search of higgs to invisible decays with the atlas detector · in atlas, the search is performed in...
Post on 22-May-2020
0 Views
Preview:
TRANSCRIPT
Search of Higgs to invisible decayswith the ATLAS detector
Stefano Rosati
INFN Roma
5 July 2014
On behalf of the ATLAS Collaboration
Stefano Rosati (INFN Roma) ICHEP 2014 - Valencia 5 July 2014 1 / 15
Motivation
Various extensions of the Standard Model (SM) allow Higgs bosondecays to a pair of stable or long-lived particles
Observation of a large Branching Ratio to invisible decays can meanevidence of New Physics BSM
For example, particles with very low interaction with SM particles, e.g.Dark Matter through the so-called Higgs-portal model
LHC data collected by the ATLAS experiment can be used to constrainthe BR of the Higgs boson @125.5 GeV to invisible particles
Limits can also be set on σ·BR to invisible particles of any additionalHiggs boson over a wide mass range
Stefano Rosati (INFN Roma) ICHEP 2014 - Valencia 5 July 2014 2 / 15
Introduction and outlineIn ATLAS, the search is performed in Higgsassociated production with a Z decaying toleptons, Z→ll (l=e,µ).Signature is a pair of isolated leptons withmass close to the Z, plus missingtransverse energyResults based on the full LHC Run-1dataset of 4.5 fb−1 at 7 TeV and 20.3 fb−1
at 8 TeVPhys. Rev. Lett. 112, 201802 (2014)
q
q
ZH χ
χ
Z
ℓ−
ℓ+
Outline of the talkEvent selectionBackgrounds and main systematic errorsLimits on the cross section and on the invisible decayDark Matter interpretation in the Higgs-portal modelIndirect limits
Stefano Rosati (INFN Roma) ICHEP 2014 - Valencia 5 July 2014 3 / 15
Event Selection - 1Di-lepton invariant mass consistent with a Z decay
I 76 < mll < 106 GeV
Missing ET >90 GeV to remove most of the Z+jets backgroundCorrelation between azimuthal angles of Emiss
T and track-based missingmomentum Pmiss
T :I Avoids Emiss
T from misreconstructed calo energy, pileupI ∆φ(Emiss
T ,PmissT ) <0.2
Eve
nts
1
10
210
310
410
510
610
710
810
→ ℓ ℓℓ
→ ℓℓ ℓ
→ ℓℓ →
-1 L dt = 20.3 fb = 8 TeV, s
ATLAS
→ ℓℓ
0 50 100 150 200 250 300 350 400 450 500Da
ta / M
C
0.5
1
1.5
/8E
ve
nts
/
1
10
210
310
410
510
610
710
→ ℓ ℓℓ
→ ℓℓ ℓ
→ ℓℓ →
-1 L dt = 20.3 fb = 8 TeV, s
ATLAS
→ ℓℓ
ϕ
0 0.5 1 1.5 2 2.5 3Da
ta / M
C
0
1
2
Stefano Rosati (INFN Roma) ICHEP 2014 - Valencia 5 July 2014 4 / 15
Event Selection - 2Require the Z momentum to balance the Emiss
TI ∆φ(pll
T,EmissT ) >2.6
Cut on the fractional pT differenceI |Emiss
T − pllT|/pll
T <0.2Exploit the boost of the Z
I ∆φ(l , l) <1.7Veto jets with pT>25 GeV and |η| <2.5
I Mostly against t t̄ and Z+jets
/8E
ve
nts
/
1
10
210
310
410
510
610
710
→ ℓ ℓℓ
→ ℓℓ ℓ
→ ℓℓ →
-1 L dt = 20.3 fb = 8 TeV, s
ATLAS
→ ℓℓ
ϕ ℓℓ
0 0.5 1 1.5 2 2.5 3Da
ta / M
C
0.5
1
1.5
/16
Eve
nts
/
10
210
310
410
510
→ ℓ ℓℓ
→ ℓℓ ℓ
→ ℓℓ →
-1 L dt = 20.3 fb = 8 TeV, s
ATLAS
→ ℓℓ
ϕ ℓ ℓ
0 0.5 1 1.5 2 2.5 3Da
ta / M
C
0.5
1
1.5
Stefano Rosati (INFN Roma) ICHEP 2014 - Valencia 5 July 2014 5 / 15
BackgroundsDi-boson processes (ZZ, WZ) dominate the backgrounds after theselection
I Determined from NLO MC, validated in control regionsInclusive Z→ ee, Z→ µµ estimated in three sideband regions withreverted ∆φ(Emiss
T ,PmissT ) and fractional pT difference cuts
Other backgrounds with an isolated same-flavour lepton pair(WW,tt,Wt,Z→ ττ )
I Estimated in a control region of events with an eµ pair
0 50 100 150 200 250 300 350
Eve
nts
/ 20
GeV
1
210
410
610
810
1010
1210 ATLAS-1 L dt = 20.3 fb∫=8 TeV, s
ZH → ℓℓ + inv.
Data
µµee, → Z
Other BG
Top quark
WW
WZ → ℓνℓℓ (incl.τ)
ZZ → ℓℓνν (incl.τ)
ZH → ℓℓ + inv., BR(H → inv.) = 1
Signal Region∆ϕ(E
missT , p
missT ) < 0.2, |E
missT − p
ℓℓT | /p
ℓℓT < 0.2
[GeV]missTE
0 50 100 150 200 250 300 350
Dat
a / M
C
0.5
1
1.50 50 100 150 200 250
Eve
nts
/ 20
GeV
1
210
410
610
810
1010
1210 ATLAS-1 L dt = 20.3 fb∫=8 TeV, s
ZH → ℓℓ + inv.
Data
µµee, → Z
Other BG
Top quark
WW
WZ → ℓνℓℓ (incl.τ)
ZZ → ℓℓνν (incl.τ)
ZH → ℓℓ + inv., BR(H → inv.) = 1
Sideband Region∆ϕ(E
missT , p
missT ) > 0.2, |E
missT − p
ℓℓT | /p
ℓℓT > 0.2
[GeV]missTE
0 50 100 150 200 250
Dat
a / M
C
2
4
Stefano Rosati (INFN Roma) ICHEP 2014 - Valencia 5 July 2014 6 / 15
Selection results and systematicsNumber of expected and observed events
For the signal, the momentum of the reconstructed Zboson is expected to be balanced by the momentum of theinvisibly decaying Higgs boson. Therefore the azimuthalseparation between the dilepton system, where the magni-tude of its transverse momentum is defined as pll
T , and theEmissT , !!!pll
T ; EmissT ", is required to be greater than 2.6. The
boost of the Z boson causes the decay leptons to be producedwith a small opening angle. The azimuthal opening angle ofthe two leptons, !!!l;l", is thus required to be less than1.7. Furthermore pll
T and EmissT are expected to be similar.
Therefore the fractional pT difference, defined asjEmiss
T ! pllT j=pll
T , is required to be less than 0.2. Finally,for the majority of the signal no additional high-pT jets areexpected to be observed in the events, while for the back-ground from boosted Z bosons and from tt̄ pairs one or morejets are expected. Thus, events are required to have noreconstructed jets with pT > 25 GeV and j"j < 2.5.After the selection requirements, the dominant back-
ground is SM ZZ production followed by SM WZproduction, as shown in Table I. These backgrounds aresimulated using MC samples normalized to NLO crosssections. The simulation of WZ events is validated bycomparing them to data events in which the third-leptonveto is replaced by an explicit third-lepton requirement.The theoretical prediction of the ZZ production is inagreement with the ATLAS cross-section measurementat
!!!s
p# 7 TeV [50].
Background contributions from events with a genuineisolated lepton pair, not originating from a Z ! ee or Z !## decay (WW, tt̄, Wt, and Z ! $$), are estimated byexploiting the flavor symmetry in the dilepton final state ofthese processes. Distributions for events with an e# pair,appropriately scaled to account for differences in electronand muon reconstruction efficiencies, can be used toestimate this background in the electron and muon chan-nels. The difference between the efficiencies for electronsand muons is estimated using the square root of the ratio ofthe numbers of dimuon and dielectron events in data withinthe mll window. Events in the e# control region notoriginating from WW, tt̄, Wt, or Z ! $$ backgrounds aresubtracted using simulated samples. Important sources of
systematic uncertainty are variations in the correction factorfor the efficiencies for electrons and muons and uncertain-ties in the simulated samples used for the subtraction. Thecombined systematic uncertainty is 23% for both the 7 and8 TeV data. The estimated background from these sourcesis consistent with the expectation from the simulation.The background from inclusive Z ! ee and Z ! ##
production in the signal region is estimated from the back-ground in three sideband regions [51].These sideband regionsare formed by considering events failing one or both of thenominal selection requirements applied to !!!Emiss
T ; pmissT "
and the fractional pT difference. Contributions from non-Zbackgrounds in the sideband regions are subtracted. Theimpact from a correlation between the above two variables isdetermined from the simulation and a correction, of at most7%, is applied to account for it. Themain uncertainties are dueto variations in this correction and differences in the shape ofthe Emiss
T distribution in the control regions. The overallsystematic uncertainty is 52% in the 7 TeV data and 59% inthe 8 TeV data.The small background from events with only one genuine
isolated lepton (inclusive W, single-lepton top pairs andsingle top production) or from multijet events is estimatedfrom data using control samples, selected by requiring twolepton candidates of which at least one fails the full leptonselection criteria. These samples are scaled with a measuredpT-dependent factor, determined from data as described inRef. [52]. Systematic uncertainties are determined followingthe procedures used in Ref. [52], yielding an uncertainty of40% in the 7 TeV data and 21% in the 8 TeV data.Systematic uncertainties on the signal and the SM ZZ
and WZ backgrounds are derived from the luminosityuncertainty, the propagation of reconstructed object uncer-tainties, and from theoretical uncertainties on the produc-tion cross sections. The luminosity uncertainty is 1.8% forthe 7 TeV data-taking period and 2.8% for the 8 TeV data-taking period [53].Lepton trigger and identification efficiencies as well as
the energy scale and resolution are determined from datausing large samples of Z events. After appropriate correc-tions to the simulation, uncertainties are propagated to the
TABLE I. Number of events observed in data and expected from the signal and from each background source forthe 7 and 8 TeV data-taking periods. Uncertainties on the signal and background expectations are presented withstatistical uncertainties first and systematic uncertainties second.
Data period 2011 (7 TeV) 2012 (8 TeV)
ZZ ! ll%% 20.0$ 0.7$ 1.6 91$ 1$ 7WZ ! l%ll 4.8$ 0.3$ 0.5 26$ 1$ 3Dileptonic tt̄, Wt, WW, Z ! $$ 0.5$ 0.4$ 0.1 20$ 3$ 5Z ! ee, Z ! ## 0.13$ 0.12$ 0.07 0.9$ 0.3$ 0.5W % jets, multijet, semileptonic top 0.020$ 0.005$ 0.008 0.29$ 0.02$ 0.06Total background 25.4$ 0.8$ 1.7 138$ 4$ 9Signal (mH # 125.5 GeV, &ZH;SM, BR!H ! inv:" # 1) 8.9$ 0.1$ 0.5 44$ 1$ 3Observed 28 152
PRL 112, 201802 (2014) P HY S I CA L R EV I EW LE T T ER Sweek ending23 MAY 2014
201802-3Main systematics uncertainties:I Luminosity uncertainty (1.8-2.8%)I Lepton trigger and reconstruction uncertainties (1.0-1.5%)I Pileup uncertainty on Emiss
T (1-2%)I Jet energy scale and resolution (3-6%)I Total uncertainties on ZZ/WZ backgrounds (8-13%)I Theoretical uncertainties on ZH cross section (3.6-5.7%) and Higgs pT boost
(1.9%)
Stefano Rosati (INFN Roma) ICHEP 2014 - Valencia 5 July 2014 7 / 15
Limits settingsThe Emiss
T distribution after the full selection is used to extract the limits onthe cross section times Branching Ratio to invisible
I Use a Maximum Likelihood fit following the CLS formalism with profilelikelihood
I Consider MH in the range 110-400 GeV
Systematics as nuisance parameters constrained by Gaussiandistributions
Data vs MC @ 7 TeV Data vs MC @ 8 TeV
100 150 200 250 300 350 400 450
Eve
nts
/ 30
GeV
-210
-110
1
10
210
310 ATLAS-1 L dt = 4.5 fb∫ = 7 TeV, s
ZH → ℓℓ + inv.
Data
ZZ → ℓℓνν (i�cl. τ)
WZ → ℓνℓℓ ( i�cl. τ)
WW, dilep. ��̄, W�, Z → ττ
µµee, → Z
W + jets, multijet, semilep. top
ZH → ℓℓ + inv., BR(H → inv.) = 1
[GeV]missTE
100 150 200 250 300 350 400 450Dat
a / E
xpec
ted
1
2
3 100 150 200 250 300 350 400 450
Eve
nts
/ 30
GeV
1
10
210
310 ATLAS-1 L dt = 20.3 fb∫ = 8 TeV, s
ZH → ℓℓ + inv.
Data
ZZ → ℓℓνν (incl. τ)
WZ → ℓνℓℓ (incl. τ)
WW, dilep. ��̄, W�, Z → ττ
µµee, → Z
W + jets, multijet, semilep. top
ZH → ℓℓ + inv., BR(H → inv.) = 1
[GeV]missTE
100 150 200 250 300 350 400 450Dat
a / E
xpec
ted
0.51
1.52
Stefano Rosati (INFN Roma) ICHEP 2014 - Valencia 5 July 2014 8 / 15
ResultsLimits on the σ·BR to invisible particles are given for MH in the range110-400 GeV
I In this case, no assumption on the production cross sectionI No significant deviation from the SM expected limit is observed
Assuming the predicted SM ZH production cross section, with MH=125.5GeV, an upper limit can be set on the BR
I Observed limit at 95% C.L. is 75%I Expected in absence of invisible decays is 62%
[GeV]Hm
150 200 250 300 350 400
inv.
) [fb
]→
H B
R(
× Z
Hσ
0
100
200
300
400
500
600
,SMZHσ
Observed 95% CL limit
Expected 95% CL limit
σ1±
σ2±
ATLAS
ZH → ℓℓ + inv.
-1 L dt = 4.5 fb∫ = 7 TeV, s-1 L dt = 20.3 fb∫ = 8 TeV, s
inv.) →H BR(
0 0.2 0.4 0.6 0.8 1
1-C
L
-210
-110
1ObservedExpected
68% CL
95% CL
ATLAS
ZH → ℓℓ + inv.
-1 L dt = 4.5 fb∫ = 7 TeV, s-1 L dt = 20.3 fb∫ = 8 TeV, s
Stefano Rosati (INFN Roma) ICHEP 2014 - Valencia 5 July 2014 9 / 15
Dark Matter in the Higgs portalThe results can be interpreted in the context of a Higgs-portal DMscenario
I Higgs as mediator between DM and SM particlesForm factor associated to Higgs nucleon coupling set to 0.33+0.3
−0.07 (Phys.Lett. B 709, 65 (2012))Assume that DM decays account for the full invisible BRVery stringent limits at low mass, degrading as the DM massapproaches MH/2
DM Mass [GeV]1 10 210 310
]2N
ucle
on c
ross
sec
tion
[cm
−D
M
-5110
-5010
-4910
-4810
-4710
-4610
-4510
-4410
-4310
-4210
-4110
-4010
-3910
-3810
-3710
σDAMA/LIBRA 3 σCRESST 2CDMS 95% CL CoGeNTXENON10 XENON100LUX ATLAS, scalar DMATLAS, vector DM ATLAS, fermion DM
ATLAS = 7 TeV,s ∫ -1Ldt=4.5 fb = 8 TeV,s ∫ -1Ldt=20.3 fbZH → ℓℓ + inv.
Higgs-portal Model
Stefano Rosati (INFN Roma) ICHEP 2014 - Valencia 5 July 2014 10 / 15
Indirect limitsIndirect limits on the BR to invisiblecan be derived from the combinationof rates in γγ,ZZ ∗ → 4l ,WW → lνlν,ττ,bb̄ channels
I Fix H couplings to SM values
Add contribution to the total widthfrom invisible, undetected decays
I ΓH = k2HΣi
ki(1−BRi,u)
ΓSMH
Introduce effective couplings to γ andgluons (possible new particles inloops)
I Fit with κγ , κg and BRi,u free
Observed limit BRi,u<0.41 at 95%C.L. (expected from SM BRi,u <0.55)If ZH→ ll + Emiss
T is included in the fitthe limit becomes BRi,u <0.37(expected BRi,u <0.39)
iBR-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
) i(B
RΛ
-2 ln
0
2
4
6
8
10
12
14 PreliminaryATLAS
-1 Ldt = 4.6-4.8 fb∫= 7 TeV, s
-1 Ldt = 20.3 fb∫= 8 TeV, s
]i
, BRgκ, γκ[ : b, bττ, ZZ*, WW*, γγ →h
,b, bττ, ZZ*, WW*, γγ →h
:Tmiss ll + E→Zh
obs. exp.
obs. exp.
i,uBR
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7)
i,u(B
RΛ
-2 ln
0
2
4
6
8
10 ATLAS Preliminary-1Ldt = 4.6-4.8 fb∫ = 7 TeV, s
-1Ldt = 20.3 fb∫ = 8 TeV, s
b,bττ,ZZ*,WW*,γγ →Combined H
> 0i,u
Limiting to physical range BR
]i,u
,BRgκ,γκ[
Observed
SM expected
ATLAS-CONF-2014-09ATLAS-CONF-2014-10
Stefano Rosati (INFN Roma) ICHEP 2014 - Valencia 5 July 2014 11 / 15
Summary and conclusions
A direct search for invisible decay modes of the Higgs boson with theATLAS experiment has been presented
I Based on the full Run 1 LHC dataset of 4.5 fb−1 at 7 TeV and 20.3 fb−1 at 8TeV collected with the ATLAS experiment
No deviation from the SM expectation is observed in the σ·BR measuredover the mass range 110-400 GeV
Assuming the SM ZH production cross section @ MH=125.5 GeV, thedirect search allows to set an upper limit of 75% at 95% C.L. on the BR toinvisible particles
I Limit at 37% from the combination of observed rates in all SM channels andupper limits from the direct search
In the Higgs-portal Dark Matter scenario, the results have beeninterpreted interpreted as limits on low-mass Dark Mattercandidates-nucleon cross sections
Stefano Rosati (INFN Roma) ICHEP 2014 - Valencia 5 July 2014 12 / 15
BACKUP SLIDES
Stefano Rosati (INFN Roma) ICHEP 2014 - Valencia 5 July 2014 13 / 15
Other selection variables
Fractional pT differenceI |Emiss
T − pllT|/pll
T <0.2
Number of jets
Eve
nts
1
10
210
310
410
510
610
→ ℓ ℓℓ
→ ℓℓ ℓ
→ ℓℓ →
-1 L dt = 20.3 fb = 8 TeV, s
ATLAS
→ ℓℓ
ℓℓ ℓℓ
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2Da
ta / M
C
0.5
1
1.5
Eve
nts
1
10
210
310
410
510
610
710
→ ℓ ℓℓ
→ ℓℓ ℓ
→ ℓℓ →
-1 L dt = 20.3 fb = 8 TeV, s
ATLAS
→ ℓℓ
0 1 2 3 4 5Da
ta / M
C
0
1
2
Stefano Rosati (INFN Roma) ICHEP 2014 - Valencia 5 July 2014 14 / 15
Dark Matter limits from the indirect search
95% confidence level limits from the combined fit of SM channels ratesand upper limits from the direct search of invisible decays
[GeV]χm1 10 210 310
]2 [c
m-Nχσ
-5710
-5510
-5310
-5110
-4910
-4710
-4510
-4310
-4110
-3910
DAMA/LIBRA (99.7% CL)CRESST (95% CL)CDMS (95% CL)CoGeNT (90% CL)XENON10 (90% CL)XENON100 (90% CL)LUX (95% CL)
Scalar WIMPMajorana WIMPVector WIMP
ATLAS Preliminary
Higgs portal model:ATLAS (95% CL) in
-1dt = 4.6-4.8 fbL∫ = 7 TeV, s-1dt = 20.3 fbL∫ = 8 TeV, s
,νlνl→WW*→4l, h→ZZ*→, hγγ→hmiss
Tll+E→bb, Zh→, hττ→h
Stefano Rosati (INFN Roma) ICHEP 2014 - Valencia 5 July 2014 15 / 15
top related