Abstract
The phenomena of gas disappearance in cathode tubes and incandescent lamps considered by us below constitute the foundations of modern industrial high-vacuum technology, which makes it possible to reduce the evacuation period to one minute or even less.
Full Text
Disappearance of Gas During an Electric Discharge
N. A. Shishakov, Moscow
The passage of an electric discharge through a tube containing gas at low pressure is usually accompanied by a decrease in the gas pressure (clean up, disappearance of gas, Aufzehrung). All tubes in which this phenomenon is observed may be divided into two types: 1) tubes with cold electrodes, for example Geissler tubes, X-ray tubes, and also electrodeless tubes, and 2) tubes with heated electrodes, for example cathode tubes or incandescent lamps. In the two cases the phenomena of gas disappearance are of a completely different character, depending above all on the difference in the characteristics of the discharge itself.
In tubes of the first type the discharge can take place only at pressures above \(10^{-3}—10^{-2}\) mm. Its action depends on the formation of positive ions, which in turn is due to the large magnitude of the cathode potential drop. Characteristic of these tubes is the large distance of the electrodes from one another. In tubes with heated cathodes and in incandescent lamps, positive ions play an insignificant role. Owing to the high temperature of the cathode and the small distance of the electrodes from one another, the transfer of electricity is carried out by means of a thermionic current. At a sufficiently high incandescence temperature, the saturation current is attained with only a small potential difference.
The phenomena of gas disappearance in tubes of the first type appear to us to be extremely complex. Despite the enormous number of investigations in this field, we still do not possess a theory that would satisfactorily explain all the phenomena belonging here. We shall therefore leave this entire field out of consideration. Our survey will deal only with phenomena in tubes of the second type. We are led to this choice not only by the considerably greater simplicity of these phenomena, but chiefly by the circumstance that at the present time they are of enormous technical importance, especially for incandescent lamps.
The phenomena of gas disappearance in cathode tubes and incandescent lamps considered below constitute the basis of the modern factory technique of high vacuum, which makes it possible to reduce the evacuation period to one minute and even less. At the present time almost
the need to use high-vacuum pumps for evacuating incandescent lamps has been completely eliminated, and evacuation only to \(0.001\)—\(0.1\) mm is sufficient. After the lamp is removed from the pump, this residual gas is made to disappear upon the first lighting of the lamp, when, owing to the sufficient voltage at the nearby ends of the filament, ionization of the gas takes place and an electrical discharge occurs. These phenomena were studied in detail by Campbell and his collaborators1. In order to avoid the influence of various disturbances that obscure the true picture of the phenomena, Campbell used, instead of ordinary incandescent lamps, cathode tubes of a peculiar shape; the cathode was a tungsten filament arranged along the axis of the tube, and the anode was a nickel wire parallel to it.
Fig. 1.
In his first works Campbell confirms the previously known fact that the disappearance of gas takes place only at the moment when the discharge is accompanied by a glow. It was therefore very important to determine what connection exists between this glow, the glow discharge, and the disappearance of gas. Campbell’s subsequent works concern the connection of the disappearance of gas with ionization and dissociation, the influence on the disappearance of gas of the walls of the vessel, the influence of the pressure and nature of the gas, and a number of other factors. We shall dwell on all this in turn.
§ 1. The potential of the glow discharge. If the tube is very well evacuated \((p \ll 10^{-3}\ \mathrm{mm})\), then the electron current, as is known, independent within wide limits of the temperature and of the electron emission, increases with the potential according to the “\(3/2\)” law. At saturation current \(i_0\), the electron current will depend only on the temperature. This is shown by curve I in Fig. 1. If the tube contains gas at a pressure of about \(10^{-3}\) mm, then the relation between \(i\) and \(V\) will be somewhat different; namely, the anode current, beginning with the ionization potential \(V_0\), will increase more rapidly with increasing potential (curve II). Owing to the small quan-
as positive ions, the saturation current will remain the same. At \(p=2\cdot 10^{-3}\) mm the characteristic at first remains the same (curve III), but at a certain potential \(V_g\) there appears a sudden jump up to the saturation current; at this moment the glow discharge occurs, which is why the quantity \(V_g\) is called the potential of the glow discharge. This potential depends on the arrangement of the electrodes and on the dimensions of the tube, and in this it differs from the ionization potential \(V_0\), which depends only on the nature of the gas. If one now begins to decrease the potential, the glow discharge will not cease at once, but only at a certain potential \(V'_g\) (curve III′). Only below this value does the current assume the value corresponding to curve III. At still higher pressures \((5\cdot 10^{-2}\) mm) the ionic current will no longer be small in comparison with the electronic current (curve IV); the irreversibility disappears.
Campbell and Rait found that the potential of the glow discharge is not affected by the temperature of the filament and by thermionic emission, but that it depends strongly on the nature and pressure of the gas. As for the latter dependence, as Fig. 2 shows, the difference \(V_g - V'_g\) decreases with increasing pressure. It is evident that with a very large potential one can attain very low pressures. Usually, in making lamps, this inequality of the quantities \(V'_g\) and \(V_g\) is used. In order not to subject the filament to a voltage dangerous for the lamp, part of the gas is first removed at a certain voltage corresponding to point \(A\); then, when at point \(B\) the discharge ceases, the voltage is raised to point \(C\), and so on.
Fig. 2.
Figure 3 presents the dependence of the potential of the glow discharge on the length of the free path of the gas molecules. It is evident that, with increasing molecular weight, the potential of the glow discharge decreases and proves, moreover, to be extremely sensitive to small admixtures of gases with large molecular weight. Especially strong in this respect
affects mercury vapor, not only at ordinary temperatures but even at \(-78^\circ\) C. This is understandable, since mercury vapors are capable of carrying current even at very small voltages. However, in the presence of mercury vapor, which strongly lowers the potential of the glow discharge of the gases, the process of disappearance proceeds quite differently than might have been expected. Kembel found that, during a discharge in pure mercury vapor, even from the walls of very clean glass an enormous quantity of gases, especially hydrogen, is evolved. Shabi and Pietsch1 first passed a discharge in pure nitrogen and then, after good evacuation, in mercury vapor. In the second case nitrogen was evolved from the walls of the tube. One may imagine that, in the presence of
Fig. 3.
Fig. 4.
Legend: \( \bullet + \times \wedge\) — bare walls. \( \circ\) — walls coated with magnesium.
mercury vapor, two processes occur simultaneously: elimination of gas owing to the lowering of the potential of the glow discharge, and reverse evolution—
depletion of the gas under the influence of bombardment of the negatively charged walls by heavy mercury ions.
§ 2. Ionization and disappearance of the gas. A special work by Campbell and Nye was devoted to this question. Using a tube with three electrodes, they were able to measure the ionic current \(i_g\) as a function of the electronic current \(i_a\), and to observe the change of \(\dfrac{i_g}{i_a}\) with time. In this way it was possible to decide the question of the number of disappearing molecules per ion, and to consider the ratio of these quantities \(\dfrac{n_1}{n_2}\) as a function of the pressure or of the mean free path of the neutral molecule \(\lambda\) (\(d\) in Fig. 4 denotes the length characterizing the dimensions of the vessel).
Fig. 5.
From the lower part of the figure it is seen that at high pressures \(\dfrac{n_1}{n_2}\) is almost independent of the pressure, and at low pressures falls rapidly; the maximum value of \(\dfrac{n_1}{n_2}\) lies somewhat below unity. In Fig. 5 the dependence of \(\dfrac{n_1}{n_2}\) on the voltage at the anode is shown. Incidentally, judging from both these figures, the results obtained by Campbell and Nye cannot be regarded as indisputable.
§ 3. Influence of the vessel walls. Campbell and Nye indicate that the disappearance of the gas depends almost neither on the material of the walls nor on the nature of the gas. It must, however, be noted that the condition of the glass surface plays an essential role. Every treatment of the glass which leads to a reduction of the gas film—for example, preliminary bombardment by electrons or mercury vapor, cooling with liquid air, coating the walls with certain elements (especially phosphorus)—leads
to a more intense disappearance of the gas. An increase in the size of the vessel acts in the same direction.
At a certain degree of loading of the walls by gas, fatigue sets in, depending on some conditions at the surface of the glass. That the glass walls are responsible for the appearance of fatigue is evident from the fact that, when the filament current is changed, the activity of the walls does not increase at all. The degree of loading of the walls lies approximately within the limits from 0.2 of the thickness of a monomolecular layer to 20 such layers (hydrogen).
§ 4. Pressure and nature of the gas. Measurement of ionization at high pressures presents great difficulties; therefore there have been very few systematic investigations on this subject. As regards low pressures, we shall mention the work of Dushman, Andrews, and Justheimer1, who found that, at constant current strength, the disappearance of the gas increases linearly with pressure. Let us turn to the behavior of various gases.
a) Noble gases. All investigations on this subject have hitherto concerned only the least rare argon and helium. These gases prove to be only slightly active. Campbell finds, for example, that for argon \(\frac{u_1}{u_2} < 0.1\), i.e. considerably lower than for nitrogen. It is further known (Shaw and Pitch) that, in a discharge in a mixture Ar—H\(_2\) (45 : 55), or in nitrogen with 1–2% Ar, hydrogen and nitrogen disappear completely, while argon remains almost entirely in the free state.
To explain the disappearance of helium, Manley2 carried out a series of experiments intended to determine the possibility of a chemical reaction of helium with the vapors of certain elements and subsequent condensation of the compound obtained on the walls. Manley found that, in the presence of mercury vapor at low pressures, the compounds Hg He and Hg He\(_{10}\) are formed. The possibility of formation of a reactive form of helium was confirmed by Joos3. Boomer4 proved the existence of the compound WHe\(_2\). He also considers possible the formation, under electron bombardment, in the vapors of Hg, J, and S, of the corresponding helides, in the form of which helium can disappear from the tube and reappear only after heating.
b) Hydrogen. In contrast to other gases, the disappearance of hydrogen may occur even without a field; for this it is necessary that the temperature of the filament be not lower than 1250–1300° K. Langmuir5 established that under such conditions hydrogen dissociates and, in atomic
DISAPPEARANCE OF GAS IN AN ELECTRIC DISCHARGE
in this state settles on the walls of the vessel, where it can be retained in considerable quantities, especially on those parts of the walls that are cooled by liquid air. If the filament is cooled and the liquid air removed, then part of the hydrogen is liberated, but already in the form of ordinary hydrogen, H₂, which, when the walls are cooled again with liquid air, no longer condenses (non-recondensible gas).
In a field, hydrogen can disappear even at a low filament temperature. Hughes and Tyndall¹) find that in a tube with an oxidized cathode the disappearance of hydrogen begins at 13 V and has a maximum at about 70 V (Fig. 6). From the fact that this disappearing gas shows the same properties as Langmuir’s NR gas, it may be concluded that here too condensation of the atomic modification of hydrogen takes place.
Fig. 6.
Schwab and Pietsch²) carried out careful experiments on the removal of hydrogen. They also used oxidized cathodes. The rate of disappearance of hydrogen proved to be independent of the pressure and dependent on the prior treatment of the glass. Of the three possible reaction products (hydrogen ions, atomic hydrogen, and water), the most probable must be considered to be water.
In fact: no NR gas is formed here; furthermore, in the presence of concentrated sulfuric acid and P₂O₅ the amount of gas that disappears increases greatly; and finally, in the presence of a small quantity of water vapor artificially introduced into the flask, the disappearance of the gas proceeds the more weakly, the greater the vapor concentration.
The question of hydrogen ions must be answered in the negative, for no change in the rate or degree of disappearance of the gas—when the reaction products are moved toward sulfuric acid through the field of a condenser at 110 V—is observed. Against atomic hydrogen speaks the circumstance that no chemical actions characteristic of
¹) Hughes and Tyndall, Phil. Mag., 6, 27, 415 (1914).
²) Pietsch, l. c.
of this active form of hydrogen. As for the oxygen necessary for the formation of water, it may be taken from the oxide films of the cathode and other metallic parts.
c) Nitrogen. The disappearance of nitrogen is accompanied by strong sputtering of the cathode, especially a tungsten one. Campbell believes that the brown deposit formed on the walls of the vessel is not the compound \(WN_2\), as Langmuir\(^{1}\) considers, but is formed through the occlusion of nitrogen by sputtered tungsten. Nitrogen may disappear in enormous quantities, and when the bulb is heated only a small part of it returns. Uzz and Tyndall believe that nitrogen, like hydrogen, may disappear in the atomic state (see Fig. 5).
d) Carbon monoxide. The disappearance of this gas was studied in detail by Campbell and his co-workers. They established that a chemical reaction takes place here, proceeding according to the equation
\[ 2CO = CO_2 + C, \]
where a neutral molecule and a carbon-monoxide ion react with one another. It is possible, however, that, alongside this principal reaction, other phenomena also take place in the tube.
§ 5. Incandescent lamps. Practitioners in the lamp industry know well that when a vacuum lamp is first lit, a blue glow appears in it, which disappears after a few seconds, and that this glow is accompanied by an improvement of the vacuum. This phenomenon of “blue glow” is completely identical with the glow discharge in cathode tubes. It also occurs as a result of the ionization of gas under the action of electrons emitted by the negative end of the filament and accelerated under the action of the potential difference between the two ends of the filament. Below the potential of the glow discharge (see p. 107) no discharge whatever is observed, and the disappearance of the gas proceeds very weakly. If a certain amount of red phosphorus is introduced into the lamp, then, as Malignani (Malignani, 1894) first observed, the disappearance of the gas is greatly accelerated. Whitney\(^{2}\) noted, for example, that some other elements (As, S, J) also exert an analogous action. Since with the aid of such auxiliary agents it became possible sometimes to attain a very high vacuum in the shortest time, it is natural that many investigators, chiefly at lamp factories, occupied themselves with this matter. Soon an enormous number of different substances were discovered which behave analogously to phosphorus.
a) Action of phosphorus. This question was studied by Hamburger,\(^{3}\) Kohlschütter and Frumkin,\(^{4}\) and especially by Campbell with
\(^{1}\) Langmuir, J. Am. Ch. Soc., 37, 417 (1915).
\(^{2}\) Whitney, Trans. A. I. E. E., 31, 921 (1912).
\(^{3}\) Hamburger, Engineering, 108, 365 (1914).
\(^{4}\) Kohlschütter und Frumkin, Zs. f. Elektrochem., 20, 110 (1914).
collaborators. Phosphorus is introduced into the lamp either in the state of white-phosphorus vapor, or in the form of red phosphorus, the latter being placed either on the walls of the bulb or deposited on a filament. Klemperer finds that, when phosphorus on a filament is evaporated while it is being heated, the quantity of gas that disappears increases with the quantity of phosphorus deposited on the filament. With slow evaporation of phosphorus from the walls of the bulb, the gas disappears considerably more slowly.
The general picture of the phenomena is as follows. The disappearance of the gas takes place during the glow discharge. In the presence of phosphorus, the disappearance of the gas at high pressures \((p > 0.04\ \mathrm{mm})\) proceeds in the same way as in the absence of phosphorus. But at excessively low pressures the gas by itself does not disappear; the admixture of phosphorus vapor (at the moment of evaporation) increases the total pressure and lowers the potential of the glow discharge. This promotes continuation of the discharge and the disappearance of the gas together with the phosphorus vapor. Under such conditions the gas pressure may fall to \(10^{-5}\ \mathrm{mm}\) and lower.
It would seem that, owing to the high reactivity of phosphorus free from the surface oxide layer, a chemical reaction could take place during the evaporation of phosphorus. However, a whole series of facts speaks against this: first, the absence of stoichiometric relations between phosphorus and any gas; further—the independence of the action of phosphorus from the nature of the gas1 and, conversely, its dependence on the state of the vessel surface (washing, etching, silvering, etc.); and, finally, the circumstance that upon heating the gas returns in the same form in which it was found before the phosphorus began to act.
Klemperer adheres to a mechanical theory of the action of phosphorus. In favor of this is the fact that, when phosphorus vapor is continuously passed through a tube, gas can disappear in enormous quantities. One may suppose that, during the discharge, all gases stick to the glass, and that with further action of the discharge part of them returns (see p. 108), and further, that as a deposit of phosphorus accumulates on the glass the latter prevents the reverse liberation of gases.
Thatcher2 considers that in the discharge phosphorus vapor is readily ionized and is deposited in the form of a countless multitude of very fine particles. At the moment of deposition, gas particles are rapidly attracted to these particles, whose surface is free from gas; compared with the phosphorus particles, these gas particles are very few. Lidbury and, in part, Dushman3 hold approximately the same view. These views find confirmation in C. T. R. Wilson’s4 experiments on the condensation of ions on the surface of small drops of water.
b) The action of other substances. There exists an enormous number of substances which behave in incandescent lamps similarly to phosphorus, for example alloys of alkali or alkaline-earth metals with other metals, halogen compounds of alkali metals, certain silicates, and many others.
The action of these substances is, for the most part, twofold. On the one hand, they contribute to improving the vacuum, as, for example, the compound \(P_4O\), which upon the first ignition of the filament decomposes into \(P\) and \(P_2O_3\), both acting as good drying agents. On the other hand, these substances can act chemically on the particles of metal sputtered by the incandescent filament, owing to which in other cases they can serve as a good protective means against coating the walls of the bulb with an opaque layer of metal. Both of these circumstances have led to the wide use of these substances in incandescent-lamp technology. As a result, it has now been possible to achieve an increase in the useful duration of burning (20% loss of luminous intensity) by a factor of 2–3 in comparison with lamps treated with phosphorus alone.
§ 6. Conclusion. The results described above, and the ideas based on them, are far from perfect. The reason for this lies primarily in the difficulties that must be overcome when excluding individual phenomena superposed on one another. In conclusion we shall point to the phenomena of gas evolution during an electric discharge, which in most cases are the principal hindrance to the precise arrangement of experiments (cf. p. 108).
Together with the phenomena of thermal evolution of gases from the walls of glass and metals, these phenomena play an enormous role throughout vacuum technology. Therefore, in practice, in order to attain the highest degree of degassing of the walls, along with heating one very often also uses electron bombardment¹ of the walls or bombardment with heavy mercury ions². This, however, does not always achieve the aim.
The giving off of gases by the metallic parts of the tube also plays a major role. Especially active in this respect are Pt, Ag, and Ni.
The gases liberated by glass consist, for the most part, of hydrogen and noble gases. Strutt³ considers that the latter appear as a result of the disintegration of elements. Ramsay⁴ admits the possibility of diffusion of He and Ne through glass. As for metals, then, besides hydrogen and gases of the carbon group, they may also give off other gases.
¹ Daudt und Ewest, ZS. f. techn. Phys., 6, 7a, 329 (1925).
² Pirani, ZS. f. Phys., 9, 327 (1922).
³ Strutt, Proc. Roy. Soc. A., 89, 499 (1914).
⁴ Ramsay, Journ. Chem. Soc. (Lond.), 103, 264 (1913).