I thought I’d give you a sense of what it takes to put together a detector like ATLAS, e.g., how much time, how many people, etc. For an overview of the ATLAS detector, please look at the ATLAS webpage and Monica’s post. Since ATLAS is huge, I will focus on just one sub-system, the Barrel Transition Radiation Tracker (TRT), which was built in the US. Its main purpose is to provide hits so that we can map the trajectory of charged particles and improve the measurement of their momentum (see Seth’s post on tracking). It can also discriminate between electrons and pions.

Figure 1: End view of the barrel TRT
You can see the barrel TRT in Fig. 1 (this is an end view where you can see the electronics and cables); more information is ATLAS website. This detector is divided into 96 modules and extends from about 50 cm to 108 cm in radius, contains about 52,000 individual wires (about 2 m long) each of which is strung inside a specially built plastic straw. As the name suggests, the barrel TRT is in the central part of ATLAS. Two other parts of the inner detector, the Pixel and the Silicon tracker, reside inside the barrel TRT 1 (that unit was being inserted into the assembly at the time this picture was taken – you can see it at the other end of the barrel).
To set the scale, the barrel TRT occupies about the half the volume of the inner detector in the barrel, which ends at a radius of about 1 m. The calorimeters, solenoid magnet and cryostat come after and go out to about 5 m in radius, and the muon system goes out to about 10 m in radius. The barrel TRT probably represents a few percent of the total cost of building the ATLAS detector, and is arguably the most sophisticated of a class of detectors called “drift chambers”; one of its selling points was that it was a low-cost way of tracking charged particles. It also has fewer electronic channels to read out (each wire is read out at both ends); in comparison, the Pixel and the Silicon tracker detectors have about 80 million electronic channels to read out.
I spoke with my colleague at Indiana University, Harold Ogren, who was one of the lead physicists on this project, as well as being the manager of the construction effort in the US. Harold and one of his colleagues originally built a similar straw tube device for an experiment that ran at the Stanford Linear Accelerator in the 1980’s. When the Superconducting Super Collider was proposed in the US, a straw tube tracker design was accepted for one of the two main experiments; when it was cancelled in 1993, he and his colleagues moved onto ATLAS, where they joined forces with the groups already working on a straw tube design.
They started building a prototype for ATLAS around 1994. Some of the groups who were on SSC joined this effort, and they had a working chamber by about 1999 that was then put in a test beam at CERN. Actual construction of the 96 modules began after the successful beam test, and it took them another 3 years to finish; each of the 52,000 wires had to be individually strung. The construction effort involved about 6-7 physicists and about 40 technicians, engineers, graduate students from Indiana, Duke and Hampton Universities and the University of Pennsylvania The electronics to read out the detector was also designed by them.
Since it was a modular design, they could ship individual modules to CERN as they were being completed, where they were put through extensive tests, e.g., each wire was scanned along every inch of its length with X-rays to check for uniformity of performance. A few wires were bad and had to be disconnected; since the bad wires are randomly distributed they don’t affect performance. These tests took another 2-3 years. All in all, the detector was ready sometime around 2006. The picture you saw above was when it was being readied to be installed in ATLAS.

Fig 2: Cosmic shower in the TRT
The barrel TRT has been running successfully and collecting cosmic ray data. In Fig 2, most likely a cosmic shower hit the TRT, the kind described in Regina’s post. You are looking at an end view of the hit wires. Each blue dot represents a single wire being hit; we can locate the position of a track within a straw with an accuracy of 0.15 mm (human hair has a thickness of about 0.1 mm). You can see curved tracks; they are curving because the magnetic field was on. You will also see that one sector, at about 8 o’clock, was (temporarily) turned off. Isolated hits are due to electronic noise; operating parameters are set so that these wires register the presence of nearby charged particles with a very high efficiency, but this also leads to 1-2% of the wires “firing” randomly; our reconstruction algorithms can easily ignore them. The tracks that you see here are then matched to hits in the Pixel and Silicon Tracker detectors to get a complete trajectory.
Now for real data!
— Vivek Jain, Indiana University
1 There is also the endcap TRT, which is on the two ends of the ATLAS detector, but that was built by other groups; it uses the same design as the one in the barrel.