From Current Literature

GENERATION AND MEASUREMENT OF ULTRAHIGH VACUUM

For a long time, the possibilities for generating and measuring high vacuum were practically limited to pressures of the order of \(10^{-7}\) mm Hg. The development of vacuum technology during this period proceeded mainly along the lines of increasing the pumping speed and of more or less substantial improvements in individual components of the vacuum system. It is true that in individual cases rarefactions as low as \(10^{-10}\) mm Hg could be obtained; however, these were rare exceptions, and such a high vacuum was achieved almost exclusively in vessels sealed off from the vacuum installation by using all kinds of “tricks” and getters.

At the same time, rarefactions of the order of \(10^{-9}\)—\(10^{-10}\) mm Hg became an urgent necessity for a whole range of scientific investigations and practical applications, for example in electronics, isotope physics, the physics of surface phenomena, etc. In this connection, in recent years a number of improvements have appeared which have made it possible to make substantial progress in the field of ultrahigh vacuum and have led at the present time to the creation of a sufficiently developed and comparatively simple method of handling the technique of obtaining and measuring pressures down to \(10^{-10}\)—\(10^{-11}\) mm Hg. The paper under review) is devoted to presenting the fundamentals of this new ultrahigh-vacuum technique; it seems expedient for us to dwell on it, although some of the methods described in it have already been discussed in the pages of UFN*.

1. MEASUREMENT OF ULTRAHIGH VACUUM

Among the various devices intended for measuring vacuum down to \(10^{-7}\) mm of mercury**), the ionization manometer has become the most widespread over the past decade. In essence, this is an ordinary triode in which the role of the anode is played by a control grid having a positive potential of 100–200 V relative to the cathode, while the anode, which is at a potential of 10–50 V negative relative to the cathode, serves as the collector of the positive ions formed as a result of

) D. Alpert, J. Appl. Phys. 24*, No. 7, 860 (1953).

**) Methods were developed for determining still lower pressures from the dependence of the work function and the rate of cathode deactivation on pressure; however, these methods proved of little suitability for practical use and were of primary importance in the process of developing ultrahigh-vacuum technique.

electron bombardment of the gas filling the space between the grid and the collector. The gas pressure is determined from the simple relation

\[ p=\frac{1}{S}\frac{i_{\mathrm{col}}}{i_{\mathrm{grid}}}\ \text{mm Hg}, \]

where \(i_{\mathrm{col}}\) is the ion current to the collector, \(i_{\mathrm{grid}}\) is the electron current to the grid, and \(S\) is the sensitivity of the manometer. In a typical case \(i_{\mathrm{grid}}\cong 10\ \text{ma}\) and

\[ S\cong 10\ \frac{1}{\text{mm Hg}}, \]

which corresponds, at a pressure of \(10^{-8}\ \text{mm Hg}\), to \(i_{\mathrm{col}}=10^{-9}\ \text{a}\).

If the effectiveness of the manometer were limited only by the possibility of measuring weak currents, then pressures of the order of \(10^{-16}\ \text{mm Hg}\) would prove measurable. In reality, however, the limits of applicability of the ionization manometer are determined by another circumstance. Namely, the anode grid of the manometer, being irradiated by an electron current, is a source of soft X-rays, which cause photoelectron emission from the ion collector, masking the principal effect already at pressures of the order of \(10^{-7}\ \text{mm Hg}\). To overcome this obstacle, the following method was originally proposed. A thin metal wire was placed in the manometer, which was briefly heated to a temperature of the order of \(1500^\circ\mathrm{K}\) in order to remove occluded gases, after which the wire was cooled to room temperature. After some time \(t\) had elapsed, the wire was again briefly heated in order to eject from it the gases adsorbed during this time. This led to an instantaneous increase in pressure, recorded by the manometer. The amount of gases adsorbed during the time \(t\) depends on the time \(t\) and on the gas pressure and can be calculated on the basis of kinetic theory. As the time \(t\) increases, saturation is reached, corresponding to the deposition of a monomolecular adsorbed layer on the surface of the filament. Therefore, by making measurements at different \(t\), one can find the value \(t_{\max}\) corresponding to saturation and, from the magnitude of \(t_{\max}\), determine the gas pressure in the vessel. Despite the simplicity of this method, it did not become widespread, owing to the long duration of the measurements and also to the substantial dependence of \(t_{\max}\) not only on the pressure but also on the nature of the gas.

A decisive step toward improving the ionization manometer was the reduction of the role of the background produced by the X-radiation of the anode grid. This was achieved in two ways. First, by reducing the probability that X-ray quanta strike the surface of the collector. For this purpose the usual triode construction was reversed, namely, the cathode was placed outside the cylindrical anode grid, and the ion collector, which was given the form of a thin filament, was placed at the center of this grid. This led to a reduction of the solid angle under which the collector is seen, and consequently of the photoelectron background, by several orders of magnitude. In addition, such a configuration of the electrodes ensured a high efficiency of the electron current with respect to gas ionization. In fact, the potential distribution in this case is such that the energy of the electrons falls from the initial value (\(\sim 150\ \text{ev}\)) below \(50\ \text{ev}\) (which corresponds to the loss by the electrons of their capacity for effective ionization) only at a distance of \(0.25\ \text{mm}\) from the collector, whereas with the former configuration, owing to the strong deceleration of the electrons, ionization could occur only in a small part of the interelectrode volume.

The second way of reducing the role of the background was to increase the ion current by preventing ions from escaping to the walls, which was achieved by enclosing the collector inside an anode grid surrounding the collector.

from all sides. The use of both methods made it possible in individual cases to measure, by means of an ionization manometer, pressures down to \(10^{-13}\) mm Hg. Such advanced designs still remain exceptions for the time being; however, ordinary manometers with an inverted arrangement of the electrodes make it possible without difficulty to measure pressures down to \(10^{-10}\)—\(10^{-11}\) mm Hg.

Images of such a manometer at various stages of its manufacture are shown in Fig. 1.

Fig. 1. Ionization manometer.

The principal shortcomings of the ionization manometer are as follows. First of all, the sensitivity of the manometer depends substantially on the nature of the gas.

As can be seen from the table (see p. 155), the sensitivity of the manometer for different gases may differ by more than a factor of 25, although in most cases it remains within the limits of a two- to threefold difference.

Thus, the use of an ionization manometer is associated with the need to take into account the composition of the gas being measured. This accounting is substantially simplified by the fact that the relative sensitivity of the manometer to different gases is almost independent of the individual features of the manometer—its design, mode of operation, etc. Particular note should be made of the difficulty of determining the sensitivity in the case of gases that exert a chemical or electrochemical action on the electrodes and walls, as well as in the case of gases decomposing under the action of electron bombardment or the high temperature of the cathode.

The second major shortcoming of the ionization manometer as a measuring instrument consists in the fact that its operation entails a decrease of the pressure in the vessel, which makes its use difficult when it is necessary to keep the pressure unchanged.

FROM CURRENT LITERATURE

Relative sensitivity of an ionization manometer for various gases
(the sensitivity for nitrogen is taken as unity)

2. PUMPING DOWN TO ULTRAHIGH VACUUM

The attainment of ultrahigh vacuum by means of diffusion pumps is impossible, since these pumps not only are incapable of evacuating a system to pressures below \(10^{-7}\)—\(10^{-8}\) mm Hg, but themselves are also sources of possible contaminants penetrating into the system. However, in order to obtain ultrahigh vacuum one may make use of the ability of a number of substances (getters) to occlude gases, and, among other things, the noted property of ionization manometers to lower the pressure in a vacuum system. This property has a dual nature. First of all, the positive ions formed by electron bombardment are carried by the electric field to the collector and the walls, where they are neutralized and adsorbed. Obviously, such a mechanism of ion trapping and pumping of the vessel will operate only in the presence of a potential difference accelerating the electrons. In addition, some gases (for example, H₂, O₂, Cl, etc.) may be “pumped out” through chemical reactions proceeding on a heated (usually to \(\sim 2300^\circ\) K) cathode, and this process is independent of the presence or absence of an accelerating potential. Thus, the action of an ionization manometer is analogous to that of a getter. If \(S_E\) denotes the pumping speed due to adsorption of ions and \(S_C\) the pumping speed due to the catalytic action of the cathode, then, assuming that the pressure in the vessel \((p)\) is much greater than the limiting pressures attainable by each of these paths, we have \((V\) is the volume of the vessel):

\[ \frac{dp}{dt}=-\frac{S_E+S_C}{V}\,p. \]

Putting

\[ \frac{V}{S_E}=\tau_E \quad \text{and} \quad \frac{V}{S_C}=\tau_C, \]

we find:

\[ p=p_0 e^{-t/\tau}, \]

where

\[ \frac{1}{\tau}=\frac{1}{\tau_E}+\frac{1}{\tau_C}. \]

Thus, \(\tau\) is the characteristic time during which the pressure decreases by a factor of \(e\) (\(\tau_E\) and \(\tau_C\) are the characteristic times for the electrical and chemical pumping processes, respectively). Using typical values of the manometer sensitivity

\[ S=10\ \frac{1}{\text{mm Hg}} \]

and \(i_{\text{electr}}=10\,\mu\text{A}\), we find that \(i_{\text{ion}}=10^{-(n+1)}\,\text{A}\) for \(p=10^{-n}\) mm Hg, which corresponds to \(i_{\text{ion}}=6\cdot 10^{18}\cdot 10^{-(n+1)}\) molecules per second at a density \(n=3.2\cdot 10^{19}\cdot 10^{-n}\) molecules per liter; hence

\[ S_E=\frac{1}{n}\frac{dN}{dt}\simeq 1\ \text{liter/minute} \]

(\(N\) is the number of molecules in the volume being pumped, \(N=nV\)), i.e. \(\tau_E\simeq 1\) minute for \(V=1\) l.

Since per unit time, per unit surface area, there arrive (in the absence of a field)

\[ \nu=\frac{n\bar v}{4} \]

molecules, where \(\bar v\) is the mean velocity of their thermal motion, the speed of chemical pumping is

\[ S_C=\frac{1}{n}\frac{dN}{dt}=\frac{n\bar v}{4}\frac{A}{n}=\frac{\bar v A}{4}, \]

where \(A\) is the area of the cathode surface. Taking \(\bar v\simeq 4\cdot 10^4\) cm/sec and \(A\simeq 0.2\) cm\(^2\), we obtain \(S_C\simeq 1\) l/sec, i.e. \(\tau_C\simeq 1\) sec for a volume \(V=1\) l. Thus, the characteristic pumping times \(\tau_E\) and \(\tau_C\) differ little in order of magnitude. The estimates given above are confirmed by experiment. Fig. 2 shows the curve of the decrease of pressure in an evacuated vessel with time as a result of the operation of an ionization manometer in the case of nitrogen, for which only the electrical pumping mechanism takes place (the data have been reduced to the conditions \(V=1\) l and \(i_{\text{electr}}=10\,\mu\text{A}\)).

Graph labels: ordinate, “Pressure in mm Hg”; abscissa, “Time in sec”; annotation, \(\tau=29.2\) sec.

Fig. 2. Pumping speed by an ionization manometer.

A very important question is the amount of gas that the manometer can pump out. It was shown that the total amount of gas that the manometer can absorb as a result of adsorption of ions corresponds to a few monomolecular layers on the adsorbing surfaces, i.e. of the order of \(10^{15}\) molecules per 1 cm\(^2\) of active surface. Consequently, at a pressure of \(10^{-4}\) mm Hg, after approximately 1 hour of operation of the manometer, saturation should occur and further pumping should cease, which is confirmed by experiments. However, at a pressure of \(10^{-9}\) mm Hg, contin-

the duration of the saturation process, i.e., the time during which the manometer is capable of operating as a pump, reaches \(10^8\) sec, or approximately three years. Manometers that operated for 75 days at such pressures still showed no signs of saturation.

As an illustration we present Fig. 3, which shows the time dependence of the readings of two ionization manometers connected to two adjacent volumes separated by a valve (see the top of the figure). The dashed line corresponds to a vessel disconnected from the pump. Owing to the action of the ionization manometer, the pressure was reduced to \(2\cdot 10^{-10}\) mm Hg and remained unchanged for 900 hours. The solid line in Fig. 3 corresponds to a system connected to a diffusion pump. Initially the pressure was reduced to \(5\cdot 10^{-10}\) mm Hg, since the ionization manometer absorbed the gases evolved by the pump; however, after about a day, saturation of the manometer began to appear, as a result of which the pressure began to rise and after 100 hours reached \(10^{-7}\) mm Hg, i.e., the vapor pressure of the oil used in the diffusion pump.

Fig. 3. Saturation of an ionization manometer.

3. VACUUM SYSTEM

As measurements have shown, in a well-degassed system the partial pressure of occluded gases is less than \(10^{-13}\) mm Hg. However, even in the absence of leaks, in a sealed system there is a continuous inflow of gas, causing on average a pressure increase of the order of \(3\cdot 10^{-11}\) mm Hg per minute, which corresponds to the rate of diffusion of atmospheric helium (present in air at a concentration of several millionths) through glass walls. Such an inflow is easily compensated by the operation of the ionization manometer. Consequently, the possibilities of obtaining ultrahigh vacuum can be limited only by 1) the presence of leaks in the system, 2) the evolution of gases during the operation of vacuum valves, and 3) the evolution of gases from liquid manometers (of the McLeod manometer type) intended for measuring low vacuum.

The use of an ionization manometer, which permits measurement of pressures down to \(10^{-10}\)—\(10^{-11}\) mm Hg, makes possible the rapid detection of leaks by the rate of pressure increase in a system disconnected from the pump (with the manometer switched off) down to inflow rates of the order of \(10^{-16}\) l/sec, i.e., several orders of magnitude smaller,

than by other methods. This makes possible rapid and effective exclusion of leaks.

To exclude from the system liquid manometers that release vapors and dissolved gases, a differential manometer, shown in Fig. 4, is used. A thin covar diaphragm (0.125 mm thick) separates the region of ultrahigh vacuum from the auxiliary volume to which the liquid manometer is connected. The diaphragm, together with an additional condenser plate, forms a capacitor included in a capacitance bridge. To measure the pressure in the high-vacuum part of the installation, the pressure in the auxiliary volume is changed until it is equalized with the pressure being measured and is measured by the liquid manometer, the equality of the pressures

Fig. 4. Manometric device for low vacuum.

being established from the equilibrium of the capacitance bridge corresponding to the unstressed state of the diaphragm. Such a design makes it possible to measure pressures above \(10^{-1}\) mm Hg with an error of about \(10^{-2}\) mm Hg (independent of the magnitude of the pressure). At the same time, such a system is readily degassed and completely guaranteed against gases entering the high-vacuum system. The same manometer may be used for calibrating an ionization manometer.

The most critical part of the vacuum system is the valves. The use of ordinary stopcocks is inevitably associated with the introduction into the system of lubricant vapors and occluded gases in quantities several orders of magnitude greater than the amount of gas in a system pumped down to limiting vacuum. Therefore, the possibility of obtaining an ultrahigh vacuum in sealed-off vessels was due to the development of a valve design free from the indicated drawback.

The valve schematically shown in Fig. 5 proved entirely satisfactory. A copper cup 35 mm in diameter has two holes 6 mm in diameter for the vacuum inlets. The cup is closed by a thin covar diaphragm capable of bending approximately 2.5 mm in both directions and provided with a covar rod, the end of which is sharpened at an angle of 45° and ground to one of the holes of the copper cup. In the assembled form, the valve parts are connected to each other by bronze (in the hydrogen flame) solder. The valve is then fitted with a cap containing a micrometer screw for inserting the valve rod. When the rod is introduced into the opening of the cup, the valve is closed, and in new valves the leakage...

gas leakage through the valve does not exceed \(10^{-10}—10^{-11}\) l/sec. (As operation proceeds, the leak increases somewhat, reaching within a year a value of \(10^{-9}—10^{-10}\) l/sec.) In the open state (the stem is withdrawn from the opening)

Labels in the diagram: diaphragm; micrometer screw; driving stem; copper cup; vacuum inlets.

Fig. 5. Diagram of the design of a high-vacuum valve.

the valve has the same throughput as large vacuum stopcocks. At the same time, it is readily degassed and completely eliminates the need to introduce contaminating lubricants into the system. In Fig. 5 below, two types of valves are shown (with a simple and a differential micrometer screw), in semi-assembled and assembled form.

As a result of the use of all the devices described, the vacuum system acquires very compact dimensions and is easily assembled from separate units (Figs. 6 and 7). The start-up of the new system is carried out as follows. After switching on the mechanical and diffusion pumps, one locates

Fig. 6. General view of the apparatus for obtaining ultrahigh vacuum.

Fig. 7. High-vacuum part of the apparatus for obtaining ultrahigh vacuum. 1 — ionization gauges—pumps, 2 — valves, 3 — tube leading to the diffusion pump.

and eliminate large leaks. Then the high-vacuum part of the system is placed entirely in a special furnace and degassed at a temperature of \(400\)—\(500^\circ\mathrm{C}\) for several hours. After this, the furnace is removed and the metal parts of the ionization gauge are heated to a temperature above \(1200^\circ\mathrm{C}\) (by enhanced electron bombardment with a positive potential on the ion collector). Next the valves are closed and small leaks are sought and eliminated, after which the ionization gauges pump the system down to ultrahigh vacuum.

The time required to carry out all these operations is characterized by the following data. After the individual parts of the system have been mounted on the frame, they are soldered together during the morning hours with a glass blower; during the second half of the day the large leaks are eliminated; degassing is carried out at night, and by the following noon the system already reaches a pressure of the order of \(10^{-10}\) mm Hg. In the absence of small leaks, pumping down to a pressure of \(10^{-9}\)—\(10^{-10}\) mm Hg takes no more than an hour.

Thus, the apparatus described makes it possible to obtain and measure ultrahigh vacuum just as simply, from the technical point of view, as ordinary high vacuum of the order of \(10^{-7}\) mm Hg.

R. G.