CMS Tracker Detector Performance Results

Contact: Janos Karancsi, Janos.Karancsi@cernNOSPAMPLEASE.ch and Viktor Veszpremi, Viktor.Veszpremi@cernNOSPAMPLEASE.ch

Presenting for approval on Apr 12, 2013

Definition of efficiency: N_valid + N_missing_with_cluster / (N_valid + N_missing), for Rechits with no other Rechits within 5 mm. N_missing_with_cluster : Missing RecHits with a cluster within 500 um.

Event selection: \x{2265}1 vertex, where |z|<25 cm, |\x{03c1}|<2.0 cm, N_dof >4. track selection: General track seeded from the Pixels, valid hits always required on "other" layers in order to minimize bias from seeding. Track p_T>1.0 GeV, N_strip_hits >10, track consistent with vertex (cut on impact parameters d0, dz - layer dependent), track separation (RecHits separated by 5mm)

The innermost modules of Disk 1 were excluded.

Hit Finding Efficiency vs Layers and Disks - 2012

Average Layer/Disk Efficiencies: SEU candidates and DAQ errors are excluded.

Average Hit Fining Efficiency vs layers and disks in 2012 Data

Dynamic Inefficiencies:

The data used in these plots only includes the highest intensity fills of 2012 (more than 1300 colliding bunches).

Instantaneous Luminosity	Level 1 Trigger Rate
Efficiency vs Instantaneous Luminosity for all Layers and Disks	Efficiency vs Level 1 Trigger Rate for all Layers and Disks

2012 runs: The plots show the variations of efficiencies throughout 2012 runs (Note: the scale is not linear in time).

Barrel Pixel	Forward Pixel
Efficiency for each barrel layer vs 2012 runs (time scale is not linear)	Efficiency for each forward disk vs 2012 runs (time scale is not linear)

Typical 2012 Fills:

Fill #2736	Fill #3239
Efficiency vs Instantaneous Luminosity in Fill #2736 (16 June 2012)	Efficiency vs Instantaneous Luminosity in Fill #3236 (28 October 2012)

All High Voltage Bias Scans - Up to 2013

Layer 1 Scans:

Efficiency	Normalized Charge MPV
Efficiency vs Bias Voltage for all Layer 1 Scans up to 2013. Points were fitted with a turnon curve.	MPV of Normalized Charge vs Bias Voltage for all Layer 1 Scans up to 2013. Points were fitted with a modified turnon curve.

Layer 2 Scans:

Efficiency	Normalized Charge MPV
Efficiency vs Bias Voltage for all Layer 2 Scans up to 2013. Points were fitted with a turnon curve.	MPV of Normalized Charge vs Bias Voltage for all Layer 2Scans up to 2013. Points were fitted with a modified turnon curve.

Layer 3 Scans:

Efficiency	Normalized Charge MPV
Efficiency vs Bias Voltage for all Layer 3 Scans up to 2013. Points were fitted with a turnon curve.	MPV of Normalized Charge vs Bias Voltage for all Layer 3 Scans up to 2013. Points were fitted with a modified turnon curve.

Full Disk 1 Scans:

Efficiency	Normalized Charge MPV
Efficiency vs Bias Voltage for all Full Disk 1 Scans up to 2013. Points were fitted with a turnon curve.	MPV of Normalized Charge vs Bias Voltage for all Full Disk 1 Scans up to 2013. Points were fitted with a modified turnon curve.

Full Disk 2 Scans:

Efficiency	Normalized Charge MPV
Efficiency vs Bias Voltage for all Full Disk 2 Scans up to 2013. Points were fitted with a turnon curve.	MPV of Normalized Charge vs Bias Voltage for all Full Disk 2 Scans up to 2013. Points were fitted with a modified turnon curve.

Summary plots - All High Voltage Bias Scans:

The Following plots were derived from the above plots: The efficiency vs Bias voltage plots were fitted with turnon function (shown above on the left plots). The Bias voltage, where the efficiency is 1% below maximum were taken from the fitted functions. Similarly, the MPV vs Bias Voltage plots were also fitted with modified turnon curves. The MPV that corresponds to the "99% Efficiency" Voltage were taken using the fitted function (shown above on the right plots).

Voltage at 99% Hit Efficiency	Norm Charge MPV at the 99% Hit Efficiency
Bias Voltage at 99% Hit Efficiency vs Total Integrated Luminosity for all Bias Scans up to 2013	Normalized Charge MPV at 99% Hit Efficiency vs Total Integrated Luminosity for all Bias Scans up to 2013

Contact: Silvia Taroni Silvia.Taroni@cernNOSPAMPLEASE.ch

Presenting for approval on Apr 12, 2013

Analized run 201278, taken 21st August 2012, 8 TeV p - p collisions, Inst Lum from 6.3 x 10 ^{33} to 2 x 10^{33) cms ^{-2}s^{-1}.

To study possible dependencies of the cluster size and charge on the instantaneous luminosity, the run has been divided into 20 Instantaneous luminosity intervals. For each interval, a Monte Carlo (MC) simulation has been performed using the pile up distributions from data to simulate the pile up in MC. This procedure allows a more precise data - MC comparison, since the two samples have the same pile up.

Few basic selections have been applied for selecting events. Vertex requests: number of Vertices > 0, ndof < 5, z < 24 cm, position rho > 2. Track requests: track pt > 0.6 GeV

Data are from Minimum Bias primary dataset requiring HLT_ZeroBias trigger.

Used CMSSW releases: CMSSW_5_3_2_patch4 for MC and CMSSW_5_3_3_patch1 for data

Cluster Size

The cluster size of each on-track cluster has been considered. We have analysed the global cluster size, the size along x and along y direction. Considering clusters having the global/x/y size from 0 to 20, we can observe only minor differences in the cluster size distributions of data and MC. The plots are normalized to the area.

* ClusterSize:

The average cluster size (considering cluster smaller than 20 pixel in total or along x or y) has been studied as a function of the instantaneous luminosity. There are differences in the three layers but no dependence on the instantaneous luminosity has been observed as in MC.

* average cluster size vs inst lim:

Cluster Charge

The collected cluster charge has been also investigated. Considering on-track clusters, the charge has been studied separately for the three layers. For each instantaneous luminosity interval, the cluster charge in each layer has been fitted using a landau distribution convoluted with a gaussian. As an example, in the below plot we show the histogram and the fit for layer 2 for inst lumi ~5.6 x 10^{33} cm{-2}s{-1}. The only difference we see is the peak position: the peak from data is shifted wrt the simulated one.

* Cluster charge fit layer 2:

Plotting the MPV as a function of the instantaneous luminosity we have observed 2 effects. The difference between layers, expected from the different dose absorbed by the three layers, and a dependence on luminosity: the peak moves to higher values with the decrease of the luminosity. This can be explained as dependent on the readout electronics and not as a real increase in charge collection efficiency: the higher occupancy produces an increase of the readout chip temperature that influence the charge value in output. This effects is not simulated in MC.

* cluster charge MPV vs Inst Lum:

Contact: Armin Burgmeier Armin.Burgmeier@cernNOSPAMPLEASE.ch

Presenting for approval on Apr 12, 2013

The pixel resolution of the pixels in PXB2 has been measured. Tracks with pT > 12 GeV having hits in all three layers of the pixel barrel detector have been selected. The tracks are re-fit without the hit in the 2nd layer. Then, the residual difference between the hit position and the interpolated track is plotted. A student-t function is fit to the distribution. Assuming the resolution is the same in all three layers the width of the function fit divided by sqrt(3/2) gives the intrinsic pixel resolution. This takes into account that the hit positions in layers 1 and 3 are smeared as well.

The following datasets and conditions were used to create these plots:

• Datasets: * /Jet/Run2012A-22Jan2013-v1/RECO * /JetHT/Run2012B-22Jan2013-v1/RECO * /JetHT/Run2012C-PromptReco-v1/RECO * /JetHT/Run2012C-PromptReco-v2/RECO * /JetHT/Run2012D-PromptReco-v1/RECO
• RecHits are being reconstructed with the template builder ("WithAngleAndTemplate")
• Global Tag: GR_R_53_V18::All * Override template tag: SiPixelTemplateDBObject_38T_v6_offline * This is equivalent to the conditions used for the Winter13 Re-Reco

Example Triplet Residual distribution

An example triplet residual distribution from run 207487

Intrinsic Pixel Resolution as a function of time

The intrinsic pixel resolution is shown as a function of integrated luminosity. There is one data point for about every 500/pb of data taken. Each data point corresponds to one collision run.

Contact: Jelena Luetic jelena.luetic@cern.ch

Presenting for approval on Apr 12, 2013

Pixel cluster parameters, cluster size and cluster charge, were determined as a function of integrated luminosity. Data points correspond to a measurement for a certified longer run taken approximately every 0.5/fb. Selected events are required to have clusters on tracks with track pt > 1GeV.

Cluster Charge

Cluster charge distribution was determined for each layer in pixel barrel and fitted to Landau distribution convoluted with Gaussian. Charge most probable value was extracted from the fit.

* Example of fit for each layer

* Cluster charge vs. integrated luminosity

Changes in cluster charge MPV at 6.5/fb and 15/fb correspond to threshold readjustments during LHC technical stops and changes at 8/fb and 18/fb correspond to implementation of new gain calibration.

Cluster Size

Cluster size as a function of integrated luminosity was determined as a mean of the distribution for global cluster size, as well as for cluster sizes in local x and y directions. Jumps in cluster size at 6.5/fb and 15/fb correspond to threshold readjustments during LHC technical stops.

Contact: Jelena Luetic jelena.luetic@cern.ch

Presenting for approval on Apr 12, 2013

Lorentz angle as a function of integrated luminosity was determined using grazing angle method for 2012 data. Each data point represents Lorentz angle for a longer run taken approximately every 0.25/fb.

Basic selection requirements include high purity muon track, cluster size in y>4, track pt>3 GeV, cluster charge < 120000 e, hit residuals < 50 micrometers and chi2/ndof<2. Edge pixels were excluded as well.

Lorentz angle vs Integrated Luminosity

Lorentz angle for special runs

At the end of the run period, in 2012, data was taken with bias voltages different from nominal bias voltage value 150 V. Approximately 25/pb was taken with bias voltages 140V and 160V and some data (only few minutes of data taking) was taken with bias voltage of 300V. Lorentz angle was determined for each of these special runs with grazing angle method and plotted together with Lorentz angle in 2012 (luminosity not to scale).

Contact: Danek Kotlinski danek.kotlinski@psiNOSPAMPLEASE.ch

Presenting for approval on Apr 12, 2013

The average pixel thresholds measured with charge injection, using the "S-Curve" method. The thresholds are averaged over pixel barrel layers 1, 2 and 3 and over forward pixels. The steps at 6.0, 12.5 and 21 fb-1 correspond to readjustment of the thresholds, which have been performed at the beginning of 2012 and during two LHC technical stops during 2012 running period.

Plot of pixel thresholds during 2011 and 2012

The average pixel thresholds in units of 1000 electrons (1 ke) for barrel layers 1 (black) , 2 (red) and 3 (blue), and for the forward pixels (green) as a function of delivered integrated luminosity in 2011 and 2012.