PROPERTIES OF OXIDE CATHODES1

John P. Blewett

EDITOR’S PREFACE

The development of electronic devices and, in particular, of electron tubes—whose production in recent times, after passing through the serial-production stage, has become one of the mass-production industries—has been made possible by the considerable improvement of the technology of oxide cathodes, both indirectly heated and directly heated; their progenitor was the Wehnelt cathode, published by him as early as 1904.

In recent years the Soviet electrovacuum industry has successfully mastered the mass production of a number of series of both receiving-amplifying and generator tubes with oxide cathodes, as well as of a number of other electrovacuum devices with oxide cathodes, such as gas-filled tubes, thyratrons, cathode oscillographs, television tubes, etc.

Unfortunately, both in the Russian and in the foreign literature, up to the present time—if one does not count the outdated reviews by Deshman, Reimann (chapter), and Schottky—there have been neither good handbooks nor even review articles devoted to questions of both the technology and the theory of oxide cathodes. The translation offered to readers of the review article by Blewett, a collaborator of the General Electric Company (G. E. C.—America), is the first work that fills this gap, though only in part. In addition, this review, supplied with a sufficiently complete bibliography, will help the workers of our institutes and factories to find their way more easily through the existing literature, which is widely scattered among many journals and often gives contradictory results on one and the same question connected with the operation of oxide cathodes.

It is necessary, however, to point out a number of points either not touched upon by the author or mentioned by him only in passing. Thus, for example, Blewett speaks only of attempts, “crowned in individual cases with success,” to use barium azide, from which barium is deposited on oxidized metallic surfaces. This, properly speaking, is that variety of oxide cathode which, in the form of barium distillation cathodes, has found wide application in so-called barium tubes, the mass production of which exists

PROPERTIES OF OXIDE CATHODES

as in our Union, so also at a number of European firms (for example, Telefunken and Philips). In this case, as the source of barium use is made not only of barium azide, which has a major shortcoming, namely a low decomposition temperature (\(\sim 120^\circ\)), but chiefly either of metallic barium in a copper or nickel envelope (Ni—Ba and Cu—Ba, see Espe and Knoll\(^{18}\)), or of thermite pellets consisting of a mixture of barium aluminates (which the author mentions only in connection with Hull’s cathode), barium oxide, barium peroxide, and aluminum dust (see also Espe and Knoll\(^{18}\)).

Despite the development of oxide cathodes and the considerable improvement of their mass-production technology, work is still proceeding in parallel on barium lamps with an economical cathode.

In connection with Blewett’s mention of a new type of oxide cathode for gasotrons, developed by Hull, it is necessary to point out the development of similar cathodes in our Union (see Shaposhnikov\(^{13,17}\)) as early as 1932–1933 and their application in gasotrons. In this connection, some additions have been made in the translation.

Moreover, in certain places, in particular in the table of properties of the alkaline-earth metals, obsolete data on melting temperatures and work functions have been replaced in the translation by newer and more accurate data.

The bibliography placed at the end of the article has been supplemented by a number of works that appeared in our journals at different times.

Moscow
B. M. Tsarev

1. INTRODUCTION

More than a third of a century has passed since Wehnelt\(^{1}\) published (1904) his observations, according to which a filament coated with an oxide of an alkaline-earth metal is a good source of thermoelectrons already at comparatively low temperatures.

The unceasing stream of publications devoted to this question since that time can characterize the technical significance of this discovery. And yet, despite the large number of carefully performed experiments, the mechanism that leads to the activation of the oxide cathode is at present still far from completely understood. The influence of unforeseen factors has led to the accumulation of a large quantity of contradictory data and to the general conclusion that the mechanism of activation of the oxide cathode is a complex process.

We shall operate chiefly with facts obtained during the last decade. During this time several review papers have appeared, the most complete of which are the works of Schottky, Rothe, and Simon\(^{7}\), Dechman\(^{9}\), Goertz\(^{11}\), Reimann\(^{13}\), and de Boer\(^{14}\). In these articles the most essential references to earlier works can be found.

The present article begins with a brief survey of the technology of oxide cathodes and then proceeds to the discussion of various particular questions.

In this work, instead of writing “barium” and implying thereby either barium, or strontium, or calcium, or any …

Table 1

Physicochemical properties of alkaline-earth metals

	Ba	Sr	Ca
Atomic weight	137.36	87.63	40.08
Melting temperature (Goffman and Schulze, 1935)	977° K	1 044° K	1 124° K
Heat of sublimation in kg cal	44.5 (Van Limpt, 1936)	37.8 (heat of vaporization, Kelly, 1935)	42.9 (Kelly, 1935)
Vapor pressure: \(\lg p\) (\(p\) in mm Hg)	\(7.83-\dfrac{9727}{T}\) (Van Limpt, 1936)	\(11.98-\dfrac{8250}{T}-\) \(-1.27\lg T\) (Kelly, 1935)	\(10.78-\dfrac{9350}{T}-\) \(-0.66\lg T-\) \(-1.5\cdot10^{-4}T\) (Kelly, 1935)
Rate of evaporation \(\lg M\) (\(M\) in g/cm² sec, calculated from vapor pressure)	\(7.67-\dfrac{9727}{T}-\) \(-0.5\lg T\)	\(11.72-\dfrac{8250}{T}-\) \(-1.77\lg T\)	\(10.35-\dfrac{9350}{T}-\) \(-1.16\lg T-\) \(-1.5\cdot10^{-4}T\)
Ionization potential in eV	5.19	5.67	6.09
Work function in eV	2.52 (Anderson, 1938; cf. also Cashman and Besse, 1939)	2.1 ? (Becker, 1935)	2.7 (Jameson and Cashman, 1936)

their combination, as was customary earlier, we shall use the symbol \((\mathrm{BaSrCa})\), which will mean “barium, or strontium, or calcium.” Correspondingly, the symbol \((\mathrm{BaSr})\) will mean “barium or strontium,” etc. In this way we avoid the possibility of attributing the results of experiments to conditions under which they were not verified.

Tables 1 and 2 give a brief summary of those properties of \((\mathrm{BaSrCa})\) and their oxides that are of special interest in connection with oxide cathodes.

2. BRIEF REVIEW OF THE TECHNIQUE OF OXIDE CATHODES

The oxides of alkaline-earth metals are chemically unstable in the presence of air; therefore oxide cathodes are usually prepared by applying alkaline-earth carbonates, nitrates, or hydroxides to a metallic core.

Table 2

Physicochemical properties of alkaline-earth oxides

	BaO	SrO	CaO
Molecular weight	153.36	103.63	56.08
Melting point (Wohmacher, 1926)	2,196°K	2,703°K	2,845°K
Heat of sublimation, in kg cal	90 (Klassen and Venemans, 1933)	117 (see section 3, d)	128 (see section 3, d)
Vapor pressure \(\lg p\) (\(p\) in mm Hg)	\(8.87-\dfrac{19700}{T}\) (Klassen and Venemans, 1933)	\(10.00-\dfrac{25500}{T}\) (see section 3, d)	\(10.40-\dfrac{2800}{T}\) (see section 3, d)
Evaporation rate \(\lg M\) (\(M\) in g/cm²·sec, calculated from vapor pressure)	\(8.73-\dfrac{19700}{T}-\) \(-0.5\lg T\)	\(9.77-\dfrac{25500}{T}-\) \(-0.5\lg T\)	\(10.04-\dfrac{28000}{T}-\) \(-0.5\lg T\)
Heat of formation, in kg cal	133	141	152
Heat of dissociation \(^{1}\) of vapors, in kg cal	147	\(\sim 125\)	\(\sim 126\)

Such a cathode is then heated in vacuum in order to convert these compounds into oxides; the gases evolved in the process are pumped off, after which the cathode is “activated.” Activation may consist of further heating, the application of an anode voltage for drawing anode current, bombardment in a gas discharge, and other similar processes. As a result, a cathode is obtained which, at a temperature of 1,000°K, emits electrons as intensely as a tungsten filament at a temperature of 2,300°K. The dependence of the emission current of an ordinary oxide cathode on temperature is given in Fig. 1, where, for comparison, the corresponding curves for tungsten, thoriated tungsten, and cathodes with a thorium coating are plotted. Along with its low operating temperature, the oxide cathode possesses high efficiency. An oxide cathode operating at 1,000°K emits approximately 100 mA from 1 cm² with an efficiency of the order of 20 mA/W.

\(^{1}\) The heat of dissociation of the vapors was calculated from a thermochemical cycle including the heat of sublimation of the oxide, the heat of dissociation of the oxide vapor, the heat of sublimation of the metal, the heat of dissociation of \(\mathrm{O}_2\) (118 kg cal), and the heat of formation of the oxide.

A tungsten cathode at \(2300^\circ\mathrm{K}\) gives approximately the same emission, but its efficiency is less than \(1\ \mathrm{mA/W}\). The high efficiency of oxide cathodes has led to their wide use in electronic tubes. The oxide coating is especially valuable in those cases where the cathode must have an equipotential surface and therefore cannot be heated by direct heating. Such an indirectly heated cathode usually takes the form of a hollow metallic cylinder with a coating applied on the outside, heated from within by an insulated tungsten filament.

Fig. 1. Thermionic emission of various cathodes

1 — oxide coating, 2 — thoriated coating, 3 — thoriated tungsten, 4 — tungsten

The useful service life of an oxide cathode operating at a temperature of \(1000^\circ\mathrm{K}\) is usually several thousand hours. At the end of this period the emission at once drops sharply to a very small value. In this case the disappearance of the oxide coating is sometimes observed, while sometimes the external appearance of the cathode remains almost the same as at the beginning of its service life.

3. THERMAL PROPERTIES OF THE OXIDES OF ALKALINE-EARTH METALS

a) Temperature measurement

The abrupt changes in most properties of oxides with changes in temperature make it necessary to strive for as accurate a determination of the latter as possible. The thermal conductivity of oxides is considerably lower than the thermal conductivity of metals. Therefore methods of temperature measurement based on bringing a thermocouple or resistance thermometer into contact with the oxide surface are associated with excessive cooling of the points of contact and lead to significant errors. For this reason, in determining the temperature of an oxide surface, methods based on the use of the radiative properties of the cathode are usually employed. The following three standard methods are the most commonly used.

1. Direct pyrometric measurement of the temperature of the coating surface. In optical pyrometry a substantial correction must be introduced for the spectral emissivity of the coating. For pyrometry by total radiation it is necessary to know the total emissivity.

2. Measurement of the heating power. If we know the total emissivity, then we may write:

\[ \text{Heating power per } 1\ \mathrm{cm}^2 \text{ of surface} = e\sigma T^4, \tag{3,1} \]

where \(e\) is the total emissivity, \(\sigma\) is Stefan’s constant, equal to \(5.72\cdot 10^{-5}\ \mathrm{erg}/\mathrm{cm}^{2}\ \mathrm{sec}\ \mathrm{deg}^{4}\).

As we shall see below, \(e\) is also not a constant, but a function of temperature.

1. Determination of the temperature of the oxide surface from the temperature of the metallic core. The temperature of the cathode core can readily be determined either by direct pyrometric measurement of the oxide-free portions of its surface, or by measuring its electrical resistance.

Table 3 gives brightness temperatures and specific resistivities as functions of the true temperature for several metals commonly used as core material. The temperature \(T\) of the oxide surface can now be calculated from the balance of supplied and radiated heat

\[ \frac{k}{d}(T' - T)=e\sigma T^{4}, \tag{3.2} \]

where \(T'\) is the temperature of the core, \(k\) is the thermal conductivity of the oxide, \(d\) is the thickness of the oxide layer, \(e\) is the total emissivity, and \(\sigma\) is Stefan’s constant.

The difference \(T' - T\), i.e., the temperature drop in the oxide layer, may reach one hundred or more degrees, so that in no case can one confidently assume a priori that the temperature of the oxide surface is equal to the temperature of the core.

b) Emissivity

All three methods of temperature determination mentioned above require knowledge of one or another radiation coefficient. Unpublished measurements made in the GEC laboratory show that the spectral emissivity for \(\lambda = 0.665\mu\) (used in optical pyrometry) lies between 0.5 and 0.7 for thin coatings of (BaSr) oxides on metallic cores. As a result, a correction of the order of \(30^\circ\) at \(1100^\circ\mathrm{K}\) is necessary in order to obtain the true temperature from the temperature read on the pyrometer.

The spectral emissivity \(e_\lambda\) can be determined from the spectral reflection coefficient \(r_\lambda\) by the relation

\[ e_\lambda + r_\lambda = 1. \]

Prescott and Morrison\(^{18}\) described a convenient method for measuring reflection coefficients. The cathode is placed in a chamber whose walls are covered with a diffusely scattering material (white velvet). Inside, the chamber is illuminated by several incandescent lamps. The cathode and the portion of the wall located behind it are observed with an optical pyrometer. If \(S\) and \(T\) are the apparent temperatures of the cathode and background, measured with the pyrometer, then the reflection coefficient can be found from the following relation, based on Wien’s formula:

\[ r_\lambda=\exp\left[\frac{C_2}{\lambda}\left(\frac{1}{T}-\frac{1}{S}\right)\right], \]

where \(C_2 = 1.432\ \mathrm{cm\ deg}\).

JOHN P. BLEWETT

Table 3

Melting temperatures, brightness temperatures, and specific resistivities of the most commonly used core materials

| \(T^\circ K\) | \multicolumn{5}{c}{Brightness temperature in \(^\circ K\) \(^{1)}\)} | \multicolumn{3}{c}{Specific resistivity in ohm·cm \(10^6\) \(^{2)}\)} |
|---:|---:|---:|---:|---:|---:|---:|---:|---:|
| | W | Mo | Ta | Pt | Ni | W | Mo | Ta |
| 273 | — | — | — | — | — | 5,00 | 5,14 | 12,56 |
| 300 | — | — | — | — | — | 5,65 | 5,78 | 13,85 |
| 400 | — | — | — | — | — | 8,06 | 8,15 | — |
| 500 | — | — | — | — | — | 10,56 | — | — |
| 600 | — | — | — | — | — | 13,23 | — | — |
| 700 | — | — | — | — | — | 16,09 | — | — |
| 800 | — | — | — | — | — | 19,00 | — | — |
| 900 | — | — | — | — | — | 21,94 | — | — |
| 1 000 | 966 | 958 | 967 | 950 | 956 | 24,93 | 23,9 | 44,1 |
| 1 100 | 1 058 | 1 049 | 1 060 | 1 037 | 1 047 | 27,94 | — | 47,3 |
| 1 200 | 1 149 | 1 139 | 1 152 | 1 124 | 1 137 | 30,98 | 29,5 | 51,0 |
| 1 300 | 1 240 | 1 228 | 1 242 | 1 211 | 1 226 | 34,08 | — | 54,8 |
| 1 400 | 1 330 | 1 316 | 1 332 | 1 296 | 1 315 | 37,19 | 35,2 | 59,0 |
| 1 500 | 1 420 | 1 403 | 1 421 | 1 381 | 1 403 | 40,36 | — | 62,4 |
| 1 600 | 1 509 | 1 489 | 1 508 | 1 466 | — | 43,55 | 41,1 | 65,8 |
| 1 700 | 1 597 | 1 574 | 1 596 | 1 551 | — | 46,78 | — | 69,3 |
| 1 800 | 1 684 | 1 658 | 1 682 | 1 634 | — | 50,05 | 47,0 | 72,5 |
| 1 900 | 1 771 | 1 741 | 1 767 | 1 717 | — | 53,35 | — | 75,8 |
| 2 000 | 1 857 | 1 824 | 1 852 | — | — | 56,67 | 53,1 | 78,9 |
| 2 100 | 1 943 | 1 905 | 1 933 | — | — | 60,06 | — | 82,0 |
| 2 200 | 2 026 | 1 986 | 2 018 | — | — | 63,48 | 59,2 | 85,2 |
| 2 300 | 2 109 | 2 065 | 2 099 | — | — | 66,91 | — | 88,3 |
| 2 400 | 2 192 | 2 143 | 2 181 | — | — | 70,39 | 65,5 | 91,3 |
| 2 500 | 2 274 | 2 220 | 2 261 | — | — | 73,91 | — | 94,4 |
| 2 600 | 2 356 | 2 297 | 2 341 | — | — | 77,49 | 71,8 | 97,4 |
| 2 700 | 2 437 | 2 373 | 2 421 | — | — | 81,04 | — | 100,2 |
| 2 800 | 2 516 | 2 448 | 2 499 | — | — | 84,70 | — | 102,9 |
| 2 900 | 2 595 | — | 2 575 | — | — | 88,33 | — | 105,6 |
| 3 000 | 2 673 | — | 2 652 | — | — | 92,04 | — | 108,7 |
| 3 200 | 2 827 | — | 2 803 | — | — | 99,54 | — | 113,9 |
| 3 400 | 2 978 | — | — | — | — | 107,2 | — | — |
| 3 600 | 3 165 | — | — | — | — | 115,0 | — | — |
| Melting temperature, \(^\circ K\) . . | 3 653 | 2 895 | 3 269 | 2 046 | 1 725 | — | — | — |

\(^{1)}\) Brightness temperatures: W—Forsythe and Worthing, Astrophys. J., 61, 146, 1925; Mo—Worthing, Phys. Rev., 28, 190, 1926; Ta—Malter and Langmuir, Phys. Rev., 55, 743, 1939; Pt and Ni—Forsythe, Int. Crit. Tab., Vol. V., p. 245.

\(^{2)}\) Specific resistivities: W—Jones and Langmuir, Gen. Elec. Rev., 30, 354, 1927; Mo—Worthing, Phys. Rev., 28, 190, 1926; Ta—Malter and Langmuir, Phys. Rev., 55, 743, 1939.

If the cathode is heated, then an additional term must be introduced into this formula, taking into account the fact that the cathode not only reflects but also emits light of the given wavelength \(\lambda\).

Unfortunately, Prescott and Morrison applied this method only to a special type of cathode, in which the oxide coating was mixed with fine nickel powder. Consequently, the value \(0.64\) obtained by them for the spectral emissive power can hardly be especially widely applicable to other types of oxide.

For the value of the integral emissive power there are a number of different data.

Dechman\(^9\) refers to the following relation between emissive power and temperature, used by Western El. Co. for its electron tubes:

\[ e = 0.4 + 2.5 \cdot 10^{-4}T \]

(for temperatures \(T\) lying in the range from 800 to \(1200^\circ\mathrm{K}\)).

This formula is applicable to coatings made of a mixture of barium and strontium oxides on cores consisting of \(95\%\mathrm{Pt}\) and \(5\%\mathrm{Ni}\).

Patai and Tomaschek\(^ {15}\) measured the integral emissive power of cathodes made by depositing a colloidal suspension and obtained the value \(e = 0.5\). This value was obtained from observations of the change in surface temperature of an oxide cathode when a gas of known thermal conductivity was introduced into the tube.

Fig. 2

Dependence on temperature of the integral radiation in watts from \(1\ \mathrm{cm}^2\) for BaO—SrO (59%—molecular content of SrO) on a nickel core. On the right the coating thicknesses in microns are indicated (Clausing and Ludwig\(^ {12}\)).

The most complete of all published investigations of the emissive power of oxides (BaSr) was carried out by Clausing and Ludwig\(^ {12}\), who studied coatings of various thicknesses of SrO and BaO—SrO on nickel strips. In these experiments the entire experimental tube, made of quartz, was heated to the temperature of the cathode, whereby it was possible to avoid a temperature drop in the oxide coating.

Clausing and Ludwig showed that in general the integral radiation can be described by an expression of the form

\[ \eta = pT^q, \tag{3,3} \]

where the quantity \(p\) lies in the interval from \(10^{-3}\) to \(10^{-8}\ \mathrm{erg}/\mathrm{cm}^2\,\mathrm{sec}\,\mathrm{deg}^q\), and \(q\) varies within the limits from 3.3 to 4.9 depending on the thickness of the oxide layer. The results of some of the measurements of the quantity \(\eta\) as a function of temperature are shown in Fig. 2. If we plot the ratio of the integral radiation to the radiation of a black body at the same

same temperature (i.e., integral emissivity) as a function of the coating thickness, the results give the curve shown in Fig. 3. From the value (at zero coating thickness) characterizing pure nickel, the emissivity increases to a maximum equal to 0.38 at a layer thickness of approximately 75 μ (weight of carbonate coating 1.0 mg/cm²), and then decreases until, at a thickness of 300 μ and more, it settles at a certain constant value equal to approximately 0.2.

Fig. 3

Dependence of the integral emissivity of BaO + SrO on the coating thickness \((T = 1100^\circ K)\) (Clausing and Ludwig¹²)

The same authors carried out measurements of the spectral emissivity \((\lambda = 0.665\,\mu)\), which gave, for thin coatings (of the order of 30 μ), a value from 0.5 to 0.6. With increasing thickness the spectral emissivity drops sharply to a value of 0.1 for coatings thicker than 150 μ.

From a comparison of the experimental data we can establish, with greater or lesser confidence, that ordinary oxide cathodes with a coating thickness of the order of \(4 \cdot 10^{-2}\) cm (which corresponds to approximately 4 mg/cm² of oxide) are characterized by the following values: spectral emissivity \(\sim 0.5\) for \(\lambda = 0.665\,\mu\); integral emissivity \(\sim 0.3\).

c) Thermal conductivity

If the emissivity of the oxide coating is known, it is sufficient to determine the heating power and the core temperature in order to calculate the thermal conductivity of the coating. The temperature of the oxide surface can be found from the equation (3,1) given above. Substituting this value and the known temperature \(T'\) of the core into equation (3,2), we can calculate the thermal conductivity \(k\).

Patai and Tomaschek¹⁵ give, for a rather dense coating obtained by depositing a colloidal suspension, a value of the thermal conductivity lying between 0.00012 and 0.0013 cal/deg·sec cm². Clausing and Ludwig¹² give only one value, equal to 0.00034 cal/deg·sec cm². There are no data on the variation of this quantity with temperature.

d) Rate of evaporation and vapor pressure

The lifetime of a cathode is often determined by evaporation of its oxide coating. It is therefore not surprising that a whole series of investigations (Arnold³, Thompson and Armstrong⁷, Rheerink (see Zwikker⁷), Claassen and Venemans¹²) was devoted to the rate of evaporation of oxides of alkaline-earth metals. The best measurements were carried out with pure BaO, since it has a vapor pressure approximately

in 1,000 times greater than SrO and CaO. For some time, the measurements made by Classen and Venemans were considered the most accurate for BaO. However, at present these results are disputed by Germann^16. Their observations agree at 1,400° K and diverge in the regions lying above and below this point. Germann’s curve corresponds to a heat of evaporation of 119 kg cal, whereas according to Classen and Venemans this quantity is only 90 kg cal. The stumbling block in measurements of this type is usually the measurement of temperature. Classen and Venemans evaporated the oxide from inside a platinum cylinder, heated by a high-frequency current, onto a cooled quartz surface.

The temperature of the outer surface of the platinum cylinder was determined pyrometrically; this temperature was also taken as the temperature of the oxide surface. Such an assumption can lead to serious errors in the presence of a temperature gradient in the oxide layer.

Germann attempted to circumvent this difficulty and pyrometered a spot of Cr₂O₃ deposited on the oxide surface. The emissive power of Cr₂O₃ is known; it is equal to 0.8 and does not change with temperature. This method, however, meets with its objections (cf. Blewett, Liebhafsky, and Hennelly^18), since the Cr₂O₃ spot, owing to its higher emissive power, is cooled below the temperature of the oxide.

Since all these considerations do not permit reliable conclusions to be drawn from the contradictory data obtained, Blewett, Liebhafsky, and Hennelly^18 applied the Knudsen method. Within the limits of experimental error their results confirm the values obtained by Classen and Venemans. The rate of evaporation of BaO is given by the equation:

\[ \lg m\sqrt{T}=-\frac{19400}{T}+8.48 \qquad (m\text{—in } \mathrm{g}/\mathrm{cm}^{2}\,\mathrm{sec}). \]

With the aid of the kinetic-theory equation

\[ p=m\left(\frac{2\pi RT}{M}\right)^{\frac{1}{2}} \]

the vapor pressure of BaO can be expressed as follows:

\[ \lg p=-\frac{19400}{T}+8.63 \qquad (p\text{—in mm Hg}). \]

Measurements of the evaporation rates of SrO and CaO were made by Classen and Venemans^12 using the same apparatus as for BaO. The results of the measurements performed give a rather considerable scatter of points, which is explained by the low, barely measurable evaporation rates.

However, the value of the heat of evaporation can be determined with considerable reliability even from a single measurement of the vapor pressure by means of the following thermodynamic relation:

The change in free energy upon evaporation of a solid is determined by the expression:

\[ \Delta F=-RT\ln p=\Delta H-T\cdot \Delta S, \]

where \(p\) is the vapor pressure, \(\Delta H\) is the heat of sublimation, and \(\Delta S\) is the change in entropy.

Using the third law of thermodynamics, the entropy values given by Kelley\(^{14}\) (Bull. 394), and the assumption that the difference in specific heats for the solid and gaseous states is equal to \(3R\) (cf. Meyer and Wintner\(^{17}\)), it proves possible to calculate \(\Delta S\), and consequently also \(\Delta H\).

We then obtain:

for SrO

\[ \Delta H=117\ \text{kg cal}, \]

for CaO

\[ \Delta H=128\ \text{kg cal}, \]

with a possible error of the order of 8%. In Fig. 4 the experimental points for the evaporation of SrO and CaO obtained by Klassen and Veenemans are plotted; however, instead of the lines drawn by the authors themselves as the best corresponding to their experiment, the figure shows lines corresponding to the calculated values of the heats of evaporation.

From a mixture of SrO and BaO, predominantly BaO evaporates, so that the surface layer remains composed of SrO alone (Klassen and Veenemans\(^{12}\)). This was also confirmed by electronographic studies by Gerthner\(^{14}\) and Derbichair\(^{17}\). The properties of mixtures of oxides are considered below, in Section 6.

Fig. 4

a — vapor pressures of alkaline-earth oxides; b — evaporation rates of alkaline-earth oxides (Klassen and Veenemans\(^{12}\))

Properties of oxide mixtures

4. ELECTRICAL PROPERTIES OF ALKALINE-EARTH METAL OXIDES

a) Electrical conductivity; theory of electrical conductivity and thermionic emission

Some idea of the confusion that exists regarding electrical conductivity can be obtained from Fig. 5, which presents literature data on the electrical conductivity of BaO or mixtures of BaO and SrO. However, taking into account certain factors, these

PROPERTIES OF OXIDE CATHODES

the results can be put somewhat in order. The early measurements of Horton² and the measurements of Spanner⁴ were made in air and therefore probably refer to a mixture of BaO with other Ba compounds. The measurements of Becker¹⁰, Reimann and his co-workers⁹˒¹⁰, and Klausing (see De Boer’s book, reference 2 on p. 285 of the 1936 Russian edition) were carried out with oxide coatings on filaments in the presence of large temperature gradients. Temperature drops, especially at low temperatures, lead to considerable polarization effects, which can distort the results obtained. Frequently observed significant deviations from Ohm’s law, i.e., dependence of the conductivity on the applied voltage, are also found.

Krocek and Lübcke⁹ used an ingenious method for determining the conductivity of an oxide layer, making use of the thermal equilibrium between the heating effect of the conductivity current in the oxide layer and the cooling effect of the emission current. The conductivity values obtained by them show a very large scatter. This apparently can be explained by the fact that different samples were prepared in different ways. As a result, the curve designated by their names in Fig. 5 conveys only the qualitative course of the results obtained. Blewitt’s measurements were carried out with nonactivated BaO in weakly compressed volumes. The measurements of Meyer and Schmitt¹¹ were made with a thin layer of BaO clamped between two nickel blocks maintained at the same temperature. Polarization disturbances were eliminated by using a high-frequency current. Thus, these measurements are apparently the best of those that could have been performed. It should be noted, however, that one cannot assign a definite value of the specific resistance to a substance such as BaO, which has not been prepared in the form of a large single crystal. The value of the conductivity is meaningful only when the particle sizes, the degree of compression of the powder, and the amount of impurities contained in it are known exactly. Krocek and Lübcke observed a decrease in the conductivity of barium oxide with increasing particle size, indicating that surface conductivity accounts for a considerable share of the total conductivity.

Fig. 5. Electrical conductivity of BaO and BaO + SrO

1 — activated BaO (Meyer and Schmitt); 2 — nonactivated BaO (Meyer and Schmitt); 3 — electrical conductivity of the cathode oxide coating. The line has been drawn through a large number of widely scattered points obtained in measurements with various cathodes (Krocek and Lübcke⁹); 4 — electrical conductivity of a BaO rod in air (Spanner⁴); 5 — activated BaO on the cathode (Klausing, see De Boer’s book, pp. 285—286); 6 — electrical conductivity of BaO on the cathode after 1,000 hours of operation. The source does not indicate in what temperature interval the measurements were carried out; applies to curves 5 and 6 (Klausing, see De Boer’s book, pp. 285—286); 7 — electrical conductivity of a weakly compressed BaO rod in vacuum (Blewitt, unpublished); 8 — electrical conductivity of a BaO + SrO coating on a filament. The source does not indicate the thickness of the oxide layer; when drawing the line it was assumed that it was equal to 0.5 mm (Reimann and Treloar¹⁰); 9 — coating of BaO + SrO on a cathode (Becker and Sears, see Becker¹⁰); 10 — electrical conductivity of a BaO rod in air (Horton²)

Little can be said about the electrical conductivity of SrO and CaO. Measurements with these oxides (Horton² and Schpanner⁴) raise many objections, since they were carried out in air. They nevertheless show that the conductivity of SrO is an order of magnitude lower than the conductivity of BaO, and that of CaO is still lower.

Measurements by Meyer and Schmitt¹¹ showed that the conductivity of barium oxide in the activated state is considerably greater than in the unactivated state. It is assumed that during activation of barium oxide an excess of free barium arises, and the resulting increase in electrical conductivity is analogous to that observed in other semiconductors in the presence of a stoichiometric excess of their electropositive component.

The question of the nature of the electrical conductivity has been discussed by many authors. The rate of activation of an oxide cathode usually increases when electron current is drawn. This suggests that an ionic current flows in the oxide layer, leading to the deposition of free (BaSrCa) on the surface of the core. Harris (see Becker⁸) found barium and strontium that had in fact penetrated into the core of an oxide cathode from which a strong current had been drawn. In those cases, however, where the cathode had merely been heated without current being drawn, no barium or strontium was found in its core. If, on the other hand, the direction of the current in the oxide layer is changed by sending electrons to the cathode from an external source, electrolysis should lead to the appearance of a layer of free barium on the surface of the oxide. In this case the same changes in cathode activity should be observed as when barium is deposited from an external source. Becker⁸ and Becker and Sears¹⁰ showed that this does in fact occur. They were also able to show that the amount of barium present on the surface is proportional to the charge that has passed through the oxide layer. This fact is illustrated by the curve of the dependence of the active cathode on the magnitude of the charge passed. For very different current strengths, values were obtained that lay on a single curve. From these measurements Becker⁸ concludes that approximately \(0.4\%\, I/i_0\) of the total current at \(800^\circ\text{K}\) is transported by Ba ions. This value is connected with the assumption that activation is determined by the presence of a monatomic layer of Ba on the surface of the oxide and with a very rough estimate of the total area of the oxide surface.

Berdennikova¹¹ carried out chemical measurements of the total amount of deposited pure Ba [for the method used, see Section 8, a)] and concluded that, under the given conditions, Faraday’s first law is valid and that the ratio of the barium-ion current to the total current is \(5\cdot 10^{-5}\) at \(1273^\circ\text{K}\).

During the process of activation of an oxide cathode, a considerable amount of oxygen is released. The evolution of oxygen was established by the experiments of Detels¹⁶, who identified this gas by spectroscopic investigation in a Geissler tube, and by Becker⁸, who detected oxygen by its effect on electron emission from a nearby tungsten cathode. Eisenweel⁶ carried out a chemical determination of the rate of oxygen evolution

in the process of electrolytic activation of the oxide coating on a platinum filament. On the basis of his measurements he concludes that, at the beginning of activation of a cathode coated with a mixture of alkaline-earth-metal oxides, approximately \(0.2\%\) of the total current in the oxide layer is carried by oxygen ions. This value falls below \(0.001\%\) when the cathode is fully activated. In a coating of barium oxide alone, the current of oxygen ions remains unchanged and is of the order of \(0.05\%\) of the total current. Pure strontium oxide gave no noticeable electrolytic evolution of oxygen at all. Becker also observed that after full activation the rate of oxygen evolution from a cathode with a mixed oxide coating became considerably smaller. He also found that oxygen evolution decreased when the current drawn was limited by space charge, i.e., when there was no electric field at the oxide surface. The potential of the auxiliary tungsten cathode did not affect the rate of oxygen deposition, from which Becker concludes that oxygen is evolved in the form of neutral atoms or molecules. However, mass-spectrometric investigation showed that a small amount of oxygen is evolved in the form of negative ions [see Section 4, c)]. The dependence of oxygen evolution on the presence of a field at the surface shows that this process is not simply electrolytic, but more complex. In view of the great electron affinity of oxygen, it is difficult to imagine why oxygen ions, upon approaching the surface, give up their extra electrons and evaporate in the form of neutral atoms. If, however, the atoms combine and form a molecule, then the necessary energy may be obtained at the expense of the heat of association. But it seems highly improbable that two negative ions could come sufficiently close to one another. It is possible that the presence of an electric field facilitates this process.

From the data cited it is evident that the electrolytic part of the conductivity is comparatively small and noticeable only during the activation process, as a result of which the oxide may be regarded as an electronic semiconductor.

In semiconductors such as barium oxide, the number of electrons producing conductivity or thermionic emission is small compared with the total number of atoms. This simplifies the theoretical treatment of the question, since the electron distribution given by Fermi statistics is practically reduced to the classical distribution.

Alkaline-earth oxides belong to the class of “semiconductors with impurities.” The impurity in the present case consists of dispersed atoms of the free alkaline metal. These impurity atoms possess localized energy levels in the forbidden region between the filled electron band and the conduction levels. As is seen from Fig. 6, the transition of an electron from an impurity level into the conduction band requires a certain energy \(Q_1\), which is less than the energy \(Q_0\) required to transfer an electron from the filled band. In the temperature range of interest to us, it should be expected that the number of electron transitions from impurity levels ...

...the impurity so greatly exceeds the number of transitions from the filled band that, in the calculation, one may confine oneself to considering only transitions of the first type.

Let us turn to the calculation of \(n_c\)—the number of electrons in the conduction band at a temperature \(T^\circ\mathrm{K}\). The number \(n_1\) of electrons passing per unit time into the conduction band from impurity levels will be proportional to the number of impurity atoms and to the Boltzmann factor

\[ n_1 \sim N_1 \exp\left(-\frac{Q_1}{kT}\right). \]

The number of electrons \(n_2\) returning per unit time from the conduction band will be proportional to the number \(n_c\) of electrons in the conduction band and to the number of free impurity levels, which is equal to \(n_c\):

\[ n_2 \sim n_c^2. \]

In the state of equilibrium \(n_2=n_1\); hence

\[ n_c^2 \sim N_1 \exp\left(-\frac{Q_1}{kT}\right) \]

and

\[ n_c \sim N_1^{\frac12}\exp\left(-\frac{Q_1}{2kT}\right). \tag{4,1} \]

Fig. 6. Distribution of energy levels for a semiconductor with an impurity

\(1\)—allowed band (conduction band), \(2\)—forbidden band, \(3\)—localized impurity level, \(4\)—filled band

Thermionic emission from the semiconductor will now occur at the expense of those electrons in the conduction band which have sufficient energy to overcome the potential barrier at the surface of the crystal. Let the height of this barrier be \(\varphi\) (see Fig. 6). This means that \(\varphi\) is the difference of the potential energies of an electron in space immediately outside the surface of the crystal and of an electron in the conduction band. The number of electrons whose energy is greater than \(\varphi\) will be proportional to the number of electrons in the conduction band and to the Boltzmann factor, so that the magnitude of the electron emission will be proportional to \(n_c \exp\left(-\frac{\varphi}{kT}\right)\), i.e., proportional to

\[ N_1^{\frac12}\exp\left[-\frac{1}{kT}\left(\varphi+\frac{Q_1}{2}\right)\right]. \]

The complete expression for the thermionic emission also includes a temperature factor and takes the form (neglecting a statistical-weight factor of order unity—cf. Schottky\({}^{14}\) or Fowler\({}^{15}\), Sec. 11, 62):

\[ I=A'DN_1^{\frac12}T^{\frac54}\exp\left[-\frac{1}{kT}\left(\varphi+\frac{Q_1}{2}\right)\right], \tag{4,2} \]

PROPERTIES OF OXIDE CATHODES

where

\[ A' = e\left\{\frac{(2\pi)^{\frac{1}{2}} k^{\frac{5}{2}} m^{\frac{1}{2}}}{h^2}\right\}^{\frac{1}{2}} \]

is a quantity equal to approximately \(10^{-6}\), if \(I\) is expressed in amperes per \(1\ \mathrm{cm}^2\), and \(D\) is the mean transparency coefficient of the potential barrier at the oxide boundary\(^1\).

We are now in a position to interpret some experimental results.

The electrical conductivity \(\sigma\) of a given material, as is known from the elementary theory of conduction, is proportional to

\[ \frac{l\cdot n_c}{\sqrt{T}}, \]

where \(l\) is the mean free path of the electrons. In comparison with \(n_c\), which strongly depends on temperature [equation (4,1)], the quantity

\[ \frac{l}{\sqrt{T}} \]

is practically constant, and the semilogarithmic plot of the dependence of \(\sigma\) on reciprocal temperature will be a straight line with a slope proportional to \(Q_1\). The graphs given in Fig. 5 show that \(Q_1\) decreases from a value of about \(5\ \mathrm{eV}\), characteristic of a stoichiometric inactive composition, to approximately \(1.6\ \mathrm{eV}\), when the oxide is fully activated by the introduction of an admixture of free barium. Knowing \(Q_1\), we can calculate from the measured values of the thermionic work function

\[ \left(\varphi+\frac{Q_1}{2}\right) \]

[cf. below, Section 10, a)] the value of \(\varphi\). The best results for

\[ \left(\varphi+\frac{Q_1}{2}\right) \]

are of the order of \(1.2\ \mathrm{eV}\) (for barium oxide), so that \(\varphi\) is approximately \(0.4\ \mathrm{eV}\) (cf. Reimann and Treloar\(^ {10}\), who measured the electron emission and electrical conductivity of one and the same specimen).

Expressions (4,1) and (4,2) show that both electrical conductivity and thermionic emission are proportional to the square root of the number of impurity atoms. It follows from this that, in the course of the activation process at constant temperature, the conductivity changes in direct proportion to the electron emission.

A linear dependence of this type was observed by Albrecht\(^ {10}\). The change in the transparency coefficient \(D\) will be discussed in Section 10, a).

\(^1\) The following formula for the emission of an oxide cathode was proposed in 1933 by G. A. Tyagunov:

\[ I = A_1 T^{\frac{5}{4}}\exp\left(-\frac{\varphi_1}{kT}\right), \]

where

\[ A_1 = \frac{ 2^{\frac{3}{4}}\pi^{\frac{1}{4}} e\, m^{\frac{1}{4}} k^{\frac{5}{4}} }{ h^{\frac{3}{2}} } N_1^{\frac{1}{2}}, \]

\[ \varphi_1=\varphi+\frac{\Delta E}{2}. \]

This formula and the results of applying it to the study of the concentration of free barium were presented by T. P. Kozlyakovskaya at the Third All-Union Conference on Semiconductors; in addition, see Tyagunov\(^ {13}\). Translator’s note.

b) Thermoelectric Effect

The theory of the thermoelectric effect in semiconductors was given by Fowler^15 (Sec. 11, 78). The expression he derived for the emf of a circuit consisting of a semiconductor and an ordinary metal, with junction temperatures \(T\) and \(T+\Delta T\), can, in the case where \(\Delta T\) is small compared with \(T\), be approximately reduced to the form:

\[ E=-2\cdot 10^{-4}\Delta T\left(7+\lg_{10}\frac{lT}{\sigma}\right)\ \mathrm{V}, \tag{4,3} \]

where \(\sigma\) is the electrical conductivity and \(l\) is the mean free path of the electrons.

Experimental investigations of the thermoelectric effect in semiconductors were carried out by Mönch^15 and other authors at the University of Erlangen (cf. Frisch^14 and Bauer^16). These authors calculated the mean free path of the electron from simultaneous measurements of conductivity and the Hall coefficient \(R\), using the theoretical relation

\[ R\sigma=2.84\cdot 10^{9}\cdot \frac{l}{\sqrt{T}}. \tag{4,4} \]

Under ordinary conditions \(l\) proves to be a quantity of the order of \(10^{-7}\ \mathrm{cm}\), so that for a temperature of \(1000^\circ\mathrm{K}\)

\[ E=-2\cdot 10^{-4}\Delta T\cdot(3-\lg\sigma)\ \mathrm{V}. \tag{4,5} \]

In the laboratory of GEC the author of the present article carried out simultaneous measurements of conductivity and thermo-emf on a sample with barium oxide. The results obtained agree with equation (4,5). Depending on the preparation and activation of the oxide, the values of the thermo-emf varied within the limits from \(9\cdot 10^{-4}\) to \(3\cdot 10^{-3}\ \mathrm{V/deg}\).

Becker and Sears^10 observed a thermo-emf of \(0.05\ \mathrm{V}\) for a temperature difference which they estimated at approximately \(20^\circ\). This corresponds to a thermo-emf of \(2.5\cdot 10^{-3}\ \mathrm{V/deg}\), which is in agreement with the measurements cited above.

c) Ionic Emission

If the oxides \((\mathrm{BaSrCa})\) are heated to a temperature of approximately \(1800^\circ\mathrm{K}\), an ionic current of the order of \(10^{-7}\ \mathrm{A/cm^2}\) can be drawn to a negatively charged collector. Mass-spectrographic analysis (Blewett and Jones^15) showed that the ionic current consists of \((\mathrm{BaSrCa})\) ions with a single charge, if one does not count the ions of potassium and sodium impurities.

This phenomenon need not necessarily be explained by the presence of electrolytic conductivity, since other processes may also lead to the liberation of alkaline-earth atoms with their subsequent ionization. It has also been found that alkaline-earth oxides emit negative ions as well (Barton^5, Blewett and Jones^15, Bechman and Carnahan^17, Broudy and Piers^18). These ions are the cause of the unpleasant phenomenon of the black spot that appears in cathode-ray tubes with magnetic control. A beam of ions, almost not

deflected by a magnetic field and calculated for action on electrons, always falls on one place in the center of the screen, which, after brief ion bombardment, loses its fluorescing ability. Mass-spectrographic analysis of the ion beam shows that it contains a large number of different components. Ions of atomic and molecular oxygen are observed, which may be emitted by the oxide.

In addition to them, however, foreign ions appear, whose mass corresponds to hydrogen, carbon, and halogen contaminants, and also a whole series of other ions that cannot be so easily identified. It is possible that some reduction of this phenomenon may be achieved by using especially pure oxide coatings on thoroughly cleaned cores, with the use of binders with respect to which one can be confident that all their constituent parts completely volatilize in the process of the initial degassing.

c) Behavior of an Incandescent Tungsten Filament in BaO Vapors

Davisson and Pidgeon³ investigated the electron emission of a tungsten filament placed, at a temperature of 1000° K and below, in a stream of BaO vapors.

They observed that, upon condensation of BaO vapors on the filament, its electron emission increases by several orders of magnitude and reaches a maximum upon deposition of approximately a monatomic layer of BaO. Thereafter the emission falls, but after this reaches a second maximum, after which it gradually decreases and becomes established at an approximately constant value. This characteristic course of the emission was also observed in our laboratory. Unpublished experiments by Tawney (G. L. Tawney), performed in the GEC laboratory, showed that the double maximum changes into a single one if the vacuum conditions are maintained especially good. His experiments were carried out with a well-gettered lamp placed in liquid nitrogen.

It is possible that this double maximum has the same nature as that observed by Taylor and Langmuir¹² when cesium was deposited on an insufficiently cleaned tungsten surface.

If the temperature of the tungsten filament is raised to 2000° K or higher, then evaporation of the BaO molecules falling on it is observed in the form of positive ions.

At high filament temperatures, approximately 5% of the BaO molecules falling on it are converted into positive ions. It seems extremely improbable that, in the present case, direct ionization of BaO molecules occurs, since their ionization potential apparently has a value of the order of 10 V.

More likely is the process of capture by the tungsten filament of an oxygen atom and one electron with the evaporation of a barium ion. The energy required for this process can be calculated from consideration of a thermochemical cycle that includes the heat

of evaporation of oxygen from the surface of tungsten (162 kg cal., Langmuir and Villars\(^{10}\)), the work function for tungsten (104 kg cal.), the heat of dissociation of barium oxide (147 kg cal., Table 2), and the ionization potential of barium (119 kg cal.). The required resulting energy is equal to:

\[ -162 - 104 + 147 + 119 = 0 \pm 10 \text{ kg cal.} \]

(A large part of the possible error arises from the inaccuracy in determining the heat of dissociation of barium oxide.) We see that this reaction is probable, since the energy required for it is very small. The temperature at which the maximum ionization is observed corresponds to an electron emission smaller than from pure tungsten, as would be the case in the presence of a layer of adsorbed oxygen.

5. CHEMICAL PROPERTIES OF THE OXIDES OF THE ALKALINE-EARTH METALS

A detailed description of the general chemical properties of the oxides of the alkaline-earth metals can be found in Mellor’s book\(^{3a}\). We shall consider only those reactions which are of interest in connection with the activation of oxide cathodes.

The most active reducing agent for the oxides (BaSrCa) is, apparently, aluminum, which enters into reactions of the “thermite” type with all the oxides. When a mixture of powders of the oxides (BaSrCa) and aluminum is heated, a vigorous reaction takes place with liberation of heat and evaporation of the reduced alkaline-earth metal.

Magnesium partially reduces CaO at high temperatures. The reaction of magnesium with SrO proceeds more energetically than with CaO, and with BaO still more energetically. Other reactions of the type

\[ (\mathrm{BaSrCa})\mathrm{O} + \text{metal} \rightleftarrows \text{metal oxide} + (\mathrm{BaSrCa}) \]

reach equilibrium in most cases under conditions of very low vapor pressure of (BaSrCa). If the process takes place with removal of the reaction products, as usually occurs under the conditions of activation of an oxide cathode, then metallic (BaSrCa) can be removed by diffusion and evaporation, and the reaction will continue until it is stopped by the formation of a layer of metallic oxide or until one of the components is completely consumed.

Investigations now being carried out under the direction of Liebhafsky and the author (see Blewett\(^{17}\)) on the endothermic reactions of BaO with Ti, Ta, Ni, Mo, and C show that at a temperature of the order of \(1400^\circ \mathrm{K}\) the initial rate of barium liberation from a mixture of Ti and BaO powders is approximately \(10^3\) times greater than the rate of evaporation of BaO.

Tantalum is about thirty times less energetic, while Ni and Mo liberate Ba at a rate lower than the rate of evaporation of BaO. The reaction of BaO with carbon leads to the formation of barium carbide

in amounts comparable with the yield of Ba in the reaction BaO + Ti. From the slope of the curves for the dependence of the logarithm of the yield on \(\frac{1}{T}\), the following values of the heats of reaction can be obtained:

\[ \begin{aligned} \mathrm{BaO} + \mathrm{Ti} \qquad & \Delta Q = 83\ \text{kg cal}\\ \mathrm{BaO} + \mathrm{Ta} \qquad & \Delta Q = 96\ \text{” ”}\\ \mathrm{BaO} + 3\mathrm{C} \to \mathrm{CO} + \mathrm{BaC}_2 \qquad & \Delta Q = 80\ \text{” ”} \end{aligned} \]

The easy activation of oxide cathodes with a core of “Konel” alloy is evidently explained by reaction with titanium (Laurie\(^9\)); the action of the “batalum” getter is based on the reaction with tantalum (Lederer and Owsley\(^ {16}\)). It is also used for obtaining barium by the method described by Benjamin and Jenkins\(^ {17}\).

The reaction of BaO with Ni was investigated by Wagner\(^9\), who, on the basis of a thermodynamic consideration, showed that at a temperature of \(1273^\circ\ \mathrm{K}\) the vapor pressure of Ba obtained in the reaction

\[ \mathrm{BaO} + \mathrm{Ni} \to \mathrm{NiO} + \mathrm{Ba}, \]

should be of the order of \(3 \cdot 10^{-17}\) atm, or approximately \(10^7\) times less than the vapor pressure of BaO at the same temperature. Accordingly, only a very slight reduction of BaO in the layer on a purely nickel core should be expected. In this connection it should be noted, however, that magnesium, which, as we have already noted above, is a very energetic reducing agent, is often present as an impurity in industrial nickel.

On the basis of Shril’s\(^ {16}\) most recent investigations, it is necessary to revise somewhat the views set forth by Mellor regarding the existence of the lower oxide of barium, \(\mathrm{Ba}_2\mathrm{O}\). These investigations showed that the substance which previous investigators had taken for \(\mathrm{Ba}_2\mathrm{O}\) is in fact simply a mixture of BaO and Ba. Shril’s work also includes a study of the mutual solubility of Ba and BaO.

In metallic barium at \(1000^\circ\ \mathrm{K}\), up to \(20\%\) BaO can dissolve. At \(1473^\circ\ \mathrm{K}\), a \(45\%\) solution of BaO in barium can be obtained. The solubility of Ba in BaO lies within the limits of \(1\%\).

6. PHYSICAL PROPERTIES OF MIXTURES OF OXIDES OF ALKALINE-EARTH METALS

Benjamin and Rooksby\(^ {12}\), by the method of X-ray diffraction, showed that BaO and SrO form a series of solid solutions. Burgers\(^ {12}\) came to the same conclusion and showed that the lattice parameter of the solid solution is a linear function of the molecular composition. Gertner\(^4\) and Derbishire\(^ {17}\) studied the surface of an oxide cathode with a mixed coating by the method of electron diffraction, and both came to the conclusion that, after calcination, the surface layer consists mainly of SrO, as was to be expected from the higher vapor pressure of BaO.

Klassen and Veneman^12 measured evaporation rates from mixtures of BaO and SrO and of BaO and CaO and showed that the evaporation rate of BaO is a monotonic, but not linear, function of molecular concentration. They also conclude that BaO and SrO, as well as BaO and CaO, form mutual solid solutions. Benjamin and Rooksby^12 assert that, for a mixed oxide coating, the highest emission is obtained at a 50% molecular content of BaO and SrO, and that this emission is greater than for each component separately.

7. PROPERTIES OF ALKALINE-EARTH METALS: THIN FILMS ON TUNGSTEN

Table 2 gives some of the most essential properties of the alkaline-earth metals. We shall consider separately the thermionic properties of thin barium films.

Becker^(8,2), Ried and Harris (Reimann^13, p. 159 ff.), Nelson^10, and Iglin^7 observed a change in the electron emission from a tungsten filament when barium was deposited on it.

Fig. 7. Thermionic emission from thin layers of barium on tungsten (Becker^8)

The quantity \(f\) is proportional to the number of barium atoms on \(1\ \text{cm}^2\) of the tungsten surface.

Figure 7 reproduces Becker’s curve, giving the dependence of the logarithm of the electron emission on the quantity \(f\), the ratio of the time of barium deposition to the time required to achieve the highest electron emission. Barium was distilled at a constant rate, and therefore the quantity \(f\) may characterize the fraction of the tungsten surface covered with barium. Until recently it was often assumed that the maximum emission, i.e., the point \(f=1\), corresponds to a monatomic layer of barium. Taylor and Langmuir^12 showed, however, that this assumption is incorrect in the analogous case of Cs on W, since the maximum emission was obtained with deposition of cesium corresponding to 0.67 of a monatomic layer. For barium, the determination of a monatomic layer is more difficult, since the arrangement of barium atoms on tungsten is unknown. It is possible, however (Blewett^17), to measure the number of barium atoms falling on \(1\ \text{cm}^2\) of tungsten surface at which the maximum electron emission occurs. This number is equal to \((5.7 \pm 0.5)\cdot 10^{14}\) atoms per \(1\ \text{cm}^2\) of the macroscopic surface of tungsten, or \((4.1 \pm 0.4)\cdot 10^{14}\) atoms per \(1\ \text{cm}^2\) of the true surface of tungsten, if the roughness factor for tungsten given by Taylor and Langmuir is applied.

The dependence of the steady-state electron emission from a tungsten cathode placed in a stream of barium vapor on temperature

shown in Fig. 8. The part of the curve belonging to the low-temperature region corresponds to emission from a thick barium layer. As the temperature rises, the evaporation of barium increases, and the barium layer retained on the cathode becomes thinner and thinner. The electron emission passes through a maximum, falls to a certain minimum, and then rises again in accordance with the emission properties of pure tungsten.

Reid and Harris determined the thermionic constants for a tungsten cathode with a barium coating corresponding to the emission maximum \((f = 1)\). The emission is determined by the formula

\[ I = 1.5\,T^2 \exp\left(-\frac{18100}{T}\right)\ \mathrm{A/cm^2}. \]

These authors also investigated the emission of a compound cathode with the structure W—O—Ba and established that the dependence of electron emission on \(f\) has the same character as in the case of barium on pure tungsten. The optimum electron emission is determined in this case by the formula

\[ I = 0.18\,T^2 \exp\left(-\frac{15600}{T}\right)\ \mathrm{A/cm^2}. \]

Fig. 8. Thermionic emission from a tungsten filament placed in a stream of barium vapor (Blewett, unpublished)

Reid and Harris studied a whole series of combinations of layers of barium and oxygen on tungsten. From a number of observed phenomena we note the interesting fact that a cathode with the structure W—Ba—O, when heated, apparently transforms into a W—O—Ba cathode.

At present there are uncertainties concerning the question of the migration of barium over the surface of tungsten. Becker\(^{8(2)}\) asserts that in a narrow temperature interval lying immediately below the temperature at which rapid evaporation of barium begins, the latter migrates with appreciable speed over the surface of tungsten.

On the other hand, Benjamin and Jenkins\(^{17}\) assert that no such migration occurs. From the published descriptions of the experiments it is apparently impossible to find a way to reconcile these contradictory results.

8. METHODS OF DETECTION (BaSrCa)

a) Chemical methods

In those cases where it is not required to separate the free metals from their compounds, the total amount of (BaSrCa) can be determined after opening the experimental lamp by means of standard microchemical methods. Upon contact with air-

...all the free alkaline-earth metals immediately pass into oxides, carbonates, and other compounds, so that there remains no possibility of separating the free metal and, for example, the evaporated oxide. Consequently, if it is necessary to measure the amount only of the free metal, the measurement must be carried out without breaking the vacuum. This can be done by quantitative analysis of the reactions that occur when certain gases are admitted into the lamp.

First of all, the following reactions suggest themselves:

\[ 2(\mathrm{BaSrCa}) + \mathrm{O}_2 \to 2(\mathrm{BaSrCa})\mathrm{O}, \tag{I} \]

in which the measure of the total amount of free alkaline-earth metals is the total amount of oxygen absorbed;

\[ \begin{gathered} \mathrm{H_2O}\\ (\mathrm{BaSrCa}) + 2\mathrm{H_2O} \to \begin{matrix} (\mathrm{BaSrCa})\mathrm{O}\\ (\mathrm{BaSrCa})(\mathrm{OH})_2 \end{matrix} +\mathrm{H}_2, \end{gathered} \tag{II} \]

\[ (\mathrm{BaSrCa}) + \mathrm{CO}_2 \to (\mathrm{BaSrCa})\mathrm{O} + \mathrm{CO}, \tag{III} \]

where the amount of \(\mathrm{H}_2\) or \(\mathrm{CO}\) is a measure of the content of free \((\mathrm{BaSrCa})\);

\[ \left. \begin{aligned} 3(\mathrm{BaSrCa}) + \mathrm{N}_2 &\to (\mathrm{BaSrCa})_3\mathrm{N}_2,\\ (\mathrm{BaSrCa})_3\mathrm{N}_2 + 6\mathrm{H}_2\mathrm{O} &\to (\mathrm{BaSrCa})(\mathrm{OH})_2 + 2\mathrm{NH}_3, \end{aligned} \right\} \tag{IV} \]

where the measure of the content of free \((\mathrm{BaSrCa})\) is the ammonia evolved.

A certain drawback of all four methods, especially the first, is the fact that other metals also inevitably take part in the reaction, in particular the electrode metals, etc. Nevertheless, reaction (II) was successfully used by Berdennikova\(^{11}\) and by Clausing (see de Boer’s book, reference 5, p. 275).

Prescott and Morrison\(^{17}\) point to a very definite advantage of reaction (III). Fritz\(^{14}\) obtained good results using reaction (IV). It should be noted, however, that special caution is necessary both in carrying out and in interpreting measurements of this type.

b) Direct weighing methods

The amount of evaporated \((\mathrm{BaSrCa})\) can be directly weighed in vacuum by mechanical methods. It is possible, for example, to make sufficiently sensitive tungsten or quartz spring balances.

The author has described lever balances (Blewitt\(^{18}\)) suitable for measuring, with sufficient accuracy in vacuum, masses of the order of \(1\ \mathrm{mg}\). When weighing methods are used, additional chemical control is, of course, required, so that the experimenter can be sure that he is actually weighing \((\mathrm{BaSrCa})\), and not any of its compounds or contaminants.

c) Optical methods

Within known limits, the opacity coefficient of the deposited layer (BaSrCa) can be used as a measure of the amount of deposited metal. It is necessary, however, to take appropriate measures against the formation of gas layers on the collector surface and against the deposition on it of other opaque substances.

At best, this method is applicable only after preliminary calibration by other independent methods.

d) Electrical methods

The most widely used methods for detecting (BaSrCa) are based on the thermionic properties of deposited layers of these metals. In the case of thick layers, methods based on resistance measurement may also be applied; however, they are complicated by the same factors as the optical methods.

The following three thermionic methods are used.

I. Measurement of thermoelectron emission. If a tungsten filament is placed in the path of the vapor stream (BaSrCa), its electron emission will change in the manner described above, in Section 7. Its dependence on temperature is illustrated by Becker’s curve (Fig. 7). The initial slope of the curve of the logarithm of the emission current as a function of temperature or time, or the time required to attain maximum emission, may serve as an absolute measure of the intensity of the vapor stream. For practical application we shall use the initial slope of the curve \(\lg I\) as a function of the magnitude \(f\) for barium on a tungsten filament at \(1100^\circ\mathrm{K}\)

\[ \frac{d \lg I}{d f} = 24.5 . \]

If \(N\) denotes the number of barium atoms on \(1\ \mathrm{cm}^2\) of the apparent surface of tungsten, then one may write \(N = 5.7 \cdot 10^{14} f\) and thus calculate the rate of deposition of barium from the initial rate of increase of the electron current \(\left(\frac{d\lg I}{dt}\right)\):

\[ \frac{dN}{dt} = 5.7 \cdot 10^{14} \frac{df}{dt} = 2.3 \cdot 10^{13} \frac{d\lg I}{dt}. \]

Usually a more reliable method is to measure the time required to attain maximum electron emission. At a not very high filament temperature (below \(1100^\circ\mathrm{K}\)), when the evaporation of barium is negligibly small, this time does not depend on temperature.

For Sr and Ca the number of atoms falling on a unit surface of tungsten at \(f = 1\) is unknown. One can only suppose that it is of the same order as for barium.

II. Determination of the contact potential. This method is a refinement of method I, carried out by Reimann\(^ {14}\) et al. The collector filament used in method I is here left at

room temperature. A negative potential relative to the cathode is applied to it, and the current of electrons reaching the filament from the space-charge region is measured. Any change in the work function of the collector filament caused by deposition \((\mathrm{BaSrCa})\) leads to a change in the contact potential and thus shifts the volt-ampere characteristic along the voltage axis.

Using this method, one can construct the curve of the dependence of the work function on the value of \(f\). At the point \(f=1\), corresponding to the maximum of the Becker curve \(\lg I = F(f)\), the work-function curve passes through a minimum.

Fig. 9. Emission of positive ions from a tungsten filament in a jet of barium vapor (Blodgett, unpublished)

At temperatures above \(2000^\circ\mathrm{K}\) this curve represents emission from pure tungsten. At temperatures below \(2000^\circ\mathrm{K}\) barium is deposited on the surface of the tungsten and remains there for a noticeable time.

This method has substantial advantages, since it is free of the drawbacks caused by evaporation of barium from the collector and by nonuniformity of the latter’s temperature due to cooling of the ends.

III. The positive-ion method. This method is especially convenient for determining Ba, although it is quite applicable also in the case of sufficiently intense fluxes of Sr or Ca. If the filament is in a stream of vapor \((\mathrm{BaSrCa})\) and its temperature is sufficiently high that the metal does not condense on it at all, then some of the atoms striking it will fly off in the form of positive ions. The relative fraction of ionized atoms is determined by the Saha–Langmuir formula

\[ \frac{n^+}{n}=\frac{\omega^+}{\omega}\exp\left(-\frac{V_i-\varphi}{kT}\right), \]

where \(n^+\) is the number of ions evaporating per second, \(n\) is the number of atoms arriving per second, \(\omega^+\) is the statistical weight of \((\mathrm{BaSrCa})^+\), \(\omega\) is the statistical weight of \((\mathrm{BaSrCa})\) \(\left[\text{for }(\mathrm{BaSrCa})\ \frac{\omega^+}{\omega}=\frac{2}{1}\right]\), \(V_i\) is the ionization potential of \((\mathrm{BaSrCa})\), and \(\varphi\) is the work function of the filament.

As the material for the collector filament, the most suitable is tungsten, which has a high melting temperature and a large work function.

When the tungsten collector filament is maintained at a temperature of \(2500^\circ\mathrm{K}\), the average fraction of alkaline-earth atoms reflected in the form of positive ions is, for Ba, \(10\%\), for Sr, \(1\%\), and for Ca, \(0.1\%\).

This method can be checked by constructing the dependence curve of the logarithm of the ion current on \(\frac{1}{T}\). In the region of sufficiently high

at which temperatures, when the filament remains completely clean, this curve turns into a straight line with a slope corresponding to the value \(V_i-\varphi\), as shown in Fig. 9.

When using this method it is necessary to take special measures against the occurrence of photocurrents from the positive-ion collector. As a result of barium deposition these currents may be quite considerable; however, they can be eliminated by means of a magnetic field of the order of 100 gauss, produced by a Helmholtz coil.

9. MANUFACTURE OF OXIDE CATHODES

There is no exact technique for manufacturing oxide cathodes. Countless patents have been taken out on methods for producing cathodes possessing high activity, a long service life, and strong adhesion of the coating to the metal of the core. However, on a number of basic points there is general agreement. The oxides of barium and strontium are considerably better emitters than calcium oxide, and barium oxide emits better than strontium oxide. A mixture consisting of 50% barium oxide and 50% strontium oxide gives a coating possessing satisfactory mechanical properties and an emission apparently exceeding somewhat the emission of barium oxide alone. As the core material, nickel has become most widely used; it possesses the necessary physical properties and is comparatively inexpensive. For special purposes a whole series of alloys has been developed. One alloy especially deserving attention is known under the name “konel.” Owing to a small admixture of titanium in it, rapid chemical reduction of the oxides takes place and a sufficient amount of free alkaline-earth metal appears, which is necessary for activation.

Oxides (BaSrCa) are unstable in an atmosphere of air and pass into more complex compounds, such as carbonates, nitrates, oxide hydrates, etc. As a consequence, cathodes are coated with one or several of these compounds, followed by their conversion into oxides upon heating. The compounds at present most widely used are the carbonates and oxide hydrates.

For better adhesion of the carbonate coating to the cathode core, organic binders are used. A mixture consisting of 1 part zapon lacquer and 20 parts amyl acetate may serve as a good binder. A suspension of the carbonate powder in this binder is applied to the cathode by immersing the latter in the suspension or by spraying. If it is necessary to obtain a thick layer, repeated application is used, with intermediate calcination at a temperature sufficiently high to burn out the binder.

A coating of oxide hydrates is applied by immersing the cathode in a bath with molten oxide hydrates. For tungsten cathodes this method is apparently the best of all. The coating obtained in this way is especially strong. After this the cathode may be subjected to calcination in air at a high temperature, during which a reaction takes place between the core metal

and coated. A cathode subjected to such treatment is called a “chemically bound” cathode (Becker\(^8\)) and in the finished state has a dull-gray appearance. Cathodes not subjected to such severe heat treatment retain the white color of the coating and are called “chemically unbound.”

The prepared cathode is mounted in a lamp, the lamp is evacuated, and it is left on the pump throughout the entire time while the cathode is being heated at a temperature of \(1400^\circ\mathrm{K}\) in order to convert the coating into oxide.

In this process, if no strong reducing agent has been used, the cathode remains “unactivated.” By this term we mean that the electron emission at the operating temperature (approximately \(1000^\circ\mathrm{K}\)) remains very low (of the order of a few microamperes). “Activation” may be carried out in a whole series of ways. In some cases prolonged heating proves sufficient. Usually, however, a voltage of the order of several hundred volts is applied to the anode of the lamp, and the activation process is accelerated. When the emission current reaches a value of the order of \(300\ \mathrm{mA}/\mathrm{cm}^2\) (at \(1100^\circ\mathrm{K}\)), further growth of emission ceases, and the cathode is ready for operation.

The method just described is the most widespread. However, several other methods also deserve attention. Numerous attempts have been made to use barium azide (\(\mathrm{BaN}_6\)), and some of them have been successful. At a temperature of approximately \(430^\circ\mathrm{K}\) a vigorous decomposition reaction occurs with the liberation of pure barium. The barium can be converted into oxide by adding oxygen or by depositing the barium on an oxidized surface.

Mixtures consisting of \(25\%\ \mathrm{BaO}\) and \(75\%\ \mathrm{Al}_2\mathrm{O}_3\) (Ramsey and Rooksby, British Patent No. 437967) or \(25\%\ \mathrm{BaO}\) and \(75\%\ \mathrm{BeO}\) (cf. Benjamin and Jenkins\(^ {17}\)) are chemically stable and have melting temperatures lower than those of each of their constituent components. This substance was used by Hull\(^ {17}\) in his migration cathodes. Barium oxalate (\(\mathrm{BaC}_2\mathrm{O}_4\)) possesses a certain advantage; it is converted into oxide with the evolution of \(\mathrm{CO}\) and \(\mathrm{CO}_2\). The gas \(\mathrm{CO}\) acts as a reducing agent and thus promotes the purification of the system from oxygen. Patay and his school in Budapest (Patay and Tomaschek\(^ {15}\)) developed a method for depositing alkaline-earth oxides from a colloidal suspension, as a result of which a somewhat smoother and denser coating is obtained than that produced by direct immersion or spraying methods. For their cathodes these authors obtained an emission that was more stable, though somewhat smaller in magnitude, than for cathodes of the usual type. Benjamin, Heck, and Jenkins also developed methods for reducing the particle size in the coating. Earlier review papers (cf. Stadt\(^6\), Simon\(^7\), Hodgson, Harley and Pratt\(^8\), Wagner\(^9\), etc.) describe a whole series of other methods of making cathodes, each of which leads to essentially identical final products.

For operation in gas-discharge devices, Hull created a cathode of an entirely different type. A convenient cathode design is

open at one end, with radial ribs arranged inside it. The oxide coating is applied to the surfaces of these ribs and to the inner surface of the cylinder. The cathode is heated by a heater arranged along its axis. The presence of gas makes it possible to draw a large electron current through the openings, while heat losses can be greatly reduced by means of special coaxial shields. In this way, a very high cathode efficiency can be achieved.

The cathode recently created by Hull, which he called a “dispenser,” i.e. a “distributing” cathode (Hull^17,18), is an improved modification of a cathode of this type.

The emitting surfaces in it consist of a series of radial ribs surrounded by a cylindrical molybdenum shield. Initially the surface remains uncoated. The cathode heater is a sleeve made of molybdenum mesh packed with the eutectic mixture BaO, Al₂O₃, which was mentioned above. When this sleeve is heated by a current passing through it, barium oxide evaporates in very small quantities onto the heat shield and onto the radial ribs and forms a very active emitting surface. By appropriate thermal shielding, the temperature is set so that the small losses of barium oxide through the openings in the shield are compensated by its continuous evaporation from the sleeve.

These cathodes give an emission current of the order of several amperes per 1 cm² and have such a long service life that, under continuous operation for three years, they show no signs of loss of emission.^1)

10. THE ACTIVATION PROCESS

Our knowledge of the activation process may be summarized as follows. Activation of an oxide cathode occurs when, as the result of a chemical reaction, or electro-

^1) Cathodes of this type may be called sublimation cathodes. They include also the cathodes developed in 1932–33 for powerful gasotrons by the employees of the “Svetlana” plant—Yu. D. Boldyrev and I. P. Polev (see Shaposhnikov^13,17 and inventors’ certificates No. 128479, class 21 d, 13 of May 10, 1933, and No. 128040, class 21 d, 13 of April 30, 1933).

In one of these cathodes, thermite barium pellets, or pellets made of alloys containing barium, are placed on the lateral surface of a molybdenum cylinder heated from within and are covered with several layers of molybdenum mesh. On the outside, the entire structure is protected by a series of molybdenum heat shields. Such a cathode, at a heating power of 300 W and at a temperature of the emitting meshes onto which barium is distilled from the pellets of about 700°, gives an emission of about 2.5 A/W.

The second type of such cathode consists of two spirals of molybdenum ribbon, between which a thermite pellet is placed. When the spiral is heated, barium is distilled onto it from the pellet, and the spiral is continuously maintained in an active state.

As we see, in principle of operation both cathodes are completely identical with Hull’s cathode and differ from the latter only in their structural design and partly in the chemical composition of the source of barium. Translator’s note.

lysis, or bombardment by positive ions, or else by the combined action of these factors, free (BaSrCa) is liberated from the oxides. This free metal is distributed throughout the entire thickness of the oxide layer and changes its properties in such a way that a strong increase appears in the electron emission from the outer surface of the oxide coating.

Let us now consider the phenomena that confirm this proposition.

a) Liberation of an alkaline-earth metal

If an oxide cathode is subjected to a sufficiently vigorous thermal or electrical action, a dark metallic deposit appears on the inner surface of the bulb of the vacuum tube. When air or oxygen is admitted, the deposit becomes white, and chemical analysis shows that it consisted of (BaSrCa).

When a strong electron current is drawn from a superheated barium oxide cathode, a greenish glow discharge arises. The spectrum of this discharge contains the characteristic lines of barium (Detels^9, Hertz^9).

A tungsten filament placed near a heated barium oxide cathode is activated in exactly the same way as in a stream of metallic barium vapor (Becker^8(1)).

The presence of free (BaSrCa) within the bulk of the cathode was established by means of the chemical methods described above, in section 8,a) (Berdennikova^11, Clausing—see de Boer’s book, p. 275, Fritz^14, Prescott and Morrison^17). It was found that in an activated cathode free (BaSrCa) is reduced from approximately 0.5% of the total number of oxide molecules.

b) Activation caused by the presence of free alkaline-earth metal

An inactive oxide cathode placed in a stream of barium vapor becomes active. Its emission becomes the same as would have been achieved by activation by one of the ordinary methods (Becker^8(1)).

Upon heating to a temperature of 1600°K, at which free (BaSrCa) should evaporate and only unreduced oxides remain in the coating, the cathode loses its activity (Koller^5).

The cathode can be deactivated by admitting gas, as a result of the reaction with which free (BaSrCa) is converted into (BaSrCa)O.

c) Mechanism of oxide reduction

There are four possible explanations of the mechanism that leads to the liberation of free alkaline-earth metal, which determines the activation of the cathode. The first, most natural assumption—that thermal dissociation takes place—must be rejected on the basis of calculations by Villars and Dushman (Dushman^9) and Becker^10. These authors showed that at

under the most favorable conditions, the dissociation pressure of BaO at \(1000^\circ\mathrm{K}\) should not exceed \(10^{-40}\) atm. Therefore, even under conditions of the best attainable vacuum, the barium liberated should so rapidly convert back into BaO that its presence could in no way be detected. For SrO and CaO the dissociation pressures must be still smaller.

There remain three possible mechanisms for the liberation of \((\mathrm{BaSrCa})\), namely: chemical reaction with the core metal, electrolysis during the passage of the emission current, and bombardment by positive ions of residual gases. It seems probable that all three of these mechanisms take part in the activation process.

The first of them can be separated from the other two by carrying out “thermal activation” of the cathode. The cathode is heated without the application of an anode voltage, and the activity is measured at low temperature by applying very short pulses of anode voltage. In using this method one may assume that the influence of the anode voltage has been reduced to a minimum.

Experiments of this type lead to the conclusion that, provided the necessary amount of heat is supplied for a sufficiently long time, almost all oxide cathodes can be thermally activated. Cathodes with a platinum core apparently require especially intensive treatment. Experiments by Toone, carried out in the GEC laboratory, show that cathodes with cores of spectroscopically pure platinum (with the exception of traces of Cu) cannot be thermally activated. Cathodes with nickel cores are thermally activated with differing degrees of ease, apparently depending on the amount of chemically active impurities present in the nickel. The inclusion in the core metal of chemically active components (as, for example, Ti in the alloy “konel”) leads to rapid thermal activation at comparatively low temperatures (Laurie\(^9\), Benjamin\(^14\)).

Any mixtures of \((\mathrm{BaSrCa})\mathrm{O}\) react with any metal in a reaction which proceeds until an equilibrium state is reached according to the formula

\[ (\mathrm{BaSrCa})\mathrm{O} + \text{metal} \rightleftarrows \text{metal oxide} + (\mathrm{BaSrCa}). \]

In most cases the equilibrium vapor pressure of \((\mathrm{BaSrCa})\) proves to be very low. If it is considerably less than the vapor pressure of \((\mathrm{BaSrCa})\mathrm{O}\), then it is obvious that the entire coating will evaporate before thermal activation is completed. Nevertheless, we can state with certainty that in all cases some portion of the free \((\mathrm{BaSrCa})\) needed for activation is produced as a result of the chemical reaction of the oxide with the core metal.

In this connection it is also necessary to consider the role of the organic binder. It is usually assumed that the binder evaporates completely and takes no part in any reactions whatsoever. However, it is quite probable that the carbon is not removed completely, and that the small amount of it remaining reacts with the oxide.

and forms a carbide, which can easily decompose with the liberation of free metal.

When lamps with oxide cathodes are broken in a moist atmosphere, the odor of acetylene often appears—a sure sign that carbide (BaSrCa) was present in the lamps.

To study the second and third possible mechanisms of reduction of (BaSrCa) (electrolysis and bombardment by positive ions), let us consider the process of “current activation,” which occurs when a positive voltage is applied to the anode. Observations of the increase in emission current show that activation in this case usually proceeds considerably faster.

If an electron current is drawn from the cathode, then the current within the oxide coating has such a direction that (BaSrCa) must be deposited at the surface of the core, while oxygen is liberated at the outer surface of the oxide layer. In Section 4, a), it was shown that both barium ions and oxygen ions participate in electrolysis. The current of oxygen ions after the completion of the activation process becomes very small; however, the current of barium ions is maintained at a value having the order of \(10^{-5}\) of the total current. Barium is liberated at the surface of the core, but then diffuses through the entire thickness of the oxide layer, since evaporation of barium from the cathode is observed throughout the whole service life. This mechanism of barium reduction probably acquires especially great importance for the continuous replenishment of the free-barium content, and thereby also for preserving cathode activity in the later stages of service life.

The third possible mechanism—the liberation of (BaSrCa) under the action of bombardment by positive ions—is difficult to separate from the other mechanisms when considered. Even at pressures of the order of \(10^{-8}\) mm Hg an amount of ions may be formed sufficient for the dissociation of a noticeable quantity of oxide (provided that the process of dissociation under the action of ion bombardment is highly efficient).

The experiments described in Section 11, 3), concerning the influence of bombardment by positive ions on cathode activity, show that this process must play some, although small, role in cathode activation. In this connection certain curves obtained by Berdennikova\(^{11}\) are of interest. For the case of maintaining a glow discharge in the gas, the curve of the amount of barium liberated per unit charge passed, as a function of time, is characterized by unusually large initial ordinates. When the gas is pumped out, the rate of barium liberation decreases and eventually settles at a certain constant value, in agreement with Faraday’s first law. It is evident that the high initial rate of barium liberation is connected with bombardment by positive ions. A systematic investigation of the action of ion bombardment, however, has not yet been carried out. Further analysis of this mechanism may be found in the papers of Koller\(^{5}\), MacNabb\(^{8}\), and Wagner\(^{9}\).

d) Source of the emitted electrons; distribution of free (BaSrCa)

In the question of the source of thermionic emission from oxide cathodes, up to 1931 there existed serious disagreement. Some authors, basing themselves on the discovery of barium content in the cores of cathodes that had undergone aging (Laurie9), or on the course of the electrical-conductivity curves of the oxide (Reimann and Murgoci9), came to the conclusion that thermionic emission comes from the surface of the oxide-cathode core. The incorrectness of this view was shown by the conclusive experiments of Becker and Sears10. These authors observed that stripping, as a result of mechanical impact, of the oxide coating from an active cathode leads to a more than thousandfold decrease in its activity. If the source of emission were the surface of the core, then the activity of the cathode should have increased or, in any case, remained constant. Becker and Sears also changed the effective dimensions of the core, thereby changing the surface area of the oxide coating. To accomplish this they inserted into the oxide layer a spiral ribbon “probe.” The cathode emission remained constant, regardless of whether the cathode potential was applied to the core or to the probe, or to both together. As stated in Section 10, b), the same authors showed that activation of a cathode by depositing barium on its surface leads to the same characteristics as activation carried out by ordinary methods. Haxford10 measured the external photoelectric work function during the process of activation of an oxide cathode and found that it coincides with the thermionic work function. These four experiments are sufficient for the conclusion that electron emission comes from the outer surface of the cathode.

Free alkaline-earth metal, on the other hand, is not localized on the surface of the oxide. We know [see Section 10, a)] that in an active cathode approximately 0.5% of the total amount of oxide is reduced to free metal. If all this amount of metal were concentrated on the surface, then, first, a change in the external appearance of the surface would occur, which in fact is not observed, and, second, the thermionic emission of the cathode would have to correspond to the emission characteristics of the pure metal. In addition, it has been established that reactions between free metal and foreign gases specially introduced into the lamp [cf. Section 8, a)] require, for their completion, a time measured in minutes and even hours. This could not occur if all the free metal were on the outer surface. All these facts, in combination with the observed parallelism of changes in electrical conductivity and thermionic activity, lead to the conclusion that the greater part of the free metal is distributed throughout the entire thickness of the oxide. It is assumed that the metal is present in the form of atomic inclusions, which facilitate the liberation of electrons and, for the formation of electro-

conductivity, and for thermionic emission. Such a mechanism, in the main, was described by Becker and Sears and theoretically developed in section 4, a).

The distribution of free metal in the oxide layer affects the emission current according to equation (4, 2) through the quantities \(N_{1}^{1/2}\) and \(Q_{1}/2\). Other quantities entering into this equation, namely \(D\) and \(\varphi\), depend on the nature of the oxide surface. The values of these quantities, as will be explained in the next section, indicate the presence of a potential barrier at the oxide boundary. The latter may be caused by the existence on the oxide surface of a monoatomic layer of free metal. At present there are still insufficient experimental data to construct further assumptions concerning the nature of the oxide surface or its changes during the activation process.

11. ELECTRICAL PROPERTIES OF OXIDE CATHODES

a) Emission constants

Above, in section 4, a), the equation was obtained

\[ I=10^{-6} D N_{1}^{1/2} T^{5/4}\exp\left[-\frac{1}{kT}\left(\varphi+\frac{Q_{1}}{2}\right)\right]\ \mathrm{A/cm^{2}}, \tag{11,1} \]

to which the thermionic emission of an oxide cathode should conform. In the past, however, it was customary to analyze data on the emission of an oxide cathode in accordance with the Richardson–Dushman formula (valid for metals)

\[ I=AT^{2}\exp\left(-\frac{\varphi_{1}}{kT}\right). \tag{11,2} \]

Whichever of these two equations is applied, the semilogarithmic characteristic turns out to be an equally correct straight line, since both relations are very little sensitive to the first temperature factor. The work function \(\varphi_{1}\), determined from the second equation, differs only by a few hundredths of an electron-volt from the quantity \(\left(\varphi+Q_{1}/2\right)\). Since the experimentally obtained data are characterized by considerably large discrepancies, the value of the work function given in the literature may be accepted without change.

The work function of oxide cathodes has been measured by many methods. It is possible to use various thermionic characteristics (Koller\(^5\), Detels\(^6\), Espe\(^6,8\), Becker\(^8\)(1), Krochek and Liubke\(^9\), Reiman and his collaborators\(^9,10\), Knipkamp and Nebel\(^11\), Maddock\(^14\), Patai and his collaborators\(^14,15\), and many others). Calorimetric measurements of the work function, associated with measurement of the cooling effect of electron emission, were carried out by Davisson and Germer\(^4\), Michel and Spannerm\(^5\), Rothe and Heinze\(^12\). Found\(^13\)

proposed a method for measuring electron emission in the absence of a field; he thus determines the true values of the thermionic constants. His method is based on the continuous supply to the cathode of positive ions from the plasma of a gas discharge. Heinze and Wagener[^17] calculate the work function on the basis of contact-potential measurements. This method is more suitable for studying changes in the work function than for determining its absolute value. Haxford[^10] measured the photoelectric work function of an oxide cathode.

Before 1915 the values of the work function for the oxides \((\mathrm{BaSrCa})\) were considered to lie between three and four electron-volts. As vacuum technique improved, lower values began to be observed, and the most recent determinations of the work function give, in most cases with an accuracy up to \(0.3\ \mathrm{eV}\), the values: \(1.1\ \mathrm{eV}\) for \(\mathrm{BaO}\), \(1.4\ \mathrm{eV}\) for \(\mathrm{SrO}\), and \(1.9\ \mathrm{eV}\) for \(\mathrm{CaO}\).

If the cathode is not completely activated, or if any impurities are present in it, higher values may be observed. The value of \(1.96\ \mathrm{eV}\) for \(\mathrm{BaO}\), obtained by Found, is determined by the presence of gas, which is necessary to create the discharge used in his method. Calorimetric measurements of the work function gave higher values than those cited above. However, thermionic measurements lower the quantities obtained by calorimetric methods, and therefore it becomes possible to suppose that, in calorimetric measurements, the cathodes were not brought to that degree of activation which is attained in the best cathodes studied by thermionic methods.

Because the coefficients \(A\) in the emission equations are very sensitive to changes in the work function, it proves more expedient to characterize an oxide cathode by its work function and by the emission current at the operating temperature, for example at \(1000^\circ\mathrm{K}\). If this is done, the results obtained by different experimenters agree considerably better than when the measured values of the coefficient \(A\) are compared. The measured emission currents usually lie within an order of magnitude of the values: \(10\ \mathrm{mA}/\mathrm{cm}^2\) for \(\mathrm{BaO}\), \(1\ \mathrm{mA}/\mathrm{cm}^2\) for \(\mathrm{SrO}\), \(0.1\ \mathrm{mA}/\mathrm{cm}^2\) for \(\mathrm{CaO}\).

On the question of the change of the emission constants in the process of activation there are serious disagreements. Este[^8] comes to the conclusion that activation is determined by the growth of the coefficient \(A\), while the work function remains unchanged. Heinze and Wagener[^17] assert that activation occurs only as a result of a decrease in the work function. Others (for example, Detels[^6], Haxford[^10]) show that during activation both the work function and the coefficient \(A\) decrease in such a way that \(\lg A\) remains a more or less linear function of the work function (Fig. 10).

We do not know the transparency coefficient \(D\) in equation (11.1), and as a consequence cannot make a direct comparison of the measured values of the coefficient \(A\) with its theoretical value. From the works of Berdennikova[^11], Clausing (see de Boer’s book, p. 275),

From Fritsch^14 and Prescott and Morrison^17 we know that \(N_1\) is approximately equal to \(0.5\%\) of \(N_0\), i.e., of the total number of molecules in \(1\ \mathrm{cm}^3\). Thus, \(N_1\) is a quantity of the order of \(10^{20}\). The Prescott–Morrison emission curve can be expressed by the equation

\[ I = 5 \cdot 10^2 T^{5/4} \exp\left(-\frac{16\,500}{T}\right). \]

Comparing it with equation (11,1) and taking \(N_1/N_0 = 0.007\), we find that \(D = 5 \cdot 10^{-2}\). This quantity is very small in comparison with values of \(D\) close to unity, observed for pure metals. Its order of magnitude, however, is close to that observed for contaminated metals (cf. Reimann’s book, reference 13 on p. 37). Still lower values of \(D\) are often observed for oxide cathodes. Thus, for example, the emission straight line given in Fig. 1 (Dejman^9) corresponds to the value \(D = 4 \cdot 10^{-4}\), if we assume, as before, that \(N_1 = 10^{20}\). Such low values of the transparency coefficient may be expected when, owing to the complex structure of the surface, not a simple potential step as shown in Fig. 6 is formed there, but a potential barrier. This case was theoretically analyzed by Fowler (reference 8 and Section 11.31 of the book Statistical Mechanics). He showed that the transparency coefficient must depend approximately exponentially on the external work function, which agrees with the observations of Detels and Haxford. The slope of the straight line expressing the dependence of \(\lg A\) on the work function obtained by Detels must correspond to a potential barrier several ångströms thick.

Fig. 10. Dependence between \(\lg A\) and \(\varphi_1\) (Detels^6)

The activation process can thus be divided into two parts: the first is the increase in the number of conduction electrons as the amount of impurity of free alkaline-earth metal increases, and the second is the decrease in the external work function associated with the formation of a surface potential barrier upon adsorption or reorientation of molecules on the surface. One or another course of the change in the coefficient \(A\) or in the work function during activation will evidently depend on the initial state of the cathode and on the method of separating out the activating impurity. In practice it proves impossible to begin with a completely inactive cathode, since already as a result of the operations of degassing and the first passage of current the activation process begins to some extent. It is therefore not surprising that there are considerable disagreements in the literature concerning the nature of the changes in the emission constants during activation.

b) Effect of the anode voltage

The study of the emission constants of an oxide cathode is further complicated by the circumstance that the curve of the dependence of the anode current on the anode voltage has no sharply expressed saturation (Fig. 11). This phenomenon cannot be attributed to the presence of residual gases, since the anode current nowhere rises above the curve corresponding to the “three-halves” law, as would occur in the presence of positive ions neutralizing, at least partially, the electronic space charge. To some extent this effect is probably determined by additional heating of the oxide layer by the current passing through it and by the associated increase in emission. In addition, the roughness of the oxide surface also plays a role here. Stronger electric fields penetrate more deeply into the depressions on the surface of the coating and cause the emergence of electron current from those parts of the surface which, in weaker fields, are completely shielded by the space charge.

Fig. 11. Thermionic current from an oxide cathode as a function of anode voltage (Blewett¹⁸)

c) Electron temperature

The first investigation of the energy distribution of electrons emitted by an oxide cathode was carried out by Kollmer⁵. He plotted the dependence of the logarithm of the anode current on the anode voltage within the range of small retarding and accelerating voltages. Although the resulting line did prove to be straight, as if the distribution were Maxwellian, its slope corresponded to a temperature of 1760° K, whereas the measured temperature of the cathode was only 1368° K. The same effect was also noted by Rother⁵ᵃ, who carried out the investigation over a whole series of different temperatures, and each time the value of the electron temperature was approximately 1.6 times greater than the corresponding cathode temperature. An explanation of this phenomenon was proposed by Nottingham¹¹˒¹⁵, who investigated an analogous effect for thoriated tungsten. Nottingham attributes the deficiency of slow electrons to the action of a special type of internal reflection, which is not observed in the case of clean surfaces. The coefficient of internal reflection \(R\), introduced by Nottingham, has the form

\[ R = \exp\left(-\frac{E}{C}\right), \]

where \(E\) is the energy of the electron, and \(C\) is a constant. Obviously, as a result of reflection of this type, predominantly slow electrons will be delayed, and the temperature of the emitted electrons will prove to be increased. The Fowler theory of the reflection of electrons from a potential barrier, mentioned in Section 11, a), leads to higher values of the reflection coefficient for lower electron energies. However, the dependence of the reflection coefficient on the electron energy differs somewhat in form from that given by Nottingham.

d) Shot effect

The shot effect in oxide cathodes has been studied in various frequency ranges by Johnson\(^5\), Kozanowski and Williams\(^9\), and others. The work of these investigators revealed anomalous effects of two types. In the region of saturation currents, at frequencies below 5000 hertz, the noise level turns out to be much higher than would be expected from the elementary theory. This effect was studied in detail by Johnson. Theoretically it was analyzed by Schottky\(^ {5a}\), who, in agreement with Johnson, explains it by fluctuations of the thermionic properties of the cathode surface. This interpretation enabled him to draw an analogy with fluctuations of light sources and to call this anomalous phenomenon the “flicker effect,” i.e., “flickering.” Schottky proposes that the fluctuation of emission corresponds to the appearance and disappearance of individual atoms of the active material, apparently barium. On the basis of Johnson’s curves he arrives at the conclusion that the number of such atoms present at any given moment on the surface is equal to approximately one third of the total number of atoms forming the surface. He determines the average lifetime of these adsorbed atoms to be approximately \(0.001\) sec.

The second type of anomalous shot effect is observed when the current strength is limited by the action of space charge. In this current region, as the anode voltage is increased, the noise level first rises to values considerably exceeding the calculated ones, passes through a maximum, and then, when the saturation current is reached, decreases to the theoretical value (provided the frequency is sufficiently high). Kozanowski and Williams explain this effect by the emission of positive ions, which turn out to be, as it were, trapped in the region of the potential minimum formed as a result of the action of space charge. Each ion, owing to its large mass and correspondingly low mobility, can free several hundred electrons from the space charge and thus create a comparatively large fluctuation of the anode current. This hypothesis is supported by further experiments of Kozanowski and Williams, in which they introduced positive ions from an external source into the region of space charge around a cathode made of pure tungsten. The course of variation of the shot effect was then almost the same as in the case of oxide cathodes.

The dependence of the temperature of the noise level, determined by the shot effect, in a commercial triode with an oxide cathode was investigated by Shepesi[^18]. A whole series of phenomena that increase the disturbances of the shot effect are rapidly intensified in those cases when the operating temperature of the cathode lies above or below the region 875–1050°K.

d) Photoeffect and secondary emission

At low temperatures the only sources of photoelectrons are the filled energy band and the unexcited impurity energy levels (see Fig. 6). The photoelectric work function, determined on the basis of the red threshold of the photoeffect, must therefore be equal to the thermionic work function, i.e. to the quantity \(\left(\varphi+\frac{Q_1}{2}\right)\). This apparently is the case. Haxford[^10] measured the photoelectric and thermionic work functions and obtained one and the same value, equal to approximately 1.3 eV.

At higher temperatures a new source of photoelectrons appears, since a considerable number of electrons are already in the conduction band. These electrons require, in order to leave the oxide, an energy equal only to the quantity \(\varphi\). With increasing temperature the number of electrons in the conduction band increases, and therefore one may expect an increase of the photocurrent. This idea was put forward by Morgulis and Nagorskii[^17] to explain the anomalous growth of the photocurrent observed by Bodeman[^8], Newbery[^8], Newbery and Lemer[^11], and others.

An analogous increase with increasing temperature was observed by Morgulis and Nagorskii[^17] also for secondary electron emission from oxide cathodes.

These authors succeeded in measuring the temperature dependence of the “anomalous” part of the secondary emission. A semilogarithmic graph of the dependence of the anomalous secondary emission on \(\frac{1}{T}\) proved to be a straight line with a slope corresponding to an energy of about 0.7 eV.

If this emission is caused by an increase in the number of electrons in the conduction band, then we may conclude that \(\frac{Q_1}{2}=0.7\) eV and \(Q_1=1.4\) eV, which is in good agreement with the value 1.6 eV derived in section 4, a) on the basis of data on electrical conductivity.

e) Change with time

If the full emission current is drawn from an active oxide cathode, the latter begins to fall, at first very rapidly, and then more and more slowly, approaching a constant final value. The initial current may exceed this final value by a factor of ten or more. At the operating temperatures of technical cathodes this decrease of emission occurs so rapidly,

that it cannot be detected without the use of special methods. For lower temperatures this decrease proceeds, however, more slowly, so that at a temperature of \(600^\circ\mathrm{K}\) several hours may be required for the current to fall to a value differing by no more than \(10\%\) from its constant final value. This phenomenon was studied by Becker\(^{8(1)}\), Becker and Sears\(^{10}\), Knipkamp and Nebel\(^{11}\), and Blewett\(^{18}\). Figure 12 shows the emission from an oxide cathode at \(880^\circ\mathrm{K}\) as a function of the time elapsed after application of the anode voltage.

Fig. 12. Fall and recovery of the emission of an oxide cathode (Blewett\(^{18}\)).

The first half of the curve shows the fall of emission when current is drawn. The second half shows the recovery of activity when the cathode is heated without drawing current.

The right-hand part of the curve shows the recovery of emission when the cathode is heated without drawing current.

It was shown (Blewett\(^{18}\)) that the fall of emission depends on the emission current and not on the anode voltage. This indicates that the effect is rather of a bulk than of a surface character. It is possible that the fall of emission is caused either by electrolytic removal of barium from the surface or by electrolytic liberation of oxygen* on the oxide surface with the corresponding neutralization of the active surface layer of free barium. Becker, and Knipkamp and Nebel, prefer the latter possibility, since it is known that during the fall of emission oxygen is liberated. Becker notes, however, that the liberation of oxygen is a surface effect depending on the anode voltage.

Because of the absence of more complete data it does not seem possible to choose between the two alternatives. In any case, the electrolytic current increases the concentration gradient in the oxide layer. Back diffusion of atoms or ions arises, and the decrease of emission ceases when the two rates of transport become equal. The temperature dependence of the recovery process shows that the heat of diffusion is approximately 17 kg cal.

With any change in the temperature of the cathode, the equilibrium between electrolysis and diffusion is disturbed, and the electron emission slowly approaches the value corresponding to the new state of equilibrium. A phenomenon of this type was observed by Davisson and Germer\(^{4}\). It is obvious that in an oxide tube operating in the saturation regime, a change in current will lag behind the changes in electrode potentials or cathode temperature that cause it. But since the majority of technical oxide cathodes operate in regimes in which the current is limited by space charge, these phenomena rarely cause interference.

g) Effect of gases on electron emission

In the presence of active gases, the emission of an oxide cathode may undergo substantial changes. Of all the ordinary gases tested so far, the most destructive effect is apparently exerted by oxygen. When present at a pressure of \(10^{-4}\) mm Hg, oxygen can reduce the emission by several orders of magnitude; a pressure of \(10^{-3}\) mm Hg is sufficient for complete poisoning of the cathode (Koller\(^5\), Reimann and Murgoci\(^9\)). If the oxygen is pumped out after complete or partial poisoning, then by strong heating or by ion bombardment in an inert gas it is still possible to restore the emission. Benjamin and Rooksby\(^ {12}\) note that a poisoned cathode with a BaO coating can always be reactivated, whereas a cathode with a BaO + SrO coating may be poisoned by oxygen so severely that its reactivation proves impossible.

Water vapor is also an active poisoning agent for an oxide cathode (Koller\(^5\)). Qualitatively its effect is analogous to that of oxygen.

According to Koller\(^5\), reducing gases such as CO and \(\mathrm{H_2}\) have a beneficial effect. Arnold\(^3\) asserts that hydrogen is a powerful healing agent, capable of restoring a strongly deactivated cathode. Reimann and Murgoci\(^9\), on the other hand, believe that hydrogen has a somewhat harmful effect on a fully activated cathode.

Neutral gases, such as \(\mathrm{CO_2}\) and \(\mathrm{N_2}\), have little effect on emission. Arnold\(^3\) believes that \(\mathrm{CO_2}\) has a weak poisoning effect. According to Koller\(^5\), however, it has a slight beneficial effect. Prescott and Morrison\(^ {17}\) showed that prolonged exposure in an atmosphere of \(\mathrm{CO_2}\) completely deactivates the cathode. Reimann (reference 13, p. 203) establishes that nitrogen is capable of restoring a deactivated cathode.

Koller\(^5\) studied the influence of argon and found that it slightly improves emission. Heating in methane (Prescott and Morrison\(^ {17}\)) or in any other organic gases or vapors promotes activation to such an extent that sometimes any further activation afterward proves superfluous.

However, apparently no gas is capable of bringing a cathode into a more active state than that which can be achieved by the best methods of activation in high vacuum.

h) Effect of bombardment by positive ions on electron emission

According to Koller\(^5\), bombardment by positive ions of \(\mathrm{CO_2}\) or other gases improves the emission of an oxide cathode. Reimann (reference 13, p. 203) also observed this effect; however, he asserts that prolonged bombardment by positive ions ultimately leads to deactivation of the cathode. Hall\(^8\) notes the harmful effect of bombardment by mercury ions and rarefied gases at ion velocities exceeding 20–25 V.

The harmful effect of prolonged or excessively intense ion bombardment in a rarefied gas may be explained by sputtering of the active surface layer of the cathode.

i) Distribution of emission over the surface

Microscopic examination of oxide cathodes shows that, as a result of the ordinary processes of preparation, the cathode surface turns out to be rough and irregular. As a consequence, both thermal nonuniformities and local concentrations of the electric field arise at the tips of small protrusions. If the coating is especially rough, then at points from which large currents are drawn, local overheating may occur, with a corresponding subsequent increase in the emission current.

Fig. 13. Optical and electron microphotographs of oxide cathodes (Heinze and Wagener^15,18)

a—optical microphotograph of an ordinary oxide cathode; b—electron microphotograph of the same cathode; c—optical microphotograph of a cathode prepared by the Latai and Thomas method; d—electron microphotograph of the same cathode

It is possible that this explains the bright spots which are observed on the surface of the cathode when large currents are drawn from it (cf. Espe^6).

However, even on comparatively smooth surfaces a very spotty emission pattern can be observed with the aid of an electron microscope.

In Fig. 13, a and b, taken from the paper by Heinze and Wagener^15, are shown optical and electron microphotographs of one and the same cathode, from which it follows that there is no correspondence between surface irregularities and points of high emission^1). An exhaustive study of this phenomenon was recently completed by Heinze and Wagener^18, who showed that the nonuniform distribution of emission over the surface is determined chiefly by differences in the work function for the different faces of crystals turned toward the surface.

They showed that the work functions for different faces of a barium-oxide crystal may differ from one another by about 0.4 eV, so that the emission of the cathode is determined almost entirely by the crystals whose low-work-function faces are turned toward the surface.

A series of brilliant electron microphotographs of an oxide cathode may be found in the papers of Brüche and Johannson^11, Richter^12,

^1) Electron-microscopic investigations by N. G. Sushkin, carried out in the MEI laboratory, do not confirm this conclusion of Heinze and Wagener. Translator’s note.

Knecht^13, Brüche^14, Heinze and Wagener^15,18, Benjamin, Haake and Jenkins^17, and many others.

From the photographs in Fig. 13 it is evident that, by means of a special fabrication technique, the nonuniformities in the distribution of emission over the surface can be reduced to such limits that they become no longer noticeable. In Fig. 13, c and d (Heinze and Wagener^18), an optical and an electron microphotograph are shown of a cathode made by the colloidal-deposition method of Patai and Tomasek^15.

k) Effect of particle size on emission

The work of Benjamin, Haake, and Jenkins^17 shows that the effectiveness of a cathode with a BaSr-oxide coating increases as the particle size of the oxide decreases.

Although, on the basis of this work, no continuous regularity can be derived, nevertheless it may apparently be assumed that, when the particle size is reduced from 100 μ to 3–7 μ, the emission increases by approximately a factor of two. Buzag^15 investigated the emission of cathodes whose coatings consist of colloidal particles, and found that, when the particle size is reduced into the region of colloidal dimensions, the cathode effectiveness falls. The maximum emission apparently corresponds to a particle size lying immediately below the upper boundary of the colloidal region. In experiments of this type the emission is measured at constant heating power, so that the temperatures of the cathodes may be unequal owing to differences in emissive power. Therefore it is quite possible that, for differently fabricated cathodes, there is in fact no real difference either in the work function or in the coefficient \(A\).

l) Loss of emission

Any fully satisfactory theory of the activation process must explain the fact that, long before the oxide supply is exhausted, the emission may fall to so low a value that the cathode becomes completely unusable. In its external appearance, a “dead” cathode may remain exactly the same as it was in the active state. Meanwhile, however, it often proves impossible to restore the activity by any of the usual methods.

Klausung believes (see the book by De Boer, reference 1 on p. 283) that such a drop in emission is caused by a decrease in the coefficient \(A\) in the emission equation. The work function apparently remains more or less constant. Evidently, further influx of free \((\mathrm{BaSrCa})\) becomes impossible, and the value \(N_{1}^{1/2}\) in equation (4,2) falls to a very low value.

For cathodes with a mixed oxide coating this phenomenon is quite understandable. Thus, for example, in a cathode with a BaO—SrO coating, chiefly BaO evaporates from the surface layer, and the relative content of the less active ...

SrO (cf. Benjamin and Rooksby\(^ {12}\)). For the phenomenon of emission decline in other types of oxide cathodes, many other explanations have been proposed. Reimann and Murgoci\(^9\) advance the supposition that free oxygen and (BaSrCa) combine within the coating and fill the pores existing in it. The coating ultimately becomes so dense that free diffusion within it proves impossible. If, in maintaining the liberation of free (BaSrCa), an essential role was played by a chemical reaction between the oxide and the core, then it is possible that, with time, the access of the chemically active impurities present in it to the boundary of the core becomes impeded, or that at this boundary a layer of oxide of the core metal is formed, separating the reacting components.

12. CONCLUSION

A large number of different compositions have been tested as materials for thermionic cathodes (cf. Dehman\(^9\)). However, not one of them has proved even approximately as effective as the oxides of the alkaline-earth metals. It is possible that emission from these substances cannot be increased by more than tenfold compared with what has already been achieved. Nevertheless, there still remains a large number of essential questions which must be resolved before the oxide cathode can be considered fully satisfactory and reliably mastered.

A major shortcoming of the experimental work in the past has been the weak study of the mechanical state of the oxide. Often determinations of bulk properties were distorted by phenomena of a purely surface origin, and, moreover, to varying degrees depending on the sizes of the individual crystals. It is obvious that among the tasks of further development there should be included the investigation of the properties of individual crystals of the oxides of alkaline-earth metals. A beginning in this direction was made by Heinze and Wagenrom\(^ {18}\) in their work on the study of thermionic emission from different crystal faces of small barium oxide crystals. There is no doubt that a complete study of the properties of single crystals of the oxides of alkaline-earth metals and the subsequent extension of the information obtained to coatings with a known particle size and with a known porosity will clarify the majority of the contradictions and discrepancies which at present hinder our understanding of the entire complex of phenomena associated with oxide cathodes.

13. BIBLIOGRAPHY

Below is a list of approximately 120 works. For the period since 1925 it is, in our opinion, sufficiently complete. References to earlier works may be found in the review articles included in our list. In parentheses after the title of an article is placed a brief indication of the material contained in it that is of special interest in connection with oxide cathodes.

REFERENCES

1. 1904

A. Wehnelt, Ann. Physik, 14, 425.

2. 1906

F. Horton, Phil. Mag., 11, 505 (Electrical conductivity of CaO and BaO).

3. 1920

H. D. Arnold, Phys. Rev., 16, 70 (Phenomena associated with oxide cathodes).

C. Davisson and H. A. Pidgeon, Phys. Rev., 15, 533 (Activation of a heated tungsten filament in BaO vapor).

3a. 1923

J. W. Mellor, Treatise on Inorganic Chemistry, 3, 660.

4. 1924

C. Davisson and L. H. Germer, Phys. Rev., 24, 666 (Calorimetric determination of the work function for an oxide coating on platinum).

H. J. Spanner, Ann. Physik, 75, 609 (Electrical conductivity and thermionic emission of BaO, SrO, and CaO).

5. 1925

H. A. Barton, Phys. Rev., 26, 360 (Release of negative oxygen ions from oxide cathodes).

J. B. Johnson, Phys. Rev., 26, 71 (“Flicker” effect in oxide cathodes; cf. W. Schottky, 1926).

L. R. Koller, Phys. Rev., 25, 671 (Electron emission of oxide cathodes).

G. Michel and H. J. Spanner, Z. Physik, 35, 395 (Calorimetric determination of the work function for BaO, SrO, and CaO).

5a. 1926

M. S. Glass, Phys. Rev., 28, 521 (Temperature dependence of the work function of an oxide cathode with a platinum core).

H. Rothe, Z. Physik, 36, 737 (Work function of oxide cathodes); Z. Physik, 37, 414 (Electron temperature in oxide cathodes).

W. Schottky, Phys. Rev., 28, 74 (Explanation of the results obtained by J. B. Johnson, 1925).

E. E. Schumacher, J. Am. Chem. Soc., 48, 396 (Melting temperature of CaO, SrO, and BaO).

6. 1927

F. Detels, Z. Hochfrequenz, 30, 10 and 52 (Formation processes in oxide cathodes).

W. Espe, Wiss. Veröff. aus der Siemens-Konzern, 5 (111), 29 and 46 (Emission processes and thermionic constants of oxide cathodes).

W. Statz, Z. techn. Physik, 8, 451 (Methods for preparing oxide cathodes).

7. 1928

J. M. Eglin, Phys. Rev., 31, 1127 (Abstract) (Emission constants of oxide cathodes).

A. W. Hull, Trans. A. I. E. E., 47, 753 (Hollow cathodes).

W. Schottky and H. Rothe, Handb. d. Experiment. Phys., Vol. 13, Pt. 2, pp. 215—232.

H. Simon, Handb. d. Experiment. Phys., 1928, Vol. 13, Pt. 2, 305—314 (Review articles).

M. de K. Thompson and W. G. Armstrong, Trans. Am. El. Chem. Soc., 54, 85 (Elasticity of BaO vapors).

C. Zwikker, Physica, 8, 241 (Elasticity of BaO vapors) (Article in Flemish).

8. 1929

J. A. Becker (1), Phys. Rev., 34, 1323 (Activation and emission of oxide cathodes); (2) Trans. Am. El. Chem. Soc., 55, 153 (“The life history of adsorbed atoms and ions”).

A. Bodemann, Ann. Physik, 3, 614 (Photoelectric effect in oxide cathodes).

W. Espe, Z. techn. Physik, 10, 489 (Thermionic constants of oxide cathodes).

P. H. Fowler, Proc. Roy. Soc., 122, 36 (Thermionic constant \(A\)).

B. Hodgson, L. S. Harley and O. S. Pratt, J. I. E. E., 67, 762 (Preparation of oxide cathodes).

A. W. Hull, Gen. Elec. Rev., 32, 213 and 390 (Phenomena in the bombardment of oxide cathodes by positive ions).

V. C. MacNabb, J. Opt. Soc. Am. and Rev. Sci. Inst., 19, 33 (Activation of oxide cathodes).

K. Newbury, Phys. Rev., 34, 1418 (Photoelectric effect in oxide cathodes).

9. 1930

N. C. Beese, Phys. Rev., 36, 1309 (Ba—Ni alloy as a material for the cores of oxide cathodes).

S. Dushman, Rev. Mod. Phys., 2, 381 (Review article; pp. 418—444 are devoted to oxide cathodes).

A. Gehrts, Z. techn. Physik, 11, 246 (Review article).

H. N. Kozanowski and N. H. Williams, Phys. Rev., 36, 1314 (Shot effect in oxide cathodes).

J. Kroczek u. E. Lübcke, Wiss. Veröff. aus der Siemens-Konzern, 9 (II), 252 (Electrical conductivity and emission of oxide cathodes).

E. F. Lowry, Phys. Rev., 35, 1367 (Role of the core metal in an oxide cathode).

A. L. Reimann and R. Murgoci, Phil. Mag., 9, 440 (Emission and electrical conductivity of oxide cathodes).

E. R. Wagner, Electronics, 1, 178 (Preparation and activation of oxide cathodes).

S. P. Gvozdov, Vestnik elektropromyshlennosti, No. 12.

A. A. Ivanov, Bulletin of the All-Union Electrotechnical Association, No. 8.

10. 1931

W. Albricht, Physica, 11, 146 (Article in the Flemish language) (Relationship between electron emission and the electrical conductivity of the oxide during activation).

J. A. Becker, Trans. Am. El. Chem. Soc., 59, 207 (Electrical conductivity of BaO + SrO).

J. A. Becker and R. W. Sears, Phys. Rev., 38, 2193 (Process of activation of oxide cathodes).

W. S. Huxford, Phys. Rev., 38, 379 (Photoemission of oxide cathodes).

H. Nelson, Physics, 1, 84 (Properties of barium films on tungsten).

K. Newbury and F. Lemery, J. Opt. Soc. Am., 21, 276 (Photoelectric effect in oxide cathodes).

A. L. Reimann and L. R. G. Treloar, Phil. Mag., 12, 1073 (Emission and electrical conductivity of oxide cathodes).

N. H. Williams and W. S. Huxford, Phys. Rev., 37, 463 (Electrical conductivity of oxide coatings).

I. Langmuir and D. S. Villars, J. Am. Chem. Soc., 53, 486.

11. 1932

T. P. Berdennikowa, Sow. Phys., 2, 77 (Chemical detection of free barium in oxide cathodes).

E. Brüche and H. Johannson, Naturwiss., 20, 353, Ann. Physik, 15, 145 (Investigation of oxide cathodes by means of an electron microscope).

A. Gehrts, Naturwiss., 20, 732 (Review article).

PROPERTIES OF OXIDE CATHODES

H. Kniepkampf and C. Nebel, Wiss. Veröff. aus der Siemens-Konzern, 11 (II), 75 (Mechanism of emission of oxide cathodes).
W. Meyer u. Schmidt, Z. Physik, 13, 137 (Electrical conductivity of BaO).
W. B. Nottingham, Phys. Rev., 41, 793 (Emission from complex surfaces).
A. A. Ivanov, Production of thermite tablets for barium lamps, Energoizdat.

12. 1933

M. Benjamin and H. P. Rooksby, Phil. Mag., 15, 810; 16, 519 (Emission and structure of oxide cathodes).
W. G. Burgers, Z. Physik, 80, 352 (Investigation of BaO + SrO cathodes by means of X-rays).
A. Classen u. C. F. Veenemans, Z. Physik, 80, 342 (Elasticities of BaO, SrO, and CaO vapors).
P. Clausing and J. B. Ludwig, Physica, 13, 193 (Radiating power of oxide coatings).
W. Heinze, Ann. Physik, 16, 41 (Calorimetric determination of the work function of oxide cathodes).
E. F. Richter, Z. Physik, 86, 697 (Investigation of oxide cathodes by means of an electron microscope).
J. B. Taylor and I. Langmuir, Phys. Rev., 44, 423 (Properties of cesium films on tungsten).

13. 1934

C. G. Found, Phys. Rev., 45, 519 (Gas-discharge methods for measuring the work function).
W. Knecht, Ann. Physik, 20, 161 (Electron-microscopic images of oxide cathodes).
L. Piatti, N. Cim., 11, 77 (Review article).
A. L. Reimann, Thermionic Emission (Wiley, 1934). (Chapter VI is devoted to oxide cathodes. There is a Russian translation, Gostekhizdat, 1940).
A. A. Shaposhnikov, Electronic and ionic devices, KUBUCh, p. 306.
G. A. Tyagunov, Svetotekhnika, No. 4.

14. 1935

P. A. Anderson, Phys. Rev., 47, 958 (Work function of barium; see Anderson, 1938).
J. A. Becker, Rev. Mod. Phys., 7, 95 (General review on thermionic emission).
M. Benjamin, Phil. Mag., 20, 1 (Influence of metallic impurities in the core of oxide cathodes).
J. H. de Boer, Electron Emission and Adsorption Phenomena, Cambridge (There is a Russian translation, ONTI, 1936).
E. Brüche, Z. Physik, 98, 77 (Investigation of oxide cathodes by means of an electron microscope).
H. Fritz, Mikrochemie, 17, 191 (Microchemical determination of the content of free barium in oxide cathodes).
H. Gaertner, Phil. Mag., 19, 82 (Electron diffraction on oxide cathodes).
K. K. Kelley, Bull. U. S. Dept. of the interior, Bureau of Mines: Bull. 383 — Vapor Pressures of Inorganic Substances; Bull. 394 — A Revision of the Entropies of Inorganic Substances.
A. J. Maddock, Phil. Mag., 19, 422 (Activation of oxide cathodes).
E. Patai u. G. Frank, Z. techn. Physik, 16, 254 (Emission constants of oxide cathodes).
A. L. Reimann, Phil. Mag., 20, 594 (Emission constants of oxide cathodes).
E. Rudberg and J. Lempert, J. Chem. Phys., 3, 627 (Elasticity of barium vapors; cf. Van Liempt, 1936).
W. Schottky, Naturwiss., 23, 116 (Theory of semiconductors).
Fritsch, Ann. Physik, 22, 375.

15. 1936

J. P. Blewett and E. J. Jones, Phys. Rev., 50, 464 (Ionic emission from heated salts).

J. H. de Boer and E. J. W. Verwey, Rec. Trav. Chim. Pays-Bas, 55, 443 (Energy and structure of molecules of alkaline-earth oxides).

A. von Buzágh, Koll. Z., 77, 172 (Influence of particle size on the emission of oxide cathodes).

A. N. Guthrie, Phys. Rev., 49, 868 (Abstract) (Surface ionization of barium on tungsten).

W. Heinze u. S. Wagener, Z. techn. Physik, 17, 645 (Surface distribution of emission of oxide cathodes).

J. A. M. van Liempt, Rec. Trav. Chim. Pays-Bas, 55, 468 (Elasticity of barium vapors).

G. Mönch, Ann. Physik, 26, 481 (Thermoelectric effect for Cu₂O; references to earlier work are included).

W. B. Nottingham, Phys. Rev., 49, 78 (Emission from composite surfaces).

E. Patai u. Z. Tomaschek, Koll. Z., 74, 253; 75, 80 (Preparation of oxide coatings from colloidal particles).

R. H. Fowler, Statistical Mechanics (Cambridge Press, second edition).

16. 1937

G. Hermann, Z. physik. Chem., 35B, 298 (Elasticity of BaO vapors).

H. Isensee, Z. physik. Chem., 35B, 309 (Release of oxygen in the process of activation of the oxide cathode).

E. A. Lederer and D. H. Walmsley, RCA Rev., 2, 117 (Reduction of BaO by tantalum).

M. Schriel, Z. Anorg. Chem., 231, 313 (Search for the lower barium oxide Ba₂O).

G. Bauer, Ann. Physik, 30, 433.

17. 1938

P. A. Anderson, Phys. Rev., 54, 753 (Contact potential difference between Ba and Mg).

C. H. Bachman and C. W. Carnahan, Proc. I. R. E., 26, 529 (Emission of negative ions from oxide cathodes).

M. Benjamin and R. O. Jenkins, Phil. Mag., 26, 1049 (Migration of barium on tungsten and nickel).

M. Benjamin, R. J. Huck and R. O. Jenkins, Proc. Phys. Soc., 50, 345 (Influence of particle size on the emission of oxide cathodes).

J. P. Blewett, Phys. Rev., 53, 935 (Abstract) (Chemical reactions with BaO).

E. U. Condon, Phys. Rev., 54, 1089 (External photoelectric effect in semiconductors).

J. A. Darbyshire, Proc. Phys. Soc., 50, 635 (Electron diffraction on oxide cathodes).

B. Gsaye u. S. Wagener, Z. Physik, 110, 145 (Contact potentials between metals and dielectrics).

W. Heinze and S. Wagener, Z. Physik, 110, 164 (Change of the emission constants of oxide cathodes in the activation process).

W. Heinze u. W. Hass, Z. techn. Physik, 19, 166 (Contact-potential method for measuring temperature).

A. W. Hull, Phys. Rev., 53, 936 (Abstract) (A new type of oxide cathode; “migration cathode”).

J. E. Mayer and I. H. Wintner, J. Chem. Phys., 6, 301 (Elasticities of vapors of halide salts of alkali metals).

N. Morgulis and A. Nagorsky, Techn. Phys. USSR., 5, 848 (Secondary emission from oxide cathodes).

C. H. Prescott and J. Morrison, J. Am. Chem. Soc., 60, 3047 (Chemical measurement of the content of free barium in active cathodes).

R. Suhrmann and C. Frühling, Naturwiss., 26, 108 (Letter) (Emission of oxide cathodes).
A. A. Shaposhnikov, Electron and Ion Devices, Svyaztekhizdat, p. 310.

1. 1933

J. P. Blewett, Phys. Rev., 55, 713 (Decline of emission of oxide cathodes).
J. P. Blewett, Rev. Sci. Inst., 10, 231 (Vacuum balance).
J. P. Blewett, H. A. Liebhafsky and E. F. Hennelly, J. Chem. Phys., 7, 478 (Evaporation rate of BaO).
L. F. Broadway and A. F. Pearce, Proc. Phys. Soc., 51, 335 (Emission of negative ions from oxide cathodes).
R. J. Cashman and E. Bassoe, Phys. Rev., 55, 63 (Photoelectric properties of Ba).
W. Heinze and S. Wagener, Z. techn. Physik, 20, 16 (Investigation of the emission of oxide cathodes by means of an electron microscope).
A. W. Hull, Phys. Rev., 55, 1145 (Abstract) (“disperser” cathode).
C. H. Prescott and J. Morrison, Rev. Sci. Inst., 10, 36 (Temperature scale for oxide cathodes).
Z. Szepesi, Wireless Eng., 16, 67 (Temperature changes of the shot effect in oxide cathodes).
Espé and Knoll, Technology of Electrovacuum Materials, Oborongiz, pp. 102–103.
A. A. Ivanov, Technology of Electrovacuum Production, Oborongiz, pp. 99–102.

1. J. Appl. Physics, 10, 668 and 831, 1939. Translated by S. V. Lobanov and N. G. Sushkin under the editorship of B. M. Tsarev. ↩