Modern Methods of Leak Detection

L. E. Levin

The rapid development of vacuum technology in the postwar years, its broad introduction into the most diverse branches of science and industry, and also the increasing complexity of the design of vacuum instruments have made it urgently necessary to solve the problem of leak detection. Leakage into vacuum apparatus is usually one of the most serious difficulties in the work of both the experimental physicist and the production engineer who must use vacuum equipment.

The increase in the volume of modern vacuum apparatus required an increase in the sensitivity of leak-detection methods. In addition, owing to the need to inspect a large number of parts, there arose an acute need to increase the productivity of testing and to reduce its cost. All this brought into being a new method for detecting leaks—the most sensitive of all known methods and one that provides high productivity—the mass-spectrometric method[^1],[^3].

The mass-spectrometric method of leak detection is associated with detecting a tracer substance in the installation being used by means of an instrument based on the principle of the mass spectrometer.

To carry out tests, the vacuum system of the leak detector is connected through a throttling valve to the previously evacuated volume under test, which is blown with the tracer gas. Particles of air and of the tracer gas that have penetrated through the leak into the volume, together with molecules of the residual gases, enter the mass-spectrometric chamber of the leak detector, are ionized, and are separated according to mass.

The magnitude of the ion current produced by the tracer substance characterizes the amount of gas that has penetrated into the volume, i.e., the size of the leak (if the work is being carried out by the method of blowing individual places) or the total leakage (if the volume under test is placed entirely in an atmosphere of the tracer gas).

In principle, it is possible to use radio-frequency mass spectrometers for leak-detection purposes; however, instruments with electromagnetic separation have become widespread.

According to the method of gas ionization, they are divided into two types:

1. Mass spectrometers with a hot cathode, in which gas particles are ionized by electrons emitted by an incandescent cathode and accelerated by an electric field \(^{2-4}\).

2. Mass spectrometers with a cold cathode, in which ionization of the test gas occurs in a gas discharge maintained at low pressures by means of a longitudinal magnetic field \(^{5,6}\).

Leak detectors of the latter type operate at elevated pressures, but have reduced sensitivity. These instruments are suitable for testing equipment for whose tightness excessively stringent requirements are not imposed. Their advantage is the very long service life of the ion source.

According to the method of focusing ion beams and, consequently, according to the shape of the magnetic analyzer, leak detectors of the sector type \(^{2}\) and leak detectors with one-hundred-and-eighty-degree focusing \(^{4,6}\) are distinguished. Tuning the instrument to ions of a definite mass is carried out by smoothly changing the accelerating voltage \(^{2}\) or the magnetic field \(^{6}\).

The overwhelming majority of leak detectors are designed to use helium as the working substance.

The advantages of helium in comparison with other gases are: its small molecular weight, which makes it possible to simplify the design of the mass spectrometer as much as possible (small working radius); the almost complete absence of it in the atmosphere and in gases released by parts of vacuum systems; and also its complete harmlessness and safety in operation. The last two circumstances compel one to give preference to helium in comparison with hydrogen, which is explosive and is evolved intensively during degassing, although the use of hydrogen as the test gas makes it possible to simplify the mass spectrometer still further \(^{7,8}\).

The mass-spectrometric leak-detection method finds wide application wherever vacuum equipment is operated. Its use, however, is not limited to testing vacuum apparatus. It proves useful in checking the tightness of sealed volumes of vacuum and non-vacuum refrigerators and reservoirs intended for storing gases under high pressure, and also for detecting leaks in underground pipelines. Depending on the conditions under which the tests are carried out and on the design features of the apparatus being tested, it is necessary to use various methods of leak detection; thus, there is the method of blowing the outer surface of the evacuated volume connected to the leak detector; the method of the helium chamber, which makes it possible—

allowing the rapid determination of the total influx into a volume; the probe method, associated with admitting helium into the apparatus under test and feeling its outer surface; the method of holding parts in a helium atmosphere with subsequent placement of them in a vacuum chamber connected to a leak detector; a special procedure for testing a large number of small parts in mass production, etc. ^9.

The sensitivity of the mass-spectrometric leak-detection method depends on the sensitivity of the mass spectrometer, on the chosen testing method, on the total influx and gas evolution of the apparatus, on the quality of the test gas, and on other factors. But the fundamental limitation on the sensitivity of the method is not the sensitivity of the mass spectrometer, which can be made very high, but the helium content in the atmosphere.

For reliable detection of a leak it is necessary that the penetration of helium through the leak cause a reading of the output instrument of the leak detector at least twice as large as the reading caused by atmospheric helium. This means that the leak detector is capable of detecting a leak passing such an amount of helium as constitutes

\[ \frac{1}{200\,000} \]

of the total flow in the vacuum system.

Alongside the mass-spectrometric method of leak detection, the method of detecting leaks by means of halogen-containing compounds has become widespread, especially in refrigerator production. This method is based on the fact that platinum heated in air to red heat is a source of positive ions, the emission from the heated anode sharply increasing in the presence of halogen-containing vapors. This phenomenon is observed at pressures from approximately 0.1 mm Hg up to atmospheric pressure. Its mechanism has not been fully studied to the present time. There is evidence according to which oxygen ionization occurs here on the surface of metals in the presence of halides ^10.

Halogen leak detectors have found wide practical application. Thus, for example, one type of industrial halogen leak detector ^11 is intended for detecting leaks in hermetically sealed volumes (similar to those used in refrigerators) that contain halogen compounds, and also for checking all systems into which some halogen compound may be introduced as a tracer. This leak detector consists of an easily movable sensor and, connected to it by a long flexible hose, a measuring unit (Fig. 1). In the sensor, made in the form of a pistol, there is placed a sensitive element consisting of two platinum cylinders: an inner one, heated to a temperature of about 900° C,

and the outer electrode, which is at a negative potential relative to the inner one. To create an air flow through the sensitive element, a small fan is placed in front of it in the pistol.

Detection of leaks with the aid of the described instrument is carried out by moving the detector along the outer surface of the volume being tested. When the pistol approaches a leak, vapors of halogen-containing compounds begin to be drawn into it together with the air. The ionic currents increase, causing an audible signal.

Fig. 1. Halogen leak detector.

Another method of leak detection is also possible. The sensitive element is connected directly to the system being tested, the outer surface of which is blown with a weak jet of Freon gas, harmless to health. During the tests, an air pressure of not less than 0.1 mm Hg is maintained in the system.^10

Halogen leak detectors can be used for testing vacuum systems, and also for detecting leaks in containers, pipelines, and systems under pressure.^12–14

In sensitivity the halogen leak detector is somewhat inferior to the mass-spectrometric one, but it is considerably more portable. A particularly important advantage of this type of leak detector becomes clear from the following. In the process of pumping, oil diffuses from the pump into the volume being tested and covers its inner surface with a thin film, partially closing the leaks. In the mass-spectrometric method this may impede their detection. Freon, however, has a strong affinity for oil, is adsorbed by it, and, on emerging from the leaks, can be detected by the halogen leak detector.^15

Along with the most widespread leak-detection methods described above, a whole series of other methods also find practical application.

The method of checking the tightness of glass vacuum equipment by means of a Tesla transformer was proposed long ago, but is widely used even at the present time.

At pressures in the apparatus from several millimeters to several microns of mercury, the discharge excited by the Tesla apparatus is concentrated at the leak, forming a brightly luminous point. The place of the leak can also be detected by a change in the color of the discharge. In the presence of leaks, the Tesla transformer excites in the apparatus a violet-red discharge, characteristic of nitrogen. Wetting the leak site with a test liquid, for example ether or acetone, and the penetration of vapors of this liquid into the apparatus causes a sharp change in the color of the discharge. The latter procedure is convenient for checking ground-glass joints, and also for detecting large leaks in metal apparatus that have glass parts.

One of the roughest methods of testing the tightness of metal parts and systems is checking them by pressure. The higher the pressure that can be created inside the object under test, the greater the sensitivity of the test proves to be. The limit in this respect is set by the mechanical strength of the system and by safety requirements. The place of a leak during such tests can be detected by the formation of bubbles when the surface of the part is coated with a soap solution or when it is immersed in water. To increase the sensitivity of the tests, the water may be treated with appropriate substances that reduce surface tension and promote the growth of bubbles[^15],[^16].

When the object under test is filled with gases containing halogens, a leak can be detected by a change in the color of the flame of a gas burner brought near it[^17]. A flame containing, for example, copper vapors then assumes a green color[^16],[^18].

Vacuum methods of checking tightness are based on the use of various manometers.

The most obvious possibility for determining leakage in vacuum systems isolated from pumping is observation of the rate of pressure change in them. The practical application of this method, however, is very strongly limited by gas evolution from the system and by the duration of the tests.

The possibility of using thermoelectric manometers to detect leaks is determined by the fact that the thermal conductivity of certain vapors and gases, which may be used as test gases, differs substantially from the thermal conductivity of air. The resistance manometer (Pirani manometer)[^19], widely used in leak-detection technology, has the greatest sensitivity in this respect.

To detect leaks in the apparatus being tested, places suspected of leakage are blown over or wetted with a test substance. As a result of a change in the composition of the gases in the volume when the test substance penetrates into it through a leak, the readings of the manometer change sharply. Wetting of the surface may be carried out, in particular, with acetone. However, the described method has the greatest sensitivity when hydrogen is used as the test gas and when a bridge circuit, balanced in the absence of hydrogen, is employed^20.

An increase in the sensitivity of tests carried out with the Pirani manometer is achieved by using the differential method^1,20,21,22 (Fig. 2). Two Pirani manometers \((M)\),

Fig. 2. Diagram of a differential leak detector.

connected to the system under test, are included in two arms of a bridge circuit. In front of one of the manometers there is placed a freezing trap \((L)\). The circuit is balanced. When hydrocarbons used as the test substance penetrate through a leak, the bridge becomes unbalanced owing to the freezing-out of the hydrocarbons in one manometer and their penetration into the other. Similar results can be obtained by chemical absorption of carbon dioxide by a trap containing calcium hydroxide^23. The latter method provides lower sensitivity.

The usual point of connection of the Pirani manometer is the fore-vacuum pipeline. Recently it has been established experimentally^24 that, for detecting leaks, a Pirani manometer installed on the high-vacuum side can be used successfully. When ether or butane is used as the test liquid, it is possible in this way to detect...

to detect very small leaks, down to \(1—10^{-3}\ \mathrm{l\cdot \mu k/sec}\), and the sensitivity turns out to depend on the volume of the apparatus, the speed with which it is pumped out, and the form of the leak.

The use of ionization and magnetic electric-discharge manometers for detecting leaks is based on the fact that the ionization efficiency of certain vapors and gases may differ substantially from the ionization efficiency of air. The use of such substances as test substances leads to the fact that, when they cover the leak, the current of the manometer decreases or increases \(^{20}\). The speed of response and the relative magnitude of the change in the readings depend on the rate at which the gas enters the system, on the speed with which this gas is pumped out, and on the ratio of the sensitivity of the manometer to this gas to its sensitivity to air.

An increase in the sensitivity of tests with the aid of an ionization manometer can be achieved by using a special compensation circuit (Fig. 3), balanced at

Fig. 3. Compensation circuit for detecting leaks by means of an ionization manometer.

the working pressure in the system and thereby allowing the instrument of high sensitivity to be used over a wide pressure range \(^{25}\).

When working with an ionization manometer, it is rational to use hydrogen as the test substance \(^{26}\).

Of very great interest for leak-detection purposes is the use of a special degassed ionization manometer with a palladium partition \(^{20,23}\). By pumping out and gettering in it, a high vacuum can be obtained. To carry out tests, the manometer is connected to the system in such a way that a palladium partition is placed between it and the system. The arrangement of the palladium leak detector is shown in Fig. 4. Palladium heated to \(700—800^\circ\mathrm{C}\) passes hydrogen well, while remaining impermeable to other gases. This makes it possible to carry out tests with sufficiently high sensitivity at comparatively high pressures in the vacuum-

system. After a leak has been detected, the hydrogen is pumped out of the manometer through the same heated partition. The partition may be heated, for example, by electron bombardment. The sensitivity of the instrument described, when carrying out leak-tightness tests, is considerably higher than that of an ordinary ionization manometer connected directly to the vacuum apparatus.

Similarly to ionization and magnetic electric-discharge manometers, an ionization manometer of the magnetron type, possessing high sensitivity, may be used for purposes of leak detection^27^. For work with it it is recommended to use

Fig. 4. Schematic representation of an ionization manometer with a palladium barrier, used as a leak detector.

a bridge circuit with two identical manometers and a trap in front of one of them, as in the differential leak detector with a Pirani manometer^20,22^, and butane or carbon dioxide as the test substance.

Among the phenomena and methods proposed in recent years for the detection of leaks, the following may be noted:

1. The dependence of the spectrum of the glow discharge in a magnetic electric-discharge manometer on the composition of the gas^28^.

2. The volume under test is coated with a fluorescent composition. The composition flows into leaks and can be detected under ultraviolet illumination^29^.

3. Without disturbing the stream of gas being pumped out, observation is made of the Schumann ultraviolet absorption spectrum. The penetration of benzene through a leak leads to strong absorption of the 1790 Å line^30^.

It is necessary to dwell on the leak-detection method associated with the use of radioactive substances. The principal merits of the method are its simplicity and high sensitivity. It is suitable for detecting leaks in objects that are difficult of access (in particular—

... in underground pipelines) and can be used to detect holes in the lead shielding of high-voltage underground cables^31,32. When working with radioactive substances, carbon monoxide containing the radioactive carbon isotope C^14, which gives β-radiation (carbon dioxide is strongly absorbed by the soil), and α-radiation emitters, in particular radon and methyl bromide containing Br^82, are used as tracer gases. The principal requirement imposed on the test gas is the possibility of detecting it at very low concentrations, which do not present a danger to the health of personnel or to the integrity of the equipment.

For the convenience of practical use of the method of detecting leaks by means of radioactive isotopes, a calculation was made of the diffusion of radioactive gases in soil^33, since the effectiveness of the method depends on the rate of diffusion of the test gas.

REFERENCES

1. R. B. Jacobs, H. F. Zuhr, J. Appl. Phys. 18, 34 (1947).
2. O. A. Nier, C. M. Stevens, A. Hustrulid und T. A. Abbot, J. Appl. Phys. 18, 30 (1947).
3. W. G. Worcester and E. G. Doughty, Trans. Amer. Inst. Elect. Enrgs. 65, 946 (1946).
4. H. A. Thomas, T. W. Williams and J. A. Hipple, Rev. Sci. Instr. 17, 368 (1946).
5. N. Warmoltz, Appl. Sci. Research B2, 61—65 (1951).
6. C. J. Milner, Vac. Phys., J. Sci. Inst., Suppl., No. 1, 29 (1951).
7. A. K. Storruste, Norske Videnskab Selskabs Forh. 22, No. 6, 18 (1949)
8. M. E. Reinders, J. Schutten, J. Kistemaker, Appl. Sci. Res. B2, No. 1, 66—70 (1951).
9. G. P. Barnard, Modern Mass Spectrometry, London, 1951.
10. Rev. Sci. Inst. 24, No. 5 (1953); J. Appl. Phys. 24, No. 7 (1953).
11. Refrig. Eng. 55, 275 (1948).
12. Refrig. Eng. 57, 816 (1949).
13. Refrig. Eng. 61, 911 (1953).
14. Brit. J. Appl. Phys. 3, No. 5 (1952).
15. J. B. Brightman and E. G. Chace, Refrig. Eng. 55, 552 (1948).
16. D. H. Crampton and C. Winnefeld, Refrig. Eng. 55, 251 (1948).
17. Refrig. Eng. 55, 593, 996 (1948).
18. Refrig. Eng. 55, 74 (1948).
19. T. E. Cuykendall, Rev. Sci. Inst. 6, 371 (1935).
20. J. Blears and J. H. Leck, Vacuum Phys., J. Sci. Inst., Suppl., No. 1, 20 (1951).
21. A. H. Turnbull, AERE—G/R—477 (March 1950).
22. J. Blears and J. H. Leck, Brit. J. Appl. Phys. 2, 227 (1951).

1. N. Ochert and Steckelmacher, Vacuum 2, 125 (1952).
2. R. N. Bloomer and M. E. Haine, J. Sci. Inst. 30, 385 (1953).
3. C. Nief, Brit. J. Appl. Phys. 3, 29 (1952).
4. J. A. Jacobs and P. R. Malmberg, Rev. Sci. Inst. 24, 399 (1953).
5. A. H. Beck and A. D. Brisbaue, Vacuum 2, 137 (1952).
6. R. Rimura, M. Sakisaka and S. Miyashiro, Bull. Inst. Chem. Research, Kyoto Univ. 27, 55 (1951).
7. E. Völcker, Zeits. f. Naturforsch. 6a, 512 (1951).
8. J. Romand, V. Schwetzoff and B. Vodar, Congr. groupment. avance méthodes anal spectrograph. produits. mét. 14, 201 (1951).
9. A. Gemant, E. Hines and Alexanderson, J. Appl. Phys. 22, No. 4, 460 (1951).
10. J. E. Johnston, J. Brit. Inst. Radio Engrs. 11, 279 (1951).
11. A. Gemant, J. Appl. Phys. 24, No. 1, 93 (1953).