1945); B. Tsarev, Collection of Articles on Vacuum Technology, Gosenergoizdat, 1948, p. 36; Contact Potential Difference, Gostekhizdat, 1949, p. 141; T. Arizumi and S. Narita, Journ. Phys. Soc. Japan 4, 300 (1949); K. Noga and G. Okoba, ibid. 4, 360 (1949); T. Arizumi and S. Kotani, ibid. 5, 70 (1950); W. Bartley and J. White, Electr. Eng. 71, 496 (1952); H. Koppen, Nachr. Techn. 2, 246 (1952).
8. M. Bernadiner and A. Nagorskii, Report of the Institute of Physics, Academy of Sciences of the Ukrainian SSR, 1941; J. Rothstein, Phys. Rev. 69, 693 (1946); C. Fay, Bell. Syst. Tech. Journ. 26, 818 (1947); B. Baker, Brit. Journ. Appl. Phys. 4, 311 (1953).
9. N. Morgulis, Izv. AN SSSR, ser. phys. 5, 536 (1941); L. Dobretsov, Electron and Ion Emission, Gostekhizdat, 1952.
10. V. Thompson, Phys. Rev. 36, 1415 (1930); N. Morgulis and I. Nakhutin, ZhETF 6, 1105 (1936); P. Drewell, Zeits. f. techn. Phys. 17, 252 (1936); A. Andrianov, Izv. AN SSSR, ser. phys. 8, 290 (1944).
11. P. Zeltsman, Diploma thesis, KSU, 1941; Proceedings of the Physics Faculty of KSU No. 6, 79 (1952); A. Fineman and A. Eisenstein, Journ. Appl. Phys. 17, 663 (1946); D. Wright, Proc. Roy. Soc. A 190, 394 (1947); N. Morgulis and V. Yagovdin, DAN SSSR 59, 247 (1948); V. Yagovdin, Proceedings of the Physics Faculty of KSU No. 6, 99 (1952); W. Danforth and D. Goldwater, Journ. Appl. Phys. 20, 163 (1949); N. Morgulis and Ya. Zingerman, DAN SSSR 81, 783 (1951).
12. J. Waymouth, Journ. Appl. Phys. 22, 80 (1951); A. Eisenstein, Journ. Appl. Phys. 22, 138 (1951); C. Eaglesfield and P. Douglas, Brit. Journ. Appl. Phys. 2, 318 (1951); G. Metson, S. Wagener, M. Holmes and M. Child, Proc. I. E. E. III 99, 69 (1952).
13. N. Morgulis, Collection Dedicated to the 70th Anniversary of Academician A. F. Ioffe, Academy of Sciences of the USSR, 1950, p. 146; N. Morgulis and Ya. Zingerman, DAN SSSR 81, 783 (1951); Ya. Zingerman, Proceedings of the Institute of Physics, Academy of Sciences of the Ukrainian SSR, issue 2, 57 (1952); A. Eisenstein, Journ. Appl. Phys. 20, 776 (1949); J. Sparks and H. Philipp, Journ. Appl. Phys. 24, 453 (1953).
14. J. Becker and W. Sears, Phys. Rev. 38, 2193 (1931); J. Blewett, Journ. Appl. Phys. 10, 678 (1939); P. Zeltsman, Diploma thesis, KSU, 1941; Proceedings of the Physics Faculty of KSU No. 6, 79 (1952); N. Morgulis and V. Yagovdin, DAN SSSR 59, 247 (1948); P. Zeltsman, N. Morgulis and V. Yagovdin, ZhTF 21, 1035 (1951); K. Noga, Journ. Phys. Soc. Japan 6, 124 (1951); E. Hensley, Journ. Appl. Phys. 23, 1122 (1952); J. Young, Journ. Appl. Phys. 23, 1129 (1952).
15. D. Wright, Nature 164, 714 (1949); Brit. Journ. Appl. Phys. 1, 150 (1950); Phys. Rev. 78, 355 (1950); 82, 574 (1951); G. Ishikawa, T. Sato, K. Okumura and T. Sasaki, Phys. Rev. 84, 371 (1951); E. Pell, Phys. Rev. 87, 457 (1952).
16. W. Tyler, Phys. Rev. 76, 179, 1887 (1949); E. Taft and J. Dickey, Phys. Rev. 78, 625 (1950); R. Sproull, Phys. Rev. 78, 630 (1950); 81, 313 (1951); W. Tyler and R. Sproull, Phys. Rev. 83, 548 (1951); Collection Semiconducting Materials, 1951, p. 122; A. Kane, Journ. Appl. Phys. 22, 1214 (1951); H. de Vorê, Phys. Rev. 85, 390 (1952); W. Dash, Phys. Rev. 92, 68 (1953).

1. E. Nishibori, H. Kawamura and K. Hirano, Proc. Phys. Math. Soc. Japan 23, 37 (1941); J. Dibiesse et R. Chompiex, C. R. 226, 239 (1948); J. Dickey and E. Taft, Phys. Rev. 80, 308 (1950); B. MacNary, Phys. Rev. 81, 631 (1951); 82, 576 (1951); H. de Vore and J. Dewdney, Phys. Rev. 83, 805 (1951); L. Apker, E. Taft and J. Dickey, Phys. Rev. 84, 508 (1951); T. Hibi and K. Ishikawa, Phys. Rev. 83, 659 (1951); 84, 1252 (1951); 87, 673 (1952); S. Ptitsyn and N. Panteleev, Collection: Proceedings of the Conference on Cathode Electronics, Academy of Sciences of the USSR (1952), p. 25; P. Carrol and E. Coomes, Phys. Rev. 85, 389 (1952); K. Noga and S. Kawamura, Journ. Phys. Soc. Japan 7, 287 (1952).

2. G. Herman and S. Wagener, Oxide Cathodes, 1949, p. 410; P. Aitchison, Nature 164, 1088 (1949); V. Stout, Phys. Rev. 77, 744 (1950); 85, 390 (1952); 89, 310 (1953); A. Rouse and J. Drahman, Phys. Rev. 85, 739 (1952).

3. H. Huber und S. Wagener, Zeits. f. techn. Phys. 23, 1 (1942); A. Eisenstein, Journ. Appl. Phys. 17, 654 (1946); 20, 776 (1949); H. Rooksby, Nature 159, 609 (1947); E. Iamaka, Journ. Appl. Phys. 22, 1087 (1951); 23, 498, 937 (1952); P. Russel, Phys. Rev. 85, 705 (1952).

4. R. Sproul, W. Dash, W. Tyler and A. Moore, Rev. Sc. Instr. 22, 410 (1951); A. Kane, Journ. Appl. Phys. 22, 1214 (1951).

5. W. Mecklenburg, Zeits. f. Phys. 120, 21 (1942); H. Morgulis and I. Deryugin, Izv. Acad. Sci. USSR, phys. ser. 15, 444 (1951); G. Spivak and E. Dubinina, Vestn. MGU No. 2, p. 27 (1953); Dokl. Acad. Sci. USSR 88, 673 (1953).

6. T. Organ and S. Parsons, J. O. S. A. 38, 191 (1948); L. Waymouth, J. Appl. Phys. 22, 80 (1951); E. Krautz, Zeits. f. Naturforsch. 6a, 16 (1951).

7. H. Schaefer and W. Walcher, Zeits. f. Phys. 121, 679 (1943); R. Sloane and C. Watt, Proc. Phys. Soc. 61, 217 (1948); R. Plumplee and L. Smith, Journ. Appl. Phys. 21, 811 (1950); L. Aldrich and C. Smith, Phys. Rev. 77, 744 (1950); P. Stier, Phys. Rev. 83, 877 (1951); L. Aldrich, Phys. Rev. 81, 320 (1951); Journ. Appl. Phys. 22, 1168 (1951); L. Nedgaard, RCA Rev. 13, 464 (1952); W. Grattidge and H. John, Journ. Appl. Phys. 23, 1145 (1952); A. Sheppierd, Brit. Journ. Appl. Phys. 4, 70 (1953).

8. A. White, Journ. Appl. Phys. 20, 856 (1949); G. Moore, H. Allison and J. Morrison, Journ. Chem. Phys. 18, 1579 (1950); R. Hughes, P. Coppola and H. Evans, Journ. Appl. Phys. 23, 635 (1952); E. Rittner, Philips. Res. Rep. 8, 184 (1953).

9. J. Beydon, L. Beaudoin, I. Challansonnet et J. Debiesse, C. R. 229, 353 (1949); G. Moore and H. Allison, Phys. Rev. 77, 246 (1950); 78, 354 (1950); J. Debiesse, L’onde électrique 30, 351 (1950); J. Debiesse, J. Challansonnet et G. Neyret, C. R. 232, 602, 2015 (1951); W. Leverton and W. Shepherd, Journ. Appl. Phys. 23, 787 (1952); R. Redington, Phys. Rev. 87, 1066 (1952); R. Beyer, Journ. Appl. Phys. 24, 1008 (1953); J. Affleck and L. Holroyd, Phys. Rev. 89, 893 (1953).

10. H. Morgulis and A. Nagorskii, ZhETF 8, 1159 (1938); M. Romerantz, Phys. Rev. 70, 33 (1946); Journ. Frankl. Inst. 241, 415 (1946); 242, 41 (1946); Proc. I. R. E. 34, 903 (1946); T. Jones, Nature 161, 846 (1948); J. Johnson, Phys. Rev. 69, 903 (1948); 73, 1058 (1948); 82, 576 (1951); 83, 49 (1951); F. Todd and L. Rueger, Phys. Rev. 76, 189 (1949); G. Hess and H. Jacobs, Phys. Rev. 75, 1312 (1949); S. Joshida, J. Takeda and H. Arata, Journ. Phys. Soc. Japan 7, 430 (1952); S. Joshida and J. Takeda, Journ. Phys. Soc. Japan 7, 639 (1952);

J. Woodes and D. Wright, Brit. Journ. Appl. Phys. 3, 323 (1952); 4, 56 (1953). Proc. Phys. Soc. B 66, 1073 (1953).

1. N. Morgulis, ZhETF 16, 959 (1946); T. Arizumi and S. Narita, Journ. Phys. Soc. Japan 6, 118 (1951); D. Wright and J. Woods, Proc. Phys. Soc. B 65, 134 (1952).

2. N. Morgulis, ZhTF 10, 1721 (1940); F. du Pré, R. Hutner and E. Rittner, Phys. Rev. 78, 567 (1950).

3. G. Germann and S. Vagner, see ¹, pp. 79 and 386.

4. T. Kozlyakovskaya, Report at the Second Conference on Semiconductors (1934); G. Tyagunov, Svetotekhnika No. 4 (1934).

5. A. Eisenstein, Journ. Appl. Phys. 17, 654 (1946).

6. R. Loosjes and H. Vink, Philips. Res. Rep. 4, 449 (1949); E. Hensley, Journ. Appl. Phys. 23, 1122 (1952); J. Young, Journ. Appl. Phys. 23, 1129 (1952).

7. T. Berdennikova, Sov. Phys. 2, 77 (1932); R. Jenkins and R. Newton, Journ. Sc. Inst. 26, 172 (1949).

8. M. Benjamin, Phil. Mag. 20, 1 (1935).

9. E. Coomes, Journ. Appl. Phys. 17, 647 (1946).

10. N. Jacobs, G. Hees and W. Croosley, Phys. Rev. 78, 356 (1950); A. Bounds and T. Briggs, Proc. I. R. E. 39, 788 (1951); T. Briggs and C. Richards, ASTM Bulletin No. 171, 66 (1951); R. Forman and R. Rouse, Journ. Res. of Nat. Burst. 46, 30 (1951); H. Poehler, Proc. I. R. E. 40, 190 (1952); J. Lander, H. Kern and A. Beach, Journ. Appl. Phys. 23, 1305 (1952); J. Levy, Proc. I. R. E. 41, 365 (1953).

11. V. Parkhomenko, Proceedings of the Conference on Cathode Electronics, Academy of Sciences of the Ukrainian SSR (1952), p. 59.

12. H. Poehler, Proc. I. R. E. 40, 190 (1952); A. Eisenstein, H. John and I. Affleck, Journ. Appl. Phys. 24, 631 (1953).

13. A. Eisenstein, Journ. Appl. Phys. 20, 776 (1949).

14. N. Morgulis, Collection Dedicated to the 70th Anniversary of Academician A. F. Ioffe, Academy of Sciences of the USSR (1950), p. 146.

15. S. Pekar, Investigations on the Electron Theory of Crystals, Gostekhizdat, 1951.

16. N. Morgulis and K. Tolpygo, ZhETF 25, 718 (1953).

17. S. Pekar and O. Tomasevich, ZhTF 17, 1393 (1947); K. Tolpygo, ZhTF 19, 1301 (1949); N. Karkhanina and N. Morgulis, ZhTF 20, 345 (1950).

18. O. Shade, Proc. I. R. E. 31, 341 (1943); E. Coomes, Journ. Appl. Phys. 17, 647 (1946); D. Wright, Proc. Roy. Soc. A 190, 394 (1947); W. Ramsay and W. Danforth, Phys. Rev. 73, 1246 (1948); D. Fuks et al., Magnetrons. Sov. Radio (1948), p. 239; W. Danforth and D. Goldwater, Journ. Appl. Phys. 20, 163 (1949); A. Eisenstein, Journ. Appl. Phys. 20, 776 (1949).

19. F. Horak, Journ. Appl. Phys. 23, 346 (1952).

20. R. Sproull, Phys. Rev. 67, 166 (1945); S. Kalashnikov, G. Kubetskii and V. Osher, ZhTF 16, 1369 (1946); S. Wagener, Proc. Phys. Soc. 61, 521 (1948); A. Eisenstein, see ¹, p. 53; S. Pticyn, see ¹, p. 58; G. Feaster, Journ. Appl. Phys. 20, 415 (1949); D. Wright, Proc. Phys. Soc. B. 62, 188 (1949).

21. G. Metson, S. Wagener, M. Holmes and M. Child, Proc. I. E. E. III, 99, 69 (1952).

22. R. Loosjes, H. Vink and C. Jansen, Journ. Appl. Phys. 21, 350 (1950); Philips. Techn. Rev. 13, 337 (1952); C. Jansen and R. Loosjes, Philips. Res. Rep. 8, 21 (1953).

23. S. Levitin, ZhTF 23, 1700, 2166 (1953).

24. M. Weinreich, Rev. Gen. d’électr. 54, 243 (1945); Journ. Appl. Phys.

20, 1256 (1949); D. Wright, Nature 160, 129 (1947); Proc. Phys. Soc. B 62, 188 (1949); T. Hanley, Journ. Appl. Phys. 19, 583 (1948); W. Danforth and F. Morgan, Phys. Rev. 79, 142 (1950); W. Danforth, Journ. Frankl. Inst. 251, 515 (1951); Phys. Rev. 86, 416 (1952); G. Mesnard, C. R. 232, 1744 (1951); 236, 904 (1953); Journ. de phys. 14, 179 (1953); Le vide 8, 1273, 1377 (1953).

1. J. Lafferty, Journ. Appl. Phys. 22, 299 (1951).

2. W. Grattidge and H. John, Journ. Appl. Phys. 23, 1145 (1952).

3. V. Espe and M. Knoll, Technology of Electrovacuum Materials. Oborongiz, 1939, p. 257; D. Fisk et al., Magnetrons. Sov. Radio (1948), p. 254; R. Hughes and P. Coppola, Journ. Appl. Phys. 23, 1261 (1952); D. Mac Nair, R. Lynch and N. Hannay, Journ. Appl. Phys. 24, 1335 (1953).

4. O. Bush, R. Vandegrift and T. Hanley, Journ. Appl. Phys. 20, 295 (1949); H. Fan, Journ. Appl. Phys. 20, 682 (1949); G. Espersen, Journ. Appl. Phys. 21, 261 (1950); L. Cronin, Chem. Eng. 57, 276 (1950); Am. Ceram. Soc. Bul. 30, 234 (1951).