Changing the TRT gas

27 April 2010

Schematics of the TRT gas system, highlighting the membrane that blew and the point of entry for the air


While ATLAS mostly sailed smoothly through the first collisions – a remarkably firm understanding of the detectors revealed the first W particles within a week – it wasn’t all easy. One of the pumps that help maintain the carbon dioxide levels in the Transition Radiation Tracker gas blew a membrane and began pumping air into the TRT at 5:20 Wednesday morning, March 24th, adding both oxygen and nitrogen to the gas mixture.

Oxygen is part of the TRT active gas, but the 120 litres of nitrogen which were pumped in the system significantly changed the gas composition. Meanwhile, the system that controlled carbon dioxide had been adding more of it to raise carbon dioxide levels back to where they should be.

By 6:00, the system was stopped. Between then and 17:00, Friday, March 26th, the TRT team replaced the faulty parts and removed the extra carbon dioxide and oxygen. Ideally, the four cubic metres of gas in the TRT is 70 per cent xenon, 27 per cent carbon dioxide, and three per cent oxygen, although nitrogen typically takes up 0.5 per cent of the mix. After the mixture stabilised, the nitrogen concentration was about 2.7 per cent. It slowly decreased to 2.3 per cent, which left the xenon at just under 68 per cent.

The excess of nitrogen affects the drift velocity of electrons freed under the passage of charged particles in the TRT active gas mixture. Anatoli Romaniouk, run coordinator for the TRT, says that at 2.3 per cent nitrogen, the total electron drift-time in the straw decreased by four nanoseconds. He and TRT gas expert Serguei Konovalov were responsible for the gas replacement operation.

Still, for late March and early April, a change of the drift time due to the continuous change of the nitrogen concentration posed little difficulty. Calibrations of the drift distance versus drift time, the so-called R-T relation, were done every 24 hours to take the gradual decrease of nitrogen into account.

“In a sense, this data can be used for particle identification with some caution, which is adequate to the initial setups we are doing now,” says Anatoli. But to fully explore the particle identification capacity of the TRT, the team uses the full information available: Transition Radiation hits, leading edge and trailing edge of the straw signal.

In order to implement the higher-precision algorithms, the TRT group will need a few months of data from a period of stable operation. Unfortunately, this makes the gradual decrease in nitrogen from 2.3 to 0.5 per cent a problem.

Normally, it takes four or five days to replace the gas. Done right away, the schedule would have conflicted with the First Physics Day with 7 TeV collisions. The TRT had more to gain by running with high nitrogen than shutting down and possibly missing these momentous collisions.

Nevertheless, the gas needed to be changed, so the TRT group looked for their opportunity and devised a way to shorten the procedure to a day and a half. “We developed a scenario which required very unusual operation of the gas system, changing the operation modes manually and forcing operation of some subsystems to the limit,” says Anatoli, noting that these are performance limits – not safety.

The procedure started at 15:00 on Wednesday, 7 April. They turned down the high voltage but left most systems running – and even calibrated some of the electronics and made some TRT DAQ software updates during this time.

The idea was to send a bubble of carbon dioxide through the gas system, pushing the old gas through the TRT detector to a specially set-up recuperation system. After the bubble formation, they started injecting xenon. This was running in parallel with the xenon removal process from the opposite side of the detector to recover the costly xenon from the old gas.

Because the gas flow was limited, the process of replacing the gas mixture inside the detector with pure carbon dioxide was not as efficient as hoped. The team had to take gas diffusion processes in the detector into account to estimate when the removal process should stop. In the end, the process continued until the carbon dioxide concentration at the detector output hit 55 per cent, as measured by a gas chromatograph.

Then began the stabilisation phase, removing carbon dioxide and adding more xenon and oxygen back into the mixture – a standard procedure, except that time constraints required a high xenon injection rate. That created another problem – the xenon started clogging at the injection point due to condensation from gas expansion.

Continuous heating of the gas supply components was required until the xenon concentration finally reached 65 per cent.  After about thirty hours, the gas exchange was completed, and the nitrogen concentration was down to an acceptable level of 0.66 per cent. 

Once the proportions of xenon, carbon dioxide, and oxygen were balanced and stable, the high voltage was reapplied under careful control. By 21:00 April 8th, the TRT was ready to take data again.

Fifty litres of xenon were lost in the process, but the rest was collected in the recuperation system. The system is able to remove the nitrogen and make the gas useable again. Now, half of the gas has been treated and the remaining concentration of nitrogen is 0.04%, though Anatoli cautions that the purity of the treated gas must be verified. With or without this gas, they have enough to keep the TRT running even in the event of another incident.

“Everything is stable now, and hopefully for a long time,” says Anatoli.

 

Katie McAlpine

ATLAS e-News