Leak detection

K. Zapfe
Deutsches Elektronen-Synchrotron DESY, Hamburg, Germany

Abstract
This paper will give an introduction to the leak detection of vacuum
systems. Various methods to detect leaks as well as the most widely used
helium leak detectors and their different applications are presented. Practical
examples in the context of accelerator vacuum systems will illustrate the
topic.

1 Introduction
Ideally a vacuum chamber should maintain the achieved vacuum pressure forever after switching off
the pumps. However, without active pumping the pressure in a real system will rise with time. This
pressure rise is produced by outgassing—the spontaneous evolution of gas molecules from the
walls—and gas molecules penetrating through leaks and entering by permeation from the outside into
the vacuum system. Figure 1 shows three typical curves for the behaviour of the pressure increase as a
function of time on a linear scale: outgassing starting with a linear pressure increase and levelling off
once outgassing and vapour pressure reach equilibrium, the linear increase due to a leak, and the
combined curve of the two effects.

Fig. 1: Pressure increase due to outgassing, a leak, and the combination of outgassing and a leak

In practice it is impossible to build a completely leak-tight vacuum system. And this is not even
necessary. On the other hand, the leak rate must be small enough to allow the required pressure level
to be reached. Therefore it is important to specify an acceptable leak rate for each vacuum system.
After manufacturing of a vacuum vessel it must be proven that the tightness specifications are
fulfilled. Further checks are necessary during as well as after assembly and installation to locate
possible leaks created during the previous steps. Thus leak detection is an important step in the
production of vacuum to guarantee that the required pressure and gas composition conditions of a
vacuum system can be reached. Adequate methods and leak detection equipment have been developed
over the past decades following the ever-increasing demands by industry.

227
 K. Z APFE

This lecture will start with some introductory remarks about leak rates and leak types followed
by various leak detection methods. The most widely used helium leak detectors and their different
applications will be described in detail. This will be illustrated by practical examples in the context of
accelerator vacuum systems.

2 Leak rate
The leak rate is defined as the pV throughput of a gas through a leak. It is a function of the type of gas,
pressure difference, and temperature. In a system of volume V the leak rate Ql is given by

Δp
Ql = V ⋅ . (1)
Δt

Here Δp is the pressure rise during the time interval Δt . Commonly used units for leak rates and their
conversion factors are given in Table 1.

Table 1: Conversion factors for leak rates in various system units [1]

mbar · l/s Torr · l/s Pa · m3/s cm3/s*

mbar l/s 1 0.75 0.1 0.99
Torr l/s 1.33 1 0.133 1.32
3
Pa m /s 10 7.5 1 ~10
3 *
cm /s 1.01 0.76 0.101 1
* o
STP - standard temperature and pressure (0 C, 1 atm)

As an example, for a high vacuum (HV) system one can take the leak rates given below as a
rule of thumb:
– Ql < 10–6 mbar l/s: very tight system

– Ql < 10–5 mbar l/s: tight system

– Ql < 10–4 mbar l/s: leaky system.

The following examples illustrate the relationship between the size of a pore, the corresponding
leak rate and the amount of gas entering into a vacuum system. For simplification it is assumed that
the pore is a straight channel of circular shape. A diameter of 0.01 mm, e.g. a hair, corresponds to a
leak rate of 10–2 mbar l/s. For a pore with a leak rate of Ql = 10–10 mbar l/s = 10–10 cm3/s, an amount of
1 cm3 of gas needs 317 years to flow through the leak channel.

3 Leak types
Leaks may appear because of various defects within the material and/or at connecting areas:
– fixed connections by brazing, welding, or gluing—especially transitions between different
materials like glass-metal, ceramics-metal, etc;

– pores and hair cracks due to mechanical—or thermal stress, which to some extent are always
present and therefore must be small enough in size and number not to cause disturbance;

228
L EAK DETECTION

– flanged connections;

– cold/warm leaks opening up at extreme temperatures—often being reversible;

– virtual leaks, where gas evaporates from inner excavations, dead holes etc., leading eventually
to long pumping times (see Sub-section 8.1) and

– indirect leaks from supply lines, for example, for cooling water or gas/liquids (He, N2) of
cryogenic systems.

A leakage, but not a leak due to a defect is the

– permeation, i.e., the natural porosity of material.

The permeation could be quite significant and even limit the detectable leak rate. For example,
at HERA large amounts of Perbunan® rubber seals with a permeation rate of 2 · 10–2 mbar l mm/s/m2
are used at the insulating vacuum tank resulting in a total permeation rate of 2 · 10–5 mbar l/s/m [2].
This long list shows that during design and manufacture great care is necessary to avoid
potential leaks.

4 Leak detection methods

The aims of a leak search are to localize a leak and/or to determine the total or local leak rate.
Depending on the size of the leak, various effects can be used for leak detection. All methods are
based on the variation of a physical property measured on one side of the vessel while the pressure or
the nature of the gas is changed on the other side. Large leaks can generate mechanical effects as
described in Sub-section 4.1, while smaller leaks require more sophisticated methods (see Sections 4.2
and 4.3). A detailed list of possible leak detection methods, their sensitivities, and references can be
found in Ref. [3]. A comprehensive review of leak detection methods and apparatus is given in
Refs. [1], [4] and [5].

4.1 Mechanical effects

Methods applying measurable mechanical effects are limited to large leaks. For example ultrasound
detectors may be used to monitor the oscillations produced by the gas in the vicinity of a leak. The
detection limit of this method is limited to leak rates of 10–2 mbar l/s. Somewhat more sensitive is the
formation of bubbles when water or even soaped water is spread on a leak with the vessel being
pressurized to over pressure. A sensitivity of 10–4 mbar l/s can be reached.
These methods are simple, very quick to carry out, cheap and one can locate a leak. However,
owing to the limited sensitivity they are restricted to the high-pressure region.

4.2 Pressure increase

In approaching the problem of a supposed leak larger than the admissible value, it is necessary to
determine first if such a leak actually exists. Taking curves of the pressure versus time as shown in
Fig. 1 will assist in determining if there is a leak and estimating the actual leak rate. First the system is
evacuated to a stable minimum pressure. When no further improvement in the pressure is evident, the
pumps are switched off or separated from the system. There will always be an initial pressure rise due
to outgassing, thus one will have to wait sufficient time until outgassing has come to equilibrium with
the vapour pressure in the gas phase. In case of a leak, the pressure will continue to increase linearly
with a different gradient instead of levelling off. The pressure curve should be taken till the shape of
the curve becomes evident. More details might be found in Refs. [3] and [6].

229
 K. Z APFE

However, it requires a lot of experience to reliably analyse such curves during initial pump
down with still a lot of outgassing or in the presence of virtual leaks. Here the application of a liquid
nitrogen trap might be useful to minimize the effect of the vapours present in the system. Small leaks
might not be visible with this method. In any case additional methods as described in the following
sub-section need to be applied for further diagnosis.

4.3 Modification of the physical properties of the residual gas

By changing locally the air composition in the vicinity of a leak by adding a gas (the tracer gas), the
composition of the residual gas and thus its physical properties are modified. As accurate and sensitive
measurement methods for many physical gas properties are available, measuring these alterations can
be used to determine the position and size of a leak. The sensitivity of such methods is sufficient to
detect even quite small leaks. In the following the most widely used methods are briefly described.
The variation of the heat conductivity can be detected using a Pirani gauge and alcohol, CO2, or
helium as tracer. The pressure will rise in case of helium, otherwise drop down. Using (heavy) noble
gases, the change in ionization cross-section can be monitored by the signal of an ion gauge or even a
sputter ion pump. A nice example of this method is described in Ref. [7].
Analysing the mass of the residual gas is the most sensitive and widespread method in locating
leaks. Using an optimized mass spectrometer and helium as tracer gas, leak rates down to
10–12 mbar l/s can be detected. A more detailed description will follow in Section 5.

4.4 Tracer gas

The tracer gas should have the following properties:
– unambiguous signal in the mass spectrum of the residual gas;

– chemically and physically inert, non explosive and cheap;

– very low content in air and

– easily removable by pumping and does not contaminate the system.

Helium is most commonly used since it fulfils all conditions mentioned above. Therefore the
leak rate is usually given as helium standard leak rate (He Std) assuming a pressure difference of 1 bar
from air to vacuum. The small diameter of the helium atoms allows the detection of very small leaks.
As the speed of helium is three times higher than the speed of air, the amount of helium entering
through a leak and thus the sensitivity is increased by a factor of 3 compared to air.
Alternatively, argon is frequently used as tracer gas. However, the signal in the mass spectrum
of the residual gas has a mixed signature of mass 40 for Ar+ and hydrocarbons and mass 20 for Ar++,
20
Ne and 18OH2. In addition the natural content of argon in the ambient air is significantly higher than
that of helium.

5 Helium leak detectors

In principle any type of residual gas analyser can be used as mass spectrometer helium leak detector.
The most sensitive and safe device, however, is a mass spectrometer with 180° magnetic sector field
optimized for the detection of mass 4.

230
L EAK DETECTION

Fig. 2: Schematic drawing of a helium leak detection cell with 180° magnetic sector field (left)
and photo of the slit system (right)

Figure 2 shows a schematic drawing and a photograph of such a helium leak detection cell. The
central parts are an ion source, a separation system, and an ion collector system. The residual gas is
ionized by bombardment with electrons. The resulting ions are accelerated into a magnetic field,
where they are deflected. A system of slits allows the helium atoms to pass to a collector unit located
at a deviation of 180°. All other ions are blocked. The magnetic field of about 1.5 kg G is typically
produced by permanent magnets. A second collector at the middle plate is used to measure the ion
current of heavy ions, which is then converted into a total pressure value.
To detect tiny leaks the currents to be measured are extremely small. At the highest sensitivity
of 10–12 mbar l/s currents as low as 10–15 A have to be measured. This is achieved thanks to the use of
electron multipliers in the most modern detectors.
In a mass spectrometer the typical path length of the ions to be collected is about 15 cm. As the
ions should pass without collisions with other gas molecules, the operating pressure should be below
10–4 mbar corresponding to a mean free path of 60 cm.

6 History of helium leak detectors

In 1910 J.J. Thomson described the first mass spectrometer used to detect neon isotopes. For several
decades such huge apparatus, easily filling a room, were operated by specialists only. The idea of
using mass spectrometers for leak detection goes back to the Manhattan Project in 1942/43 [8]. The
leak-tightness requirements needed for the apparatus of the uranium enrichment plants triggered the
need for very sensitive and reliable leak detection systems. This led finally to the choice of a mass
spectrometer tuned on the mass of helium designed by Dr. A.O. Nier [9]. The first publication dates
back to 1947.
During the 1950s and 1960s the original choice of glass as material was replaced by a metal
solution since it was much more robust and adequate for industrial use. Also the first compact units
including a pump station—weighing about 200 kg—came onto the market. While today the leak
detection cell is still very similar to the original design, the pumping systems have considerably
improved replacing the original diffusion pumps by turbomolecular or even dry pumps. The
sensitivity of helium leak detectors has increased from an initial value of 10–6 mbar l/s by 6 orders of
magnitude down to 10–9 mbar l/s around 1970 and 10–12 mbar l/s today.

231
 K. Z APFE

7 Helium leak detection

When searching for leaks at vacuum vessels the system to be tested is continuously evacuated. The
tracer gas penetrating from the outside into the system is pumped through a leak detector where its
concentration is measured. For leak location the tracer gas is sprayed locally, while for total leakage
measurement, the vacuum vessel is completely enclosed by the tracer gas as shown schematically in
Fig. 3.

Fig. 3: Leak detection of vacuum vessels using the tracer method – left: leak location, right: total
leakage measurement

For pressurized systems the detector method is usually applied. The device to be investigated is
pressurized with the tracer gas. For leak location a sniffer probe is connected to a conventional helium
leak detector by a capillary tube or needle valve to reduce the pressure from atmosphere to the
maximum admissible pressure of about 10–4 mbar (see Fig. 4, left). The detection limit of the ‘sniffer
method’ of about 10–7 mbar l/s is determined by the natural He content of air. For total leakage
measurement the vessel is placed in an evacuated test chamber with detector unit as shown in the right
part of Fig. 4.

Fig. 4: Leak detection of pressure vessels using the detector method – left: leak location using a
sniffer, right: total leakage measurement

The sniffer method can also be applied in case of very large leaks of vacuum systems, when the
use of any classical leak detection method is impossible as the pressure can not be sufficiently
lowered. In that case the sniffer can be used to detect the presence of helium in the exhaust line of a
roughing pump while spraying helium onto the system.

7.1 Direct flow method

In principle the leak detector can be connected to a vacuum system such that the tracer gas flows
completely or only partially through the detection cell. Using the direct-flow method, both the leak
detection cell and pumps are directly connected to the vacuum system as indicated in Fig. 5. In this
case the helium leak rate QHe is given by the effective helium pumping speed Seff, He of the high-
vacuum pump and the helium partial pressure pHe within the cell:
QHe = pHe ⋅ Seff, He . (2)

232
L EAK DETECTION

Fig. 5: Layout for the direct flow method with cold trap (left) and throttle valve (right)

The ratio between the partial pressure increase in the leak detection cell and the helium flow
through the leak defines the intrinsic sensitivity of a leak detector. In order to increase the sensitivity,
the pumping speed for the tracer gas has to be reduced [see Eq. (2)]. This must be done without
diminishing the pumping speed for the other gases in order to keep an appropriate operating pressure
for the filament emitting the ionizing electrons. As leak detection usually takes place in unbaked
systems, selective pumping is therefore needed to provide a high pumping speed for water and a low
pumping speed for helium.
Originally diffusion pumps were most commonly used to produce high vacuum. They were
combined with a liquid nitrogen trap to effectively condense water (left part of Fig. 5). The trap also
impeded back-streaming of oil vapour from the diffusion pump to the leak detection cell. This
arrangement has been very successful for many years, although their operation had several
disadvantages. The cold trap needed to be refilled periodically during operation thus requiring an easy
access to a liquid nitrogen source. The diffusion pumps had to be operated in quite well defined ways
as they were sensitive to misuse such as inadequate venting. Since the 1980s turbomolecular pumps
started to replace the diffusion pumps and cold traps.
In order to increase the sensitivity, a throttle valve could be installed as indicated in the right
part of Fig. 5 to reduce the gas flow to the pumps. According to Eq. (2), decreasing the pumping speed
by a factor of 10 decreases the minimum detectable leak rate by the same factor. The direct flow
method is quite fast and very sensitive. Nevertheless, in most commercial leak detectors it is
nowadays replaced by counter-flow detectors.

7.2 Counter-flow method

In 1968 W. Becker proposed a different arrangement of the pumps and leak detection cell [10]. As
shown schematically in the left part of Fig. 6 the leak detection cell is no longer connected directly to
the vacuum system, but to the inlet of a high-vacuum pump. However, the realization of the counter-
flow method still took several years. Only in the middle of the 1970s were adequate turbomolecular
pumps available on the market [11, 12].

233
 K. Z APFE

Fig. 6: Layout for the counter-flow method: single-stage high-vacuum pump (left) and two-stage
turbomolecular pump (right)

The method is based on the fact that the compression ratio K of turbomolecular (and diffusion)
pumps increases rapidly with the mass of the pumped gas. Hence by injecting the gas from the tested
vessel at the exhaust of the pump, it is possible to obtain at its inlet a backstreaming flux largely
enriched for lighter gases. Typical values for the compression ratio of a turbomolecular pump running
at full speed are KHe = 50, K H2O = 4000 and K N 2 = 30 000 resulting in a suppression by a factor of 80
for water and 600 for nitrogen. Compared to the direct flow method the helium partial pressure within
the detection cell is reduced by the turbomolecular pump by the factor K resulting in a leak rate QHe
of:
QHe = pHe ⋅ Seff, He ⋅ K . (3)

A major drawback of this simple method is the direct connection of the tested vacuum vessel to
the roughing pump and thus the risk of its contamination by oil vapour. In addition, the stability of the
pumping characteristics is important to ensure the necessary stability for accurate leak detection. The
typically small pumping speed of the roughing pump significantly increases the time constant for leak
detection.
As a remedy to the problems described above, more sophisticated commercial leak detectors
have been developed using specially designed turbomolecular pumps. For example, installing a
second turbomolecular pump between vacuum vessel and roughing pump ensures clean pumping with
high pumping speed and thus a short time constant. Nowadays two-stage turbomolecular pumps are
most widely used having an outlet flange between the two stages to the leak detection cell (right part
of Fig. 6). Alternatively, a simple counter-flow leak detector can be connected to the outlet of the
turbomolecular pump in a pump station. With the recent developments of dry pumps, more and more
counter-flow leak detectors using dry pumps are available on the market, thus avoiding
contaminations from oil vapour.
The counter-flow detectors offer several advantages compared to the direct-flow detectors.
They do not need cold traps. The tracer gas no longer flows directly through the detection cell, but just
a part of it by backstreaming through the high vacuum pump. Hence for the same flow, the pressure in
the cell is lower and consequently the detection can start earlier, i.e., at a higher pressure level of

234
L EAK DETECTION

about p ≅ 10–1 mbar. In addition the leak detection cell and thus the filament are better protected in
case of a sudden pressure increase. As a consequence, however, they are less sensitive (1 · 10–10 mbar
l/s) than direct-flow detectors, the latter one being still unbeatable in the case of very small leaks.

7.3 Characteristics
Various parameters characterize a leak detector. Most important are the sensitivity, detection limit,
response time, and maximum allowable inlet pressure.
According to Eqs. (2) and (3) the intrinsic partial pressure sensitivity s for the direct flow (DF)
and counter flow (CF) detector are given by the helium partial pressure and the helium pumping speed
at the connecting flange of the vacuum system:
pHe 1
sDF = = (4)
QHe Seff
and
pHe 1
sCF = = . (5)
QHe Seff ⋅ K He

For the overall sensitivity the intrinsic sensitivity and the sensitivities of the mass spectrometer and
the amplifier need to be taken into account [1].
The smallest detectable leak or detection limit is given by the ratio of the minimum detectable
signal and the overall sensitivity. This means that in practice the total pressure in the detection cell
needs to be taken into account. Thus using the same leak detector, the minimum detectable leak is
different when testing a small vessel or a 100 m long accelerator section. The detection limit could
also be limited due to a significant helium level in the atmosphere surrounding the leak detector. This
could lead to a permanent high helium background signal from, for example, helium entering by
permeation through O-rings, small internal leaks in the detector, or via the exhaust line into the
roughing pump and thus being stored in the pump oil. To counteract the latter case the roughing pump
could be flushed by running with gas ballast for some period.
Another important characteristic is the time-dependent behaviour of the helium signal. This
depends not only on the leak detector itself but also on the system under investigation. After spraying
helium onto a vessel of volume V the helium partial pressure pHe will start to increase to its
equilibrium value QHe/Seff according to

QHe ⎛ ⎞
t
−
pHe = ⎜1 − e τ ⎟ . (6)
Seff ⎜⎝ ⎟
⎠
Here τ is the time constant
V
τ= . (7)
Seff

Usually the response time is defined as the time to reach 95% of the equilibrium value. Three
examples for the time-dependent behaviour of the detector signal are shown in Fig. 7. In the case of
curve 1 the signal quickly reaches its equilibrium value. Increasing the volume by a factor of 10 the
response time will be 10 times longer (curve 2). Increasing the pumping speed will decrease the
response time, but according to Eq. (2) also reduce the signal by the same factor as shown in curve 3.
One has to note that the signal will start to rise after a dead time only.
When removing the test gas from the system, the recovery behaviour for the three cases is
similar as plotted on the right side of Fig. 7.

235
 K. Z APFE

Fig. 7: Time-dependent behaviour of the helium signal for a leak (left) and during pump out
(right)

Before leak detection is started, the system must be pumped down such that the maximum inlet
pressure of the leak detector has been reached. The warm-up or start-up time defines the period
needed till stable operation of the mass spectrometer is possible.
In order to perform quantitative measurements, helium leak detectors need to be calibrated with
a standard leak. This needs to be repeated regularly as the long-term stability is usually insufficient,
especially when working on large vacuum systems. Drifts from the electronics of the detector need to
be checked and corrected regularly where necessary. Details of the procedures and calibration
equipment may be found in Refs. [5] and [13].

8 Practical experience and examples

For practical applications the choice of a proper leak detector is important. Depending on the volume
to be checked (e.g., small set-ups versus accelerator sections), vacuum level (e.g., O-ring-sealed
versus metal-sealed systems), tolerable/expected leak rates (e.g., HV versus UHV systems) or
frequency (mass production versus individual measurement), various leak detector systems are
available on the market.
A leak search of larger systems requires systematic work. Checking the proper functioning and
leak tightness of the leak detector first are mandatory. Working with reduced gas flow from helium
bottles, the system should be searched from top to bottom and against an existing air flow when
working, for example, in a tunnel. One should also keep in mind that small leaks might initially be
blocked by water and only open up after sufficient pumping time or even after bake-out. This may
require several leak checks at various vacuum levels.
Recording the pressure and other relevant parameters during pump-down and operation of a
vacuum system can save a lot of time in analysing and diagnosing the behaviour of the system. Any
deviation from the usual pump-down curve is a hint of a possible leak. If the system is pumped for the
first time, more experience is necessary to evaluate the pressure curve. As indicated in Fig. 8, the
pressure of a tight system decreases linearly with time in a log-log scale, while in the presence of a
leak the pressure tends to level off at a value given by the ratio of the leak flow to the pumping speed
[7]. Similar information can be taken from the pressure increase versus time when isolating the
vacuum system from the pumps as described in more detail in Section 4.2 and shown in Fig. 1.

236
L EAK DETECTION

Fig. 8: Pressure versus time during pump-down with and without a leak [7].

As it is much faster and easier to find and repair leaks in a laboratory than in an accelerator, a
careful check of all components prior to installation is strongly recommended. This should include the
use of metallic gaskets on all sealing surfaces.

8.1 Virtual leaks

A virtual leak looks like a real leak in a pump-down or pressure rise curve. However, there will be no
signal in a leak detector and it cannot be located from outside. Usual causes of such leaks are a poor
design and/or poor manufacture of the vacuum vessel, e.g., excavations, dead holes for screws, air
enclosures in a weld or narrow slits leading to excessive outgassing. Diagnosis could be done using a
residual gas analyser and measuring the gas composition before and after venting with argon. After
such a venting, the argon replaces the nitrogen in the virtual leak leading to an argon enhancement in
the residual gas after the second evacuation.

8.2 Cold leaks

When operating a vacuum system at extreme temperatures (hot or cold), additional leaks could open
up. A typical arrangement for testing a vacuum vessel, e.g., a superconducting cavity at temperatures
below 4.5 K, is shown in Fig. 9. The vessel is immersed in liquid helium, while in general pumps,
vacuum gauges, leak detector, and residual gas analyser are operated in the room temperature part of
the system. In the presence of a ‘cold leak’ He atoms will directly enter from the liquid into the
vacuum vessel and at least be partially absorbed onto the cold walls. Thus the delay between the
atoms entering the system and a significant He signal in the detectors might be very long. In practice
such systems are warmed up to about 10 K after a certain time period to release the He atoms from the
walls to measure the total leak rate. Leak location requires the system to be warmed up to room
temperature; however it is not exceptional that cold leaks close reversibly and/or are not detectable
after warm up.

237
 K. Z APFE

Alternatively, hot extractor gauges can be operated at liquid helium temperatures as indicated in
Fig. 9. As these gauges are very sensitive to helium, they can be used for in situ leak detection as
described in more detail in Ref. [14].

Fig. 9: Set-up for testing a superconducting cavity in liquid helium (LHe) at temperatures below 4.5 K

8.3 Total leakage measurement

As described in Section 7, for total leakage measurement the vacuum vessel is completely enclosed by
a tracer gas. By leaving the system without active pumping and thus accumulating the tracer gas, very
small leaks can be detected, especially on big volumes. For example, in the HERA proton ring large
parts of the beam pipe are surrounded by liquid helium from the superconducting magnets. The
integral leak rate of a section of about 1.4 km was determined by looking for helium during a warm-
up of the beam pipe after a longer period at liquid helium temperature. Using a residual gas analyser
no helium signal could be detected resulting in a total leak rate below 10–15 mbar l/s at 4.5 K [15].
Examples for total leakage measurements of pressurized systems are given elsewhere in these
proceedings [16].

9 Helium leak detectors - state of the art

There is a wide range of commercial leak detectors on the market. Very compact and even portable
units are available. Triggered by the semiconductor industry, an increasing number of oil-free leak
detectors are available.
Standard detectors use the counter-flow method. A flexible system of valves allows the
injection of the helium flow into the detection cell from various locations of the pumping system, thus
giving a huge range of sensitivity from quite large to very small leaks. The lowest quoted values for
the minimum detectable leak are 5 · 10–12 mbar l/s and 5 · 10–8 mbar l/s for sniffers. The mass selection
system is optimized for 4He as the standard tracer gas, mostly including the option to use masses 2
(H2) and 3 (3He).

238
L EAK DETECTION

The electronics is usually equipped with semiconductors and thus sensitive to radiation. For
applications requiring a radiation-hard solution, there is no alternative to old-fashioned tubes. Typical
start-up times are of the order of a few minutes, response times less than one second. A high degree of
automation is standard, usually including calibration, tuning, and regular correction of the zero line.

10 Concluding remark
Accelerator vacuum systems do have demanding requirements for leak tightness. Adequate leak
detection methods, appropriate leak detectors, and well trained personnel are required to prove that the
leak tightness specifications are fulfilled and to detect and locate the leaks. The various methods
presented during this lecture are well suited to measure and localize the full range of leaks from very
large to tiny. A variety of suitable commercial leak detectors are available on the market. There has
been great progress in operational availability, performance, and size since 1950 when the first
devices came onto the market. During the past 20 years improvements have been mainly in handling
and automation. Today’s leak detection equipment is robust, easy to use, and offers a large range of
sensitivity. Nevertheless, finding and repairing leaks is a time-consuming process, especially in the
case of emergency leak testing of a running accelerator system. It still requires well-trained and
experienced technicians with good knowledge of the system to be checked. Therefore it is most
important to prevent leaks during all stages of design, manufacture, and assembly.

Acknowledgements
I would like to thank my colleagues of the vacuum group MVP at DESY for their patience in
introducing me to the field of vacuum techniques and skills of leak detection. Special thanks are due
to N. Mildner and A. Wagner for preparing part of the drawings, S. Holm for taking some of the
photographs as well as performing measurements for this lecture, and N. Hilleret of CERN for careful
reading of the manuscript.

References
[1] J.M. Lafferty, Foundations of Vacuum Science and Technology (Wiley, New York, 1998).
[2] D. Trines et al., Proc. XVth Int. Conf. on High Energy Accelerators, Int. J. Mod. Phys. A (Proc.
Suppl.) 2A (1993) 347.
[3] A. Roth, Vacuum Technology (Elsevier, Amsterdam, 1990).
[4] F.F. O’Hanlon, A User’s Guide to Vacuum Technology (Wiley, New York, 1989).
[5] M. Wutz, H. Adam and W. Walcher, Handbuch Vakuumtechnik (Vieweg, Braunschweig,
1997).
[6] C.A. Schalla, J. Vac. Sci. Technol. 12 (1975) 430.
[7] N. Hilleret, Leak detection, S. Turner (Ed.), Proc. CAS Vacuum Technology, Snekersten, 1999,
CERN 99/05 (CERN, Geneva, 1999), p. 203.
[8] A. Nerken, J. Vac. Sci. Technol. A9 (1991) 2036.
[9] A.O. Nier, C.M. Stevens, A. Hustrulid and T.A. Abott, J. Appl. Phys. 18 (1947) 30.
[10] W. Becker, Vak. Techn. 8 (1963) 203.
[11] M.H. Hablanaian and W.E. Briggs, Proc. 7th Int. Vac. Congress, Vienna, 1977, p. 199.
[12] W. Becker and W. K. Huber, Proc. 7th Int. Vac. Congress, Vienna, 1977, p. 203.

239
 K. Z APFE

[13] http://www.iso.ch ISO 3530:1979.

[14] M.G. Rao, Adv. Cryog. Eng. 41 (1996) 1783.
[15] D. Trines et al., Proc. XVth Int. Conf. on High Energy Accelerators, Int. J. Mod. Phys. A (Proc.
Suppl.) 2A (1993) 344.
[16] K. Zapfe, Large systems commissioning, these proceedings.

240