CMS-DP-2018/015

CMS ECAL Performance in 2017

Abstract: CMS ECAL and ES alignment, calibration and performance in 2017. EG L1 performance in 2017. Noise projections for LHC Run3.

CDS entry

iCMS entry


Figure CaptionSorted ascending
pdf version
Timing History 2017 abs Final2.png
Average timing evolution during the 2017 data taking for the positive and the negative sides of ECAL Barrel and Endcaps. A clear evolution is visible, showing a mild recovery of the timing when the detector is not irradiated. Timing variations within 200 ps have a negligible impact on the reconstruction, therefore during the 2017 data taking the timing conditions used in the reconstruction were updated after 200 ps shifts.
pdf version
meeincat0011 EBhighR9 final2.png
pdf version
meeincat0011 EEhighR9 final2.png
Dielectron invariant mass distribution for the 2017 data taking period using Z→ee electrons in the ECAL barrel and endcaps. Only electrons with low bremsstrahlung (R9>0.94) are considered. A dedicated re-calibration using the full 2017 dataset was performed. The histograms show the distributions of di-electron invariant mass in data with and without the re-calibration applied.
pdf version
IC precision EB Zee EoP pi0 2.png
ECAL crystals intercalibration using the data collected in 2017. Precision of the channel inter-calibration, using energy deposits, as a function of the pseudo-rapidity η in the ECAL barrel detectors. The precision for measuring the inter-calibration constants from Z→ee, π→ γγ decays, and electrons arising from W and Z boson decays compared to the tracker response (E/p), is shown as a function of η in EB using 2017 data. The precision of the Z→ee and photon inter-calibrations is at the level of the systematic errors. The precision of the E/p inter-calibrations is still dominated by the statistical errors for η> 1. The black points represent the precision of the combination of the three methods (weighted average). Reference: JINST 8 (2013) P09009
pdf version
idark vs rmsg12 fit.png
Every ECAL Barrel crystal is equipped with a pair of Avalanche Photodiodes (APDs). The plot shows the measurement of the ECAL electronic noise in the most sensitive amplification range (gain=12) as a function of the measured leakage current of an APD pair. The points below 10uA are measured in CMS in situ, while the points at higher current are based on lab measurement of APDs irradiated with neutrons at different fluences, as indicated in the figure. The pink line is a fit to the data with a square root function.
pdf version
idark prediction vs lumi.png
Every ECAL Barrel crystal is equipped with a pair of Avalanche Photodiodes (APDs). The plot shows the prediction of the leakage current in the ECAL Barrel APDs as a function of delivered luminosity. The leakage current increases due to the hadron fluence.The blue curve is the prediction for the central region of the ECAL barrel (eta=0), while the red curve is for the ECAL barrel border (eta=1.45).The prediction is for an operational temperature of 18 degrees. The vertical grey bands indicate the Long Shutdowns. The model uses the fit of the APD leakage current measurement in situ vs time and measurements of the delivered luminosity and it includes a permanent damage component and a slow and a fast recovering components. The extrapolation assumes 68fb-1 of delivered luminosity in 2018 and 80 fb-1 in each of the three years of Run3: 2021-2022-2023
pdf version
pi0MassEBxtal 30003 final.png
pdf version
pi0MassEExtal 8155 final.png
Examples of the invariant mass of photon pairs with one photon in one fixed crystal of the ECAL Barrel at η=0.03 (top plot), and of the ECAL Endcap at η=1.8 (bottom plot), in the mass range of the π0. Data collected in 2017 and corresponding to an integrated luminosity of approximately 9.8 fb-1 are used. These events are collected by CMS with a dedicated trigger at a rate of 7 (2) kHz in the Barrel (Endcap). The high trigger rate is made possible by a special clustering algorithm that saves only a minimal amount of information of the events, in particular energy deposits in the ECAL crystals surrounding a possible π0 candidate. For the candidates in the Endcaps in the region covered by the Preshower (ES, 1.7<η<2.55), information from the ES is also used.
These events are used as prompt feedback to monitor the effectiveness of the laser monitoring calibration and to inter-calibrate the energy of ECAL crystals.
The energy scale is uncalibrated for these π0 invariant mass plots.
pdf version
spikeratint 306456 oldped final.png
pdf version
Table1.png
This plot shows the fraction of trigger primitives (TPs), above a given transverse energy (ET) threshold, that are due to spikes, from Run 306456, with a peak pileup of 59. It is the ratio of the two plots shown before. It shows that the rate of mis-identified spikes becomes more important at larger ET. The table below shows the fraction of TPs due to mis-identified spikes, for several representative ET thresholds.
pdf version
spikeratint 306459 oldped final.png
pdf version
Table2.png
This plot shows the fraction of trigger primitives (TPs), above a given transverse energy (ET) threshold, that are due to spikes, for a lower pileup run (306459) in the same fill, with average pileup ranging from 15-40. The table below shows the fraction of TPs due to mis-identified spikes, for several representative ET thresholds. The values are slightly lower than for Run 306456, indicating a mild pileup dependence in the Level-1 spike-killer efficiency
pdf version
Run2017F 306456 spikerat ped final.png
pdf version
Table3.png
This plot shows the fraction of trigger primitives (TPs), above a given transverse energy (ET) threshold, that are due to spikes, from run 306456, recorded in late 2017, with a peak pileup of 59. The two curves show the residual spike fraction estimated for two sets of pedestal values. The pedestals have been observed to drift with time, and they are explicitly used in the L1 spike killer.
The spike contamination represented by the continuous line uses pedestals that were recorded at the end of 2016, from Run 296917. These were used in the L1 spike killer for Run 306456.
The spike contamination represented by the dashed line uses pedestals from the end of 2017, from Run 305848. The trigger primitives were re-emulated using these new pedestals.
The table below shows the fraction of TPs due to mis-identified spikes, for several representative ET thresholds: it improves when the pedestals are more up-to-date.
pdf version
Run2017F 306459 spikerat ped final.png
pdf version
Table4.png
This plot shows the fraction of trigger primitives (TPs), above a given transverse energy (ET) threshold, that are due to spikes, for a lower PU run (306456), with a peak pileup of 59. The two curves show the residual spike fraction estimated for the same two sets of pedestal values: 2016 pedestals from run 296917, 2017 pedestals from Run 305848. The table below shows the fraction of TPs due to mis-identified spikes, for several representative ET thresholds. It improves when the pedestals are more up-to-date. The pedestals are observed to drift with time. The improvement is similar to the higher PU case.
pdf version
noise prediction vs lumi.png
Prediction of the noise in the ECAL Barrel electronics as a function of the integrated luminosity. The noise increases due to the radiation-induced increase of the APD leakage current. The blue curve is the prediction for the central region of the ECAL barrel (eta=0), while the red curve is for the ECAL barrel border (eta=1.45). The prediction is for an operational temperature of 18 degrees. The vertical grey bands indicate the Long Shutdowns. The model for the prediction of the noise uses the fit of the noise vs leakage current, based on the measurement in situ and on lab measurements of few heavily irradiated APDs.The model also uses the fit of the APD leakage current based on the leakage current measurement in situ vs luminosity. The extrapolation assumes 68fb-1 of delivered luminosity in 2018 and 80 fb-1 in each of the three years of Run3: 2021-2022-2023. The ECAL electronics is equipped with a sampling ADC and 10 samples are retained per bunch crossing. The plot shows the RMS of the fluctuations of the single sample for the most accurate gain range of the electronics (gain=12). The ECAL pulse amplitude reconstruction uses a fitting technique exploiting all 10 samples and the pulse amplitude is affected by a noise which is about 50% of the fluctuation of the RMS of the single sample shown in this plot. The ADC scale is converted to energy through a calibration constant, and one ADC count corresponds approximately to 40 MeV.
pdf version
cDEta EB final.png
pdf version
cDEta EE final.png
Relative alignment of the ECAL crystals and the CMS tracker, measured using electrons and positrons from Z→ee events, separately for the ECAL barrel and endcaps. The plots show the residual difference in Δη between the position of the ECAL supercluster and the extrapolated track position, using the point of closest approach to the supercluster. The distribution of Δη is shown for data before (red triangles) and after (blue dots) the ECAL alignment procedure has been carried out. The distribution for perfectly aligned Monte Carlo events is also shown (black histogram). Technical details: the alignment procedure uses low bremsstrahlung electrons from Z→ee collected at the beginning of the 2017 data-taking period (4.6 fb-1). The distance between position measurements provided by ECAL and the track extrapolated to ECAL with respect to three dimensional translations (x,y,z) (and three Euler angles in the endcaps) is minimized. Conclusion: a relative ECAL-tracker precision of 1x10-3 (in η units) in the barrel and 3x10-3 in the endcaps has been achieved. This meets the ECAL alignment goals of 4x10-3 in η units for electron ID and di-photon resonance reconstruction
pdf version
cDPhi Electrons EB final.png
pdf version
cDPhi Electrons EE final.png
Relative alignment of the ECAL crystals and the CMS tracker, measured using electrons only from Z→ee events, separately for the ECAL barrel and endcaps. The plots show the residual difference in Δφ between the position of the ECAL supercluster and the extrapolated track position, using the point of closest approach to the supercluster. The distribution of Δφ is shown for data before (red triangles) and after (blue dots) the ECAL alignment procedure has been carried out. The distribution for perfectly aligned Monte Carlo events is also shown (black histogram). Technical details: the alignment procedure uses low bremsstrahlung electrons from Z→ee collected at the beginning of the 2017 data-taking period (4.6 fb-1). The distance between position measurements provided by ECAL and the track extrapolated to ECAL with respect to three dimensional translations (x,y,z) (and three Euler angles in the endcaps) is minimized. Conclusion: a relative ECAL-tracker precision of 3x10-3 (in φ units) in the barrel and 6x10-3 in the endcaps has been achieved. This meets the ECAL alignment goals of 20x10-3 in φ units for electron ID and di-photon resonance reconstruction
pdf version
cDPhi Positrons EB final.png
pdf version
cDPhi Positrons EE final.png
Relative alignment of the ECAL crystals and the CMS tracker, measured using positrons only from Z→ee events, separately for the ECAL barrel and endcaps. The plots show the residual difference in Δφ between the position of the ECAL supercluster and the extrapolated track position, using the point of closest approach to the supercluster. The distribution of Δφ is shown for data before (red triangles) and after (blue dots) the ECAL alignment procedure has been carried out. The distribution for perfectly aligned Monte Carlo events is also shown (black histogram). Technical details: the alignment procedure uses low bremsstrahlung electrons from Z→ee collected at the beginning of the 2017 data-taking period (4.6 fb-1). The distance between position measurements provided by ECAL and the track extrapolated to ECAL with respect to three dimensional translations (x,y,z) (and three Euler angles in the endcaps) is minimized. Conclusion: a relative ECAL-tracker precision of 3x10-3 (in φ units) in the barrel and 6x10-3 in the endcaps has been achieved. This meets the ECAL alignment goals of 20x10-3 in φ units for electron ID and di-photon resonance reconstruction
pdf version
ResolutionPaper 2017 0 final.png
pdf version
ResolutionPaper 2017 2 final.png
Relative electron (ECAL) energy resolution, unfolded in bins of pseudo-rapidity (η) for the barrel and the endcaps. Electrons from Z→ee decays are used. The resolution is shown separately for very low bremsstrahlung electrons (named golden, with R9>0.94 with R9 = E3x3 / ESC) and for bremsstrahlung electrons (R9 < 0.94). The relative resolution σE/E is extracted from an unbinned likelihood fit to Z→ee events, using a Voigtian (Landau convoluted with Gaussian) as the signal model. The resolution is plotted separately for data and MC events. The ECAL conditions used in the simulation reflect the status of the detector as predicted after 25/fb of data-taking in 2017. Conclusions: the resolution is affected by the amount of material in front of the ECAL and is degraded in the vicinity of the eta cracks between ECAL modules (indicated by the vertical lines in the plot). Also, the resolution improves significantly after a dedicated calibration using the full 2017 dataset (blue points) with respect to the end-of-year (EOY) 2017 calibration (gray points) for which only time dependent effects were corrected for.
pdf version
histories 2011-2012-2015-2016-2017-2018 180509.png
Relative response to laser light (440 nm in 2011 and 447 nm from 2012 onwards) injected in the ECAL crystals, measured by the ECAL laser monitoring system, averaged over all crystals in bins of pseudorapidity (h), for the 2011, 2012, 2015, 2016, 2017 and 2018 data taking periods, with magnetic field at 3.8 T. The response change observed in the ECAL channels is up to 10% in the barrel and it reaches up to 50% at η ~ 2.5, the limit of the tracker acceptance. The response change is up to 90% in the region closest to the beam pipe. The recovery of the crystal response during the periods without collisions is visible. These measurements, performed every 40 minutes, are used to correct the physics data. This is an update of the plots appearing in CMS-DP-2012/007, CMS-DP-2012/015, CMS-DP-2015/016, CMS-DP-2015/063, CMS-DP-2016/031, CMS-DP-2017/003 and CMS-DP-2017/023 and includes measurements taken up to May 2018. The bottom plot shows the instantaneous LHC luminosity delivered during this time period.
pdf version
medianmeeinEBcat11vstime final2.png
Scale stability of the di-electron invariant mass distribution for the 2017 data taking period using Z→ee electrons in the ECAL barrel. A dedicated re-calibration using the full 2017 dataset was performed. The graph shows the stability of the Z peak when the two electrons are reconstructed in the barrel. Each point on the left panel is obtained by taking the median Mee for the respective time bin. The histogram on the right shows the spread of median Mee. The Mee variable is stable within 0.1% during the year.
pdf version
medianmeeinEBcat00vstime final2.png
Scale stability of the di-electron invariant mass distribution for the 2017 data taking period using Z→ee electrons in the ECAL endcaps. A dedicated re-calibration using the full 2017 dataset was performed. The graph shows the stability of the Z peak when the two electrons are reconstructed in the endcaps. Each point on the left panel is obtained by taking the median Mee for the respective time bin. The histogram on the right shows the spread of median Mee. The Mee variable is stable within 0.2% during the year.
pdf version
pi02 stability.png
Stability of the relative energy scale measured from the invariant mass distribution of π0→ γγ decays in the ECAL barrel (positive z, EB+). The energy scale is measured by fitting the invariant mass distribution of approximatively 500k photon pairs in the mass range of the π0 meson. Each point is obtained from a fit to approximatively 5 minutes of data taking. The energy scale is plotted as a function of time over the 2017 data taking period. The plots show the data with (green points) and without (red points) light monitoring (LM) corrections applied. The right-hand panel shows the projected relative energy scales.
pdf version
medianR9inEBcat11vstime final2.png
Stability of the shape of the electromagnetic deposits in the ECAL barrel for electrons from Z decays. A dedicated re-calibration using the full 2017 dataset was performed. The plot shows the stability of the variable R9, defined as E3x3/ESC, where E3x3 is the energy deposited in a 3x3 crystal matrix around the seed crystal and ESC is the supercluster energy. R9 is responsive to changes in pedestals and noise. Each point on the left panel is obtained by taking the median of the R9 histogram for the respective time bin. The histogram on the right shows the spread of median R9. The R9 variable is stable within 0.3% during the year.
pdf version
medianR9inEBcat00vstime final2.png
Stability of the shape of the electromagnetic deposits in the ECAL endcaps for electrons from Z decays. A dedicated re-calibration using the full 2017 dataset was performed. The plot shows the stability of the variable R9, defined as E3x3/ESC, where E3x3 is the energy deposited in a 3x3 crystal matrix around the seed crystal and ESC is the supercluster energy. R9 is responsive to changes in pedestals and noise. Each point on the left panel is obtained by taking the median of the R9 histogram for the respective time bin. The histogram on the right shows the spread of median R9. The R9 variable is stable within 0.3% during the year.
pdf version
AlignmentResidualBA mFront iter1 final.png
pdf version
AlignmentResidualBA mRear iter1 final.png
pdf version
AlignmentResidualBA pFront iter1 final.png
pdf version
AlignmentResidualBA pRear iter1 final.png
pdf version
ESAlignment2D before final.png
pdf version
ESAlignment2D after final.png
The CMS preshower is a silicon strip detector situated between the strip tracker and the ECAL crystals in the endcaps. Each side consists of two disks (front and rear) set to different directions, thus the front (rear) plane is sensitive to the X(Y)-axis. The preshower records the position of each charged particle which passes through. The coordinate of each plane is aligned offline with respect to the aligned tracker. The preshower alignment algorithm starts from matching the expected hit point of reconstructed tracks with the trajectory extrapolation to the data. A perturbative method minimizing the Χ^2 value in the parameter space is then performed for several iterations until reaching a stable result with respect to the tracker. The 1dim plot shows the residual distribution before (black dots) and after (red dots) alignment for the four planes. The aligned result is sitting within the resolution of the silicon strips, +/- 0.055 cm. The 2dim distributions show the average residual for each sensor module, before and after alignment. These figures exhibit the final good alignment of the ES, as well as the rotation of the disks.
pdf version
PedTreeEBMean final.png
pdf version
PedTreeEEMean final.png
The ECAL signal is readout with a multi gain ADC (gain 12, 6, and 1). Crystal energy deposits up to about 150 GeV are read with gain 12. Gain 12 pedestals mean history in the ECAL Barrel (top) and Endcaps (bottom) for the 2017 data taking period is shown. A long-term, monotonic drift upwards is visible. In the short term (in-fill) luminosity related effects are visible. The short term variations are smaller when the LHC luminosity is lower (for instance in August with respect to July). In November, when LHC produced collisions at low luminosity between heavy ions, the in-fill effects almost vanish. The long term drift depends on the integrated luminosity, while the short term effects depend on the instantaneous luminosity.
pdf version
RMSTreeEBMean final.png
The ECAL signal is readout with a multi gain ADC (gain 12, 6, and 1). Crystal energy deposits up to about 150 GeV are read with gain 12. Gain 12 pedestals RMS (noise) history in the ECAL Barrel for the 2017 data taking period is shown. A long-term, monotonic drift upwards is visible, which is related to the increase of the APD dark current due to irradiation. No short term (in-fill) luminosity related effects are visible.
pdf version
MIP2017F Z0 P0 X06 Y07 bis.png
pdf version
DegradeByLumi 2017 Front final.png
pdf version
DegradeByLumi 2017 Rear final.png
The two planes of the Preshower (ES), coupled with the EE crystals, form a sampling calorimeter. The ES essentially counts the number of charged particles passing through the layers of silicon, which is an estimate of the amount of energy deposited in the ES lead absorbers. We use charged particles with momentum close to minimum ionizing to calibrate the ES, so for simplicity we refer to them as “MIPs”. They are collected with the preshower operated in high-gain mode. The design-goal accuracy of the channel-by-channel calibration is set to 5%. This corresponds to a contribution of about 0.25% to the overall EE+ES energy resolution for high-energy electrons since only a few percent of electron/photon energy is deposited in ES. The sources of response variation (sensor-to-sensor and channel-to-channel) are the sensor thickness seen by the incident particles (depends on angle of incidence), gain of the front-end electronics chain, and charge collection efficiency which varies with the radiation damage.
The top plot shows the energy distribution for a silicon sensor in ADC counts: the average MIP energy deposit is 80.4 keV. The black histogram represents the real data while the red line represents the fit. The hits are selected using the extrapolated trajectory, provided by the tracker. The distribution is fitted to a Landau function (to model the signal) convoluted with a Gaussian function (to model the intrinsic noise). The fitting range was limited around the peak to avoid disturbance from other sources. The presence of a second MIP causes a bump near 100 ADC counts.
The other two plots show the decreasing rate of MIPs during the 2017 data-taking (from 2017A to 2017F) in 5 h regions (the interval in h is 0.5), of the front plane (middle plot) and rear plane (bottom plot). The rate on ES+ is similar to one on ES-, thus the averages of those rates are used. The MIPs decrease faster in the higher h region. This result implies that the ES sensors in high h regions are more affected by the radiation damage.
pdf version
TPdist 306456 oldped final.png
This plot shows a zoom of the previous plot above 16 GeV corresponding to 32 ADC counts. Note that the saturation scale of the trigger primitives is 255 ADC = 127.5 GeV
pdf version
TPint 306456 oldped final.png
This plot shows the integral of the trigger primitive transverse energy spectrum, plotted as a function of transverse energy (ET), from Run 306456, with a peak pileup of 59. The trigger primitive spectrum is shown in black. The yellow histogram corresponds to trigger primitives matched to spikes identified offline. It shows that the rate of mis-identified spikes becomes more important at larger ET. This is largely because spikes have a higher ET distribution than the underlying minimum bias spectrum.
pdf version
spikerat 306456 oldped final.png
This plot shows the residual spike contamination in the trigger primitives, as a function of transverse energy (ET), from Run 306456, with a peak pileup of 59. It is computed by taking the ratio of the two distributions of the previous plot. It shows that the spike contamination grows with ET. This is largely because the ET distribution for spikes is harder than the underlying minimum bias spectrum.
pdf version
TPdist all 306456 oldped final.png

This plot shows the transverse energy (ET) distribution of ECAL trigger primitives recorded during an LHC collision run taken in November 2017, Run 306456, with a peak pileup of 59. The events were recorded using a Zero Bias trigger which fires on the BPTX beam counters located either side of CMS. The trigger primitive spectrum is shown in black: each ADC count corresponds to 0.5 GeV transverse energy, the discontinuity at 16 GeV is due to the online rejection of direct APD signals (“spikes”) which are suppressed above 16 GeV. The yellow histogram corresponds to trigger primitives matched to spikes identified offline, using topological (Swiss-cross) and timing criteria.


Reference:
http://iopscience.iop.org/article/10.1088/1742-6596/404/1/012043

-- AndreaMassironi - 2018-05-29


This topic: CMSPublic > WebPreferences > EcalDPGResultsCMSDPS2018015
Topic revision: r1 - 2018-05-29 - AndreaMassironi
 
This site is powered by the TWiki collaboration platform Powered by PerlCopyright &© 2008-2021 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