CMS Tracker Detector Performance Results

Link to the Tracker Material budget categorization plots for phase-I pixel detector

Link to the estimation of phase-0 pixel tracker material budget with triplet method

Link to 2016 Strip performance plots approved in 2017

Link to 2017A Signal Over Noise

Link to 2018 Strip offline performance

Link to Strip offline performance in run 2 approved in 2019

Link to Simulation of the Silicon Strip Tracker pre-amplifier in early 2016 data

Link to 2017 strip radiation plots

Link to 2018 strip radiation plots

Link to 2019 strip radiation plots

Link to the Tracker Alignment plots: E/p validation on 2012 data and simulation

slides with explanations: pdf

individual plots, all clusters are on tracks:

|

Typical Landau distribution of cluster charge with example fit for MPV

Cluster size in local X coordinate in number of pixels, typical distribution

Cluster size in local Y coordinate in number of pixels, typical distribution

Total cluster size in number of pixels, typical distribution

Mean cluster size in local X and Y coordinates vs. integrated luminosity in 2012

Mean cluster size vs. integrated luminosity in 2012 for central tracks

Mean cluster size vs. integrated luminosity in 2012 for forward tracks

MPV of cluster charge vs. integrated luminosity in 2012 for central tracks

MPV of cluster charge vs. integrated luminosity in 2012 for forward tracks

slides with explanations: pdf

individual plots:

|

Subdetector Signal Vs Delay

Optimal Delay Vs Subdetector Layer

Tracker Status Map

slides ( pdf),

individual plots: |

Tracker Material in units of radiation length vs. eta

Tracker Material in units of interaction length vs. eta

strip module hit efficiency

pixel module hit efficiency

pixel barrel resolution in the z coordinate

Measurement of Pixel hit resolution from early 2011 data based on the Pixel Triplet method. The slides of the approval talk can be found here.

|

Resolution in rphi

Resolution in z vs. DIP angle

Results from the Laser Alignment System in the Strip Tracker using 2011 data. Find the slides from the approval meeting here. They contain a description of the system and supporting plots. Here are the approved plots for the public:

TEC+ laser beam profile

TOB laser beam profile

Online monitoring of calibration events

Laser parameter scan for calibration

Stability within a run; changes with respect to last 5 minutes

TIB stability projections

TEC+ stability projections

TEC- stability projections

Individual deviations: torsion in TIB

Individual deviations: chi2 in TIB

Individual deviations: temperature

Illustration for Lorentz Angle measurement

Measurement of Lorentz drift during Magnet ramp

On track strip clusters Signal to Noise ratio S/N for each part of the strip tracker, operated in deconvolution mode (nominal conditions). Update of plots approved in 2010.

TIB S/N

TOB S/N

This is the situation towards the end of the 2010 pp runs, where 10 of the 11 inefficient modules have been recovered.

Strip efficiency summary

Strip efficiency tacker map

Contact: Michael Segala, Thomas Speer
Presented for approval on 13 October 2010 at the general meeting: slides
Results from 3.9 pb-1 of data, all resolutions in microns.

Resolution as a function of strip pitch (eps)

Resolution for each sensor type (eps)

Tracker Map of resolutions (in microns)

Contact: Loic Quertenmont loic.quertenmont@cernNOSPAMPLEASE.ch
Presented for approval on June 23th, 2010 at general Meeting.
More details on the used technique are described in: CMS AN-2010/020.

High Purity Tracks with P>2.0GeV and 8 Single-Sides Hits have been used.

Data	MC
		Distribution of the Most Probable Value (extracted) from a Landau fit on the pathlength-corrected Charge Distribution for each SiStrip APV. Mean value of the MPV distribution depends on the thickness of the sensor and of the geometrical position (higher particle spectrum illumination). Effect well visible in MC
Tickmark Calibrated	Ticklark+Particle Calibrated	Data
		Pathlength Corrected Charge Distribution for a SiStrip Tracker calibrated for optical response by internal reference signal (tickmark) and for a SiStrip Tracker calibrated with Particles (on top of the previous calibration). The dashed line represent the inter-calibration value (300 ADC/mm) used by the particle gain calibration
Un-calibrated	Particle Calibrated	MC
		Pathlength Corrected Charge Distribution for an un-calibrated SiStrip Tracker and for a SiStrip Tracker calibrated with Particles. The dashed line represent the inter-calibration value (300 ADC/mm) used by the particle gain calibration.

Contact: Loic Quertenmont loic.quertenmont@cernNOSPAMPLEASE.ch
Presented for approval on June 23th, 2010 at general Meeting.

More details on the used technique are discribed in: CMS AN-2009/020, CMS AN-2010/020, CMS PAS TRK-10-001 as well as on the dEdx twiki page

Only collision events have been used: BPTX(TT0), BSC(TT41) and beam halo veto (! TT36||TT37||TT38||TT39)
A Good vertex must be present in the event
High Purity Tracks with |d0|<0.25cm and |dZ|<0.5cm and 12 Single-Sides Strip Hits have been used. SiStrip Cluster charge has been calibrated using gain with particles.

Harmonic-2 Estimator	Truncated-40 Estimator
		dE/dx Distribution for tracks with P>5.0GeV using the Pixel-Less Truncated-40 and Harmonic-2 estimators.
Data-Harmonic-2 Estimator	MC-Harmonic-2 Estimator
		Distribution of dE/dx vs P using the Pixel-Less Harmonic-2 estimators. Kaon and Proton bands are well visible in both the Data and MC. The Deuteron band is supressed in the MC, because PYTHIA is not producing deuterons.
Data-Truncated-40 Estimator	MC-Truncated-40 Estimator
		Distribution of dE/dx vs P using the Pixel-Less Truncated-40 estimators. Kaon and Proton bands are well visible in both the Data and MC. The Deuteron band is supressed in the MC, because PYTHIA is not producing deuterons.
Harmonic-2 Estimator	Truncated-40 Estimator
		Distribution of the reconstructed mass using momentum and dE/dx information of the tracks with P5.0MeV/cm using the Pixel-Less Truncated-40 and Harmonic-2 estimators. Difference between Data and MC for the Deuteron peak is due to the non-generation of Deuteron particles by PYTHIA.
Data-Asymmetric Smirnov Discriminator	MC-Asymmetric Smirnov Discriminator
		Distribution of dE/dx vs P using the Pixel-Less Asymmetric Smirnov discriminator. Kaon and Proton bands are well visible in both the Data and MC. The Deuteron band is supressed in the MC, because PYTHIA is not producing deuterons.
Data-Asymmetric Smirnov Discriminator	MC-Asymmetric Smirnov Discriminator
		dE/dx Template used by the discriminators to assign a probability that an observed cluster is compatible with a Minimum Ionizing Particle (MIP). What is shown is the probability that a MIP of a given pathlength (X-axis) would have release less charge than what was observed (Y-Axis). Tracks of at least 5.0GeV and more than 5 Single-Sided SiStrip Hits have been used to compute this dE/dx Templates. Note: This plot is relatively technical and should then be shown by expert only.

Contact: Carlotta Favaro, Hella Snoek
Presented for approval on 9 February 2011. A general description of the result can be found here. A detailed summary of the analysis is given in this presentation for the Tracker DPG meeting.

Transverse and longitudinal hit resolution as a function of the cluster length (png)

Individual plots are pasted below. Find more informations and names of contact persons in the following approval talks: Results for ICHEP 2010 are here: http://indico.cern.ch/conferenceDisplay.py?confId=81387

Results for kinks and bows are here: http://indico.cern.ch/conferenceDisplay.py?confId=99220 (with plots in pdf format currently at https://frmeier.web.cern.ch/frmeier/alignment/)

Results for updated FPIX DMR, Tracker tilts, PV validation in 2010 are here https://indico.cern.ch/conferenceDisplay.py?confId=123710

* TK alignment - KinksAndBows - Plot 1 - Cosmic track Chi2 probability:

* TK alignment - KinksAndBows - Plot 2 - TIB local residuals:

* TK alignment - KinksAndBows - Plot 3 - TOB local residuals:

* TK alignment - KinksAndBows - Plot 4 - Sagitta for 3 BPIX ladders:

• TK alignment - Tracker Tilt - example of Chi2 scan as a function of Theta_{X}:
  

• TK alignment - Tracker Tilt - example of Chi2 scan as a function of Theta_{Y}:
  

• TK alignment - Tracker Tilt - Results of Tracker Tilt alignment on 2010 data:
  

• TK alignment - Tracker Tilt - Results of Tracker Tilt alignment on MC: tiltbandmc.eps
  

• TK alignment - PV validaiton 2010 - DeltaZ separation of BPIX half shells vs time in 2010 (measured with PV residuals):
  

• dmr_FPIX_x.eps: TK alignment - DMR 2010 - DMR-u in FPIX after EoY2010 alignment

• dmr_FPIX_y.eps: TK alignment - DMR 2010 - DMR-v in FPIX after EoY2010 alignment

Results on PV validation, Z mass validation and kinks and bows were approved at the Run Organization Meeting on August 16, 2011, slides with little more explanation are attached to the agenda of the following CMS General Meeting. Individual plots are listed below.

• Primary vertex delta z validation: 2011_alignment_PVvalidation.eps or 2011_alignment_PVvalidation.pdf.
• Mean of (some) determined bow parameters
  • bow along x for single sensor modules: 2011_alignment_bowW20_singleSens.eps or 2011_alignment_bowW20_singleSens.pdf.
  • bow along x for double sensor modules: 2011_alignment_bowW20_doubleSens.eps or 2011_alignment_bowW20_doubleSens.pdf.
  • bow along y for single sensor modules: 2011_alignment_bowW02_singleSens.eps or 2011_alignment_bowW02_singleSens.pdf.
  • bow along y for double sensor modules: 2011_alignment_bowW02_doubleSens.eps or 2011_alignment_bowW02_doubleSens.pdf.
  • delta alpha for double sensor modules: 2011_alignment_bowDalpha.eps or 2011_alignment_bowDalpha.pdf.
• Z mass validation (positively charged muons):
  • mass vs eta: 2011_alignment_ZmassVsEtaPlus.pdf or with different marker 2011_alignment_ZmassVsEtaPlus_newMarkers.pdf or
    
  • mass vs phi: 2011_alignment_ZmassVsPhiPlus.pdf or with different marker 2011_alignment_ZmassVsPhiPlus_newMarkers.pdf or
    

Contact: Roberto Castello (roberto.castello@cernNOSPAMPLEASE.ch), Gero Flucke (gero.flucke@desyNOSPAMPLEASE.de)
Presented for approval on May 2nd, 2012 at Weekly General Meeting. Explanation of the plots is contained in the linked slides.

Control of weak modes and global deformations
pdf	pdf	Twist deformation

pdf

pdf

Skew deformation

pdf

pdf

Sagitta deformation

Tracker Surface deformations
pdf	pdf	Surface deformations in TIB
pdf	pdf	Surface deformations and kinks in TOB
pdf	pdf	Surface deformations and kinks in TEC
pdf		Surface deformations in the BPIX
pdf		Chi2 probabilty vs d0

Alignment precision in Pixel detector
pdf	pdf	DMR for Barrel pixels ( Data/MC design/MC misalignment)
pdf	pdf	DMR for Forward pixels ( Data/MC design/MC misalignment)

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

Presented for approval on May 13, 2011

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: ≥1 vertex, where |z|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>0.6 GeV, N_strip_hits >10, track consistent with vertex (cut on impact parameters d0, dz - layer dependent), track separation (RecHits separated by 5mm)

Efficiency vs Layers and Disks - first 2011 runs:

Efficiency vs layers and disks - systematic error: 0.002, statistical error: of order 10^-5

High Voltage Bias Scan - 2011 March:

BmO Sector6 Layer1 HV1	BpO Sector2 Layer3 HV1	BmI Disk1 ROG1 HV1
Dependence of efficiency on the applied bias voltage in the BPix HV group 1 BmO Sector6 Layer1	Dependence of efficiency on the applied bias voltage in the BPix HV group 1 BpO Sector2 Layer3	Dependence of efficiency on the applied bias voltage in the FPix HV group 1 BmI Disk1 Blade 1-3

Plots approved mainly for expert presentations:

2010 runs:

The initial average number of protons per bunch and the number of bunches per orbit (colliding at CMS) are shown in green. The observed efficiency loss is qualitatively understood, but not verified quantitatively. It is not due to detector degradation. Some dynamic efficiency loss is expected due to high occupancy because readout buffers can become full which may cause losses of data. The innermost layer, Layer 1, is the most susceptible to this effect Efficiency on FPix stays within the systematics uncertainty of 0.002

Barrel Pixel	Forward Pixel
Efficiency for each barrel layer vs dates of the largest 2010 runs (time scale is not linear)	Efficiency for each disks vs dates of the largest 2010 runs (time scale is not linear)

Integrated over all 2010 runs. Modules used in track seeding are excluded (eg Disk1 Panel1 Plaquette 2). X axis goes along global z, Y axis along -phi.

Hatched areas: inactive modules, White areas: no tracks or otherwise excluded from efficiency calculation

Statistical error - BPix: between 6.10-5 and 0.001, FPix: between 9.10-5 and 0.007. Systematic uncertainty: 0.002

Barrel Pixel	Readout chip efficiency - Layer1	Readout chip efficiency - Layer2	Readout chip efficiency - Layer3
Forward pixel	Readout chip efficiency - Inner FPix (+X side)	Readout chip efficiency - Inner FPix (-X side)

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

Presented for approval on May 13, 2011

Average cluster size (black) and efficiency (red) versus the relative phase between the LHC clock and the pixel clock.

The average cluster size is calculated using clusters attached to tracks with charge (uncorrected for incidence angle) greater than 12k electron. The the clock phase delay of the pixel clock is chosen to be 13 ns. This operating point is well within the maximum efficiency plateau.

Average on-track cluster size and sensor efficiency in the pixel detector at various delays of the pixel clock (delays w.r.t the LHC clock)

Barrel Pixel	Forward Pixel
Average on-track cluster size and sensor efficiency in the pixel detector at various delays of the pixel clock (delays w.r.t the LHC clock) - Barrel Pixel	Average on-track cluster size and sensor efficiency in the pixel detector at various delays of the pixel clock (delays w.r.t the LHC clock) - Forward Pixel

Barrel Pixel	Zoomed in around optimal point - Layer 1	Zoomed in around optimal point - Layer 2	Zoomed in around optimal point - Layer 3
Forward pixel	Zoomed in around optimal point - Disk -2	Zoomed in around optimal point - Disk -1	Zoomed in around optimal point - Disk -1	Zoomed in around optimal point - Disk -2

Contact: Maxime Gouzevitch, Maxime.Gouzevitch@cernNOSPAMPLEASE.ch and Giacomo Sguazzoni, Giacomo.Sguazzoni@cernNOSPAMPLEASE.ch.

Presented for approval on 10 June 2011.

Slides with explanations here and raw plots below.

• 2T vs. 3.8T difference vs. z in the fitted Beam Pipe center position x and y:
  

• Nuclear interactions xy map in the beam pipe region ofr -20cm<z<20cm:
  

Presented on 11- June-2011 at General Meeting

Approved plots

Presented on 21- June-2011 at Run Coordination meeting: were they finally approved ?

Approved (?) plots

Presented on 11- June -2011 at Tracker DPG meeting

Plots of Hit efficiency for 2011; Tracker mao with hybrid and sensor temperature; SS2 leak rate evolution

Approved (?) plots

Note also that the pixel online group approved many plots for PIXEL2010. Slides are on indico: http://indico.cern.ch/conferenceDisplay.py?confId=99220 http://indico.cern.ch/conferenceDisplay.py?confId=99218

Most of the approved material can also be found on afs (/afs/cern.ch/cms/performance/tracker/activities/results_2010/Approved/) but should not be considered without the accompanying material in the Detector Performance Note.

Presented for approval in run meeting on August 27 by Ben Kreis.

The thresholds represent absolute thresholds, determined in s-curve run 131247. If you use numbers, please quote the numbers provided in the plots 'with stats'.

Detector	with stats	without stats
BPIX	png pdf	png pdf
FPIX	png pdf	png pdf
combined BPIX andFPIX	png pdf	png pdf
overlayed BPIX and FPIX	png pdf	png pdf

Contact: Urs Langenegger urs.langenegger@psiNOSPAMPLEASE.ch

Contact: Andreas Jaeger andreas.jarger@cernNOSPAMPLEASE.ch

Presented for approval April 30th, 2010

Cluster charge for barrel and endcaps, compared to the MC prediction. This is an update of the results presented in the CRAFT08 paper: http://dx.doi.org/10.1088/1748-0221/5/03/T03007 There is a good description of data by MC, both in barrel and endcaps.

Pixel cluster charge in the barrel region

Pixel cluster charge in the endcap region

Cluster size in barrel and endcaps. There is a good description of data by MC, both in barrel and endcaps. The difference in cluster size between barrel and endcap is solely due to geometric effects (crossing angle of tracks in the barrel). This is an update of the results presented in the CRAFT08 paper: http://dx.doi.org/10.1088/1748-0221/5/03/T03007

Pixel cluster size in the barrel region

Pixel cluster size in the endcap region

Contact: Christophe Delaere, Thomas Speer, Laura Borrello christophe.delaere@cernNOSPAMPLEASE.ch, thomas.speer@cernNOSPAMPLEASE.ch, Laura.borrello@cernNOSPAMPLEASE.ch

Presented for approval April 30th, 2010

On track strip clusters Signal to Noise ratio S/N for each part of the strip tracker, operated in deconvolution mode (nominal conditions). These plots give the performances prior the final tuning of the sampling time. It will improve further for data following April 28th.

TIB	TID	TOB	TEC thin	TEC thick
19.4	18.5	22.5	19.4, 19.1	23.9, 23.4
TIB S/N	TID S/N	TOB S/N	TEC+ thin S/N TEC- thin S/N	TEC+ thick S/N TEC- thick S/N

Contact: Viktor Veszpremi, Viktor.Veszpremi@cernNOSPAMPLEASE.ch

Presented for approval on June 22, 2010

• DP note: CMS-DP-2010/022

Definition of efficiency: N_hits_found / N_hits_expected computed in the fiducial regions. Layer 1 eff. underestimated by ~1.5% due to tracks originating outside Layer 1 (secondaries)

Used 7 TeV collision data. Event selection: ≥1 vertex, where |z|4. Track selection: seeded from the Pixels, valid hits always required on „other” layers in order to minimize bias from seeding. Track p_T>0.9 GeV, N_strip_hits >10, track consistent with vertex, where |dz|<1.0 cm and |d0|<0.5 cm.

Sensor efficiency in each pixel layer

Effect of the applied HV bias on sensor efficiency. Efficiency curve expected to change in future due to aging and radiation effects. Current operating points are 150 V in BPix and 300 V in FPix (consistent with module testing).

Dependence of efficiency on the applied bias voltage in the BPix HV group BpO sector 2 Layer 3

Dependence of efficiency on the applied bias voltage in FPix BmI disk 1 port cards 1-3 (HV1)

Contact: Viktor Veszpremi, Viktor.Veszpremi@cernNOSPAMPLEASE.ch

Presented for approval on June 22, 2010

Average cluster size (black) and efficiency (red) versus the relative phase between the LHC clock and the pixel clock. During data taking the clock+trigger signal sent to the pixels was delayed in 2 ns steps.

The efficiency is calculated as above, but tracks are seeded from the strip tracker instead, and no valid hits are required (explicitly) on the other layers.

The average cluster size is calculated using all clusters with charge (uncorrected for incidence angle) greater than 12k electron. The advantage of this quantity is that it is totally independent of the tracking. The the clock phase delay of the pixel clock is chosen to be 10 ns in order to maximize the size of the clusters read out by the pixel detectors. This operating point is also well within the maximum efficiency plateau.

Average cluster size and sensor efficiency in the pixel detector at various delays of the pixel clock (delays w.r.t the LHC clock)

---

Plots approved mainly for expert presentations:

---

Pixel efficiency per layer. In FPix Disk 1, the maximum efficiency region has a negative slope due to misalignment of the clock phase among the read-out groups. An additive correction factor to the clock phase delay for each read-out group was derived such that after re-adjusting the clock+trigger delays, the average cluster size is maximized at a common clock phase delay (of ~10 ns) in BPix / FPix.

Pixel efficiency per layer at various delays of the pixel clock

Average cluster size at various delays of the pixel clock. FPix: at late clock phase, the average is biased towards large clusters, because the small (1-2 pixel) clusters with large average charge per pixel are no longer read out due to the timewalk effect.

The largest relative difference in the clock phases among the read-out groups was found to be 8 ns in the FPix (the clock+trigger signal arrives to this part of the pixel detector the latest.) The first plot shows the efficiency vs clock delay curve of this (black) and two other read-out groups. The second plot shows the same curves after the clock+trigger delay adjustment is applied. Note that the second plot is merely a derivative of the first one. (In order to test a new detector configuration, one would need to perform another delay scan measurement.) The merit of the second plot is that it demonstrates how well the correction works that is derived from the average cluster size measurements.

Efficiency vs clock delay in three read-out groups in the FPix measured during the delay scan

Efficiency vs clock delay after the internal alignment of the FPix.

The largest relative difference in the clock phases among the BPix read-out groups was found to be 3 ns (the clock+trigger signal arrives to this part of BPix the earliest.) The first (second) plot shows the efficiency vs clock delay curve of this and two, well-aligned groups before (after) the internal alignment of the BPix.

Efficiency vs clock delay in three read-out groups in the BPix measured during the delay scan

Efficiency vs clock delay after the internal alignment of the BPix.

Contact: Christophe Delaere christophe.delaere@cernNOSPAMPLEASE.ch

Presented for approval on April 30th, 2010

The timing is measured independently for each partition. For endcaps, where thin and thick sensors are mixed, each type of sensor is considered separately. The dashed line represents the working point determined in December 2009, measuring the timing of the TOB partition only (see TRKApprovedResults900GeV). The optimal point is obtained by considering the maximum of the fitted pulse shape, using the analytical deconvoluted signal description.

From the plot on sees the good state of the initial timing. There is a 2ns effect between thin and thick sensors, just as expected from the difference of drift time in the bulk of the silicon. The differences from partition to partition are less than 2ns, which is compatible with the resolution of the initial cross-partition timing. The various partitions will be corrected on April 28th, so that data following the April technical stop should have all partitions and sensors better in sync.

Details of the procedure are presented in CMS Note 2008/007.

Strip delay scan

Strip efficiency