Table Of ContentProceedings of the PIC 2012, Sˇtrbsk´e Pleso, Slovakia
HIGGS BOSON SEARCHES AT THE TEVATRON
3
E.W. VARNES
1
University of Arizona
0
2 1118 E. 4th St, Tucson AZ 85721, USA
E-mail: [email protected]
n
a The search for the Higgs boson, both in the context of the standard model and
J
extensions to it,has been akey focus duringRunII of the Tevatron. I reporton
2 the status of these searches, which are highlighted by evidence at the 3 standard
1 deviationlevelfortheSMHiggsinitsb¯bdecaymode,thestrongestdirectevidence
todateforfermioniccouplingsoftheHiggsboson.
]
x
e 1 Introduction
-
p
e TheHiggsmechanism[1]isoneofthecornerstonesofthestandardmodel(SM)[2].
h While the SM has passed many experimental tests since its formulation, direct
[
detection of the physical Higgs boson predicted by the Higgs mechanism proved
2 elusive. In the period between the shutdown of LEP2 and the advent of the LHC,
v theTevatronpp¯colliderwastheonlyfacilitypotentiallycapableofproducingHiggs
7
bosons. This report summarizes the results of the Higgs boson searchat the Teva-
9
tron, using data collected by both the CDF and D0 experiments. Some of these
1
4 resultsarenowsupersededwiththe discoveryofaHiggs-likebosonatthe LHC [3],
. but some (in particular the evidence for the b¯b decay mode of this new particle)
2
remain highly relevant.
1
2
1
:
v 2 General features of Tevatron searches
i
X
r Statistics areatapremiumwhensearchingforthe Higgsbosonatthe Tevatron,so
a every effort is made to extract maximal information from the data. This generally
means that multivariate discriminants are used to distinguish signal from back-
ground,withboosteddecisiontreesandneuralnetworksbeingthemostcommonly-
used. When setting limits on the Higgs production cross section, SM branching
ratios are assumed and the modified frequentist method is used.
3 Searches for the SM Higgs boson
Since the mass of the Higgs boson (m ) is not predicted, the search program at
H
theTevatronisstructuredforsensitivityfromtheLEP2limitof114GeVuptothe
end of the Tevatron’s kinematic reach, ≈200 GeV, using different production and
decay modes in the various mass ranges.
(cid:13)c Instituteof Experimental Physics SAS,Koˇsice, Slovakia 1
2 E.W. Varnes
3.1 Searches optimized for high masses
For m >≈ 135 GeV, the most sensitive search mode at the Tevatron is H →
H
WW → ℓνℓ′ν, where ℓ is an electron or muon. Using this channel, the Tevatron
wasabletoexcludevaluesform near160GeV[4],markingthefirstexclusionafter
H
LEP2. Both CDF and D0 have now updated their searches to use the entire Run
IIdatasample. Figure1showsthedistributionofthefinaldiscriminantvaluesina
subsetofthedata(theanalysisisperformedinseparatesubchannelswithdifferent
signal to background (S/B) ratios to maximize the sensitivity) for both CDF [5]
and D0 [6]. The data is in good agreement with the background-only hypothesis,
and there is no significant indication of the presence of the Higgs boson in this
channel.
Events / 0.05Events / 0.05111100002323COMDHSF =R0 1uJ6ne5 It IsG P, erHeVil/igcmh2i nSa/rBy (a) ∫ L =WtWZDWDWHt ZYaW9jZWγt.aW7 ×f b10-1 vents/0.02vents/0.02111100002323 D(bØ), 8.6 fb-1, eµ + ET DZWtt+a+jtejaetsts SDMBiikugbglnot.ia sjseloytnst.
1100 EE1100
11
11
1100--11
1100--22 1100--11
00 00..22 00..44 00..66 00..88 11
--11 --00..88 --00..66 --00..44 --00..22 00 00..22 00..44 00..66 00..88 11 FFiinnaall DDiissccrriimmiinnaanntt 00jjeett
NN Output
Figure 1. Examples of final discriminant outputs for the search for H → WW. Plot (a) shows
thedistributionofneuralnetworkoutputvaluesfromtheCDFdatacomparedtotheexpectation
from backgrounds and from a 165 GeV Higgs boson, for events with 0 jets, while (b) shows a
similarcomparisonusingD0events intheeµchannel.
The combined CDF and D0 data samples are used to set limits on the Higgs
production cross section as a function of m [7], as shown in Fig. 2. Higgs boson
H
masses between 147 and 180 GeV are excluded by the Tevatron data.
3.2 Searches optimized for low mass
For m <≈ 135 GeV the dominant decay mode is to b¯b, but a search requiring
H
only two b jets in the final state, as expected for gg → H →b¯b production, would
be overwhelmed by multijet background. Therefore we search for the associated
productionofthe Higgsbosonwith aW oraZ boson,whichcomprisesonlyabout
10%oftheHiggsbosonproductioncrosssection,butwhichcanincludehardleptons
from the decay of the W or Z. Requiring such a lepton in addition to the b¯b pair
greatly suppresses background. This results in three search modes: WH → ℓνb¯b,
ZH → ℓℓb¯b, and ZH → νν¯b¯b. The S/B ratio at the Tevatron is better than that
Tevatron Higgs Searches 3
Tevatron Run II Preliminary H→WW, L ≤ 10.0 fb-1
M
Observed
S
mit/1100 Expected w/o Higgs
Li ±1 s.d. Expected
L ±2 s.d. Expected
C
%
5
9
11
SM=1
June 2012
110 120 130 140 150 160 170 180 190 200
m (GeV/c2)
H
Figure2. 95%C.L.upperlimitontheHiggsbosonproductioncrosssectionasafunctionofHiggs
bosonmass,normalizedtotheSMexpectation, asdeterminedusingcombinedCDFandD0data
intheWW decaychannel.
at the LHC in these channels, allowing the Tevatron results for H →bb to remain
competitive with those from the LHC.
Amongthethreesearchmodes,ZH →ℓℓb¯bhasthecleanestsignatureandallows
the full kinematic reconstruction of the final state, but has the smallest expected
yield, WH →ℓνb¯bhas largeryield andbackground,andZH →ννb¯b benefits from
alargersignalandfromthecontributionofWH eventsinwhichthechargedlepton
escapes detection, but must overcome a challenging multijet background.
BeforedrawingconclusionsabouttheconsistencyoftheTevatrondatawiththe
presenceofaHiggsboson,itis importantto verifythe analysistechniquesby mea-
suringthecrosssectionofaknownprocessthatleadstothesamefinalstatesashas
a comparable event yield. Diboson (WZ and ZZ) production where one Z boson
decaystob¯b,isanexcellentverificationprocess,asitresultsinsignaturessimilarto
WH andZH withthe exceptionthattheb¯bmassdistributionpeaksattheZ pole.
The expected diboson event yield is close to an order of magnitude larger that for
SM associated Higgs boson production. However, backgrounds are also somewhat
larger at the Z boson mass than at m . The b¯b mass distribution observed in the
H
CDF and D0, with all SM processes other than WZ and ZZ subtracted, is shown
in Fig. 3. A clear peak is present at the Z mass, with a significance of more than
4.5standarddeviations(s.d.),andameasuredcrosssectionofσWZ+ZZ =3.9±0.9
pb, to be compared to the SM expectation of 4.4±0.3 pb.
Turning the analysis apparatus to the Higgs boson search, CDF and D0 both
find an excess over the SM background predictions in the range 120 < m < 145
H
4 E.W. Varnes
2) 1200 Tevatron Preliminary, L ≤ 9.5 fb-1
V/c 1+2 b-Tagged Jets int
e1000
G
Data - Bkgd
20 800 WZ
s / ( 600 ZHZiggs Signal
nt m =120 GeV/c2
e H
v 400
E
200
0
-200
0 50 100 150 200 250 300 350 400
Dijet Mass [GeV/c2]
Figure3. Background-subtracted dijetmassdistributionforevents selectedusingthesametools
usedfortheWH andZH searches,reconfiguredtotreatWZ andZZ assignal. Theplotsshows
thecombinationofCDFandD0datainallsearchchannels
GeV [8,9], as shown in Fig. 4(a) and (b). These excesses have a significance of 2.5
s.d. forCDFand1.5s.d. forD0. Theconsistencyofthedatawiththebackground-
only and signal plus background processes can be visualized by combining events
frombothexperimentsacrossallanalysischannels[10]bybinningthemaccordingto
the S/B ratio implied by their multivariate discriminant value, as shown in Fig. 5.
The Higgs boson production cross section limit implied by the data is shown in
Fig. 4 (c).
The significance of the excess can be quantified by calculating the probability
for the observeddistribution ofevents to arisefrombackgroundonly, asa function
ofthe m valueassumedinthemultivariatediscriminant. As showninFig.6,this
H
probability falls below 10−3 near 135 GeV, corresponding to a significance of 3.3
s.d. Accounting for the look-elsewhere effect brings the global significance to 3.1
s.d. If we assume that the particle observed at the LHC is in fact the SM Higgs,
and therefore that any excess relevant to the Higgs should be calculated at 125
GeV, a significance of 2.8 s.d. is found. This is the most significant evidence to
date for Higgs boson decays to fermions.
The measured value of the Higgs boson cross section times b¯b branching ratio
is 0.23+0.09 pb, larger than (but not inconsistent with) the SM expectation of
−0.08
0.12±0.01 pb, assuming that m = 125 GeV. The cross section times branching
H
ratio as a function of m is shown in Fig. 7.
H
While the WW and b¯b decay modes provide the greatest sensitivity to the
Higgs boson at the Tevatron, searches have also been conducted in other modes,
such as H → γγ, H → ττ, WH → WWW, and tt¯H production. While none of
these channels are sensitive to SM Higgs production individually, including them
increases the overall sensitivity at the Tevatron by 10 - 20% [8,11], depending on
m . TheresultofthecombinationofCDFandD0searchesinallmodes[7]isshown
H
in Fig. 8. A SM Higgs boson with mass between 147 and 180 GeV is excluded at
Tevatron Higgs Searches 5
M
Limit/S1100 sion OEExxbppseeeccrvtteeeddd w±1/o s H.di.ggs (a) (b) for
% CL F exclu EExxppeecctteedd w±2/m sH.d=.125 GeV/c2 w/o Higgs
95 CD
11
SM=1
90 100 110 120 130 140 150
m (GeV/c2)
H
Tevatron Run II, SM H→bb, L ≤ 9.7 fb-1
int
R95 (c) Observed Expected for mH=125 GeV/c2
Expected w/o Higgs
1100 ±1 s.d.
±2 s.d.
11
SM=1
100 105 110 115 120 125 130 135 140 145 150
m (GeV/c2)
H
Figure4. 95%C.L.upperlimitontheHiggsbosoncrosssection(normalizedtotheSMprediction)
obtainedfromthecombinationofsearchesintheVH →b¯bchannelsatCDF(a),D0(b),andthe
combinationofthetwo(c).
95% C.L., and at lower masses the limits are less restrictive than expected in the
background-only scenario, indicating an excess of events. The significance of the
excess, shown in Fig. 9 is 2.5 s.d. (including look-elsewhere effect), driven largely
by the b¯b decay mode.
4 Searches for the Higgs boson in extensions to the SM
In addition to searching for the Higgs boson in the context of the SM, the CDF
and D0 collaborations carried out searches in non-SM scenarios. In some cases
these searches involved a reinterpretation of the results of the SM searches (as the
expectedproductioncrosssectionanddecaybranchingfractionswouldbemodified
bythenewphysics),whileinothercasestheyinvolvedsearchesfornovelsignatures.
6 E.W. Varnes
Tevatron Run II Preliminary H→bb, L ≤ 10.0 fb-1
s
vent105 mH=125 GeV/c2 TBeavcaktgrroonu Dnda tFait (a)
E104 Signal
Tevatron Run II H→bb±, L ≤ 9.7 fb-1
103 25150 Data-background (b)
0.
102 nts/100 SM Higgs signal, mH=125 GeV/c2
e
v ±1 s.d. on background
10 E
50
1
0
10-1 2510
0.
-2 -50 nts/5
10 Eve0
10-3 -100 -5
-4 -10-1 -0.75 -0.5 -l0o.g25(s/b)0
10 -150 10
-4 -3 -2 -1 0 -3 -2.5 -2 -1.5 -1 -0.5 0
June 2012 log (s/b) log (s/b)
10 10
Figure5. (a)DistributionofVH →b¯bcandidatesfromCDFandD0combined,comparedtothe
expectationsfrombackgroundandsignal. EventsarebinnedaccordingtotheexpectedS/Bratio.
(b)Thedistributioninthehighest-S/B binsaftersubtractingthebackground.
e 102
alu TTeevvaattrroonn RRuunn IIII,, LLiinntt ≤≤ 99..77 ffbb--11 1-CLb Observed
v
- 1-CL Expected
p 10 b
d ±1 s.d.
n
u
o 1 ±2 s.d.
r
g
k
ac10-1 1σ
B
2σ
10-2
10-3 3σ
100 105 110 115 120 125 130 135 140 145 150
m (GeV/c2)
H
Figure6. Probabilityforthebackground-onlyhypothesistodescribethecombinedCDFandD0
dataasafunctionofHiggsbosonmass.
Tevatron Higgs Searches 7
Tevatron Run II, L ≤ 9.7 fb-1
int
660000
)
b Measured Expected for m =125 GeV/c2
–bb) (f 550000 ±1 s.d. EAxspseucmteindg fboers mt fiHt =ra1te2 5a tG meHV=1/c225 GeV/c2
→ H
H ±2 s.d. Assuming SM rate
(
Br 440000 Predicted
x
)
H
Z330000
σ
+
H
W
σ 220000
(
110000
00
100 105 110 115 120 125 130 135 140 145 150
m (GeV/c2)
H
Figure7. MeasuredVHproductioncrosssection,withone-andtwo-s.d. errorbands,asafunction
ofHiggsbosonmass,usingthecombinedCDFandD0data. Theredbandshowstheexpectation
fromtheSM,withitsuncertainty.
One example of the former category is the Higgs search under the assumption
of the existence of a fourth generation of quarks. In this scenario, there would be
additionalfermionloopscontributingtothegg →H crosssection. Sincethefourth
generation quarks must be more massive than the top, each additional quark loop
contributes an amplitude similar to that of a top quark, resulting in a production
cross section ∼ 9 times larger than the SM prediction. Thus the upper limit on
the gg →H production cross section can constrain the existence of a fourth quark
generation. Under the assumption that the fourth-generation leptons have masses
that just exceed the current limits, a fourth quark generation is excluded by the
combined CDF and D0 data for 124 < m < 286 GeV [12], as shown in Fig. 10.
H
This implies that if the new particle seenat 125GeV atthe LHC is in fact the SM
Higgs boson, then a fourth chiral quark generation is excluded.
One can can also search for deviations from the SM Higgs couplings. One
example of this is the search for a fermiophobic Higgs. In this scenario the gluon
fusion production mechanism is greatly suppressed (since Higgs bosons cannot be
created via a fermion loop), and the branching fractions to WW, ZZ, and γγ are
enhanced (particularly at low m ), since there is no possibility of H → b¯b. CDF
H
andD0havebothperformedsearchesoptimizedforsensitivitytosuchfermiophobic
HiggsbosonsintheWW andγγchannels,andfindnoevidencefortheirproduction.
The resulting limits on the cross section in the fermiophobic model obtained from
the combined CDF and D0 data [13] are shown in Fig. 11. Fermiophobic Higgs
masses below 119 GeV are excluded at 95% C.L.
8 E.W. Varnes
CDF Run II Preliminary, L ≤ 10 fb-1
M (a) M
L Limit/S1100 EO±±12xbσσps eeEEcrxxvtppeeeeddcctteedd CExDcFlusion σσ / on S10 (bD)Ø PSrMel iHmiigngasr yC, oLimnt b≤i n9a.7ti ofbn-1 EEOxxbppseeeccrvtteeeddd ±w1/o s .Hdi.ggs
% C mit Expected ±2 s.d.
5 Li
9 L
C
%
5
9
11
SM=1 1
LEP Exclusion
DØ Exclusion
February 27, 2012
100 110 120 130 140 150 160 170 180 190 200 100 110 120 130 140 150 160 170 180 190 200
m (GeV/c2) Higgs Boson Mass (GeV/c2)
H June 2012
Tevatron Run II Preliminary, L ≤ 10.0 fb-1
M
95% CL Limit/S1100 atron + LEP Exclusion LEP Exclusion LEP+ATLAS ExclusionATLAS Exclusion ATLAS Exclusion OE±±12xb pssse..eddcr..vt eEEeddxx ppwee/ccott eeHddiggs TEe+xvAcalTutLrsoAionSn+CMS AExTcLlAu(Ssci+o)CnMS
v
e
T
11
SM=1
CMS Exclusion AExTcLlAuSsi+oCnMS June 2012
100 110 120 130 140 150 160 170 180 190 200
m (GeV/c2)
H
Figure8. 95%C.L.upperlimitontheHiggsbosoncrosssection(normalizedtotheSMprediction)
obtainedfromthecombinationofsearchesinallchannelsatCDF(a),D0(b),andthecombination
ofthetwo(c).
In supersymmetry the Higgs sector is extended to include at least two com-
plex doublets, resulting in five physical bosons (three neutral, two charged) af-
ter electroweak symmetry breaking. Under the Minimal Supersymmetric Model
(MSSM) [14] the production of neutral Higgs bosons (here generically referred to
as φ) is proportionalto tan2β. The dominant production mode is pp¯→φb modes,
and the dominant decay of the φ is to b¯b, resulting in a bbb final state. The main
complication of a search in the bbb mode is understanding the large multijet back-
ground,whichis difficult to simulate andsubjectto largetheoretical uncertainties.
This background is modeled using a data-driven approach that considers the dis-
tribution of jet transversemomenta and the number of taggedb jets in eachevent.
The results are presented in Fig. 12. The CDF data sample has an excess of
Tevatron Higgs Searches 9
e 1-CL Observed
u TTeevvaattrroonn RRuunnIIII PPrreelliimmiinnaarryy b
al 1-CL Expected
-v 10 LL ≤≤ 1100..00 ffbb--11 Expebcted ±1 s.d.
d p Expected ±2 s.d.
n 1
u
gro10-1 1σ
k
ac 2σ
B10-2
10-3 3σ
10-4
4σ
10-5
100 110 120 130 140 150 160 170 180 190 200
Higgs Boson Mass (GeV/c2)
June 2012
Figure9. Probabilityforthebackground-onlyhypothesistodescribethecombinedCDFandD0
dataasafunctionofHiggsbosonmass.
44
n
o Expected
dicti TLe ≤v a8t.r2o nfb R-1un II Preliminary Observed
s) Pre 33 ±±12 ss..dd.. EExxppeecctteedd
s
a
m 4G(High mass)
w
o
G(l 22
4
mit/
Li
L. 11
C.
%
5
9 4G(Low mass)=1
00
120 140 160 180 200 220 240 260 280 300
m (GeV)
H
Figure10. 95%C.L.upperlimitonHiggsbosonproductionundertheassumptionoftheexistence
ofafourthgenerationofchiralquarks,usingcombinedCDFandD0data.
events over background at the 2.8 s.d. level for a φ mass of 120 GeV [15], while
D0’s largest excess is at the 2.5 s.d. level for m = 150 GeV [16]. In both cases
φ
accounting for the look-elsewhere effect reduces the overall significance to 2 s.d.
10 E.W. Varnes
Tevatron Run II Preliminary L ≤ 8.2 fb-1
M
FH1100 EOxbpseecrvteedd
t/ ±1 s.d. Expected
mi ±2 s.d. Expected
Li
L
C
% 11
5 FHM = 1
9
--11
August 9, 2011
1100
100 110 120 130 140 150 160 170 180 190 200
m (GeV/c2)
H
f
Figure11. 95%C.L.upperlimitonHiggsbosonproductionundertheassumptionthattheHiggs
doesnotcoupletofermions,usingcombinedCDFandD0data.
The combined limits [17] are also shown in Fig. 12. The fact that the CDF and
D0 excesses are at different mass hypotheses results in a combined excess at the 2
s.d. level in the 120<m <150 GeV region, no larger than the excess seen in the
φ
individual experiments.
Some extensions to the SM, including left-right symmetric [19] and Little
Higgs [20] models, predict the existence of doubly-charged Higgs bosons. D0
searches for these particles using the mode pp¯→ ℓ±ℓ±τ∓τ∓ where the taus decay
hadronically, representing the first search for the tau pair decay of doubly-charged
Higgs bosons [21]. The data is consistent with expectations from the SM, and the
resulting lower limits on the mass of a doubly-charged Higgs boson depend upon
the specific model being considered. An example, under the assumption that the
H±± decays only to tau pairs, is shown in Fig. 13.
Another class of extensions to the SM, the “Hidden Valley” models [22], posits
the existence of a new gauge sector, which is not directly observable. However
there are new “messenger” particles that couple to both the exotic and SM gauge
sectors, and the coupling of these particles to the Higgs boson can result in dis-
tinctive signatures. CDF has searched for Higgs bosons decays under two Hidden
Valley scenarios. In the first of these, the messenger particles are long-lived and
decay predominantly to b quark pairs. Thus Higgs decay to a pair of messenger
particles results in two pairs of b jets, each originating form a vertex that is typ-
ically centimeters away from the primary interaction vertex. The CDF data does
not indicate an excess for this signature over the SM expectations [23], resulting
