ATLAS Internal Note MUON-N0-104 Eect of Vacuum Grease on Helium Leak Rate Through O-Rings T. Ferbel and A. Manz Max Planck Institute, Munich January 10, 1995 Abstract: Because of the concern about long-term silicon contamination of MDTs, we have tested the extent to which vacuum grease on O-rings is essential for producing gas tight muon chambers. We nd that, for trial periods of up to six hours, a small amount of grease decreases the helium leak rate by only about a factor of 2{3. Thus, it appears that it may be acceptable to construct MDTs without vacuum grease on O-rings of endplugs. We also comment on the matter of checking MDTs for gas tightness. The Nature of the Studies: The \monitored' drift tubes (MDTs) used for tracking trajectories of muons through the toroids of the ATLAS detector are to be operated at a gas pressure of three atmospheres [1]. The fundamental element of the system is an aluminum tube that contains a centered anode wire, which is held in place (under tension) by endplugs lodged in at the ends of the tubes, and has locators positioned in the middle for the long tubes. In the current design, each endplug contains a groove for crimping the tube into proper position, and a channel that contains an O-ring that provides the gas seal. The operating gas, the high voltage (HV) for the anode, and the grounding of the tube, are provided through plastic feedthroughs embedded in the aluminum endplugs. Using vacuum grease on O-rings can certainly provide a better gas seal for MDTs, and can ease the insertion of endplugs into their tubes. Nevertheless, silicon is known to accelerate deterioration of the performance of wire chambers, especially in high-radiation environments, as well as produce an increase in outgassing time. In addition, the long-term deterioration of silicon grease also poses some concern. We have consequently attempted to determine whether foregoing the use of vacuum grease can be tolerated in the construction of MDTs. To this end, we constructed tubes with standard endplugs, but without their normal feedthroughs. The specic endplug design is sketched in Fig. 1. We compared the leak rate through the O-rings, using both greased and dry O-rings, as the tubes were pressurized with helium gas. (The 30 cm aluminum tubes, provided by the Menziken Co., were crimped to endplugs using a mechanical crimping technique developed previously at the Institute.) 1
Figure 1: Endplug design and conguration used in these studies. 2
The assembled tubes were checked for leaks using a helium snier, and then inserted into a steel-bellows vacuum chamber. (None of the tubes had initial leak rates that exceeded 10 9 mbar-l/sec.) The more quantitative tests commenced with the insides of the tubes open to the atmosphere through the small inlets shown in Fig. 1. The chamber and the outside walls of the tubes were rst pumped down with a Leybold UL500 helium leak detector. Typically, after about 10 minutes of pumping, the apparent leak rate stabilized at a value of < 10 9 mbar-l/sec. Subsequently, after another 5 minutes of pumping, 1 bar (overpressure) of helium gas was admitted into the tube, and the leak rate through the O-rings was then mapped as a fuction of time. After 30 minutes, the helium overpressure was raised to 6 bars, and readings were continued for about an additional six hours. To check the eect of introducing normal feedthroughs into the endplugs, we outtted the endplugs of one tube with feedthroughs for HV and gas inlets, but not one for the anode wire. In fact, the gas outlet on the innermost endplug was closed with solder, and no holes were drilled for the anode wire in either endplug. Helium gas was admitted into this tube through the gas inlet on the outer endplug. In addition, to gauge the response of the UL500 to an unacceptably large leak, we prepared another tube (without the above feedthroughs) that was scratched in the region of the O-ring. Prior to the mechanical crimping of the tube about its endlpugs, the tube was checked for leakage using a helium snier, and found to have a leak of approximately 10 6 mbar-l/sec. This somewhat damaged tube was then put through a test procedure similar to that used for the standard tubes. Finally, we also investigated the response of the UL500 when, in the presence of a tube with feedthroughs (same as the ones described at the beginning of this paragraph), we introduced an external leak of 5 10 8 mbar-l/sec directly into the cylidrical vacuum chamber. These last three tests were all performed using dry O-rings. Results and Conclusions: Figure 2 displays the time dependence of the observed leak rate for seven tubes (including those with feedthroughs) that did not have vacuum grease on the O-rings, and Fig. 3 shows the data for six tubes that used grease. The start time is dened by the time of pressurization of each tube with 1 bar (gauge) of helium. In general, about 10 minutes after pressurization, the leak rate for both dry and greased O-rings started exceeding 10 9 mbar-l/sec. There were small but signicant changes in the slope of the leak rate with time that we attribute to exing and relaxation of the O-rings under pressure. The greatest changes were observed at the transition to 6 bars. At this junction, the leak rate often fell rather dramatically (more so for the case of greased O-rings), but then subsequently continued its initial rise with what appeared to be a displacementof 10{15 minutes in the time origin. An exception to this trend occurred for one of the tubes with greased O-rings, where, after the application of 6 bars of helium, 3
Figure 2: The helium leak rate as a function of time for seven tubes with dry O-rings. The two sets of data corresponding to measurements using tubes with feedthroughs, but without holes for the anode wire in their endplugs, are included in the gure. The tube with an inital leak rate of 5 10 8 mbar-l/sec had an external leak of that order introduced directly into the vacuum chamber (see text). An expanded time scale is given in the lower part of the gure. One bar of helium was admitted at t=0, and six bars at t=30 minutes. 4
Figure 3: The helium leak rate as a function of time for six tubes with greased O-rings. An expanded time scale is provided in the lower part of the gure. One bar of helium was admitted at t=0, and six bars at t=30 minutes. 5
the leak rate dropped below 10 9 mbar-l/sec, and remained at that level for about two hours. It rose subsequently to 10 8 mbar-l/sec over the following two hours, and then jumped suddenly to > 10 6 mbar-l/sec, and remained (although unstable) in the range of (1:4 5:8) 10 6 mbar-l/sec. Such behavior may be attributable to our having used helium gas in the tests. Helium is known to penetrate and to be easily absorbed in vacuum grease, and therefore small adjustments in the position of the O-ring could aect the apparent leak rate. We do not expect such eects with normal operating gas. For all tubes, the leak rate after six hours ranged between 2:2 10 6 and 5:0 10 6 mbar-l/sec, and between 4:8 10 6 and 8:4 10 6 mbar-l/sec, respectively, for the greased and dry O-rings. As we will see shortly, these rates reect primarily the diusion of helium through the O-rings, but must also reect any leakage around them. Although the statistics are poor, there is clear evidence that the greased O-rings provide better gas tightness. Nevertheless, the observed dierence using helium gas is not dramatic, and may prove to be negligible using the Ar/N 2 =CH 4 gas mixture currently favored by ATLAS [2]. The results for the two tubes that contained feedthroughs were given in Fig. 2. We see that, without an external leak, the response is comparable to that for tubes that had no feedthroughs. The nal leak rate is on the high side of tubes without grease, but only a factor of four worse than found for the most gas-tight tubes (with greased O-rings) in Fig. 3. The response for the case of the small external leak appears to be just a superposition of a constant level (achieved prior to pressurization to 1 bar) and the typical increase of leak rate with time found in Fig. 2. The results for the damaged tube (not shown) were totally dierent than for the other tubes. The leak rate grew rapidly with increasing helium pressure. At the start, it took longer (over an hour) to stabilize the readings of the UL500 to < 10 9 mbar-l/sec, and the leak rate went up immediately to about 4 10 6 mbar-l/sec as the helium overpressure was set to just 0.05 bar. The leak rate then remained stable at (3 4) 10 6 mbar-l/sec for 30 minutes, and rose by an order of magnitude when the helium overpressure was raised at that time to 0.5 bar. Over the next hour, the leak rate fell by almost a factor of two. Then, upon another increase in helium pressure to an overpressure of 1 bar, the leak rate increased by another order of magnitude, and remained relatively stable at about 2 10 4 mbar-l/sec. In summary, the results from these studies indicate that the dierence between using greased O-rings in MDTs rather than leaving them dry appears to have only a small eect on the helium leak rate of a tube after several hours of pressurization. Consequently, we expect no great gain from using vacuum grease on the O-rings. Also, because of their dierent response to helium, it should be relatively straightforward to identify damaged MDTs [3]. That is, since the apparent helium leak rate on tubes that are not gas tight increases immediately with the slightest application of helium gas, a simple test using a helium snier 6
should suce to determine if a tube is gas tight. We plan shortly to examine these same issues, and the matter of back-diusion of gas contaminants into MDTs, using mass spectroscopy with the nominal operating gas. Acknowledgements: We thank Reinhard Hofmann for preparing the tubes and performing the extensive measurements, Karlheinz Ackermann for many helpful discusions, Horst Stieg and Winfried Tribanek for their continued interest in this work, and Sabine Naujoks for help with drawings of the equipment. One of us (T.F.) also wishes to acknowledge the support of the University of Rochester and the Alexander von Humboldt Foundation, and to thank Friedrich Dydak and the members of the Max Planck Institute for their hospitality. References: 1. ATLAS Technical Proposal, CERN/LHCC/94-43 (1994). 2. Private communication from Michael Treichel. 3. Carl Bromberg has often emphasized that there is nothing subtle about the functioning of such tubes: they either work beautifully, or not at all. 7