1
17/05/2017
Seminario finale di dottorato
Giuseppe Fasanella
“La Sapienza” University of Rome & INFN Roma 1Co-supervision with
ULB Université Libre de Bruxelles
Search for new physics in dielectron and diphoton final states at CMS
Direttori di tesi: Dr Paolo Meridiani Prof. Barbara Clerbaux
Giuseppe Fasanella, ULB and INFN Roma I
2
Outline
● Introduction
● Motivation
● LHC and the CMS detector
● Heavy (neutral) resonance searches- dielectron channel- diphoton channel
● Conclusions
Giuseppe Fasanella, ULB and INFN Roma I
3
The present picture
+ Higgs boson (2012)
The standard model (SM) of particle physics includes:● all known leptons and quarks ● the force carriers of 3 of the 4 fundamental forces of Nature● The Higgs boson (which accounts for the particles masses)
The question is: “Is there something else?”
Giuseppe Fasanella, ULB and INFN Roma I
4
MotivationThe searches for new physics in the dielectron/diphoton final state are● Theoretically motivated: new heavy neutral resonances predicted in several
BSM scenarios (GUT, extra-dimensions, theories with non-minimal Higgs sectors)● Experimentally motivated: very clean channels, hence high discovery potential
Signal Signature: Excess of events (peak) or a distortion in the invariant dielectron/diphoton invariant mass distribution compared to the SM processes
Giuseppe Fasanella, ULB and INFN Roma I
M [GeV]
# Ev
en
ts
# Ev
en
tsM [GeV]
5
The LHC
The LHC (Large Hadron Collider), located under the Geneva area, is:● Hadron accelerator and collider (only proton-proton collisions considered here)● Design center-of-mass energy of 14 TeV● 27 km of circumference● It hosts 4 main experiments (ALICE, ATLAS, CMS, LHCb)● After the Higgs discovery, the goal is the search for new physics beyond the SM
In my thesis:● The searches for new physics in the dielectron and diphoton are presented● Different datasets are analyzed colled by the CMS experiment during 2015-2016
Giuseppe Fasanella, ULB and INFN Roma I
6
The CMS detector
The main detector for the final states targeted in these analyses is the e.m. calorimeter (ECAL):made-up of ~76k scintillating crystals
Giuseppe Fasanella, ULB and INFN Roma I
7
Particles in CMS
Giuseppe Fasanella, ULB and INFN Roma I
● Each particle has a specific signature in the CMS detector.● Both electrons/photons release most of their energy in the ECAL. ● Electrons also leave hits in the silicon Tracker.● Quarks appear in the detector as “jets” (spray of collimated particles)
with high hadronic activity.
8
Electrons and photons in CMS
Giuseppe Fasanella, ULB and INFN Roma I
Electrons (photons) identification based● on isolation: search for isolated ECAL deposit with (or without)
associated track.● on shape: the energy of the impacting particle is shared by a cluster of
crystals, with a peculiar form and a certain spread in due to the magnetic field
9
Search for high mass resonances (Z ') decaying in dielectron final state
Giuseppe Fasanella, ULB and INFN Roma I
10
Event Selection● High energy electron pairs (HEEP) selection is used● Cut-based selection designed to be highly efficient at high ET
● Events categories: Barrel-Barrel (BB) or Barrel-Endcap (BE)● The pair with the highest invariant mass is selected● The selection efficiency is studied with data-driven techniques
Giuseppe Fasanella, ULB and INFN Roma I
11
Z' to dielectron: backgroundsThree main types of SM backgrounds (BG) in the di-electron channel→ All 3 of them resulting in a falling and continuous BG vs mass
● The most significant one is the irreducible SM Drell-Yan (DY) process● Background predicted using simulations
● The second most important BG comes from real electrons in processeswith W and Z bosons involved (WW, tt, tW, ZZ,...)● These processes are flavor-symmetric:● Verify the simulations by looking at the e-mu spectrum
● The third type of background is the jet background, where one or more jet is misidentified as an electron (di-jets events, W + jets …)● Estimated directly with data (Fake Rate method)
Giuseppe Fasanella, ULB and INFN Roma I
12
Energy scale & resolution corrections
Giuseppe Fasanella, ULB and INFN Roma I
For narrow resonances, experimental peak width dominated by calorimeter energy resolution → need to control scale & resolution of the ECAL
Derived in 2 steps:● Data/simulation comparisons at the Z-peak region (pT~45 GeV)
● ~15 M Zee events: it is possible to define several kinematical regions● the scale in data is matched to the one in simulation ● the resolution in simulation is broaden to match the one in data
EBEB EBEE
13
Energy scale & resolution corrections
Giuseppe Fasanella, ULB and INFN Roma I
For narrow resonances, experimental peak width dominated by calorimeter energy resolution → need to control scale & resolution of the ECAL
Derived in 2 steps:● Data/simulation comparisons at the Z-peak region (p T~45 GeV)● On top of that, need to control the pT spectrum going from Z-peak region to
unknown territory (several effects could lead to non-linearity: electronics, shower containements, energy losses, ….)
● A clear trend was exposed at high-pT due to events acquired with different gains → solved
14
Dielectron mass spectra
● Observed mass spectra are compatible with the SM-only hypothesis
Giuseppe Fasanella, ULB and INFN Roma I
15
Observed limits● 95% confidence level limit are set on the (normalized) production cross-section for
a heavy resonance decaying in dielectron final state● Two models considered: a Z' particle with couplings identical to the SM ones (Z'SSM)
and a Z' particle coming from GUT theory (Z'ψ)
Giuseppe Fasanella, ULB and INFN Roma I
16
Search for high mass resonances decaying in diphoton final state
Giuseppe Fasanella, ULB and INFN Roma I
17
Event selection● Many similarities with the dielectron case:
– Cut based selection targeting high-pT photons
● Main difference is the requirement that the cluster in the calorimeter should not be associated to a track in tracker
● Analysis results are interpreted in terms of a spin-0 and spin-2 resonance hypothesis
p1T
75 GeV
p2T
75 GeV
||max
<2.5
||min
<1.45
categorization EB-EBEB-EE
Giuseppe Fasanella, ULB and INFN Roma I
18
Backgrounds
Giuseppe Fasanella, ULB and INFN Roma I
Irreducible backgrounds from processes with real photons:
Reducible backgrounds: from +jets and di-jet processes:
Differently w.r.t the dielectron case, the background is fitted direcly from data, by modeling the mass distribution with an analytical function.
A bit of history
Giuseppe Fasanella, ULB and INFN Roma I
At mid-Dicember 2015, some excess of events (~3 standard deviations in the local p-value) have been reported by both ATLAS and CMS around the mass region ~750 GeV
2016
20
Diphoton invariant mass spectra(latest 2016 results)
EBEEEBEE
Giuseppe Fasanella, ULB and INFN Roma I
● Background fitted directly from data● Observed mass spectra are compatible with the SM-only hypothesis
21
Observed limits 2016
Resonance widthResonance width
Giuseppe Fasanella, ULB and INFN Roma I
● 95% confidence level limit are set on the production cross-section for a heavy resonance decaying in diphoton final state
22
Results: p-values 2015+2016
SPIN0
SPIN2
WIDTHWIDTH
Spin 0
Spin 2
23
Conclusions
● Searches for new resonances both in dielectron and diphoton final states have been presented
● The analyses, thanks to the increase in the center-of-mass energy from 8 TeV to 13 TeV, explored a way wider region w.r.t Run1 data-taking
● No significant excesses seen in data over the SM-only expectations
● Dielectron resonance search:● Mass below ~ 4 TeV are excluded at 95% CL
● Diphoton resonance search:● The 2015 dataset showed a mild excess of events (~2.9 standard
deviations in the local p-value) in the mass region ~750 GeV● It turned out to be simply a statistical fluctations, which
disappeared with more statistics
Giuseppe Fasanella, ULB and INFN Roma I
24
Back-up
Giuseppe Fasanella, ULB and INFN Roma I
Energy corrections vs ET
Giuseppe Fasanella, ULB and INFN Roma I
● Residual scale corrections vs ET are inside the 1% syst. Uncertainty band
● Note: scale corrections at the level of 0.5 % (for EB 3.8 T) vs 1.5 % (for EB 0T)→ extra 1% syst. uncertainty on scale is added for 0T dataset to take into account possible mismodelling of the difference in energy scale between the two dataset
● Smearing corrections are roughly constant vs ET
26
Interpretation of results
Three datasets considered:• 19.7/fb at √s=8TeV• 3.3/fb at √s=13TeV (2015)● 12.9/fb at √s=13TeV (2016)
Combinations:● 2015 + 2016● 8 + 13 TeV
27
General calibration strategy:
● Identification and reconstruction of leptons/photons with very high transverse momentum (pT)
● Since we are searching for a peak at high mass, it is important to control the electron energy resolution in this regime
The strategy is divided in 2 steps:1) Calibrate the detector response at the Z peak (pT~45 GeV):
Comparison of the data-MC Z peak (scale and broadness)2) After applying 1), we inspect the high mass region:
Either via MC-only or looking at boosted Z events
Giuseppe Fasanella, ULB and INFN Roma I
28
Giuseppe Fasanella, ULB and INFN Roma I
29
Energy scale & resolution corrections
Giuseppe Fasanella, ULB and INFN Roma I
Important input for both analyses: the signal shape is convoluted with the detector resolution
Derived in 2 steps:● Data/simulation comparisons at the Z-peak region (p T~45 GeV)● On top of that, inspect the high-pT region → the target is to control the
energy scale at the level of ~1% for pT >~ 300 GeV ● with the level of statistics cumulated in 2015, the region up-to pT >~150
GeV was inspected
30
Energy scale & resolution corrections
Giuseppe Fasanella, ULB and INFN Roma I
For narrow resonances, experimental peak width dominated by calorimeter energy resolution → need to control scale & resolution of the ECAL
Derived in 2 steps:● Data/simulation comparisons at the Z-peak region (p T~45 GeV)● On top of that, need to control the pT spectrum going from Z-peak region to
unkown territory (several effects could lead to non-linearity: electronics, shower containements, energy losses, ….)
● A clear trend was exposed at high-pT
31
Observed limits● 95% confidence level limit are set on the (normalized) production cross-section for
a heavy resonance decaying in dielectron final state (left)● Two models considered: a Z' particle with couplings identical to the SM ones (Z'SSM)
and a Z' particle coming from GUT theory (Z'ψ)
Giuseppe Fasanella, ULB and INFN Roma I
Combined with dimuon channel
32
Event selection● Simple set of requirements
– Fixed pT cuts, at least one photon in the barrel region (EB: ||<1.45).
– Events categorized in barrel-barrel (EBEB) and barrel-endcap (EBEE) configurations.
– Efficient cut-based photon identification criteria.
● Per-photon efficiency in the barrel: 90%.
● Per-photon efficiency in the endcaps: 85%.
– Analysis results interpreted in terms of a spin-0 and spin-2 resonance hypothesis
p1T
75 GeV
p2T
75 GeV
||max
<2.5
||min
<1.45
categorization EB-EBEB-EE
Giuseppe Fasanella, ULB and INFN Roma I
33
Energy corrections
● Quite similar technique to the dielectron one (electrons and photons, expecially at high-pT , are quite similar objects)
● Data/simulation comparison shows very good agreement
Giuseppe Fasanella, ULB and INFN Roma I