Eur. Phys. J. C (2009) 59: 463–495DOI 10.1140/epjc/s10052-008-0746-8

Review

Higgs bosons at the LHC

Karl Jakobsa

Physikalisches Institut, Universität Freiburg, Hermann-Herder-Str. 3, 79104 Freiburg, Germany

Received: 7 July 2008 / Published online: 14 October 2008© Springer-Verlag / Società Italiana di Fisica 2008

Abstract The investigation of the dynamics responsible forelectroweak symmetry breaking is one of the prime tasks ofthe experiments at the CERN Large Hadron Collider (LHC).In this article, the potential of the ATLAS and CMS exper-iments for the discovery of a standard model Higgs bosonand for Higgs bosons in the minimal supersymmetric exten-sion is summarized. Emphasis is put on those studies whichhave been performed recently by the experimental collab-orations using a realistic simulation of the detector perfor-mance. This includes a discussion of the search for Higgsbosons using the vector boson-fusion mode, a discussion onthe measurement of Higgs boson parameters as well as a de-tailed review of the MSSM sector for different benchmarkscenarios.

1 Introduction

While the standard model of the electroweak and strong in-teractions is in excellent agreement with the many exper-imental measurements, the dynamics responsible for elec-troweak symmetry breaking is still unknown. Within thestandard model, the Higgs mechanism [1–4] is invoked, i.e.a doublet of complex scalar fields is introduced of which asingle neutral scalar particle, the Higgs boson, remains aftersymmetry breaking.

The Higgs boson is the only standard model particle thathas not been discovered so far. The direct search at thee+e− collider LEP has led to a lower bound on its massof 114.4 GeV/c2 [5]. Indirectly, high-precision electroweakdata constrain the mass of the Higgs boson via their sensitiv-ity to loop corrections. Assuming the overall validity of thestandard model, a global fit [6] to all electroweak data leadsto mH = 87+36

−27 GeV/c2. On the basis of the present theo-retical knowledge, the Higgs sector in the standard modelremains largely unconstrained. While there is no direct pre-diction for the mass of the Higgs boson, an upper limit of∼1 TeV/c2 can be inferred from unitarity arguments [7–9].

a e-mail: karl.jakobs@uni-freiburg.de

Further constraints can be derived under the assumptionthat the standard model is valid only up to a cutoff energyscale Λ, beyond which new physics becomes relevant. Re-quiring that the electroweak vacuum is stable and that thestandard model remains perturbative allows one to set up-per and lower bounds on the Higgs boson mass [10–20]. Fora cutoff scale of the order of the Planck mass, the Higgsboson mass is required to be in the range 130 < mH <

190 GeV/c2. If new physics appears at lower mass scales,the bound becomes weaker, e.g. for Λ = 1 TeV the Higgsboson mass is constrained to be in the range 50 < mH <

800 GeV/c2.The minimal supersymmetric standard model [21–23]

contains two complex Higgs doublets, leading to five phys-ical Higgs bosons after electroweak symmetry breaking:three neutral (two CP-even, h,H , and one CP-odd, A) and apair of charged Higgs bosons, H±. At tree level, the Higgssector of the MSSM is fully specified by two parameters,generally chosen to be mA, the mass of the CP-odd Higgsboson, and tanβ , the ratio of the vacuum expectation valuesof the two Higgs doublets. Radiative corrections modify thetree-level relations significantly. Loop corrections are sensi-tive to the mass of the top quark, to the mass of the scalarparticles and in particular to mixing in the stop sector. Thelargest values for the mass of the Higgs boson h are reachedfor large mixing, characterized by large values of the mix-ing parameter Xt := At − μ cotβ , where At is the trilinearcoupling and μ is the Higgs mass parameter. If the full one-loop and the dominant two-loop contributions are included[24–27], the upper bound on the mass of the light Higgs bo-son h is expected to be around 135 GeV/c2. While the lightneutral Higgs boson may be difficult to distinguish fromthe standard model Higgs boson, the other heavier Higgsbosons are a distinctive signal of physics beyond the stan-dard model. The masses of the heavier Higgs bosons H,A

and H± are often almost degenerate.If the global fit to the electroweak precision data is re-

peated within the MSSM and constraints from heavy fla-vor physics and from the abundance of cold dark matter

464 Eur. Phys. J. C (2009) 59: 463–495

in the universe are included [28], the mass of the lightestHiggs boson can be tightly constraint to be mH = 110+10

−8 ±3 (theo.) GeV/c2.

At present, the Fermilab Tevatron pp collider with acenter-of-mass energy of 1.96 TeV continues to be the lead-ing accelerator exploring the energy frontier and its preem-inence will continue until the LHC experiments have col-lected sufficient amounts of good quality and well under-stood data. Until the end of 2007 the two experiments CDFand DØ have analyzed data corresponding to an integratedluminosity of ∼2.0 fb−1. So far no evidence for the produc-tion of Higgs bosons has been found [29]. The data takingperiod is expected to be continued until the end of 2009,with a possible extension into the year 2010, leading to anincrease in the integrated luminosity to 5.5 or 7.0 fb−1, re-spectively. Based on the present experience and assumingsome improvements in the analysis techniques, both Teva-tron experiments combined are expected to reach sensitiv-ity to the production of a standard model Higgs boson upto masses of 180 GeV/c2 at 95% confidence level (C.L.).Within the context of the MSSM, also most of the parame-ter space can be tested at 95% C.L. However, even with thefull integrated luminosity of 7 fb−1 and after combining allchannels and both experiments, a 5σ discovery will not bepossible at the Tevatron, leaving it to the LHC to explore anddiscover Higgs bosons over the full parameter range of boththe standard model and the MSSM.

The Higgs boson production and decay processes and thestatus of the calculation of higher-order QCD correctionsare presented in Sect. 2 of this article. The setup of the twogeneral purpose experiments ATLAS and CMS at the LHCare briefly discussed in Sect. 3. The expected performanceof these experiments for the search for the standard modelHiggs boson is presented in Sect. 4. The expectations for the

measurement of important Higgs boson parameters are dis-cussed in Sect. 5, before the discovery potential of MSSMHiggs bosons in various benchmark scenarios is summa-rized in Sect. 6.

2 Higgs bosons at hadron colliders

2.1 Higgs boson production

At hadron colliders, Higgs bosons can be produced via fourdifferent production mechanisms:

• gluon fusion, gg → H , which is mediated at lowest orderby a heavy quark loop;

• vector boson fusion (VBF), qq → qqH ;• associated production of a Higgs boson with weak gauge

bosons, qq → W/ZH (Higgs Strahlung, Drell–Yan-likeproduction);

• associated Higgs boson production with heavy quarks,gg,qq → t tH , gg,qq → bbH (and gb → bH ).

The lowest-order production cross sections for the fourdifferent processes are shown in Fig. 1(left) for the LHC col-lider as a function of the Higgs boson mass [30]. The domi-nant production mode is the gluon-fusion process, followedby vector boson fusion. In the low mass region it amountsat leading order to about 20% of the gluon-fusion cross sec-tion, whereas it reaches the same level for masses around800 GeV/c2. The associated WH , ZH and t tH productionprocesses are relevant only for the search of a light standardmodel Higgs boson with a mass close to the LEP limit.

For all production processes higher-order QCD correc-tions have been calculated. In particular, significant progresshas been made during the last years in the calculation of

Fig. 1 Leading-order (left) and next-to-leading order (right) produc-tion cross sections for a standard model Higgs boson as a function ofthe Higgs boson mass at the LHC. In the cross section calculation the

CTEQ6L1 and CTEQ6M structure function parametrization have beenused respectively (The calculations have been performed by Spira [30])

Eur. Phys. J. C (2009) 59: 463–495 465

QCD corrections for the gluon-fusion and for the associatedt tH and bbH production processes.

Already more than ten years ago, the next-to-leading or-der (NLO) QCD corrections to the gluon-fusion processhave been calculated and have been found to be large[31–36]. Their calculation appears to be challenging, sincealready in the leading-order diagram a massive one-looptriangle appears. The NLO calculation showed a signifi-cant increase of the predicted total cross section by about50–100%. These large corrections stimulated the calcula-tion of the next-to-next-to-leading order (NNLO) correc-tions, to which many authors contributed [37–42] and whichhas been completed in the heavy top-quark limit (mt → ∞).In this limit the calculation of the Feynman diagrams is con-siderably simplified. The approximation has been tested atNLO and has been found to agree within 5% with the fullNLO calculation up to mH = 2mt , if the NLO result ob-tained with (mt → ∞) is rescaled with the ratio of the LOcross sections calculated with a finite top mass and with(mt → ∞). For larger Higgs boson masses still a surpris-ingly good agreement is found, e.g. the approximation devi-ates from the exact result by only 10% at mH = 1 TeV/c2, ifthe same rescaling is applied [43, 44]. Based on these obser-vations, it appears to be reasonable to apply the heavy top-quark approximation also at NNLO. The results for the totalcross section show a modest increase between the NLO andNNLO calculation at the level of 10–20%, indicating that anicely converging perturbative series seems to be emerging.The results of the LO, NLO and NNLO cross section calcu-lations are shown in Fig. 2(left).

In particular, a clear reduction of the uncertainty due tohigher-order corrections has been observed, which is esti-mated to be about 15%, based on variations of the renormal-ization scale [43, 44].

Another important and challenging theoretical calcula-tion constitutes the NLO calculation of the QCD correctionsto the associated t tH production. Results have been pub-lished recently for a finite top-quark mass [46–50]. Alsoin this case a dramatic reduction in the variation of thecross section prediction with the renormalization and fac-torization scale μ has been found [46–50], as can be seenin Fig. 2(right), where the LO and NLO cross sectionsare shown as a function of the scale. For the choice μ =mt + mH /2, for example, the NLO correction is found toresult in a small increase of the cross section, of the orderof 20%.

In the standard model, the cross section for producing aHiggs boson in association with b quarks is relatively small.However, in a supersymmetric theory with a large valueof tanβ , the b-quark Yukawa coupling can be strongly en-hanced and Higgs boson production in association with b

quarks becomes the dominant production mechanism. Crosssection calculations including next-to-leading order correc-tions have been presented in two different approaches [51].In the so-called four-flavor scheme, no b quarks are presentin the initial state and the lowest-order processes are thegg → bbh and qq̄ → bbh tree-level processes. Due to thesplitting of gluons into bb pairs and the small b-quark mass,the inclusive cross section is affected by large logarithmsand the convergence of the perturbative expansion may bepoor. Depending on the final state considered, the conver-gence can be improved by summing the logarithms to allorders in perturbation theory through the use of b-quarkparton distributions, i.e. moving to a five-flavor scheme[52–54]. In this scheme, the lowest-order inclusive processis bb → h. The first-order correction to this process includesthe process gb → bh.

In many analyses two high-pT b quarks are experimen-tally required to improve the signal-to-background ratio.

Fig. 2 (Left): Gluon-fusion production cross section for a standardmodel Higgs boson at the LHC at leading (dotted), next-to-leading(dashed) and next-to-next-to-leading order (solid). The upper (lower)curve of each pair corresponds to a choice of the renormalization andfactorization scale of μR = μF = mH /2 (μR = μF = 2mH ) (from

Ref. [45]; see also Refs. [40–42]). (Right): variation of the t tH pro-duction cross section at the LHC with the renormalization and factor-ization scale μ = μR = μF , varied around the value μ0 = mt + mH /2in leading and next-to-leading order (from Refs. [46–50])

466 Eur. Phys. J. C (2009) 59: 463–495

The leading subprocess in this region of phase space isgg → bbh. The relevant production cross section, imple-menting parton-level cuts on the b quarks that closely re-produce the experimental cuts, have been computed at NLOin Refs. [55, 56]. The NLO corrections modify the leading-order predictions by less than 50% at the LHC. For a cutof pT > 20 GeV/c and |η| < 2.0 (2.5) for the b quarks andusing μ = (2mb + mh)/4 as factorization and renormaliza-tion scale, the NLO corrections are negative for small Higgsboson masses around 120 GeV/c2 and positive for largemasses, with a cross-over point around 300 GeV/c2.

For the cases of one and no tagged b-jets in the final state,results for the relevant cross sections can be calculated inthe four- and five-flavor schemes. The two calculations rep-resent different perturbative expansions of the same physicsprocess and should agree at sufficiently high order. The NLOfour-flavor result is obtained by integration over one of the b

quarks in the gg → bbh calculation [55, 56]. For the one-jetcase, this calculation is found to agree within their respec-tive scale uncertainties with the NLO calculation performedin the five-flavor scheme [57]. For the inclusive cross section(no tagged b jets) the five-flavor calculation has been per-formed to NLO in Refs. [58, 59] and to NNLO in Ref. [60].Again, within the large respective uncertainties, agreementbetween the NLO four- and NNLO five-flavor calculation isfound for small Higgs boson masses, while for large Higgsboson masses the five-flavor scheme tends to yield largercross sections. This represents major progress compared tothe situation of several years ago, when large discrepanciesbetween the NLO bb → h and the LO gg → bbh calcula-tion had been reported [61–64].

Next-to-leading order calculations of the productioncross sections are also available for the remaining two pro-duction mechanisms: WH,ZH and qqH production. Forthe vector boson-fusion process the NLO corrections are

found to be moderate [65–67], i.e. at the level of 10%.The NLO QCD corrections for the associated productionof a Higgs boson with a vector boson can be derived fromthe Drell–Yan process and give a 30% increase with re-spect to the leading-order prediction [68, 69]. Recently, theNNLO QCD corrections for this process have been calcu-lated [70, 71]. For the Drell–Yan type corrections, a mod-erate increase of the NLO cross sections of up to 3% isfound for the LHC for a Higgs boson mass in the range100 < mH < 300 GeV/c2. For the ZH associated produc-tion, additional gluon-fusion contributions appear at NNLO(a triangular diagram gg → Z∗ → ZH and a box diagramwhere the Z and H bosons are emitted from internal quarklines). These contributions have been found to increase theZH production cross section at the LHC in the low massregion by about 10%.

The production cross sections for the four differentprocesses including the NLO QCD corrections are shown inFig. 1(right) as a function of the Higgs boson mass [30]. Fora more detailed review of the theoretical aspects of Higgsboson production the reader is referred to Refs. [75–77].

2.2 Higgs boson decays

The branching fractions of the standard model Higgs bosonare shown in Fig. 3(left) as a function of the Higgs bosonmass. They have been calculated taking into account bothelectroweak and QCD corrections [72, 73]. The latter in-clude logarithmic corrections which are sizable in partic-ular for decays into b or c quarks. Large logarithms areresummed by using both the running quark mass and thestrong coupling constant evaluated at the scale of the Higgsboson mass. When kinematically accessible, decays of thestandard model Higgs boson into vector boson pairs WW or

H

Fig. 3 Branching fractions (left) and total decay width (right) of the standard model Higgs boson as a function of Higgs boson mass (fromRefs. [72, 73])

Eur. Phys. J. C (2009) 59: 463–495 467

Fig. 4 Branching fractions of the MSSM Higgs bosons as a function of their masses for tanβ = 3 and tanβ = 30 assuming a SUSY mass scaleof 1 TeV/c2 and vanishing mixing (from Refs. [72, 73])

ZZ dominate over all other decay modes. Above the kine-matic threshold, the branching fraction into t t can reach upto 20%. All other fermionic decays are only relevant forHiggs boson masses below 2 mW , with H → bb̄ dominatingbelow 140 GeV/c2. The branching fractions for H → ττ

and H → gg both reach up to about 8% at Higgs bosonmasses between 100 and 120 GeV/c2. Decays into two pho-tons, which are of interest due to their relatively clean ex-perimental signature, can proceed via fermion and W loopswith a branching fraction of up to 2 × 10−3 at low Higgsboson masses.

Compared to the mass resolution of hadron collider ex-periments, the total decay width of the standard modelHiggs boson is small at low masses and becomes signifi-cant only above the threshold for decays into ZZ, as shownin Fig. 3(right). At mH = 1 TeV/c2, the Higgs resonanceis broad with a width of about 600 GeV/c2. In this massregime, the Higgs field is coupling strongly, resulting inlarge NNLO corrections. With increasing coupling, the peakposition of the Higgs boson resonance does not increase be-yond a saturation value close to 1 TeV/c2, as found in bothperturbative and non-perturbative calculations [74].

Within the MSSM, the branching fractions of five phys-ical Higgs bosons have to be considered as a function oftheir masses as well as tanβ and the masses of the SUSY

particles. In Fig. 4 the branching fractions of all MSSMHiggs bosons are shown assuming that supersymmetric par-ticles are heavy enough to be neglected [72, 73]. The neutralHiggs bosons decay dominantly into bb̄ and τ+τ− at largetanβ or for masses below 150 GeV/c2 (up to 2mt for theCP-odd Higgs boson A). Decays into WW , ZZ and pho-tons are generally suppressed by kinematics as well as theHiggs couplings and become relevant only in the decouplinglimit mA → ∞, where the light CP-even Higgs boson h ef-fectively behaves like a standard model Higgs boson whileall other MSSM Higgs bosons are heavy. Charged Higgsbosons preferably decay into tb if accessible. For massesbelow mt + mb , the decay H± → τν dominates, with smallcontributions from H± → cb and H± → cs. In addition,both charged and heavy neutral Higgs bosons can decay intolighter Higgs bosons: H± → Wh, WA as well as H → hh,AA, ZA and A → Zh. Generally, the branching fractions ofthese decay modes are significant only at small tanβ .

Supersymmetric particles can influence the phenomenol-ogy of Higgs decays either via loop effects or, if lightenough, as viable decay modes. In particular, Higgs bosondecays to charginos, neutralinos and third generation sfermi-ons can be relevant if kinematically allowed. With R-parityconserved, Higgs boson decays into the lightest neutralinoare not directly detectable (“invisible decay”).

468 Eur. Phys. J. C (2009) 59: 463–495

3 The LHC experiments

After a construction time of more than ten years, the LargeHadron Collider (LHC) has started operation in 2008 andwill collide protons at a center-of-mass energy of 14 TeV(after an initial commissioning run at 10 TeV). This ma-chine will open up the possibility to explore the TeV en-ergy range, which plays a key role in the investigation ofthe electroweak symmetry breaking. Two experiments, AT-LAS and CMS, have been designed and optimized as gen-eral purpose pp detectors, capable of running at high lumi-nosity (L = 1034 cm−2 s−1) and detecting a variety of final-state signatures. For details on the detector layout and per-formance, the reader is referred to Refs. [78, 79]. Only abrief summary of the main design features and the status ofthe commissioning is given in the following.

Both experiments use a superconducting solenoid aroundthe inner detector cavity to measure the track momenta. Pat-tern recognition, momentum and vertex measurements areachieved with a combination of high-resolution silicon pixeland strip detectors. In the ATLAS experiment, tracking isperformed in a 2 T magnetic field. Electron identificationis enhanced with a continuous straw-tube tracking detectorwith transition radiation capability in the outer part of thetracking volume. Due to the presence of silicon strip andpixel detectors, both experiments will be able to perform b

quark tagging using impact parameter measurements and thereconstruction of secondary vertices.

The calorimeter of the ATLAS experiment consists of aninner barrel cylinder and end-caps, using the intrinsicallyradiation resistant Liquid Argon (LAr) technology. Over

the full length, this calorimeter is surrounded by a novelhadronic calorimeter using iron as absorber and scintillatingtiles as active material. The barrel part of the LAr calorime-try is an electromagnetic accordion calorimeter [78], witha highly granular first sampling (integrated preshower de-tector). In the end-cap region (1.5 < |η| < 3.2) the LArtechnology is used for both electromagnetic and hadroniccalorimeters. In order to achieve a good jet and missingtransverse energy (/ET ) measurement, the calorimeter cov-erage is extended down to |η| < 4.9 using a special for-ward LAr calorimeter with rod-shaped electrodes in a tung-sten matrix. The CMS calorimeter system [79] consists ofa high-resolution lead tungstate (PbWO4) crystal calorime-ter, complemented by a hadronic copper-scintillator sand-wich calorimeter. Both the electromagnetic and a large frac-tion of the hadronic calorimeter are located inside the coilof the solenoid. The choice of the detection technique is to alarge extent motivated by the search for the Higgs boson inthe H → γ γ decay channel, which requires both an excel-lent electromagnetic energy and angular resolution.

In addition to the inner solenoid, the ATLAS detector haslarge superconducting air-core toroids consisting of inde-pendent coils arranged outside the calorimetry. This allowsfor a stand-alone muon momentum measurement with threestations of high-precision tracking chambers at the inner andouter radius and in the middle of the air-core toroid. In theCMS experiment the magnetic flux of the large solenoid isreturned through a 1.8 m thick saturated iron return yoke(1.8 T) which is instrumented with muon chambers. A sin-gle magnet thus provides the necessary bending power for

Fig. 5 Picture of theinstallation of the ATLASdetector before the endcapcalorimeter system is movedinto the toroid system (May2006). Four of the eightsuperconducting toroid coils arevisible

Eur. Phys. J. C (2009) 59: 463–495 469

Fig. 6 Picture of the CMSinstallation; the inner tracker islowered down into theexperimental hall (December2007)

precise tracking in the inner detector and in the muon spec-trometer. The magnetic field in the central cavity, in whichthe inner detectors are located, is 4 T.

It is assumed that after the commissioning phase a lu-minosity of 1033 cm−2 s−1, hereafter called low luminos-ity, can be achieved. This value is supposed to increase af-ter a few years of operation to the design luminosity of1034 cm−2 s−1, hereafter called high luminosity. Integratedluminosities of 10 and 100 fb−1 should therefore be col-lected at the LHC after about one to two and four to fiveyears of data taking, respectively.

The various subdetector components of both experimentshave been constructed over the past years and nearly all ofthem are meanwhile, i.e. Fall 2008, installed in the experi-mental halls at CERN. As an example the insertion of theATLAS endcap calorimeter into the toroid system is shownin Fig. 5. The CMS experiments has first been assembledand many detector components have been commissioned inthe surface hall, before the various slices of the experimentswere lowered down into the experimental hall. In Fig. 6 thedetector is shown during installation in the underground hall,while the inner tracker is lowered down through the accessshaft. It is expected that both detectors will be ready to takefirst collision data in 2008.

4 Search for the standard model Higgs boson

The standard model Higgs boson is searched for at the LHCin various decay channels, the choice of which is givenby the signal rates and the signal-to-background ratios inthe different mass regions. The search strategies and back-ground rejection methods have been established in many

studies over the past ten years [80–82]. Originally, inclusivefinal states had been considered; among them the well estab-lished H → γ γ and H → ZZ(∗) → decay channels.More exclusive channels have been considered in the lowmass region by searching for Higgs boson decays in bb orγ γ in association with a lepton from the decay of an accom-panying W or Z boson or a top quark. The search for a stan-dard model Higgs boson in the intermediate mass region canbe extended by using the vector boson-fusion mode and ex-ploiting forward jet tagging, which had been proposed in theliterature several years ago [83–87]. In this section, a briefsummary of the standard model Higgs boson discovery po-tential at the LHC is given.

Due to the enormous progress in the calculation ofhigher-order QCD corrections for signals and backgroundsover the past few years, the LHC physics studies have startedto use these calculations. Only in cases where the correctionsfor the major backgrounds are not available or in older stud-ies, performance figures are evaluated by using Born-levelpredictions for both signals and backgrounds.

In addition, more detailed and better understood recon-struction methods have been established by the experimentalcollaborations. They are partially based on test beam mea-surements of various detector components or rely on exten-sive Monte Carlo simulations of the detailed detector re-sponse [88].

4.1 Inclusive Higgs boson searches

Several important channels for Higgs boson discovery atthe LHC have been discussed extensively in the literature.Among those channels are the H → γ γ decay mode, thegold plated decay channel H → ZZ(∗) → 4 as well as the

470 Eur. Phys. J. C (2009) 59: 463–495

decay channel H → WW(∗) → νν. If no additional parti-cles except the Higgs boson decay products are searched for,the production via gluon fusion provides the largest contri-bution to the signal event yields.

4.1.1 H → ZZ(∗) decays

The decay channel H → ZZ(∗) → provides a ratherclean signature in the intermediate mass region115 GeV/c2 < mH < 2mZ . In addition to the irreduciblebackgrounds from ZZ∗ and Zγ ∗ production, there are largereducible backgrounds from t t and Zbb production. Dueto the large production cross section, the t t events domi-nate at production level, whereas the Zbb events contain agenuine Z boson in the final state and are therefore moredifficult to reject. In addition, there is background fromZZ continuum production, where one of the Z bosonsdecays into a τ pair, with subsequent leptonic decays ofthe τ leptons, and the other Z decays into an electron ormuon pair. It has been shown that in both LHC experimentsthe reducible backgrounds can be suppressed well belowthe level of the irreducible background from ZZ∗ → 4.Calorimeter and track isolation together with impact para-meter measurements can be used to achieve the necessarybackground rejection [80–82]. Assuming an integrated lu-minosity of 30 fb−1, the H → ZZ∗ → 4 signal can beobserved with a significance of more than 5σ in the massrange 130 < mH < 180 GeV/c2, except for a narrow regionaround 170 GeV/c2, where the branching ratio is suppresseddue to the opening up of the WW decay mode.

For Higgs boson masses in the range 180 < mH <

700 GeV/c2, the H → ZZ → 4 decay mode is the most re-liable one for the discovery of a standard model Higgs bosonat the LHC. The expected background, which is dominatedby the continuum production of Z boson pairs, is smallerthan the signal. In this mass range the Higgs boson width

grows rapidly with increasing mH , and it dominates overthe experimental mass resolution for mH > 300 GeV/c2.The momenta of the final-state leptons are high and theirmeasurement does not put severe requirements on the de-tector performance. The H → ZZ → 4 signal would beeasily observable above the ZZ → 4 continuum back-ground after less than one year of low luminosity opera-tion for 200 < mH < 600 GeV/c2. As examples of signalreconstruction above the background, the expected signalsfrom Higgs bosons with masses of 150 and 300 GeV/c2 areshown in Fig. 7 for an integrated luminosity of only 10 fb−1

in the ATLAS experiment. It should be mentioned that theinterference effect between the resonant signal and the non-resonant background, as discussed in Ref. [89], has not beentaken into account.

For larger values of mH , the Higgs boson signal be-comes very broad and the signal rate drops rapidly. In thehigh mass region, the decay modes H → ZZ → νν andH → ZZ → jj provide additional discovery potential[80–82] and allow one to extend the 5σ discovery range upto ∼1 TeV/c2.

4.1.2 H → γ γ decays

The decay H → γ γ is a rare decay mode, which is only de-tectable in a limited Higgs boson mass region between 80and 150 GeV/c2, where both the production cross sectionand the decay branching ratio are relatively large. Excel-lent energy and angular resolution are required to observethe narrow mass peak above the irreducible prompt γ γ con-tinuum. In addition, there is a large reducible backgroundresulting from direct photon production or from two-jet pro-duction via QCD processes. Using a realistic detector sim-ulation, it has been demonstrated [80–82] that the requiredrejection can be achieved and that the residual jet–jet and γ –jet backgrounds can be brought well below the irreducibleγ γ background.

Fig. 7 Expected H → ZZ(∗) → 4 signal above the background for mH = 150 and 300 GeV/c2 and for an integrated luminosity of 10 fb−1 inthe ATLAS experiment

Eur. Phys. J. C (2009) 59: 463–495 471

Fig. 8 Reconstructed γ γ invariant mass distribution of H → γ γ sig-nals at different masses (mH = 115, 120, 130 and 140 GeV/c2) andthe various backgrounds for an integrated luminosity of 1 fb−1 in theCMS experiment. It should be noted that the Higgs signals are scaledby a factor 10 (from Refs. [81, 82])

It should be stressed that due to the large amount of ma-terial in the central detectors of the LHC experiments (from∼40% of a radiation length at η ∼ 0 to 140% towards thebarrel-endcap transition region [78, 79]), a large fractionof converted photons appears and results in a degraded γ γ

mass resolution as compared to unconverted photons.On the theoretical side, the large K factors for the gluon-

fusion process lead to an improved signal significance. Inaddition, it is anticipated that multivariate analysis strate-gies like neural networks and likelihood methods can beused to increase the signal significance. For a Higgs bosonwith a mass of 130 GeV/c2, for example, a signal signif-icance of around 6σ is expected for both experiments, as-suming data corresponding to an integrated luminosity of30 fb−1 and NLO cross sections. Despite the intrinsicallybetter energy resolution of the CMS calorimeter, the ex-pected performance for a Higgs boson discovery in the γ γ

decay mode is found to be comparable between the two ex-periments. As an example of signal reconstruction abovebackground, the expected signal from a Higgs boson witha mass of 130 GeV/c2 in the CMS experiment is shown inFig. 8, assuming an integrated luminosity of 100 fb−1.

4.1.3 H → WW(∗) decays

For Higgs boson masses around 170 GeV/c2, for which theZZ∗ branching ratio is suppressed, the discovery potentialcan be enhanced by searching for the H → WW(∗) → νν

decay [80–82, 90]. In this mode it is not possible to recon-struct a Higgs boson mass peak. Instead, an excess of eventsabove the expected backgrounds can be observed and usedto establish the presence of a Higgs boson signal. Usually,the transverse mass computed from the leptons and the miss-

ing transverse momentum, mT =√

2pT /ET (1 − cosφ),

is used to discriminate between signal and background. TheWW , t t and single-top production processes constitute se-vere backgrounds and the signal significance depends criti-cally on their absolute knowledge. In addition, a so-called jetveto is applied; i.e., it is required that there is no jet activityin the central region of the detector. An estimate of the sig-nal significance relies therefore heavily on a reliable MonteCarlo description of this jet activity. It is anticipated that af-ter relaxing cuts, the Monte Carlo predictions for those back-grounds can be normalized to the data in regions where onlya small fraction of signal events is expected.

It has recently been shown that the gg → WW back-ground contribution increases the background in particularin the regions of phase space selected by the selection cri-teria, i.e. at small values of φ between the two final-stateleptons [91, 92]. Although this background constitutes only5% of the qq → WW background before cuts, its fractionincreases to about 30% after the final selection criteria.

Due to the large signal rate this channel may provide foran early discovery if the backgrounds can be reliably con-trolled. The CMS experiment claims that for a Higgs bosonwith a mass of 165 GeV/c2 a signal with a 5σ significancecan already be claimed with data corresponding to an inte-grated luminosity of only ∼2 fb−1 [81, 82].

For Higgs boson masses beyond ∼800 GeV/c2 the decaymode H → WW → ν jj provides additional discovery po-tential [80–82]. The branching ratio is about 150 times largerthan for the H → ZZ → 4 decay channel, thus providingthe largest possible signal rate with at least one charged lep-ton in the final state. An important contribution to the pro-duction cross section of such a heavy Higgs boson is the vec-tor boson-fusion process, resulting in the production of twofinal-state quark jets in the high rapidity regions in associ-ation with the Higgs boson (see Sect. 4.2). This productionmechanism provides an important additional signature forisolating the signal from the background. The experimentalchallenge in the reconstruction of these events is twofold:(i) the reconstruction of high-pT W decays to two jets and(ii) the tagging of the forward jets in the presence of largepile-up. As an example, for a Higgs boson mass of 1 TeV/c2

and an integrated luminosity of 100 fb−1, about 60 signalevents above a background of 20 events are expected in theATLAS experiment [80]. However, there is no substantialdifference between the shapes of the signal and backgrounddistributions and some years of running may be needed be-fore a signal could be established.

4.2 Higgs boson searches using vector boson fusion

In recent studies it has been demonstrated that, not only inthe high mass but also in the intermediate mass range, thediscovery potential can be significantly increased by per-forming a search for Higgs boson production in the vector

472 Eur. Phys. J. C (2009) 59: 463–495

boson-fusion mode [81–87, 93]. Although the contributionto the cross section in the intermediate mass range amountsat leading order only to about 20% of the total productioncross section, the additional event characteristics can be ex-ploited to suppress the large backgrounds. In vector boson-fusion events, the Higgs boson is accompanied by two jetsin the forward regions of the detector, originating from theinitial quarks that emit the vector bosons. On the other hand,central jet activity is suppressed due to the lack of color ex-change between the initial-state quarks. This is in contrastto most background processes, where color flow appears inthe t-channel. Jet tagging in the forward region of the de-tector together with a veto of jet activity in the central re-gion are therefore powerful tools to enhance the signal-to-background ratio.

The performance of the detectors for forward jet tagginghas been studied in a detailed simulation [81, 82, 93]. In thestudy presented in Ref. [93], the two tag jets are searchedfor over the full calorimeter coverage (|η| < 4.9). The jetswith the highest pT in the positive and negative regions ofpseudorapidity are taken to be the tag jet candidates. Eachjet is required to have a transverse energy of at least 20 GeV.The pseudorapidity distribution and the separation η be-tween the two tag jets, as found from the parton-level in-formation, is shown in Fig. 9 for signal events with mH =160 GeV/c2. For the tagging algorithm described above,distributions at parton and reconstruction level are in goodagreement. For comparison, the corresponding distributionsfor tag jets as reconstructed in t t background events are su-perimposed on the figure. From these distributions it canbe concluded that a large pseudorapidity separation can be

used for the discrimination between signal and backgroundsproduced via QCD processes. The detailed simulations per-formed have demonstrated that tag jets can be reliably re-constructed in the LHC experiments.

As pointed out above, a veto against jets in the central re-gion will be an important tool to suppress background fromQCD processes. It should be noted that a reliable estimate ofthe jet veto efficiency is difficult to obtain. Present estimatesare based on leading-order parton shower Monte Carlo cal-culations and might be affected by sizable uncertainties.

In the following, the Higgs boson search in the vectorboson-fusion mode for various final states is discussed.

4.2.1 qqH → qqWW(∗)

According to Monte Carlo studies, the LHC experimentshave a large discovery potential in the H → WW(∗) →+−/P T decay mode [80–82, 94]. The additional signa-tures of tag jets in the forward and of a low jet activityin the central regions of the detector allow for a signif-icant reduction of the background. This results in a bet-ter signal-to-background ratio compared to the inclusiveH → WW(∗) channel, which is dominated by the gluon-fusion process. As a consequence, the signal sensitivity isless affected by systematic uncertainties on the predictionsof the background. As an example, the reconstructed trans-verse mass distribution for a Higgs boson signal with a massof 160 GeV/c2 is shown on top of the backgrounds from t t ,Wt , and WW production in Fig. 10(left). This figure demon-strates the better signal-to-background ratio as compared tothe inclusive case.

Fig. 9 (Left) Pseudorapidity distribution of the tag jets in signal eventswith mH = 160 GeV/c2 and for t t background events. (Right) Sepa-ration η between the tag jets for the same types of events. The fullhistograms show the distributions at parton level; the dots represent

the distributions at reconstruction level, after the tagging algorithm hasbeen applied. The corresponding distributions for jets identified as tagjets in t t events are superimposed as dashed histograms. All distribu-tions are normalized to unity (from Ref. [93])

Eur. Phys. J. C (2009) 59: 463–495 473

Since neutrinos appear in the final state, the transversemass, defined as

MT =√(

ET + Eνν

T

)2 − ( �p T + /P T

)2,

is used for the mass reconstruction. For Higgs boson massesbelow ∼ 2mW , the W bosons are mostly at rest in the Higgsboson center-of-mass system, resulting in m = mνν , suchthat for both the dilepton and the neutrino system the trans-verse energy can be calculated as [83–87]

ET =

√(P

T

)2 + m2 and Eνν

T =√

(/P T )2 + m2.

After all cuts, a signal cross section in the eμ channelfor a Higgs boson with a mass of 160 GeV/c2 of the orderof 4.6 fb is expected above a total background expectationof 1.2 fb. Due to this large signal-to-background ratio, thischannel alone has a good discovery potential for a Higgs bo-son with a mass around 160 GeV/c2 and is not very sensitiveto systematic uncertainties on the background. This com-pares favorably with the gg → WW(∗) channel discussedin Sect. 4.1.3. It is interesting to note that an application oflooser kinematical cuts on the final-state leptons allows fora better discrimination between the signal and background

shape [93]. The corresponding transverse mass distributionis shown in Fig. 10(right). In this case, the background ex-tends to higher MT values and a background normalizationoutside the signal region is possible.

The presence of a signal can also be demonstrated in thedistribution of φ, the difference in azimuthal angle be-tween the two leptons in the final state. This distribution isshown in Fig. 11 for eμ final states passing the looser cuts,i.e., without cuts on the spatial separation of the leptons. De-pending on MT the event sample has been split into two sub-samples: (a) a so-called signal sample (MT < 175 GeV/c2)and (b) a control sample (MT > 175 GeV/c2). For eventsin the signal region (Fig. 11(left)), a pronounced structureat small φ is seen, as expected for a spin-0 resonance.This behavior is not present for events in the control sam-ple (Fig. 11(right)), where the t t and WW backgrounds areexpected to dominate. Therefore, the unbiased φ distribu-tion, resulting from relaxed kinematical cuts, can be used forboth a demonstration of the consistency of the signal with aspin-0 hypothesis and for an additional background normal-ization. This normalization can be performed in the high φ

region using events directly below the peak.In addition, it has been studied whether the larger

hadronic branching ratio of the W boson can be exploited

Fig. 10 (Left) Distribution ofthe transverse mass MT for aHiggs boson with a mass of160 GeV/c2 and thebackgrounds in the eμ-channelafter all cuts. (Right) The same,after relaxing kinematical cutson the reconstructed leptons.The accepted cross sectionsdσ/dMT (in fb/10 GeV/c2)including all efficiency andacceptance factors are shown inboth cases (from Ref. [93])

Fig. 11 Distributions of theazimuthal opening angle φ

between the two leptons forevents in the signal region(MT < 175 GeV/c2) (left) andevents outside the signal region(MT > 175 GeV/c2) (right).The accepted cross sectionsdσ/dφ (in fb/0.13 rad)including all efficiency andacceptance factors are shown inboth cases (from Ref. [93])

474 Eur. Phys. J. C (2009) 59: 463–495

and the process qq → qqH → qqWW(∗) → qq ν qq canbe detected. The cross section times branching ratio is about4.3 times larger than for the dilepton channel. However,since only one lepton is present in the final state, W+jetproduction is a serious additional background, with a crosssection more than two orders of magnitude larger than thesignal cross section. A detailed study has shown that after allcuts the signal rates and the signal-to-background ratio aremuch lower than the corresponding numbers in the dileptonchannel. Given these results, this channel cannot be consid-ered a Higgs boson discovery channel at the LHC.

4.2.2 qqH → qqττ

It has been shown that in the mass region 110 < mH <

140 GeV/c2 the ATLAS and CMS experiments are sensitiveto the ττ decay mode of the standard model Higgs bosonin the vector boson-fusion channel [80–82, 94]. Searchesfor H → ττ decays using the double leptonic decay modeqqH → qq ττ → qq νν̄ ν̄ν and the lepton–hadron decaymode qqH → qq ττ → qq νν had ν seem to be feasible.In these analyses, Z + jet production with Z → ττ consti-tutes the principal backgrounds. The ττ invariant mass canbe reconstructed using the collinear approximation. In sig-nal and background events, the H and Z bosons are emittedwith significant pT , which contributes to large tau boostsand causes the tau decay products to be nearly collinear inthe laboratory frame. Assuming that the tau directions aregiven by the directions of the visible tau decay products(leptons or hadrons from tau decays, respectively), the taumomenta and therefore the ττ invariant mass can be recon-structed. As an example, distributions of the reconstructedττ invariant mass for the eμ and for the lepton–hadron fi-nal states are shown in Fig. 12 for Higgs boson masses of120 and 135 GeV/c2. A discovery based on a combinationof these final states would require an integrated luminos-ity of about 30 fb−1. It should, however, be stressed thatthis assumes that the background from Z → ττ decays in

the signal region is known with a precision of ±10%. Morestudies are needed to establish that this precision can indeedbe achieved. The detection of the H → ττ decay mode isparticularly important for a measurement of the Higgs bo-son couplings to fermions and for Higgs boson searches inMSSM scenarios, as discussed in Sects. 5.3 and 6.4.

4.2.3 qqH → qqγ γ

In addition, the prospects for discovering a standard modelHiggs boson in the H → γ γ decay mode have been evalu-ated [81, 82, 94]. Due to the small H → γ γ decay branch-ing ratio the expected signal rate in the vector boson-fusionmode is small. As in the inclusive analysis, there are signif-icant backgrounds from non-resonant γ γ production withjets and from QCD multijet and direct photon production,e.g. γ γjj (QCD and electroweak), γjjj and jjjj . After ap-plying a similar selection as in the inclusive case and re-quiring tagging jets with a large separation in pseudorapid-ity, a signal significance of 2.2σ for mH = 130 GeV/c2 hasbeen found, assuming an integrated luminosity of 30 fb−1.Although the significance is lower than in the inclusive case,a much larger signal-to-background ratio with values around0.5 can be reached. On the other hand, this decay mode ap-pears to be considerably more sensitive to the backgroundresulting from QCD processes with jets misidentified asphotons.

4.2.4 qqH → qqbb

The detection of the fully hadronic decay mode qqH →qqbb is extremely challenging in a hadron collider environ-ment, given the huge background from QCD jet production.Already at the first trigger level the acceptance for signalevents is low, since it is foreseen the experiments will be runwith relatively high ET thresholds for pure jet triggers [94].Even if the trigger issue is ignored, a very low signal-to-background ratio is expected. First parton-level studies have

Fig. 12 (Left) Thereconstructed ττ invariant massfor a Higgs boson signal of120 GeV/c2 in the eμ channelabove all backgrounds afterapplication of all cuts except themass window cut around theHiggs boson mass. The numberof signal and background eventsare shown for an integratedluminosity of 30 fb−1 (fromRef. [93]). (Right) The same, forthe lepton–hadron channel andfor a Higgs boson withmH = 135 GeV/c2 (fromRefs. [81, 82])

Eur. Phys. J. C (2009) 59: 463–495 475

shown that the dominant bbjj background is about 300times larger than the signal. To claim a signal, the back-ground must be known with accuracies at the per mille level,which will be very hard to achieve.

4.3 Higgs boson searches using the associated WH , ZH

and t tH production

4.3.1 H → γ γ decays

The γ γ decay mode has also been studied for the asso-ciated production of a Higgs boson with a W or Z bo-son or a t t pair [80–82, 94]. In both production modes atleast one additional lepton is required from the vector bo-son or top-quark decay. Those channels have been found tohave a better signal-to-background ratio than the inclusiveH → γ γ channel, e.g., assuming an integrated luminosityof 100 fb−1, 13.2 signal events for mH = 120 GeV/c2 areexpected in the ATLAS experiment from the sum of WH ,ZH and t tH production above a total background of 5.7events [80]. The background is dominated by irreduciblecontributions from Wγγ , Zγγ and t tγ γ production, whichsum up to an irreducible fraction of 70–80%. The statisticalsignificance is found to be around 4.3σ for masses in therange between 100 and 120 GeV/c2. An observation in thischannel would therefore represent independent confirmationof the possible discovery of a light Higgs boson at the LHC.

4.3.2 H → bb decays

For low Higgs boson masses the discovery potential mightbe increased by searching for the H → bb̄ decay mode,which has the largest branching ratio in this mass range.Due to the huge backgrounds from QCD jet production inthis decay mode, only the associated production modes canbe used. It has already been demonstrated that the discoverypotential for a standard model Higgs boson in the WH pro-duction mode at the LHC is marginal [80]. Also the extrac-tion of a Higgs boson signal in the t tH , H → bb̄ channel

appears to be difficult [81, 82] and is extremely sensitive tothe precise knowledge of the large backgrounds. The strat-egy is to fully reconstruct the two top quarks which calls foran excellent b-tagging capability of the detectors. A criti-cal item is the knowledge of the shape of the main residualbackground from t tjj and t tbb production.

The CMS collaboration has recently updated their stud-ies using a more realistic simulation of the b-tagging per-formance of the detector [95]. In addition, explicit matrixelement calculations based on the ALPGEN Monte Carlo[96] have been used to estimate the above mentioned back-grounds. It has been found that the t tjj background dom-inates; in addition, there is a large t tbb background and acombinatorial background (ambiguities in the association ofthe tagged b jets to the top quarks and Higgs boson) fromthe signal itself. The problem is illustrated in Fig. 13(left),where the reconstructed invariant bb mass distribution forthe t tH → νb qqb bb signal with mH = 115 GeV/c2

and for background events is shown for the CMS exper-iment, assuming an integrated luminosity of 30 fb−1. Asmentioned above, the signal significance depends stronglyon the uncertainty on the absolute knowledge of the back-ground. As shown in Fig. 13(right), only for background un-certainties below 5%—which is difficult to reach—a signalsignificance exceeding 2σ can be reached [95]. The signal-to-background conditions have recently been confirmed infull detector simulation studies by the ATLAS collabora-tion [94].

4.3.3 H → WW(∗) decays

The signal significance in the H → WW(∗) → ν ν de-cay mode can still be enhanced in the mass region around160 GeV/c2 by searching for the associated productionmode WH with W → ν, leading to a three-lepton fi-nal state accompanied by missing transverse momentum[97, 98]. The third lepton allows for a significant suppres-sion of the background and therefore for a better signal-to-background ratio than in the gg → H → WW(∗) channel.

Fig. 13 (Left): Reconstructedbb invariant mass distribution ofa t tH → νb qqb bb signal(light) with mH = 115 GeV/c2

and the background (shaded) inthe CMS experiment for anintegrated luminosity of30 fb−1. (Right): The Higgsboson discovery significance asa function of the uncertainty onthe background for thecombination of all t t decaychannels (t t → νb νb,

νb qqb and qqb qqb) in theCMS experiment for anintegrated luminosity of 30 fb−1

476 Eur. Phys. J. C (2009) 59: 463–495

Fig. 14 ATLAS sensitivity for the discovery of a standard modelHiggs boson for an integrated luminosity of 30 fb−1 over the full massregion. The signal significances are plotted for individual channels, aswell as for the combination of channels. Systematic uncertainties onthe background have been included for the vector boson-fusion chan-nels (±10%) and for the H → WW(∗) → ν ν channel (±5%)

Furthermore, Higgs bosons decaying to WW can besearched for in the associated t t production mode. Experi-mentally this channel leads to striking signatures with multi-lepton and multijet final states. In a recent study, final stateswith two like-sign leptons or with three leptons have beeninvestigated [94]. In order to reject the large backgroundsfrom heavy flavor decays into leptons, strict isolation crite-ria have to be applied. The estimates of these backgroundssuffer from large uncertainties, and more studies are neededto obtain reliable estimates on the potential discovery poten-tial in this channel. It should be noted that this channel—if it can be measured—will provide important informationfor the determination of the Higgs boson couplings to topquarks.

4.4 Combined signal significance

The combined ATLAS Higgs boson discovery potential overthe full mass range, 100 < mH < 1000 GeV/c2, assumingan integrated luminosity of 30 fb−1 is shown in Fig. 14.The full mass range up to ∼1 TeV/c2 can be covered witha signal significance of more than 5σ with several discov-ery channels available at the same time. The vector boson-fusion channels play an important role for the discovery ofa Higgs boson at the LHC. It should be noted that in the AT-LAS evaluation no K factors have been included. Updated

Fig. 15 CMS sensitivity for the discovery of a standard model Higgsboson for an integrated luminosity of 30 fb−1 over the full mass region.The signal significances are plotted for individual channels, as well asfor the combination of the channels. Systematic uncertainties on thebackgrounds have been included and the results have been derived in-cluding next-to-leading order predictions for signals and backgrounds(from Refs. [81, 82])

results from the ATLAS Collaboration are expected to bepublished soon [94].

A similar performance has also been established for theCMS experiment. The corresponding discovery potential isshown in Fig. 15. In this updated plot the t tH channel hasno longer been included as a discovery channel.

The various discovery channels available at the LHC arecomplementary both as regards physics and detector as-pects. The gluon- and vector boson-fusion channels test dif-ferent production mechanisms. This complementarity alsoprovides sensitivity to non-standard Higgs models, such asfermiophobic models [99].

Also from the experimental point of view the Higgs bo-son discovery potential at the LHC is robust. The searchesare complementary in the sense that different detector com-ponents are important for different channels. The H → γ γ

decays require excellent electromagnetic calorimetry. In theidentification of vector boson-fusion the measurement ofjets, in particular the reconstruction of the forward tag jets,is essential.

5 Measurement of Higgs boson parameters at the LHC

After a possible observation of a Higgs boson at the LHCit will be important to establish its nature. Besides a precisemeasurement of its mass, which enters electroweak preci-sion tests, the determination of the spin and CP quantum

Eur. Phys. J. C (2009) 59: 463–495 477

numbers is important. To establish that the Higgs mecha-nism is at work, measurements of the couplings of the Higgsboson to fermions and bosons as well as a demonstration ofthe Higgs boson self-coupling are vital. The LHC potentialfor a measurement of these parameters is discussed in thefollowing.

5.1 Mass and width

A precise measurement of the Higgs boson mass can be ex-tracted from those channels where the invariant mass canbe reconstructed from electromagnetic calorimeter objects,as in the case of H → γ γ and H → ZZ(∗) → 4 decays[80–82].

The expected accuracy in the CMS experiment after col-lecting data corresponding to 30 fb−1 is better than 0.3% formasses smaller than about 300 GeV/c2, taking into accountstatistical errors only [81, 82]. With an integrated luminos-ity of 300 fb−1, each experiment would measure the Higgsboson mass with a precision of ∼0.1% over the mass range100–400 GeV/c2. The measurement error is determined bythe absolute knowledge of the lepton energy scale, which isassumed to be ±0.1%. The precision could be slightly im-proved in the mass range between ∼150 and 300 GeV/c2 ifa scale uncertainty of ±0.02% could be achieved. For largermasses, the precision deteriorates, because the Higgs bo-son width becomes large and the statistical error increases.However, even for masses around 700 GeV/c2 a precisionof about 1% can be reached.

The width of a standard model Higgs boson can be mea-sured directly only for masses larger than about 200 GeV/c2,where the intrinsic width of the resonance is comparable toor larger than the experimental mass resolution. This is themass region covered mainly by searches for H → ZZ → 4

decays. It has been estimated [80] that over the mass range300 < mH < 700 GeV/c2 the precision of the measurementof the Higgs boson width is approximately constant and isof the order of 6%.

In the low mass region, mH < 200 GeV/c2, the Higgsboson width can be constrained only indirectly by usingthe visible decay modes, as proposed and discussed inRefs. [100, 101].

5.2 Spin and CP eigenvalue

After finding a resonance, one of the first priorities must bea determination of the spin and the CP eigenvalues. Theycan be inferred from the angular distributions of the lep-tons in the H → ZZ → 4 decay mode [81, 82, 102–108].The CMS Collaboration has considered the case that the ob-served scalar boson φ is a mixture of a CP-even (H ) andCP-odd (A) boson according to φ = H + ξA. The precisionof the determination of ξ is shown in Fig. 16(left). Using thesame observables the ATLAS Collaboration has found thatthe hypothesis of non-standard model CP and spin combi-nations can be distinguished from the standard model valuesfor masses above 250 GeV/c2 and integrated luminositiesabove 100 fb−1 [108].

The coupling of a scalar boson to vector bosons can beparametrized in a model independent way via three terms:a CP-even standard model-like contribution, which only ex-ists if the scalar field has a non-vanishing vacuum expec-tation value, and CP-even and CP-odd anomalous terms.The contributions and admixtures can be determined in thevector boson-fusion process using the azimuthal separationφjj of the tagging jets [109]. The expected spectra areshown in Fig. 16(right) for the different coupling hypothe-ses, assuming a Higgs boson with a mass of 160 GeV/c2

in the WW decay mode. For an integrated luminosity of10 fb−1 anomalous purely CP-even or CP-odd effectivecouplings can be excluded with a significance of approxi-mately 5σ . For a Higgs boson with a mass of 120 GeV/c2 asensitivity of 2σ can be reached for an integrated luminostyof 30 fb−1 using the H → ττ decay mode [110].

Fig. 16 (Left) Expectedprecision for the determinationof ξ (see text) for a boson φ

with a mass of 300 GeV/c2 inthe CMS experiment assumingan integrated luminosity of60 fb−1. (Right) Expecteddistribution of φjj in vectorboson-fusion events withH → WW → νν

(mH = 160 GeV/c2) in theATLAS experiment includingbackground for an integratedluminosity of 10 fb−1

478 Eur. Phys. J. C (2009) 59: 463–495

5.3 Couplings to bosons and fermions

For a Higgs boson with mH < 2 mZ , the experiments atthe LHC will be able to observe the dominant decays intobosons and heavy fermions over mass ranges for which thebranching ratios in question are not too small. Decays tolight fermions, i.e. to electrons, muons or to quarks lighterthan the b-quark, will, however, not be observable either dueto the small decay rate or due to the overwhelming back-ground from QCD jet production.

In that mass range, the total width is expected to be smallenough so that the narrow width approximation can be usedto extract the couplings. The rates σ · BR(H → f f̄ ) mea-sured for final states f f̄ are to a good approximation pro-portional to Γi · Γf /Γ , where Γi and Γf are the Higgs bo-son partial widths involving the couplings at production anddecay, respectively, and Γ is the total width of the Higgsboson.

The strength of the LHC in the coupling measurementsis based on the simultaneous information which, for a givenHiggs boson mass, is available in the various production anddecay modes. Recently, a study has been performed in whichthe full information of all accessible production and decaychannels, as listed in Table 1, is used to fit the coupling para-meters [101]. In this study, the correlations among the var-ious channels as well as experimental and theoretical sys-tematic uncertainties are taken into account.

Assuming that the measured values correspond to thestandard model expectations, a likelihood function is formedwhich, for a given integrated luminosity, is based on theexpected Poisson distribution of the event numbers and on

Table 1 List of all studies used in the global likelihood fit performedin Ref. [101]; each channel has been used in the mass range indicated

Production mode Decay mode Mass range

(GeV/c2)

Gluon fusion H → ZZ(∗) → 110–200

H → WW(∗) → ν ν 110–200

H → γ γ 110–150

Vector boson H → ZZ(∗) → 110–200

fusion H → WW(∗) → ν ν 110–190

H → ττ → νν νν 110–150

H → ττ → νν had ν 110–150

H → γ γ 110–150

t t H production H → WW(∗) → ν ν (ν) 120–200

H → bb 110–140

H → γ γ 110–120

WH production H → WW(∗) → ν ν (ν) 150–190

H → γ γ 110–120

ZH production H → γ γ 110–120

the estimated systematic errors. These errors include a 5%uncertainty on the integrated luminosity, uncertainties onthe reconstruction and identification efficiencies of leptons(2%), photons (2%), b-quarks (3%) and on the forward jettagging and jet veto efficiency (5%). In addition, theoreti-cal uncertainties on Higgs boson production (20% for ggH ,15% for t tH , 7% for WH and ZH and 4% for vector boson-fusion) and on branching fractions (1%) have been takeninto account.

Under the assumption that only one scalar CP-even Higgsboson exists, relative branching ratios, which are identical toratios of partial decay widths, can be measured. The decayH → WW is used as normalization, since it can be mea-sured over the full intermediate mass range with a relativelysmall error. In Fig. 17(left) the expected relative errors onthe measurement of ratios of Higgs boson branching ratiosare shown, assuming an integrated luminosity of 300 fb−1.

In particular, the ratios ΓZ/ΓW can be measured with anaccuracy of the order of 10–20% for Higgs boson massesabove 130 GeV/c2. The ratios Γτ/ΓW and Γb/ΓW are lessconstrained, with errors expected to be of the order of 30to 60%. It should be noted that these results are based oninformation from the decay only and they do not exploit anyinformation from the production.

If additional theoretical assumptions are made, informa-tion from the production can be used as well, and in par-ticular a measurement of the top-Yukawa coupling becomespossible via the strong dependence of the gluon-fusion andt tH production cross sections on this coupling. Assumingthat only the known particles of the standard model cou-ple to the Higgs boson and that no couplings to light fermi-ons are extremely enhanced, all accessible Higgs boson pro-duction and decay modes at the LHC can be expressed interms of the Higgs boson couplings gW ,gZ,gt , gb and gτ .The production cross sections depend on the square of thesecouplings. The exact dependence has to be calculated the-oretically and put into the fit with corresponding system-atic uncertainties. The gluon-fusion production cross sec-tion, for example, is not strictly proportional to the top-Yukawa coupling squared but receives additional contribu-tions from the interference with diagrams containing a b-loop. Using the program of Ref. [111], they have been foundto be at the level of 7% for mH = 110 GeV/c2 and 4%for mH = 190 GeV/c2. These additional contributions areignored and it is assumed that the b-coupling is not en-hanced by a factor of 10 or more compared to the standardmodel. Furthermore, the interference effect between W - andZ-fusion in the vector boson-fusion is expected to be small(<1%) [30] and is ignored.

Similarly, the Higgs boson branching ratios are pro-portional to g2/Γ , and again, the proportionality factorsare taken from theory, assuming a relative uncertainty ofabout 1%. The decay H → γ γ proceeds either via a W or

Eur. Phys. J. C (2009) 59: 463–495 479

Fig. 17 The expected relative errors for the measurement of relativebranching ratios (left) and relative couplings (right), normalized tothose of the H → WW decay and assuming an integrated luminosity

of 300 fb−1 for one experiment. The dashed lines give the expectedrelative error without systematic uncertainties (from Ref. [101])

a top-quark loop, with destructive interference between thetwo. The relative contributions are taken from Ref. [112].

In Fig. 17(right) the relative errors on the measurementof relative couplings are shown for an integrated luminosityof 300 fb−1 for one experiment. Due to the large contribu-tions of the gluon-fusion and t tH production modes, the ra-tio of the top-Yukawa coupling to the Higgs boson couplingto W bosons can be well constrained with an estimated un-certainty of the order of 10 to 20%.

Recently, the determination of Higgs boson couplings atthe LHC has been discussed for general multi-Higgs dou-blet models [113]. Imposing the additional constraint thatthe HWW and HZZ couplings are bound from aboveby their standard model values (g2

W/Z(SM)), i.e. g2W/Z <

1.05 ·g2W/Z(SM), an absolute measurement of the Higgs bo-

son couplings to the vector bosons and the τ lepton, andb and t quarks becomes possible. The constraints imposedare theoretically motivated and are valid in particular for theMSSM. Any model that contains only Higgs doublets and/orsinglets satisfies the relation g2

W/Z < g2W/Z(SM). The extra

5% margin allows for theoretical uncertainties in the transla-tion between couplings-squared and partial widths and alsofor small admixtures of exotic Higgs states, as for exam-ple SU(2) triplets [113]. Contributions of additional parti-

cles running in the loops for H → γ γ and gg → H areallowed for and their contribution to the partial width is fit-ted. Assuming an integrated luminosity of 300 fb−1 (for allchannels, except vector boson fusion, for which 100 fb−1

has been assumed) and a combination of two experiments,typical accuracies for the absolute measurement of the cou-plings of about 20 to 30% can be achieved for Higgs bosonmasses below 160 GeV/c2. For masses above the W -pairthreshold the measurement of the W and Z partial widthcan be performed with an accuracy at the level of ±10%.

5.4 Higgs boson self-coupling

To fully establish the Higgs mechanism, it must be demon-strated that the shape of the Higgs potential VH has the formrequired for electroweak symmetry breaking. The potentialcan be expressed in terms of the physical Higgs field H by

VH = m2H

2 H 2 + m2H

2vH 3 + m2

H

8v2 H 4, where v = (√

2GF )−1/2 =246 GeV/c2 is the Higgs vacuum expectation value. The co-efficients of the second and third term are proportional tothe strength of the Higgs trilinear (λHHH = 3m2

H /v) andquartic (λ′

HHHH = 3m2H /v2) Higgs boson self-coupling,

respectively. In order to extract information on these cou-plings, multiple Higgs boson production must be measured

480 Eur. Phys. J. C (2009) 59: 463–495

[114–119]. Since the quartic coupling is about two orders ofmagnitude smaller than the trilinear coupling, present stud-ies have focused on the determination of λHHH .

The coupling strength λHHH enters the production rateof Higgs boson pairs. At LHC energies, the inclusiveHiggs boson pair production is dominated by gluon fu-sion [120, 121]. Other processes have cross sections whichare factors of 10 to 30 smaller. For mH > 140 GeV/c2,H → WW is the dominant decay mode and WW WW finalstates are produced with the largest branching ratio. In or-der to suppress the large backgrounds from t t and multiplegauge boson production, like-sign lepton final states offerthe best sensitivity [122–125]. Studies have shown that the(±νjj) (±νjj) final state has the highest sensitivity for ex-tracting information on the self-coupling parameter λHHH .

The studies conclude that with data corresponding to theultimate luminosity expected at the LHC of 300 fb−1 per ex-periment, a determination of the Higgs boson self-couplingwill not be possible. A measurement with reasonable errorswill require a luminosity upgrade, i.e. the realization of theso-called Super LHC (SLHC).

In the analysis of Ref. [123, 124] it is concluded that theLHC may rule out the case of a non-vanishing λHHH us-ing the four W final state for a mass range of 150 < mH <

200 GeV/c2 with a confidence level of 95%. For the SLHC,assuming an integrated luminosity of 3000 fb−1 per experi-ment, it is claimed that a measurement of λHHH with a pre-cision of 20% will be possible. In this study NLO K factorshave been used.

These conclusions still need to be confirmed in a fullsimulation of the detector performance. First preliminarystudies [125] confirm the sensitivity at the SLHC; however,some background contributions might have been underesti-mated. Further studies to clarify these issues are currently inprogress.

The mass region mH < 140 GeV/c2 is considered to bevery challenging for a measurement of the Higgs boson self-coupling, even at the SLHC [123, 124]. Recently, it has beenproposed to use the rare decay mode HH → bb γ γ to in-vestigate the self-coupling in that mass region [126]. At theLHC, assuming an integrated luminosity of 300 fb−1 perexperiment and including NLO K factors, only six signalevents are expected for the most optimistic case of mH =120 GeV/c2. Using this decay mode would require a lumi-nosity upgrade to rule out λHHH = 0 at the 90% confidencelevel [126]. Also, in this case, a detailed simulation of thedetector performance has not yet been completed.

6 Search for MSSM Higgs bosons at the LHC

In addition to the excellent prospects for standard modelHiggs searches, the LHC experiments have a large potential

in the investigation of the MSSM Higgs sector. The experi-mental searches carried out at LEP and presently continuedat the Tevatron can be extended to much larger Higgs bo-son masses. In the search for the light, standard model-likeHiggs boson h the same channels as in the search for thestandard model Higgs boson will be used. Heavier Higgsbosons will be searched for in additional decay channels,which become accessible in certain regions of the MSSMparameter space due to enhanced couplings, e.g. the de-cay mode H/A → ττ at large tanβ . Decays into τ lep-tons also contribute to the search for charged Higgs bosons,which at the LHC can be extended to masses beyond thetop-quark mass. A comprehensive and complete study ofthe LHC discovery potential in the MSSM has first beenpresented in Ref. [127]. In that study, the discovery po-tential had been determined for two benchmark scenarioswith different assumptions on mixing in the stop sector, asdefined at LEP [128]. In a so-called no-mixing scenariothe trilinear coupling in the stop sector At is set to zero,whereas in the so-called maximal mixing scenario the valueAt = √

6 mSUSY, with mSUSY = 1 TeV/c2, is used. In thefollowing, the search strategies for MSSM Higgs bosonsat the LHC are briefly discussed and the discovery poten-tial is presented for different MSSM benchmark scenarios.Throughout this chapter, all discovery contours for MSSMHiggs bosons in the (mA, tanβ) parameter space are givenfor a signal significance at the 5σ level. In most analysesSUSY particles are considered to be heavy enough to playa negligible role in the phenomenology of Higgs boson de-cays. This assumption is abandoned in Sect. 6.6, where theinterplay of the Higgs sector and SUSY particles is dis-cussed.

6.1 Search for the light CP-even Higgs boson h

The channels discussed in Sect. 4 can be used to searchfor the lightest MSSM CP-even Higgs boson h. Resultsfor the discovery potential obtained in studies carried outby the CMS Collaboration are shown in the (mA, tanβ)

plane in Fig. 18(left). In these studies, maximal mixing,mSUSY = 1 TeV/c2, and integrated luminosities of 30 and60 fb−1 have been assumed [81, 82]. In addition to the in-clusive h → γ γ channel the vector boson-fusion channelqqh → qqττ contributes significantly to the discovery po-tential. For an integrated luminosity of 60 fb−1 nearly thewhole parameter space can be covered with a significanceof more than 5σ by a single experiment.

In the maximal mixing scenario, the neutral Higgs bosonsΦ = h,H and A are nearly degenerate in mass in the uncov-ered region up to mA ∼125 GeV/c2. At large tanβ the en-hanced coupling to b quarks can be used to search for theseHiggs bosons in the pp → bbΦ → bb μμ channel. It hasbeen shown that signals can be extracted with a significance

Eur. Phys. J. C (2009) 59: 463–495 481

Fig. 18 (Left): The 5σ discovery potential in the CMS experiment forthe light CP-even MSSM Higgs boson h from the inclusive h → γ γ

channel and for the light and heavy scalar bosons, h and H , in theqqh(H) → qqττ → qq − had channel in the (mA, tanβ) plane as-suming maximal mixing. Integrated luminosities of 30 and 60 fb−1 are

assumed (from Refs. [81, 82]). (Right): Discovery contour plot for theneutral Higgs bosons, Φ = h,H,A, in the process pp → bbΦ , withΦ → μμ. The significance inside the grey area exceeds 5σ with anintegrated luminosity of 30 fb−1 (from Refs. [81, 82])

above 5σ in the mass region around mA ∼ 125 GeV/c2 andtanβ > 25 [81, 82, 94]. The 5σ discovery contour in the(mA, tanβ) plane is shown in Fig. 18(right) for the CMSexperiment. The signal extraction is challenging due to theoverwhelming background from Z → μμ production. Itwill therefore be essential to have a good b-tagging perfor-mance and a good dimuon mass resolution as well as to con-trol the background from data, using Z → control sam-ples. It should be noted that this particular part of the para-meter space can also be covered with a significance exceed-ing 5σ by searches for charged Higgs bosons (see Sects. 6.3and 6.4).

6.2 Search for the heavy MSSM Higgs bosons H and A

6.2.1 The large tanβ region

At large tanβ , where the couplings of the heavy Higgsbosons H and A to down-type fermions are enhanced, theassociated Higgs boson production with a bb pair is thedominant production mode. The decay into τ pairs has asignificant branching ratio over a large region of parameterspace, so that the process bb H/A → bb ττ plays a keyrole. This channel can be complemented by the H/A → μμ

decay mode, for which the much smaller branching ratio iscompensated by the better signal-to-background ratio. In ad-dition, a search in the bbbb final state has been proposed inthe literature [129, 130], but it suffers from sizable back-ground from QCD jet production and therefore has only alimited discovery potential [80–82].

For the bb ττ final states both excellent b-tagging andtau identification performance are essential to suppress thelarge backgrounds from γ ∗/Z, QCD multijet and W+jet

production. Detailed studies have shown that both leptonicand hadronic tau decays (τhad) can be used to isolate sig-nals from heavy Higgs boson production [80–82]. For mA <

400 GeV/c2 the final state with one leptonic and onehadronic tau decay (−τhad) has the largest discovery reach.At large mA the double hadronic decay mode contributessignificantly, while for small Higgs boson masses its reach islimited by the low trigger efficiency. In all decay modes theττ invariant mass can be reconstructed using the collinearapproximation, as discussed in Sect. 4.2. An example of areconstructed H/A signal with mH/A = 300 GeV/c2 in the − τhad decay mode is shown in Fig. 19(left), based on asimulation of the CMS detector performance and assumingan integrated luminosity of 30 fb−1. The discovery contoursfor the various ττ decay modes are shown in the (mA, tanβ)

plane in Fig. 19(right) for the CMS experiment.As mentioned above, also the μμ decay mode contributes

to the discovery potential. Given the good mass resolutionfor dimuon final states, this channel can eventually be usedto separate nearby Higgs boson resonances using a precisemeasurement of the shape of the mass distribution. For smallmA it covers regions of parameter space not yet excludedby the LEP experiments (see the discussion at the end ofSect. 6.1).

6.2.2 The small tanβ region

In the region of small tanβ heavy Higgs bosons can besearched for via their decays into the lightest Higgs boson.This allows for a simultaneous observation of two Higgsbosons, for example via the decay modes H → hh andA → Zh.

The final states of interest for a search in the H → hh

decay mode are bb γ γ , bb ττ and bb bb. The decays

482 Eur. Phys. J. C (2009) 59: 463–495

Fig. 19 (Left): The reconstructed ττ invariant mass distribution inthe final state with one lepton and one hadronic tau ( − τhad) fromthe process gg → bb H/A → bb ττ . The H/A signal (dark) formA = 300 GeV/c2 and tanβ = 25 is shown on top of the backgroundfor an integrated luminosity of 30 fb−1. (Right): The 5σ discoverypotential for the neutral MSSM Higgs bosons Φ = h,H and A in

the (mA, tanβ) plane assuming the maximal mixing scenario. Thebb associated Higgs production is considered with the Φ → μμ andΦ → ττ decay modes. For the ττ final states the three different taudecay modes are considered for integrated luminosities of 30 fb−1

( − and − τhad) and 60 fb−1 (τhad − τhad) (from Refs. [81, 82])

into bb γ γ can reliably be triggered and offer good kine-matic constraints and mass resolution for the reconstructionof mH . However, due to the small branching ratio into γ γ ,the discovery potential in that channel is limited by the smallsignal rate. For an integrated luminosity of 300 fb−1, an ob-servation is possible in the parameter region tanβ < 4 and2mh < mH < 2mt [80].

Searches in both the bb ττ and bb bb final states havebeen found to be less promising [80–82]. For the bb ττ case,the signal extraction is difficult due to large backgroundsfrom W+jets and t t production and due to the poor massresolution for the signal. A detection of the H → hh decayin the bb bb mode requires a four-jet trigger with as lowa pT threshold as possible and excellent b-tagging perfor-mance to control the overwhelming background from four-jet events.

The A → Zh decay mode has so far been studied in thefinal states Zh → bb and Zh → bb bb [80–82]. A searchin the latter decay mode is as challenging as the H → hh →bb bb channel discussed above. The decays into bb caneasily be triggered and offer the largest rates apart from thedominant bb bb decays. Given that the signal rate is rapidlyfalling with increasing tanβ , the A → Zh → bb channelcan only be observed at small tanβ and for 200 GeV/c2 <

mA < 2mt .For mA > 2mt and tanβ ∼ 1, the H → t t and A → t t

branching ratios are close to 100%. Since the H and A

bosons are almost degenerate in mass in the relevant regionof parameter space, the two decays cannot be distinguishedexperimentally. As discussed in the literature [131, 132],a signal from H/A → t t would appear as a peak in the t t in-variant mass spectrum above the t t continuum background.The interference of the signal and background amplitudes

leads to a strong suppression of the signal at high masses.For a Higgs boson mass of 370 (450) GeV/c2, the signal-to-background ratio is expected to be of the order of 9% (1%)only. Under the optimistic assumption that the backgroundt t mass spectrum will be known to better than 1% from ex-perimental data and theory, a signal extraction would be pos-sible in a limited region of parameter space for 2mt < mA <

450 GeV/c2.

6.3 Charged Higgs bosons

At the LHC, a charged Higgs boson can be detected in sev-eral scenarios. For mH± < mt − mb , top-quark productionrepresents a copious source via the decay t → H±b. If thecharged Higgs boson is heavier than the top quark, it canbe produced via the process gg → H±tb (or gb → H±t)[133–138]. Additional contributions come from the Drell–Yan type process gg,qq → H+H− and from the associatedproduction with a W boson, qq → H±W∓ [139–148]. As-suming a heavy SUSY mass spectrum, the charged Higgsboson decays into standard model particles only. For smalltanβ and mH± < mt , the main decay channels are H± →τν, cs,Wh and t∗b. For mass values above the top-quarkmass, the H± → tb decay mode becomes dominant, in par-ticular at small tanβ . At larger tanβ , also the H± → τν

decay has a significant branching ratio, which, however, de-creases with increasing Higgs boson mass (see Fig. 4 fordetails).

Due to the large t t production cross section, chargedHiggs bosons with mH± < mt − mb can be detected nearlyup to the kinematic limit. The LHC discovery potentialhas been studied for both the dominant H± → τν andthe H± → qq ′ channel, via the decay chain pp → t t →

Eur. Phys. J. C (2009) 59: 463–495 483

Fig. 20 The 5σ discovery contours (left) and the 95% C.L. exclusion contours (right) for the combination of all charged Higgs search channelsin the (mH+ , tanβ) plane in the ATLAS experiment for different assumptions on the integrated luminosity (from Ref. [94])

H±b W∓b̄ [80–82]. In the region 5 < tanβ < 10, thehadronic decay mode contributes significantly and increasesthe discovery potential.

Following the studies presented in Refs. [149–155], bothLHC Collaborations have recently evaluated their discov-ery potential for heavier charged Higgs bosons, i.e. mH± >

mt − mb [81, 82, 94, 156]. For the process gb → H±t →tb t → νbb qq̄b, the large backgrounds can be sufficientlysuppressed if three b-tags are required and both top quarksare reconstructed. In addition, it has been studied whetherthe discovery range in the H± → tb decay mode can be ex-tended if four b-jets are required in the final state (via theprocess gg → H±tb). Although this requirement reducesthe background, the reconstruction of the charged Higgs bo-son is rendered more difficult due to the combinatorics inthe reconstruction of the multijet final state.

More favorable signal-to-background conditions arefound in the H± → τν decay mode for the productionprocess gb → H±t → τν qqb, with τ → had ν. Both theassociated top quark and the tau lepton are required to de-cay hadronically. This has the advantage that the transversemass of the τhad −/P T system can be used to discriminate be-tween signal and background. The signal distribution is ex-pected to show a Jacobian peak structure at the Higgs bosonmass. The channel must be triggered by a hadronic tau+/P T

or hadronic tau + multijet trigger, both of which are fore-seen in the two LHC experiments. The backgrounds fromW + jet, t t,Wt and QCD multijet production can be sup-pressed by requiring in addition to an identified hadronic tauexactly three high-pT jets, one of which must be b tagged.The signal can be enhanced by exploiting the τ polariza-tion in one prong tau decays [154, 155], which leads to aharder spectrum of single pions when the tau originates from

an H± rather than from W decay. After a transverse masscut, the backgrounds are small and a charged Higgs bosoncan be discovered with a significance of more than 5σ formH± > mt in the large tanβ region.

The 5σ discovery contours as well as the 95% C.L. ex-clusion contours for the combination of all channels, as de-termined in recent studies by the ATLAS Collaboration, areshown in the (mH+ , tanβ) plane in Fig. 20.

Within the MSSM, the sensitivity to the H± → W(∗)h

decay mode is weaker [156]. For Higgs boson masses be-low the top-quark mass, the extraction of the signal is onlypossible in the small tanβ region (1.5 < tanβ < 2.5). ForHiggs boson masses above the top-quark mass, the discov-ery potential in this channel is marginal. However, in ex-tended models, as for example in the NMSSM [157–164],where the Higgs sector of the MSSM is extended by a com-plex singlet scalar field, the H± → W(∗)h channel shows aviable signal [156].

6.4 The MSSM discovery potential in various benchmarkscenarios

Different benchmark scenarios have been proposed for theinterpretation of MSSM Higgs boson searches [165]. In theMSSM, the masses and couplings of the Higgs bosons de-pend, in addition to tanβ and mA, on the SUSY parame-ters through radiative corrections. In a constrained model,where unification of the SU(2) and U(1) gaugino masses isassumed, the most relevant parameters are At , the trilinearcoupling in the stop sector, the Higgs mass parameter μ, thegaugino mass term M2, the gluino mass mg and a commonscalar mass MSUSY. Instead of the parameter At , the stopmixing parameter Xt := At − μ cotβ can be used. In par-

484 Eur. Phys. J. C (2009) 59: 463–495

ticular, the phenomenology of the light Higgs bosons h de-pends on the SUSY scenario, for which the following havebeen considered in a recent study [166, 167].

1. mh-max scenario: the SUSY parameters are chosen suchthat for each point in the (mA, tanβ) parameter space aHiggs boson mass close to the maximum possible valueis obtained. For fixed M2, μ, mSUSY and mg this isachieved by adjusting the value of Xt . This scenario issimilar to the maximal mixing scenario discussed above.

2. No mixing scenario: in this scenario vanishing mixing inthe stop sector is assumed, i.e. Xt = 0. This scenario typ-ically gives a small mass for the lightest CP-even Higgsboson h and is less favorable for the LHC.

3. Gluophobic scenario: the effective coupling of the lightHiggs boson h to gluons is strongly suppressed for a largearea of the (mA, tanβ) plane. This requires large mix-ing in the stop sector, leading to cancellations betweentop-quark and stop loops such that the production crosssection for gluon fusion is strongly suppressed.

4. Small α scenario: the parameters are chosen such thatthe effective mixing angle α between the CP-even Higgsbosons is small. This results in a reduced branching ratiointo bb and ττ for large tanβ and intermediate valuesof mA.

The parameters for the four scenarios are summarized inTable 2 [166, 167]. For a given point in parameter space,couplings and branching ratios of the Higgs bosons havebeen calculated using the program described in Refs. [168–170], which is based on a two-loop diagrammatic approachin an on-shell renormalization scheme [24–26]. In this cal-culation, the full one-loop radiative corrections and all dom-inant two-loop corrections are included.

In the evaluation of the discovery potential for the lightHiggs boson h all production modes, i.e. the gluon fu-sion, the vector boson fusion as well as the associated bbh,t th and Wh production have been used. The decay modesh → γ γ , qqh → qqττ , qqh → qqWW , bbh → bb μμ,t th with h → bb and Wh → νbb have been considered.The discovery potential for the light CP-even Higgs bosonh for the individual channels in the four benchmark scenar-ios is shown in Fig. 21 for the ATLAS experiment alone,assuming an integrated luminosity of 30 fb−1.

The study shows that already at low luminosity the full(mA, tanβ) plane can be covered in all benchmark scenar-ios considered, apart from a region at small mA. However,in that particular area the searches for heavier Higgs bosonshave a sensitivity, such that at least one MSSM Higgs bo-son would be discovered at the LHC already with a mod-erate integrated luminosity of 30 fb−1 (see Fig. 22 and thediscussion below). In the area not yet excluded by the LEPdata, the light CP-even Higgs boson is almost guaranteedto be discovered in all four scenarios via the vector boson-fusion channels. For an integrated luminosity of 30 fb−1 thediscovery potential in the large mA region is dominated bythe vector boson-fusion channel with h → ττ . In the mh-max scenario, the qqh → qqWW channel also contributes.For small mA and large tanβ , the bbh associated produc-tion with h → μμ dominates. Since many complementarychannels are available at the LHC, the loss in sensitivity dueto suppressed couplings in certain benchmark scenarios canbe compensated for by other channels. In the small α sce-nario, for example, the effect of the suppressed branchingratio into τ leptons (visible in the region tanβ > 20 and200 < mA < 300 GeV/c2) is nicely compensated for by theh → WW contribution. For large integrated luminosities,the h → γ γ and h → ZZ∗ → 4 channels provide addi-tional sensitivity.

For the heavier Higgs bosons, the production cross sec-tions and decay branching ratios are similar for the var-ious benchmark scenarios, in particular for large valuesof mA. Therefore, a similar discovery potential as presentedin Sects. 6.2 and 6.3 is expected for all scenarios considered.

The overall discovery potential for the four scenarios ispresented in Figs. 22 and 23 for integrated luminosities of30 and 300 fb−1, respectively. Already with a modest inte-grated luminosity of 30 fb−1, the full parameter space can becovered for all benchmarks. However, in the region of mod-erate tanβ and large mA only one MSSM Higgs boson, thestandard model-like Higgs boson h, can be discovered, evenif the large integrated luminosity of 300 fb−1 is assumed.The exact location of that region in the parameter plane de-pends on the details of the model considered. A further in-crease in integrated luminosity will only marginally reducethat region [122]. As discussed in Sect. 6.6, some sensitiv-ity to heavier Higgs bosons might, however, be provided viatheir decays into SUSY particles.

Table 2 Values of the SUSYparameters as used inRefs. [166, 167] for the fourbenchmark scenarios

mSUSY (GeV/c2) μ (GeV/c2) M2 (GeV/c2) Xt (GeV/c2) mg (GeV/c2)

mh-max 1000 200 200 2000 800

No mixing 1000 200 200 0 800

Gluophobic 350 300 300 −750 500

Small α 800 2000 500 −1100 500

Eur. Phys. J. C (2009) 59: 463–495 485

Fig. 21 The 5σ discovery contours for the light CP-even Higgs bo-son h in the (mA, tanβ) plane after collecting 30 fb−1 in the ATLASexperiment for the four benchmark scenarios considered: the mh-max

scenario, the no mixing scenario, the gluophobic and small α scenario(see text for details). The cross-hatched yellow region is excluded bysearches at LEP (from Refs. [166, 167])

Finally, it should be noted that Higgs boson decays into

standard model particles can be strongly suppressed in cer-

tain MSSM scenarios. An example has been presented in

Ref. [171], where the light Higgs boson decays predom-

inantly into light bottom squarks leading to multijet final

states. In such scenarios the detection of a light Higgs bo-

son might be difficult at hadron colliders.

6.5 CP violating scenarios

All studies presented so far have been performed assumingCP conservation in the Higgs sector. Although CP conser-vation is present at Born level, CP violating effects mightbe introduced via complex mixing parameters At,Ab ormass terms. As a consequence the mass eigenstates H1,H2

and H3 are mixtures of the CP eigenstates h,H and A.

486 Eur. Phys. J. C (2009) 59: 463–495

Fig. 22 The overall 5σ discovery potential for MSSM Higgs bosons atthe LHC for an integrated luminosity of 30 fb−1 in the four benchmarkscenarios (see text). In the shaded area (cyan) only the light CP-evenHiggs boson h can be observed. In the blue left-hatched area the heavy

neutral Higgs bosons H and/or A, and in the red right-hatched area thecharged Higgs bosons H± can be detected. The cross-hatched yellowregion is excluded by searches at LEP (from Refs. [166, 167])

A special scenario considered is the so-called CPX scenario[172], which is characterized by large phases: arg(At ) =arg(Ab) = arg(Mgluino) = 90◦. Due to the decoupling of theZ resonance in this scenario no lower mass limits from LEPapply for the mass of the lightest Higgs boson H1 and un-covered regions in parameter space exist for low H1 masses[173]. The exact positions of these holes depend also on the

precise value of the top-quark mass and on the theoreticalmodel considered [173].

A preliminary study of the discovery potential at the LHChas indicated that in the CPX scenario a large fraction of theparameter space can be covered using the standard searchchannels. However, even at the LHC, uncovered regions atmedium tanβ ∼ 5 and small mH+ around 150 GeV/c2—

Eur. Phys. J. C (2009) 59: 463–495 487

Fig. 23 The same as Fig. 22, but for an integrated luminosity of 300 fb−1 (except for the A/H → ττ decay mode, for which an integratedluminosity of 30 fb−1 has been assumed) (from Refs. [166, 167])

which correspond to small H1 masses—remain. The situa-

tion is illustrated in Fig. 24. It needs to be studied whether

they can be covered by exploiting additional search chan-

nels via t t production, e.g. t t → Wb H+b → νb WH1b

with H1 → bb.

A similar situation may occur in the context of the next-

to-minimal-supersymmetric standard model (NMSSM)

[157–164]. It needs to be studied whether these scenarios

can be covered at the LHC by exploiting additional searchchannels.

6.6 The interplay between the Higgs sector and SUSYparticles

SUSY particles have an impact on the Higgs boson dis-covery potential via their appearance in the decay chain(mostly for H and A) and in loops (mostly for production

488 Eur. Phys. J. C (2009) 59: 463–495

Fig. 24 Overall ATLAS discovery potential for Higgs bosons in the CPX scenario for data corresponding to an integrated luminosity of 300 fb−1.The cross-hatched area has been excluded by the LEP experiments

via gluon fusion and for H → γ γ decays) [174]. In partic-ular, the heavy Higgs bosons H , A and H± could be de-tected via their decays into neutralinos and charginos, usingmultilepton final states. Due to present experimental con-straints, the decay of the light Higgs boson h into the lightestSUSY particles is kinematically forbidden over a large frac-tion of the constrained MSSM parameter space. Instead, thisHiggs boson could appear at the end of the decay cascadeof SUSY particles, for example in the decay χ0

2 → χ01 h,

where χ01 and χ0

2 are the lightest and second lightest neu-tralinos.

6.6.1 Search for the heavy MSSM Higgs bosons in SUSYdecay modes

As discussed in Sect. 6.4, the heavy MSSM Higgs bosonscannot be detected via their decays into standard modelparticles in the parameter range of intermediate tanβ andlarge mA. It has been shown that decays into charginos andneutralinos can be used in some areas of SUSY parameterspace for the detection of MSSM Higgs bosons [81, 82]. Ifthe masses of these particles are small enough, the branch-ing ratios for the decays H/A → χ0

2 χ02 and H± → χ0

2,3χ±

are sizable. If in addition sleptons are light, these decayscan be searched for in multilepton final states via the decaychain χ0

2 → ̃ → χ01 . Due to the invisible neutralinos,

the Higgs boson mass reconstruction is only possible if the

masses of χ01 and ̃ are known, for example from the analy-

sis of first or second generation squark cascade decays [175].In the search for the decay channel H/A → χ0

2 χ02 →

χ01 χ0

1 , lepton isolation, a large missing transverse en-ergy and a jet veto are used to suppress the standard modelbackgrounds from ZZ, Zbb, Zcc and t t production as wellas the background from SUSY (q̃, g̃) production [81, 82].The discovery reach depends strongly on the choice of theSUSY parameters. Examples for the 5σ discovery reach areshown in Fig. 25(left).

A similar analysis has been performed within the moreconstrained minimal supergravity (mSUGRA) [176–180]scenario. This model is determined by five parameters: com-mon masses m0 and m1/2 for scalars (squarks, sleptons andHiggs bosons) and gauginos and higgsinos at the GUT scale,a common trilinear Higgs–sfermion–sfermion coupling A0,as well as tanβ and the sign of the Higgs mass parameter μ.Using renormalization group equations, the masses and mix-ings of all SUSY and Higgs particles are determined at theelectroweak scale. The parameter space has been scannedfor fixed values of m0 = 50, 100, 150, 200 and 250 GeV/c2

in the range of m1/2 = 100–300 GeV/c2 and tanβ = 1.5–50 with A0 = 0 [80]. The resulting 5σ discovery contoursare shown in Fig. 25(right), projected onto the (mA, tanβ)

plane, for fixed values of m0, a negative sign of μ and as-suming an integrated luminosity of 300 fb−1.

In the search for charged Higgs bosons, three-lepton finalstates from the decay chain gb → H±t , H± → χ0

2,3χ±1,2 and

Eur. Phys. J. C (2009) 59: 463–495 489

Fig. 25 (Left) The 5σ discovery contours in the (mA, tanβ) planefor H/A → χ0

2 χ02 → χ0

1 χ01 decays for different SUSY para-

meters and integrated luminosities (from Refs. [81, 82]). (Right) The5σ discovery contours in the (mA, tanβ) plane for H/A → χ0

2 χ02 →

χ01 χ0

1 in the mSUGRA scenario for fixed m0 = 50, 100 and200 GeV/c2, A0 = 0 and μ < 0. The parameter m1/2 has been scannedin the range between 100 and 300 GeV/c2. An integrated luminosityof 300 fb−1 is assumed (from Ref. [80])

χ02 → χ0

1 , χ±1 → χ0

1 ν can be exploited [81, 82, 94]. Theaccompanying top quark is required to decay hadronically.The discovery potential in this channel is limited to a smallregion of parameter space where sleptons are light and |μ| ≤150 GeV/c2, i.e. close to the lower limit set by the LEPexperiments.

6.6.2 Search for h → bb in SUSY cascade decays

The lightest Higgs boson might appear at the end of decaycascades of SUSY particles. One copious production sourcemay be the decay of the second lightest neutralino into thelightest neutralino, χ0

2 → χ01 h. The former is produced with

large rates in the decays of squarks and gluinos. In R-parityconserving SUSY models the lightest Higgs boson h is inthis decay mode accompanied by missing transverse energy,carried away by the lightest SUSY particle. The presence ofmissing transverse energy and several energetic jets from thesquark or gluinos cascades can be used to obtain a samplethat consists mainly of events containing SUSY particles. Inthis case, the discovery of the Higgs boson h in its dominantdecay mode h → bb becomes possible, without requiringthe presence of an additional lepton.

Both collaborations, ATLAS and CMS, have studiedthe observability of the h → bb signal in the mSUGRAparameter space [80–82, 181]. As an example, the 5σ

discovery contours, as obtained in Refs. [81, 82], are

Fig. 26 The 5σ discovery contours in the (m0,m1/2) plane for theCP-even Higgs bosons h via the decay h → bb from χ0

2 → χ01 h

decays in squark and gluino cascades for mSUGRA scenarios withtanβ = 30, A0 = 0 and μ > 0 for 10 and 100 fb−1. The regionsexcluded by experimental and theoretical constraints are indicated aswell (from Refs. [81, 82])

shown in Fig. 26 in the (m0,m1/2) plane with A0 = 0,tanβ = 30 and μ > 0 for integrated luminosities of 10 and100 fb−1.

490 Eur. Phys. J. C (2009) 59: 463–495

6.6.3 Search for heavy MSSM Higgs bosons in SUSYcascade decays

In a recent study four possible sources for the detection ofheavy MSSM Higgs bosons in SUSY cascade decays havebeen considered [181].

• Cascade decays of squarks and gluinos via the heavychargino χ±

2 and neutralinos χ03,4 with subsequent decays

into the lighter chargino χ±1 or neutralinos χ0

1,2 and Higgsbosons:

pp → g̃g̃, q̃q̃, q̃g̃

→ χ±2 , χ0

3 , χ04 + X

→ χ±1 , χ0

2 , χ01 + h,H,A,H± + X.

• Direct decays of squarks and gluinos into the lightestchargino and the next-to-lightest neutralino, with sub-sequent decays into the lightest neutralino and Higgsbosons:

pp → g̃g̃, q̃q̃, q̃g̃

→ χ±1 , χ0

2 + X

→ χ01 + h,H,A,H± + X.

• Direct decays of heavy stop and sbottom squarks into thelighter ones and Higgs bosons (in the case of large enoughsquark mass splitting):

pp → t̃2 t̃2, b̃2b̃2, with

t̃2(b̃2) → t̃1(b̃1) + h,H,A or b̃1(t̃1) + H±.

• Top quarks originating from SUSY particle cascades, de-caying into H± bosons:

pp → g̃g̃, q̃q̃, q̃g̃ → t + X → H± + X.

Four representative SUSY scenarios have been discussed:squarks are considered to be either lighter or heavier thangluinos and either light gaugino- or higgsino-like charginosand neutralinos are assumed. Using a fast simulation of theperformance of the CMS detector, signals of heavy neutral(charged) Higgs bosons have been reconstructed in the bb

(τν) decay mode. Backgrounds from both standard modeland SUSY processes have been taken into account. It hasbeen demonstrated that light Higgs bosons (mΦ < 200–250 GeV/c2) appearing in the cascade decays can be de-tected with a significance exceeding 5σ in some represen-tative MSSM scenarios. As an example, in Fig. 27 the re-constructed bb invariant mass is shown for the SUSY para-meter point with mg̃ = 1200 GeV/c2, mq̃ = 800 GeV/c2,M2 = 350 GeV/c2 and μ = 150 GeV/c2.

Fig. 27 The reconstructed bb invariant mass spectrum for aSUSY scenario with M2 = 350 GeV/c2, μ = 150 GeV/c2,mg̃ = 1200 GeV/c2 and mq̃ = 800 GeV/c2 in a simulation of theCMS experiment, assuming an integrated luminosity of 30 fb−1. Thetwo mass peaks result from decays of the lighter h and the heavierH/A Higgs bosons. They are shown on top of the backgrounds fromSUSY processes (dark) and from standard model t t production (light)(from Ref. [181])

For large M2 values there is sufficient phase space for thedecay of the heavier neutralinos and charginos, with masses∼M2, into the lighter higgsino states, with masses mχ± ∼mχ0

1∼ mχ0

2∼ |μ|, and Higgs particles with masses mΦ <

200 GeV/c2. In the distribution two mass peaks resultingfrom the decays of the light Higgs boson h and the heavierH and A are visible above the backgrounds. The discoverypotential at that parameter point for heavy Higgs bosons isshown in Fig. 28 for an integrated luminosity of 100 fb−1. Inthis scenario, the Higgs bosons A,H (H±) can be detectedvia the cascade decays for masses up to mA ∼ 220 GeV/c2

(200 GeV/c2) for all tanβ .As discussed in Ref. [181], this study is not meant to be

exhaustive but should rather be considered as a preliminaryinvestigation of a few representative scenarios to illustratethe discovery potential via cascade decays. In particular, amore detailed experimental simulation for a larger numberof SUSY parameter points is necessary and a discussion ofsystematic uncertainties on the backgrounds and their im-pact on the signal extraction needs to be performed. In addi-tion to the bb decay mode, also the ττ decay mode shouldbe considered for heavy neutral Higgs bosons. As discussedin Sect. 6.7.3, the ratio of the two signals could contributeto establish the SUSY nature of a Higgs boson signal in cas-cade decays.

It should be stressed that the cascade processes will beextremely useful to measure the couplings of supersymmet-ric particles to Higgs bosons, which again would be an es-sential ingredient to reconstruct the parameters of the un-derlying SUSY model in a global fit. If cascade decays via

Eur. Phys. J. C (2009) 59: 463–495 491

Fig. 28 The 5σ discovery contours for MSSM Higgs bosons in the(mA, tanβ) plane for an integrated luminosity of 100 fb−1 for theSUSY parameter point with mg̃ = 1200 GeV/c2, mq̃ = 800 GeV/c2,M2 = 350 GeV/c2 and μ = 150 GeV/c2. In the region withmA < 220 GeV/c2 the heavier Higgs bosons H and A can be detectedin cascade decays of squarks and gluinos (from Ref. [181])

heavier neutralinos are kinematically allowed and the rele-vant branching ratios are sufficiently large, the decay chainχ0

3,4 → h,H,A + χ01 → bb + χ0

1 provides important infor-mation to reconstruct the masses of the heavier neutralinos,which in turn helps to constrain the SUSY model signifi-cantly.

6.7 Determination of MSSM parameters

Assuming that non-standard model Higgs bosons will bediscovered at the LHC, it is important to extract the para-meters of the underlying model. For this, all relevant mea-surements from searches for SUSY and Higgs particles willbe taken into account in a global fit. The masses, productionrates and branching ratios of the observed Higgs bosons areexpected to contribute significantly to constrain the model.The accuracy which can be reached in the measurement ofthese input parameters to the global fit is discussed in thefollowing.

6.7.1 Measurement of the Higgs boson masses

The precision of the mass measurement of MSSM Higgsbosons has been determined under the same assumptions asdescribed in Sect. 5.1. While in addition to statistical uncer-tainties experimental systematic uncertainties on the back-ground subtraction and on the knowledge of the absoluteenergy scale have been included, no theoretical errors havebeen considered. As in the case of the standard model, lep-ton and photon final states provide the highest precision forthe Higgs boson mass measurements.

The mass of the lightest CP-even Higgs boson The lightCP-even Higgs boson h can be detected in the h → γ γ

decay mode over the full mass range of interest, i.e. be-tween ∼90 and ∼135 GeV/c2. For an integrated luminosityof 300 fb−1, a precision of the mass measurement of theorder of 0.1–0.5% can be achieved [80]. The larger valueis reached for parameter values close to the 5σ discoverycontour. A mass reconstruction is also possible in the vec-tor boson-fusion mode qqh → qqττ using the collinear ap-proximation (see Sect. 4.2); however, the precision is aboutan order of magnitude worse.

The masses of the heavier neutral Higgs bosons For mA <

2 mt and small values of tanβ , the mass of the H boson canbe determined in the H → ZZ(∗) → 4 and H → hh →bbγ γ decay modes. Assuming an integrated luminosity of300 fb−1, a precision of the order of 0.1–0.3% for the 4

and of about 1% for the bbγ γ channel can be achieved[80]. While in the latter case the measurement is limited bysystematic uncertainties, the range in the 4 channel resultsfrom the strong variation of the signal rate with tanβ overthe discovery region.

The pseudoscalar Higgs boson A can be discovered forsmall tanβ in the A → Zh → bb and in the A → γ γ de-cay mode. The expected precision of the mass measurementis of the order of 1–2% for bb and of the order of 0.1%for γ γ final states.

For large values of tanβ , the heavy Higgs bosons H andA will be discovered in the ττ and μμ decay modes. Overa large part of the parameter space they cannot be disentan-gled from each other, being almost degenerate in mass andhaving almost identical decay modes. In the region of pa-rameter space where both H/A → ττ and H/A → μμ de-cays are observable, the precision of the mass measurementis determined by the μμ decay mode and is estimated to beat the level of 0.1%. For moderate values of tanβ , where thediscovery reach of the ττ channel extends further than thatof the μμ channel, the precision is degraded to 1–7%.

The mass of the charged Higgs boson Both the H± → τν

and H± → tb decay modes can be used to extract infor-mation on the mass of the charged Higgs boson. In τν de-cays the mass is determined from the shape of the trans-verse mass distribution using a likelihood technique [156].For an integrated luminosity of 300 fb−1, the precision ofthe mass measurement varies between 0.8 and 1.8% forcharged Higgs boson masses in the range between 200 to500 GeV/c2 [80, 156]. In this estimate, uncertainties on theshape and on the rate of the background and on the energyscale are taken into account. Due to the larger backgrounds,the precision of the mass measurement is found to be worsein the tb decay mode [156].

492 Eur. Phys. J. C (2009) 59: 463–495

6.7.2 Measurement of tanβ

It has been suggested to determine the MSSM parametertanβ from the measurement of the production cross sectionof heavy neutral and charged MSSM Higgs bosons [182]. Atlarge tanβ the cross sections are approximately proportionalto tan2 β; however, loop corrections involving SUSY parti-cles modify this behavior. Due to potentially large radiativecorrections to the bottom-Yukawa coupling, the results ob-tained from cross section measurements in this way corre-spond to a measurement of an effective parameter tanβeff

[183–186]. The extraction of the fundamental tanβ parame-ter requires additional knowledge of the sbottom and gluinomasses as well as of the Higgs boson mass parameter μ.

The method has been applied by the ATLAS [80] andmore recently by the CMS Collaboration [187]. In the latteranalysis, the precision of the tanβ measurement has beendetermined by taking into account a systematic uncertaintyof ±20% on the next-to-leading order cross sections and of±3% on the branching ratios. Since the cross section de-pends on the Higgs boson mass, the uncertainty on the massmeasurement has also been included. In order to determinethe sensitivity of the tanβ measurement to SUSY parame-ters, they have been varied in the range of ±20% aroundtheir nominal values (M2 = 200 GeV/c2, μ = 300 GeV/c2,MSUSY = 1 TeV/c2 and At = 2450 GeV/c2). Over the re-gion of parameter space where a discovery is possible, such

a variation changes the cross section by at most 11%, whichleads to a ±6% uncertainty on tanβ [187]. For the determi-nation of the experimental numbers, a 2–3% uncertainty onsignal efficiencies, mainly related to uncertainties in τ - andb-tagging, and a ±5% uncertainty on the luminosity havebeen assumed.

In the analysis, all combinations of tau decay channels(, τhad and τhadτhad) have been used to extract the finalmeasurement. In the parameter range close to the 5σ dis-covery contour, where the signal rate is smallest, a statisti-cal uncertainty of the order of 11–12% for tanβ is found. InFig. 29(left) the statistical and the combined statistical plussystematic uncertainty on tanβ is shown as a function of mA

for tanβ = 10, 20, 30 and 40, assuming an integrated lumi-nosity of 30 fb−1. The total uncertainty ranges from 12 to19% depending on tanβ and mA.

As mentioned above, also the cross section measurementwill be used in a global fit together with other relevant mea-surements to determine the SUSY parameters simultane-ously.

6.7.3 Discrimination between the standard model andthe MSSM

The observation of the light Higgs boson h in various searchchannels might allow for a discrimination between the stan-dard model Higgs sector and its extensions via the measure-ment of e.g. ratios of branching ratios in the same production

Fig. 29 (Left): Relative uncertainty on the tanβ measurement in theCMS experiment assuming an integrated luminosity of 30 fb−1. Theuncertainty is given for fixed values of tanβ as a function of mA

including statistical (lower curves) and statistical plus systematic un-certainties (upper curves) (from Ref. [187]). (Right): Sensitivity for the

discrimination between the standard model and the MSSM from themeasurement of (see text) in the mh-max scenario. The discoverycontours are shown for a luminosity of 300 fb−1, for the vector bosonfusion channels only 30 fb−1 are assumed

Eur. Phys. J. C (2009) 59: 463–495 493

mode, which is advantageous as several systematic uncer-tainties cancel. The sensitivity of a discrimination betweenthe standard model and the MSSM has been estimated usingthe ratio R of branching ratios measured in the vector boson-fusion production mode: R = BR(h → ττ)/BR(h → WW).The red (black) area in Fig. 29 displays the regions inwhich , defined as = (RMSSM − RSM)/σexp is largerthan 1 (2). Here σexp denotes the expected error on the ratioR in this particular point of MSSM parameter space. Onlystatistical uncertainties have been taken into account and mh

is assumed to be known with high precision. A similar study,based on the analysis used in the determination of the Higgsboson couplings, reaches the same conclusion [101].

6.8 Search for an invisibly decaying Higgs boson

Some extensions of the standard model predict Higgs bosonsthat decay into stable neutral weakly interacting particles(see e.g. Ref. [188] for more details). In a collider detector,such decays would lead to invisible final states and trigger-ing and detection would only be possible if the Higgs bosonis produced in association with other particles.

At the LHC, the search for invisibly decaying Higgsbosons is much more difficult than at e+e− colliders, sincethe missing mass technique cannot be applied [188]. Higgsbosons can be searched for in the associated productionmodes with vector bosons (WH,ZH), t t pairs (t tH ) andjets (vector boson fusion mode qqH ). All three processeshave already been suggested in the literature and feasibilitystudies have been performed [189–193]. In all cases missingtransverse energy is used as a key signature for the presenceof an invisible Higgs boson decay.

The LHC Collaborations have carried out preliminaryanalyses which indicate that the experiments at the LHChave some potential to detect excesses of events with largemissing transverse energy above the background from stan-dard model processes [188]. For a reliable measurement anormalization of the backgrounds in the experiment will benecessary. In the associated ZH and t tH production modesthe expected signal cross sections are small, in the range ofa few fb, and the normalization might be affected by largeuncertainties. However, those channels are nevertheless ofinterest since they can be reliably triggered, and in addition,the t tH channel does not rely on the vector boson coupling,which might be suppressed in certain scenarios. It shouldalso be mentioned that all studies performed so far only con-sidered backgrounds from standard model processes. Theimpact of non-standard model backgrounds, which might bepresent if Higgs bosons decay invisibly, remains to be stud-ied.

7 Conclusions

It has been demonstrated in numerous experimental stud-ies that the experiments at the CERN Large Hadron Col-lider have a huge discovery potential for the standard modelHiggs boson. If the Higgs mechanism is realized in naturethe corresponding Higgs boson should not escape detectionat the LHC. The full mass range can be explored and theHiggs boson can be detected in several decay modes. In ad-dition, the parameters of the resonance can be measured withadequate precision to establish Higgs boson-like couplingsto bosons and heavy fermions. For both the discovery andthe parameter measurements the vector boson-fusion modeplays an important role.

In the minimal supersymmetric standard model, Higgsbosons can be detected across the entire parameter spacewith a significance of more than 5σ for established CP con-serving benchmark scenarios.

Acknowledgements I would like to thank the numerous colleaguesfrom both the theoretical and experimental community for providingsuch an impressive wealth of results to review. In particular, gratitudeis expressed to Filip Moortgat, Sacha Nikitenko, Elzbieta Richter-Was,Markus Schumacher, Michael Spira and Dieter Zeppenfeld for veryuseful discussions and comments and to all my colleagues from theATLAS Higgs working group for the excellent collaboration over manyyears.

References

1. P.W. Higgs, Phys. Rev. Lett. 12, 132 (1964)2. P.W. Higgs, Phys. Rev. 145, 1156 (1966)3. F. Englert, R. Brout, Phys. Rev. Lett. 13, 321 (1964)4. G.S. Guralnik, C.R. Hagen, T.W. Kibble, Phys. Rev. Lett. 13, 585

(1964)5. ALEPH, DELPHI, L3, OPAL Collaborations, Phys. Lett. B 565,

61 (2003)6. LEP Collaborations ALEPH, DELPHI, L3 and OPAL,

LEP Electroweak Working Group, SLD Electroweak andHeavy Flavor Groups, hep-ex/0312023. Updated numbersfrom the LEP Electroweak Working Group, March 2008.http://lepewwg.web.cern.ch/LEPEWWG

7. B.W. Lee et al., Phys. Rev. Lett. 38, 883 (1977)8. M. Quiros, hep-ph/97034129. A. Ghinculov, T. Binoth, Acta Phys. Pol. B 30, 99 (1999)

10. L. Maiani, G. Parisi, R. Petronzio, Nucl. Phys. B 136, 115 (1979)11. N. Cabibbo et al., Nucl. Phys. B 158, 295 (1979)12. R. Dashen, H. Neunberger, Phys. Rev. Lett. 50, 1897 (1983)13. D.J.E. Callaway, Nucl. Phys. B 233, 189 (1984)14. M.A. Beg et al., Phys. Rev. Lett. 52, 883 (1984)15. M. Lindner, Z. Phys. C 31, 295 (1986)16. G. Altarelli, G. Isidori, Phys. Lett. B 337, 141 (1994)17. J.A. Casas, J.R. Espinosa, M. Quiros, Phys. Lett. B 342, 171

(1995)18. J.A. Casas, J.R. Espinosa, M. Quiros, Phys. Lett. B 383, 374

(1996)19. B. Grzadkowski, M. Lindner, Phys. Lett. B 178, 81 (1986)20. T. Hambye, K. Riesselmann, Phys. Rev. D 55, 7255 (1997)21. H.P. Nilles, Phys. Rep. 110, 1 (1984)22. H.E. Haber, G.L. Kane, Phys. Rep. 117, 75 (1985)

494 Eur. Phys. J. C (2009) 59: 463–495

23. S.P. Martin, in Perspectives on Supersymmetry, ed. by G.L. Kane.(World Scientific, Singapore, 1998), p. 1. hep-ph/9709356

24. S. Heinemeyer, W. Hollik, G. Weiglein, Eur. Phys. J. C 9, 343(1999)

25. S. Heinemeyer, G. Weiglein, J. High Energy Phys. 10, 72 (2002)26. G. Degrassi, S. Heinemeyer, W. Hollik, P. Slavich, G. Weiglein,

Eur. Phys. J. C 28, 133 (2003)27. S. Heinemeyer, hep-ph/040724428. O. Buchmüller et al., arXiv:0707.344729. TEVNPH Working Group, for the CDF and DØ Experiments,

arXiv:0712.238330. M. Spira, private communication31. A. Djouadi, M. Spira, P.M. Zerwas, Phys. Lett. B 264, 440

(1991)32. S. Dawson, Nucl. Phys. B 359, 283 (1991)33. D. Graudenz, M. Spira, P.M. Zerwas, Phys. Rev. Lett. 70, 1372

(1993)34. M. Spira, A. Djouadi, D. Graudenz, P.M. Zerwas, Phys. Lett. B

318, 347 (1993)35. M. Spira, A. Djouadi, D. Graudenz, P.M. Zerwas, Nucl. Phys. B

453, 17 (1995)36. S. Dawson, R.P. Kauffman, Phys. Rev. D 49, 2298 (1994)37. R.V. Harlander, Phys. Lett. B 492, 74 (2000)38. R.V. Harlander, W.B. Kilgore, Phys. Rev. D 64, 013015 (2001)39. S. Catani, D. de Florian, M. Grazzini, J. High Energy Phys. 0105,

025 (2001)40. R.V. Harlander, W.B. Kilgore, Phys. Rev. Lett. 88, 201801

(2002)41. C. Anastasiou, K. Melnikov, Nucl. Phys. B 646, 220 (2002)42. V. Ravindran, J. Smith, W.L. van Neerven, Nucl. Phys. B 665,

325 (2003)43. M. Krämer, E. Laenen, M. Spira, Nucl. Phys. B 511, 523 (1998)44. R.V. Harlander, in Proc. Hadron Collider Physics (Springer,

Berlin, 2002)45. R. Harlander, private communication46. W. Beenakker, S. Dittmaier, M. Krämer, B. Plümper, M. Spira,

P.M. Zerwas, Phys. Rev. Lett. 87, 201805 (2001)47. W. Beenakker, S. Dittmaier, M. Krämer, B. Plümper, M. Spira,

P.M. Zerwas, Nucl. Phys. B 653, 151 (2003). hep-ph/021135248. L. Reina, S. Dawson, Phys. Rev. Lett. 87, 201804 (2001)49. L. Reina, S. Dawson, D. Wackeroth, Phys. Rev. D 65, 053017

(2002)50. S. Dawson, L.H. Orr, L. Reina, D. Wackeroth, Phys. Rev. D 67,

071503 (2003)51. J. Campbell et al., in Proc. Les Houches Workshop, 2003.

hep-ph/040530252. R.M. Barnett, H.E. Haber, D.E. Soper, Nucl. Phys. B 306, 697

(1988)53. F.I. Olness, W.K. Tung, Nucl. Phys. B 308, 813 (1988)54. D. Dicus, S. Willenbrock, Phys. Rev. D 39, 751 (1989)55. S. Dittmaier, M. Krämer, M. Spira, Phys. Rev. D 70, 074010

(2004). hep-ph/030920456. S. Dawson, C. Jackson, L. Reina, D. Wackeroth, Phys. Rev. D

69, 074027 (2004)57. J. Campbell, R.K. Ellis, F. Maltoni, S. Willenbrock, Phys. Rev.

D 67, 095002 (2003). hep-ph/0204093v258. D. Dicus, T. Stelzer, Z. Sullivan, S. Willenbrock, Phys. Rev. D

59, 094016 (1999)59. C. Balazs, H.J. He, C.P. Yuan, Phys. Rev. D 60, 114001 (1999)60. R. Harlander, W.B. Kilgore, Phys. Rev. D 68, 013001 (2003)61. M. Carena, J.S. Conway, H.E. Haber et al., hep-ph/001033862. G.C. Blazey et al., hep-ex/000501263. V. Barger, C.E.M. Wagner et al., hep-ph/000315464. D. Rainwater, M. Spira, D. Zeppenfeld, in Proc. Les Houches,

2001. hep-ph/0203187. http://wwwlapp.in2p3.fr/conferences/LesHouches/Houches2001

65. T. Han, G. Valencia, S. Willenbrock, Phys. Rev. Lett. 69, 3274(1992)

66. T. Figy, C. Oleari, D. Zeppenfeld, Phys. Rev. D 68, 073005(2003). hep-ph/0306109

67. E.L. Berger, J. Campbell, Phys. Rev. D 70, 073011 (2004).hep-ph/0403194

68. T. Han, S. Willenbrock, Phys. Lett. B 273, 167 (1991)69. A. Djouadi, M. Spira, Phys. Rev. D 62, 014004 (2000)70. R. Hamberg, W.L. van Neerven, T. Matsuura, Nucl. Phys. B 359,

343 (1991)71. O. Brein, A. Djouadi, R. Harlander, Phys. Lett. B 579, 149

(2004)72. A. Djouadi, J. Kalinowski, M. Spira, Comput. Phys. Commun.

108, 56 (1998). hep-ph/970444873. A. Djouadi, M. Spira, P.M. Zerwas, Z. Phys. C 70, 427 (1996).

hep-ph/951134474. A. Ghinculov, T. Binoth, J.J. van der Bij, Phys. Lett. B 427, 343

(1998)75. A. Djouadi, hep-ph/050317276. A. Djouadi, hep-ph/050317377. R. Harlander, Acta Phys. Pol. B 38, 693 (2007)78. G. Aad et al. (ATLAS Collaboration), J. Instrum. 3, S08003

(2008)79. S. Chatrchysn et al. (CMS Collaboration), J. Instrum. 3, S08004

(2008)80. ATLAS, Collaboration, Detector and physics performance tech-

nical design report, CERN/LHCC/99-15, 199981. CMS, Collaboration, CMS physics technical design report, II:

physics performance, CERN/LHCC/2006-021, 200682. CMS Collaboration. J. Phys. G: Nucl. Part. Phys. 34, 995 (2006)83. D.L. Rainwater, D. Zeppenfeld, J. High Energy Phys. 12, 5

(1997)84. D.L. Rainwater, D. Zeppenfeld, K. Hagiwara, Phys. Rev. D 59,

014037 (1999)85. T. Plehn, D.L. Rainwater, D. Zeppenfeld, Phys. Rev. D 61,

093005 (2000)86. D.L. Rainwater, D. Zeppenfeld, Phys. Rev. D 60, 113004 (1999)

[Erratum Phys. Rev. D 61 (2000) 099901]87. N. Kauer, T. Plehn, D.L. Rainwater, D. Zeppenfeld, Phys. Lett.

B 503, 113 (2001)88. R. Brun, F. Carminati, CERN Program Library Long Writeup

W5013, 199389. U. Baur, E.W.N. Glover, in Proc. of the Large Hadron Collider

Workshop, Aachen, 1990, ed. by G. Jarlskog, D. Rein, CERN90-10/ECFA 90-133, vol. II, p. 570

90. M. Dittmar, H.K. Dreiner, Phys. Rev. D 55, 167 (1997)91. T. Binoth, M. Ciccolini, N. Kauer, M. Krämer, J. High Energy

Phys. 03, 065 (2005)92. M. Dührssen, K. Jakobs, P. Marquardt, J.J. van der Bij, J. High

Energy Phys. 05, 064 (2005)93. S. Asai et al., Eur. Phys. J. C 32, 209 (2003). hep-ph/040225494. ATLAS Collaboration, Performance of the ATLAS experiment—

detector, trigger, physics, CERN-OPEN-2008-020 (to appear)95. D. Benedetti et al., J. Phys. G: Nucl. Part. Phys. 34, 221 (2007)96. M. Mangano et al., J. High Energy Phys. 07, 001 (2003)97. H. Baer, J.D. Wells, Phys. Rev. D 57, 4446 (1998)98. K. Jakobs, A study of the associated production WH , with W →

ν and H → WW(∗) → νν, ATLAS note, ATL-PHYS-2000-008

99. L. Brücher, R. Santos, Eur. Phys. J. C 12, 87 (2000), and refer-ences therein

100. D. Zeppenfeld, R. Kinnunen, A. Nikitenko, E. Richter-Was,Phys. Rev. D 62, 013009 (2000)

101. M. Dührssen, Prospects for the measurement of Higgs bosoncoupling parameters in the mass range from 110–190 GeV/c2,ATLAS note, ATL- PHYS-2003-030

Eur. Phys. J. C (2009) 59: 463–495 495

102. J.R. Dell’Aquila, C.A. Nelson, Phys. Rev. D 33, 101 (1986)103. A. Skold, P. Osland, Phys. Lett. B 329, 305 (1994)104. J.F. Gunion, X.G. He, Phys. Rev. Lett. 76, 4468 (1996)105. S.Y. Choi, D.J. Miller, M.M. Muhllleitner, P.M. Zerwas, Phys.

Lett. B 553, 6 (2003)106. R.M. Godbole, D.J. Miller, M.M. Muhlleitner, J. High Energy

Phys. 0712, 031 (2007)107. P.S. Bhupal, A. Djouadi, R.M. Godbole, M.M. Muhlleitner, S.D.

Rindani, arXiv:0707.2878108. C.P. Buszello et al., Eur. Phys. J. C 32, 209 (2004)109. T. Plehn, D. Rainwater, D. Zeppenfeld, Phys. Rev. Lett. 88,

051801 (2002)110. C. Ruwiedel et al., Eur. Phys. J. C 51, 385 (2007)111. M. Spira, hep-ph/9510347112. B. Kniehl, Phys. Rep. 240, 211 (1994)113. M. Dührssen et al., Phys. Rev. D 70, 113009 (2004).

hep-ph/0406323114. D. Dicus, C. Kao, S. Willenbrock, Phys. Lett. B 203, 457 (1988)115. E.W.N. Glover, J.J. van der Bij, Nucl. Phys. B 309, 282 (1988)116. F. Boudjema, E. Chopin, Z. Phys. C 73, 85 (1996)117. V.A. Ilyin et al., Phys. Rev. D 54, 6717 (1996)118. A. Djouadi, W. Kilian, M. Mühlleitner, P.M. Zerwas, Eur. Phys.

J. C 10, 27 (1999)119. D.J. Miller, S. Moretti, Eur. Phys. J. C 13, 459 (2000)120. T. Plehn, M. Spira, P.M. Zerwas, Nucl. Phys. B 479, 46 (1996)121. A. Djouadi, W. Kilian, M. Mühlleitner, P.M. Zerwas, Eur. Phys.

J. C 10, 45 (1999)122. G. Azuelos et al., J. Phys. G 28, 2453 (2002). hep-ph/0204087123. U. Baur, T. Plehn, D. Rainwater, Phys. Rev. Lett. 89, 151801

(2002)124. U. Baur, T. Plehn, D. Rainwater, Phys. Rev. D 68, 033003 (2003)125. F. Mazzucato, A. Blondel, A. Clark, Studies on the measurement

of the SM Higgs self-couplings, ATLAS note, ATL-PHYS-2002-029

126. U. Baur, T. Plehn, D. Rainwater, Phys. Rev. D 69, 053004 (2004).hep-ph/0310056

127. E. Richter-Was et al., Int. J. Mod. Phys. A 13, 1371 (1998)128. M. Carena, P.M. Zerwas et al., in Proc. of the LEP2 Workshop,

ed. by G. Altarelli, T. Sjostrand, F. Zwirner, CERN, 1995129. J. Dai, J.F. Gunion, R. Vega, Phys. Lett. B 345, 29 (1995)130. J. Dai, J.F. Gunion, R. Vega, Phys. Lett. B 387, 801 (1996)131. K.J.F. Gaemers, G. Hoogeveen, Phys. Lett. B 146, 347 (1984)132. A. Dicus, A. Stange, S. Willenbrock, Phys. Lett. B 333, 126

(1994)133. A.C. Bawa, C.S. Kim, A.D. Martin, Z. Phys. C 47, 75 (1990)134. L.G. Jin, C.S. Li, R.J. Oakes, S.H. Zhu, Eur. Phys. J. C 14, 91

(2000)135. A. Belyaev, D. Garcia, J. Guasch, J. Sola, J. High Energy Phys.

06, 059 (2002)136. S.H. Zhu, hep-ph/0112109137. T. Plehn, Phys. Rev. D 67, 014018 (2003)138. E.L. Berger, T. Han, J. Jiang, T. Plehn, hep-ph/0312286139. J. Gunion, H.E. Haber, F. Paige, W. Tung, S. Willenbrock, Nucl.

Phys. B 294, 621 (1987)140. D.A. Dicus, J.L. Hewett, C. Kao, T.G. Rizzo, Phys. Rev. D 40,

787 (1989)141. W. Hollik, S.H. Zhu, Phys. Rev. D 65, 075015 (2002)142. S. Willenbrock, Phys. Rev. D 35, 173 (1987)143. S.H. Zhu, C.S. Li, C.S. Gao, Phys. Rev. D 58, 055007 (1998)144. S. Moretti, K. Odagiri, Phys. Rev. D 59, 055008 (1999)145. O. Brein, W. Hollik, Eur. Phys. J. C 13, 175 (2000)146. O. Brein, W. Hollik, S. Kanemura, Phys. Rev. D 63, 095001

(2001)147. A.A. Barrientos Bendezu, B.A. Kniehl, Phys. Rev. D 59, 015009

(1999)

148. A.A. Barrientos Bendezu, B.A. Kniehl, Phys. Rev. D 64, 035006(2001)

149. M. Drees, D.P. Roy, Phys. Lett. B 269, 155 (1991)150. J.F. Gunion, Phys. Lett. B 322, 125 (1994). hep-ph/9312201151. V.D. Barger, R.J. Phillips, D.P. Roy, Phys. Lett. B 324, 236

(1994)152. K. Odagiri, Phys. Lett. B 452, 327 (1999)153. S. Moretti, D.P. Roy, Phys. Lett. B 470, 209 (1999)154. D.P. Roy, Phys. Lett. B 277, 183 (1992)155. D.P. Roy, Phys. Lett. B 459, 607 (1999)156. K.A. Assamagan, Y. Coadou, A. Deandrea, hep-ph/0203121157. H.P. Nilles, M. Srednicki, D. Wyler, Phys. Lett. B 124, 337

(1983)158. J.M. Frere, D.R.T. Jones, S. Raby, Nucl. Phys. B 222, 11 (1983)159. J.R. Ellis, J.F. Gunion, H.E. Haber, L. Roszkowski, F. Zwirner,

Phys. Rev. D 39, 844 (1989)160. M. Drees, Int. J. Mod. Phys. A 4, 3635 (1989)161. U. Ellwanger, M. Rausch de Traubenberg, C.A. Savoy, Phys.

Lett. B 315, 331 (1993)162. S.F. King, P.L. White, Phys. Rev. D 52, 4183 (1995)163. M. Drees, M. Guchait, D.P. Roy, Phys. Lett. B 471, 39 (1999)164. U. Ellwanger, J.G. Gunion, C. Hugonie, J. High Energy Phys.

0507, 041 (2005)165. M. Carena, S. Heinemeyer, C.E. Wagner, G. Weiglein, Eur. Phys.

J. C 26, 601 (2003)166. M. Schumacher, hep-ph/0410112167. M. Schumacher, private communication168. S. Heinemeyer, W. Hollik, G. Weiglein, Phys. Rev. D 58, 091701

(1998)169. S. Heinemeyer, W. Hollik, G. Weiglein, Phys. Lett. B 440, 296

(1998)170. S. Heinemeyer, W. Hollik, G. Weiglein, FEYNHIGGS program.

http://www.feynhiggs.de171. E.L. Berger, C.W. Chiang, J. Jiang, T.M.P. Tait, C.E.M. Wagner,

Phys. Rev. D 66, 095001 (2002). hep-ph/0205342172. M. Carena et al., Phys. Lett. B 495, 155 (2000)173. E. Accomando et al., hep-ph/0608079174. H. Baer, M. Bisset, X. Tata, J. Woodside, Phys. Rev. D 46, 46

(1992)175. M.M. Nojiri, G. Polesello, D.R. Tovey, hep-ph/0312317176. L. Alvarez-Gaume, J. Polchinski, M.B. Wise, Nucl. Phys. B 221,

495 (1983)177. L. Ibanez, Phys. Lett. B 118, 73 (1982)178. J. Ellis, D.V. Nanopolous, K. Tamvakis, Phys. Lett. B 121, 123

(1983)179. K. Inoue et al., Prog. Theor. Phys. 68, 927 (1982)180. A.H. Chamseddine, R. Arnowitt, P. Nath, Phys. Rev. Lett. 49,

970 (1982)181. A. Datta, A. Djouadi, M. Guchait, F. Moortgat, Nucl. Phys. B

681, 31 (2004). hep-ph/0303095182. J.F. Gunion et al., in Snowmass 96. hep-ph/9703330183. L.J. Hall, R. Rattazi, U. Sarid, Phys. Rev. D 50, 7048 (1994)184. M. Carena, M. Olechowski, S. Pokorski, C.E. Wagner, Nucl.

Phys. B 426, 269 (1994)185. M. Carena, D. Garcia, U. Nierste, C.E. Wagner, Nucl. Phys. B

577, 88 (2000)186. J. Guasch, P. Häflinger, M. Spira, hep-ph/0305101187. R. Kinnunen et al., CMS NOTE 2004/027188. V. Büscher, K. Jakobs, Int. J. Mod. Phys. 20, 2523 (2005)189. D. Choudhury, D.P. Roy, Phys. Lett. B 322, 368 (1994).

hep-ph/9312347190. S. Frederiksen, N. Johnson, G. Kane, J. Reid, Phys. Rev. D 50,

4244 (1994)191. J.F. Gunion, Phys. Rev. Lett. 72, 199 (1994). hep-ph/9309216192. O.J.P. Eboli, D. Zeppenfeld, Phys. Lett. B 495, 147 (2000)193. H. Davodiasl, T. Han, H.E. Logan, hep-ph/0412269