Calorimeter systems at collider experiments

36
Click to edit Master title style IEEE short course on: Calorimetry Calorimeter systems at collider experiments Erika Garutti (DESY) 21/10/2011 [email protected] 1

description

Calorimeter systems at collider experiments. Erika Garutti (DESY). Outline. From single calorimeter detectors to calorimeter in a detector system Calorimeters for jets Particle flow algorithms to improve jet energy resolution Highly granular calorimeters - PowerPoint PPT Presentation

Transcript of Calorimeter systems at collider experiments

Page 1: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 1

Calorimeter systems at collider experiments

Erika Garutti(DESY)

21/10/2011

Page 2: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 2

• From single calorimeter detectors to calorimeter in a detector system

• Calorimeters for jets

• Particle flow algorithms to improve jet energy resolution

• Highly granular calorimeters - techniques for analog and digital calorimetry

Outline

21/10/2011

Page 3: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 3

From single calorimeters to a HEP detector

CMS ECAL Endcap

ATLAS barrel HCAL and coil

Calorimeters are in general one component of a complex detector system

Typical of collider detector is the onion-likeStructure of the detector system

21/10/2011

Page 4: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 4

Detectors for collider experiments

Typical onion-like structure for most of modern collider detectors- The tracking system comes first (minimum material budget) - The calorimeter stops (most of) the particles so has to come second- Muons can escape the calorimeter and require an extra detector

CMS

21/10/2011

Page 5: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 5

Particles are not kind!

The distinction between electromagnetic and hadronic calorimeter is not rigorousfor a hadron

~30-40% of first hard interaction of a hadron happen in the EM-calo

W

The choice of a high Z material for the EM-calo minimizes the hadron interactions before the Had-calo:

~30 X0 to stop an EM shower =1 lint of Tungsten (W) or 3 lint of Iron (Fe)

Fe

21/10/2011

Page 6: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 6

Particles are not kind!

About 11-12 lint are needed to containhadrons with energy ~100 GeV

~1.2 m of W or 2.2 m of Fe

WThe choice of a high Z material for the Had-calo minimizes its depth

Fe[c

m]

21/10/2011

Page 7: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 7

ideal calo ideal calo system

Ideal calorimeter

e- 100 GeV

p- 100 GeV

= k x 100 GeV

= k x 100 GeV

Implications:• e/p = 1• L 30 X0 && L 11 lint

L

[g/cm3] int [cm] L [m]

PbWO4 BGOFePb W

8.28 7.137.8711.34 19.25

19.521.8816.717.69.9

2.12.41.81.91.1

Calorimeter system requirements

• g identification (EM/Had segment.)• separation of jets (lateral segment.)• calo contained inside magnetic coil

21/10/2011

Page 8: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 8

Why not using tracker (has better

resolution)?

Particles are not alone!

• Jets are a collimated group of particles that result from the fragmentation of quarks and gluons

• They are measured as clusters in the calorimeter

• momentum of cluster is correlated to the momentum of the original quark

At collider experiments particles come typically in “jets”

21/10/2011

Page 9: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 9

• Measure charged + neutral particles

• Performance of calorimeters improves with energy

• DE/E 1/ + const.• while in a magnetic spectrometer • Dp/p p

• Obtain information on energy flow: total (missing) transverse energy, incoming direction (with high segmentation)

• Obtain information fast (<100ns feasible) recognize and select interesting events in real time (trigger)

Why are jets measured in the calorimeter?

At high energy calorimetry is a must

magn.spectr.

particle E or p [GeV]

21/10/2011

Page 10: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 10

Phenomenology of jets

• Partons (quark/gluon) are produced from the interaction of beam particles

• Partons fragment into hadrons

• Jets clustering algorithm:– Typically uses a geometric

assumption to group particles from the same parton (cone)

• A fraction of the parton energy can be lost (out of the cluster)

Jet = sum of many particles (e,g,p,p,n,K,…)technically: (EEM CAL + EHAD CAL )clusters + muon momentum + Emiss

21/10/2011

Page 11: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 11

Jet versus calorimeter energy scale

• Jets are complicated processes

• EM and Had Calo calibrations are generally not sufficient to get calibrated jet energy– More work needs to be done!!

• Jet energy scale is crucial for many important measurements:– Top quark mass (used to constrain Higgs boson)– Higgs searches / branching ratio– Search for beyond physics the standard model

• Measurements often performed by comparing real data with simulations– Need to get both physics and detector simulation right

21/10/2011

Page 12: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 12

Absolute jet energy scale

• Response to single particles non-linear (in test beam)

• However, jets are identified as one single objects by clustering algorithm

• For a 50 GeV jet: calibration is not the same whether:– one 50 GeV pion– 10 times 5 GeV pionsor whether:– one 50 GeV p0 or p+/-

CMS test beam

21/10/2011

Page 13: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 13

Absolute jet energy scale

• Response to single particles non-linear (in test beam)

• However, jets are identified as one single objects by clustering algorithm

• For a 50 GeV jet: calibration is not the same whether:– one 50 GeV pion– 10 times 5 GeV pionsor whether:– one 50 GeV p0 or p+/-

Solution:• Get the average energy scale:

Simulate an “average” particle configuration inside jet

• Use test beam information to get calibration factor for single particles

21/10/2011

Page 14: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 14

What is inside a jet?

Eparticle/Ejet

𝜎 𝑗𝑒𝑡

𝐸 𝑗𝑒𝑡= 1𝐸 𝑗𝑒𝑡

[ 𝑓 h𝑐 𝑎𝑟𝜎 h𝑐 𝑎𝑟⊕ 𝑓 𝑒𝑚𝜎 𝑒𝑚⊕ 𝑓 h𝑎𝑑𝜎 h𝑎𝑑 ]= 𝑎√𝐸

⊕𝑏⊕ 𝑐𝐸

?

There are wide variations to the average particle energy inside a jet

… but also on the energy carried by different type of particles in a jet

These fluctuations add uncertainty to the jet energy scale determination

21/10/2011

Page 15: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 15

Jet energy resolution at LHC

Stochastic term for hadrons only: ~93% and 42% respectively

jet jet

21/10/2011

Page 16: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 16

ideal calo ideal calo system

Ideal calorimeter

e- 100 GeV

p- 100 GeV

= k x 100 GeV

= k x 100 GeV

Calorimeter system requirements

• g identification (EM/Had segment.)• separation of jets (lateral segment.)• calo contained inside magnetic coil

Calorimeter system

e- 100 GeV

p- 100 GeV

Sampling calorimeters can have highest density Different material in EM/Had segments Different layer thickness in the same materialExtra material (support/cables) between calos

differentsampling factors

21/10/2011

Page 17: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 17

Sampling Method• Weights applied to different calorimeter compartments• Enlarged cone size yields increased electronic noise

H1 Method• Weights applied directly to cell energies• Better resolution and residual nonlinearities

Energy weighting for jets

HADEMHADEMPSjet EEEEEE 3g

CCjHADj

jHADHADjEMj

jEMEMPSjet EEE ,,,, )()(

Back-to-back dijet events

|h|=0.3

ParameterSampling Method H1 MethodDR=0.4 DR=0.7 DR=0.4 DR=0.7

a (%GeV1/2) 66.0 ± 1.5 61.2 ± 1.3 53.9 ± 1.3 51.5 ± 1.1b (%) 1.2 ± 0.3 1.4 ± 0.2 1.3 ± 0.2 2.5 ± 0.2

2 prob. (%) 1.6 0.8 27.3 66.7

ATLASCan the

jet energy resolution be better?

21/10/2011

Page 18: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 18

LEP-like

LEP-like detector

E60%/ΔΕ jet Mj1

j2

Mj3j4

Precision jet physics

Require jet energy resolution improvement by a factor of 2 Worse jet energy resolution (60%/E) is equivalent to a loss of ~40% lumi

Jet1

Jet2Jet3

Jet4

ILC design goal

W Z0

Mj1

j2

Mj3j4

jjjet E30%/ΔΕ

sjet ~3%

Perfect Note due to Breit-Wigner tails best possible separation is 96 %

reasonable choice for LC jet energy resolution:minimal goal sE/E < 3.5 %

lepton machine (ILC: e+ e- @ 0.5-1 TeV, CLIC: @ 1-3 TeV )

build a detector with excellent jet energy resolution

At the Tera-scale, we will do physics with W’s and Z’s as Belle and Babar do with D+ and Ds

Brqq~70%

21/10/2011

Page 19: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 19

Calorimeter for Particle Flow

• Jet energy resolution is worse than (or at most as good as) hadron resolution [world best: ZEUS HCAL shad~35%/E]

• How to improve on jet energy resolution:

Resolution in hadronic calorimeter limited by “fluctuations” : number of p0

produced & amount of invisible energy in one nuclear interaction Two approaches:- measure the shower components in each event

access the source of fluctuations (Dual/Triple Readout)

- minimize the influence of the calorimeter (in particular hadronic one) use combination of all detectors

21/10/2011

Page 20: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 20

The first idea: Energy flow First algorithm developed by ALEPH (LEP) in the early 90ies:• Combine energy measurement from the calorimeter with the momentum measurement from the tracking

p=20 GeVEcalo= 25 GeV

En = 5 GeV

Energy of neutral hadron obtained by subtraction: En = Ecalo – ptrack

BUT: shad ~ 60% E Ehad = 25 ± 3 GeV En = 5 ± 3 GeV

Calorimeter resolution important in the subtraction method

• To not double count the energy: energy deposited in the calorimeter by the tracks has to be masked Generally granularity of had. (and em) calorimeter is the limiting factor

21/10/2011

Page 21: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 21

Particle Flow paradigm reconstruct every particle in the event

up to ~100 GeV Tracker is superior to calorimeter use tracker to reconstruct e±,m±,h± (<65%> of Ejet )

use ECAL for g reconstruction (<25%>)(ECAL+) HCAL for h0 reconstruction (<10%>)

HCAL E resolution still dominates Ejet resolutionBut much improved resolution (only 10% of Ejet in HCAL)

PFLOW calorimetry = Highly granular detectors + Sophisticated reconstruction software

Typical single particle energy at LC

21/10/2011

Page 22: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 22

Particle Flow expectations at LCGoal Jet energy resolution:

Current Pflow performance (PandoraPFA + ILD)uds-jets (full GEANT 4 simulations)

EJETsE/E = /√Ejj |cosq|<0.7 sE/Ej

45 GeV 25.2 % 3.7 %100 GeV 29.2 % 2.9 %180 GeV 40.3 % 3.0 %250 GeV 49.3 % 3.1 %

Equivalent stochastic term shown for comparison PFA resolution is not stochastictails in Gaussian distribution = CONFUSION

Benchmark performance using jet energy resolution in Z decays to light quarks:

21/10/2011

Page 23: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 23

State of the art of Particle Flow algorithm

Currently best performing algorithm: PandoraPFA High granularity Particle Flow reconstruction is highly non-trivial

Clustering Topological Association

30 GeV12 GeV

18 GeV

Iterative Reclustering

9 GeV9 GeV

6 GeV

Photon ID Fragment ID

Mark Thomson, NIM 611 (2009) 24-40 For more details:

many complex steps (not all shown)

21/10/2011

Page 24: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 24

Confusion in Particle Flow

If these hits are clustered together withthese, lose energy deposit from this neutralhadron (now part of track particle) and ruin energy measurement for this jet.

Level of mistakes, “confusion”, determines jet energy resolution not the intrinsic calorimetric performance of ECAL/HCAL

Three types of confusion: i) Photons ii) Neutral Hadrons iii) Fragments

Failure to resolve photonFailure to resolve neutral hadron

Reconstruct fragment asseparate neutral hadron

21/10/2011

Page 25: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 25

Technical aspects of Particle Flow

Use calorimeter measurement to “guide” the clustering:• re-cluster if Ecluster differs too much

from track momentum

Back to an “Energy Flow” methodbut much higher sophistication

Hadronic calorimeter resolutioneffects the clustering performance (second order effect)

21/10/2011

Page 26: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 26

Detector design at ILC

“no” material in front – calorimeter inside the solenoidlarge radius and length – to better separate the particleslarge magnetic field – to sweep out charged trackssmall Moliere radius – to minimize shower overlapsmall granularity – to separate overlapping showers

ILD: International Large Detector

HCAL

ECAL

PandoraPFA currently used to optimize the ILD detector design

ECAL:• SiW sampling calorimeter • longitudinal segmentation: 30 layers • transverse segmentation: 5x5 mm2 pixels

• Steel-Scintillator tile sampling calorimeter• longitudinal segmentation: 48 layers (6 lI)• transverse segmentation: 3x3 cm2 tiles

HCAL:

21/10/2011

Page 27: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 27

Optimization of HCAL

• 3cm x 3cm tiles looks reasonable (5M ch. vs 50M for 1x1cm and 500k ch for 10x10cm)• for low-energetic jets the confusion term of PFA is less sensitive to tile size

Material X0/cm rM/cm lI/cm X0/lI

Fe 1.76 1.69 16.8 9.5

Cu 1.43 1.52 15.1 10.6

W 0.35 0.93 9.6 27.4

Pb 0.56 1.00 17.1 30.5

?

• Maximum containment inside the solenoid small lI

• HCAL will be large: absorber cost/structural properties important

• small granularity – to separate overlapping showers

21/10/2011

Page 28: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 28

Understand Particle Flow performance

21/10/2011

Page 29: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 29

Time structure of the hadronic shower

Steel HCAL

Timing for 250 GeV jet (corrected for time of flight)

• 95 % of energy in 10 ns• 99 % in 50 ns

• In steel suggests optimal timing window in range >10 ns

How is the situation in W?

Previous studies performed assuming a r/o electronics gate of 200ns

21/10/2011

Page 30: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 30

Time structure of the hadronic shower• both #n and #p far from closed shells• naively would expect more nuclear interactions with W• Problem: expect longer time profile (decays, secondary interactions)• Furthermore: not clear how well modeled in Geant 4

Tungsten HCAL Steel HCAL

Tungsten is much “slower” than Steel• only 80 % of energy in 25 ns• only 90 % in 100 ns• how much due to thermal n ?

single KLs (QGSP_BERT)

0.3 MiP cut

21/10/2011

Page 31: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 31

Particle Flow performance vs time cutTungsten HCAL Steel HCAL

• For no time cut (1000 ns) peformance of CLIC_ILD very good- somewhat better than ILD (thicker HCAL, larger B)

• For high(ish) energy jets – strong dependence on time cut- suggests time window of > 10 ns- need something like 50 ns to get into “flat region”

21/10/2011

Page 32: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 32

Summary on Particle Flow Algorithm

• Interplay of highly granular detectors and sophisticated pattern recognition (clustering) algorithms

• Basic detector parameters thoroughly optimized using PandoraPFA

• Time structure of hadronic shower is an important parameter in the feasibility study & in the design of the readout electronics

needs validation

A PFLOW detector is not cheap: do we believe in simulations ?

21/10/2011

Page 33: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 33

The zoo of PFLOW calorimeters

21/10/2011

Page 34: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 34

Energy deposited by a charged particle in the active material of a sampling calorimeter follows a Landau distribution

Long-tail Therefore large fluctuations in energy deposition for a single particle

Typical calorimeters have multiple particles crossing each cell• analogue readout – including Landau fluctuations A sufficiently high granularity calorimeter may only have a single particle crossing each cell• possibility of digital readout, i.e. count charged particles – insensitive to Landau fluctuations

Analogue .vs. Digital readout

21/10/2011

Page 35: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 35

Analogue .vs. Digital readout

Non-linear behaviorfor dense showers

photon analysis

ECAL: Analog readout required

S.Magill (ANL)

hadron analysis

HCAL: either Analog or Digital readout

Slope = 23 hits/GeV

Calorimeter cell size 1x1cm2

iNEg ih NE

21/10/2011

Page 36: Calorimeter systems  at collider experiments

Click to edit Master title styleIEEE short course on: Calorimetry

[email protected] 36

The zoo of PFLOW calorimeters

* Credit: the following slides are based on work done by the CALICE collaboration21/10/2011