Searches for Invisible Decay Modes of the Higgs Boson with
the CMS Detector
Daniele Trocino – Northeastern University on behalf of the CMS Collaboration
XXXVII International Conference on High Energy Physics Valencia (Spain) – July 4, 2014
Why Invisible Higgs?
●
Recent measurements of the branching fractions of the 125 GeV Higgs boson are compatible with the Standard Model expectations, but with large uncertainties
● can still accommodate significant non-SM decays
● moreover, additional Higgs bosons with exotic decays are still possible
●
Many models predict Higgs boson(s) decaying to heavy, undetectable particles
● fourth-generation neutrinos, neutralinos, graviscalars in extra-dimensions, etc
● some of these constitute viable dark matter (DM) candidates
● in “Higgs-portal” models, the DM does not couple to any SM particles but the Higgs
– Higgs as a mediator between the SM and DM sectors
CMS-PAS-HIG-13-005
updated results in M. Chen's talk
Indirect and Direct Searches
Γ
tot= Γ
totSM⋅ ∑
obsXκ
2X⋅ B
SMX1−B
BSM,
κX2 = ΓobsX ΓSMX
●
Non-SM decays modify the total Higgs width
● indirect constraints on BBSM can be inferred by fitting the observed decay modes X
●
Direct searches are conducted in production channels where the Higgs boson recoils
against a visible system
● vector boson fusion (VBF):
qq → qq VV → qq H → 2 jets + MET
● associated production with a Z boson, decaying to a pair of leptons or bb
qq → Z* → ZH → ℓ
+ℓ
–+ MET qq → Z* → ZH → bb + MET
CMS-PAS-HIG-13-005
68% CL 95% CL B
BSM< 0.52
updated results in M. Chen's talk
inv. H Z
VBF: Signal and Selection
Signature and selection
➢ Two forward jets, separated by a large rapidity gap, with high invariant mass
– 2 jets with pT > 50 GeV/c, |η| < 4.6 – η1·η2 < 0, Δη > 4.2, M(jj) > 1100 GeV/c2 – central-jet veto (CJV): no additional jets
with η1 < η < η2 (pT > 30 GeV/c)
➢ Large missing transverse energy (MET)
– MET > 130 GeV
Backgrounds and rejection
➢ Z(νν) + jets
➢ W(ℓν) + jets, DY(ℓℓ) + jets
– lepton veto: no ℓ with pT > 10 GeV/c
➢ Single top, tt, diboson
➢ QCD multijets
jet
jet MET
arXiv:1404.1344
VBF: Backgrounds
Z(νν) + jets estimated from Z(μμ) + jets events
● same selection as for the signal +
● exactly 2 muons with M(μμ) = 60-120 GeV/c2
● control→signal extrapolation factor from MC
NZ( ν ν) =
99
±29 (stat)
±25 (syst)
W(ℓν) + jets estimated from single e / μ / τ
hadevents
● control→signal extrapolation factor from MC
NW(eν ) =
63
±9 (stat)
±18 (syst)
NW(μ ν) =
67
±5 (stat)
±16 (syst)
NW(τ ν) =
53
±18 (stat)
±18 (syst)
QCD estimated from sidebands of two uncorrelated variables: MET and CJV (“ABCD” method)
MET 130 GeV
CJV
pass fail
signal
B A
D C
control
NQCD =
30.9
±1.6 (stat)
±23.0 (syst)
CMS-PAS-HIG-13-013
VBF: Results
● Mass independent selection, cut-and-count approach
● Using a CLS method, we set limits on σ × B(H→invisible) and σ × B(H→invisible) / σVBF
● Assuming SM VBF production cross section and acceptance:
observed 95% CL limit on B(H→inv) for mH = 125 GeV/c2 : 0.65 (expected: 0.49)
Final yields
125 GeV/c
125 GeV/c
arXiv:1404.1344
Z(ℓℓ) + H(inv): Signal and Selection
Signature and selection
➢ Two isolated leptons from Z decay
– 2 leptons (e, μ) with pT > 20 GeV/c, v|η| < 2.5 – M(ℓℓ) = MZ ± 15 GeV/c2
➢ Large missing transverse energy (MET)
– MET > 120 GeV
Backgrounds and rejection
➢ ZZ(ℓℓνν) + jets, WW(ℓνℓν) + jets
➢ WZ(ℓνℓℓ) + jets
– 3rd lepton veto: no e / μ with pT > 10 GeV/c
➢ DY(ℓℓ) + jets
– MET cut
– requirements on MET-ℓℓ balance
➢ tt, single top, W(ℓν) + jets, QCD
– ≤ 1 jet (pT > 30 GeV/c)
– no b-tagged jets (pT > 20 GeV/c)
MET
arXiv:1404.1344
MET > 120 GeV DY
Z(ℓℓ) + H(inv): Backgrounds
DY(ℓℓ) + jets estimated from photon + jets events
● resembles high-pT Z production in all relevant aspects – production mechanism, hadronic recoil, pile-up,
underlying event conditions
● re-weighting of pT(γ) to match Z spectrum in data
WW, single top, tt, Z(ττ) estimated cumulatively from e μ events
● eμ → ee and eμ → μμ extrapolation factors are computed from the side-bands (SB) of the Z peak (m(ℓℓ) = 40-70 + 110-200 GeV/c2 )
αee = NeeSB/NeSBμ, Neesign = αee⋅Nesignμ αμ μ = Nμ μSB/NeSBμ, Nμ μsign = αμ μ⋅Nesignμ
ZZ(ℓℓνν) + jets and WZ(ℓνℓℓ) + jets backgrounds are estimated from MC predictions
Z(ℓℓ) + H(inv): Results
Limits computed from a fit to 2D distributions mT vs Δφ(ℓℓ)
(only mT in 7 TeV dataset) Azimuthal angle Δφ(ℓ+ℓ–)
mT=
√
2pTETmiss [1−cosΔ ϕ (Z, ETmiss)]Assuming SM ZH cross section, the 95% CL observed limit on B(H→inv) for mH = 125 GeV/c2
is 0.83 (expected: 0.86)
iv:1404.1344
0-jet and 1-jet events analyzed separately (different S/B)
Final yields
Z(bb) + H(inv): Signal and Backgrounds
Signature and selection
➢ Two b-jets from Z decay
– 2 b-tagged jets with
pT > 30 / 60 GeV/c, |η| < 2.5 – pT (jj) > 100-130 GeV/c
➢ Large MET
– three boost categories:
100-130, 130-170, >170 GeV
Backgrounds and rejection
➢ Z(νν) + jets, W(ℓν) + jets
➢ tt, single top
➢ ZZ (ννbb), WZ (ℓνbb)
– no isolated leptons with pT >15 GeV/c
➢ QCD multijets
– requirements on MET quality
MET
b-jet
b-jet
● Background modeled with MC, normalized from data
● Seven control regions defined for the main backgrounds – Z + jets (0, 1, 2 b-jets), W + jets (0, 1, 2 b-jets), tt
The Combined Secondary Vertex (CSV) algorithm tags b-jets combining information from secondary vertices and
track impact parameters
Z + bb enriched region arXiv:1404.1344
Z(bb) + H(inv): Results
High-pT region
●
Final discrimination is obtained with a multivariate analysis (BDT), separately for each Higgs mass hypothesis and in each boost category
●
The BDT output distributions are used to set limits on σ × B(H→inv) / σ
ZH●
95% CL observed limit on σ × B(H→inv) / σ
ZHfor m
H= 125 GeV/c
2: 1.82 (expected: 1.99)
arXiv:1404.1344
Final yields
different boost categories analyzed separately
(different S/B)
VBF + ZH Combination
●
Individual limits on σ × B(H→invisible) / σ
SMfrom VBF and ZH channels are combined
●
Assuming SM VBF and ZH cross sections, they are interpreted as limits on B(H→invisible)
For m
H= 125 GeV/c
2at 95% CL
VBF: 0.65 (expected 0.49) ZH (ℓℓ + bb): 0.81 (expected 0.83) VBF + ZH: 0.58 (expected 0.44)
Comparable with recent indirect limits from measurement of visible decays
arXiv:1404.1344
Higgs-Portal Dark Matter
●
Limits on the Higgs invisible branching fraction can be interpreted in the context of Higgs-portal DM models
● direct-detection experiments measure the DM-nucleon elastic interaction, mediated by a Higgs
● if the DM has mass below mH /2, the decay width Γinv can be translated to spin-independent DM-nucleon elastic cross section and compared to results from direct detection
λ
hχχλ
hχχf
Ndirect detection
production at colliders
from effective field theory (EFT)
scalar DM
vector DM
fermion DM
parametrizes the DM-Higgs interaction
f
N=0.33
−0.07+0.30Limits on DM-Nucleon Cross Section
Using the combined limits from CMS VBF, Z(ℓℓ)H, and Z(bb)H searches for m
H= 125 GeV/c
2B(H→inv) < 0.51 at 90% CL
Results are competitive with current results from direct-detection experiments in the low DM mass region
arXiv:1404.1344
we set limits on the
spin-independent DM-nucleon cross section for 3 scenarios:
scalar, fermion, vector DM
Conclusions
●
Direct searches for invisible decay modes of the Higgs boson(s) are performed using CMS 2011 and 2012 data sets, in a Higgs mass range of 105-400 GeV/c
2● VBF, Z(ℓℓ)H, and Z(b
b
)H channels were considered● no significant excess observed over the expected background
● combined limits from the three searches are comparable with the indirect constraints from visible decay channels for the 125 GeV Higgs boson, assuming SM production cross sections
– 95% CL observed limits: B(H→inv) < 0.52 (indirect search), < 0.58 (direct search)
●
The limits on invisible branching fraction of the 125 GeV Higgs boson are interpreted as upper bounds on the DM-nucleon cross section in a Higgs-portal DM scenario
● results competitive with DM direct-detection experiments for DM masses < mH/2
References
●
CMS-PAS-HIG- 13‐005
http://cds.cern.ch/record/1542387?ln=en
●
arXiv:1404.1344, CMS-HIG-13-030, CERN-PH-EP-2014-051
http://arxiv.org/abs/1404.1344
submitted to European Physical Journal C
●
CMS-PAS-HIG-13-013
http://cds.cern.ch/record/1596283?ln=en
Backup
VBF: Signal and Selection
Signature and selection
➢ Two forward jets, separated by a large rapidity gap, with high invariant mass
– 2 jets with pT > 50 GeV/c, |η| < 4.6 – η1·η2 < 0, Δη > 4.2, M(jj) > 1100 GeV/c2
➢ Large missing transverse energy (MET)
– MET > 130 GeV
Backgrounds and rejection
➢ Z(νν) + jets
➢ W(ℓν) + jets
– veto leptons with pT > 10 GeV/c
➢ QCD multijets
– no additional jets with pT > 30 GeV/c
and η1 < η < η2 (central-jet veto, or CJV) – Δφj j < 1.0 rad
➢ Single top, tt, diboson, DY + jets
jet
jet MET
CMS-PAS-HIG-13-030
VBF: Backgrounds
Z(νν) + jets estimated from Z(μμ) + jets events
● exactly 2 muons with M(μμ) = 60-120 GeV/c2
● MET > 130 GeV (removing muons)
● other backgrounds subtracted using MC
● control→signal extrapolation factor from MC
NZ( ν ν) =
99
±29 (stat)
±25 (syst)
W(ℓν) + jets estimated from single e / μ / τ
hadevents
● control→signal extrapolation factor from MC
NW(eν ) =
63
±9 (stat)
±18 (syst)
NW(μ ν) =
67
±5 (stat)
±16 (syst)
NW(τ ν) =
53
±18 (stat)
±18 (syst)
QCD estimated from sidebands of two uncorrelated variables: MET and CJV (“ABCD” method)
● electroweak backgrounds subtracted using MC
● possible correlation between the two variables is accounted for in the systematic uncertainty
MET 130 GeV
CJV
pass fail
signal
B A
D C
control
NQCD =
30.9
±1.6 (stat)
±23.0 (syst)
CMS-PAS-HIG-13-013
VBF: Results
● Mass independent selection, cut-and-count approach
● Using a CLS method, we set limits on σ × B(H→invisible) and σ × B(H→invisible) / σVBF
● Assuming SM VBF production cross section and acceptance:
observed 95% CL limit on B(H→inv) for mH = 125 GeV/c2 : 0.65 (expected: 0.49)
Final yields Main systematics
CMS-PAS-HIG-13-030
Z(ℓℓ) + H(inv): Signal and Selection
Signature and selection
➢ Two isolated leptons from Z decay
– 2 leptons (e, μ) with pT > 20 GeV/c, |η| < 2.5 – M(ℓℓ) = MZ ± 15 GeV/c2
➢ Large missing transverse energy (MET)
– MET > 120 GeV
Backgrounds and rejection
➢ ZZ(ℓℓνν) + jets, WW(ℓνℓν) + jets
➢ WZ(ℓνℓℓ) + jets
– no 3rd e/μ with pT > 10 GeV/c
➢ DY(ℓℓ) + jets
– Δφ(MET, ℓℓ) > 2.7 rad – |MET – pT(ℓℓ)|/pT(ℓℓ) < 0.25
➢ tt, single top, W(ℓν) + jets, QCD
– at most 1 jet with pT > 30 GeV/c – no b-tagged jets with pT > 20 GeV/c
MET
CMS-PAS-HIG-13-030
Z(ℓℓ) + H(inv): Backgrounds
DY(ℓℓ) + jets estimated from photon + jets events
● resembles high-pT Z production in all relevant aspects
● production mechanism, hadronic recoil, pile-up, underlying event conditions
● re-weighting of pT(γ) to match Z spectrum in data
WW, single top, tt, Z(ττ) estimated cumulatively from e μ events
● no Z peak in dilepton invariant mass ⇒ scale factors from eμ to ee/μμ final states are computed from the side-bands (SB) of the Z peak: m(ℓℓ) = 40-70 GeV/c2 and 110-200 GeV/c2
αee = NeeSB/NeSBμ, Neesign = αee⋅Nesignμ αμ μ = Nμ μSB/NeSBμ, Nμ μsign = αμ μ⋅Nesignμ
ZZ(ℓℓνν) + jets and WZ(ℓνℓℓ) + jets backgrounds are estimated from MC predictions
Z(ℓℓ) + H(inv): Results
● Events with 0 jets and 1 jet analyzed separately, due to the different S/B
● Main systematic uncertainties
● theory (ZZ / ZH)
– PDFs: 5.0 / 5.7%
– QCD scales: 6.4 / 7.0%
● DY normalization from data: 5.4%
Limits computed from a fit to 2D distributions mT vs Δφ(ℓℓ)
(only mT in 7 TeV dataset) Azimuthal angle Δφ(ℓ+ℓ–)
mT=
√
2pTETmiss [1−cosΔ ϕ (Z, ETmiss)]Assuming SM ZH cross section, the 95% CL observed limit on B(H→inv) for mH = 125 GeV/c2
is 0.83 (expected: 0.86)
iv:1404.1344
Z(bb) + H(inv): Results
High-pT region
● Final discrimination is obtained with a multivariate analysis (BDT) using a set of kinematic variables after full selection and background normalization, separately for each Higgs mass hypothesis and in each boost region
● Expected and observed events in the three boost regions
● Main systematics
● signal cross section: PDF and QCD scales (6%) and NNLO QCD + NLO EWK corrections (7%)
● background normalization from data: 8%
● b-tagging efficiency effect on background shape: 7%
● The BDT output distributions are used to set limits on σ × B(H→inv) and σ × B(H→inv) / σZH
● Assuming SM ZH cross section, the 95% CL observed limit on B(H→inv) for mH = 125 GeV/c2 is 1.82 (expected: 1.99)
arXiv:1404.1344
Higgs-Portal Dark Matter
●
Limits on the Higgs invisible branching fraction can be interpreted in the context of Higgs-portal DM models
● hidden sector with stable neutral particle χ coupling directly to the Higgs boson
● direct-detection experiments measure the nuclear recoil from DM-nucleon elastic interaction, mediated by the Higgs boson
● if χ has mass below mH /2, H→χχ decay width Γinv can be translated to spin-independent DM-nucleon elastic cross section and compared to results from direct detection
λ
hχχλ
hχχf
Ndirect detection
production at colliders
Three scenarios for χ
scalar
vector
fermion
parametrizes the
DM-nucleon interaction
f
N=0.33
−0.07+0.30