calorimetry for fcc-hh
TRANSCRIPT
Calorimetry for FCC-hh
Martin Aleksa on behalf of the FCC-Calo Group
• Introduction: FCC-hh and FCC-hh Reference Detector• Calorimetry of the FCC-hh Reference Detector • Single Particle Performance• Conclusions & Outlook
Based on Material from the FCC CDR Summary Volumes (https://fcc-cdr.web.cern.ch/) and Studies by the FCC-hh Calorimetry Group (https://indico.cern.ch/category/8922/)
November 25, 2019 CHEF 2019 — M. Aleksa (CERN) 1
Kawachi Wisteria Garden, Kitakyushu, Fukuoka, Japan
Introduction – FCC and the FCC-hhReference Detector
November 25, 2019 CHEF 2019 — M. Aleksa (CERN) 2
Scope of FCC Study• International collaboration
(hosted at CERN) to study:– ~100 km tunnel infrastructure in
Geneva area, linked to CERN
– e+e- collider (FCC-ee), potential first step
– pp-collider (FCC-hh), long-term goal, defining infrastructure requirements
~16T 100TeV pp in 100km
– HE-LHC with FCC-hh technology
– Ions and lepton-hadron options with hadron colliders
• CDR for European Strategy Update 2019/20
CHEF 2019 — M. Aleksa (CERN)November 25, 2019 3
FCC Conceptual Design Report• Used as input for Update of European
Strategy in Particle Physics 2019– 4 CDR volumes submitted to EPJ in
December 2018
• Experiment & Physics Studies using FCCSW based on – 150 TB of full simulated data
– 100 TB of Delphes events
• https://fcc-cdr.web.cern.ch/
CHEF 2019 — M. Aleksa (CERN)November 25, 2019 4
Baseline Parameters for FCC-hh• Present working hypothesis:
– Peak luminosity:
• baseline: 5x1034
• ultimate: ≤ 30x1034
– Integrated luminosity:
• baseline ~250 fb-1 (per year)
• ultimate ~1000 fb-1 (per year)
• An operation scenario with…
– 10 years baseline, 2.5 ab-1
– 15 years ultimate, 15 ab-1
• … would result in a total of
O(20) ab-1 over 25 years of
operation.
CHEF 2019 — M. Aleksa (CERN)November 25, 2019 5
Parameter FCC-hh HE-LHC HL-LHC
collision energy cms [TeV] 100 27 14
dipole field [T] 16 16 8.33
circumference [km] 97.75 26.7 26.7
beam current [A] 0.5 1.1 1.1
bunch intensity [1011] 1 1 2.2 2.2
bunch spacing [ns] 25 25 25 25
synchr. rad. power / ring [kW] 2400 101 7.3
SR power / length [W/m/ap.] 28.4 4.6 0.33
long. emit. damping time [h] 0.54 1.8 12.9
beta* [m] 1.1 0.3 0.45 0.15 (min.)
normalized emittance [mm] 2.2 2.5 2.5
peak luminosity [1034 cm-2s-1] 5 30 16 5 (lev.)
events/bunch crossing 170 1000 460 132
stored energy/beam [GJ] 8.4 1.4 0.7
Requirements for FCC-hh Detector• ID Tracking target: achieve σpT / pT = 10-20% @ 10 TeV
• Muons target: σpT / pT = 5% @ 10 TeV
• Keep calorimeter constant term as small as possible (and good sampling term)
– Constant term of <1% for the EM calorimeter and <2-3% for the HCAL
• High efficiency b-tagging, τ-tagging, particle ID!
• High granularity in tracker and calos
• Pseudorapidity (η) coverage:
– Precision muon measurement up to |η|<4
– Precision calorimetry up to |η|<6
• Achieve all that at a pile-up of 1000! Granularity & Timing!
• On top of that radiation hardness and stability!
CHEF 2019 — M. Aleksa (CERN)November 25, 2019 6
Used in Delphesphysics simulations
VBF jets η-distr.
A Possible FCC-hh Detector – Reference Design for CDR
CHEF 2019 — M. Aleksa (CERN)November 25, 2019 7
• Reference design for an FCC-
hh experiment developed to
demonstrate feasibility of an
experiment exploiting full
physics potential
• Input for radiation
simulations
• Input for Delphes physics
simulations
• Room for other ideas, other
concepts and different
technologies
Forward detectors up to η=6
Barrel HCAL: σE/E≈50%/√Ē⊕3%
Barrel ECAL: σE/E≈10%/√Ē⊕0.7%
Tracker: σpT/pT≈20% at 10TeV (1.5m radius)
Central Magnet:B=4T, 5m radius
23m
9m
Muon System: σpT/pT≈5% at 10TeV
1 MeV Neutron Equivalent Fluence for 30ab-1
CHEF 2019 — M. Aleksa (CERN)November 25, 2019 8
Central tracker:• first IB layer (2.5 cm ): ~5-6 1017 cm-2
• external part: ~5 1015 cm-2 Calorimeter gap: from 1016 cm-2 to 1014 cm-2
Forward calorimeters:~5 1018 cm-2 for both the EM and the HAD-calo
Barrel calorimeter:EM-calo: 4 1015 cm-2
HAD-calo: 4 1014 cm-2
End-cap calorimeter:EM-calo: 2.5 1016 cm-2
HAD-calo: 1.5 1016 cm-2
Generally ~10-30 times worse than HL-LHC (similar for TID)Exception: Forward calorimeter goes to higher η bigger factor
FCC-hh Detector – Calorimetry
November 25, 2019 CHEF 2019 — M. Aleksa (CERN) 9
FCC-hh Calorimetry
CHEF 2019 — M. Aleksa (CERN)November 25, 2019 10
• Good instrinsic energy resolution
• Radiation hardness• High stability• Linearity and uniformity• Easy to calibrate
• High granularity
Pile-up rejection
Particle flow
3D/4D/5D imaging
FCC-hh Calorimetry
FCC-hh Calorimetry
CHEF 2019 — M. Aleksa (CERN)November 25, 2019 11
Reference DetectorInspired by ATLAS calorimetry: • Excellent conventional
calorimetry• In addition high granularity to
optimize for Particle Flow techniques, pile-up rejection, boosted objects….
• ECAL, Hadronic EndCap and Forward Calo (≥30X0):
LAr / Pb (Cu) • HCAL Barrel and Extended
Barrel (≥10λ):Scintillating tiles / Fe(+Pb) with SiPM
FCC-hh Electromagnetic Calorimeter (ECAL)
CHEF 2019 — M. Aleksa (CERN)November 25, 2019 12
ZOOM
• Compared to ATLAS, FCC-hh Calo needs finer longitudinal and lateral granularity– Optimized for particle flow– 8 longitudinal compartments, fine lateral granularity – Granularity: Δη x Δφ ≈ 0.01 x 0.01; first layer Δη x Δφ ≈ 0.0025 x
0.02 ~2.5M channels
• Noble liquid (LAr) as active material– Radiation hardness, linearity, uniformity, stability
• Possible only with straight multilayer electrodes (no accordion)– Straight absorbers (Pb + stainless steel sheets in EM section)– Readout and HV on straight multilayer electrodes (PCBs, 7 layers,
1.2mm thick)
• EM Barrel: Absorbers 50∘ inclined with respect to radial direction – Sampling fraction changes with depth fsampl ≈ 1/7 to 1/4 – LAr gap 2 x 1.15mm to 2 x 3.09mm– Longitudinal segmentation essential to be able to correct!
• EM EndCap (EMEC): Straight Pb + steel absorbers and multilayer electrodes perpendicular to the beam axis
• Hadronic EndCap (HEC) and Forward Calorimeter: Straight Cu absorbers and multilayer electrodes perpendicular to the beam axis
EM Barrel
EM and HadronicEndCap
How to Achieve High Granularity?Realize read-out electrodes as multi-layer PCBs (1.2mm thick)• Signal traces (width wt) in dedicated signal
layer connected with vias to the signal pads• Traces shielded by ground-shields (width ws)
forming 25Ω – 50Ω transmission lines• capacitance between shields and signal
pads Cs will add to the detector capacitance Cd
• Ccell = Cs + Cd ≈ 100 – 1000pF• The higher the granularity the more shields
are necessary Ccell increases
CHEF 2019 — M. Aleksa (CERN)November 25, 2019 13
Serial noise contribution proportional to capacitance Ccell
4 – 40MeV noise per read-out channel assuming ATLAS-like electronics
Hadronic showers: Energy sums over O(500-10000) cells
Performance of ECAL• Sampling fraction changes as function of the shower depth
– Longitudinal segmentation essential to be able to correct!– 8 longitudinal layers found to be optimal
• Required energy resolution achieved – Sampling term ≤ 10%/√Ē, only ≈300 MeV electronics noise despite
multilayer electrodes
• Challenge: Up to <µ> = 1000 interactions per bunch-crossing– Pile-up noise is dominating (up to 1.3GeV of pile-up noise)
– Out-of-time pile-up can be corrected for by clever reconstruction algorithms (already applied for ATLAS Phase I Upgrade) and by optimizing cluster size
– In-time pile-up can only be reduced by optimizing cluster size, using timing information or/and efficient pile-up suppression using the inner tracker
CHEF 2019 — M. Aleksa (CERN)November 25, 2019 14
σpile-up noise = 3GeV (large cluster, <µ>=1000)σpile-up noise = 1.3GeV (small cluster, <µ>=1000)
Sampling fractionvs depth
smal
lclu
ster
larg
e cl
ust
er
Performance of ECAL
• Position resolution important to perform combination between tracks measured by inner tracker and calorimeter clusters
• Fine φ segmentation in the first layer (Δη x Δφ ≈ 0.0025 x 0.02) and fine η-segmentation of other layers (Δη = 0.01) crucial for π0
rejection (H𝛾𝛾)– Excellent results obtained using
MVA with up to 15 variables– Deep Neural Network based
analysis shows similar results
CHEF 2019 — M. Aleksa (CERN)November 25, 2019 15
500µm500µm
FCC-hh Hadronic Calorimeter Barrel (HCAL)
Barrel HCAL:• ATLAS type
– Scintillator tiles – steel
• Higher granularity than ATLAS– Δη x Δφ = 0.025 x 0.025– 10 instead of 3 longitudinal layers– Steel –> stainless Steel absorber
(calorimeter inside magnetic field), Pb spacers
• SiPM readout faster, less noise, less space
• Total of 0.3M channels• Calibration of scintillating tiles
(+SiPMs) using Cs source• Calibration of SiPMs using LEDs or
laser
CHEF 2019 — M. Aleksa (CERN)November 25, 2019 16
e/h ratio very close to 1 achieved using steel absorbers and Pb spacers (high Z material)
ZOOM
First Tests with SiPMs in the Lab• First measurements in the lab using Sr90 source
and cosmics• SiPM (Hamamatsu S13360-1325CS)• Measured S/N ratio as a function of distance from
the wavelength shifting fibre
CHEF 2019 — M. Aleksa (CERN)November 25, 2019 17
Components used for tests: • Scintillating tiles made of
polystyrene doped with 1.5% pTp and 0.04% POPOP and double cladding Y11
• Wavelength-shifting fibres from Kuraray
• SiPM: Multi-Pixel Photon Counter (MPPC) - Hamamatsu, type S12571-015C
e/h Ratio – Standalone π± Resolution• Steel absorbers partially substituted by lead
(all spacers)– Sci : Pb : Steel = 1 : 1.3 : 3.3
• By adding Pb decreasing non-compensation by suppression of EM response:– Pb: X0 = 0.6 cm, λ=17.6 cm– Fe: X0 = 1.8 cm, λ=16.8 cm
• Reduction of depth by 0.4 λ (at η = 0)
CHEF 2019 — M. Aleksa (CERN)November 25, 2019 18
Combined Calorimeter ResolutionSingle π± resolution obtained from simulation with simple reconstruction “benchmark method”
• Correcting for energy lost in dead material between calorimeters by exploiting correlation between energy loss and geometric mean of neighboring layer energies
CHEF 2019 — M. Aleksa (CERN)November 25, 2019 19
Results:• Sampling term of 44% (48%) achieved for B=0T (B=4T) respectively. • Constant term of ≈2% achieved• More sophisticated analysis (including local weighting) expected to further
improve this resolution• Ultimate energy resolution will be achieved using 4D imaging and particle flow
Combined Performance for Hadrons• FCC-hh calorimetry: High granularity of EM and HCal
• Ideal for particle flow techniques (but not yet implemented in SW)
• Which resolution can be achieved with calorimetry only, if using the full shower-imaging information?
• First results obtained using a convolutional neural network
– Training with 8M events (without any noise for the moment)
– Excellent results obtained Sampling term of 37%/√Ē (!!!)
– Very promising – further improvement expected using particle flow in combination with the inner tracker (being implemented into SW)
CHEF 2019 — M. Aleksa (CERN)November 25, 2019 20
Noble Liquid Calorimetry for FCC-ee?
CHEF 2019 — M. Aleksa (CERN)November 25, 2019 21
HCALECAL
Drift chamber
• FCC-ee experiment calorimetry requirements:– Excellent jet energy resolution (~30%/√Ē)– Photon measurement down to 300MeV low noise– Excellent particle ID
• High granularity calo for particle flow• Noble liquid calorimeter (as described before) and HCAL has
been integrated into FCC-ee Experiment in FCC SW• First implementation:
– Inner Tracker: DREAM-like drift chamber– ECAL: FCC-hh style LAr/Pb calorimeter– HCAL: FCC-hh style Tile calorimeter (Fe absorbers)– Solenoid coil outside the calorimetry
• Options which will be studied– Inner tracker: Si tracker– ECAL: W absorbers, LKr as active material, granularity optimization– HCAL: CLD-like HCAL– Solenoid coil in front of the ECAL, integrated into LAr cryostat (need thin
coil and low-material cryostat, see back-up)
• EP R&D work-package, also H2020 EoI has been submitted
Conclusions and Outlook• Noble liquid + scintillating Tile calorimetry has been successfully
used for LHC experiments (ATLAS Calorimeters) – Strengths: Stability, pointing resolution, particle ID, radiation hardness
• Similar calorimeter concept is proposed for the reference detector of FCC-hh– Simulations show that such a calorimeter would fulfil physics
requirements for FCC-hh– Excellent standalone performance for electrons and single hadrons– Physics objects performance will be shown in the next talk
• Noble liquid calorimetry is also an interesting option for an FCC-ee experiment!
CHEF 2019 — M. Aleksa (CERN)November 25, 2019 22
Back-Up
November 25, 2019 CHEF 2019 — M. Aleksa (CERN) 23
Criteria for Physics Potential of Future Colliders – FCC-hh
CHEF 2019 — M. Aleksa (CERN)November 25, 2019 24
M. Mangano, Sept. 2018
• Guaranteed Deliverables:– Study of Higgs and top quark properties, and exploration of EWSB phenomena, with unmatchable
precision and sensitivity• Sensitivity to the shape of the Higgs potential (Higgs self coupling)
– Ultimate precision standalone and in combination with FCC-ee and FCC-eh
• Exploration Potential:– Mass reach enhanced by factor ~ E / 14 TeV
• will be 5–7 at 100 TeV, depending on integrated luminosity
– Sensitivity to rare processes enhanced by orders of magnitude– Benefit from indirect precision probes at low and high Q2
• Provide YES/NO Answers:...to questions like…– Is the SM dynamics all there is at the TeV scale?– Is there a TeV-scale solution to the hierarchy problem?– Is DM a thermal WIMP?– Was the cosmological EW phase transition 1st order? – Could baryogenesis take place during the EW phase transition?
What Limits Granularity in ATLAS LAr?• In the ATLAS LAr calorimeter
electrodes have 3 layers that are glued together (~275µm thick)– 2 HV layers on the outside– 1 signal layer in the middle
• All cells have to be connected with fine signal traces (2-3mm) to the edges of the electrodes– Front layer read at inner radius– Middle and back layer read at
outer radius
• limits lateral and longitudinal granularity
• maximum 3 long. layers
CHEF 2019 — M. Aleksa (CERN)November 25, 2019 25
Requirements for FCC-ee Calorimetry
November 25, 2019 CHEF 2019 — M. Aleksa (CERN) 26
FCC-ee
CHEF 2019 — M. Aleksa (CERN)November 25, 2019 27
“IDEA”“CLIC-detector revisited - CLD”• Calorimetry requirements:– Excellent jet energy resolution (~30%/√Ē)– Particle ID
• High granularity calorimetry based on particle flow– Same technologies as for CLIC/ILC under study
• e.g. Si/W ECAL and Scintillator/Iron HCAL
– On top of that fibre-sampling dual-readout calorimetry could be a very interesting option for future leptonic colliders • Fine transverse granularity• Need longitudinal segmentation to separate e/𝛾
from π±! Idea with fibres of different length• Excellent hadronic resolution
(simulation ~35%/√Ē)
– Noble liquid based calorimetry• Calo only: Simulation with DNN calib. for
LAr ECAL + TileCal HCAL: ~37%/√Ē• Fine granularity particle flow reco with ID
Requirements for FCC-ee Calorimetry
CHEF 2019 — M. Aleksa (CERN)November 25, 2019 28
Dimensions: Example CLD• Energy range of particles: – All particles ≤ 182.5 GeV
• 22X0 and 5-7𝜆 sufficient
– Measure particles down to 300 MeV (e.g. photons)• Little material in front of the calorimeter• Low noise (noise term dominant at small
energies, b≪ 300 MeV)!
• EM resolution as good as possible (≤ 15%/√Ē)– e.g. for CLFV τ decays τ → µ𝛾
• Jet resolution must be excellent (~ 30%/√Ē) to separate W and Z decays
• Position resolution of photons: σx = σy = (6 GeV/E⊕ 2) mm Particle ID: – e±/π± separation, – τ decays with collimated final states, separate
different decay modes with minimal overlap (e.g. π0 close to π±)
Could Noble Liquid Calorimetry Work? (1/2)• Dimensions:
– ECAL: 22X0: ~60 cm radial space for FCC-hh type LAr/Pb calorimeter (including 15cm cryostat thickness)• Could be further reduced to ~45cm by using W instead of Pb
– ECAL+HCAL: 5-7λ: ~1.2 m radial dimension for TileCal (FCC-hh style) – Calorimetry within Δr = 1.65m (W ECAL + TileCal) to Δr = 1.8m (Pb ECAL + TileCal).– Only slightly bigger than CLD (ΔrCLD = 1.4m)
CHEF 2019 — M. Aleksa (CERN)November 25, 2019 29
• EM resolution: – Sampling term of ≈ 10%/√Ē quite easily achievable– Resolution at low energies dominated by noise term
• Noise term ≈ 300MeV achieved with FCC-hh calorimeter• Noise term of ≤ 100 MeV seems possible for low energy deposits
(optimized cluster size, see back-up)• In addition for large cell capacitances: Could probably gain factor 1.5
when going to slower shaping times τ ≥ 100 ns (FCC-hh τ = 45ns)
– To measure down to 300 MeV need to minimize material in front of the calorimeter ”thin” Al cryostat or carbon fibre cryostat (R&D program at CERN has been launched, see next slide)
R&D on Low Mass Composite Cryostats
CHEF 2019 — M. Aleksa (CERN)November 25, 2019 30
See
WG
4 c
on
trib
uti
on
at
EP R
&D
WS
htt
ps:
//in
dic
o.c
ern
.ch
/eve
nt/
743
66
1/c
on
trib
uti
on
s/3
15
41
80
/att
ach
men
ts/1
72
15
46
/27
81
79
9/2
01
809
25
_WG
4_D
etec
tor_
Mec
han
ics.
pd
f
Could Noble Liquid Calorimetry Work? (2/2)• Jet resolution
– Requirement: (~ 30%/√Ē) to separate W and Z decays– Calo only: hadron resolution (π-) achieved in FCC-hh (LAr ECAL + TileCal HCAL):
• Benchmark reconstruction: ~48%/√Ē• Simulation with DNN calibration ~37%/√Ē• Already close to the required 30%/√Ē
– Particle flow will further reduce the jet resolution (not simulated yet)– Future jet energy reconstruction will probably use machine learning to
combine tracker information and full 3D imaging information
• Particle ID:– Fine granularity in η, φ, depth can be adjusted according to needs (PCB)– E.g. FCC-hh first calorimeter layer 5mm x 20mm cells (x 9cm depth)– Similar segmentation of read-out electrode would lead to cell sizes of 2.5mrad
x 2.5mrad for FCC-ee
• Position resolution:– For FCC-hh achieved ση ≈ 2x10-3/ √Ē– ≈ 4.4mm/√Ē ~1mm for 20GeV (2mm for 5GeV) sufficient
• Timing:– ATLAS LAr: Timing resolution of 150ps achieved for EM showers ≥30GeV, 65ps
for EM showers ≥100GeV– Possibility to further improve this resolution – studies needed and planned
CHEF 2019 — M. Aleksa (CERN)November 25, 2019 31
Example: Optimize Noise for FCC-hh Calorimeter
CHEF 2019 — M. Aleksa (CERN)November 25, 2019 32
Challenge: Good photon energy resolution down to 300MeV: – 30% resolution at 300MeV if noise level can be kept below 100MeV
– Cluster size optimization (only first layers have significant energy deposits) less cells per cluster less electronics noise
Courtesy A. Zaborowska
Only 3.5% of the energy in layers 5 to 8 (for 300 MeV photon)