Calorimeter Calibration and Jet Energy Scale

Jean-Francois Arguin

November 28th, 2005

Physics 252B, UC Davis

Outline

Quick remainder of calorimetry Calibration before the experiment starts: test beam Calibration when the experiment is running:

Hardware calibration Collider data

Measuring jets at high-energy colliders Example of a physics measurement: top quark

mass

Basics of Calorimetry

Incident particle creates a shower inside material Shower can be either

electromagnetic or hadronic

Energy is deposited in material through ionization/excitation

Basics of Calorimetry II

Basic principle of calorimetry: deposited energy is proportional to incident energy

Calorimeter calibration translate detector response to incident energy

Great feature of showers for detector use: length is proportional to logE

Electromagnetic showers

Created by incident photon and electron electrons emit

bremstrahlung photons undergo pair

production

Length of shower expressed in term of X

0

X0 depends on material

95% containment requires typically about 20X

0

Created by incident charged pion, kaon, proton, etc Typical composition:

50% EM (e.g. ) 25% Visible non-EM energy 25% invisible energy (nuclear break-ups)

Requires longer containment (expressed in λ)

Hadronic showers

0

Calorimeter detectors

Detector hardware must: Favor shower development Collect deposited energy

Can do both at the same time (e.g. BaBar/Belle crystal calorimeters)

Or have calorimeters with alternating passive and sensitive material

Example of electron shower with lead absorber:

Sampling calorimetry (Ex.: CDF)

Scintillators (sensitive material) emit lights with passage of ionizing particles

Collect light deposited in sensitive material using wavelength shifter (WLS)

WLS → photomultipliers that convert light into electric signal

CDF Calorimeters Segmentation

Calorimeter is segmented into towers that are read-out independently

Lead (iron) interspersed with scintillators for EM (HAD) calorimeters

Each central tower covers

Each tower has an EM calorimeter followed by an HAD calorimeter

151.0

CDF Calorimeters

Three regions: central, wall and plug

Use “projective” geometry Designed to measure

electrons, photons, quarks, gluons, hadrons, neutrinos

Note: design of calorimeter performed with a simulation of the most important processes you plan to measure (ex.: Higgs at LHC)

Construction

Go on and built the thing after it is designed! Many institutions in the world participate

First calibration: test beam

Take one calorimeter “wedge”, send beam of particles with known energy

Obtain correspondence detector response → energy in GeV

A few towers only submitted to test beam Set absolute scale for all

towers Relative scale for other

towers obtained later

Wedge getting ready to receive beam:

How does the test beam works (Ex.: plug calorimeter)

Performed at Fermilab meson beam facilities

Beams characteristics: Various types for EM and HAD

showers: electrons, pions, muons Various energy: 5-120 GeV (electrons),

5-220 GeV (pions)

Beams can be contaminated → bias the calibration constants E.g. use Cherenkov detector in front of

calorimeter to identify proton contamination in pion beam

WhyMuons?

Calorimeter response linearity

Extract calibration constant for many energy point

Can test linearity of calorimeter

Can add “artificial” material in front of calorimeter to simulate tracker+magnet material

Send pions and electrons to hadronic calorimeter Why sending

electrons in hadron

calorimeter?

Performance determined from test beam

• From RMS of tower response to same beam energy → measure calorimeter resolution

• Can test tower transverse uniformity (influences resolution)

• Stochastic term resolution:– EM:

– HAD: EEE /%80/

EEE /%14/

Final detector assembly: getting ready for physics!

The Tevatron

Proton-antiproton collisions at Most energetic collider in the

world Collisions every 0.4 μs Circumference of 6.3 km

TeVs 96.1

The CDF Detector

CDF II: general purpose CDF II: general purpose solenoidal detectorsolenoidal detector

7 layers of silicon tracking – Vertexing, B-tagging

COT: drift chamber– coverage – Resolution:

Muon chambers– Proportional chamber

interspersed with absorber– Provide muon ID up-to

Calorimeters Central, wall, plug calorimeter

1|| %1.0/ 2 Tp p

T

5.1||

Calibration when the detector is installed

Only a few towers saw test beam, how to calibrate the whole thing?? Test beam sets the absolute scale as a function energy Two solutions:

Hardware calibrations Physics calibration (using collider data)

These calibrations need to: Cross-check absolute scale (e.g. test beam not 100% realistic) Track detector response through time

Expected degradation of scintillator and PMT PMT sensitive to temperature

Uniform response through all towers

Hardware calibration

Can use radioactive sources that have very well defined decay energy Cobalt 60 (2.8 MeV) Cesium 137 (1.2 MeV)

Source calibration can be performed between colliders run

Sources are movable and can expose one tower at a time

Check uniformity over all towers and over time

Sources are sensitive to both scintillator and PMT responses

Laser calibrations

The lasers are connected directly to PMTs Skip scintillator/WLS steps

Used to uniformize PMTs response over towers and time

Physics calibrations

Use real collider data For calibration, you have to have some “known” and

some “unknown” (the calorimeter response) Examples of “known” information:

Mass of a well-known particles Ex.: Z→ee (Z mass measured at LEP)

Energy deposited by muons over a given length Muon sample

Energy measured in tracker (assuming tracker in calibrated) Redundant to energy measured in calorimeter for electrons

Example: Z boson mass

Z mass measured with great accuracy at LEP using beam energy

Background is very small for Z→ee

Sample is relatively small, but good enough

Z mass peak:

Example: E/p of electrons

Used for relative scale over towers

Cannot be used in forward region (no tracker)

In plug: rely on sourcing and lasers

Example: muons for HAD calorimeters

Muon calibrate detector response to ionizing energy

Use muon from J/ψ for identification (mass not used like Z boson)

Again, not used for PHA (rely on sourcing, laser)

Physics with photons/electrons

Calorimeter calibration not the only issue

Electron/photon physics also rely on tracking

Removal of background E.g. remove pion

background by studying shower shape

Search for new physics: Z' candidate:

Precision measurement: W mass:

What are jets?

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 of towers is correlated with the momentum of the original quark and lepton

Why not using tracker

(has better resolution)?

Phenomenology of jets

Quark/gluon produced from ppbar interaction

Fragmentation into hadrons Jets clustering algorithm:

Adds towers inside cone

Fraction of energy is out-of-cone

Underlying event contributes

Jet versus calorimeter energy scale

Jets are complicated processes Previous calorimeter calibrations are 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) Jet cross-sections (comparison to QCD predictions)

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

Relative energy scale

Jet energy measurement depend on location in detector

True even after all previous calibrations!

How come? Jets are wide Some regions of CDF

calorimeter are not instrumented

Relative energy scale: Use QCD dijet events Should have equal

transverse momentum

Absolute energy scale

Response to single pion non-linear (in test beam)

However, jets are identified as one single objects

For a 50 GeV jet: calibration is not the same whether: One 50 GeV pion 10 times 5 GeV pions

Solution: Get the average energy scale

Simulate an “average” particles configuration inside jet

Use test beam information to get calibration factor for single particles

Out-of-cone energy

Cone of fixed radius used to identify jets

Need to correct for fraction of energy out-of-cone (typically 15%)

This is mostly physics related How well is the physics

generator representing fragmentation?

Underlying event energy

Proton/antiproton remnants splash energy in calorimeter

Spoils jet energy measurement

Depends on the number of ppbar interaction per event

Extracted from “minimum bias” events

Small effect: ~0.4 GeV per jet

Final jet energy scale uncertainty

Estimate of jet energy scale uncertainty is important to estimate systematic uncertainties of measurements

Dominated by out-of-cone (low-pT) and absolute energy scale (high-pT)

Ranges from 10% to 3% energy uncertainties

Example physics measurement: top quark mass

Top produced in pairs at Tevatron

Top decays to W boson and b-quark 100% of time in SM

Typical event selections: Well-identified electron(s) or

muon(s) Large missing ET Several reconstructed jets

identified in calorimeters

Note: 4 jets in final state!

Identification of b-quark jets

• Complicated final state:

• Which jets come from which parton?

• Can identify b-quark jets using one characteristic: – Long b-quark lifetime

• Note: lots of semileptonc B-hadrons decay (involving neutrino)– Require special b-jets

calibration

jjjjltt

Top mass reconstruction

Event-by-event kinematic fitter (assumes event is ttbar)

Attempts all jet-parton assignments Assign b-tag jets to b-

quarks

The one most consistent with ttbar hypothesis is kept

More correct combinations with b-tags!

The strategy

Construct reconstructed top mass distributions for many true top mass So-called “templates”

Compare distribution reconstructed in data with templates Using likelihood fit

Account for background contamination Dominated by W+jets

production

The measurement (spring 2005)

Using 138 candidate ttbar events, fit yields:

Mtop= 173.2 +2.9/-2.8 (stat.) +/- 3.4 (syst.) GeV/c2

By shifting by JES uncertainty defined before: Mtop changes by 3.1 GeV/c !

JES uncertainty limiting factor for Mtop measurement

Improvement: W→jj calibration

Inside ttbar events, invariant mass of two jets from W boson decay should equal MW

Can use W→jj decays to further constraint JES

Use same data for measurement and calibration… cheating?? No: Mjj (almost) independent

of Mtop

Remaining correlations are accounted for

The measurement(adding W→jj information)

Using same dataset as previously:

Mtop= 173.5 +2.7/-2.6 (stat.) +/- 2.8 (syst.) GeV/c2

Total Mtop uncertainty improved by 10%

JES uncertainty decreased by 20%

Good prospect for future

Impact of Mtop measurement

Mtop, MW connected to Higgs boson mass through radiative corrections

MH< 186 GeV/c2 @ 95%C.L.

Can constrain mass of supersymmetric particles

Conclusion

Detector calibration needed to translate detector response in energy

Various techniques used for calorimetry: Test beam Radioactive sources Lasers Collider data

Calorimeter can be used to measure: Electrons, photons, jets, missing ET

Good calorimeter and jet calibration needed for measurements like top quark mass