4.2.1 SCT installation and integration in ATLAS
The construction of the Semiconductor Tracker has been a worldwide undertaking with the participation of 31 institutes from 13 different countries [67]. After the construction of the modules in several locations, the macro-assembly took place in
2005 at Oxford, UK for the barrel, Nikhef, Holland for end-cap A and Liverpool, UK for end-cap C. The SCT was later integrated and tested, together with the TRT, in the SR1 building at Point 1, CERN. The SCT and TRT barrels were simultaneously installed in the ATLAS cavern in August 2006 and the SCT barrel was fully com-missioned by May 2007. End-cap A and C were inserted into the cavern during May and June 2007 and were commissioned during January and February 2008. After the complete system was fully installed, the SCT joined the ATLAS combined Milestone cosmic ray run (M6) in March 2008, which included most of the ATLAS sub-detectors as well as all trigger levels.
The SCT barrel was off due to safety reasons when the first proton beam was circulated in the LHC on September 10th 2008. The main concern was that particles from the unfocused beam might traverse the barrel modules at a small angle and damage the modules. The end-caps, however, were operating at reduced bias voltage (20 V) and high threshold and could detect particle events from beam splashes created in the collimators 140 m upstream from ATLAS. As the first beam splashes were detected by ATLAS, the SCT crew shouted with excitement as the SCT end-caps lit up like light bulbs on the Atlantis event display in the ATLAS control room. The event display, together with some cheerful SCT collaborators, can be seen in Figure 4.4.
Figure 4.4: Picture of part of the SCT crew in the ATLAS control room, September 10th, 2008, aka. “First Beam Day”. The Atlantis event display in the background shows one of the first beam splash events in ATLAS. The SCT displays a large amount of recorded hits in the end-caps.
4.2.2 SCT performance results with cosmic ray data
Since the SCT was fully installed about 18 months before the first collisions, it was properly tested and analyzed through the combined cosmic data-taking as well as stand-alone and calibration runs. During the time of commissioning, the SCT had about 99.3% of its 4088 modules fully calibrated and operational. This included the 13 modules excluded due to the non-operational cooling loop, as was discussed in section 3.3.1. The SCT performance results using cosmic ray and single beam data can be found in [68], while more updated collision performance results are available in [69].
Hit efficiency
A hit is defined as a cluster with signal on at least one strip. The hit efficiency for a given threshold is measured through expected hits; tracks passing through an active region of silicon should result in a hit being read out. The SCT barrel hit efficiency was accurately measured with the cosmic data and the requirements for the tracks analyzed include having 10 SCT hits, 30 TRT hits and χ2/DoF < 2. The silicon efficiency of the barrel modules for a threshold of 1 f C was found to be 99.75%
with the solenoid field on. The hit efficiency for the end-caps was measured to be between 99.0% and 99.5% with cosmic ray data, with the larger uncertainties due to lack of statistics as well as not being accurately timed in with respect to the trigger.
More precise barrel and end-cap hit efficiency results have now been completed with collision data (see [69]). The hit efficiency has been measured to be 99.89% in the barrel, 99.75% in end-cap A and 99.76% in end-cap C for combined tracks.
Noise occupancy
A hit is defined as noise if it is not part of a space point, where a space point is a hit on each side of the module. Noise occupancy can be measured with cosmic data runs using random triggers or during stand-alone calibrations. For the SCT barrel and end-caps, the noise occupancy measured per channel at the operating voltage of 150 V is significantly less than the design specification of 5·10−4 (see Figure 4.5).
A mean of 1.5·10−5 for inner barrel, 2.2·10−5 for outer barrel, 3.6·10−5 for outer end-cap and 2.5·10−5 for middle end-cap modules was measured for a threshold of 1 f C.
Alignment and timing
The alignment of the Inner Detector is accomplished by track-based offline alignment algorithms minimizing a χ2 function defined from the track residuals. A residual is defined as the measured hit position minus the expected hit position from the
Figure 4.5: Noise occupancy measured in the barrel and end-cap modules in the calibration period during September 2009. A mean of 1.5·10−5 for the inner barrel, 2.2·10−5 for the outer barrel, 3.6·10−5 for the outer end-cap and 2.5·10−5 for the middle end-cap modules was measured, which is significantly less than the design specification of 5·10−4 [68].
track extrapolation. Three different levels of alignment are used for the SCT, from barrel and end-cap structures to the individual module level. Figure 4.6 shows the unbiased residual distribution at module level integrated over all hits-on-tracks in the SCT barrel for the nominal geometry and the preliminary aligned geometry for 2008 cosmic ray data. A single Gaussian fit was performed and the resolution of the fit was compared with the nominal geometry as well as the Monte Carlo simulation of a perfectly aligned detector. The results indicated satisfactory alignment for the barrel. The final accuracy will later be achieved with careful studies of LHC collision data. Due to the geometry of the cosmic tracks, the alignment of the end-caps was not possible with cosmic data and was first achieved with collision data.
Lorentz angle studies
Analyzing the Lorentz angle provides a deeper understanding of the detector align-ment and spatial resolution and is required by the data simulation to allow an accu-rate comparison with data. Details of the SCT Lorentz angle analysis from the 2008 cosmic ray data can be found in [70].
The Lorentz angle originates from the Hall effect. In the presence of electric and magnetic fields, the charge carriers created in the silicon sensor drift along a direction slightly shifted in comparison to that expected without magnetic field. Depending
x residual [mm]
-0.4 -0.3 -0.2 -0.1 -0 0.1 0.2 0.3 0.4
number of hits on tracks
0
Figure 4.6: Unbiased residual distribution integrated over all hits-on-tracks in the SCT barrel for the nominal geometry and the preliminary aligned geometry for 2008 cosmic ray data [68].
on the incidence angle of the particles, the charge is spread over several strips. Since this spread is minimal for an incidence angle equal to the Lorentz angle (θL), it can be extracted from the minimum of the average strip cluster-size as a function of the incidence angle of the track to the module surface accounting for the binary electronics for a given threshold. The angle has been measured with cosmic ray data in the SCT barrel to be θL= 3.93±0.03(stat)±0.10(syst), which is in good agreement with the predicted value of θL= 3.69±0.26(syst) (see Figure 4.7).
4.2.3 SCT with collision data
The SCT is currently fully operational and performing very well. A persistent issue is, as mentioned in section 3.3, the dying optical links (TXs) but this fortunately does not have any impact on the offline data. The cooling loop affecting 13 modules is still not operational at the time of writing, but there are otherwise no significant issues affecting the SCT performance. The intrinsic efficiency of the modules is very high; approximately 99.8% overall and 99.0% of the SCT configuration was fully functional as of May 2010 (including the disabled cooling loop) [69]. Table 4.2 shows the configuration of the disabled modules, chips and masked strips in the SCT as of that date. At the time of writing (August 2011) only 29 SCT modules are disabled in the entire SCT (13 modules due to the leaking cooling loop and 16 modules due to HV and LV errors), while there are 81 dead TX channels using the redundancy readout path.
Figure 4.7: Measurement of mean cluster size as a function of incidence angle with 2008 cosmic ray data. The measurements are shown with and without magnetic field together with the Monte Carlo predictions. The value of the Lorentz angle is extracted from the position of the minimum of the field ON data [70].