1

Search for the Higgs Boson at Search for the Higgs Boson at DDØØ in the Trilepton Channel in the Trilepton Channel

James KrausJames KrausUniversity of MississippiUniversity of Mississippi

2

Outline

The Standard Model and the Higgs Boson Fermilab Tevatron and the DØ experiment Higgs Production and Higgs Searches at DØ

– Motivation for the trilepton Higgs search Search for the Higgs in the trilepton + Missing

Transverse Energy (MET) final state Discussion of DØ and Tevatron combinations

– Comparison with LHC results

3

Standard Model

Describes 3 of the 4 Fundamental Forces– Strong Force

• Force carrier is gluon• Binds quarks into protons, neutrons• Holds protons & neutrons in nucleus

– Electromagnetic Force• Force carrier is photon• Acts on charged particles• Responsible for electricity,

chemical bonds

4

Standard Model

Describes 3 of the 4 Fundamental Forces– Strong Force– Electromagnetic Force– Weak Force

• Force carriers W±, Z0

• Responsible for some radioactive decays, nuclear fission

– Gravity• Planetary motion, falling apples

137

55 Cs 137

56 Ba

νe

e-

5

Standard Model

Describes 3 of the 4 Fundamental Forces– Strong Force– Electromagnetic Force– Weak Force

• Force carriers W±, Z0

• Responsible for some radioactive decays, nuclear fission

– Gravity• Planetary motion, falling apples

137

55 Cs 137

56 Ba

νe

e-

Gravity is not included in Standard Model

6

Standard Model The Standard Model arranges 6 quark flavors and

6 lepton flavors into 3 generations of matter

Quarks experience strong, weak, and electromagnetic forces

Leptons do not experience the strong force. Neutrinos only interact via the weak force

The generations are arranged by increasing mass, but are otherwise similar

7

Electroweak Theory Salam, Glashow, & Weinberg develop Electroweak

theory in 1967 (Nobel prize in 1979)– Describes EM and Weak forces with a single theory– Predicted existence of massive, neutral Z boson

Electroweak symmetry must be broken to give us the observed electromagnetic and weak forces

With electroweak symmetry unbroken, particles are massless...

8

A Massive Problem

e

e

Top Quark173 GeV

u d s c b Z Boson91 GeV

W Boson80 GeV

(Photon & Gluon = Massless)

1 GeV ≈ mass of the proton

9

A Massive Solution

2010 Sakurai Prize... for "elucidation of the properties of spontaneous symmetry

breaking in four-dimensional relativistic gauge theory and of the mechanism for the consistent generation of vector boson masses."

Englert Brout Higgs Guralnik Hagen KibblePRL 13, 321-323

(1964)PRL 13 508-509

(1964)PRL 13, 585-587

(1964)

10

The Higgs Boson in the Standard Model

Higgs mechanism responsible for electroweak symmetry breaking

Interactions with Higgs fieldgives particles their masses– No prediction of the strength of

this interaction The Higgs is the last undiscovered particle predicted

by the Standard Model– Standard Model does not predict the mass of the Higgs

11

Fermilab

Fermilab is a national lab located in Batavia, IL, outside Chicago

− Is home to US-based high energy physics experiments, including Tevatron

To Chicago

12

The Fermilab Tevatron

Run II of Fermilab Tevatron March 2001 to Sept. 2011

− p-p collisions with energy of 1.96 TeV = 1960 mp

One collision every 396 ns− >1.7 million collision/min

Technology− Superconducting magnets

Size − 4 mile circumference

DØ

CDF

13

Fermilab Site Home to a healthy herd of bison

14

Fermilab at Night

15

The DØ Experiment A multipurpose

particle detector Innermost detectors are

the trackers, followed by calorimetry and muon chambers

Using information from

all subdetectors, we can reconstruct the pp collision

Pseudorapidity η = -ln(tan(θ/2))

16

Electrons, Photons Lighter particles (e, γ) are

stopped in the first few layers of the the calorimeter

Hadrons make it farther into the detector

Only charge particles leave tracks

Track matching allow us to distinguish e from γ

17

Hadronic Jets When a high momentum

quark or gluon is produced in a collision, results in a jet of hadronic particles into the detector

18

Muons Muons can pass through our

entire detector– µ are massive enough not to

be stopped like e or γ in EM calorimeter

– Don't feel strong force, so hard interactions are rare

Muons are charged, so they leave tracks

Place trackers outside the calorimeter to detect passage, and match to a track in inner tracker

19

Missing Transverse Energy Neutrinos do not interact

with our detector– Infer their presence through

missing transverse energy– Transverse = ⊥ to beamline

To conserve momentum,

vector sum pT = 0

– If non-zero, then either• Energy Mis-measurement• Missed one or more particles (usually neutrinos)

electron

METMET

jet

muons

20

Higgs Production at the Tevatron The Higgs at Tevatron

produced by gluon fusion (gg→H), associated production (qq→W/Z+H), and vector boson fusion (qq→qqH)

21

Higgs Decay Modes

The Higgs decays primarily to bottom quarks at low mass, WW at higher mass

The searches for the Higgs at the Tevatron focus on H→bb at low mass, H→WW at high mass.

Also use H → ττ, H → γγ due to low backgrounds

22

Higgs BackgroundsOr, You've been looking for over 10 years, why haven't you found it already?

The rate of Higgs production is small compared to backgrounds− Bottom decays swamped by

QCD background except for associated ZH/WH production

− WW cleaner, but still large backgrounds

23

Trilepton Higgs Search

H0

New search channel at DØ – not in July 2011 combination

ZH and WH production followed by H → WW decay

For WH → WWW → lνlνlν, where l = e, µ– Ratio of eee:eeµ:µµe:µµµ = 1:3:3:1

For ZH → ZWW → lllνX, ratio is 1:2:2:1 Currently, only considering the eeµ and µµe

final states, with others to be added in future

24

Higgs Decay Modes

The expected Higgs signal from VH→VWW is between one tenth and one hundreth that from H→WW

However, eliminates almost all W, Z, Wγ, and WW backgrounds

So there is a gain in the signal over background in the three lepton channels compared to the 2 lepton channels

25

Trilepton Backgrounds

Principle backgrounds for the trilepton search– WZ → lνll– ZZ → llll

Also have backgrounds with 2 real leptons and a jet or photon fake– Z + jets– Z + γ– tt→lνblνb

26

Z + jets

Z→l+l- + jets cross section is the largest of all the backgrounds

However, it is suppressed by low j→l fake rates

– Z + jet fake rate is estimated by passing simulated Z+jet events through our MC detector model

Z + jet events have no real missing energy– Variables that are sensitive to the missing energy

separate this background from our signal

q

q

Z/γ*

g

Fakes lepton

27

Z + γ

Z + γ significant background for µµe final state Only difference between

electrons and photons in our detector is a track

28

Z + γ

Z + γ significant background for µµe final state Only difference between

electrons and photons in our detector is a track

Sometimes the photon undergoes interaction with the tracker, leading to γ→ee

– If one electron gets most of the photon energy, will be reconstructed as an electron

γe+

e-

29

Estimating the Z + γ Background

Correctly estimating the γ→e fake rate in MC difficult

We use a data-driven method to estimate this background

– Calculate the γ→e fake ratio from a photon sample in data

– Reconstruct µµγ events, where the photon is not matched to a track

– Apply the fake ratio to the data events to get the background estimate

30

Top Background

Like Z + jets, tt requires jet to fake lepton– tt events have real missing energy– b-quark jets have a higher fake rate than light quark jets– The tt background is the smallest SM background, but

tends to be more signal-like, so it is significant

Fakes lepton

31

WZ and ZZ Backgrounds

WZ → lνll and ZZ → llll are estimated from MC– Currently, there is no dedicated

4l Higgs search at DØ, so these events are included in our search

– Of all backgrounds, WZ most closely mimics signal

32

Trilepton Selection Cuts

We have requirements on both eeµ and µµe channels– 3 good leptons must be found – M(ee) or M(µµ) > 15 GeV– One lepton with pT > 15 GeV,

all with pT > 10 GeV

– All leptons from same pp collision– between

any two leptonsΔ R=√Δη2+Δϕ2>0.3

33

Reducing Backgrounds in µµe Channel

The µµe channel has larger backgrounds than eeµ– From Z backgrounds – jet → e fake rate higher than

jet→µ fake rate, γ fake e but not µ Have additional cuts on µµe to reduce backgrounds

– Remove final state radiation (FSR)– Photon is radiated off a muon after

Z decay, so µµγ will have Z mass– Also, no real missing energy, so

can keep most signal in the Z mass range by cutting only low MET events

34

Reducing Backgrounds in µµe Channel

The µµe channel has larger backgrounds than eeµ– From Z backgrounds – jet → e fake rate higher than

jet→µ fake rate, γ fake e but not µ Have additional cuts on µµe to reduce backgrounds

35

Reducing Backgrounds in µµe To reduce background

more, turn to variables related to MET

Special MET– If ∆φ between MET and

nearest object < 90°. specialMET = MET×sin∆φ, otherwise = MET

MET Significance– Based on the uncertainty on

the energy measured in the detector

36

Reducing Backgrounds in µµe We require either

– Special MET > 15 GeV or

– MET Significance > 3 Removes most Z + jets and

Zγ, leaves most signal

37

Event Yields After all cuts, have 175 trilepton events

Less than 3 expected signal events

S/B ~ 0.15

Errors reflect uncertainty due to sample size. With all statistical and systematic errors, expected and observed agree within 1σ

38

Signal vs. Background

Once we have a good background model, we look for differences between signal and background

Example – ∆φ between leptons The Higgs is a spin-0

particle; W is spin-1⇒ small opening angle in

leptons from the W decays

Leptons from Z tend to be back-to-back

39

Signal vs. Background

Once we have a good background model, we look for differences between signal and background

Example – ∆φ between leptons

40

Signal vs. Background

Once we have a good background model, we look for differences between signal and background

Example – ∆φ between leptons The signal to background ratio is so small that any

one variable will not be sufficient to isolate the Higgs

Instead, we look at many variables associated with each event at once – multivariate analysis

41

Multivariate Analysis

Take multiple variables, each of which has some separating power

Combine them into one variable that separates the signal and background

42

Limit Setting

Multivariate discriminants are used to hunt for signal If there is a significant amount of signal, will see

pile-up of events near one, in excess of background

43

How We Set Limits Based on log-likelihood ratio

– L is the Poisson likelihood that the s+b (b) correctly models the data

– θ represents the systematic uncertainties on the measurements (luminosity, energy scale, etc.)

Calculate probability density of LLR based on models of signal and background

44

How We Set Limits Based on log-likelihood ratio

CLb and CLs+b are given by the integrals of the b and s+b LLR distributions above observed LLR– Represents how well the given model agrees with data

We vary the signal content of the s+b model to find where CLs = CLs+b/CLb < 0.05

– We exclude those points at the 95% Confidence Level (CL)

45

Limit Setting vs Mass

No prediction of Higgs mass, so calculate LLR at multiple mass points– Set limits at 5 GeV intervals between 100 and 200 GeV– Solid black observed, dashed black b, dashed red s+b

for signal set to SM expectation

46

Limit Setting vs Mass

Data consistent with background, so set limits– 95% CL limits shown for each mass point, adjusted for

expected Higgs cross section at that mass– If observed limit falls below 1, we exclude SM Higgs at

that mass point

47

Trilepton Higgs Limits

We expect a 95% CL limit of 12x SM expectation at mH = 125 GeV, measure a limit of 18x SM expectation

48

DØ Combination

Trileptons only one channel out of many

49

DØ Combination

Trileptons only one channel out of many– Others include ZH→llbb,

50

DØ Combination

Trileptons only one channel out of many– Others include ZH→llbb, WH→lνbb,

51

DØ Combination

Trileptons only one channel out of many– Others include ZH→llbb, WH→lνbb, ZH → ννbb,

52

DØ Combination

Trileptons only one channel out of many– Others include ZH→llbb, WH→lνbb, ZH → ννbb,

H→WW→lνlν,

53

DØ Combination

Trileptons only one channel out of many– Others include ZH→llbb, WH→lνbb, ZH → ννbb,

H→WW→lνlν, VH→l+l+X,

54

DØ Combination

Trileptons only one channel out of many– Others include ZH→llbb, WH→lνbb, ZH → ννbb,

H→WW→lνlν, VH→l+l+X, H→ττ,

55

DØ Combination

Trileptons only one channel out of many– Others include ZH→llbb, WH→lνbb, ZH → ννbb,

H→WW→lνlν, VH→l+l+X, H→ττ, H → γγ, and more

56

DØ Combination

Trileptons only one channel out of many– Others include ZH→llbb, WH→lνbb, ZH → ννbb,

H→WW→lνlν, VH→l+l+X, H→ττ, H→γγ, and more– Below 145 GeV, H→bb searches dominate, but

H→WW still contributes significantly

57

DØ Combination

We expect a 95% CL limit of 1.85x SM expectation at mH = 125 GeV, measure a limit of 2.53x SM expectation

Exclude 159 to 166 GeV

58

Previous Higgs Search

The Tevatron Higgs search limits from July `11

156–177 GeV

59

Current Large Hadron Collider Results

The highest energy collider in the world is now the LHC at CERN– 122.5 < mH < 127.5 allowed at 95% CL

60

Tevatron Combination

The CDF experiment does similar Higgs searches– CDF sees larger excess at low mass than DØ

149–178 GeV

61

Tevatron Combination

We expect a 95% CL limit of 1.1x SM expectation at mH = 125 GeV, measure a limit of 2.2x SM expectation

147 – 179 GeV

62

Tevatron and LHC results

63

Magnitude of Excess Most significant local excess

– Tevatron: 2.7σ at 120 GeV– CMS: 2.8σ at 125GeV– ATLAS: 2.6σ at 126 GeV

• H→γγ+H→ZZ channels 3.4σ

64

Global Excesses

However, need to consider look-elsewhere effect– The more independent mass points examined, the more

likely it is that at least one of them will be found to be high due to a statistical fluctuation

– The look-elsewhere factor scales the probability for the number of independent mass points

After application of look-elsewhere factor– Tevatron global excess of 2.1σ over 100 GeV – 200 GeV– ATLAS global excess of 2.2σ over 110 GeV – 146 GeV– CMS global excess of 2.1σ over 110 GeV – 145 GeV

65

Plans for Improving Results

Tevatron no longer taking data, but we are still refining our search to improve our limits

For the trilepton search, we are looking into – Improving our multivariate analysis by including jet

information– Adding eee and µµµ channels– Improving jet modeling

More is still to come – Tevatron is close to having the expected limits less than the SM expectation between 100 and 180 GeV

66

Conclusions Have set limits on the Higgs in

the trilepton channel at DØ– First time this has been done at

DØ was in Feb. 2012 Tevatron excludes Higgs masses

147 – 179 GeV @ 95% CL– 110 – 122.5 GeV, 127 – 600 GeV

excluded by LHC experiments The global excesses are

– Tevatron: 2.1σ for 100–200 GeV

– ATLAS: 2.2 σ for 110 – 146 GeV

– CMS: 2.1σ for 110 – 145 GeV

67

Backup Slides

68

photons

Feynman Diagrams

A pictorial way to depict particle interactions = quark, lepton, or hadron = gamma, W Z = gluon

becomes

e+ e-

time

e+

e-time

69

The Higgs Field

The Higgs field is everywhere The Higgs field impeded the

progress of a particle, slowingit from speed of light– Effectively giving it mass

The Higgs boson is the quantaof the Higgs field

70

Indirect Measurements

Because of quantum mechanics, the Higgs bosons can influence other particles without being directly detected

More mass more interactions ⇒with virtual Higgs bosons

Measuring the mass of the W boson and top quark can tell us about the Higgs boson

tt

71

Indirect Measurements

Fit to the precision EW data

Prefers a low mass Higgs

⇒ MH ≈ 94 GeV

⇒ MH ≤ 152 GeV at 95% CL

tt

72

Look-Elsewhere Effect

Consider a simple case – flipping a coin 100 times– Expect it to come up heads about 50 times– Not exactly 50 each time – probability distribution of

number of heads– If we perform this task 1000

times (100, 000 coin flips) we may have a case where we find 65 heads, and not 50

– Does not mean coin is broken– The probability of getting 65 heads is 0.00087 – but if you

perform the experiment 1000 times, there is a 87% chance this will happen at least once – “look elsewhere effect”

73

eeμ/μμe Systematics Uncertainties

• Apply both flat and shape systematics; shape systematics include Z-pT smearing, electron pT smearing, and muon pT smearing

74

The Standard Model of Particle Physics

The Standard Model describes the strong, electromagnetic, and weak forces

– Strong force – holds quarks together in protons and neutrons, protons and neutrons in atomic nuclei

– Electromagnetic force – runs your iPod, etc.

– Weak force – responsible for some nuclear decays

75

The Standard Model of Particle Physics

The Standard Model (SM) describes the electromagnetic, weak, and strong forces

Only quarks feel the strong force

Force carrier is the gluon

76

The Standard Model of Particle Physics

The Standard Model (SM) describes the electromagnetic, weak, and strong forces

All charged particles feel the electromagnetic force

Force carrier is the photon

77

Searching For The Higgs at DØ

In all decay modes, there are at least 100 background events for every expected signal event

There is no one variable, such as dilepton mass or missing transverse energy, that can separate the Higgs signal from background

However, by looking at multiple variables, we can separate the signal and background sufficiently to possibly find the Higgs

78

The Weak Force

137

55 Cs 137

56 Ba

“missing” energy(neutrino)

β particle(electron)

Henri Becquerel discovered radioactivity in 1896

Enrico Fermi proposes model for beta decay in 1934– Point interaction with strength

GF = 1.166x10-5 GeV-2

p

e-

nn GF

ν

79

The Weak Force Fermi's point interaction fails at high energies

– Weak force is short range: Massive force carrier– EM is long range: photon massless

e

e

e- e-

p pν

p

e-

n W

W

W- ( q )

αe2

q 2+M γ2 =

αe2

q 2

αW2

q 2+MW2

If q≪MW ⇒ αW2

MW2 ≃ GF

Assume αe = αW M⇒ W ≈ 100 GeV

( q )

80

The Weak Force

137

55 Cs 137

56 Ba

“missing” energy(neutrino)

β particle(electron)

Henri Becquerel discovered radioactivity in 1896

Enrico Fermi proposes model for beta decay in 1934– Point interaction with strength

GF = 1.166x10-5 GeV-2

p

e-

nn GF

ν

81

The Weak Force Fermi's point interaction fails at high energies

– Weak force is short range: Massive force carrier– EM is long range: photon massless

e

e

e- e-

p pν

p

e-

n W

W

W- ( q )

αe2

q 2+M γ2 =

αe2

q 2

αW2

q 2+MW2

If q≪MW ⇒ αW2

MW2 ≃ GF

Assume αe = αW M⇒ W ≈ 100 GeV

( q )

82

The Standard Model of Particle Physics

The Standard Model describes the electromagnetic, weak, and strong forces

However, we have not yet found the last particle predicted by the Standard model, the Higgs Boson

83

WZ, ZZ, and top Backgrounds

WZ → lνll and ZZ → llll are estimated from MC– Currently, there is no dedicated 4l Higgs search at DØ,

so these events are included in our search Like Z + jets, tt requires jet to fake lepton

– tt events have real missing energy– b-quark jets have a higher fake rate than light quark jets– The tt background is the smallest SM background, but

tends to be more signal-like, so it is significant