s = 7 TeV with the ATLAS detector

ATLAS-CONF-2012-089 07July2012

ATLAS NOTE

ATLAS-CONF-2012-089

July 7, 2012

Search for Displaced Muon Jets from light Higgs boson decay in proton-proton collisions at √

s = 7 TeV with the ATLAS detector

The ATLAS Collaboration

Abstract

A search is presented for collimated muon pairs produced in the decay of long-lived neutral particles in proton-proton collisions at

s=

7 TeV center of mass energy, with the ATLAS detector at the LHC. In a sample of 1.94 fb

of collision events collected during 2011, the observed data are consistent with the Standard Model background expectations.

Limits on the product of the production cross section and the branching fraction of a Higgs boson decaying to Hidden Valley neutral long-lived particles are derived as a function of the particles’ mean cτ.

c

Copyright 2012 CERN for the benefit of the ATLAS Collaboration.

Reproduction of this article or parts of it is allowed as specified in the CC-BY-3.0 license.

1 Introduction

A search is presented for collimated muon pairs produced in the decay of long-lived neutral particles in proton-proton collisions at

s

7 TeV center of mass energy with the ATLAS detector at the LHC.

The benchmark model considered in this analysis consists of a Higgs boson decaying to a new hidden sector of particles which finally produces two sets of collimated muon pairs, but the search described is equally valid for other, distinct models such as heavier Higgs boson doublets, singlet scalars or a Z

, that predominantly decay to a hidden sector and eventually produce collimated muon pairs.

The Higgs boson [1–3] may have branching fractions that di

er from the predictions of the Standard Model (SM). A particularly interesting possibility is that the Higgs boson, with a mass m

in the range 100 -140 GeV, decays into states residing in a new light hidden sector of particles. The phenomenology of light hidden sectors has been studied extensively over the past few years [4–8]. Possible characteristic topological signatures of such extensions of the SM are the “Lepton Jets” (LJs). A LJ is a cluster of highly collimated particles: electrons, muons and possibly pions [5, 9–11]. These arise if light unstable particles with masses in the MeV to GeV range (for example dark photons,

) reside in the hidden sector and decay predominantly to SM particles. At the LHC, hidden sector particles may be produced with large boosts, causing the visible decay products to form jet-like structures. Hidden sector particles such as

may be long-lived, resulting in decay lengths comparable to, or larger than, the detector dimensions.

The production of LJs can occur through various channels. For instance, in supersymmetric models, the lightest visible superpartner may decay into the hidden sector. Alternatively, a scalar particle that couples to the visible sector may also couple to the hidden sector through Yukawa couplings or the scalar potential. This analysis is focused on the case where the Higgs boson decays, possibly through couplings to supersymmetric particles, to the hidden sector [12, 13]. The SM Higgs boson has a narrow width into SM final states when m

2m

. Consequently, any new (non-SM) couplings to additional states, which reside in a hidden sector, may easily dominate the Higgs branching fraction. Even with new couplings, the total Higgs boson width is typically small, well below a GeV. If a SM-like Higgs boson is discovered, it will remain important to constrain possible rare decays into LJs. Therefore, searches for both prompt and long-lived dark photons decaying into LJs form an important component of the Higgs boson search program. Although a Higgs boson decaying to displaced LJs has never been searched for directly, the OPAL experiment conducted a decay mode independent search at LEP [14].

Neutral particles with large decay lengths and collimated final states represent, from an experimental point of view, a challenge both for the trigger and for the reconstruction capabilities of the detector.

Collimated particles in the final state can be difficult to disentangle due to the finite granularity of the detectors; moreover, in the absence of the central tracking detector information and of the main vertex constraint, it is difficult to reconstruct charged tracks from decay vertices far from the interaction point (IP). The ATLAS detector [15] is equipped with a muon spectrometer (MS) with high granularity tracking detectors and allows for the reconstruction of charged tracks in a “standalone” configuration using only the muon detector information. This is a crucial feature for detecting charged particles not originating from the primary interaction vertex.

The search presented in this analysis focuses on neutral particles decaying to muon jets (MJ) and it is the first ATLAS search for non-prompt MJ; prompt MJ searches have been performed both at the Tevatron [16, 17] and at the LHC [18, 19]. Other searches for displaced decays of a light Higgs to heavy fermion pairs have also been performed at the LHC [20].

The benchmark channel used for this analysis is a simple scenario where the Higgs boson decays to

a pair of hidden fermions ( f

) each of which decays to one long lived

and one stable hidden fermion

( f

), resulting in two LJs from the

decays in the final state (see Figure 1). The benchmark mass of

the

(0.4 GeV) is chosen to maximize its decay branching fraction to muons [12]. This choice allows

the analysis strategy to be tuned for the relatively clean environment of the muon spectrometer.

h

γ_d γ_d

f_d1 (LSP) f_d2

f_d2

f_d1 (LSP)

Figure 1: Schematic picture of the Higgs boson decay chain, H→2( f

f

), used for the benchmark sample. The Higgs boson decays to two hidden fermions ( f

). Each hidden fermion decays to a

and to a stable hidden fermion ( f

), resulting in two muon jets from the

decay in the final state.

2 The ATLAS Detector

ATLAS is a general purpose collider detector [15] consisting of an inner tracking system (ID) embed- ded in a superconducting solenoid, which provides a 2 T magnetic field parallel to the beam direction, electromagnetic and hadronic calorimeters and a muon spectrometer using three air core toroidal magnet systems

. The trigger system has three levels [21] called Level-1 (L1), Level-2 (L2) and Event Filter (EF). The (L1) is a hardware-based system using information from the calorimeter and muon spectrom- eter, and defines one or more Regions of Interest (RoIs), geometrical regions of the detector, identified by

and

coordinates, containing interesting physics objects. The L2 and EF (globally called High Level Trigger, HLT) are software-based systems and can access information from all subdetectors. The ID, consisting of silicon pixel and micro strip detectors and a straw tube tracker, provides precision tracking of charged particles for

2.5. The electromagnetic and hadronic calorimeter system covers

|η| ≤

4.9 and, at

0, consists of 22 radiation lengths (in the electromagnetic part) and 9.7 interaction lengths. The MS provides a trigger (|

2.4) and momentum measurements (|

2.7) for charged particles entering the spectrometer. It consists of one barrel and two endcap parts, each with 16 sectors in

(MDTs) are used for precision tracking in the region

2.0 and Cathode Strip Chambers for 2.0

|η| ≤

2.7. The MS detectors are arranged in three stations of increasing distance from the IP: inner, mid- dle and outer. The air core toroidal field allows an accurate charged particle reconstruction independent from the ID information. The trigger chambers (Resistive Plate Chambers in the barrel and the Thin Gap Chambers in the endcaps) are located in middle and outer (only in the barrel) stations. The L1 muon trigger requires hits in the middle station to create a low p

muon RoI or hits in both the middle and outer stations for a high p

RoI. The muon RoIs have a spatial extent of 0.2×0.2 (

and of 0.1×0.1 in the endcap. The L1 RoIs are used by the HLT to provide sharp thresholds for muon momenta up to 40 GeV.

1ATLAS uses a right-handed coordinate system with its origin at the nominal interaction point (IP) in the center of the detector and thez-axis coinciding with the beam pipe axis. Thex-axis points from the IP to the center of the LHC ring, and they-axis points upward. Cylindrical coordinates (r,φ) are used in the transverse plane,φbeing the azimuthal angle around the beam pipe. The pseudorapidity is defined in terms of the polar angleθasη=-ln tan(θ/2).

3 Signal and background simulation

The set of parameters used to generate the benchmark samples is listed in Table 1. In the signal Monte Carlo the Higgs boson is generated using the gluon-gluon fusion production mechanism. The gluon- gluon fusion Higgs boson production cross section in pp collisions at

s

7 TeV, estimated at the currently highest known order (NNLO) [22], is

24.0 pb for m

100 GeV and

12.0 pb for m

The OPAL decay mode independent search at LEP implies m

82 GeV [14]. Furthermore, it is most probable for the Higgs boson to dominantly decay into the hidden sector when m

2m

since a heavier Higgs boson has a large partial width into WW and ZZ final states. Therefore, the most interesting parameter space consists of the range m

82

161 GeV, and the representative values m

100, 140 GeV are chosen for the benchmark points. The masses of f

and f

are chosen to be light relative to the Higgs boson mass, and far from the kinematic threshold at m

m

m

.

Higgs mass mfd2 mfd1 γ_dmass cτ (GeV) (GeV) (GeV) (GeV) (mm)

100 5.0 2.0 0.4 47

140 5.0 2.0 0.4 36

Table 1: Parameters of the benchmark samples.

For the dark photon mass chosen in the benchmark model, the

decay branching fractions are expected to be [12]: 45% e

e

, 45%

, 10%

. Thus about 20% of the Higgs H

X decays are expected to have a four-muon final state.

For the signal Monte Carlo simulation the

generator [23] is used, linked together with Mad- Graph 4.4.2 [24] and BRIDGE [25], for gluon-gluon fusion production of the Higgs boson and the subsequent decay to hidden sector particles. The cτ values of the

in the Monte Carlo simulation is chosen to have 80% of the decays inside the whole sensitive ATLAS detector volume, i.e. up to 7 m in radius and 13 m along the z axis

around the interaction point (IP). The simulated events are processed through the full ATLAS simulation chain based on GEANT4 [26, 27].

Potential backgrounds include all the processes which lead to real prompt muons in the final state, like the SM processes W

Z

t¯ t, WW, WZ, and ZZ. However, the main contribution to the background is expected from processes giving a high production rate of secondary muons, such as decays in flight of K/

and heavy flavor decays in multi-jet production or muons due to cosmic rays, which do not point to the primary vertex position. The SM background samples are generated using

(W

and Z

Jets) and

(t¯ t, WW, WZ, and ZZ). For the multi-jet background evaluation, a data driven method is used. The cosmic ray background evaluation is also done on data. Additional pp interactions in the same and nearby bunch crossings (pile-up) are included in the simulation. All MC samples are re- weighted to reproduce the observed distribution of the mean number of interactions per bunch crossing in the data.

4 The kinematics of the sample

In order to understand the main characteristics of the signal sample that will define the analysis strategy a set of plots at the Monte Carlo generator level of the most relevant kinematics variables is presented.

Figure 2 shows the p

distribution of the leading

(right) and the

angle between the two

(left).

Due to the small mass of the

a boost factor of the order of 50 should be expected. The

pair is

27 m in radius and 13 m along thezaxis are the locations of the muon trigger system.

emitted approximately back-to-back, with a distortion of the distribution due to the emission of f

. The distribution of

R

(

(

between the two muons from the

and the p

of the muons are presented in Figure 3. Due to the small mass of the

the

below 0.1. The p

is relatively low but high enough to reach, in a large fraction of the cases, the MS, for which a minimum p

of 4 GeV is needed.

(GeV)

γd

leading pT

0 10 20 30 40 50 60 70 80 90 100 /GeV dγfraction of

0 0.005 0.01 0.015 0.02 0.025 0.03

=100GeV mH

=140GeV mH

Preliminary ATLAS

Simulation

γd

γd

φ

∆

-4 -3 -2 -1 0 1 2 3 4

pairs/0.08rad dγfraction of

0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08

=100GeV mH

=140GeV mH

Preliminary ATLAS

Simulation

Figure 2: Distributions of p

of the leading

(left) and

between the two

(right) for the signal MC samples with m

100 GeV and m

140 GeV.

µ

Rµ

∆ 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4

µµR∆ / 0.004 dγfraction of

0 0.02 0.04 0.06 0.08 0.1 0.12 0.14

=100GeV mH

=140GeV mH

Preliminary ATLAS

Simulation

(GeV)

µ

pT

0 20 40 60

/ GeVµfraction of

0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08

=100GeV mH

=140GeV mH

Preliminary ATLAS

Simulation

Figure 3:

R between the two muons from the

(left) and the p

distribution of the muons (right) from both

decay for the signal MC samples with m

100 GeV and m

140 GeV.

Finally Figure 4 shows the transverse missing energy E

distribution of the Monte Carlo sample

computed at the event generator level. Since the two f

are emitted similarly to the two

, back to back

in

E

is quite small and it will not be considered in the signal selection.

(GeV)

miss

ET

0 50 100 150 200 250

fraction of events/GeV

0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02 0.022 0.024

=100GeV mH

=140GeV mH

Preliminary ATLAS

Simulation

Figure 4: E

in the event at the event generator level, for the signal MC samples with m

100 GeV and m

140 GeV.

5 Data samples and trigger selection

The dataset used for this analysis was collected during the first part of 2011 at

s

7 TeV. Only periods in which all ATLAS subdetectors were operational are used. The total integrated luminosity [28, 29] is 1.94 fb

with a 3.7 % uncertainty. All events are required to have at least one reconstructed vertex along the beam line with at least three associated tracks, each with p

0.5 GeV. The primary interaction vertex is defined as being the vertex whose constituent tracks have the largest sum of p

.

This analysis deals with

displaced decays with final states containing only muons. Signal events are therefore characterized by a four-muon final state, with the four muons coming from two displaced decay vertices. Due to the relatively low p

of the muons and to the displaced decay vertex, a low p

multi muon trigger with muons reconstructed only in the MS is needed. In order to have an acceptably low trigger rate at a low p

threshold, a multiplicity of at least three muons must be required. Candidate events are collected using an unprescaled HLT trigger with three reconstructed muons of p

6 GeV, seeded by a L1 request of three muon RoIs. These muons are reconstructed only in the MS, since muon tracks, originating after the pixel detector, will not have a matching track in the ID tracking system. The trigger efficiency for the Monte Carlo signal samples, defined as the fraction of events passing the trigger requirement with respect to the events selected applying the analysis selection criteria (defined in Section 6) is 0.32

0.01

for m

0.01

for m

The main reason for the relatively low trigger efficiency is the small opening

R between the two muons of the

decay. From the

R distribution on the signal Monte Carlo

R

0.1 is expected, as shown in Figure 3. This value of

R is often smaller than the L1 trigger granularity; in this case the L1 will produce only one RoI. Therefore the trigger only fires if at least one of the

produces two distinct L1 RoIs. The single

RoI e

ciency,

(

), defined as the fraction of

passing the o

ine selection that give two (one) trigger RoIs is 0.296

0.004 (0.626

0.004 ) for m

100 GeV and 0.269

0.003 (0.653

0.003) for m

as function of the dark photon

and of the

R of the two muons from the

decay. The di

erent trigger granularity between the barrel and the endcap and the consequent change in

as function of

is clearly visible.

The systematic uncertainty on the trigger efficiency is estimated using a sample of J/ψ

with

∆

R

0.1 from collision data and on a corresponding sample of Monte Carlo events, using the tag and probe method. The tag is a combined muon, defined as a “standalone” muon combined at the ID level, associated with the trigger feature and with a “good ID track”, defined as a track having the number of hits in the pixel detector N

1, the number of hits in the Semi Conductor Tracker N

6 and the number of hits in the Transition Radiation Tracker N

6. The probe is a “good ID track” which, when combined with the tag track, makes an invariant mass within the J/ψ mass window. A muon RoI, different from the one associated with the tag, that matches in

and

with the probe, is searched for. The ratio between the number of probes with matched RoI and the number of probes without match gives the

ratio.

The values of

0.42

0.05 for the J/ψ

data and

0.39

0.05 for the corresponding Monte Carlo sample are obtained. The relative statistical uncertainty on the di

erence between these two estimates has been assumed as systematic uncertainty on the trigger efficiency, and it amount to 17%.

γd

η

-2 -1 0 1 2

2RoI∈

0.0 0.2 0.4 0.6 0.8 1.0 1.2

= 100 GeV mH

= 140 GeV mH

Preliminary Simulation ATLAS

µ

Rµ

∆ 0.00 0.01 0.02 0.03 0.04 0.05 0.06

2RoI∈

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

= 100 GeV mH

= 140 GeV mH

Preliminary Simulation ATLAS

Figure 5:

as a function of the

of the

and of the

R of the muon pair for the 100 GeV and 140 GeV Higgs boson mass Monte Carlo samples. The e

ciency decrease at small values of

R and the different trigger granularity between the barrel and the endcap are clearly visible.

6 Muon Jets reconstruction and event selection

MJs from

displaced decays are characterized by a pair of muons in a narrow cone, produced away from the primary vertex of the event. Consequently tracks reconstructed only in the MS with a good quality fit [30] are used. MJs are identified using a simple clustering algorithm that associates all the muons in cones of

R

0.2, starting with the muon with highest p

. The size

R

0.2 of the cone takes into account the multiple scattering of the muons in the calorimeters. All the muons found in the cone are associated to a MJ. After this procedure, if any muons are unassociated with a MJ the search is repeated for this remainder, again starting with the highest p

muon. This continues until all possible MJs are formed. The MJ direction and momentum are obtained from the vector sum over all muons in the MJ. Only MJs with two reconstructed muons are accepted and only events with two MJs are kept for the subsequent analysis.

The possible contribution to the background of SM processes which lead to real prompt muon pairs

in the final state, is evaluated using Monte Carlo samples. After the trigger and the requirements of

having two MJ in the event their contributions have been found to be negligible. The only significant

(GeV)

isol

ET

0 5 10 15 20

/ 0.6GeVµfraction of

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40

Ldt = 1.94fb-1

∫

=7TeV, s data, µ µ

→ Z

= 100 GeV mH

= 140 GeV mH

Preliminary ATLAS

Figure 6: Calorimetric isolation around the MJ direction. The calorimetric isolation variable E

is defined as the difference between the transverse energy E

in a cone

R

0.4 around the highest p

of the MJ and the E

in a cone

R

0.2. The distribution of the isolation variable E

for the two signal Monte Carlo samples with m

100 GeV and m

140 GeV is compared with the isolation around the muons of a Z

sample from collision data.

background sources are expected from processes giving a high production rate of secondary muons, such as decays in flight of K

and heavy flavor decays in multi-jet production or muons not pointing to the primary vertex due to cosmic rays.

In order to separate the signal from the background a number of discriminating variables have been studied. The multi-jet background can be significantly reduced by using calorimeter isolation around the MJ direction. The calorimetric isolation variable E

is defined as the difference between the transverse calorimetric energy E

in a cone

R

0.4 around the highest p

muon of the MJ and the E

in a cone

∆

R

0.2, see Figure 6.

The isolation modeling is validated for real isolated muons with a sample of muons coming from Z

decays. The distribution obtained on the Z

sample shows a good agreement with the distribution obtained on the signal Monte Carlo samples, as shown in Figure 6. To further improve the signal to background ratio, two additional discriminating variables are used:

between the two MJs and

p

for the MJs, defined as the sum of the transverse momentum of the tracks, identified with the ID, inside a cone

R

0.4 around the direction of the MJ

.

For the background coming from cosmic rays muons (mainly pairs of parallel cosmic muons crossing the detector) a cut on the impact parameters of the muon tracks with respect to the primary interaction vertex is used.

After the optimization of the cuts, the final set of selection criteria are the following:

•

topology cut: events are required to have two MJs, N

2

•

MJ isolation: require MJ isolation with E

5 GeV for both MJs in the event (see Figure 6)

3The muon tracks of MJ in the ID, if any, are not removed from the isolation sum, so that prompt muons, which give a reconstructed track in both the ID and MS, will contribute to theΣp^ID_T . As a consequence a cut onΣp_T^IDof a few GeVwill remove prompt MJs or MJs with very short decay length.

(GeV)

ID

pT

Σ

0 5 10 15 20 25 30

Entries/GeV

0 20 40 60 80 100 120

=140 GeV mH

data

=140 GeV mH

data ATLAS

= 7 TeV

s Preliminary

(2 MJ cut)

(GeV)

ID

pT

Σ

0 5 10 15 20 25 30

Entries/GeV

0 20 40 60 80 100 120

=140 GeV mH

data

=140 GeV mH

data ATLAS

= 7 TeV

s Preliminary

(isolation cut)

(GeV)

ID

pT

Σ

0 5 10 15 20 25 30

Entries/GeV

0 20 40 60 80 100 120

=140 GeV mH

data

=140 GeV mH

data ATLAS

= 7 TeV

s Preliminary

cut ) φ

∆ (

(GeV)

ID

pT

Σ

0 5 10 15 20 25 30

Entries/GeV

0 20 40 60 80 100

=140 GeV mH

data

=140 GeV mH

data ATLAS

= 7 TeV

s Preliminary

(impact par. cut)

Figure 7: Distribution of

p

of the MJ at di

erent steps of the event selection criteria: after the requirement of the signal topology (top left), after the isolation cut (top right), after the

cut (bottom left), and after the impact parameters cut (bottom right). The black dots are the data, the blue histogram is the signal Monte Carlo with m

140 GeV.

•

require

2 between the two MJs (see Figure 2)

•

require opposite charge of the two muons in a MJ

•

require a cut on the transverse and longitudinal impact parameters of the muons with respect to the primary vertex :

200 mm and

270 mm

•

require

p

3 GeV for both MJ (see Figure 7)

The result of the selection at different steps of the cut flow is shown in Figure 7 and summarized in Table 2. No events survive the selections in the data sample whereas the expected number of signal events from Monte Carlo simulation are 75 and 48 for Higgs boson masses of 100 and 140 GeV, respectively.

The method for the estimation of the background yields, quoted in Table 2, will be discussed in Section 6.

The resulting single

reconstruction e

ciency at the end of the selection, defined as the ratio between the number of

in the spectrometer acceptance (|

2.4) with both muons having p

6 GeV and the number of

at the end of the o

ine selection, is shown in Figure 8 as function of

the opening

R of the two muons from the

decay and of the decay length in the transverse plane, L

, of the

. The drop of the reconstruction efficiency at very short L

is a consequence of the

p

cut.

The systematic uncertainty on the reconstruction e

ciency is evaluated using a tag and probe method

by comparing the reconstruction e

ciency

on a sample of J/ψ

from collision data and

on a J/ψ

Monte Carlo sample. The tag is a combined muon (MS track with a combined

reconstruction in the ID); the probe is a “good ID track” which, when combined with the tag track,

makes an invariant mass within the J/ψ mass window. To measure the reconstruction e

ciency the ID

probe track is associated to a “standalone” muon track, different to the one associated to the tag. The

result is shown in Figure 9.

sample cosmic rays multi-jet total background mH =100 GeV mH=140 GeV DATA

NMJ=2 3.0±2.1 N/A N/A 135±11⁺₋₂₁²⁹ 90±9⁺¹⁷₋₁₃ 871

E^isol_T ≤5 GeV 3.0±2.1 N/A N/A 132±11⁺₋²⁸₂₁ 88±9⁺₋¹⁷₁₃ 219

|∆(φ)|>2 3.0±2.1 153±18±9 154.5±18.1±9 123±11⁺₋₁₉²⁶ 81±9⁺₋₁₂¹⁵ 104 ch=0 1.5±1.5 57±15±22 58.5±15.5±22 121±11⁺₋₁₉²⁶ 79±8⁺₋₁₂¹⁵ 80 d^µ₀,z^µ₀ 0⁺₋₀^1.64 111±39±63 111±39±63 105±10⁺₋₁₆²² 66±8⁺₋₁₀¹² 70 Σp^ID_T <3 GeV 0⁺₋₀^1.64 0.06±0.02⁺_−0.06^0.66 0.06±0.02⁺_−0.02^0.66 75±9⁺₋₁₂¹⁶ 48±7⁺⁹₋₇ 0

Table 2: Cut flow for the signal selection on signal Monte Carlo, the corresponding cosmic rays back- ground, the multi-jet background estimatation from the ABCD method (described in Section 6) and the data; all the numbers of events are normalized to an integrated luminosity of 1.94 fb

. The first uncer- tainties are statistical, and the second systematic.

The relative difference between the result obtained from the J/ψ

data and the J/ψ

MC sample in the same range of

R

0.1 of the signal is taken as systematic uncertainty on the reconstruction efficiency.

7 Multi-jet and cosmic rays backgrounds evaluation

In order to estimate the multi-jet background contamination in the signal region we use a data driven method, a 2D ABCD method slightly modified to cope with the problem of the very low statistics in the control regions. The ABCD method assumes that two variables can be identified, which are relatively uncorrelated, and which can each be used to separate signal and background. The assumption is made that the multi-jet background distribution can be factorized in the MJ isolation and the MJ

plane.

The region B, defined by E

5 GeV and

2, is the signal region (see Figure 10). Neglecting the signal contamination in region A, C and D (E

5 GeV or

2) the number of multi-jet background events in the signal region can be evaluated as N

N

N

. The problem of the very low statistics in the control regions is handled by not directly counting the number of events surviving the cut on the last discriminating variable

p

in the 3 control regions but by integrating from zero up to the nominal cut on the

p

a functional form describing the

p

distributions with parameters fitted in each region. The functional model has been varied to estimate systematic uncertanties.

The values of N

, N

, and N

as a function of the cut on the final discriminant variable

p

, are extracted by modeling the

p

distributions with bifurcated gaussian templates, with parameters fitted from the data in the corresponding regions, and by integrating the fitted function in the range 0

< Σ

p

3 GeV. The extracted yields are N

(7.1

1.5

)

10

, N

(1.81

1.0

)

10

and N

(1.51

0.07

)

10

and the estimated number of multi-jet background events in the signal region is N

0.06

0.02

. Possible sources of systematic uncertainties related to the background estimation method are also evaluated. The pdf template model is changed and the procedure to estimate the number of multi-jet background events in the signal region is repeated. The difference in N

is taken as systematic uncertainty of the modeling of the multi-jet background shape and it amounts to

The e

ect of a possible signal leakage in the background regions is also considered and it is found to be negligible.

The background induced by muons from cosmic rays showers is evaluated using events collected by

the same trigger that has been used on collision events, active when there are no collisions (empty bunch

γd

η

-2 -1 0 1 2

rec∈

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

= 100 GeV mH

= 140 GeV mH

Preliminary Simulation ATLAS

µ

Rµ

∆

0.00 0.01 0.02 0.03 0.04 0.05 0.06

rec∈

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

= 100 GeV mH

= 140 GeV mH

Preliminary Simulation ATLAS

xy (m) L

0 1 2 3 4 5 6 7

rec∈

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50

= 100 GeV mH

= 140 GeV mH

Preliminary Simulation ATLAS

Figure 8:

reconstruction efficiency

as a function of the

R and of the transverse decay length of the

for m

100 GeV and m

140 GeV. The reconstruction e

ciency is defined as the ratio between the number of

in the spectrometer acceptance (|

2.4) with both muons of p

6 GeV and the number of

at the end of the offline selection. The drop of the reconstruction efficiency at very short decay lengths is a consequence of the

p

cut.

crossings) and rescaling the events by the collision to empty ratio (1.26) and of the active time (1.18)

. No events survived the requirements on the impact parameters with respect to the primary vertex (|d

200 mm and

270 mm), resulting in a cosmic rays contamination estimate of 0

. The final yields for the different background sources are summarized in Table 2.

8 Systematic uncertainties

The following effects are considered as possible sources of systematic uncertainties:

4The trigger in the empty bunch crossing was not active in all the runs.

Tag-Probe

∆R

0.0 0.1 0.2 0.3 0.4

TP rec∈

0.0 0.2 0.4 0.6 0.8 1.0

data ψ J/

ψ MC J/

= 100 GeV mH

= 140 GeV mH

Preliminary ATLAS

= 7TeV s

Ldt = 1.94fb-1

∫

Figure 9: Tag and probe reconstruction efficiency

as a function of the

R between the two muons, evaluated on a sample of J/ψ

from collision data and on a corresponding sample of Monte Carlo events. For completeness the

for the signal Monte Carlo, evaluated with a similar tag and probe method is also shown.

Figure 10: Definition of regions used for the ABCD method.

• Luminosity

The overall normalization uncertainty of the integrated luminosity is 3.7 % [28, 29].

• Muon momentum resolution

The systematic uncertainty on the momentum resolution for standalone muons has been evaluated by shifting the momenta of the muons by scale factors derived from data-Monte Carlo agreement, and by observing the effect of this shifts on the signal efficiency. The overall effect of the muon momentum resolution uncertainty has been found to be negligible.

• Trigger

The systematic uncertainty on the single

trigger e

ciency, evaluated using a tag and probe

method is 17%.

• Reconstruction efficiency

The systematic uncertainty on the reconstruction e

ciency, evaluated using a tag and probe method from the single

reconstruction efficiency, is 13%.

• Effect of pile-up

The systematic uncertainty related to the e

ect of pile-up on the signal e

ciency is evaluated by comparing the number of signal events after all the selection criteria on the signal Monte Carlo sample, with and without the pile-up re-weighting applied. This systematic uncertainty has been found to be negligible.

• Effect ofΣ

p

Since the

p

cut could a

ect the minimum cτ value that can be excluded, the e

ect of this cut on the signal Monte Carlo has been studied. A variation of 10% on the

p

cut results in a relative variation of

1% on the signal.

• Background evaluation

The systematic uncertainties that can affect the background estimation are related to the data driven method used. The functional model used to fit

p

distribution is varied to evaluate the systematic related to the modeling shape, which also describe the e

ect of the

p

cut on the background estimation. This systematic uncertainty amount to

0.66 events. The effect of signal leakage is also evaluated and it has been found to be negligible.

9 Results and interpretation

The acceptance e

ciency for the selection criteria described above is evaluated for the signal Monte Carlo benchmark samples (see Table 1) as a function of mean cτ of the

. From the signal acceptance efficiency, the number of expected signal events assuming 100% branching fraction for H

X are predicted. These numbers, together with the expected number of background events (multi-jet and cosmic rays), and with the number of observed events surviving the selection criteria in 1.94 fb

, are used as input to obtain 95% upper confidence limits. Using the CLs method [31], the expected 95%

upper limits on the the cross section times branching fraction for the process H

X as function of the dark photon mean cτ for m

100 GeV and m

140 GeV are set. These limits are plotted on Figure 11 as a function of the mean cτ of the dark photon.

Assuming 100% branching fraction of H

X, a wide range of

mean cτ are excluded at the 95% CL, as shown in Table 3.

Higgs boson mass excluded cτrange

(GeV) (mm)

100 1 mm≤cτ≤670 mm 140 1 mm≤cτ≤430 mm

Table 3: Ranges of

mean cτ excluded at 95% CL for the signal with m

with m

140 GeV, assuming 100% branching fraction of H

X.

10 Conclusions

A search for a light Higgs boson decaying into a pair of hidden fermions ( f

), each of which decays to

a

and to a stable hidden fermion ( f

), resulting in two muon jets from the

decay in the final state,

[mm]

τ Dark photon c

1 10 10² 10³

+X) [pb] dγ dγ→BR(H×σ95% CL Limit on

0 10 20 30 40 50

= 100 GeV mH

(stat.+syst.) σ

± 2 expected

(stat.+syst.) σ

± 1 expected

ATLAS Preliminary

Ldt = 1.94 fb-1

∫

=7 TeV, s

observed limit expected limit

×100%

σSM

+X) = γd

γd

→

×BR(H σ

[mm]

τ Dark photon c

1 10 10² 10³

+X) [pb] dγ dγ→BR(H×σ95% CL Limit on

0 5 10 15 20 25

= 140 GeV mH

(stat.+syst.) σ

± 2 expected

(stat.+syst.) σ

± 1 expected

ATLAS Preliminary Ldt = 1.94 fb-1

∫

=7 TeV, s

observed limit expected limit

×100%

σSM

+X) = γd

γd

→

×BR(H σ

Figure 11: The expected 95% upper limits on the cross section times branching fraction for the process H

X as function of the dark photon mean cτ for the benchmark sample with m

100 GeV (left) and with m

curve shows the observed limit.

with the ATLAS detector is presented. In 1.94 fb

sample of

s

7 TeV proton-proton collisions no excess of events consistent with the light Higgs boson decay is observed. The observed data are consistent with the Standard Model background expectations. Limits are set on the cross section times branching fraction H

X. The

mean cτ ranges from 1 mm to 670 mm for the signal with m

m

CL assuming 100% branching fraction for H

X.

References

[1] F. Englert and R. Brout, Phys. Rev. Lett.

[2] P. Higgs, Phys. Lett.

, (1964) 132.

[3] C. H. G.S. Guralnik and T. Kibble, Phys. Rev. Lett.

[4] M. J. Strassler and K. M. Zurek, Echoes of a Hidden Valley at Hadron Colliders, Phys. Lett. B651 (2007) 374.

[5] N. Arkani-Hamed and N. Weiner, LHC Signals for a Superunified Theory of Dark Matter, JHEP 12 (2008) 104.

[6] T. Han, Z. Si, K. M. Zurek, and M. J. Strassler, Phenomenology of Hidden Valleys at Hadron Collid- ers, JHEP 07 (2008) 008.

[7] S. Gopalakrishna, S. Jung, and J. D. Wells, Higgs Boson Decays to Four Fermions Through an

Abelian Hidden Sector, Phys. Rev. D78 (2008) 055002.

[8] M. J. Strassler and K. M. Zurek, Discovering the Higgs Through Highly-Displaced Vertices, Phys.

Lett. B661 (2008) 263-267.

[9] M. Baumgart, C. Cheung, J. T. Ruderman, L. T. Wang, and I. Yavin, Non-Abelian Dark Sectors and Their Collider Signatures, JHEP 04 (2009) 014.

[10] C. Cheung, J. T. Ruderman, L. T. Wang, and I. Yavin, Lepton Jets in (Supersymmetric) Electroweak Processes, JHEP 04 (2010) 116.

[11] Y. Bai and Z. Han, Measuring the Dark Force at the LHC, Phys. Rev. Lett. 103 (2009) 051801.

[12] A. Falkowski, J. T. Ruderman, T. Volansky, and J. Zupan, Hidden Higgs Decaying to Lepton Jets, JHEP 05 (2010) 077.

[13] A. Falkowski, J. T. Ruderman, T. Volansky, and J. Zupan, Discovering Higgs Decays to Lepton Jets at Hadron Colliders, Phys. Rev. Lett. 105 (2010) 241801.

[14] G. Abbiendi et al. [OPAL Collaboration], Decay mode independent searches for new scalar bosons with the OPAL detector at LEP, Eur. Phys. J. C

, 311 (2003)[hep-ex/0206022].

[15] The ATLAS Collaboration, JINST

[16] V.M. Abazov et al. [D0 Collaboration], Search for Dark Photons from Supersymmetric Hidden Valleys, Phys. Rev. Lett. 103, 081802 (2009).

[17] V.M. Abazov et al. [D0 Collaboration], Search for Events with Leptonic Jets and Missing Trans- verse Energy in pp Collisions at

s

1.96 TeV , Phys. Rev. Lett. 105, 211802 (2010).

[18] The CMS Collaboration, Search for Light Resonances Decaying into Pairs of Muons as a Signal of New Physics, J. High Energy Phys. 07 (2011) 098.

[19] The ATLAS Collaboration, A Search for Lepton-Jets with Muons at ATLAS, ATLAS-CONF-2011- 076, 2011.

[20] The ATLAS Collaboration, Search for a light Higgs boson decaying to long-lived weakly- interacting particles in proton-proton collisions at

s

7 TeV with the ATLAS detector, Phys. Rev.

Lett. 108, 251801 (2012).

[21] The ATLAS Collaboration, Eur. Phys. J. C

(2012) 1849.

[22] LHC Higgs Cross Section Working Group, S. Dittmaier, C. Mariotti, G. Passarino, and R. Tanaka (Eds.), Handbook of LHC Higgs Cross Sections: 1. Inclusive Observables, CERN-2011-002 (CERN, Geneva, 2011).

[23] S. Mrenna, T. Sjostrand and P. Z. Skands, JHEP

[24] J. Alwall, M. Herquet, F. Maltoni, O. Mattelaer and T. Stelzer, MadGraph 5 : Going Beyond, FERMILAB-PUB-11-448-T. Jun 2011. 37 pp., JHEP 1106 (2011) 128, e-Print: arXiv:1106.0522 [hep-ph].

[25] P. Meade and M. Reece,“Bridge: Branching Ratio Inquiry

Decay Generated Events,arXiv:hep- ph/0703031.

[26] S. Agostinelli et al., Nuclear Instruments and Methods

[27] The ATLAS Collaboration, Eur. Phys. J. C

(2010) 823.

[28] The ATLAS Collaboration, Eur. Phys. J. C

(2011) 1630.

[29] The ATLAS Collaboration, ATLAS-CONF-2011-116, 2011.

[30] The ATLAS Collaboration, Expected Performance of the ATLAS Experiment Detector, Trigger and Physics, 2009. arXiv:0901.0512 [hep-ex].

[31] Read, Alexander L., Presentation of search results: The CL(s) technique, J. Phys., G28, 2002,

2693-2704.