LEAK DETECTION IN VACUUM SYSTEMS
Before assuming that a vacuum system leaks, it is useful to consider if any other problem is present. The most
important tool in such a consideration is a properly maintained log book of the operation of the system. This is particularly the case if several people or groups use a single system. If key check points in system operation are recorded weekly, or even monthly, then the task of detecting a slow change in performance is far easier.
Leaks develop in cracked braze joints, or in torchbrazed joints once the flux has finally been removed.
Demountable joints leak if the sealing surfaces are badly scratched, or if a gasket has been scuffed, by allowing
the flange to rotate relative to the gasket as it is compressed.Cold flow of Teflon or other gaskets slowly reduces the compression and leaks develop. These are the easy leaks to detect, since the leak path is from the atmosphere into the vacuum chamber, and a trace gas can be used for detection.
A second class of leaks arise from faulty construction techniques; they are known as virtual leaks. In all of these,a volume or void on the inside of a vacuum system communicates to that system only through a small leak path.Every time the system is vented to the atmosphere, the void fills with venting gas, then in the pumpdown this gas flows back into the chamber with a slowly decreasing throughput, as the pressure in the void falls. Pressures and pumping speeds developed by a steady throughput of gas (Q) through a vacuum chamber, conductance (C) and pump.the system pumpdown.
A simple example of such a void is a screw placed in a blind tapped hole. A space always remains at the bottom of the hole and the void is filled by gas flowing along the threads of the screw. The simplest solution is a screw with a vent hole through the body,providing rapid pumpout. Other examples include a double O-ring in which the inside O-ring is defective, and a double weld on the system wall with a defective inner weld. A mass spectrometer is required to confirm that a virtual leak is present. The pressure is recorded during a
routine exhaust, and the residual gas composition is determined as the pressure is approaching equilibrium. The system is again vented using the same procedure as in the preceding vent, but the vent uses a gas that is not significant in the residual gas composition; the gas used should preferably be nonadsorbing, such as a rare gas.After a typical time at atmospheric pressure, the system is again pumped down. If gas analysis now shows significant vent gas in the residual gas composition, then a virtual leak is probably present, and one can only look for the culprit in faulty construction.
Leaks most often fall in the range of 10¯4 to 10¯6 torr-L/s. The traditional leak rate is expressed in atmospheric cubic centimeters per second, which is 1.3 torr-L/s. A variety of leak detectors are available with practical sensitivities varying from around 1X10¯3 to 2 X 10¯11 torr-L/s.The simplest leak detection procedure is to slightly pressurize the system and apply a detergent solution, similar to that used by children to make soap bubbles, to the outside of the system. With a leak of ~1X10¯3 torr-L/s, bubbles should be detectable in a few seconds. Although the lower limit of detection is at least one decade lower than this figure, successful use at this level demands considerable patience.
A similar inside-out method of detection is to use the kind of halogen leak detector commonly available for
refrigeration work. The vacuum system is partially backfilled with a freon and the outside is examined using a sniffer hose connected to the detector. Leaks the order of 1X10¯5 torr-L/s can be detected. It is important to avoid any significant drafts during the test, and the response time can be many seconds, so the sniffer must be moved quite slowly over the suspect area of the system. A far more sensitive instrument for this procedure is a dedicated helium leak detector (see below) with a sniffer hose testing a system partially back-filled with helium.A pressure gauge on the vacuum system can be used in the search for leaks. The most productive approach applies if the system can be segmented by isolation valves. By appropriate manipulation, the section of the system containing the leak can be identified. A second technique is not so straightforward, especially in a nonbaked system. It relies on the response of ion or thermal conductivity gauges differing from gas to gas. For example, if the flow of gas through a leak is changed from air to helium by covering the suspected area with helium, then the reading of an ionization gauge will change, since the helium sensitivity is only ~16% of that for air. Unfortunately, the flow of helium through the leak is likely to be ~2.7 times that for air, assuming a molecular flow leak, which partially offsets the change in gauge sensitivity. A much greater problem is
that the search for a leak is often started just after exposure to the atmosphere and pumpdown. Consequently outgassing is an ever-changing factor, decreasing with time.
Thus, one must detect a relatively small decrease in a gauge reading, due to the leak, against a decreasing background pressure. This is not a simple process; the odds are greatly improved if the system has been baked out, so that outgassing is a much smaller contributor to the system pressure.
A far more productive approach is possible if a mass spectrometer is available on the system. The spectrometer is tuned to the helium-4 peak, and a small helium probe is moved around the system, taking the precautions described later in this section. The maximum sensitivity is obtained if the pumping speed of the system can be reduced by partially closing the main pumping valve to increase the pressure, but no higher than the mid-10¯5 torr range, so that the full mass spectrometer resolution is maintained. Leaks in the 1X 10¯8 torr-L/s range should be readily detected.
The preferred method of leak detection uses a standalone helium mass spectrometer leak detector (HMSLD).Such instruments are readily available with detection limitsof 2 X10¯10 torr-L/s or better. They can be routinely calibrated so the absolute size of a leak can be determined. In many machines this calibration is automatically performed at regular intervals. Given this, and the effective pumping speed, one can find, using Equation 1,whether the leak detected is the source of the observed deterioration in the system base pressure.
In an HMSLD, a small mass spectrometer tuned to detect helium is connected to a dedicated pumping system,
usually a diffusion or turbomolecular pump. The system or device to be checked is connected to a separately pumped inlet system, and once a satisfactory pressure is achieved, the inlet system is connected directly to the detector and the inlet pump is valved off. In this mode, all of the gas from the test object passes directly to the helium leak detector. The test object is then probed with helium, and if a leak is detected, and is covered entirely with a helium blanket, the reading of the detector will provide an absolute indication of the leak size. In this detection mode, the pressure in the leak detector module cannot exceed ~10¯4 torr, which places a limit on the gas influx from the test object. If that influx exceeds some critical value, the flow of gas to the helium mass spectrometer must be restricted, and the sensitivity for detection will be reduced. This mode, of leak detection is not suitable for dirty systems, since the gas flows from the test object directly to the detector, although some protection is usually provided by interposing a liquid nitrogen cold trap.
An alternative technique using the HMSLD is the socalled counterflow mode. In this, the mass spectrometer
tube is pumped by a diffusion or turbomolecular pump which is designed to be an ineffective pump for helium
(and for hydrogen), while still operating at normal efficiency for all higher-molecular-weight gases. The gas
from the object under test is fed to the roughing line of the mass spectrometer high-vacuum pump, where a higher pressure can be tolerated (on the order of 0.5 torr).
Contaminant gases, such as hydrocarbons, as well as air, cannot reach the spectrometer tube. The sensitivity of an HMSLD in this mode is reduced about an order of magnitude from the conventional mode, but it provides an ideal method of examining quite dirty items, such as metal drums or devices with a high outgassing load.
The procedures for helium leak detection are relatively simple. The HMSLD is connected to the test object for
maximum possible pumping speed. The time constant for the buildup of a leak signal is proportional to V/S, where V is the volume of the test system and S the effective pumping speed. A small time constant allows the helium probe to be moved more rapidly over the system.
For very large systems, pumped by either a turbomolecular or diffusion pump, the response time can be
improved by connecting the HMSLD to the foreline of the system, so the response is governed by the system
pump rather than the relatively small pump of the HMSLD. With pumping systems that use a capture-type
pump, this procedure cannot be used, so a long time constant is inevitable. In such cases, use of an HMSLD and helium sniffer to probe the outside of the system, after partially venting to helium, may be a better approach.Further, a normal helium leak check is not possible with an operating cryopump; the limited capacity for pumping helium can result in the pump serving as a low-level source of helium, confounding the test.
Rubber tubing must be avoided in the connection between system and HMSLD, since helium from a large
leak will quickly permeate into the rubber and thereafter emit a steadily declining flow of helium, thus preventing use of the most sensitive detection scale. Modern leak detectors can offset such background signals, if they are relatively constant with time. With the HMSLD operating at maximum sensitivity, a
probe, such as a hypodermic needle with a very slow flow of helium, is passed along any suspected leak locations, starting at the top of the system, and avoiding drafts. Whenever a leak signal is first heard, and the presence of a leak is quite apparent, the probe is removed, allowing the signal to decay; checking is resumed, using the probe with no significant helium flow, to pinpoint the exact location of the leak. Ideally, the leak should be fixed before the probe is continued, but in practice the leak is often plugged with a piece of vacuum wax (sometimes making the subsequent repair more difficult), and the probe is completed before any repair is attempted. One option, already noted, is to blanket the leak site with helium to obtain a quantitative
measure of its size, and then calculate whether this is the entire problem. This is not always the preferred procedure, because a large slug of helium can lead to a lingering background in the detector, precluding a check for further leaks at maximum detector sensitivity.
A number of points need to be made with regard to the detection of leaks:
1. Bellows should be flexed while covered with helium.
2. Leaks in water lines are often difficult to locate. If the water is drained, evaporative cooling may cause
ice to plug a leak, and helium will permeate throughthe plug only slowly. Furthermore, the evaporating
water may leave mineral deposits that plug the hole.A flow of warm gas through the line, overnight, will
often open up the leak and allow helium leak detection.Where the water lines are internal to the system,
the chamber must be opened so that the entire line is accessible for a normal leak check. However,
once the lines can be viewed, the location of the leak is often signaled by the presence of discoloration.
3. Do not leave a helium probe near an O-ring for more than a few seconds; if too much helium goes into
solution in the elastomer, the delayed permeation that develops will cause a slow flow of helium into the system, giving a background signal which will make further leak detection more difficult.
4. A system with a high background of hydrogen may produce a false signal in the HMSLD because of
inadequate resolution of the helium and hydrogen peaks. A system that is used for the hydrogen isotopes
deuterium or tritium will also give a false signal because of the presence of D2 or HT, both of which have their major peaks at mass
4. In such systems an alternate probe gas such as argon must be used, together with a mass spectrometer which can be tuned to the mass 40 peak.
Finally, if a leak is found in a system, it is wise to fix it properly the first time lest it come back to haunt you!
Before assuming that a vacuum system leaks, it is useful to consider if any other problem is present. The most
important tool in such a consideration is a properly maintained log book of the operation of the system. This is particularly the case if several people or groups use a single system. If key check points in system operation are recorded weekly, or even monthly, then the task of detecting a slow change in performance is far easier.
Leaks develop in cracked braze joints, or in torchbrazed joints once the flux has finally been removed.
Demountable joints leak if the sealing surfaces are badly scratched, or if a gasket has been scuffed, by allowing
the flange to rotate relative to the gasket as it is compressed.Cold flow of Teflon or other gaskets slowly reduces the compression and leaks develop. These are the easy leaks to detect, since the leak path is from the atmosphere into the vacuum chamber, and a trace gas can be used for detection.
A second class of leaks arise from faulty construction techniques; they are known as virtual leaks. In all of these,a volume or void on the inside of a vacuum system communicates to that system only through a small leak path.Every time the system is vented to the atmosphere, the void fills with venting gas, then in the pumpdown this gas flows back into the chamber with a slowly decreasing throughput, as the pressure in the void falls. Pressures and pumping speeds developed by a steady throughput of gas (Q) through a vacuum chamber, conductance (C) and pump.the system pumpdown.
A simple example of such a void is a screw placed in a blind tapped hole. A space always remains at the bottom of the hole and the void is filled by gas flowing along the threads of the screw. The simplest solution is a screw with a vent hole through the body,providing rapid pumpout. Other examples include a double O-ring in which the inside O-ring is defective, and a double weld on the system wall with a defective inner weld. A mass spectrometer is required to confirm that a virtual leak is present. The pressure is recorded during a
routine exhaust, and the residual gas composition is determined as the pressure is approaching equilibrium. The system is again vented using the same procedure as in the preceding vent, but the vent uses a gas that is not significant in the residual gas composition; the gas used should preferably be nonadsorbing, such as a rare gas.After a typical time at atmospheric pressure, the system is again pumped down. If gas analysis now shows significant vent gas in the residual gas composition, then a virtual leak is probably present, and one can only look for the culprit in faulty construction.
Leaks most often fall in the range of 10¯4 to 10¯6 torr-L/s. The traditional leak rate is expressed in atmospheric cubic centimeters per second, which is 1.3 torr-L/s. A variety of leak detectors are available with practical sensitivities varying from around 1X10¯3 to 2 X 10¯11 torr-L/s.The simplest leak detection procedure is to slightly pressurize the system and apply a detergent solution, similar to that used by children to make soap bubbles, to the outside of the system. With a leak of ~1X10¯3 torr-L/s, bubbles should be detectable in a few seconds. Although the lower limit of detection is at least one decade lower than this figure, successful use at this level demands considerable patience.
A similar inside-out method of detection is to use the kind of halogen leak detector commonly available for
refrigeration work. The vacuum system is partially backfilled with a freon and the outside is examined using a sniffer hose connected to the detector. Leaks the order of 1X10¯5 torr-L/s can be detected. It is important to avoid any significant drafts during the test, and the response time can be many seconds, so the sniffer must be moved quite slowly over the suspect area of the system. A far more sensitive instrument for this procedure is a dedicated helium leak detector (see below) with a sniffer hose testing a system partially back-filled with helium.A pressure gauge on the vacuum system can be used in the search for leaks. The most productive approach applies if the system can be segmented by isolation valves. By appropriate manipulation, the section of the system containing the leak can be identified. A second technique is not so straightforward, especially in a nonbaked system. It relies on the response of ion or thermal conductivity gauges differing from gas to gas. For example, if the flow of gas through a leak is changed from air to helium by covering the suspected area with helium, then the reading of an ionization gauge will change, since the helium sensitivity is only ~16% of that for air. Unfortunately, the flow of helium through the leak is likely to be ~2.7 times that for air, assuming a molecular flow leak, which partially offsets the change in gauge sensitivity. A much greater problem is
that the search for a leak is often started just after exposure to the atmosphere and pumpdown. Consequently outgassing is an ever-changing factor, decreasing with time.
Thus, one must detect a relatively small decrease in a gauge reading, due to the leak, against a decreasing background pressure. This is not a simple process; the odds are greatly improved if the system has been baked out, so that outgassing is a much smaller contributor to the system pressure.
A far more productive approach is possible if a mass spectrometer is available on the system. The spectrometer is tuned to the helium-4 peak, and a small helium probe is moved around the system, taking the precautions described later in this section. The maximum sensitivity is obtained if the pumping speed of the system can be reduced by partially closing the main pumping valve to increase the pressure, but no higher than the mid-10¯5 torr range, so that the full mass spectrometer resolution is maintained. Leaks in the 1X 10¯8 torr-L/s range should be readily detected.
The preferred method of leak detection uses a standalone helium mass spectrometer leak detector (HMSLD).Such instruments are readily available with detection limitsof 2 X10¯10 torr-L/s or better. They can be routinely calibrated so the absolute size of a leak can be determined. In many machines this calibration is automatically performed at regular intervals. Given this, and the effective pumping speed, one can find, using Equation 1,whether the leak detected is the source of the observed deterioration in the system base pressure.
In an HMSLD, a small mass spectrometer tuned to detect helium is connected to a dedicated pumping system,
usually a diffusion or turbomolecular pump. The system or device to be checked is connected to a separately pumped inlet system, and once a satisfactory pressure is achieved, the inlet system is connected directly to the detector and the inlet pump is valved off. In this mode, all of the gas from the test object passes directly to the helium leak detector. The test object is then probed with helium, and if a leak is detected, and is covered entirely with a helium blanket, the reading of the detector will provide an absolute indication of the leak size. In this detection mode, the pressure in the leak detector module cannot exceed ~10¯4 torr, which places a limit on the gas influx from the test object. If that influx exceeds some critical value, the flow of gas to the helium mass spectrometer must be restricted, and the sensitivity for detection will be reduced. This mode, of leak detection is not suitable for dirty systems, since the gas flows from the test object directly to the detector, although some protection is usually provided by interposing a liquid nitrogen cold trap.
An alternative technique using the HMSLD is the socalled counterflow mode. In this, the mass spectrometer
tube is pumped by a diffusion or turbomolecular pump which is designed to be an ineffective pump for helium
(and for hydrogen), while still operating at normal efficiency for all higher-molecular-weight gases. The gas
from the object under test is fed to the roughing line of the mass spectrometer high-vacuum pump, where a higher pressure can be tolerated (on the order of 0.5 torr).
Contaminant gases, such as hydrocarbons, as well as air, cannot reach the spectrometer tube. The sensitivity of an HMSLD in this mode is reduced about an order of magnitude from the conventional mode, but it provides an ideal method of examining quite dirty items, such as metal drums or devices with a high outgassing load.
The procedures for helium leak detection are relatively simple. The HMSLD is connected to the test object for
maximum possible pumping speed. The time constant for the buildup of a leak signal is proportional to V/S, where V is the volume of the test system and S the effective pumping speed. A small time constant allows the helium probe to be moved more rapidly over the system.
For very large systems, pumped by either a turbomolecular or diffusion pump, the response time can be
improved by connecting the HMSLD to the foreline of the system, so the response is governed by the system
pump rather than the relatively small pump of the HMSLD. With pumping systems that use a capture-type
pump, this procedure cannot be used, so a long time constant is inevitable. In such cases, use of an HMSLD and helium sniffer to probe the outside of the system, after partially venting to helium, may be a better approach.Further, a normal helium leak check is not possible with an operating cryopump; the limited capacity for pumping helium can result in the pump serving as a low-level source of helium, confounding the test.
Rubber tubing must be avoided in the connection between system and HMSLD, since helium from a large
leak will quickly permeate into the rubber and thereafter emit a steadily declining flow of helium, thus preventing use of the most sensitive detection scale. Modern leak detectors can offset such background signals, if they are relatively constant with time. With the HMSLD operating at maximum sensitivity, a
probe, such as a hypodermic needle with a very slow flow of helium, is passed along any suspected leak locations, starting at the top of the system, and avoiding drafts. Whenever a leak signal is first heard, and the presence of a leak is quite apparent, the probe is removed, allowing the signal to decay; checking is resumed, using the probe with no significant helium flow, to pinpoint the exact location of the leak. Ideally, the leak should be fixed before the probe is continued, but in practice the leak is often plugged with a piece of vacuum wax (sometimes making the subsequent repair more difficult), and the probe is completed before any repair is attempted. One option, already noted, is to blanket the leak site with helium to obtain a quantitative
measure of its size, and then calculate whether this is the entire problem. This is not always the preferred procedure, because a large slug of helium can lead to a lingering background in the detector, precluding a check for further leaks at maximum detector sensitivity.
A number of points need to be made with regard to the detection of leaks:
1. Bellows should be flexed while covered with helium.
2. Leaks in water lines are often difficult to locate. If the water is drained, evaporative cooling may cause
ice to plug a leak, and helium will permeate throughthe plug only slowly. Furthermore, the evaporating
water may leave mineral deposits that plug the hole.A flow of warm gas through the line, overnight, will
often open up the leak and allow helium leak detection.Where the water lines are internal to the system,
the chamber must be opened so that the entire line is accessible for a normal leak check. However,
once the lines can be viewed, the location of the leak is often signaled by the presence of discoloration.
3. Do not leave a helium probe near an O-ring for more than a few seconds; if too much helium goes into
solution in the elastomer, the delayed permeation that develops will cause a slow flow of helium into the system, giving a background signal which will make further leak detection more difficult.
4. A system with a high background of hydrogen may produce a false signal in the HMSLD because of
inadequate resolution of the helium and hydrogen peaks. A system that is used for the hydrogen isotopes
deuterium or tritium will also give a false signal because of the presence of D2 or HT, both of which have their major peaks at mass
4. In such systems an alternate probe gas such as argon must be used, together with a mass spectrometer which can be tuned to the mass 40 peak.
Finally, if a leak is found in a system, it is wise to fix it properly the first time lest it come back to haunt you!
Posts
8 comments:
Very informative blog. I was searching for something like this. your blog helped me a lot. Thank you so much for sharing.
leak detection equipment
I need a good vacuum cleaner?
Pump repairs
This is easier and surely gives comfort to internet users. Thanks for sharing. Post like this offers great benefit. Thank you!
www.bestvacuumfortilefloors.com
It's a nice article! There are many types of leak detection, such as thermal leak detection, gas leak detection, and etc.
regards,
airless spray
Very informative site, i must bookmark it, keep posting interesting articles.
Buy 5+ and Get 15% Off cheap parts
Thank you for this excellent post. I have just started to read this article carefully. Keep up posting these type of post contuniously.
Leak Detection Switch
Hi..
I appreciate your informative post and thanks for sharing.
Vacuum Leak Tester
Post a Comment