Cosmics and Calibration Approved Pixel Plots

Introduction

The Pixel detector integration, commissioning, cosmics data taking and calibration plots below are approved to be shown by ATLAS speakers at conferences and similar events. Procedure to approve plots is illustrated here.

• Figures come mainly from either the cosmic runs, from dedicated pixel calibration runs and from integration and installation of the IBL in 2014
• 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
• Plots on the Pixel performances with collisions can be found here

Please do not add figures on your own. Contact the Pixel Project Leader Martin Kocian or Claudia Gemme in case of questions and/or suggestions.

Performance Plots tracking from June 2015

Public Link	CDS record (ATLAS)	Full Title	Update Date	Contacts
PIX-2015-006	ATL-INDET-INT-2015-001	IBL Thermal stability during M9 runs	2015/05/12	Antonello Miucci
PIX-2015-008	ATL-COM-INDET-2015-091, ATL-INDET-INT-2016-008, ATL-INDET-INT-2016-009, ATL-COM-INDET-2016-034, ATL-INDET-INT-2016-012, ATL-INDET-INT-2016-018	Results from the ATLAS IBL low voltage current task force X-ray irradiation campaigns	2016/08/24	Malte Backhaus, Alessandro La Rosa, Karola Dette, Sultan DMS
PIX-2016-002	ATL-INDET-INT-2016-004	Leakage current vs fluence for planar and 3D sensors in IBL	2016/02/03	Nicholas Stuart Dann
PIX-2016-003	ATL-INDET-INT-2016-006	Pixel L1/L2 upgrade Optoscan in SR1	2016/02/18	Federico Meloni, Geoffrey Mullier
PIX-2016-005	ATL-INDET-INT-2016-003	IBL Calibration Plots and Drift of Calibration with Luminosity in 2015	2016/02/04	Antonello Miucci, Laura Jeantly
PIX-2016-006	ATL-INDET-INT-2016-014, ATL-INDET-INT-2016-025, ATL-INDET-INT-2017-010	Evolution of LV currents and sensor leakage currents in the IBL	2017/06/21	Nicholas Stuart Dann
PIX-2016-010	ATL-INDET-INT-2016-022	FEI4 LV current measurements - Data taken with Bern cyclotron	2016/09/01	Marco Rimoldi
PIX-2017-002	ATL-INDET-INT-2017-004	IBL Calibration Plots and Drift of Calibration with Luminosity in 2016	2017/02/16	Satoshi Higashino
IDTR-2017-002	ATL-INDET-INT-2017-008	Comparison Between Lorentz Angle at Beginning of Run1 and Run 2	2017/05/21	Fares Djama
PIX-2018-001	ATL-INDET-INT-2018-001	IBL construction paper	2018/02/24	Editors
PIX-2018-006	ATL-INDET-INT-2018-009	Study of threshold drop in FE-I4 by SEU	2018/06/20	Yohei Yamaguchi
PIX-2018-008	ATL-INDET-INT-2018-011, ATL-INDET-INT-2019-005, ATL-INDET-INT-2019-008	Measurements and predictions of Leakage Current	2019/03/26	Aidan Grummer, Jennet Dickinson
PIX-2018-009	ATL-INDET-INT-2018-003, ATL-INDET-INT-2019-010	IBL ToT & Threshold Calibration Evolution in 2017 and 2018	2019/06/06	Xiaotong Chu
PIX-2019-001	ATL-INDET-INT-2019-012 , ATL-INDET-INT-2019-016	IBL Leakage current and Depletion voltage in Run 2	2019/11/30	Jennet Dickinson, Ben Nachman

INDET and PHYS PUB NOTES

Reference	Full Title	Publication Date
ATL-PHYS-PUB-2015-012	Cluster Properties and Lorentz Angle Measurement in the 4-Layer Pixel Detector Using Cosmic Rays	2015/06/01
ATL-INDET-PUB-2014-006	ATLAS Pixel IBL: Stave Quality Assurance	2014/09/09
ATL-INDET-PUB-2014-004	A Leakage Current-based Measurement of the Radiation Damage in the ATLAS Pixel Detector	2014/08/27

Recent additional notes with approved plots

August 2014 Approved

• IBL Module Plots
  (Malte, ATL-INDET-INT-2014-003, Preliminary); Plots in the record
• IBL 3D Plots
  (Nanni, ATL-INDET-INT-2014-004, Preliminary); Plots below
• IBL stave test plots in pit
  (Karole Dette, ATL-INDET-2014-005, Preliminary) Plots below and here
• IBL Stave loading
  (Antonello, ATL-INDET-INT-2014-006, Preliminary); Plots in the below

• Pixel Module Failures
  (Kerstin, ATL-COM-INDET-2014-035, Preliminary) Plots in the record
• IBL connectivity test results
  (Javier, ATL-INDET-INT-2014-007, Preliminary) Plots here
• IBL Modules
  (Kazuki,ATL-COM-INDET-2014-046, Preliminary); Plots below
• IBL Stave QA
  (Jennifer, ATL-COM-INDET-2014-049, Preliminary) Plots below

June 2014 Approved

• IBL Stave QA selected plots
  (ATL-INDET-PUB-2014-002) Plots in the record and here (see link to the ATL-INDET-006)

Pixel detector and IBL 2015 Plots

border=1 cellpadding=10 cellspacing=10>

Data transmission efficiency of the optical to electrical transceivers (RX) designed for upgraded ATLAS pixel L1/L2 operation, shown as a function of the optical input power (renormalised to the nominal operating settings, corresponding to 1 on the x axis) and the RX signal discrimination thresholds. For each setting combination, the efficiency is defined as the ratio between the number of channels successfully receiving a 2000 bytes message in a loopback test run at a rate of 80 MHz and the total number of channels sending the message. Contact: Geoffrey Mullier, June 2015

eps, pdf version

Time over threshold (TOT) mean values vs injected charge for a FE-I4 module [LI_S18_C_M1_C2] of the SR1 system test. The mean values are obtained from a TOT scan, the fit is performed with a second order polynomial: TOT = p0(p1+Q)/(p2+Q) where Q is the charge, p1 and p2 are the two fit parameters. [The proportionality is not linear due to secondary order effects like time-walk and non constant discharge rate.] Contact: Maria Elena Stramaglia, June 2015

eps, pdf version

Pixel detector Yield and IBL commissioning Plots

3-layer Pixel detector: end of Run 1 and after reinstallation

border=1 cellpadding=10 cellspacing=10>

Number of disabled modules of the Pixel Detector at the end of Run1 classified by the type of the failure. (approved??) Contact: Kerstin Lantzsch, Beatrice Mandelli, June 2014 Reference: ATL-INDET-SLIDE-2014-388 internal	eps, pdf version
Number of modules of the pixel Detector to be disabled after refurbishment and re-installation in ATLAS (May 2014) during LS1 classified by failure mode (HV/ LV/ Data In/ Data Out) and the phase of causing problems (Run1/ Surface / After re-installation). Modules having issues but being operable are not included. Contact: Kerstin Lantzsch, Hideyuki Oide, June 2014 Reference: ATLAS-COM-CONF-2014-043 internal	eps, pdf version
Percentage of disabled modules at the end of Run 1 and after the re-insertion of Pixel Detector into the ATLAS Experiment for disk and three layers. Contact: Kerstin Lantzsch, Beatrice Mandelli, June 2014 Reference: ATL-INDET-SLIDE-2014-388 internal	eps, pdf version

IBL detector Module production plots

border=1 cellpadding=10 cellspacing=10>

Vbd measured on wafer at foundry on CNM and FBK tiles. For CNM the measurement is done by 3D guard ring and for FBK by means of temporary metals. All “green” tiles enter in the histogram. The bin at 100 V includes overflow of tiles having breakdown voltage higher than 100 V. Approval date: August 2014 Contact: Andrea Gaudiello, Nanni Darbo Reference: ATL-INDET-INT-2014-004 internal	eps, pdf version
Vbd measured on assembly level (ASSY, after bump-bonding), versus Vbd measured on the 3D guard ring at wafer level (before dicing) for a subset of CNM production. Measurements at assembly level were done with a current compliance of 100 µA. Approval date: August 2014 Contact: Ivan Lopez Paz, Nanni Darbo Reference: ATL-INDET-INT-2014-004 internal	pdf version
Vbd measured on assembly level (ASSY, after bump-bonding), versus Vbd measured on wafer at foundry on FBK tiles, using temporary metal. Only tiles later assembled in a module enter into the plot. Approval date: August 2014 Contact: Andrea Gaudiello, Nanni Darbo Reference: ATL-INDET-INT-2014-004 internal	eps, pdf version
Ratio of hit occupancy of the first and second row in a source scan for a module with a 3D FBK Sensor. The 90Sr beta source is placed at ~4 cm from the surface and 12 million events are recorded. The ratio provides and indication of the extra hits collected in the first row as consequence of charge collected outside the pixel geometrical area. This is a feature observed in FBK sensors also at test beam. CNM sensors do not have the same behaviour and the ratio is around 1. The result is consistent with measurements in the Stave QC and at test-beam Approval date: August 2014 Contact: Andrea Gaudiello, Nanni Darbo Reference: ATL-INDET-INT-2014-004 internal	eps, pdf version
A 3D module from FBK illuminated by a 90Sr source during the IBL Stave QC at CERN. SMD components and metal lines on the flex-hybrid are visible, shadowing the source. The module is A8-2 on stave ST11. Approval date: August 2014 Contact: Beatrice Mandelli, Nanni Darbo Reference: ATL-INDET-INT-2014-004 internal	eps, pdf version Zoom on top rows pdf version
Occupancy distribution for 3D sensor pixels in 90Sr source scan during the IBL stave QC at CERN. A module with 3D sensor from FBK and one from CNM are compared. Histograms are for pixel inside first/last row/column, called “normal” pixels, and or first/last rows or columns in case of FBK and for first/last rows and columns for CNM. Histogram are normalized to the same number of entries. The ratio for the mean values of the FBK external columns to normal pixel is 1.1±0.3, while for external row is 1.6±0.3. CNM Sensors do not show increase of counts in the periphery pixels. The staves were tuned to a threshold of 3000 e at 22◦C. Approval date: August 2014 Contact: Beatrice Mandelli, Nanni Darbo Reference: ATL-INDET-INT-2014-004 internal	pdf version
Cluster size as function of pseudo-rapidity. The comparison is between two 3D FBK modules. One irradiated to 6x10^15 neq/cm2 and the other un-irradiated. The cluster size is compared with what would be expected by “geometrical” assumption if charge is collected in the full thickness of the silicon substrate. The cluster analysis was not done using the tracking for the telescope: the alignment is not done yet due to difficulties encountered with noise and other systematic effects. The two detectors were operated with a threshold of 1600 e. The irradiated samples with a Vbias of -160 V and the un-irradiated with a Vbias of -20 V. Approval date: August 2014 Contact: Clara Nellist, Nanni Darbo Reference: ATL-INDET-INT-2014-004 internal	eps version, pdf version
Module Production Trends: from top to bottom for 3D Single Chip modules, IBL production only (up to batch 13) 3D Single Chip modules, Planar Double Chip modules, IBL production only (up to batch 12) Planar Double Chip modules. The Yellow line shows the number of modules accepted or lower quality sent to University of Geneva for loading. Contact: Kazuki Motohashi, Claudia Gemme, August 2014 Reference: ATL-COM-INDET-2014-046 internal	Single Chip Modules eps, pdf version Single Chip Modules Production only eps, pdf version Double Chip Modules eps, pdf version Double Chip Modules Production-only eps, pdf version
Yield of IBL module production for sensor types of Planar (PPS) and 3D (CNM, FBK) per production batch group. In the top panel for failure modules, “B.B. Fail.” stands for large bump-bonding failure, “Bare Fail.” stands for the module not assembled due to mainly mechanical damages, and “Other Fails.” stands for both electrical and sensor failures discovered after assembly. Within the same batch group, similar configuration of the laser condition and the bump bonding was applied. In the bottom panel, the first batch group was excluded from the average yield in the plot because they were largely affected by bumpbonding problems that were fixed in the following batches. The total average bad module fraction for PPS, CNM, FBK is 0.36, 0.50, 0.44, respectively. Contact: Claudia Gemme, Hideyuki Oide, June 2014 Reference: ATLAS-COM-CONF-2014-043 internal	eps, pdf version

IBL stave loading and QA plots

border=1 cellpadding=10 cellspacing=10>

Graphic showing the number of modules loaded, reworked and left after production. Modules are divided by technology. Contact: Antonello Miucci, August 2014 Reference: ATL-INDET-INT-2014-006	eps, pdf version
Graphic showing the numbers of modules for each category of rework. Reworks were mainly motivated by accidents during the loading or modules re-bonds and FE failures during the early testing. A minority where replaced after the stave QA due to some failing registers and reworked after a dedicated re-bond operation done at DSF laboratories at CERN. Contact: Antonello Miucci, August 2014 Reference: ATL-INDET-INT-2014-006	eps, pdf version
Number of modules replaced at stave level divided by category. Reworks were mainly motivated by failures during the loading or FE failures during the early testing. 2 modules were replaced after the stave QA due to some failing registers. The total number of loaded modules is 400 (240 planar + 72 FBK + 88 CNM) 6 modules replaced “after DSF” NOT included in the plot Contact: Tayfun Ince, Alessandro La Rosa, Beatrice Mandelli, June 2014 Reference: ATL-INDET-SLIDE-2014-388 internal	eps, pdf version
Residual values of the modules fiducial marks with respect to their nominal position for single chips (SC) and double chips (DC) in x and y directions, where the y axis is along the beam direction. Fourfiducial marks where measured per module when possible for the 20 produced staves. Contact: Javier Bilbao De Mendizabal, June 2014 Reference: ATL-INDET-INT-2014-002 internal	Single Chip Modules eps, pdf version Double Chip Modules eps, pdf version
Due to non-flatness of the stave surface, Planarity is defined as the maximal excursion of the stave out of its plane. Graphic on the top shows the planarity of the bare staves before and after the thermal cycle. The bottom one shows the difference between the measurements after the Thermal Cycle and before the Thermal Cycle. Data for stave 17 before the Thermal Cycle are unavailable. Contact: Antonello Miucci, August 2014 Reference: ATL-INDET-INT-2014-006	eps, pdf version
Graphic of temperature during the equipped staves in thermal cycle in the Loading QA. Contact: Javier Bilbao De Mendizabal, June 2015	eps, pdf version
Graphic showing the loading work flow at the loading site, quality assurance site and at CERN DSF laboratories. Contact: Antonello Miucci, August 2014 Reference: ATL-INDET-INT-2014-006	eps, pdf version
Example of a source scan hit map during the stave QA. Regions with a lower number of hits are clearly visible and match precisely the areas where passive components are mounted on the module flex PCB. Each normal pixel has around 150 to 200 hits while long pixels in the outer columns have nearly twice as many hits, which is expected due to their size. The difference in number of hits between working and not working pixels is large enough to readily identify dead and disconnected pixels. Many more plots of the stave QA are available in the document ATL-INDET-PUB-2014-002. Contact: Tayfun Ince, Beatrice Mandelli, June 2014 Reference: ATL-INDET-SLIDE-2014-388 internal	eps, pdf version
IV curve for all DCS groups on ST12. M4A and M4C are CNM sensors, the others are planar sensors. Many more plots of the stave QA are available in the document ATL-INDET-PUB-2014-002. Contact: Jennifer Jentzsch, August 2014 Reference: ATL-COM-INDET-2014-049 internal	eps, pdf version
Operational fraction of pixels in the $\eta$-$\phi$ plane for the 14 installed staves. Resolution: 128 bins in $\eta$ from -3.03 to 3.03 that correspond to a bin width of. 0.0473. 56 bins in $\phi$ from 0 to 2\,$\pi$ that correspond to a bin width of 0.112. Many more plots of the stave QA are available in the document ATL-INDET-PUB-2014-002. Contact: Jennifer Jentzsch, August 2014 Reference: ATL-COM-INDET-2014-049 internal	eps, pdf version
Threshold distribution of pixels on one chip before and after tuning for an example IBL chip for 3000~\e\ threshold tuning at 22$^{\circ}$C powering group temperature. A tuned threshold distribution typically has a standard deviation of less than 100~\e\ . Many more plots of the stave QA are available in the document ATL-INDET-PUB-2014-002. Contact: Jennifer Jentzsch, August 2014 Reference: ATL-COM-INDET-2014-049 internal	eps, pdf version

IBL Tests in SR1

border=1 cellpadding=10 cellspacing=10>

Time over threshold (TOT) mean values vs injected charge for a FE-I4 module [LI_S18_C_M1_C2] of the SR1 system test. The mean values are obtained from a TOT scan, the fit is performed with a second order polynomial: TOT = p0(p1+Q)/(p2+Q) where Q is the charge, p1 and p2 are the two fit parameters. [The proportionality is not linear due to secondary order effects like time-walk and non constant discharge rate.] Approval date: May 2015 Contact: Maria Elena Stramaglia

eps, pdf version

IBL Tests in the Pit

border=1 cellpadding=10 cellspacing=10>

a) Threshold chip-to-chip variation among the 14 IBL production staves. Data was taken with a configuration targeting 3000 e and a 10 ToT target response for 16000 e that was obtained during the QA at 20 C. The individual pixel data have first been averaged over each chip. The plots show, for each chip position on the stave, the mean and scatter of the 14 data points (one from each stave). The error bars show the RMS spread, while the solid boxes show the minimum and maximum values. b) Correspondent FE threshold mean values distribution for each technology. c) and d) show the difference of the values obtained in the RCE measurements and the QA. The shift to lower thresholds is caused by the lower module temperature compared to the temperature when the tuning was done. Contact: Karola Dette, August 2014 Reference: ATL-INDET-INT-2014-005 internal	eps, pdf version
a) Chip-to-chip variation of average Time over Threshold (ToT) in each pixel from injections of a 16000 e charge. ToT is measured in in units of bunch crossings, each of which represents 25 ns. Data was taken with a configuration targeting 3000 e and a 10 ToT target response for 16000 e that was obtained during the QA at 20oC. The plots show, for each chip position on the stave, the mean and scatter of the 14 data points (one from each stave). The error bars show the RMS spread, while the solid boxes show the minimum and maximum values. b) Correspondent ToT mean values distribution for each technology. c) and d) show the difference of the values obtained in the RCE measurements and the QA Contact: Karola Dette, August 2014 Reference: ATL-INDET-INT-2014-005 internal	eps, pdf version
a) Noise chip-to-chip variation among the 14 IBL production staves. Data was taken with a configuration targeting 3000 e and a 10 ToT target response for 16000 e that was obtained during the QA at 20oC. The individual pixel data have first been averaged over each chip. The plots show, for each chip position on the stave, the mean and scatter of the 14 data points (one from each stave). The error bars show the RMS spread, while the solid boxes show the minimum and maximum values. b) shows the difference of the values obtained in the RCE measurements and the QA. The higher noise on the A-side in the QA measurements was caused by a small noise on the HV line of the setup and the sensitivity of FBK modules, which were more frequently chosen for loading on A-side, to such noise. The noise on the outer 3D modules is generally higher than on the planar modules. Contact: Karola Dette, August 2014 Reference: ATL-INDET-INT-2014-005 internal	eps, pdf version
a) Threshold chip-to-chip variation among the 14 IBL production staves after retuning all pixels to a target threshold of 3000 e and to a 10 ToT target response for 16000 e. The individual pixel data have first been averaged over each chip. The plots show, for each chip position on the stave, the mean and scatter of the 14 data points (one from each stave). The error bars show the RMS spread, while the solid boxes show the minimum and maximum values. b) Correspondent FE threshold mean values distribution for each technology. c) and d) show the difference of the values obtained in the RCE measurements and the QA Contact: Karola Dette, August 2014 Reference: ATL-INDET-INT-2014-005 internal	eps, pdf version
a) Chip-to-chip variation of average Time over Threshold (ToT) in each pixel from injections of a 16000 e charge. ToT is measured in in units of bunch crossings, each of which represents 25 ns. The pixels were retuned to 3000 e target threshold and to a 10 ToT target response for 16000 e. The plots show, for each chip position on the stave, the mean and scatter of the 14 data points (one from each stave). The error bars show the RMS spread, while the solid boxes show the minimum and maximum values. b) Correspondent ToT mean values distribution for each technology. c) and d) show the difference of the values obtained in the RCE measurements and the QA Contact: Karola Dette, August 2014 Reference: ATL-INDET-INT-2014-005 internal	eps, pdf version
a) Noise chip-to-chip variation among the 14 IBL production staves after retuning all pixels to a target threshold of 3000 e and to a 10 ToT target response for 16000 e. The individual pixel data have first been averaged over each chip. The plots show, for each chip position on the stave, the mean and scatter of the 14 data points (one from each stave). The error bars show the RMS spread, while the solid boxes show the minimum and maximum values. b) shows the difference of the values obtained in the RCE measurements and the QA. The higher noise on the A-side in the QA measurements was caused by a small noise on the HV line of the setup and the sensitivity of FBK modules, which were more frequently chosen for loading on A-side, to such noise. The noise on the outer 3D modules is generally higher than on the planar modules. Contact: Karola Dette, August 2014 Reference: ATL-INDET-INT-2014-005 internal	eps, pdf version
Single channel measurement of delay (left Y-axis) and duty cycle (right-red Yaxis) depending on the delay-setting. Approval date: Summer 2014 Contact: Maria Elena Stramaglia	eps, pdf version
Distribution of the delay overall the almost complete set of tested channels. Approval date: Summer 2014 Contact: Maria Elena Stramaglia	eps, pdf version
Occupancy histogram generated by the FE emulator after 100000 triggers and 188 hits. Automatic hit generation. Approval date: Summer 2014 Contact: Maria Elena Stramaglia	eps, pdf version

Radiation damage plots

Depletion Voltage

border=1 cellpadding=10 cellspacing=10> <!--
<tr>
<td bgcolor=#eeeeee>
Evolution of the average depletion voltage with time. the inset on the top shows the average temperature during the period with clear indication of the high temperature period corresponding to short and long cooling stops.
</td><td align="center">
<img src="/twiki/pub/AtlasPublic/ApprovedPlotsPixel/timevsdeplvDeplV-July2012.png" alt="timevsdeplvDeplV-July2012.png" width='320' height='230' />


eps version
</td>
</tr>

<tr>
<td bgcolor=#eeeeee>
Evolution of the average depletion voltage with luminosity. the inset on the top shows the average temperature during the period with clear indication of the high temperature period corresponding to short and long cooling stops.
</td><td align="center">
<img src="/twiki/pub/AtlasPublic/ApprovedPlotsPixel/lumivsdeplvDeplV-July2012.png" alt="lumivsdeplvDeplV-July2012.png" width='320' height='230' />


eps version
</td>
</tr>
--> <!--
<tr>
<td bgcolor=#eeeeee>
Evolution of the depletion voltage with fluence for lower Z modules (inner) and higher Z modules (outer). Each point represents the measured depletion voltage for a single module at the given fluence. Fluence is larger for modules closer to the interaction point.
</td><td align="center">
<img src="/twiki/pub/AtlasPublic/ApprovedPlotsPixel/colorfluencevsdeplvallDeplV.png" alt="colorfluencevsdeplvallDeplV.png" width='320' height='230' />


eps version
</td>
</tr>
-->

After radiation damage. Plot of the crosstalk scan at the 11.05.2011 for a bias voltage of 46V. Module L2_B01_S1_A6_M2 is shown.	eps version
After radiation damage. Plot of the crosstalk scan at the 15.08.2011 for a bias voltage of 10V. Module L2_B01_S1_A6_M2 is shown	eps version
After radiation damage. Plot of the crosstalk scan at the 15.08.2011 for a bias voltage of 42V. Module L2_B01_S1_A6_M2 is shown.	eps version
Results of the depletion voltage scan showing the reduction of cross-talk while increasing the high voltage. An error function fit has been performed to estimate the depletion voltage. The depletion voltage is defined as 90% of the maximum value. Module L0_B08_S1_C7_M6C is shown.	eps version
Depletion Voltage distribution in several scans for all modules in the b-layer that have passed the module selection.	eps version
Result of a depletion voltage scan on 22.06.2011. An error function fit has been performed to estimate the depletion voltage. The depletion voltage is defined as 90% of the maximum value. Module L0_B08_S1_C7_M6C is shown.	eps version
Depletion Voltage distribution on 22.06.2011 for all modules in the b-layer that have passed the module selection.	eps version
Evolution of the depletion voltage with time. the inset on the top shows the average temperature during the period with clear indication of the high temperature period corresponding to short and long cooling stops. Each point represents the measured depletion voltage for a single module at the given time.	eps version
Evolution of the depletion voltage with luminosity. the inset on the top shows the average temperature during the period with clear indication of the high temperature period corresponding to short and long cooling stops. Each point represents the measured depletion voltage for a single module at the given luminosity.	eps version
Change of the depletion voltage as a function of the module position in Z. Fluence is larger for modules closer to the interaction point. Reduction in depletion voltage is slightly larger for inner modules	eps version
Depletion voltage residual for the b-Layer; view in 2D (module position and stave number). Fluence is larger for modules closer to the interaction point. Reduction in depletion voltage is slightly larger for inner modules and as seen most of the modules of the b-Layer are already type inverted (July 2012, ~12 fb-1).	eps version

Depletion Voltage updates on Set 2014

border=1 cellpadding=10 cellspacing=10>

Evolution of the average depletion voltage with time. the inset on the top shows the average temperature during the period with clear indication of the high temperature period corresponding to short and long cooling stops. Contact: Andre Lukas Schorlemmer, September 2014 Reference: ATL-COM-INDET-2014-062 internal	eps, pdf version
Evolution of the average depletion voltage with luminosity. the inset on the top shows the average temperature during the period with clear indication of the high temperature period corresponding to short and long cooling stops. Contact: Andre Lukas Schorlemmer, September 2014 Reference: ATL-COM-INDET-2014-062 internal	eps, pdf version
Evolution of the depletion voltage with fluence for lower Z modules (inner) and higher Z modules (outer). Each point represents the measured depletion voltage for a single module at the given fluence. Fluence is larger for modules closer to the interaction point. Contact: Andre Lukas Schorlemmer, September 2014 Reference: ATL-COM-INDET-2014-062 internal	eps, pdf version
Mean value of the effective depletion voltage as a function of the 1 MeV neutron equivalent fluence. Measurements for all three detector layers are shown. The model prediction is superimposed for each layer separately. Contact: Andre Lukas Schorlemmer, September 2014 Reference: ATL-COM-INDET-2014-062 internal	eps, pdf version

PP0 level High Voltage Current Monitoring

border=1 cellpadding=10 cellspacing=10>

HV current monitoring plot vs. date in linear scale. Plot of the averaged high voltage current for all modules in the different Barrel layers as a function of time. Linear scale. Current measured by the ISEG power supplies and constantly monitored by DCS. Leakage current evolves with time; steps consistent with annealing during detector warm-up periods. Prediction is based on delivered luminosity, the expected fluence by barrel layer from Phojet + FLUKA simulations, and the “Dortmund” model [O. Krasel thesis]. Qualitatively very good agreement; however, the prediction underestimates the data and had to be scaled up: Layer 0 +15%, Layer 1 and 2 +25%.	eps version
HV current monitoring plot vs. date in log scale. Plot of the averaged high voltage current for all modules in the different Barrel layers as a function of time. Logarithmic scale. Current measured by the ISEG power supplies and constantly monitored by DCS. WARNING: updated only until December 2011	eps version
HV current monitoring plot vs. integrated luminosity (1/pb) in linear scale. Plot of the averaged high voltage current for all modules in the different Barrel layers as a function of the integrated luminosity in 1/pb. Linear scale. Current measured by the ISEG power supplies and constantly monitored by DCS. Leakage current follows integrated luminosity; steps consistent with annealing during detector warm-up periods. Prediction is based on delivered luminosity, the expected fluence by barrel layer from Phojet + FLUKA simulations, and the “Dortmund” model [O. Krasel thesis]. Qualitatively very good agreement; however, the prediction underestimates the data and had to be scaled up: Layer 0 +15%, Layer 1 and 2 +25%.	eps version
HV current monitoring plot vs. fluence in linear scale. Plot of the averaged high voltage current for all modules in the different Barrel layers as a function of the fluence. Linear scale. Current measured by the ISEG power supplies and constantly monitored by DCS.	eps version

Module level High Voltage Current Monitoring with CMBs

border=1 cellpadding=10 cellspacing=10> <!-- COMMENTED BY CG ON 12/8/2014 on request by Rui
<tr>
<td bgcolor=#eeeeee>
HV current monitoring plot vs. date in linear scale average data. Plot of the the average high voltage currents measured using the specialized Current Monitoring Boards at HVPP4 for all modules already equipped. Linear scale. Current measured by the CMB boards and constantly monitored by DCS. The model used is the same as for similar plots obtained using the ISEG current information.
</td><td align="center">
<img src="/twiki/pub/AtlasPublic/ApprovedPlotsPixel/hvpp4_CorCur_0_1new.png" alt="hvpp4_CorCur_0_1new.png" width='600' height='385' />


eps version
</td>
</tr>
<tr>
<td bgcolor=#eeeeee>
HV current monitoring plot vs. integrated luminosity in linear scale average data. Plot of the the average high voltage currents measured using the specialized Current Monitoring Boards at HVPP4 for all modules already equipped. Linear scale. Current measured by the CMB boards and constantly monitored by DCS. The model used is the same as for similar plots obtained using the ISEG current information.
</td><td align="center">
<img src="/twiki/pub/AtlasPublic/ApprovedPlotsPixel/hvpp4_CorCur_lumi_0_2new.png" alt="hvpp4_CorCur_lumi_0_2new.png" width='600' height='385' />


eps version
</td>
</tr>
END COMMENTED BY CG ON 12/8/2014 on request by Rui
-->

HV current monitoring plot vs. integrated luminosity in linear scale average data. The plot shows the average of the high voltage currents measured using the specialized Current Monitoring Boards at HVPP4 over all modules equipped. Annealing periods can be seen clearly. Currents are corrected to Tref = 0 C, band gap Eg = 1.21eV. Includes beam introduced ionization current correction. Currents match with the Hamburg Model prediction [Olaf Krasel, Univ. Hamburg Ph.D. dissertation, 2004]. Approval date: August 2014 Contact: Rui Wang Reference: ATL-INDET-SLIDE-2014-512 internal

Total eps version Detailed in phi eps version Detailed in eta eps version Temperature Profile for B-layer modules eps version

Pixel level High Voltage Current Monitoring with Monleak scans

border=1 cellpadding=10 cellspacing=10>

Increase of the pixel leakage current for all pixels in the b-layer. The increase is measured taking as a reference a scan taken when the total delivered luminosity was 1 pb-1. The measured points are fitted to a gaussian to obtain the average increase in the current, and the errors reported are the fit errors (no systematic uncertainties are included).

eps version

Average increase of the pixel leakage current for all b-layer pixels as a function of the total integrated luminosity. Every point corresponds to a single scan, and is obtained by fitting to a gaussian the pixel-by-pixel current difference using as a reference a scan taken when the total delivered luminosity was 1 pb-1. Systematic errors are not shown and fit errors are too small to be appreciated in the plot's scale.

eps version

Module level High Voltage Current Monitoring with Monleak scans

border=1 cellpadding=10 cellspacing=10>

Increase of the module leakage current for all modules in the b-layer. The increase is measured taking as a reference a scan taken when the total delivered luminosity was 1 pb-1. The measured points are fitted to a gaussian to obtain the average increase in the current, and the errors reported are the fit errors (no systematic uncertainties are included).	eps version
Increase of the module leakage current for all modules in the Layer 1. The increase is measured taking as a reference a scan taken when the total delivered luminosity was 1 pb-1. The measured points are fitted to a gaussian to obtain the average increase in the current, and the errors reported are the fit errors (no systematic uncertainties are included).	eps version
Increase of the module leakage current for all modules in the Layer 2. The increase is measured taking as a reference a scan taken when the total delivered luminosity was 1 pb-1. The measured points are fitted to a gaussian to obtain the average increase in the current, and the errors reported are the fit errors (no systematic uncertainties are included).	eps version
Increase of the module leakage current for all modules in the Disks. The increase is measured taking as a reference a scan taken when the total delivered luminosity was 1 pb-1. The measured points are fitted to a gaussian to obtain the average increase in the current, and the errors reported are the fit errors (no systematic uncertainties are included).	eps version
Average increase of the module leakage current for all the different detector layers as a function of the total integrated luminosity. Every point corresponds to a single scan, and is obtained by fitting to a gaussian the module-by-module current difference using as a reference a scan taken when the total delivered luminosity was 1 pb-1. Systematic errors are not shown and fit errors are too small to be appreciated in the plot's scale.	eps version

Threshold setting at 3500 e used since the start of data taking with LHC

border=1 cellpadding=10 cellspacing=10>

Noise, threshold and threshold / noise for normal (red), ganged (green) and long and interganged pixels (blue). The noise is typically about 170 electrons while the threshold is 3500 electrons. The threshold / noise is about 25 for all pixels apart from the ganaged pixels where it is about 10. Ganged pixels are basically a connection of two pixels within one readout channel resulting thus in about twice the noise of a single "normal" pixel.

eps version linear scale png version linear scale eps version eps version linear scale png version linear scale eps version eps version linear scale png version linear scale eps version

Noise, threshold and threshold / noise for all pixels.

eps version eps version eps version

In-time Threshold setting at 3500 e with hit doubling

border=1 cellpadding=10 cellspacing=10>

In-time Threshold for all pixels when the hit doubling feature is enabled. The resulting in time threshold is reduced from 4800 e to 3700 e for a set threshold of 3500 e.

eps version

In-time Threshold setting at 3500 e with no hit doubling (therefore for pure documentation being obsolete)

border=1 cellpadding=10 cellspacing=10>

In-time Noise, threshold and threshold / noise for normal (red), ganged (green) and long and interganged pixels (blue). These plots are before applying the hit doubling setting in the front-end chip, therefore the in time threshold value is around 4800 e. These figures are for reference, but they are obsolete as we are using hit doubling. They may become actual in a far future if we would have to switch off the hit doubling feature due to very high occupancy

eps version linear scale png version linear scale eps version eps version linear scale png version linear scale eps version

Noise, threshold and threshold / noise for all pixels. The noise is typically about 170 electrons while the threshold is 3500 electrons. The threshold / noise is about 25 for all pixels.

eps version eps version

Figures summarizing the detector performance versus time in the 2008 and 2009 data taking

border=1 cellpadding=10 cellspacing=10>

The integrated number of tracks with at least one pixel hit as function of day for the running between September 14th and December 1st 2008. Shown are both the data with solenoid on and off. Between October 22nd and November 26th no cosmic ray data were taken and a full n ew calibration of the detector was performed which was then used end of November. The number of tracks with at least 4 pixel hits is about a factor 2 lower than this. The total number of tracks is about 240,000 without magnetic field and 190,000 with magnetic field. The eps files for 2008 and 2009 data taking are also available.
Fraction of modules that was enabled during cosmics data taking 2008 and 2009. The main features in 2008 are: At day 20 three cooling loops (corresponding to 36 modules) in the disks were turned off Before day 70 a new tuning of the optical links was performed and some new TX plugins were installed resulting together in the recovery of about 3% of the modules. In 2009 the previously leaky loops were recovered and the fraction od disabled modules seen is mostly due to the inoperational TX plugins that were then recovered in August 2009. The eps files for 2008 and 2009 are also available.
The rate of tracks with at least one pixel hit as function of day for the 2008 and the 2009 running. In 2008 the rate was increased substantially in early October, mostly due to significant improvements in the muon (RPC) trigger timing and due to using track triggers at L2. In November a new L1 trigger of the TRT was developed which improved the rate further to about 0.5Hz which is consistent with the rate of pixels crossing the detector at all. In the beginning of the 2009 data taking the track rate was incorrectly calculated in the online monitoring which is reflected in the plot.
The number of noise hits in the entire Pixel detector per event per bunch crossing for the 2009 data taking before and after applying the offline noise mask. It can be seen that the noise was very stable.	n/a

General Figures characterising the performance of the detector

border=1 cellpadding=10 cellspacing=10>

Display of cosmic ray event going through the pixel detector taken on October 18th 2008. Shown are the XY view (of SCT and pixels and of pixels alone) and an RZ view. The track has a hit in each of the layers in both the upper and the lower hemisphere. In the bottom of L0 there are even two hits due to a module overlap. Apart from the signal hits there is only one other hit in the pixel detector demonstrating the very low noise level in the detector. eps file
The cluster charge in the pixel detector per 250 microns using the 2008 cosmic ray data taken with the solenoid on (top) and off (below). This was was obtained by scaling the raw cluster charge by the theoretical dependence on the propagation distance through silicon. Clusters containing ganged pixels, more than one pixel in the long pixel direction or containing pixels at the edge of the module were excluded. Clusters are required to be part of a reconstructed track, the only reconstructed cluster in the module and only contain 2 pixels in the short pixel direction. BOn eps file BOff eps file
The width of the cluster versus the incidence angle of the cluster in azimuth for pixel clusters on tracks. Shown are both the data for the solenoid off and the solenoid on. The minimum of this distribution determines the Lorentz angle and the fit result is shown. As expected the Lorentz angle is consistent with 0 for the data without magnetic field. Using the data with magnetic field the preliminary measurement of the Lorentz angle is 205 +/- 0.5 mrad. The expectation is about 225 mrad. Lorentz eps version
The fraction of pixels that are masked due to high noise. Shown is the fraction masked online and offline as function of run number during the cosmic ray data taking in Fall 2008. It can be seen that the fraction of these problematic pixels is well below 0.02%. The spike in the middle at run 90250 is due to different calibration settings for that run. eps file
The occupancy of pixels in a sample with no tracks in the pixel detector which measures the noise occupancy in the pixel detector in each of the three barrels and the disks after the noisiest pixels are masked (see above). The noise occupancy is about 10^-10. This is determined using the officially reprocessed data with the final noise masking. eps file
Time over Threshold (ToT) versus injected charge for the average of all pixels from one frontend showing a nearly linear relationship. Also shown is the parameterization used for the calibration. eps file
The efficiency for attaching hits on tracks in active modules in the pixel barrels. It is about 99.8% in all three layers using the official alignment of ATLAS. This plot is based on cosmic ray data. Disabled modules are not considered for the denominator of the efficiency definition.
Fraction of disconnected bump bonds per module. On average about 0.1% of the bump bonds are disconnected per module. eps file
Pixel module timing: shown is average cluster arrival time for each module in units of bunch crossings. 8 BC were read out during the cosmic ray data taking. The clusters have been selected as ToT >15 to reduce time walk. The timing distribution peaks at 3 BC and it is seen that all modules are timed in within +/- 1 bunch crossing. It is planned to reduce the readout window to 5 BC for the initial collision data taking and ultimately to 1 BC at LHC design luminosity. eps file
ToT distribution for pixels with an injected charge of 20,000 electrons with tuning performed during module production (blue) and the new tuning in the ATLAS Pit (red). After tuning the mean value is about 30 with a resolution of 0.65.
Noise, theshold and threshold / noise for normal (red), ganged (green) and long and interganged pixels (blue). The noise is typically about 200 electrons while the threshold is 4000 electrons. The threshold / noise is about 25 for all pixels apart from the ganaged pixels where it is about 10. Ganged pixels are basically a connection of two pixels within one readout channel resulting thus in about twice the noise of a single "normal" pixel. Black-and-white versions as well as eps files for all plots are available in the table of attachments.
Most Probable Value (left) and FWHM (right) of the corrected charge distribution for pixel clusters on tracks as a function of the track momentum obtained for cosmics data and simulation. The Distributions are fitted with a Landau convoluted with a Gaussian function. The cluster charge has been corrected for the track length in the Silicon sensor which is 250 um thick. Tracks with spatial incident angle <0.93 with respect to the normal to the module surface have been selected. The increase of the MPV in data (Montecarlo) is 7.5% (7.2%) compatible with the theoretical increase expectation of ~8.5% in the same momentum range for muons . The Ratio of Data/simulation is ~ 98.5% for the Most Probable Value and 95% for the FWHM: thus the released charge in data is a bit lower than in Montecarlo. The reason for this offset can be either due to the non-perfect calibration of the injection circuitry used for determining the threshold and ToT relationship or an inadequate of the energy loss model implemented in simulation. Here is the eps file for writeups.

Detailed Figures mostly suggested for detailed Pixel Detector specific presentation

border=1 cellpadding=10 cellspacing=10>

Before radiation damage. HV scan on L2_B01_S1_A6_M2A. ( eps version)
HV distribution histo. Caption provided soon. ( eps version)
Xtalk 2D histo. Caption provided soon. ( eps version)
Noise 2D histo. Caption provided soon. ( eps version)
Measurement of the leakage current in nA measurement for the two different pixel bump bonds, AMS and IZM. There is no leakage current yet present for the overwhelming majority of the pixels as expected. There is a small tail towards large values which is different for the two different productions (AMS and IZM) which is not yet understood. At the end of the nominal LHC running the leakage current is expected to grow up to 25 nA per pixel for an operating temperature of -5C.
Charge Sharing calibration: the cluster position is determined by the charge weighted average of the pixel hits contributing to the cluster. Here the pixel residual is shown assuming a "digital readout" without taking the charge measurement into account. A clear dependence on the relative charge sharing of the 2 pixels is observed. The offset and slope of this is determined for each module with sufficient statistics and then used to determine the cluster position. This is known as the "analog" resolution.
Comparison of the residuals in eta direction (non-precision direction) for all clusters and for those at large eta with ("analog") and without ("digital") the charge sharing calibration. A clear improvement is observed due to the calibration in particular at high eta.
Comparison of the residuals in phi direction (the precision direction) for all clusters and for those at large eta with ("analog") and without ("digital") the charge sharing calibration. A small improvement is observed due to the calibration. However, the residual width is still too broad due to residual misalignments still present in the data so that the expected improvement due to using the analog measurements is not yet visible.
Time walk: Delay of the measured time of the pixel hit versus charge for one example module using calibration data. At high charge the asymptotic value of the delay is about 53 ns for this module. At low charge the hit arrives later as seen by the smaller delay due to a longer rise time of the frontend preamplifier. This is determined for each module and each pixel type from a dedicated calibration, and the timing can then be optimized to collect the maximal charge within one bunch crossing.
Map of pixel column number versus row number for the module with the largest fraction of disconnected bump bonds. The disconnected bump bonds are shown in blue and are concentrated in the frontends on the left. eps file
Column number versus row number of disconnected bump bonds integrated over all modules. It is seen that the disconnects mostly occur at the boundaries of the frontends. The module above is again seens as a large area on the right. No other large regions are seen. eps file
The measured Lorentz Angle versus the temperature of the Pixel Detector Module. A dependence of 0.78+/-0.18 mrad/K observed is consistent with the expectation of -0.74 mrad/K. 2008 data version
The measured Lorentz Angle versus the temperature of the Pixel Detector Module. A dependence of -0.74+/-0.06 mrad/K observed is consistent with the expectation of -0.74 mrad/K. 2009 data version with more points in temperature and therefore higher precision. ( eps version)
Residual between the track incidence angle and the pixel cluster position in the precision direction (local x) versus the z position of the hit for two example staves. A stave consists of 13 modules aligned in the z-direction and is a separate mechanical entity. E.g. in the B-layer 22 staves make up the full barrel. It is seen that there is no significant dependence on z in the first example but there is a significant bow with a sagitta of 500 microns in the 2nd example. These bows have been corrected for in the alignment procedure. The plots here are shown before the in situ alignment is performed but after survey constants are applied. The surveys were not sensitive to these bows.
Time over Threshold (ToT) versus injected charge for all pixels. Note, that the overwhelming majority (in red area) of the pixels shows a near linear relationship. There is a small fraction of pixels that contribute to the tail (in blue area). eps file
The MPV of the pixel cluster charge as a function of the propagation distance through silicon using the 2008 cosmic ray data taken with the solenoidal magnetic field on. The MPV is determined, in bins for propagation distance, from a fit of a convolution of a Landau and a Gaussian.The fit is performed separately on clusters containing either 1, 2 or 3 pixels. The number of pixels refers to the cluster size in the short pixel direction. Clusters containing ganged pixels, more than one pixel in the long pixel direction or containing pixels at the edge of the module were excluded. Clusters are required to be part of a reconstructed track with p > 5 GeV and the only reconstructed cluster in the module. The data is shown with solid points and the simulation with open points.The black lines indicate the results of a fit to theoretical expectations of the charge with a single free parameter, an overall scale factor C, that allows the normalization to float. If C has a value of 1, then then the charge observed is consistent with expectations. The systematic uncertainties on C are ~3%. eps file
The MPV of the pixel cluster charge as a function of the propagation distance through silicon using the 2008 cosmic ray data taken with the solenoidal magnetic field off. The MPV is determined, in bins for propagation distance, from a fit of a convolution of a Landau and a Gaussian.The fit is performed separately on clusters containing either 1, 2 or 3 pixels. The number of pixels refers to the cluster size in the short pixel direction. Clusters containing ganged pixels, more than one pixel in the long pixel direction or containing pixels at the edge of the module were excluded. Clusters are required to be part of a reconstructed track and the only reconstructed cluster in the module. The data is shown with solid points and the simulation with open points.The black lines indicate the results of a fit to theoretical expectations of the charge with a single free parameter, an overall scale factor C, that allows the normalization to float. If C has a value of 1, then then the charge observed is consistent with expectations. The systematic uncertainties on C are ~3%. eps file
Examples of the fits to the cluster charge for clusters containing 2 pixels performed in a certain bin of track propagation distance. The range of propagation distance is indicated on each plot. The fit has four parameters: the MPV of the Landau distribution (MPV_L), the width of the Landua (sigma_L), the width of the Gaussian (sigma_G) and a normalisation constant (N). The MPV of the convolution is obtained numerically from these 4 parameters. eps file eps file
The number of pixel clusters as a function of the incidence angle of the track with respect to the normal vector to the module surface using the 2008 cosmic ray data taken with the solenoidal magnetic field on. The number of clusters is shown for all cluster sizes (black) and clusters containing 1, 2 or 3 pixels. The number of pixels refers to the cluster size in the short pixel direction. Clusters containing ganged pixels, more than one pixel in the long pixel direction or containing pixels at the edge of the module were excluded. Clusters are required to be part of a reconstructed track with p > 5 GeV and the only reconstructed cluster in the module. The data is shown with solid points and the simulation with open points.For this range of incidence angles, the agreement between the data and simulation is excellent. eps file
The number of pixel clusters as a function of the incidence angle of the track with respect to the normal vector to the module surface using the 2008 cosmic ray data taken with the solenoidal magnetic field off. The number of clusters is shown for all cluster sizes (black) and clusters containing 1, 2 or 3 pixels. The number of pixels refers to the cluster size in the short pixel direction. Clusters containing ganged pixels, more than one pixel in the long pixel direction or containing pixels at the edge of the module were excluded. Clusters are required to be part of a reconstructed track and the only reconstructed cluster in the module. The data is shown with solid points and the simulation with open points. For this range of incidence angles, the agreement between the data and simulation is excellent. eps file
The average wafer thickness measured using a wafer gauge instrument as a function of the distance from the centre of the pixel sensor wafer. Results are shown separately for the two different sensor vendors: ON and CiS. The results of a fit with a r^4 dependence is shown. This parametrisation was used to determine values for the average sensor thickness. eps file

Links

• Pixel main page

Major updates:
-- BeniaminoDiGirolamo - 20-May-2010 -- BeniaminoDiGirolamo - 01-Jun-2011 -- BeniaminoDiGirolamo - 18-Oct-2011 -- BeniaminoDiGirolamo - 07-May-2012

Responsible: Pixel Project Leader (Beniamino di Girolamo)
Last reviewed by: Never reviewed