Laboratory Technique

Methods for Attaining High Vacuum

The purpose of the present article is to give a survey of the methods for producing high vacuum that are currently used both in scientific and in technical practice. After the invention, exceptionally important in significance, of the diffusion pump, many of the pumps known earlier acquired only historical interest, and therefore they will not be touched upon. To this number must be assigned all mercury pumps, both manual and automatic (Tepler, Sprengel, and others), as well as the well-known, comparatively recently constructed molecular pump of Gaede, which has lost its significance as a consequence of the dizzying rapidity of modern technical progress.

On the other hand, such pumps as Gaede’s mercury and oil pumps still play a large role in pumping technique; but in view of the fact that they are generally known, they will not be touched upon in the proposed survey.

Pumps producing high vacuum are characterized by three parameters: 1) the magnitude of the necessary fore-vacuum (Vorvakuum), 2) the pumping speed, and 3) the limiting attainable vacuum.

The exposition will proceed in the following order:
1) Gaede’s diffusion pump.
2) Langmuir’s condensation pump.
3) Dewar’s method.

In addition, we consider it necessary to indicate briefly the method of degassing large metal parts sealed into various physicotechnical instruments subjected to high pumping, such as the Coolidge tube, high-power thermionic lamps, and so forth.

Gaede’s diffusion pump (1).

In its simplest form the diffusion pump consists of a vessel C (Fig. 1), the upper part of which is a porous partition. The vessel is covered by a cap B and is cooled from below by a jet of water supplied through tubes a and b. The water vapor obtained in A flows around vessel C and creates between B and C a space saturated with water vapor and free of air. The vapor diffuses into vessel C and, in condensed form, accumulates through d in E. Air diffuses in the opposite direction, as a result of which the mercury in the barometer rises. Such a pump can give a rarefaction down to the vapor pressure of water at room temperature, i.e., down to 10 mm.

The basic scheme of the diffusion pump, implemented in the simplest diffusion pump described above, was carried out by Gaede and subsequently repeatedly—

...a liquid-mercury type, a pump in which the porous partition was replaced by a narrow slit. The design of this pump is as follows (Fig. 2).

Mercury \(Q\) is heated by a gas burner or an electric furnace, and the vapors flow from \(A\), between the steel cylinder \(b\) and the tube \(a\), to the condenser \(c\); cold water is supplied at \(m\) and discharged at \(n\). The steel cylinder \(b\) is immersed in the ring \(d\), filled with mercury, so that the space \(A\) is separated from \(B\) and communication between them is maintained through the slit \(l\) in the cylinder \(b\). The mercury vapors pass through the slit \(l\), condense on the walls of the condenser \(c\), and then, descending as drops into the ring \(d\), overflow back into \(Q\) over its edge. The air located in space \(B\) flows in the opposite direction through the slit \(l\) and is carried by the mercury vapors through the tube \(a\), open at the top, into another condenser \(C\), communicating with the Vorvakuum through \(g\). Since theory shows that the pumping speed has an optimum at a certain pressure of the mercury vapor, a thermometer \(h\) is introduced into the pump to determine the corresponding pressure; \(K\) is a mercury pin, allowing the pump to be disassembled and reassembled; \(f\) is the pin for the vessel being evacuated.

Fig. 1.

Fig. 2.

\(V\) is a valve separating the Vorvakuum from the vessel being evacuated. The pumping speed in this vessel is equal to the difference between the quantities of gas: 1) flowing from \(B\) to \(A\), and 2) entrained by the mercury vapors and transferred back from \(A\) to \(B\).

From theoretical considerations it follows that the width of the slit should be of the order of the mean free path of a molecule of the gas being evacuated in mercury vapor. In Gaede pumps it was of the order of \(0.1\)–\(0.05\) mm.

The existence of an optimum in the mercury-vapor pressure follows from simple considerations. At a very high pressure, the vapor will offer too great a resistance to the gas flowing against it, while at a very low pressure the gas from the Vorvakuum will flow back into the vessel being evacuated. Hence the necessity of measuring the temperature of the steel cylinder having the slit. Hence also the existence...

at a certain maximum pumping speed, which is visible from Fig. 3, where the abscissa axis gives the temperature \(t\), and the ordinate axis gives the pumping speed.

The drawback of the basic scheme, namely the directly opposite direction of the vapor and gas streams, gives rise to the following two disadvantages: 1) the need for a limiting fore-vacuum, which must be of an order not higher than \(0.1\) mm, and 2) a low pumping speed of about \(80 \frac{\mathrm{cm}^3}{\mathrm{sec}}\). The existence of a sharp maximum in the pumping speed as a function of temperature also constitutes a substantial inconvenience of Gaede’s pump. Fluctuations of \(10^\circ\) already cause considerable changes in the speed.

Its advantage is the constancy of the speed and the absence of a limiting attainable pressure. In view of the fact that at the present time there are pumps more advanced than Gaede’s pump, the latter has also lost its practical importance; but owing to the fruitfulness of its idea, which subsequently gave rise to a whole series of attempts to eliminate the shortcomings inherent in Gaede’s pump, it should be given considerable attention.

Langmuir’s condensation pump (2).

Following Gaede’s work, in 1917 there appeared Langmuir’s work, in which there was proposed a pump exceptional in its simplicity, giving a high vacuum and devoid of all the shortcomings inherent in Gaede’s diffusion pump. It may be said that the problem of constructing a very simple pump, operating from a very insignificant fore-vacuum (the so-called water-jet pump), with a very high speed and making it possible to obtain an arbitrarily high vacuum, was brilliantly solved by Langmuir in his condensation pump.

Its arrangement is as follows (Fig. 4):

Mercury vapors from the vessel \(A\), heated by a Bunsen burner or an electric furnace, pass through the tube \(B\) to the outlet opening \(L\), where, entraining the gas being pumped out, they condense on the walls cooled by flowing water \((K_1, K_2)\) and flow back in drops through \(D\) into \(A\). The vessel being evacuated is connected to the tube \(F\), and the fore-vacuum is connected to \(V\). The width of the annular slit in Langmuir’s pump reaches from 1 to 3 mm; the pumping speed in this pump can be obtained from 1500 to 3000 \(\frac{\mathrm{cm}^3}{\mathrm{sec}}\). In contrast to the preceding Gaede pump,

Fig. 3.

Fig. 4.

for a condensation pump there is no critical temperature, and hence no limiting Vorvakuum.

As early as 1919 the author, together with Papalexi at the Moscow Physical Institute, demonstrated the possibility of attaining a high vacuum by means of two Langmuir pumps connected in series directly to a water-jet pump. For this it is necessary to raise the temperature of vessel \(A\) to \(330^\circ C\), and also to heat the supply tube \(B\) with an electric current, thereby raising the vapor pressure in the tube at the outlet opening \(L\).

At the present time, in the West and in America, quartz Langmuir pumps are being manufactured in order to pump directly from a water-jet pump.

The limit of the attainable vacuum is determined by the pressure of mercury vapor in vessel \(C\), i.e., when it is cooled with liquid air this limit theoretically reaches \(10^{-27}\) (practically \(10^{-8}\) cm.). In addition to Langmuir’s work, the idea of a diffusion pump gave rise to a whole series of other attempts to improve Gaede’s diffusion pump. These include pumps proposed by Williams (3), Knipps (4), Jones and Russel (5), and others. At present, Langmuir pumps with heating by means of a mercury arc fused directly into it are manufactured by Siemens and Halske.

Gaede’s diffusion pump, as well as Langmuir’s, pumps out all gases and vapors, including water vapor, and therefore does not require desiccants.

Dewar’s Method.

In 1875 Dewar discovered the possibility of obtaining a vacuum by making use of the ability of charcoal to absorb gases and vapors. With decreasing temperature this ability increases greatly, as is evident from the appended table. The first column gives the absorbed volume per 1 g of charcoal at a temperature of \(0^\circ C\), and the second at the temperature of liquid air. All data refer to normal pressure.

Baerwald’s investigations (7) showed that high absorptive capacity is possessed by charcoals: 1) from the pith of elder, 2) linden, 3) from the shell of the coconut, and 4) from the kernel of the same nut.

The greatest absorptive capacity is possessed by the first, i.e., charcoal from elder pith. For practical purposes, however, the most convenient are charcoals either of linden or from the shell of the coconut.

K. Scheel and W. Heuse (8) recommend preparing charcoals for absorbing gases and vapors in the following manner. Coconut shell or another material is subjected to burning in a closed porcelain vessel for \(3/4\) hour, after which the heating is continued in a vacuum desiccator. The charcoal prepared in this way is introduced into a glass vessel soldered to the vessel in which a strong rarefaction is to be obtained. Then, at low pressure, the charcoal is heated to \(500^\circ C\), during which it gives off all absorbed gases and vapors. After this it is cooled and placed in a vessel with liquid air. Depending on the initial pressure at which the adsorption process begins, the vacuum obtained will be different. Thus, K. Scheel and W. Heuse, carrying out a preliminary pumping with a water-jet pump down to 12 mm pressure, were able to bring the pressure by Dewar’s method down to 0.00153 mm. In the same case, when the Vorvakuum was created by a mercury Gaede pump, the vacuum reached 0.000007 mm.