New Electrovacuum Devices with a Hot Cathode in a Rarefied Gas

N. D. Morgulis, Kiev

The modern development of the physics of electronic and ionic processes, and in particular the physics of electrical discharges in gases, has led to the development of new types of electrovacuum devices, at the same time considerably broadening the scope of their technical application. Whereas previously the field of application of electrovacuum devices was limited chiefly to low-current electrical engineering (electron tubes), modern electrovacuum devices, while expanding their range of application here as well, are opening new prospects in high-current technology (power mercury rectifiers, both ordinary and controlled; gazotrons, thyratrons, tubes with magnetic control, amplifiers with grid current—Wandstromverstärker, etc.). In the present article we shall confine ourselves to considering only the most interesting new class of electrovacuum devices, namely two- and three-electrode tubes with a hot cathode in an atmosphere of rarefied gas, in particular mercury vapor, known in our country under the names gazotron and thyratron, the development of which is associated chiefly with the names of Langmuir and Hull.

1. Physical Foundations

If, after heating the cathode, one gradually raises the anode potential $v_a$ of a two-electrode vacuum tube, then the strength of the electron current will gradually increase, following the $3/2$ law up to a constant value corresponding to saturation (Fig. 1); this increase occurs, as is known, at the expense of the gradual destruction of the electron space charge located near the cathode.

If, however, there is a rarefied gas in the bulb at a pressure of the order of $1$–$100$ bars and we, after heating the cathode, begin to raise the anode potential, then at first the strength of the electron current will increase according to the $3/2$ law ($AB$ in Fig. 2), and only at a certain moment, when the anode potential reaches

NEW ELECTROVACUUM DEVICES

will reach the value of the so-called ignition potential; an electric discharge will ignite in the lamp, and the current strength will at once increase to the value corresponding to saturation (branch \(CD\), Fig. 2).

In this case, when the discharge is ignited, intense ionization of the gas occurs, and the positive ions that arise thereby compensate the electronic space charge near the cathode. In this process the potential difference in the discharge space is distributed in such a way that almost all of it is concentrated in a narrow region near the cathode; in the rest of the space the potential is almost unchanged and

Fig. 1. Characteristic of a xenotron.

Fig. 2. Characteristic of a gasotron.

corresponds approximately to the potential of the anode. In such a case, according to Langmuir, one distinguishes a space with intensely ionized gas and an almost unchanged potential, called “plasma” (“plasma”), and a narrow region where the main potential drop is concentrated—in the present case near the cathode—the “sheath” (“sheath”).

If we turn to the plasma, then, owing to the strong ionization, it contains large quantities of electrons and ions, which are distributed in such a way that their space charges mutually compensate one another. Owing to causes not yet fully clarified up to the present time

processes the electron velocities, and in the first approximation the ions as well, are established in accordance with the Maxwell–Boltzmann velocity distribution law, under which both electrons and ions may be assigned a certain temperature \(T_e\) and \(T_p\). In that case, if their concentration is denoted by \(n_e\) and \(n_p\), their mass by \(m_e\) and \(m_p\), and the current of the unordered motion by \(I_e\) and \(I_p\), one can obtain for the electrons

\[ n_e=\sqrt{\frac{2\pi m_e}{kT_e}}\frac{I_e}{e} =4.03\cdot10^{13}\frac{I_e}{T_e^{1/2}}. \tag{1} \]

(\(I_e\) in amperes).

As for the ions, their temperature is approximately half the electron temperature, i.e. \(T_p\sim \frac{1}{2}T_e\), and since, on the other hand, their motion at the plasma boundary occurs only in the direction from the plasma toward the film, they can be present in half the quantity, i.e.

\[ n_p=\frac{1}{2}\sqrt{\frac{2\pi m_p}{kT_p}}\frac{I_p}{e} =2.02\cdot10^{13}\frac{I_p}{T_p^{1/2}}. \tag{2} \]

Thus

\[ \frac{I_e}{I_p} =\frac{1}{2}\left(\frac{m_pT_e}{m_eT_p}\right)^{1/2} =\left(\frac{m_p}{2m_e}\right)^{1/2}. \tag{3} \]

For mercury \(\left(\frac{m_p}{m_e}\right)^{1/2}=607\), and therefore from (3) \(\frac{I_e}{I_p}=429\), whereas experiment gives \(\frac{I_e}{I_p}=411\pm17\).

If a metallic surface—a collector, which for simplicity we shall take to be plane—is introduced into the plasma space, then, depending on the potential of this collector \(v_k\) relative to the plasma, we shall have three regions:

I. The collector potential \(V_k\) is considerably more negative than the potential of the discharge space at the given point (branch \(AB\), Fig. 3).

In this case the collector is surrounded by a film of ionic space charge. Between the magnitude of the ion current to the collector \(I_p\), its potential \(V_k\), and the thickness of the film \(d\), there exists a relation given by the usual formula applied in the case of space charge:

\[ I_p=2.33\cdot10^{-6}\frac{V_k^{3/2}}{d^2\sqrt{\frac{m_p}{m_e}}}. \tag{4} \]

(The magnitude of the ionic current \(I_p\) is determined by the rate of replenishment of the stock of ions in the film from the plasma region, i.e., it depends on the degree of its ionization, which in turn depends on the magnitude of the electron current between the main electrodes and on the value of the cathode potential jump, which determines their velocities. Thus, when the collector potential is changed, the thickness of the film will change, being proportional to \(d \sim v^{1/k}\). Exceptionally important in this case is the circumstance that the entire collector potential is compensated by the potential jump in the film of ionic space charge surrounding it, so that the electric field in the space outside the film does not depend on the collector potential and cannot be changed by it; in this case the charge on the collector is equal to the charge concentrated in this film. When the collector potential is changed, in accordance with (4) only the thickness of the film will change, which under certain conditions may reach a considerable value, on the order of up to \(1\ \mathrm{cm}\). This film near the negative collector can sometimes even be observed, and its thickness determined experimentally, since it has the appearance of a dark space—the Langmuir dark space (to some extent analogous to the Crookes dark space in a glow discharge), with a fairly sharply defined outer boundary. In Fig. 4, at right, is shown the Langmuir probe used for such a determination of the thickness of the film of the lamp, and at left is shown its photograph during operation, from which it is evident how clearly the contours of this dark space are visible at the top.

Fig. 3. Typical current-voltage curve of the collector.

In Table 1 a comparison is presented of the experimentally obtained values of \(d\) by Langmuir with the calculated ...

theoretically by formula (4) for a discharge in argon at a pressure of 0.00422 mm of mercury with a tungsten collector.

II. The collector potential is slightly more negative than the space potential (branch \(BD\), Fig. 3). In this case, upon

Fig. 4.

the ionic current of the collector will be superposed by the current of electrons having, according to the Maxwell–Boltzmann distribution, relatively the greatest velocities. The magnitude of just one

TABLE

electron current can be determined by subtracting from the total current to the collector the extrapolated value of the ionic current \(BC\) in Fig. 3. The value obtained in this way

the electron current in this region must obey the Maxwell–Boltzmann law, i.e.

\[ i = S I_0 e^{\frac{eV_k}{kT_e}} \tag{5} \]

(\(S\) is the collector surface), or

\[ \ln i = \mathrm{const} + \frac{eV_k}{kT_e}. \tag{6} \]

In Fig. 5, graphs of this kind are presented, from whose slope one can determine the mean energy or temperature of the electrons from the relation \(\frac{1}{2}mv^2 = eV_e = \frac{3}{2}kT_e\).

Fig. 5. Semilogarithmic current–voltage curve.

(1 volt corresponds to 3750° K); these electron temperatures are of the order of tens of thousands of degrees, decreasing with increasing pressure. Such an increase of the electron current with \(V_k\) continues until the potential \(V_k\) becomes equal to the potential of the surrounding plasma \(V_0\); at this moment, as is seen from Fig. 5, we observe a break in the straight line, from whose position one can determine the potential of the discharge space in which our collector is located (in the case shown in Fig. 5, \(V_0 = -12.5\) V).

III. \(V_k > V_0\). In this case the collector will be surrounded by a film of negative space charge, in which the conditions are determined analogously to the first case, again by equation (4), and the magnitude of the electron current will be approximately constant, corresponding to the current due to random-

of motion in the plasma \(l_c \sim \left(\dfrac{m_e}{2m_p}\right)^{1/2} I_p\). And in this case it is sometimes possible to observe this film near the collector in the form of a dark space. With a further increase of \(V_k\), ionization may begin in the space near the collector, which will become covered with a luminous film (analogous to anodic glow). If, with a further increase of \(V_k\), the number of ions produced by ionization near the collector reaches \(\sim \left(\dfrac{m_e}{m_p}\right)^{1/2}\) of the number of electrons in the plasma, the electron space charge in the film will be compensated by them, and the strength of the electron current will at once increase greatly—a peculiar breakdown of the film will occur (arrow in Fig. 3).

Another set of questions important for us is connected with the sputtering of adsorbed films located on the surface of an incandescent cathode operating in an atmosphere of rarefied gas. It is known that, under the influence of bombardment by positive ions, the cathode of a discharge tube is gradually sputtered; in this case the degree of sputtering depends on the cathode material, the kind of gas, and the velocities of the ions bombarding it. Experiments have shown that the amount of cathode material sputtered per unit time is expressed approximately by the formula:

\[ m=a(V-V_0), \tag{7} \]

where \(a\) is a constant, \(V\) is the energy of the ion striking the surface of the cathode, and \(V_0\) is the critical potential below which sputtering practically ceases. The destructive action of bombardment of the cathode surface by positive ions must naturally be especially strong for cathodes activated by means of an adsorbed electropositive film, in particular thorium or barium.

TABLE 2

In the investigation of this important question, works by Kingdon and Langmuir, on the one hand, and Hull, on the other, were published. The first authors showed that, in the case of the operation of a cathode made of thoriated tungsten in an atmosphere of highly rarefied gas, relation (7) is applicable with the critical potentials \(V_0\) given in the second column of Table 2. In the following—

In Hell’s investigations it turned out that the values of these critical potentials, which are very important for practical use, lie somewhat lower; these values are given in the third column of the table.

In Fig. 6 are shown, according to Hell, the volt-ampere characteristics of a cathode made of thoriated tungsten, operating in an atmosphere of mercury vapor at a pressure of \(0.005\) mm, at different cathode temperatures. We see how the electron current at first gradually increases, following the \(3/2\) power law; at a certain

Fig. 6. Characteristics of a thoriated filament in mercury vapor at \(0.005\) mm pressure.

moment, when the velocity of the ions striking the cathode, which have arisen owing to ionization of the gas by electron impact, reaches a critical value, sputtering of the thorium film begins—the cathode is deactivated, and the emission rapidly falls. A process of the same kind will also occur with an oxide cathode having on its surface an adsorbed film of barium.

II. Gasotron

In our terminology, a gasotron is a two-electrode tube—a diode with a heated cathode in mercury vapor at low pressure. The unipolar character of the conductivity of the ordinary vacuum diode—the kenotron, whose characteristic was presented in Fig. 1, long ago led to its use as an alternating-current rectifier. When a kenotron operates in an a.c. circuit equal to \(E\), between its electrodes in

in the active direction there will be some potential difference \(V_a\), necessary for the current \(I\) to pass through it; in this case its efficiency may be approximately determined from the formula:

\[ \eta = 1 - \frac{W_n + I V_a}{IE}, \tag{7} \]

where \(W_n\) is the power consumed for heating. To obtain a higher efficiency, it would be necessary to use economical cathodes of the thoriated or oxide type, on the one hand, and to develop a design that would make it possible to obtain considerable currents at small potentials \(V_a\). The first approach, in powerful kenotrons, encounters considerable difficulties because of the danger of sputtering of the active cathode film under bombardment by ions of residual gases, which appear during operation of the kenotron,—owing to the large \(V_a\) of the kenotron, the effectiveness of ion bombardment in this case will be very considerable. On the other hand, obtaining large electron currents in kenotrons requires the presence of very high anode potentials \(V_a\), as is seen from Fig. 1, which are necessary for extracting the corresponding number of electrons from the space-charge zone, which also has a very unfavorable effect on its efficiency. The use, for rectification purposes, of a mercury arc, which gives a good efficiency, has a number of inconveniences connected primarily with the need for artificial cooling, the relative instability of the arc discharge, and difficulties of manufacture. A radical way out of the situation is the introduction into the kenotron of a small quantity of gas or mercury vapor at a pressure of the order of \(0.01\) mm—a gasotron, the characteristic of which has already been given in Fig. 2 and discussed above. The fundamental difference between a gasotron and a kenotron is as follows: in a kenotron, a change in the strength of the electron current \(I\) occurs when the anode potential changes,—for a given kenotron design, the value of \(I\) is determined only by it (of course, in the absence of saturation). In a gasotron, however, upon ignition of the discharge the strength of the anode current at once increases to saturation; intermediate values of the current strength can be obtained, as already indicated, by stabilizing the discharge with an external resistance,—thus the magnitude of the anode current will be determined only by this external resistance, while the potential difference between the electrodes throughout this entire region will be approximately constant and will not depend on the current strength. Further, it proves possible to use oxide cathodes in gasotrons with a design that ensures exceptionally high economy. From what has been indicated above it is clear that if, during operation of the gasotron, the potential difference

between the electrodes, concentrated, as we know, in the film around the cathode, does not exceed the critical value of sputtering, which for mercury vapor corresponds to 22 V, then the energy of the ions will prove insufficient to sputter the barium film, and the oxide cathode can operate calmly. This turns out to be quite possible to do, and thus gasotrons with a low ignition potential within the limits of 12–20 V

Fig. 7. Directly heated cathodes. Indirectly heated cathodes.

operate quite stably. Further, owing to the fact that the electron current here is caused chiefly by compensation of the electronic space charge by ions, in this case it proves possible to provide special cathode designs with good thermal insulation, ensuring exceptionally great economy and the possibility of obtaining very large electron currents at the same low ignition potential. The arrangement of such cathodes is shown in Fig. 7, where cathodes with ordinary heating are depicted at the top, and indirectly heated cathodes at the bottom; thanks to reliable thermal insulation and the use of the energy of the dissipated ray emission, it is possible with cathodes of the latter type to obtain up to 1 A of electron current at 1 W

incandescence, whereas the use of such cathodes in kenotrons is completely impossible because of the very large internal resistance that a kenotron would have in this case. Moreover, in some cases it is possible to maintain the cathode at operating incandescence exclusively by means of the emission current itself, as is clearly seen from Fig. 8, where the abscissa gives the electron-current intensity, and the ordinate gives the incandescence voltage (solid curve) and incandescence power (dotted curve).

It is clear from all that has been said that the gazotron has enormous advantages over the kenotron; this is emphasized still more by the data of a comparison of two rectifier circuits for three-phase current with six tubes, each circuit delivering 180 kW of rectified-current power, the first operating with kenotrons and the second with gazotrons, as given in Table 3.

Fig. 8. Average value of the constant anode current. Voltage curve in the heating circuit.

current, with six tubes, in circuits each delivering 180 kW of rectified-current power, the first operating with kenotrons and the second with gazotrons, as given in Table 3.

Exceptionally important for stable and reliable operation of a gazotron is the choice of the corresponding mercury-vapor pressure. The point is that, with an increase in the mercury-vapor pressure or in the condensation temperature of the mercury drop, the burning potential of the gazotron will decrease, which is undoubtedly advantageous. However, it is necessary to take into account here that at the same time there will also decrease the

TABLE 3

potential of reverse ignition, which is permissible only up to certain limits. Thus the upper limit of the elasticity of mercury vapor is set by the reverse-ignition potential, which, as is generally accepted, must have a certain safety margin; the lower limit of the vapor elasticity is set by the burning potential in the active direction, which must not exceed 22 V, at which sputtering of the active cathode begins.

In Fig. 9 are presented curves of the dependence of the burning potential \(II\) and the reverse-ignition potential \(I\) on the temperature of the condensing mercury, for the gasotron UV-869 of the General Electric Co, designed for 20,000 V; from the figure it is seen that the working region lies approximately at \(5—50^\circ\) C ambient temperature, i.e., fortunately, it lies precisely at the usual temperatures of workrooms. Further, it is very important to select properly the heating of the cathode and its operating regime, since with underheating of the cathode or with its overload we may enter the region \(CD\) of Fig. 2; in this case the potential jump in the film may turn out to be greater than the critical sputtering potential, and the active film on the cathode will be destroyed. With this brief description of the gasotron we shall confine ourselves.

Fig. 9. Dependence of the voltage drop (curve \(II\)) and reverse-ignition voltage (curve \(I\)) on temperature.

III. Tyratron

If a grid is placed in the space between the anode and the cathode of a gasotron, then the resulting triode with mercury vapors is called a tyratron; this name is given from the Greek \(\theta\acute{\upsilon}\rho\alpha\)—door, because of its characteristic feature, indicated

below. However, the role of the grid in this case differs radically from its role in an ordinary vacuum triode; this is evident from consideration of the characteristics of the triode (a) and the thyratron (c) shown in Fig. 10. If a negative bias is applied to the grid \(V_c\) and a positive potential to the anode \(V_a\), and the negative potential \(V_c\) is then gradually decreased, the electron current in the triode will appear at the moment corresponding to the condition

\[ V_c=-D V_a, \tag{8} \]

where \(D\) is the permeability, and will then gradually increase, reaching the value corresponding to saturation; when the grid potential is varied back, the points lie exactly on the curve obtained in the forward direction, so that we do not obtain even a hysteresis loop indicating a poor vacuum in the triode.

Fig. 10. Characteristics of a vacuum amplifying tube (a), a gas-filled amplifying tube (b), and an ion tube (c).

The situation will be entirely different in the case of the thyratron. If, again, after heating the cathode of the thyratron, a considerable negative bias is applied to the grid and then a positive potential to the anode, then, as in the case of the triode, the grid will lock the electrons at the cathode, and no current will flow through the thyratron. If, however, the negative grid potential is then decreased, then at the moment corresponding to condition (8) an electron current will appear which, in a very short time, will ignite the discharge in the thyratron. The ions that appear in this process will be attracted to the negative grid, which, playing the role of a negative collector, will immediately become covered with a film of ionic space charge in such a way that the entire potential drop \(V_c\) will be concentrated in this film and will no longer affect the discharge. Owing to this, the anode current at the moment of ignition corresponding to condition (8) will at once rise to its full value, which, as we already know, will be determined not by saturation but by the magnitude of the external resistance in the anode circuit. The time required for such complete ignition of the discharge in the thyratron is, depending on the conditions, of the order of 1–10 microseconds. Further change of the grid potential in either direction po-

can no longer affect the discharge, and will cause only a change in the thickness of its film of space charge in accordance with formula (4). Thus the fundamental difference between the thyratron and the triode consists in the fact that, with the usual supply of the electrodes by direct current, by means of the grid in the thyratron we can only switch on the discharge, but thereafter we can influence neither the magnitude of the anode current nor even switch it off (Fig. 10,c); this characteristic property determines its name. To switch off an ignited discharge in a thyratron there is only one possibility—to switch off the anode voltage or make it negative; in that case the positive ions and electrons present in the discharge space will disappear on the walls and electrodes, and we shall again arrive at the initial state. Here one should pay attention to the following circumstance: the cessation of the discharge at a negative grid potential greater than (8) can occur only if, during the time when the anode potential was switched off, the ions and electrons have time to diffuse to the walls and electrodes and disappear on them; if this time is made so small that at some moment it proves insufficient for the disappearance of the ions and electrons, then the discharge, after such a brief switching-off of the anode potential, will again continue. This time, necessary for the disappearance of ions and electrons after the current in the thyratron is switched off, is called the deionization time; according to Hull’s measurements it is of the order of 10 to 1000 microseconds, and depends on the geometry and potential of the electrodes and on the gas pressure, and is expressed approximately by the following empirical formula:

\[ \tau=\frac{0.0012\,pI^{0.7}}{V_c^{1/2}\,x}\ \text{sec.}, \tag{9} \]

where \(p\) is the gas pressure, in bars, \(I\) is the current strength in amperes, \(V_c\) is the grid potential with respect to the discharge space—the plasma—in volts, and \(x\) is the distance between the grid and the anode.

The ignition of a discharge in a thyratron is thus determined by a condition analogous to (8), i.e., it occurs at the moment when the potential on the grid reaches the value

\[ V_з=-DV_a; \tag{10} \]

in this case the quantity \(D\), in contrast to the case of the triode, depends not only on the geometry of the electrodes alone, but also on the conditions of gas ionization and its elasticity. Practically, however, this quantity may be regarded as constant, but in order to emphasize the essential difference between its role in the case of the thyratron and in the case of the triode, it is here called the “fac-

grid-control ratio” — f. c. r. (“grid-control ratio”).

In Fig. 11 is shown the experimentally obtained dependence of the ignition potential of the grid \(V_3\) on \(V_a\) for a thyratron, from which it is clear that, in accordance with (16), it has an actually linear character, by whose angular coefficient the f. c. r. can be determined; only at small \(V_a\) is a deviation from the linear dependence observed, connected, apparently, with a decrease in the probability of ionization by the impact of an electron having a low velocity.

Fig. 11. Control characteristics of a thyratron, representing the grid voltage at which a current arises, as a function of anode voltage.

We see that, with the ordinary circuit for supplying the electrodes of a thyratron, as also in a direct-current triode, the possibility of controlling the anode current by means of the grid potential is completely excluded—in this way one can only switch on the current, and cannot not only regulate its magnitude, but even switch it off. If, however, the electrodes of the thyratron are supplied with alternating current, then the possibility arises of regulating the strength of the anode current by means of the grid potential; for this there are two possibilities:

1. Let us apply an alternating voltage to the anode; for simplicity let us assume it sinusoidal, \(V_a = V_{a0}\sin \omega t\), the oscillogram of which is shown in Fig. 12.

In such a case the ignition potential of the grid \(V_3\) (10) will also have a sinusoidal character:

\[ V_3 = -D V_{a0}\sin \omega t. \]

and is directed to the side opposite to \(V_a\); in the same figure its curve is shown by the broad dashed line. Let us also apply to the grid an alternating voltage \(V_c\) with some phase difference \(\varphi\) with respect to the anode voltage. If now, at some instant, this grid voltage \(V_c\) becomes equal to or more positive than the ignition voltage \(V_z\), i.e., in the figure the curve \(V_c\) intersects \(V_z\), then at that instant a discharge will be ignited, which will continue throughout this entire half-period as long as the anode voltage is positive—it will go out only when this positive half-period ends and \(V_a\) becomes negative. At the instant of ignition of the discharge, the anode voltage immediately drops from the value it had at that moment to the normal burning value of 12–20 V and remains at this level all the time while the discharge continues. A direct-current instrument, for example a d’Arsonval meter, in the anode circuit will show, as is evident from the figure, some value that will correspond to the mean value of those current pulses which flow during each positive half-period upon ignition; the oscillogram of the anode current is given below the oscillograms of the voltages. It is now quite clear that if the phase difference between the grid and anode voltages is changed, then the curves \(V_c\) and \(V_z\) will intersect at different places, and thus one can change the magnitude of the current pulse flowing during each positive half-period, and thereby, consequently, the mean value of the anode current. In Fig. 12 are shown the graphs for three cases, when the phase difference between \(V_c\) and \(V_a\) is \(0^\circ\), \(60^\circ\), and \(120^\circ\), from which this idea of regulating the strength of the anode current is quite clear—in contrast to the triode, here we can regulate only the mean value of the anode-current strength, and not the instantaneous value, as in a triode.

Fig. 12. Ignition displacement with a change in the phase of the alternating voltage on the grid.

In Fig. 13a is shown the dependence of the mean and effective value of the anode-current strength on the phase difference between the grid and anode voltages, whence it is evident that these quantities vary smoothly within the interval between \(0^\circ\) and \(180^\circ\), and then at \(180^\circ\) undergo an abrupt jump from zero to their full value, thereafter remaining unchanged up to \(360^\circ\).

These dependences may be represented by the expressions:

\[ \overline{I}=\frac{I_{\max}}{2\pi}(\cos\varphi+1);\qquad I_{ef}=\frac{I_{\max}}{2}\sqrt{1+\frac{\varphi}{2}+\frac{\sin 2\varphi}{2\pi}}. \tag{11} \]

The case when \(\varphi=180^\circ\) is also interesting because here, by small changes of the phase difference, it is possible either to increase or to switch off the output current completely at once—the relay case.

1. Another method for regulating, again, the mean value of the anode-current strength in a thyratron is as follows. Let us apply to the anode and grid alternating voltages with some phase difference, say \(90^\circ\); the ignition conditions in this case are shown in Fig. 13, b.

If now some constant voltage is superposed on this alternating grid voltage, then, if this additional bias is positive, the intersection of the curves \(V_c\) and \(V_3\) will be pro-

Fig. 13. Displacement of ignition when a voltage superposed on an alternating voltage is applied to the grid. Constant bias on the grid: \(a\) positive, \(b\) equal to zero, and \(c\) negative.

Fig. 14. Control curves of an ionic tube with the grid voltage shifted in phase.

will occur more readily than before, the current impulse and its mean value, as is seen from Fig. 13, a, will increase. If, however, the added constant bias is negative, then the current will decrease (Fig. 13, c). In this way, by varying in the usual manner the constant grid bias, one can gradually vary the magnitude of the anode current, with a constant phase difference between the alternating grid and anode voltages.

Fig. 15 gives the dependence between the mean value of the anode current and the constant grid potential, i.e., to a certain extent the ordinary characteristic of a thyratron, for various phase differences between the alternating grid and anode voltages.

A very convenient form is possessed by the characteristic at \(120^\circ\)—an almost rectilinear change of the current from zero to the full value, which here, as in the case of the gasotron, is determined by the external resistance. The characteristic at \(180^\circ\) indicates the convenience of using in this case the thyratron as a relay, when, with a small change in the grid potential, it is possible at once to switch on and switch off (in contrast to the case of supplying the electrodes with direct current) the thyratron current from zero to its full value.

Fig. 15. Control curves of an ionic tube when combining a constant voltage with sinusoidal alternating voltages of various phases on the grid.

A photograph of two thyratrons with opened electrodes is given in Fig. 16.

Let us dwell briefly on a device similar to a certain degree to the thyratron, the so-called grid-current amplifier (“Wandstromverstärker” according to Shotky), whose construction is shown in Fig. 17 on the left, and its photograph on the right.

Between the main electrodes \(K\) and \(RA\) a discharge is ignited in mercury vapor—the cathode here may be either an incandescent spot on liquid mercury or an ordinary heated oxide surface. At the side walls of the bulb, in the plasma, a grid \(G\) and a working anode \(A\) are placed close to one another. If a constant positive potential is applied to the anode, and a constant, say negative, potential to the grid, then, in the presence of a discharge between the main electrodes, the grid is covered with a sheath of ions of definite thickness. The field of the work-

as a result of which the anode will attract to itself a certain number of electrons, which, moving toward it, will pass through the turns of the grid through the space between the sheaths of positive ions surrounding the neighboring turns. If the negative potential of the grid is increased more and more, then, in accordance with (4), the thickness of the sheath will become ever greater, the space between neighboring sheaths ever narrower, and the electron current to \(A\) will decrease. Finally, at a large negative grid potential the thickness of the sheath around a turn of the grid may become so considerable that neighboring ones will merge into a single whole; between

Fig. 16. A typical thyratron with a heated cathode.

the anode and the plasma there will then be, as it were, a continuous screen with zero potential, and the current to the anode will cease. An exceptionally essential condition for this is that, in the space between the grid and the working anode, no additional ionization of the gas occur by the electrons flying to the anode, since then in this space we shall obtain an additional discharge, supplying the grid with an additional quantity of ions—the strength of the ion current to the grid will increase, the thickness of the sheath according to (4) will decrease, and we shall no longer be able, as in the tyratrone, to regulate the strength of the anode current. This condition will be fulfilled if the distance between the grid and the anode \(A\) is made very small, less than the free path length of an electron in the gas of the tube. Such a tube makes it possible,

It is entirely analogous to a triode to regulate the instantaneous value of the anode-current strength by means of the grid potential,—its characteristic is given in Fig. 10, b, and the parameters have the following order of magnitude: transconductance \(S \sim 0.5\) ampere/volt, internal resistance \(R_i \sim 70\,\Omega\), permeability \(D \sim 2—5\%\).

IV. Applications of Tyratrons

From consideration of the features of tyratrons it is clearly evident that the field of their application must differ from the field of application of electron tubes, and, unlike them, may lie not only in the field of electrical engineering of weak currents, but also of strong currents. First of all, the possibility of their application as relays catches the eye,—if in an ordinary powerful triode, when the grid \(V_c\) is changed by 1 V, the electron-current strength \(I\) changes by approximately several milliamperes, then in a tyratron, with an appropriate choice of the operating point, by changing \(V_c\) by 0.1 V one can switch on and off currents of the order of tens of amperes, using for this at least the characteristic at a phase difference between the grid and anode potentials of \(180^\circ\) (Fig. 15). On the other hand, using a tyratron as an alternating-current rectifier, one can regulate the strength of the rectified current, at least by a corresponding change in the phase difference between the grid and anode potentials \(V_c\) and \(V_a\). In both of these cases, for changing the conditions in the grid circuit it is very convenient to use photoelements connected in the appropriate manner. Without touching on other numerous possibilities for the application of tyratrons, we shall dwell only on two that are of interest and characterize the field of its application in technology and in the methodology of laboratory experiment, namely on its application as a converter of direct current into alter-

Cross section of a gas amplifier.

External view of an amplifier with a metallic, water-cooled anode.

Fig. 17.

...variable and for use in the counting and recording of rapidly occurring processes.

1. The problem of converting direct current into alternating current is one of the urgent problems of modern high-current electrical engineering. Questions of the electrification of large regions, in particular the Urals–Kuzbass, put on the agenda the question of transmitting electrical energy by direct current at extra-high voltage, which has a number of advantages over transmission by alternating current. However, this question comes down to the necessity of designing reliable and economical converters of direct current into alternating current—inverters; and in this task the thyratron, evidently, will play a considerable role. We shall therefore consider one of the possible inverter circuits, shown in Fig. 18.

Fig. 18. Circuit for converting direct current into alternating current.

Let us suppose that at some instant one of the thyratrons, say the left one, is ignited, while the right one is extinguished, and that the grid of the left thyratron is positive, that of the right one negative. If the e.m.f. of the direct-current source is \(E\), and the voltage drop across the ignited thyratron is \(V_a\) (approximately 15 V), then the capacitor \(C\) will charge to the voltage \((E - V_a)\), its left plate having the potential \(V_a\), and its right plate \(+E\). After a time interval corresponding to a half-period of the alternating grid voltage, the right grid becomes positive and the left negative, and as a result the right thyratron is ignited. At this moment the right plate of \(C\) assumes the lower potential \(V_a\), i.e., it undergoes a drop of \((E - V_a)\) volts, and a similar potential drop must occur instantaneously on the left plate as well, whose potential becomes equal to \((-E + 2V_a)\) volts. As a result, a potential \((-E + 3V_a)\) is applied for a short time to the anode of the left thyratron, and the thyratron is extinguished; if during this time, which is determined by the time constant \(CR\), deionization has time to occur, then by the moment when the potential on the left plate has leveled off and reaches its normal value, now equal to \(E\), the left thyratron will no longer ignite—

will ignite. After half a period, the grid of the left thyratron becomes positive and that of the right one negative; a process analogous to that described will occur, as a result of which the left thyratron will again ignite and the right one will go out, and so on. Owing to this alternating ignition and extinction of each of the thyratrons in the primary circuit of the transformer, we obtain a current of alternating direction with a frequency corresponding to the frequency of the alternating \(V_c\), whose curve depends on the load and has a large number of harmonics. The circuit of such an inverter can be simplified still further if, instead of a transformer supplying the alternating grid voltage, a corresponding discharge circuit is arranged in the grid circuit, whose time constant \(CR\) will determine the time necessary for the grid potential of the extinguished thyratron to again reach the ignition potential, i.e., by this the frequency of the alternating current obtained will be determined. Naturally, the upper limit of the frequency obtained by means of an alternating-current inverter will be limited by the deionization time.

Fig. 19. Direct-current converter.

With the aid of such an inverter and at least a gasotron for rectifying alternating current, it is easy to realize the construction of a direct-current transformer, one of the possible circuits of which is shown in Fig. 19.

Here, in the input circuit, the supplied direct current is converted by the inverter into alternating current, which is transformed to the required voltage and again rectified.

Thanks to the high efficiency (for example, an inverter with thyratrons has \(\eta = 99.87\%\)), such circuits with thyratrons, after certain improvements, will probably find significant application in engineering. According to the literature, the largest thyratron was constructed in 1929.

Hollom was designed for \(100\ \mathrm{A}\) and \(20\,000\ \mathrm{V}\); it had a metallic housing, which itself served as the grid,—if we recall that the voltage drop across the thyratron is of the order of \(12\)—\(20\ \mathrm{V}\), then the efficiency of such a thyratron will be exceptionally high; in Holl’s opinion there are no difficulties in manufacturing units of 10 and even 100 times greater power.

Fig. 20.

1. As a second illustration of the application of the thyratron, let us describe one interesting use of it in the technique of physical experiment, for the counting and recording of rapidly occurring processes. This method was proposed by Wynn-Williams and was applied by him for counting \(\alpha\)-particles.

In Fig. 20 its circuit is shown; the thyratrons \(P, Q, R, S\), and \(T\) are connected successively in a ring,—the impulse being recorded acts simultaneously on the grids of all the thyratrons. A large negative bias is applied to the grid of each of the thyratrons, so that at first they are all extinguished. If one of the thyratrons is ignited, say \(Q\) (for example, by grounding its grid for a moment), then, owing to the voltage drop across the resistance in its circuit, the grid potential of the following thyratron \(R\) changes, becoming close to the ignition threshold of the discharge in it, while the left plate,

New Electrovacuum Devices

Fig. 21.

Notes visible in the diagram:
- Direct-current path
- Shift to the grid
- \(I_{0m}=+220\) volts
- \(I_{0m}=0\)
- \(H+\), \(H-\)
- \(3\) to \(6\) volts; \(30\) to \(50\) volts; \(60\) volts
- \(100000\ \Omega\), \(10000\ \Omega\), \(20000\ \Omega\), \(1000\ \Omega\), \(300\ \Omega\)
- \((0.01\ \mu f)\), \((0.1\ \mu f)\), \((0.25\ \mu f)\), \(2\ \mu f\)
- \(A, B, C, D, E, F\); \(R_1, R_2, R_3, R_4, R_5, R_6, R_7\); \(C_a, C_b, C_c, C_d, C_e, C_f, C_g\); \(G_b, G_c, G_d, G_e, G_f\); \(K_b, K_c, K_f\); \(S, S_a, S_b\)

capacitor \(K_{QK}\) and the right plate \(K_{TM}\) are charged to the potential \((E - V_a)\), while their other plates are at zero potential. If now the signal to be registered arrives, it is sent to it and the thyratron that has been prepared for it, i.e. \(R\), is ignited. At this moment the right plate of the capacitor \(K_{QK}\) is charged to the potential \((E - V_a)\), and, consequently, the left plate is for a short moment, determined by the time constant \(RK_{QK}\), at the potential \(2(E - V_a)\), i.e. at the cathode of thyratron \(Q\) there is applied a considerable positive potential equal to \((E - 2V_a)\), owing to which thyratron \(Q\) is extinguished. Thus, in such a process, as a result of the action of the registered pulse, thyratron \(R\) is ignited, \(Q\) is extinguished, and \(S\) is prepared to receive the next pulse. The thyratrons are connected in a ring in such a way that thyratron \(T\) is connected with thyratron \(X\), in whose anode circuit there is a mechanical counter, and with thyratron \(P\) for closing the ring. It is obvious that in this way the resolving power of the counter is greatly increased, since the actual number of signals is as many times greater than the number registered by it as there are thyratrons in the ring. With such a circuit it proved possible to bring the resolving power of the circuit (by this should be understood the shortest time possible for separating and separately registering two pulses) to \(1/600\)—\(1/700\) sec.

Another circuit, proposed by Wynn-Williams, is presented in Fig. 21. It consists of three cascades with two thyratrons in each. The idea of this circuit is as follows: each cascade operates quite analogously to the circuit shown in Fig. 18, i.e. owing to the incoming pulses, first one and then the other thyratron will be ignited alternately. Each following cascade is connected to one of the thyratrons of the preceding one, owing to which the resolving power, with the addition of a new cascade, is doubled. In fact, if signals \(1, 2, 3, 4, 5\ldots\), etc., are applied to the circuit, then thyratron \(A\) registers signals \(2, 4, 6\ldots\), and thyratron \(B\)—\(1, 3, 5\ldots\); the thyratrons of the second cascade register only the signals that ignite thyratron \(A\), i.e. thyratron \(C\)—\(4, 8, 12\ldots\), and thyratron \(D\)—\(2, 6, 10\ldots\); the thyratrons of the third cascade register only the signals that ignite thyratron \(C\), i.e. thyratron \(E\)—\(8, 16, 24\ldots\), and thyratron \(F\)—\(4, 12, 20\ldots\). Thus, with three cascades, the actual number of signals is \(2^3\), or 8 times, greater than the number of registrations of the mechanical counter included in the circuit of thyratron \(E\). In this way the resolving power of such a circuit was brought to \(1/250\) sec.

Thus, as is clear from all the foregoing, the appearance of the thyratron should be regarded as a serious success in

field of study and practical application of one of the most serious areas of modern physics—the electrophysics of discharge in a rarefied gas; despite their youth, the prospects for their application both in technology and in the methodology of laboratory research are very considerable, and undoubtedly their ubiquitous, broad use should be expected in the near future.

Literature

1. K. Compton and I. Langmuir, Rev. Mod. Physics, 2, 123, 1930 (for the Russian translation, see Uspekhi Fizich. Nauk, 1931).
2. L. Davies, Electrician, 106, 936, 1931.
3. L. Dunoyer et P. Toulon, Journ. de phys., 5, 257, 289, 1924.
4. A. Glaser, E. T. Z., 52, 829, 1931.
5. A. Güntherschulze, Elektrische Gleichrichter und Ventile, 1929.
6. A. Güntherschulze, Zs. f. Phys., 62, 619, 1930.
7. A. Hull and W. Winter, Phys. Rev., 21, 211, 1923.
8. A. Hull, Journ. Am. Inst. El. Eng., 47, 798, 1928.
9. A. Hull, Trans. A. J. E. E., 47, 753, 1928.
10. A. Hull, Gen. El. Rev., 32, 213, 390, 1929.
11. A. Hull and J. Langmuir, Proc. Nat. Acad. Sc., 15, 218, 1929.
12. A. Hull and L. Snoddy, Phys. Rev., 37, 1791, 1931.
13. A. Hull, Physics, 2, 409, 1932.
14. K. Kingdon and Langmuir, Phys. Rev., 22, 148, 1923.
15. K. Klemperer und E. Lübcke, Arch. f. Elektrotechn., 26, 67, 1932.
16. I. Langmuir, Journ. Frankl. Inst., 196, 751, 1923.
17. I. Langmuir and H. Mott-Smith, Gen. El. Rec., 27, 449, 36, 616, 782, 810, 1924.
18. I. Langmuir, Zs. f. Phys., 46, 271, 1927.
19. I. Langmuir, Phys. Rev., 33, 954, 1929.
20. E. Lübcke, Zs. f. techn. Phys., 8, 445, 1927.
21. E. Lübcke und W. Schottky, Wiss. Ver. Siemens-Konzern, 9, 390, 1930.
22. E. Lübcke, E. T. Z., 52, 1513, 1931.
23. Mazda-gyarporh., Journ. Sc. Inst., 8, 296, 1931.
24. G. Mierdel, Phys. Zs., 28, 344, 1927.
25. H. Mott-Smith and I. Langmuir, Phys. Rec., 28, 727, 1926.
26. W. Nottingham, Phys. Rev., 37, 1790, 1931.
27. W. Nottingham, Journ. Frankl. Inst., 211, 271, 1931.
28. W. Nottingham, Journ. Frankl. Inst., 211, 71, 1931.
29. D. Prince, Gen. El. Rev., 13, 347, 19–S.
30. A. Samuel, Rev. Sc. Instr., 2, 532, 1931.
31. S. Sashoff, Electr. Journ., 27, 486, 1930.
32. M. Schenkel und J. v. Issendorff, Siemens Zs., 11, 142, 1931.
33. R. Seeliger, Phys. Zs., 30, 527, 1929.
34. R. Seeliger und G. Mierdel, Selbständige Entladungen in Gasen, 1927.
35. R. Seeliger, Phys. Zs., 33, 273, 1932.
36. K. Sixtus and L. Tonks, Phys. Rev., 37, 390, 1931.
37. M. Sitnikov and A. Chernyshev, Elektrichestvo No. 12, 614, 1931.
38. H. Steiner and H. Maser, Proc. Inst. Rad. Eng., 18, 67, 1930.
39. J. Warner, Proc. Inst. Rad. Eng., 19, 1561, 1931.
40. V. Vologdin, Rectifiers, 1932.
41. C. Wynn-Williams, Proc. Roy. Soc. A., 132, 295, 1931.
42. C. Wynn-Williams, Proc. Roy. Soc. A., 136, 312, 1931.