Public Pixel Tracker Plots for Collision Data
- Introduction
- Public Plots tracking since 2020 with Glance
- Performance Plots tracking June 2015 - 2019
- PHYS and INDET PUB NOTES
- Additional notes with approved plots
- Data 2012 - Module Error Rate Plots
- Data 2012 - Depletion Depth Plots
- Data 2011 - Depletion Depth Plots
- Data 2011 - MC Comparison Plots
- Lorentz Angle plots
- Noise occupancy and association efficiency
- Cluster properties and charge sharing
- Timing plots
- Energy loss in Pixel
- Beam background studies in Pixel
- Links
Introduction
The Pixel detector collision plots below are approved to be shown by ATLAS speakers at conferences and similar events. Procedure to approve plots is illustrated here.- Previous (before 2020) Plots on the Pixel cosmics and calibration can be found here.
- Figures come mainly from collisions data taking.
- Plots related to the alignment of the Pixel detector and the entire inner tracking detector can be found here
- Other Publications on the Pixel detector can be found here
Public Plots tracking since 2020 with Glance
Since 2020 we list below all the Pixel Public plots, calibration, cosmics and collisions. IMPORTANT Glance allows an easy access to the ATLAS Plots. This is the link
to the full list of the Pixel Plots since 2018.
| Public Link | Full Title | Publication Date | Glance link | Contact Editor |
|---|---|---|---|---|
| PIX-2020-001 |
Studies on SEE in the IBL detector | 2020/03/19 | PLOT-PIXE-2020-01 |
Juan Antonio Garcia Pascual |
| PIX-2020-002 |
Silicon effective band gap energy in Pixel and IBL | 2020/05/28 | PLOT-PIXE-2020-02 |
Aidan Grummer |
| PIX-2020-003 |
Pixel Hit-on-track Efficiency, Average Occupancy and Fraction of Disabled Modules for Run2 | 2020/07/27 | PLOT-PIXE-2020-03 |
Kanae Asai |
| SEE FEI4 Papers |
Measurements of Single Event Upset in ATLAS IBL (overwrites preliminary results) | 2020/06/24 | Paper |
Sasha Rozanov |
| IDET-2020-001 |
Measurements of Sensor Radiation Damage in the ATLAS Inner Detector using Leakage Currents | 2020/07/16 | PLOT-IDET-2020-01 |
Ben Nachman |
| PIX-2020-004 |
IBL Low Voltage current evolution | 2020/09/22 | PLOT-PIXE-2020-04 |
Sasha Rozanov |
| PIX-2022-001 |
Pixel HV Scans LS2 | 2022/07/05 | PLOT-PIXE-2022-01 |
Marco Battaglia |
Performance Plots tracking June 2015 - 2019
PHYS and INDET PUB NOTES
| Reference | Full Title | Publication Date |
|---|---|---|
| ATL-INDET-PUB-2014-005 |
Using the size of clusters in the ATLAS pixel detector to reject spurious clusters and provide initial estimate of track's direction and position along the beam line | 2014/09/05 |
| ATL-INDET-PUB-2016-001 |
IBL Efficiency and Single Point Resolution in Collision Events | 2016/08/03 |
Additional notes with approved plots
- Pixel Clusters Properties in 2012
(ATL-COM-INDET-2014-010) - Stability of the Proton Mass measured using the Pixel ionization in Run 1
(ATL-COM-INDET-2014-007) - Module failures: impact on b-tagging performance and analysis
(ATL-COM-INDET-2013-079) - Mapping the material in the ATLAS Inner Detector using secondary hadronic interactions in 7 TeV collisions
(ATLAS-CONF-2010-058)
Updated plots are available in ATL-COM-PHYS-2011-296 and ATL-COM-PHYS-2011-584
- dE/dx measurement in the ATLAS Pixel Detector and its use for particle identification
(ATLAS-CONF-2011-016)
Data 2012 - Module Error Rate Plots
border=1 cellpadding=10 cellspacing=10>| The average number of modules with readout errors per event in the innermost pixel layer (consisting of 286 modules), as a function of the time withing a run in luminosity block units (one luminosity block corresponds approximately to one minute of data-taking). Beacuse of single event upset a module may get stuck in a permanent error state. Therefore an automatic recovery procedure has been developed which detects modules in such state and reset them. The data are taken from a run before (black points) and after (blue points) the recovery procedure was introduced. After the automatic recovery, the number of modules in error state has decreased significantly. |
SynchErrors_208354_206369.eps |
Data 2012 - Depletion Depth Plots
border=1 cellpadding=10 cellspacing=10>|
Dependence between track depth and angle of incidence. Single track depth values are shown in the scatter plot. The corresponding size of the cluster in which the track depth was calculated is indicated by a colour code. The maximum track depth values are superimposed. They are extracted from error function fits for different angle slices.
(Bottom) Example of a track depth distribution for one angle slice. The incidence angle varies between 1.3 rad and 1.4 rad. The corresponding error function fit is also shown. The minimum and maximum track depth values are defined by the inflection points of the error function fit. Contact: Andre Lukas Schorlemmer, September 2014 Reference: ATLAS-COM-CONF-2014-059 internal
|
02.07.2012 
eps, pdf version 02.07.2012, zoom 
eps, pdf version |
|
Cluster depth distribution for the bias voltage scan taken at the 02.07.2012, 26.09.2012, 01.11.2012, 22.01.2013 and the
29.01.2013.
The bias
voltage of layer 0 was increased stepwise during the scan until the sensor was fully
depleted. Tracks from collision data were used to calculate the cluster depth. Only
clusters located on modules with a module position on stave of -3 in layer 0 are used in
this plot. The most probable value of the cluster depth increases with increasing bias
voltage until the sensors are fully depleted.
Contact: Andre Lukas Schorlemmer, September 2014 Reference: ATLAS-COM-CONF-2014-059 internal
|
02.07.2012 
eps, pdf version 26.09.2012 
eps, pdf version 01.11.2012 
eps, pdf version 22 and 29.01.2013 
eps, pdf version |
|
Depletion depth as a function of the Module Position on Stave. Only pixel layer 0 is shown. The bias voltage
scan was taken at the 02.07.2012. At the beginning of July 2012 it was expected that modules in layer 0
start to undergo type-inversion. The bias voltage of layer 0 was increased stepwise during the scan to a
maximum of 150 V. At the date of the scan this maximum voltage was definitely high enough to fully deplete the sensors. The depletion depth increases with increasing bias voltage until full depletion is reached, as
expected after type-inversion. It is clearly visible that the depletion voltage is very small right after type-inversion, because the depletion depth does not increase significantly for bias voltages higher than 20 V.
The amount of statistics is too low for a measurement in the centre of the detector at low bias voltages.
The error bars are purely statistical and
they are in general smaller than the systematic uncertainty, which varies between 10 and
20 μm.
In the bottom plot, a zoom into the region with higher bias voltages is shown. Contact: Andre Lukas Schorlemmer, September 2014 Reference: ATLAS-COM-CONF-2014-059 internal
|
02.07.2012 
eps, pdf version 02.07.2012, zoom 
eps, pdf version |
|
Depletion depth as a function of the Module Position on Stave. Only pixel layer 0 is shown
and the bias voltage scan was taken at the 26.09.2012. The bias voltage of layer 0 was
increased stepwise during the scan to a maximum of 150 V. At the date of the scan this
maximum voltage was definitely high enough to fully deplete the sensors. The amount of
statistics is too low for a measurement in the centre of the detector. The error bars are
purely statistical and they are in general smaller than the systematic uncertainty, which
varies between 10 and 20 μm.
In the bottom plot, a zoom into the region with higher bias voltages is shown. Contact: Andre Lukas Schorlemmer, September 2014 Reference: ATLAS-COM-CONF-2014-059 internal
|
26.09.2012 
eps, pdf version 26.09.2012, zoom 
eps, pdf version |
|
Depletion depth as a function of the Module Position on Stave. Only pixel layer 0 is shown
and the bias voltage scan was taken at the 01.11.2012. The bias voltage of layer 0 was
increased stepwise during the scan to a maximum of 150 V. At the date of the scan this
maximum voltage was definitely high enough to fully deplete the sensors. The amount of
statistics is too low for a measurement in the centre of the detector. The error bars are
purely statistical and they are in general smaller than the systematic uncertainty, which
varies between 10 and 20 μm.
In the bottom plot, a zoom into the region with higher bias voltages is shown. Contact: Andre Lukas Schorlemmer, September 2014 Reference: ATLAS-COM-CONF-2014-059 internal
|
01.11.2012 
eps, pdf version 01.11.2012, zoom 
eps, pdf version |
|
Depletion depth as a function of the Module Position on Stave. Only pixel layer 0 is shown.
The bias voltage scan was taken between the 22.01.2013 and the 29.01.2013.
The bias voltage of layer 0 was
increased stepwise during the scan to a maximum of 150 V. At the date of the scan this
maximum voltage was definitely high enough to fully deplete the sensors. The amount of
statistics is too low for a measurement in the centre of the detector. The error bars are
purely statistical and they are in general smaller than the systematic uncertainty, which
varies between 10 and 20 μm.
In the bottom plot, a zoom into the region with higher bias voltages is shown. Contact: Andre Lukas Schorlemmer, September 2014 Reference: ATLAS-COM-CONF-2014-059 internal
|
Jan.2013 
eps, pdf version Jan.2013, zoom 
eps, pdf version |
|
Depletion depth as a function of the Module Position on Stave. The depletion depth is
shown for pixel layer 1. The bias voltage scan was taken at the 01.11.2012. The bias voltage of layer 1 was
increased stepwise during the scan to a maximum of 150 V. At the date of the scan this
maximum voltage was definitely high enough to fully deplete the sensors. The amount of
statistics is too low for a measurement in the centre of the detector. The error bars are
purely statistical and they are in general smaller than the systematic uncertainty, which
varies between 10 and 20 μm.
In the bottom plot, a zoom into the region with higher bias voltages is shown. Contact: Andre Lukas Schorlemmer, September 2014 Reference: ATLAS-COM-CONF-2014-059 internal
|
01.11.2012 
eps, pdf version 01.11.2012, zoom 
eps, pdf version |
|
Depletion depth as a function of the Module Position on Stave. The depletion depth is
shown for pixel layer 1.
The bias voltage scan was taken between the 22.01.2013 and the 29.01.2013.
The bias voltage of layer 1 was
increased stepwise during the scan to a maximum of 150 V. At the date of the scan this
maximum voltage was definitely high enough to fully deplete the sensors. The amount of
statistics is too low for a measurement in the centre of the detector. The error bars are
purely statistical and they are in general smaller than the systematic uncertainty, which
varies between 10 and 20 μm.
In the bottom plot, a zoom into the region with higher bias voltages is shown. Contact: Andre Lukas Schorlemmer, September 2014 Reference: ATLAS-COM-CONF-2014-059 internal
|
Jan.2013 
eps, pdf version Jan.2013, zoom 
eps, pdf version |
|
Depletion depth as a function of the Module Position on Stave. Only pixel layer 0 is shown.
The bias voltage scan was taken at the 01.11.2012. The amount of statistics is too low for
a measurement in the centre of the detector at low bias voltages.
For proton-proton collisions the muon stream has been used to measure the depletion depth. However, it is possible to use several other streams. For this measurement the minbias stream was used as a crosscheck.. The error bars are purely statistical and they are in general
smaller than the systematic uncertainty, which varies between 10 and 20 μm.
In the bottom plot, a zoom into the region with higher bias voltages is shown. Contact: Andre Lukas Schorlemmer, September 2014 Reference: ATLAS-COM-CONF-2014-059 internal
|
01.11.2012 
eps, pdf version 01.11.2012, zoom 
eps, pdf version |
|
Depletion depth as a function of the Module Position on Stave. The depletion depth is only
presented for pixel layer 0. Monte Carlo data with two different depletion depths is shown
in the figure. The Monte Carlo data samples have been produced to validate the method.
Contact: Andre Lukas Schorlemmer, September 2014 Reference: ATLAS-COM-CONF-2014-059 internal
|
eps, pdf version |
Data 2011 - Depletion Depth Plots
border=1 cellpadding=10 cellspacing=10>| The Depletion Depth of the Atlas Pixel Detector is shown as a function of the track incidence angle in the long pixel direction. Results are shown for two 2011 data runs and compared with Monte Carlo simulations. Two sets of Monte Carlo simulations have been produced one with an active sensor thickness S = 250 μm and a second one with S = 200 μm to validate the method. The detector ran fully depleted in 2011. Therefore, the measured depletion depth is in agreement with the active sensor thickness of S ~ 250 μm. The shown errors bars are purely the statistical errors on the fit to the track depth distribution of each angular slice. An additional error from the surface point correction contributes to the mean value of the Depletion Depth. The statistical error of the surface point is ~ 8 μm for MC Simulations and ~ 3 μm for beam data. The systematic uncertainty is 10 μm. |
Alldepl.eps |
| The plot with only data. The reference document for these plots can be found in CDS: https://cdsweb.cern.ch/record/1428448?ln=en
|
Datadepl.eps |
Data 2011 - MC Comparison Plots
border=1 cellpadding=10 cellspacing=10>| Data-MC comparison of cluster size versus angle for 2011 data and Pythia8 |
ClusAng.eps |
| Pixel Cluster size in eta - Data-MC comparison - data 2011, Pythia8 |
PixDeltaCol.eps |
| Pixel Cluster size in phi - Data-MC comparison - data 2011, Pythia8 |
PixDeltaRow.eps |
| Reconstructed primary vertices per bunch crossing - All clusters - Data-MC comparison - data 2011, Pythia8 |
AllClus.eps |
| Reconstructed primary vertices per bunch crossing - On-track clusters - Data-MC comparison - data 2011, Pythia8 |
TrackClus.eps |
Lorentz Angle plots
border=1 cellpadding=10 cellspacing=10>| *Pixel cluster width as a function of the track incident angle in Rphi direction.*
Dataset 900 GeV: runs of period 10th of December 2009 to 14th of December 2009 run numbers: 142165, 142166, 142191, 142193, 142195, 142383 MinBias stream, Threshold: 4000 e Dataset 7 TeV: run at 30/3/2010 (run number: 152166) MinBias stream, Threshold: 3500 e Magnetic Field is ON x-axis corresponds to the track incidence angle on the pixel module (r-phi plane) in local reference frame Only clusters on tracks are considered Cuts applied:
Entries for 900 GeV plot: ~1M clusters (after applying all cuts) Entries for 7 TeV plot: ~3.9M clusters (after applying all cuts) Both positive and negative tracks are taken into consideration Kink for 7 TeV plot at low angles, due to limited statistics at this region Slight difference between 900 GeV and 7 TeV plots is due to the difference in threshold Fit function: 
have a look at reference Reference: http://cdsweb.cern.ch/record/1248606
Lorentz angle values, derived from fit:
|
eps version 
eps version 
eps version | ||||||
| Pixel cluster width as a function of the track incident angle in Rphi direction.
Tracks are selected with the cuts of the cosmics note on the number of hits: nPixel > 1 and 5*nSCT+nTRT > 29, using only barrel hits. Some selection on the pixel hit on track is also done as in the pixel Lorentz note (track extrapolation inside the pixel cluster, at least two rows and columns between the cluster and the border, no ganged pixels, and pseudorapidity of the track less than 1). Data of two different runs and the 900 GeV simulation (with r871 tag, i.e. day-1 misalignments) are compared. The agreement in cluster size vs angle is good, the small difference is due to the fact that the Lorentz angle in the simulation is slightly larger than in the data. |
LorentzDataVsSimBonOff.eps |
Noise occupancy and association efficiency
border=1 cellpadding=10 cellspacing=10>| *Noise occupancy*
Pixel Detector occupancy in randomly triggered events with empty bunches. Noise rate is dominated by few pixels (300-1500 out of 80M) which are detected on a run-by-run basis by offline prompt calibration and masked during the bulk processing. For the bulk processing, the remaining noise occupancy is <10-9 hit/ pixel/BC, corresponding to <0.2 noise hits per event when reading out 5 BC. The runs shown correspond to the data taking period 18th April – 9th May 2010 |
eps version |
| *Efficiency*
Efficiency for a track to have an hit associated when crossing a Pixel Detector layer. Data shown are for express stream on 7 TeV run 155112 (15th May 2010), values taken from DQ monitoring after bulk reconstruction. Dead modules are excluded from the association efficiency computation, but otherwise dead regions contribute to the inefficiency. The full efficiency of B-Layer is due to the track selection, the lower efficiency for the most external disks is mainly due to inefficient regions on some modules. Error bars are smaller then marker sizes. |
eps version |
| *Inefficiency sources*
The lower efficiency of the most external endcap disks can be attributed to dead regions and dead FE of few modules. The top plot show the efficiency for all modules of EndCap A. Empty bins are dead modules. The right plots show, for three low efficiency modules, the hitmap collected on runs 152166-153200, corresponding to the first two weeks of operation. Inefficiency is due to a front-end failure for the top module, and region of disconnected bumps at the module edge. |
eps version 
eps version 
eps version 
eps version |
Cluster properties and charge sharing
border=1 cellpadding=10 cellspacing=10>| *Cluster size as a function of track incident angle*
The size of pixel clusters depends on particle incident angle w.r.t. the detector module surface. Cosmic rays and particles originating in collisions feature different incoming direction w.r.t. the detector modules. In this plot the cluster size along the precise pixel direction is represented for hits measured in the barrel. The incident angle value has been corrected by subtracting the value of the Lorentz angle. The full angular range [-90°; 90°] is covered by cosmic rays, while only a smaller range is covered by particles originating in collisions |
eps version b/w png version b/w eps version 
eps version b/w png version b/w eps version |
| *Cluster size as a function of track incident angle*
In this plot the cluster size along the beam direction is represented for hits measured in the barrel. The full pseudorapidity range is covered by particles originating in collisions while only a smaller range is covered by cosmic rays. |
eps version b/w png version b/w eps version 
eps version b/w png version b/w eps version |
| *Fit of charge sharing correction*
Residual between track extrapolation and the centre-of-cluster position in the Pixel Detector for two-pixel clusters in the local y direction and different incident angles. Residuals are plotted as a function of the charge sharing among pixels in the cluster and the slopes of the distribution are fit. The measured slopes (!) are used to improve the position resolution with respect to the purely binary readout according to the formula 
|
eps version |
| *Value of charge sharing correction as a function of track incident angle*
The charge sharing correction (\Delta – described in previous plot) depends on the cluster size and on the particle incident angle w.r.t. the detector module surface. During the “offline calibration loop” such correction is computed and parameterized for each direction of the pixel clusters in order to reach the optimal resolution in the detector. Here the charge correction is drawn for clusters made of two rows of pixels, as a function of the track incident angle, corrected for the Lorentz angle. A comparison between the correction computed on cosmic ray data and on collision data is reported. For the angular range in which both datasets feature sufficient statistics for the computation of the correction, results are nicely compatible, with small differences due to different running conditions (read-out threshold). |
eps version |
| *Value of charge sharing correction as a function of track incident angle*
Here the charge correction is drawn for clusters made of two rows of pixels, as a function of the track incident angle, corrected for the Lorentz angle. A comparison between the correction computed on cosmic ray data and on collision data is reported. For the angular range in which both datasets feature sufficient statistics for the computation of the correction, results are nicely compatible, with small differences due to different running conditions (read-out threshold). |
eps version |
| *Value of charge sharing correction as a function of track incident angle*
Here the charge correction is drawn for clusters made of three rows of pixels, as a function of the track incident angle, corrected for the Lorentz angle. A comparison between the correction computed on cosmic ray data and on collision data is reported. For the angular range in which both datasets feature sufficient statistics for the computation of the correction, results are nicely compatible, with small differences due to different running conditions (read-out threshold). |
eps version |
| *RMS of residuals as a function of track incident angle (Collision data)*
The RMS of residuals along the precise pixel direction is presented for collision data, as a function of the track incident angle, corrected for the Lorentz angle. The angular range available for particles originating in collisions is smaller with respect to cosmic rays. Nevertheless, the improvement in resolution due to charge sharing based correction is visible in the range where two-rowclusters are the most probable. The overall RMS is higher with respect to cosmic ray data, due to the average lower momentum of particles. Updated in March 2011. Top plot: DATA, bottom plot: MC. Same overall behaviour as in data, more statistical fluctuations in MC. Sample: data10_7TeV.00167844.physics_JetTauEtmiss.merge.DESDM_TRACK.r1774_p327_p333, PIXEL/PixReco tag; PixelOfflineReco-7TeV-000-04; Autumn reprocessing (much improved alignement); DESDM (richer in high pT tracks). Plots obtained using PixelCalibAlgs-00-04-15 Tracks considered: pT ≥ 5 GeV; 5*nSCTHits + nTRTHits ≥ 30; Cluster accepted if √((1000/GeVTrkPt)2 + ResCut 2) ≥ ResDigital, with ResCut = 80 μm for φ and ResCut = 400 μm for η. RMS : computed only for points where the statistics is sufficient ( > 50 clusters); weighted RMS, considering only bins within [mean - 3*RMS(residuals histogram), mean + 3*RMS(residuals histogram)]; Residual histograms have 100 bins, and range [-150, 150] μm in phi, [-400, 400] μm in η. |
eps version 
eps version |
| *RMS of residuals as a function of track incident angle (Collision data)*
The RMS of residuals along the beam direction is presented for collision data. The improvement determined by charge sharing based correction is clearly visible in the rage where two-column clusters are present. Plot updated in March 2011. Top plot: DATA, bottom plot: MC. Same overall behaviour as in data, more statistical fluctuations in MC. Sample: data10_7TeV.00167844.physics_JetTauEtmiss.merge.DESDM_TRACK.r1774_p327_p333, PIXEL/PixReco tag; PixelOfflineReco-7TeV-000-04; Autumn reprocessing (much improved alignement); DESDM (richer in high pT tracks). Plots obtained using PixelCalibAlgs-00-04-15 Tracks considered: pT ≥ 5 GeV; 5*nSCTHits + nTRTHits ≥ 30; Cluster accepted if √((1000/GeVTrkPt)2 + ResCut 2) ≥ ResDigital, with ResCut = 80 μm for φ and ResCut = 400 μm for η. RMS : computed only for points where the statistics is sufficient ( > 50 clusters); weighted RMS, considering only bins within [mean - 3*RMS(residuals histogram), mean + 3*RMS(residuals histogram)]; Residual histograms have 100 bins, and range [-150, 150] μm in phi, [-400, 400] μm in η. |
eps version 
eps version |
| *RMS of residuals as a function of track incident angle (Cosmic ray data)*
The resolution of the Pixel Detector is determined, among other aspects, by the incident angle of the particles. In these plots, to study the resolution of the detector, the residuals between cluster position and track extrapolation are presented. The intrinsic resolution of the detector and the track extrapolation uncertainty sum up in determining the RMS of residual distributions The RMS of residuals along the precise pixel direction is presented for cosmic ray data, as a function of the track incident angle, corrected for the Lorentz angle. A comparison between the two algorithm used to determine the cluster position in the detector is visible: the charge sharing algorithm improves resolution, in particular for incident angles in the range (-25 deg;-5 deg) and (5 deg;20 deg). This is in fact the range where two-row clusters are the most probable. |
eps version |
| *RMS of residuals as a function of track incident angle (Cosmic ray data)*
The RMS of residuals along the beam direction is presented for cosmic ray data. The improvement determined by charge sharing based correction is clearly visible, in the rage where two-column clusters are present. Due to the cosmic rays angular distribution, low statistics is available for high values of pseudorapidity. |
eps version |
Timing plots
border=1 cellpadding=10 cellspacing=10>| *LVL1A clusters ON track*
Dataset: data10_7TeV. 00155160.express_express.merge.NTUP_TRKVALID.x10_m469 17th May 2010 run, sqrt(s) = 7 TeV Read out window: 4 BCs Threshold: 3500 e Clusters associated to tracks:
Differences in entries for the less populated bins (L1A =0, L1A =2 and L1A =3) are due to timewalk effect |
eps version 
eps version 
eps version |
| *LVL1A clusters OFF track*
Dataset: data10_7TeV. 00155160.express_express.merge.NTUP_TRKVALID.x10_m469 17th May 2010 run, sqrt(s) = 7 TeV Read out window: 4 BCs Threshold: 3500 e All Pixel clusters selected without association to tracks
|
eps version 
eps version 
eps version |
| *Result of fine scans*
The plots show the results of the two timing scans we performed during collision data taking. A global shift of the clock phase was applied to the Pixel Detector in fine steps (1 ns for the left plot and 0.5 ns for the right) and the time for which the hits started to be registered out of time (i.e. registered in the previous bunch crossing) was measured for individual modules. The plots show the last scan point in which all the module hits were still in time. The fine adjustment of the clock phase of individual modules was performed in between for each module using the delay circuitry in the off-detector readout electronics. The sigma of the plots show the good timing alignment of the modules after the adjustment, however the position of the mean requires a correction to prevent that several modules would register hits out of time. |
eps version 
eps version |
| *Late hits*
Clusters corresponding to late BC are not associated to tracks (7 TeV, run 155160). The distribution of the position inside a module shows that they are mostly:
|
eps version b/w png version b/w eps version |
| *Timewalk from collision data*
The timewalk can also be observed directly on data looking at the average relative BC in which a hit is observed as a function of the collected charge. In such a case the resolution is limited by the B granularity, and the response is flat above 10 000 e because almost all hits are detected in the correct BC. The ToT -Charge relationship cannot be well calibrated near to the threshold region, therefore sometimes low ToT hits are reconstructed with a charge lower than the 3 500 e threshold and the timewalk profile is strongly deformed. This plot is obtained from run 155073 |
eps version |
| *Synchronization*
Module synchronization can be assessed by checking the average bunch crossing detected for high charge (>15 000 e) single-pixel clusters. After timing adjustement on collision data, on all the barrel module the dispersion is 0.007 BC, corresponding to 0.17 ns. The same plot for cosmic-rays, obtained before the module-by-module time adjustment was carried out, was giving a RMS of 0.17 BC, corresponding to 4.25 ns The bin width is 0.004 BC = 0.1 ns |
eps version |
| For low luminosity data-taking, the Pixel Detector readout window covers 5 bunch crossings, centered around the trigger (bunch crossing=2 in figure). Even before correcting for a few out-of-time modules, the standard data-taking mode for the Pixel Detector readout for early 2010 data, which corresponds to a three bunch- crossing window, will be 99.95% efficient for clusters on tracks.
From their time distribution, most clusters not associated to tracks are not due to noise but to low momentum particles generated by pp interactions and not reconstructed. The late clusters in the off track distribution are due to low charge deposits that are either near the edge of the active region or fakes due to a readout ambiguity in the region between two front-end chips. |
PixelLVL1A_SolOn.eps |
| A timing monitoring histogram for a single non-T0-timed Pixel module. That means that for this module the sampling clock phase relative to collision timing is not optimized for collection of pixel hits in a single bunch crossing. The histogram shows Pixel ToT versus relative bunch crossing of the Pixel readout for all hits sampled by the online ROD monitoring during a single run with 900 GeV collisions. The size of readout window was 5 bunch crossings long; the small fraction of noise hits with predominantly lower Pixel ToT can be observed in the first bin. Because the module was not properly T0-timed the high amplitude signals (i.e. with high Pixel ToT) fall into the relative bunch crossing bin 1, as a result of the timewalk effect. |
eps version |
| A timing monitoring histogram for a single T0-timed Pixel module. This implies that for this module the sampling clock phase relative to the collision timing is optimized for collection of pixel hits in a single bunch crossing. The histogram shows Pixel ToT versus relative bunch crossing of Pixel readout for all hits sampled by the online ROD monitoring during a single run with 900 GeV collisions. The size of readout window was 5 bunch crossings long; due to very low noise occupancy of this module no hits are seen in the first two bins. The timewalk effect is clearly seen - high amplitude signals (i.e. with high Pixel ToT) are detected earlier, i.e. they have lower relative bunch crossing. |
eps version |
Energy loss in Pixel
border=1 cellpadding=10 cellspacing=10>| *Energy loss in Pixel*
The track dE/dx is calculated starting from the charge collected in the pixel clusters associated to the track itself. To have a good measurement at the track level, a selection at the cluster level has to be applied in order to be sure that the charge has been properly collected. Need to exclude the edge of the detector and the ganged region where the collected charge is lower. This is achieved by requiring the position of the cluster in a fiducial region excluding the border of the sensor and the ganged pixels:
The fraction of clusters that survives this cut (Good clusters) is 91%. But the number of tracks for which the track dE/dx is not measurable (assuming that at least two clusters are required) is reduced only by 3% | |
| *dE/dx scatter plots*
The track dE/dx is defined as an average of the individual cluster dE/dx measurements (charge collected in the cluster, corrected for the track length), for all the Good Pixel Clusters associated to the track. To reduce the Landau tails, the average is evaluated after having removed the cluster(s) with the highest charge: one cluster is removed for tracks with 2,3,4 good clusters; two clusters for tracks with 5 or more good clusters. To properly measure the track momentum, a cut on the number of SCT hits is required. Plots of track dE/dx vs momentum are presented according to the number of Good Pixel Clusters associated to the track (=1,=2,=3,=4,=5,>=6 and >=3) for the full detector acceptance and for the Barrel only (|h|<1.8). The distribution for tracks with only one good cluster is shown as well, but in this case dE/dx is not considered a useful measurement. For not specialistic Pixel talks, we suggest to use the plot with at least 3 Good Pixel Clusters (first 2). Bands for pions, kaons and (anti-)protons are clearly visible. Deuterons are visible as well. The resolution of the track dE/dx is shown later in this table Run Number: 152166, 152214, 152221, 152441 (Ecm = 7 TeV) Data format: data10_7TeV.0015xxxx.physics_MinBias.merge.NTUP_MINBIAS.f239_p127_tid125123_00) For a complete note use dE/dx note
|
eps version 
eps version 
eps version 
eps version 
eps version 
eps version 
eps version 
eps version 
eps version 
eps version 
eps version 
eps version 
eps version 
eps version |
| *Track dE/dx resolution*
A track dE/dx resolution of ~12% is measured using particles with p > 3 GeV ( to minimize spread due to low beta ionization), with at least 6 SCT hits (to assure track quality) and requiring at least 3 Good clusters (to properly measure the Track dE/dx). Mean and sigma are obtained from a gaussian fit of the data. Samples:
|
eps version |
| Energy loss in Pixel (900 GeV)
This plot shows the dependence of the dE/dx on the track momentum and charge.
Notice how protons and kaons are easily identifiable at low p for the higher energy deposit. Tracks are selected to have at least three barrel pixel hits, some cuts are imposed on the track angles in order to maximize the probability that the charge collection is complete. The dE/dx is measured per track as the mean of the cluster charge properly weighted for the track length in silicon. Indeed the highest cluster charge associated to the track is excluded in the calculation of the mean, to reduce the Landau tails. The plot is based on about 180 000 tracks, i.e. about 10% of the collision data. |
eps version |
| *Pixel cluster charge*
Pixel cluster charge corrected for the path length, tracks with pT>0.1 GeV have been used. The black data points with errors are superimposed to the MC red line. |
eps version |
| Proton Mass Stability in Run 1
Proton mass calculated from the momentum measured in the Inner Detector and specific energy loss measured in Pixel Detector. The data, spanning from 2010 to 2012 and over 5 orders of magnitude in luminosity, are subdivided in periods of similar data taking conditions to illustrate the stability of the method. A typical run for each period is selected for the measurement. Vertical lines separate the years of data-taking. The reported values are the fitted peaks of the mass measurements for tracks with pt > 400 MeV /C, "good pixel clusters" >=2, nSCT >=6, d0GeV/C, dE/dx >1.9 MeV cm^2/g in a specific run. Only the statistical errors are shown. Clusters on the sensor edges or attached to shallow tracks are not considered as good pixel clusters.
The red horizontal line represents the nominal proton mass value and the gray horizontal lines show the +/- 1 % error band. The full scale of the plot corresponds to the 1 \sigma mass resolution (~12%). The plot demonstrates that the calibration of the dE/dx is stable at the 1 % level over data-taking conditions and detector settings. |
eps version |
| dE/dx during 25 ns bunch spacing run in 2012
Distribution of the pixel dE/dx in data (black) and Monte Carlo (full yellow) for tracks with |eta| < 2.5, at least 6 SCT associated hits (to assure track quality), at least 2 "good pixel clusters" (to properly measure the dE/dx). Both data and MC have bunch spacing of 25 ns. Clusters on the sensor edges or attached to shallow tracks are not considered as good pixel clusters.
Blue (red) line show the fit obtained with a Landau convoluted with a Gaussian function for data (MC). The ionization in MC is slightly larger: the Most Probable Value is 1.12 MeV cm^2/g in MC and 1.05 MeV cm^2/g in data. Data sample: group.detTindet.data12_8TeV.00216416.physics_MinBias.TrkD3PD.25nsReco.06 group.detTindet.mc12_8TeV.159000.ParticleGenerator_nu_E50.TrkD3PD.e1284_s1469_s1470_r4948.25nsReco.05 |
eps version |
| dE/dx during one of the 50 ns bunch spacing run in 2012
Distribution of the pixel dE/dx in data (black) and Monte Carlo (full yellow) for tracks with |eta| < 2.5, at least 6 SCT associated hits (to assure track quality), at least 2 "good pixel clusters" (to properly measure the dE/dx). Both data and MC have bunch spacing of 50 ns. Clusters on the sensor edges or attached to shallow tracks are not considered as good pixel clusters.
Blue (red) line show the fit obtained with a Landau convoluted with a Gaussian function for data (MC). The ionization in MC is slightly larger: the Most Probable Value is 1.12 MeV cm^2/g in MC and 1.08 MeV cm^2/g in data. Data Sample: group.detTindet.data12_8TeV.00*.physics_MinBias.TrkD3PD.50nsReco.06 group.detTindet.mc12_8TeV.159000.ParticleGenerator_nu_E50.TrkD3PD.e1284_s1469_s1470_r4844.50nsReco.05 |
eps version |
Beam background studies in Pixel
border=1 cellpadding=10 cellspacing=10>| *Pixel cluster row width (phi) with solenoid on / off*
The row width of pixel clusters in the phi direction is plotted for all clusters before tracking and for clusters associated with tracks, for two 2009 data runs in which solenoid magnetic field was on (run 142193) and off (run 141994). The histograms are normalized to the number of entries to give the fraction of pixel clusters for comparison. Fake clusters and clusters with ganged pixels are excluded to avoid edge effects. For all clusters before tracking, the phi width tail decreases when the solenoid is off. This effect is not apparent in the tail for clusters on tracks. The decrease is thought to be due to a lack of deflection in the absence of a magnetic field of low transverse momentum tracks, primarily from the beam background. The low cluster size region is expanded from the previous plot. At small cluster sizes, the phi width increases when the solenoid is off for all clusters and clusters on tracks. This is expected from the increased spread of the drifting charge carriers, which are no longer focused by the Lorentz angle effect in the absence of the magnetic field. |
eps version 
eps version |
| *Pixel cluster column width (eta), with solenoid on / off*
The column width of pixel clusters in the eta direction is plotted for all clusters before tracking and for clusters associated with tracks, for two 2009 data runs in which solenoid magnetic field was on (run 142193) and off (run 141994). The histograms are normalized to the number of entries to give the fraction of pixel clusters for comparison. Fake clusters and clusters with ganged pixels are excluded to avoid edge effects. For all clusters before tracking, the eta width tail increase when the solenoid is off. This effect is not apparent in the tail for clusters on tracks. The increase is thought to be from low transverse momentum tracks, primarily from the beam background, producing longer column width clusters in the barrel region in the absence of the magnetic field. At small cluster sizes, turning off the solenoid has little effect on the pixel cluster column width. This is expected because the solenoid magnetic field is parallel to the long pixel direction in the barrel, so there is no change in the spread of the drifting charge carriers in the eta direction, contrary to the case for the phi direction. |
eps version 
eps version |
| *Pixel cluster width correlations – Solenoid on, colliding bunches*
The correlations of the pixel cluster size column, the pixel row width (eta) and the pixel column width (eta) are plotted for all clusters before tracking for a 2009 data run in which solenoid magnetic field was on (run 142193). Fake clusters and clusters with ganged pixels are excluded to avoid edge effects. The plot includes all clusters before tracking, for colliding bunches only.
|
eps version b/w png version b/w eps version 
eps version b/w png version b/w eps version 
eps version b/w png version b/w eps version |
| *Pixel cluster width correlations – Solenoid on, non-colliding*
The correlations of the pixel cluster size column, the pixel row width (eta) and the pixel column width (eta) are plotted for all clusters before tracking for a 2009 data run in which solenoid magnetic field was on (run 142193). Fake clusters and clusters with ganged pixels are excluded to avoid edge effects. The plot includes all clusters before tracking, for non-colliding bunches only.
|
eps version b/w png version b/w eps version 
eps version b/w png version b/w eps version 
eps version b/w png version b/w eps version |
| *Pixel cluster width correlations for track clusters – Solenoid on, colliding bunches*
The correlations of the pixel cluster size column, the pixel row width (eta) and the pixel column width (eta) are plotted for clusters on tracks for a 2009 data run in which solenoid magnetic field was on (run 142193). Fake clusters and clusters with ganged pixels are excluded to avoid edge effects. The plot is for clusters associated with tracks, for colliding bunches only.
|
eps version b/w png version b/w eps version 
eps version b/w png version b/w eps version 
eps version b/w png version b/w eps version |
| *Pixel cluster width correlations – Solenoid off, colliding bunches*
The correlations of the pixel cluster size column, the pixel row width (eta) and the pixel column width (eta) are plotted for all clusters before tracking for a 2009 data run in which solenoid magnetic field was off (run 141994). Fake clusters and clusters with ganged pixels are excluded to avoid edge effects. The plot includes all clusters before tracking, for colliding bunches only.
|
eps version b/w png version b/w eps version 
eps version b/w png version b/w eps version 
eps version b/w png version b/w eps version |
| *Pixel cluster width correlations – Solenoid off, non-colliding*
The correlations of the pixel cluster size column, the pixel row width (eta) and the pixel column width (eta) are plotted for all clusters before tracking for a 2009 data run in which solenoid magnetic field was off (run 141994). Fake clusters and clusters with ganged pixels are excluded to avoid edge effects. The plot includes all clusters before tracking, for non-colliding bunches only.
|
eps version b/w png version b/w eps version 
eps version b/w png version b/w eps version 
eps version b/w png version b/w eps version |
| *Conclusions*
In the previous plots the correlations between the total pixel cluster size and the cluster widths in the eta (column) and phi (row) directions are plotted for various combinations of the conditions:
The plots show that most clusters contain a small number of pixels (< 10), whilst at larger cluster sizes, a distinct V-shape is revealed in the correlation between the pixel cluster eta and phi widths versus the total pixel cluster size. The phase space can be divided into two regions, thought to originate from collisions and from beam backgrounds (halo/gas), for the following reasons. The lower region of the V-shape is present for colliding bunches and is absent in non-colliding bunches. This effect is independent of the state of the solenoid magnetic field. The lower region of the V-shape is also present for clusters associated with tracks, reconstructed from collisions. The conclusion is that the lower region is predominantly from clusters associated with collisions. The upper region of the V-shape is present in both colliding and non-colliding bunches, suggesting a common source, such as beam background. The shape of the region for non- colliding bunches, was also found to match beam background Monte Carlo simulations. A comparison of the correlation of the cluster widths in the eta versus phi directions for the solenoid on and off, confirm the conclusions of the plots on slides 2 and 4. When the solenoid is switched off, the cluster eta width increases and the cluster phi width decreases, at large cluster sizes, due to the lack of deflection in the absence of a magnetic field of low transverse momentum tracks, primarily from the beam background. | |
| *Cluster eta width vs pseudorapidity, B field on*
The pixel cluster column width (eta) is plotted by Pixel barrel layer and end-caps for all clusters before tracking for a 2009 data run in which solenoid magnetic field was on (run 142913). Fake clusters and clusters with ganged pixels are excluded to avoid edge effects. The plot includes all clusters before tracking, for colliding bunches only. The distribution in eta is due to most tracks emerging from the collision point (+ constant from beam background). The slight asymmetry in global pseudorapidity is due to a displacement of interaction point by = -7mm. |
eps version |
| *Cluster width vs pseudorapidity, B field on, unpaired bunches*
The pixel cluster column width (eta) is plotted by Pixel barrel layer and end-caps for all clusters before tracking for a 2009 data run in which solenoid magnetic field was on (run 142913). Fake clusters and clusters with ganged pixels are excluded to avoid edge effects. The plot includes all clusters before tracking, for non-colliding bunches only. The distribution is flat in eta within errors, as expected from beam backgrounds. |
eps version |
Links
Major updates:
-- BeniaminoDiGirolamo - 21-Oct-2011 Responsible: Beniamino Di Girolamo - Pixel Project Leader
Last reviewed by: Never reviewed
Topic revision: r85 - 2022-07-05 - ClaudiaGemme
ATLAS 
Internal TWiki
Changes
Notifications
RSS Feed
Account 
Copyright &© 2008-2023 by the contributing authors. All material on this collaboration platform is the property of the contributing authors. or Ideas, requests, problems regarding TWiki? use Discourse or Send feedback