precision timing detectors for particle physics
TRANSCRIPT
Precision Timing Detectors for Particle Physics
25.05.2017 Adi Bornheim
Caltech
Physics at HL-LHC
03.02.2017 Adi Bornheim, Precision Timing 2
The Challenge of HL-LHC
24.05.2017 3
• HL-LHC will accumulate 3 ab-1 of p-p collisions to achieve its physics goal.
• This will require operation at 200 collisions per bunch crossing.
• Achieving physics performance very challenging at these PU rates.
20 fb-1 3000 fb-1
140 PU
200 PU
Time tagging of tracks
• Removing PU hits in isolation cones • Improve reconstruction of taus
03.02.2017 Adi Bornheim, Precision Timing 4
Photon Timing in H→γγ
24.05.2017 A. Bornheim - CMS ECAL Precision Timing 5
Vertex ID uses ΣpT to find hard vertex.
At HL-LHC, many PU vertices with high ΣpT.
With a timing measurement for two photons, can calculate the z and t location of the vertex.
H→γγ important mode for precision Higgs physics. Clean signal with large visible cross section makes it a key
driver for measurements of Higgs production and decays dynamics.
Standard candle for di-Higgs to bbγγ.
4D Triangulation with Photon Timing
24.05.2017 A. Bornheim - CMS ECAL Precision Timing 6
t⋅cγ1
t⋅cγ2
With two time and position measurements eg. from two photons and with the constraint from the beam axis x and y location, the vertex x and t can be calculated analytically.
Equivalent to GPS with two satellites.
Physics impact of Photon Vertexing
24.05.2017 A. Bornheim - CMS ECAL Precision Timing 7
H→γγ primary vertex selection efficiency reduced from 80% in current LHC data to 40% at 140 PU and even further at 200 PU.
Dramatic reduction of effective mass resolution, equivalent to a loss in effective luminosity for statistically limited measurements.
Precision timing in the ECAL, combined with precision timing tags on tracks, will recover with deterioration almost fully.
Fast Timing Fundamentals
∆U
∆t
For good time resolution, need: 1. fast rise time (trise) ⇒ primary signal rise time (scintillation : LYSO ~30
ps, Si sensors ~1ns) 2. low Signal-to-Noise (∆U/U) ⇒ primary signal amplitude : LYSO ~30k
photons/MeV (1.07 MeV/mm MIP) , Si sensors ~30k e/h pairs in 300 µ for a MIP
3. more time samples (nsamples)
03.02.2017 Adi Bornheim, Precision Timing 8
9
Construction & Installation: Barrel
Timing Performance of CMS ECAL
03.02.2017 10 Adi Bornheim, Precision Timing
Large PbWO crystal calorimeter. Results from pp collision data at LHC : Electron showers from Z→ee decay ∆tTOF :
~270 ps, single channel : ~190 ps, without path length correction : ~380 ps
Constant term of resolution : ~20 ps in test beam, ~70 ps in situ (same clock).
Studies on jet timing vertex resolution suggest very promising performance.
Scintillation Light Time Spectrum Timing information is extracted from the leading edge of the signal – the
rise time of the light output is important. Crystal scintillators feature intrinsic signal rise times of few 10 ps. For
PbWO4 60 ps have been measured in a cube of 2x10x10 mm3 . S. E. Derenzo, M. J. Weber, W. W. Moses and C. Dujardin. “Measurements
of the intrinsic rise times of common inorganic scintillators.” IEEE TNS, 47:860-864, 2000.
23.05.2017 A. Bornheim - CMS ECAL Precision Timing 11
12
Optical Transit Time Spread Effect of the scintillation photon arrival at the photo detector we
refer to as Optical Transit Time Spread.
Experimental program to explore ultimate timing resolution, in particular the impact of the optical transit time spread.
γ x
γ x
t1
t2
EM shower propagation snapshot
Scintillation light propagation cS < c
100 GeV γ
23 cm
Time evolution of a shower from photon in CMS ECAL PbWO crystal (25 cm long).
1.5 [ns] 0.0 0.5 1.0 12
03.02.2017 Adi Bornheim, Precision Timing
Simulation : Conversion Depth vs Time
GEANT simulation of 100 GeV photon : Correlation between conversion depth and scintillation photon arrival at crystal near and far side face.
Stronger correlation at the near side face.
03.02.2017 Adi Bornheim, Precision Timing 13
Z γ
16 cm
Conversion depth pos Z [cm]
phot
on a
rriv
al ti
me
[ps]
Near side crystal face Far side crystal face
Shower Depth Effects in Data Near side correlation is visible in data. On this data set, using correlation to correct shower depth
effects improves resolution significantly.
03.02.2017 14 Adi Bornheim, Precision Timing
∆tN -∆tF
∆t re
f -∆
t N
∆t re
f -∆
t F
∆tN -∆tF
Full Size Crystal Timing
• Measurements on large (10 cm & 20 cm) LYSO crystals. • Light & shower propagation effects in large crystals. • Time resolution of ~50 ps achieved with 32 GeV.4 • CALOR2014 , CALOR2014 talk
03.02.2017 15 Adi Bornheim, Precision Timing
CMS ECAL Timing Studies
03.02.2017 Adi Bornheim, Precision Timing 16
CMS ECAL, CALOR 2016
Front side readout of CMS PbWO crystal with 2 SiPMs.
Limited to ~70 ps.
NINO chip readout with DRS sampling of the digital output.
Shower depth fluctuations limit precision.
PbWO Light Timing Structure
CMS ECAL MC studies for Phase II upgrade.
Light extraction at one end of the crystal.
Precise understanding of the pulse shape needed to optimize readout electronics.
03.02.2017 Adi Bornheim, Precision Timing 17
Shower and light propagation, c = 1
Shower, light propagation + refractive index PbWO
Shower, light propagation, refractive index + tSCINT PbWO
200 ps/bin
200 ps/bin
amp.
[au]
am
p. [a
u]
PbWO with MCP
LYSO with MCP
CMS PbWO xtal
MCP as photo sensor reflective tube coupling
MCP as photo sensor clear cookie coupling
2x2x2 cm LYSO xtal
Rise time (19%-90%) ~1 ns for both, limited by digitizer MCP-PMT rise time ~150 ps.
PbWO vs LYSO pulse shapes.
CMS ECAL Barrel HL-LHC Upgrade
03.02.2017 Adi Bornheim, Precision Timing 19
• ECAL timing from LHC Run I : 150 ps global (70 ps local). • New Very Front End (VFE) with Trans-Impedance Amplifier (TIA)
& Oversampling (optimal with 160 MS/s). • Full detector readout, L1 accept rate 750 kHz @ 12.5 µs latency. • Timing performance limited by the APD/cable to VFE. • Timing resolution expected at 30 ps down to about 20 GeV. • More details see eg. TIPP2017 .
CMS ECAL Laser Timing
Calorimetric Timing in CMS
Boundary conditions from CMS Phase II Calorimeter Upgrade : • PbWO ECAL Barrel : Replacement of Very Front End and :
– Cope higher L1 rates (750 kHz) – Mitigate “nuclear counter” induced noise in APDs (spikes) – Reduce impact of increased noise from APD dark current
• Endcap Calorimeter replacement with Silicon/Tungsten & Silicon/Steel calorimeter (EE & FH HGCAL) + Scintillator/Steel Backing HCAL : – Current CMS Endcap will not withstand HL-LHC radiation doses. – Cope higher L1 rates (750 kHz) – Higher transverse granularity in the hadronic section. – Longitudinal granularity
03.02.2017 Adi Bornheim, Precision Timing 20
03.02.2017 21
LYSO Shashlik Calorimeter • Stacked tiles of LYSO & tungsten • WLS fiber light readout • LYSO : radiation resistant, very high yield • Large light yield implies:
more stable pulse shape, better S/N better time measurement
• Measured 10%/sqrt(E) and ~1% constant term.
Adi Bornheim, Precision Timing
Shashlik Test Beam
Adi Bornheim, Precision Timing 22
• Proof of concept demo of calorimetric precision timing with shashlik fiber prototype.
• Optical signal integration and transport.
03.02.2017
Pulse Shapes from WLS fibers
23
• Pulse shape (rise-time) has significant dependence on type of WLS fiber
• DSB1 fibers have significantly faster rise-time than Y11 • ~2ns (DSB1) compared to ~7ns (Y11) • Direct contact with LYSO crystal gives ~1ns
Zoom-in version
03.02.2017 Adi Bornheim, Precision Timing
Shashlik Timing Performance
• Performance of solid LYSO cube and LYSO/W scales with the rise time difference due to the WLS.
• Few 10 ps resolution achievable with LYSO based calorimeter, reaching ~32 ps at 32 GeV equivalent signal.
• SH cell timing extracted from same light signal which allows excellent energy resolution.
• http://dx.doi.org/10.1016/j.nima.2015.04.013
24 03.02.2017 Adi Bornheim, Precision Timing
SiPM & Capillary Readout
03.02.2017 Adi Bornheim, Precision Timing 25
Quartz capillary with liquid WLS (DSB) B. W. Baumbaugh, R. Ruchti, et al, IEEE2015
Clear plastic light guide to couple WLS CAP
DSB WLS Plastic fiber
SH Cell Timing with SiPM
03.02.2017 Adi Bornheim, Precision Timing 26
Capillary and plastic fiber timing the same at fixed amplitude. Time resolution per fiber similar for fixed particle energy. SH cell timing around 60 ps per fiber (48 ps four fibers combined). SiPM timing with ps laser around 10 ps. Reference timing detector not unfolded, around 10 ps resolution.
03.02.2017 Adi Bornheim, Precision Timing 27
Shashlik Time Resolution • Observe 1/√E scaling vs energy (number of photons) • Achieve ~50ps resolution at the highest energy • Time resolution limited by S/N (no amplifier on the SiPM output).
DSB
Micro Channel Plates • MCPs achieve <10 ps resolution for MIPs. • Crucial tool to study timing detectors. • Operation in secondary emission mode
avoids the need for a photo cathode. • Significant R&D for large area MCPs (LAPPD). • For LHC, would need large area, radiation
hard, segmented and high rate capable. • NIM-A 795 (2015) 288-292.
03.02.2017 Adi Bornheim, Precision Timing 28
σT : 7 ps (single device : 9.6/√2)
Pixelated MCP, multiple measurements on a shower
EM shower temporal coherence
Single Si Pad Timing
• Si pad : Hamamatsu, 6x6 mm, 325 µm, no gain. • 0.2 mm steel box, 1.5 cm “thick” • ORTEC VT120C pre-amplifier • Hamamatsu C5594 amplifier
03.02.2017 Adi Bornheim, Precision Timing 29
Experimental Setup
03.02.2017 Adi Bornheim, Precision Timing 30
Trigger counter 2x2 mm
x-y stager
Energy and Timing Measurements
• Correlation between beam energy and signal amplitude (recall, only one 6x6 mm pad).
• Timing resolution improves with signal amplitude.
• More details see http://dx.doi.org/10.1016/j.nima.2015.11.129 NIM-A
03.02.2017 Adi Bornheim, Precision Timing 31
At 6 X0
Amplitude Dependence • Signal amplitude spread varies due to limited containment on
single Si sensor. Use to look at timing resolution vs amplitude. • Timing resolution strictly scales with the signal amplitude. Here,
reaching 16 ps at the highest amplitude bin for 32 GeV beam.
03.02.2017 Adi Bornheim, Precision Timing 32
Timing with Si/W Calorimeter
• CMS plans to replace endcap calorimeter with a Si/W – Si/Fe (HGCAL) + Scint/Brass (BH) calorimeter for HL-LHC upgrade.
• It has been demonstrated that a <20 ps resolution can be achieved with a single Si pad sensor w/o gain if placed in an EM shower.
• An EM shower will have a few 10 pads contributing to the measurement.
03.02.2017 Adi Bornheim, Precision Timing 33
HGCAL
BH
⇒ Calorimeters can provide precise timing for energy deposits of a (10s) few GeV.
MIP Timing Detector
03.02.2017 Adi Bornheim, Precision Timing 34
For HL-LHC upgrades – with ~todays technology: • Assuming a timing layer between tracker and calorimeter, choose
granularity such that each track can be time tagged and occupancy below 10% to avoid double hits.
• Use CMS HGCAL simulation (1 cm2 Si pads) to estimate occupancy. • 1 cm2 granularity sufficient up to eta ~2.4. • Impact point taken from tracker to associate time tag to track. • CMS Timing Layer : Barrel area ~40 m2 , Endcap ~9 m2
CMS ECAL Barrel Timing Layer • LYSO/SiPM : 15 ps for MIP demonstrated in TB. • Granularity ~1 cm2 (4 per PbWO crystal), ~300000 channels, LYSO 3 - 4 mm,
entire detector few (~2.5) cm thick, attached to the tracker support tube. • CMS approved to include in Phase II scope in April, presented to LHCC in
May. • Endcap timing layer investigating LGAD based solution.
03.02.2017 Adi Bornheim, Precision Timing 35
M. Lucchini et al, CALOR2016, CALOR2016
1 tray, 2 half trays
Module
03.02.2017 Adi Bornheim, Precision Timing 36
Barrel Timing Layer
Test Beam, CERN T9, May 2017
Timing Layer Test Beam at FNAL
Ongoing test at FNAL with LYSO/SiPM, LGAD and CdTe.
03.02.2017 Adi Bornheim, Precision Timing 37
BTL Schedule
June CMS week discussion on project status. Further test beams in June and September.
03.02.2017 Adi Bornheim, Precision Timing 38
Sensors
Electronics
Mechanics
Timing with CdTe
23.05.2017 Adi Bornheim, Precision Timing with CdTe Sensors 39
EM Showers are photon rich CdTe has high sensitivity for low energy photons
MIP signal 50k electrons in 300 µm.
CdTe density ~5.8 g/cm3 Sensor available in large
thickness (here : 1 mm). CdTe more suitable for
calorimetric application.
Test Beam Setup
23.05.2017 40 Adi Bornheim, Precision Timing with CdTe Sensors
Tests performed at CERN test beam facilities T9 (up to 7 GeV) and H2 (up to 200 GeV)
DRS based fast digitizer as readout with MCP as timing reference (<10 ps precision).
Pulse Shapes
23.05.2017 41 Adi Bornheim, Precision Timing with CdTe Sensors
Example pulse shapes from 100 GeV electron beam Pulse shape have fast rise time : ~1 ns. Excellent signal/noise due to high ionization density in
the shower. Two time constants : electron- and hole-induced pulse.
Energy Response
23.05.2017 42 Adi Bornheim, Precision Timing with CdTe Sensors
Energy response scales well with beam energy. Single sensor test (1x1 cm2) does not allow detailed
performance study. High density of CdTe allows rather homogeneous
sandwich calorimeter.
Corrected Timing
23.05.2017 43 Adi Bornheim, Precision Timing with CdTe Sensors
Timing resolution depends on the impact point relative to the diode contacts.
Corrected resolution across sensor : 25 ps.
Best resolution : 20 ps.
Single Mip Timing
23.05.2017 44 Adi Bornheim, Precision Timing with CdTe Sensors
Ongoing test beam at FNAL. New boards and housing to reduce noise will allow
access to smaller signal, eventually single MIP timing.
Summary
• Precision timing detectors with a few 10 ps resolution can have a major impact on HL-LHC physics performance.
• A range of detector technologies available to achieve this performance.
• System integration aspects a challenge. • Many groups involved in R&D, exciting work, ready
for a large scale application. • CMS is moving towards a holistic event
reconstruction with full timing coverage.
03.02.2017 45 Adi Bornheim, Precision Timing
03.02.2017 Adi Bornheim, Precision Timing 46
Backup
03.02.2017 Adi Bornheim, Precision Timing 47
Scintillation Light Time Spectrum Scintillating crystals get often classified in fast and slow by their light output
decay constants. This is often 10s of ns – PWO, LYSO : ~40 ns. Timing information is extracted from the leading edge of the signal – the rise
time of the light output is important. LYSO scintillation light properties :
Light output rise time tR < 75 ps, 35000 photons/MeV, tD = 33 ns. See : S Seifert, J H L Steenbergen, H T van Dam and D R Schaart, 2012
JINST 7 P09004. doi:10.1088/1748-0221/7/09/P09004
03.02.2017 Adi Bornheim, Precision Timing 48
Low Gain Avalanche Diodes • Recent results from arxiv 1608.08681 • Low-Gain Avalanche Detectors (LGAD) design, employing n-on-p
silicon sensors with internal charge multiplication in a thin, low-resistivity diffusion layer below the junction.
• UFSD used in this test belongs to the first production of thin (50 μm) sensors, with an pad area of 1.4 mm2.
• With three sensors aligned on a track, 15 ps are achieved.
03.02.2017 Adi Bornheim, Precision Timing 49
Deep Depleted APD • Deep Depleted APD with mesh readout • Large MiP signal (3600 eh pairs × 3 internal APD gain of few 100) • Weighting field controlled with sintered Au (bottom) and mesh
(top) layer, Landau contribution limited, may offer additional benefit due to its electrical properties.
• Eg. S. White, Elba, Fast Timing Detector R&D for the HL-LHC era
03.02.2017 Adi Bornheim, Precision Timing 50
Timing Calorimeter with CdTe
03.02.2017 Adi Bornheim, Precision Timing 51
EM Showers are photon rich CdTe has high sensitivity for low energy photons
10x10x1 mm3 CdTe, Acrorad Ltd.
• CdTe sensors can be thick (few mm). • 50k electrons / 300 µm • Density ~2.5x larger than Si • Larger Z results in larger cross
section for soft x-ray photons.
CdTe Pulse Shapes
03.02.2017 Adi Bornheim, Precision Timing 52
• Charge carrier mobility in CdTe is very different than in Si.
• Holes move slower, electrons vary with local field.
• Pulse rise times of ~1.3 ns observed.
Signal Dynamics in CdTe
03.02.2017 Adi Bornheim, Precision Timing 53
• Excellent timing resolution down to 20 ps achieved.
• Resolution varies across sensor. 25 ps average across sensor.
• Further studies ongoing for different sensor sizes, thickness, etc.
• Submitted to NIM.
Time walk Correction
23.05.2017 54 Adi Bornheim, Precision Timing with CdTe Sensors
Time response depends on amplitude : change in S/N, electronic chain typically not perfectly linear.
Still observe very good time resolution of ~110 ps at the lowest energy shower (2 GeV, 2 X0)
JetMET with Precision Timing
03.02.2017 Adi Bornheim, Precision Timing 55
• High pile-up has major impact on JetMET reconstruction.
• At low pT (40 GeV) PU energy in the jet cone exceeds the jet energy by a factor 2 or more.
• Each PU event adds 3 GeV in quadrature to the MET resolution, resulting in about 40 GeV for 200 GeV.
Track Time Tagging • At 200 PU, increasingly difficult to separate nearby vertices. • Vertex merging causes degradation in isolation and MET variables. • ΣpT of merged vertices promotes PU events to be ranked higher
than hard interaction in vertex ID. • Further studies ongoing on rate of PU track merging to vertices. • For more details see 2016 ACES workshop
03.02.2017 Adi Bornheim, Precision Timing 56
2
Further track timing applications
03.02.2017 Adi Bornheim, Precision Timing 57
As vertices start to overlap within effective tracking resolution, rate of pileup tracks associated to hard interaction vertex increases.
Corresponding degradation of charged isolation, b-tagging, Jet/MET performance.
Physics Case for Timing at HL-LHC
03.02.2017 Adi Bornheim, Precision Timing 58
03.02.2017 Adi Bornheim, Precision Timing 59
Time Walk Correction
23.05.2017 60 Adi Bornheim, Precision Timing with CdTe Sensors
Only partial containment with a 1x1 cm2 sensor. Large spread in contained energy also results in
sizeable effect of the time walk correction for fixed energy.
Timing Response Uniformity
23.05.2017 61 Adi Bornheim, Precision Timing with CdTe Sensors
Timing response of the sensors shows a strong dependence on impact point.
Note : Sensor exposed to a shower after 6 X0 tungsten.
4D Triangulation with Photon Timing
03.02.2017 Adi Bornheim, Precision Timing 62
t⋅cγ1
t⋅cγ2
• With two time and position measurements eg. from two photons and with the constraint from the beam axis x and y location, the vertex x and t can be calculated.
• Equivalent to GPS with two satellites.