Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 1222
Searches for a Charged Higgs Boson in ATLAS and Development of Novel Technology for Future Particle Detector Systems
DANIEL PELIKAN
ACTA UNIVERSITATIS UPSALIENSIS ISSN 1651-6214 ISBN 978-91-554-9153-6 UPPSALA urn:nbn:se:uu:diva-242491 2015 Dissertation presented at Uppsala University to be publicly examined in Polhemssalen, Ångströmlaboratoriet, Lägerhyddsvägen 1, Uppsala, Friday, 20 March 2015 at 10:00 for the degree of Doctor of Philosophy. The examination will be conducted in English. Faculty examiner: Prof. Dr. Fabrizio Palla (Istituto Nazionale di Fisica Nucleare (INFN) Pisa).
Abstract Pelikan, D. 2015. Searches for a Charged Higgs Boson in ATLAS and Development of Novel Technology for Future Particle Detector Systems. Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 1222. 119 pp. Uppsala: Acta Universitatis Upsaliensis. ISBN 978-91-554-9153-6.
The discovery of a charged Higgs boson (H±) would be a clear indication for physics beyond the Standard Model. This thesis describes searches for charged Higgs bosons with the ATLAS experiment at CERN’s Large Hadron Collider (LHC). The first data collected during the LHC Run 1 is analysed, searching for a light charged Higgs boson (mH± Keywords: Charged Higgs boson, Matrix Method, ATLAS, 60 GHz, future particle detector Daniel Pelikan, Department of Physics and Astronomy, High Energy Physics, 516, Uppsala University, SE-751 20 Uppsala, Sweden. © Daniel Pelikan 2015 ISSN 1651-6214 ISBN 978-91-554-9153-6 urn:nbn:se:uu:diva-242491 (http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-242491) List of papers This thesis is based on the following papers, which are referred to in the text by their Roman numerals. I The ATLAS Collaboration, + Search for charged Higgs bosons decaying√ via H → τν in top quark pair events using pp collision data at s = 7 TeV with the ATLAS detector, JHEP 1206 (2012) 039 II The ATLAS Collaboration, Search for charged Higgs bosons through the violation√ of lepton universality in ttbar events using pp collision data at s = 7 TeV with the ATLAS experiment, JHEP 03 (2013) 076 III The ATLAS Collaboration, Estimation of non-prompt and fake lepton backgrounds in√ final states with top quarks produced in proton-proton collisions at s = 8 TeV with the ATLAS detector, ATLAS-CONF-2014-058 IV Daniel Pelikan, Nils Bingefors, Richard Brenner, Dragos Dancila, Leif Gustafsson, Wireless data transfer with mm-waves for future tracking detectors, 2014 JINST 9 C11008 V Daniel Pelikan, Nils Bingefors, Richard Brenner, Dragos Dancila, Leif Gustafsson, Radial transfer of tracking data with wireless links, PoS(TIPP2014)095 List of papers not included in this thesis: VI The ATLAS Collaboration, Study of discriminating variables for charged Higgs boson searches in tt¯ events with leptons, using 35/pb of data from the ATLAS detector, ATLAS-CONF-2011-018 VII The ATLAS Collaboration, Search for a charged Higgs boson decaying via H+ → τlep + ν in tt¯ events with one or two light√ leptons in the final state using 1.03/fb of pp collision data recorded at s = 7 TeV with the ATLAS detector, ATLAS-CONF-2011-151 VIII The ATLAS Collaboration, ± Search for charged Higgs bosons decaying√ via H → τν in tt¯ events using 4.6 f b−1 of collision data at s = 7 TeV with the ATLAS detector, ATLAS-CONF-2012-011 IX Daniel Pelikan, Search for charged Higgs bosons in ATLAS, EPJ Web of Conferences 28, 12057 (2012) Hadron Collider Physics Symposium (HCP) 2011 Reprints were made with permission from the publishers. Contents 1 Introduction ...... 9 2 Theoretical background ...... 11 2.1 The Standard Model of particle physics ...... 11 2.1.1 The particles of the Standard Model ...... 11 2.1.2 The Higgs mechanism ...... 14 2.2 Beyond the Standard Model (BSM) physics ...... 16 2.2.1 Phenomena not explained by the Standard Model ...... 16 2.2.2 Supersymmetry ...... 17 2.2.3 The charged Higgs bosons ...... 20 2.2.4 Production and decay of charged Higgs bosons ...... 22 3 The CERN laboratory ...... 34 3.1 Scientific achievements made by CERN ...... 34 3.2 The accelerator infrastructure ...... 34 3.2.1 Linac-2 ...... 36 3.2.2 Linac-3 ...... 36 3.2.3 Linac-4 ...... 36 3.2.4 Proton Synchrotron Booster (PSB) ...... 36 3.2.5 Proton Synchrotron (PS) ...... 36 3.2.6 Super Proton Synchrotron (SPS) ...... 36 3.3 The Large Hadron Collider (LHC) ...... 37 3.4 The four experiments at the LHC ...... 38 4 The ATLAS detector system ...... 39 4.1 Conventions ...... 39 4.2 The inner detector ...... 41 4.3 The calorimeters ...... 42 4.4 The muon spectrometer ...... 44 4.5 The trigger system ...... 46 5 Physics analysis in ATLAS ...... 47 5.1 Charged Higgs boson searches in ATLAS ...... 47 5.1.1 Measurement of discriminating variables for charged Higgs boson searches ...... 47 5.1.2 Charged Higgs boson searches in the single- and di-lepton channels ...... 48 5.1.3 Combination of the most sensitive channels for charged Higgs boson searches ...... 49 5.1.4 Charged Higgs boson searches through the violation of lepton universality ...... 50 5.1.5 Charged Higgs boson searches with the 2012 dataset ... 51 5.2 The Matrix Method ...... 51 5.2.1 The Matrix Method related to top quark and charged Higgs boson physics in ATLAS ...... 52 5.2.2 Physics analyses ...... 53 6 Future data readout challenges at LHC ...... 58 6.1 Introduction ...... 58 6.2 Wireless technology in future detector systems ...... 59 6.2.1 Choice of the frequency band for wireless data links ... 60 6.2.2 Connection concept ...... 61 6.2.3 Requirements on the antenna system ...... 62 6.3 Radiation damage to antenna substrate ...... 63 6.3.1 Effect of changes in the dielectric constant ...... 63 6.3.2 Radiation damage studies ...... 64 6.4 Outlook ...... 65 7 Wireless technology ...... 66 7.1 Electromagnetic waves ...... 66 7.2 The scattering parameter ...... 67 7.3 Antenna characteristics ...... 71 7.4 The transmission line ...... 72 7.4.1 Transmission line types ...... 72 7.4.2 Characteristic impedance of a transmission line ...... 73 7.4.3 Matched and unmatched transmission line ...... 74 7.4.4 The Smith chart ...... 75 7.4.5 Impedance matching ...... 76 7.5 Signal modulation ...... 78 7.6 The patch antenna ...... 82 7.6.1 Design steps for a patch antenna ...... 82 7.6.2 Impedance matching ...... 85 7.6.3 Design of the feeding-line ...... 86 7.6.4 Antenna simulation ...... 87 7.7 Fabrication of the antennas ...... 90 7.8 Test setup for passive repeater antennas ...... 92 8 Summary of papers ...... 95 9 Conclusion ...... 97 10 Summary in Swedish - Sammanfattning på svenska ...... 99 11 Acknowledgements ...... 102 References ...... 107 1. Introduction One of the great philosophical questions of mankind is “Where do we come from?”. Science is trying to answer this question with the help of evolution in biology, the Big Bang in physics and many other theories. The question “Where do we come from?” can be reduced to the question “What are we made of?”. This leads to the topic of particle physics, the study of the smallest building blocks of our universe and their interaction with each other. In order to answer the question “Where do we come from?” we first need to answer a set of more fundamental questions. Part of this thesis is trying to answer some of them. The most fundamental particles that we believe exist, and their interactions with each other, are summarized in the Standard Model of particle physics. The Large Hadron Collider (LHC) and the detectors ATLAS and CMS were built to search for new physics and hunt the last missing particle of the Stan- dard Model, the Higgs boson. It was discovered in 2012 and the discovery was honoured with a Nobel Prize one year later. But there are more open questions which need to be answered. The Standard Model is not believed to be the final theory and sci- entists expect to find new physics beyond the Standard Model. For instance, the Standard Model does not explain gravity and is not a Grand Unified The- ory (GUT). One possible extension of the Standard Model is supersymmetry, which can include gravity, can explain dark matter, the other 95% of matter which are not Standard Model Matter, furthermore supersymmetry has the po- tential to be a Grand Unified Theory. Supersymmetry requires at least two Higgs doublets leading to five Higgs bosons, where two are charged and three are neutral, compared to one neutral state in the Standard Model. This the- sis presents searches for charged Higgs bosons with the ATLAS experiment. Their discovery would be a clear indication for physics beyond the Standard Model. Searches for light charged Higgs bosons (lighter than the top quark mass) were performed in different channels from the beginning of LHC Run 1 until its end. With more and more data being accumulated, better exclusion limits were set and the search for the heavy charged Higgs bosons was started. The ATLAS detector system is an excellent detector for muon and electron identification, therefore many of the search channels that were chosen had leptons in their final state. One key ingredient in physics searches is the esti- mation of backgrounds. Some of the backgrounds are well modelled in simu- lation, while some others must be derived using data. One of these data driven estimated backgrounds is the amount of misidentified leptons in the final se- lection. This background is very difficult and computing intensive to model, 9 therefore it is estimated directly from data. The method used for the estima- tion is called Matrix Method. It was optimized to fit the changing conditions of the ATLAS data, e.g. increasing pile-up and new detector calibrations. In order to probe physics beyond the Standard Model, colliders and exper- iments eventually have to reach even higher energies and interaction rates. This opens the possibility to produce particles with high masses and rare pro- duction probabilities. As a result the technology of the existing and future particle detectors has to be upgraded and research has to be done to develop these new technologies. One major problem that modern detectors face is the huge amount of information generated per collision in these systems. The high granularity of modern detectors is needed to cope with the huge amounts of generated tracks and energy deposits happening during collisions. In order to reconstruct the whole event, and not to get blinded by all the pile-up in the detector, a high resolution is essential. High granularity means many readout channels, a lot of electronics and cables needed to process or bring out the data. The electronics and cables introduce a lot of non sensitive material into the detector system and should be kept to a minimum. One way to reduce the amount of cables and connectors is the usage of wireless technology in order to read out the detector system. Other scenarios are envisaged, for example where wireless based communication between tracking hardware is placed in- side the detector. In such a scenario, decisions on the interest of an event are already made before the detector is read out. With such an approach, a lot of bandwidth would be saved. The detector only needs to be read out completely if the event is really of interest for physics. This thesis consists of two main parts, physics searches and technology devel- opment for future detectors. The physics search is introduced in chapter 2 with some theoretical background about particle physics and especially charged Higgs bosons. CERN with its accelerator infrastructure is introduced in chap- ter 3, and the detector system ATLAS used for the physics studies is described in chapter 4. Chapter 5 is about the physics analysis of charged Higgs bosons. The relation between physics analysis and hardware developement is ex- plained in chapter 6, where the use of wireless technology in future detector systems is motivated by the search for new physics. Requirements on future experimental facilities and readout scenarios are presented as well as require- ments on the wireless technology itself. Chapter 7 gives an introduction into wireless technology and antenna design. A summary of the papers listed in this thesis is given in chapter 8, underlining the contributions of the author. The thesis is concluded in chapter 9,followedbyasummaryinSwedishin chapter 10 for non experts. 10 2. Theoretical background 2.1 The Standard Model of particle physics ȡȕȧȧ → Α Β ȅșȎ ΐ ΑΖΔ ǿșȎ ΐΖΒ ǿșȎ Ώ ȗȜȕȦțș → ΑΎΒ ΑΎΒ ΑΎΒ Ώ ț ȧȤȝȢ → ΐΎΑ ȩȗΐΎΑ ΐΎΑ Ȩ ΐ ȩȤ ȗȜȕȦȡ ȨȣȤ țȠȩȣȢ Γ Η ȅșȎ ΘΔ ȅșȎ Γ ΐΗ ǿșȎ Ώ ΐΑΔ Θ ǿșȎ ΐΎΒ ΐΎΒ ΐΎΒ Ώ γ Ώ ΐΎΑ Ș ΐΎΑ ȧ ΐΎΑ Ȗ ΐ Ώ Ȁ ȉȩȕȦȟȧ Ȁȝțțȧ ȘȣȫȢ ȧȨȦȕȢțș ȖȣȨȨȣȡ ȤȜȣȨȣȢ ȖȣȧȣȢ Δΐΐ ȟșȎ ΐΏΔ Ζ ȅșȎ ΐ ΖΗ ǿșȎ Θΐ ΐΘ ǿșȎ ΐ ΑΎΒ μ ΑΎΒ τ Ώ ΐΎΑ ș ΐΎΑ ΐΎΑ ΐ Ȓ șȠșȗȨȦȣȢ ȡȩȣȢ Ȩȕȩ Ȓ ȖȣȧȣȢ ǶΑ Α șȎ ǶΏ ΐΘ ȅșȎ ǶΐΗ Α ȅșȎ ΗΏ ΒΘ ǿșȎ Ώ ν Ώ νμ Ώ ντ ɍΐ ΐΎΑ ș ΐΎΑ ΐΎΑ ΐ ȏ ȄșȤȨȣȢȧ șȠșȗȨȦȣȢ ȡȩȣȢ Ȩȕȩ ȢșȩȨȦȝȢȣ ȢșȩȨȦȝȢȣ ȢșȩȨȦȝȢȣ ȏ ȖȣȧȣȢ ǿȕȩțș ǺȣȧȣȢȧ Figure 2.1. Schematic illustration of the particle content of the Standard Model [1]. 2.1.1 The particles of the Standard Model The Standard Model (SM) of elementary particles [2, 3, 4] was developed in the 1960s and 1970s. It is a theory describing the fundamental constituents, the building blocks of matter and their interactions through electromagnetic, strong and weak forces. The matter particles are elementary, which means that they do not have a substructure. They are called fermions and have half integer spin, they can be further categorized into leptons and quarks. The force carriers are bosons, particles with an integer spin. A schematic illustration of the particle content of the Standard Model is given in figure 2.1. 11 The fermions are organized in three families. The LEP experiments showed that there are no more than three families with light neutrinos (mν < mZ/2) in the SM [5]. Each family consists of two quarks and two leptons. The quarks can be categorized into one up-type quark and one down-type quark, the leptons into a electron like lepton and a neutrino. The up-type quark carries + 2 − 1 an electric charge of 3 e and the down type quark an electric charge of 3 e. The massive leptons (electron, muon and tau) have an electric charge of −1e. Neutrinos are electrically neutral and much lighter than the charged leptons. All particles with an electric charge interact electromagnetically. All fermions interact weakly, but the neutrinos have the weak force as their only way to interact. Quarks, unlike leptons, also carry a colour charge, which is the quanta of the strong interaction. Quarks come in three colours, “red”, “green” and “blue”, additionally they only exist in a confined state for which the sum of the colour charges is white. Baryons are bound states of three quarks, or three anti- quarks. Mesons are composed of a quark and anti-quark pair. In other words, the colour white can be achieved by combining one of each red, green and blue (or anti-red, anti-green and anti-blue) quarks, or by adding a colour and the corresponding anti-colour. The gauge bosons are the force carriers. The photon, γ is the carrier of the electromagnetic force, the W and Z bosons are the carriers of the weak force and the gluons, g are the carriers of the strong force. The Higgs boson is the last boson in the SM, and was discovered in 2012 by the ATLAS and CMS collaborations [6, 7]. It is different from the gauge bosons and its role in the SM will be explained in 2.1.2. Figure 2.2 shows the distribution of the four-lepton invariant mass m4l measured by ATLAS and the invariant mass of diphoton candidates measured by CMS, as a selection of channels in which the Higgs boson with a mass of around 125 GeV was discovered. These are the golden channels for Higgs boson seaches, since the backgrounds in these channels are very low. In order to get enough sensitivity to claim a discovery, both experiments combined several channels (H → γγ, H → ZZ, H → WW, H → ττ, H → bb). The local p0 value [8] with respect to the Higgs boson mass of the com- bination is shown in figure 2.3 for both ATLAS and CMS. The p0 value is a measure of the significance at which the null hypothesis (SM without a Higgs boson) would be rejected. The smaller the p0-value, the stronger the evi- dence to favour an alternative hypothesis. Lines of sigma levels (the number of statistical fluctuations of the background) are shown in the plot to guide the translation from p0-value to sigma level [9]. As can be seen almost 6σ in ATLAS and 5σ in CMS are reached. The 5σ level means that a new particle was discovered, with 99.9999% probability and < 0.0001% probability it is just a background fluctuation. 12 &06 V 7H9/ IE V 7H9/ IE Data 9 ATLAS H (*) 8QZHLJKWHG 25 Background ZZ (*) * H→ZZ →4l Background Z+jets, tt Signal (m =125 GeV) 20 H YHQWV Events/5 GeV Syst.Unc. ( ∫ -1 15 s = 7 TeV: Ldt = 4.8 fb P γγ *H9 s = 8 TeV: ∫Ldt = 5.8 fb-1 10 'DWD 6%)LW %)LW&RPSRQHQW 5 ±σ ± σ 0 6 6% :HLJKWHG(YHQWV*H9 100 150 200 250 P γγ *H9 m4l [GeV] (a) ATLAS (b) CMS Figure 2.2. Four-lepton invariant mass, m4l, for the selected candidates, measured in ATLAS (left). The data with error bars is shown in black while the background and signal expectation from simulation are shown in colours [6]. Invariant mass distribu- tion of di-photon candidates in CMS (right) [7]. Note that these plots are not with the full 2012 dataset, these are the plots from the discovery paper, plots with the full 2012 dataset can be found in: [10, 11]. CMS s = 7 TeV, L = 5.1 fb-1 s = 8 TeV, L = 5.3 fb-1 1 1σ 2σ 0 10-2 ATLAS 2011 - 2012 3σ 1 -1 10 10-4 σ Local p -2 4 10 2 σ Local p-value -3 10 3 σ -6 ∫ -1 10 σ 10-4 s = 7 TeV: Ldt = 4.6-4.8 fb 5 -5 -1 4 σ 10 s = 8 TeV: ∫Ldt = 5.8-5.9 fb -8 Combined obs. 10-6 10 Exp. for SM H 6σ 10-7 5 σ H γγ→ Sig. Expected → 10-8 -10 H ZZ Observed 10 H → WW -9 H ττ→ 10 6 σ -10 H → bb 7σ 10 10-12 110 150 200 300 400 500 110 115 120 125 130 135 140 145 m [GeV] H mH (GeV) (a) ATLAS (b) CMS Figure 2.3. Combined search results of the observed (solid) local p0 as a function of the Higgs boson mass mH and the expectation (dashed) for a SM Higgs boson signal hypothesis at the given mass for ATLAS (left) [6]. On the right side the observed local p0 for the five decay modes (H → γγ, H → ZZ, H → WW, H → ττ, H → bb)which were combined and the overall combination as a function of the SM Higgs boson mass [7]. Note that these plots are not with the full 2012 dataset, these are the plots from the discovery paper. 13 2.1.2 The Higgs mechanism In a local gauge theory the gauge boson fields are massless, because of the gauge symmetry. From experiments it is known [12, 13, 14, 15] that the gauge bosons, except the photon and gluons, have masses. Introducing explicit mass terms in the SM generates problems with gauge invariance and renormaliza- tion, making it impossible to predict observables with good accuracy. Hence, there has to be a different mechanism to introduce mass into the SM. This can be done by introducing spontaneous symmetry breaking [16, 17]. Particle physics is based on Quantum Field Theory (QFT) [18] and the Higgs mecha- nism will be derived on the principals of QFT in the following: A complex Higgs SU (2) doublet Φ (x) Φ(x)= a (2.1) Φb(x) is included in the SM, with a SU (2) ×U(1) invariant Lagrangian: H μ † L (x)=[D Φ(x)] [Dμ Φ(x)] −V(Φ), (2.2) where Φ† is the Hermitian adjoint of Φ and V (Φ) is the Higgs potential of the form: V (Φ)=μ2(Φ†(x)Φ(x)) + λ(Φ†(x)Φ(x))2. (2.3) With the boundary conditions μ2 < 0andλ > 0, the potential becomes the Mexican hat shaped potential shown in figure 2.4. Dμ is the gauge covariant derivative, containing the interaction terms which lead to the gauge bosons W, Z and γ. μ and λ are real parameters. Figure 2.4. The Higgs potential, the minimum has a non zero value. The Mexican hat potential has a continuum of minima. The electroweak symmetry can be broken by choosing one minimum using the freedom of a 14 global SU (2) rotation. A convenient choice for such a vacuum state is: Φ0 μ2 Φ = a = √1 0 = − 0 Φ0 with v λ (2.4) b 2 v Expanding around the vacuum state gives: η ( )+ η ( ) Φ( )= √1 1 x i 2 x x + σ( )+ η ( ) (2.5) 2 v x i 3 x where ηi(x) and σ(x) are scalar fields. Using this equation in the Lagrangian yields mass terms, by spontaneous symmetry breaking, for the gauge bosons. After diagonalising, the W and Z bosons aquire masses, while the photon γ remains massless. The surviving μ2 field σ(x), with a vacuum expectation value of v = − λ ≈ 246 GeV [1], the average expected value in the vacuum, gives rise to a massive, electrically neutral, spin-0 particle, called the Higgs boson. It should be noted that the vacuum state is not SU (2) gauge invariant, even though the Lagrangian is. This is called spontaneous symmetry breaking. More details of the derivation can be found in [18]. Yukawa couplings, the interactions between a scalar field and a Dirac field, i.e. between the Higgs field and the fermions, indicates the mass scale of the quarks and leptons in the SM. The fermion masses can be generated using the scalar field Φ, with hyper- ∗ charge Y = 1, and the isodoublet Φ˜ = iτ2Φ with a hypercharge Y = −1. τ2 is the second Pauli matrix1. For any fermion generation, the SU (2)L × U(1)Y invariant Yukawa La- grangian can be introduced [19]: LF = −λeL¯ΦeR − λdQ¯ΦdR − λuQ¯Φ˜ uR + h.c. (2.6) where L¯ and Q¯ are the isodoublets of the left-handed fermions for leptons and quarks, respectivly and eR, uR and dR are the isosinglets of the right- handed fermions for leptons, up-type quarks and down-type quarks. After gauge transformation the Lagrangian can be written in the form: 1 LF = −√ λ f (v + σ) f¯L fR + h.c. (2.7) 2 The constant term in front of the fermion isosinglets f¯L fR is the mass of the fermion: λev λuv λdv me = √ , mu = √ , md = √ (2.8) 2 2 2 01 0 −i 10 1Pauli matrices: τ = , τ = , τ = 1 10 2 i 0 3 0 −1 15 2.2 Beyond the Standard Model (BSM) physics Even though the SM is a very successful theory, it is not believed to be the final theory for describing matter and the fundamental interactions. It is regarded as a theory which is valid up to a certain energy level. There are other theories be- yond the SM which try to explain phenomena which can not be accounted for by the SM. Example of these theories are supersymmetric models (see sec- tion 2.2.2) like the Minimal Supersymmetric Standard Model (MSSM) [20] and Next-to-Minimal Supersymmetric Standard Model (NMSSM) [21], Little Higgs models [22], String Theory [23], Technicolour [24] and Extra Dimen- sion [25]. All these theories contain the SM, but are more general. Which of these theories best describes physics beyond the SM, can only be verified experimentally. 2.2.1 Phenomena not explained by the Standard Model There are several phenomena which can not be explained by the SM but with a possible explanation or solution in one of the extensions of the SM. The most prominent ones are mentioned in the following: Dark matter and dark energy: Observations of the cosmic microwave back- ground show that the SM can explain around 5% of the energy content in the universe. It is believed that around 27% is dark matter and 68% is dark en- ergy [26]. The existence and properties of dark matter are studied from its gravitational effects on visible matter, radiation and large-scale structures of the universe. Dark Energy is used to explain the acceleration of the expansion of the universe. Matter anti-matter asymmetry: The universe is mainly made of matter, e.g. particles, not anti-particles. The SM predicts that matter and anti-matter were produced in almost equal amounts in the Big Bang. Charge Parity (CP) Violation in the SM, the effect that matter does not exactly behave in the same way as anti-matter under CP transformation, can account for some but not the full SM matter in our universe. Hence the reason for the matter anti-matter asymmetry must be phenomena not explained by the SM [27]. Three generations: From the Z-boson measurements it is known that there are exactly three generations of matter particles. However, the SM does not explain why [5]. Grand Unification Theory (GUT): There are three gauge symmetries in the SM corresponding to the three fundamental forces. The coupling constants of these interactions vary with energy and become approximately equal around 1016 GeV. That is why it is expected that the three symmetries unite into one single gauge symmetry above this energy. Below the threshold of around 1016 GeV the symmetry is spontaneously broken into the three SM symme- tries [28]. This unification can be included in supersymmetric models, see figure 2.5. 16 (a) Standard Model (b) SUSY Model Figure 2.5. Evolution of the SU(3) × SU(2) ×U(1) gauge couplings to high energy scales, using the one-loop renormalization group equation of the SM (left) and the supersymmetric generalization of the SM (right) [28]. Hierarchy problem: In the SM, masses of particles are introduced by spon- taneous symmetry breaking caused by the Higgs field. Due to the presence of virtual particles, the Higgs mass requires large quantum corrections in the SM. These corrections are significantly larger than the measured Higgs boson mass, which requires that the mass parameter of the Higgs boson in the SM is fine-tuned in order to cancel the quantum corrections. This requirement of fine-tuning is considered unnatural. Quantum gravity: Gravity is not included in the SM. Adding gravity to the SM is not possible as it generates unphysical effects like particle velocities higher than the speed of light, and further modifications would be needed. The most successful theory (General Relativity) is incompatible with the SM [29]. Gravity acts on a macroscopic scale, therefore the SM predictions are not af- fected by it. Neutrino mass: In the SM neutrinos are massless and cannot change flavour. The observed number of electron neutrinos arriving from the sun is much smaller than the number predicted by the Standard Solar Model [30], this is called the solar neutrino problem. To solve this problem neutrinos need to have mass, in order to oscillate between flavours [31]. 2.2.2 Supersymmetry Some of the problems described in section 2.2.1 can be solved by supersym- metry (SUSY). SUSY is a proposed extension of space-time described by the Poincaré group, which relates boson and fermion fields to each other. Each particle from one group has a “superpartner” in the other group, with a spin difference of a half-integer. This generates at least a doubling of the SM par- ticle spectrum. If SUSY exists, and would be unbroken, every particle in the 17 SM would have a superpartner. This superpartner would have exactly the same mass as its SM counterpart and should have therefore already been discovered. This means that, assuming SUSY exists, it has to be broken. Superpartners could exist with higher masses compared to SM particles. Considering SUSY as a true symmetry of nature, the Hierarchy problem would be solved since the quantum corrections are cancelled by those from the corresponding super- partners above the SUSY breaking scale. Also the dark matter problem can be solved since SUSY can provide a dark matter candidate particle. The light- est supersymmetric particle, provided there exists a conservation law which forbids its decay into SM particles, could be a good candidate for dark mat- ter. SUSY is also a Grand Unification Theory, since within SUSY the weak, strong and electromagnetic interactions can be unified at high-energy [28]. Also quantum gravity can be included in Local Supersymmetric Models. Minimal Supersymmetric Standard Model The Minimal Supersymmetric Standard Model (MSSM) is the most economic implementation of SUSY beyond the SM in terms of new particles and addi- tional parameters. It is one of the best studied candidates for physics beyond the SM and was introduced in 1981 to stabilize the weak scale and solving the hierarchy problem [32]. As described in the previous section, SUSY associates bosons with fermions. In the MSSM, the supersymmetric scalars are named after the corresponding fermion in the SM. For each fermion in the SM, there are two scalar fermions (sfermion) in the MSSM to account for left-handed and right-handed states in the SM. For the gauge bosons in the SM, there are half-integer spin gaugions in the MSSM. These superpartners are named after the gauge eigenstate in the SM with an added “-ino” suffix: Bino, Wino and gluino. The Bino and the Wino mix with the half-integer spin superpartners of the Higgs boson, the χ 0 ,χ0,χ 0,χ 0 Higgsinos. There are four Neutralinos mass eigenstates ( 1 2 3 4 ), which are fermions and are electrically neutral. They are a mixture of the Bino, the neutral Wino and the neutral Higgsinos. The two Chargino mass eigenstates χ ±,χ ± ( 1 2 ) are electrically charged fermions. They are a mixture of the charged Wino and charged Higgsinos. A single Higgsino would lead to a gauge anomaly, i.e. not a good symme- try. If two Higgsinos are added, this gauge anomaly is solved. The simplest approach is one with two Higgsinos and two scalar Higgs doublets. Also in order to have Yukawa couplings between the Higgs boson and both down- and up-type quarks, two scalar Higgs doublets are needed. The strength of the mixing of left-handed and right-handed sfermions is given by the mass of the corresponding SM fermion. The particle content of the MSSM is given in table 2.1. So far no evidence for SUSY has been observed [34, 35, 36, 37, 38, 39, 40] as can be seen in figure 2.6. 18 h aes[ papers the 2.6. Figure ATLAS SUSY Searches* - 95% CL Lower Limits ATLAS√ Preliminary Status: ICHEP 2014 s =7,8TeV − e,μ,τ,γ Emiss L dt[fb 1] Model Jets T Mass limit Reference MSUGRA/CMSSM 0 2-6 jets Ye s 20.3 q˜, g˜ 1.7 TeV m(q˜)=m(g˜) 1405.7875 MSUGRA/CMSSM 1 e,μ 3-6 jets Ye s 20.3 g˜ 1.2 TeV any m(q˜) ATLAS-CONF-2013-062 34 iiso UYsace rmALS h lti oplto from compilation a is plot The ATLAS. from searches SUSY on Limits MSUGRA/CMSSM 0 7-10 jets Ye s 20.3 g˜ 1.1 TeV any m(q˜) 1308.1841 χ0 χ˜0 st . nd . q˜q˜, q˜→q ˜1 0 2-6 jets Ye s 20.3 q˜ 850 GeV m( 1)=0 GeV, m(1 gen ˜q)=m(2 gen ˜q) 1405.7875 , χ0 χ˜0 g˜g˜, g˜→qq¯ ˜1 0 2-6 jets Ye s 20.3 g˜ 1.33 TeV m( 1)=0 GeV 1405.7875 35 ± ± χ ±χ0 ,μ χ˜0 < χ˜ χ˜0 g˜g˜, g˜→qq ˜1 →qqW ˜1 1 e 3-6 jets Ye s 20.3 g˜ 1.18 TeV m( 1) 200 GeV, m( )=0.5(m( 1)+m(g˜)) ATLAS-CONF-2013-062 χ0 ,μ χ0 g˜g˜, g˜→qq(/ν/νν) ˜ 2 e 0-3 jets - 20.3 g˜ 1.12 TeV m( ˜1)=0GeV ATLAS-CONF-2013-089 , 1 GMSB (˜ NLSP) 2 e,μ 2-4 jets Ye s 4 . 7 g˜ 1.24 TeV tanβ<15 1208.4688 36 GMSB (˜ NLSP) 1-2 τ +0-1 0-2 jets Ye s 2 0 . 3 g˜ 1.6 TeV tanβ>20 1407.0603 γ χ0 GGM (bino NLSP) 2 - Ye s 20.3 g˜ 1.28 TeV m( ˜1)>50 GeV ATLAS-CONF-2014-001 , 0 GGM (wino NLSP) 1 e,μ+ γ - Ye s 4 . 8 g˜ 619 GeV m(χ˜ )>50 GeV ATLAS-CONF-2012-144 Inclusive Searches 1 37 γ χ0 GGM (higgsino-bino NLSP) 1 b Ye s 4.8 g˜ 900 GeV m( ˜1)>220 GeV 1211.1167 GGM (higgsino NLSP) 2 e,μ(Z) 0-3 jets Ye s 5 . 8 g˜ 690 GeV m(NLSP)>200 GeV ATLAS-CONF-2012-152 , − Gravitino LSP 0 mono-jet Ye s 10.5 1/2 scale 645 GeV m(G˜)>10 4 eV ATLAS-CONF-2012-147 38 F χ0 χ˜0 < g˜→bb¯ ˜1 03b Ye s 2 0 . 1 g˜ 1.25 TeV m( 1) 400 GeV 1407.0600 , 0 0 → χ˜ 0 7-10 jets Ye s 20.3 g˜ 1.1 TeV m(χ˜ ) <350 GeV 1308.1841 gen. g˜ tt¯ 1 1 39 med. 0 0 → ¯χ˜ 0-1 e,μ 3 b Ye s 2 0 . 1 g˜ 1.34 TeV m(χ˜ )<400 GeV 1407.0600 rd g˜ tt 1 1 ˜ g + 0 3 χ ,μ χ < g˜→bt¯˜1 0-1 e 3 b Ye s 20.1 g˜ 1.3 TeV m( ˜1) 300 GeV 1407.0600 , 0 χ0 40 ˜ ˜ ˜ → χ˜ 0 2 b Ye s 2 0 . 1 b˜ 100-620 GeV m( ˜ )<90 GeV 1308.2631 b1b1, b1 b ±1 1 1 χ ,μ ˜ χ± χ0 b˜1b˜1, b˜1→t ˜1 2 e (SS) 0-3 b Ye s 20.3 b1 275-440 GeV m( ˜1 )=2 m( ˜1) 1404.2500 ± 0 ]. → χ ,μ ˜ χ t˜1t˜1(light), t˜1 b ˜1 1-2 e 1-2 b Ye s 4.7 t1 110-167 GeV m( ˜1)=55 GeV 1208.4305, 1209.2102 χ0 ,μ ˜ χ˜0 << χ˜± t˜1t˜1(light), t˜1→Wb˜1 2 e 0-2 jets Ye s 20.3 t1 130-210 GeV m( 1)=m(t˜1)-m(W)-50 GeV, m(t˜1) m( 1 ) 1403.4853 0 0 ˜ ˜ ˜ → χ˜ 2 e,μ 2jets Ye s 20.3 t˜ 215-530 GeV m(χ˜ )=1 GeV 1403.4853 t1t1(medium), t1 t 1± 1 1 → χ χ0 < χ± χ0 t˜1t˜1(medium), t˜1 b ˜1 0 2 b Ye s 20.1 t˜1 150-580 GeV m( ˜1) 200 GeV, m( ˜1 )-m( ˜1)=5 GeV 1308.2631 χ0 ,μ ˜ χ˜0 t˜1t˜1(heavy), t˜1→t ˜1 1 e 1 b Ye s 20 t1 210-640 GeV m( 1)=0 GeV 1407.0583 gen. squarks χ0 ˜ χ˜0 t˜1t˜1(heavy), t˜1→t ˜1 0 2 b Ye s 20.1 t1 260-640 GeV m( 1)=0 GeV 1406.1122 rd χ0 χ0 ˜ ˜ ˜ → ˜ 0 mono-jet/c-tag Ye s 20.3 t˜1 90-240 GeV m(t˜ )-m( ˜ )<85 GeV 1407.0608 3 direct productiont1t1, t1 c 1 1 1 ,μ χ0 t˜1t˜1(natural GMSB) 2 e (Z) 1 b Ye s 2 0 . 3 t˜1 150-580 GeV m( ˜1)>150 GeV 1403.5222 → + ,μ χ0 t˜2t˜2, t˜2 t˜1 Z 3 e (Z) 1 b Ye s 20.3 t˜2 290-600 GeV m( ˜1)<200 GeV 1403.5222 χ0 ,μ χ0 ˜L,R˜L,R, ˜→ ˜1 2 e 0 Yes 20.3 ˜ 90-325 GeV m( ˜1)=0 GeV 1403.5294 + − + χ± 0 ± 0 χ˜ χ˜ , χ˜ →ν˜ (ν˜) 2 e,μ 0 Ye s 20.3 ˜ 140-465 GeV m(χ˜ )=0 GeV, m(,˜ ν˜)=0.5(m(χ˜ )+m(χ˜ )) 1403.5294 +1 −1 +1 ±1 1 1 1 χ χ χ →τν τν τ - χ˜ χ˜0 τ, ν χ˜± χ˜0 ˜ 1 ˜1 , ˜1 ˜ ( ˜) 2 Ye s 2 0 . 3 1 100-350 GeV m( 1)=0 GeV, m(˜ ˜)=0.5(m( 1 )+m( 1)) 1407.0350 χ±χ0 ,μ χ±, χ0 χ± χ0 χ0 χ± χ0 ˜ ˜ →˜ ν˜ (˜νν),ν˜˜ (˜νν) 3 e 0 Ye s 20.3 ˜ ˜ 700 GeV m( ˜1 )=m( ˜2), m( ˜1)=0, m(,˜ ν˜)=0.5(m( ˜1 )+m( ˜1)) 1402.7029 EW 1 2 L L L 1 2 ± ± 0 ± direct χ χ0 χ0 χ0 ,μ χ , χ χ χ0 χ0 ˜ ˜ →W ˜ Z ˜ 2-3 e 0Yes20.3˜ ˜ 420 GeV m( ˜1 )=m( ˜2), m( ˜1)=0, sleptons decoupled 1403.5294, 1402.7029 ±1 2 1 1 1± 2 ± χ χ0→ χ0 χ0 1 ,μ χ˜ , χ˜ 0 χ˜ χ˜0 χ˜0 ˜ 1 ˜2 W ˜1h ˜ 1 e 2 b Ye s 20.3 1 2 285 GeV m( 1 )=m( 2), m( 1)=0, sleptons decoupled ATLAS-CONF-2013-093 χ0χ0 χ0 → ,μ χ˜ 0 χ˜0 χ˜0 χ˜0 ,˜ ν χ˜0 χ˜0 ˜ 2 ˜ 3, ˜ 2,3 ˜R 4 e 0 Yes 20.3 2,3 620 GeV m( 2)=m( 3), m( 1)=0, m( ˜)=0.5(m( 2)+m( 1)) 1405.5086 + − ± ± χ˜ χ˜ χ˜ χ˜ χ˜± χ˜0 = τ χ˜± = Direct 1 1 prod., long-lived 1 Disapp. trk 1 jet Ye s 20.3 1 270 GeV m( 1 )-m( 1) 160 MeV, ( 1 ) 0.2 ns ATLAS-CONF-2013-069 χ0 Stable, stopped g˜ R-hadron 0 1-5 jets Ye s 2 7 . 9 g˜ 832 GeV m( ˜1)=100 GeV, 10 μs<τ(˜g)<1000 s 1310.6584 τ χ0→τ , μ τ ,μ 1-2 μ -- χ˜ 0 10 Scalar gluon pair, sgluon→qq¯ 0 4jets - 4.6 sgluon 100-287 GeV incl. limit from 1110.2693 1210.4826 Scalar gluon pair, sgluon→tt¯ 2 e,μ(SS) 2 b Ye s 14.3 sgluon 350-800 GeV ATLAS-CONF-2013-051 WIMP interaction (D5, Dirac χ) 0 mono-jet Ye s 1 0 . 5 M* scale 704 GeV m(χ)<80 GeV, limit of<687 GeV for D8 ATLAS-CONF-2012-147 Other √ √ √ s =7TeV s =8TeV s =8TeV 10−1 1 full data partial data full data Mass scale [TeV] 19 *Only a selection of the available mass limits on new states or phenomena is shown. All limits quoted are observed minus 1 σ theoretical signal cross section uncertainty. Table 2.1. The proposed particles in the MSSM [33]. Names Spin R-parity (PR) Gauge Eigenstates Mass Eigenstates + 0 0 + − 0 0 0 ± Higgs bosons 0 1 Hu Hd Hu Hd h H A H u L u R dL dR (same) squarks 0 −1 s L s R c L c R (same) tL tR bL bR t1 t2 b1 b2 e L e R νe (same) sleptons 0 −1 μ L μ R νμ (same) τL τR ντ τ1 τ2 ντ / − 0 0 0 0 χ 0 χ 0 χ 0 χ 0 neutralinos 1 2 1 B W Hu Hd 1 2 3 4 / − ± + − χ ± χ ± charginos 1 2 1 W Hu Hd 1 2 gluino 1/2 −1 g (same) goldstino 1/2 − (gravitino) (3/2) 1 G (same) 2.2.3 The charged Higgs bosons Charged Higgs bosons are predicted by many models beyond the SM. The Two-Higgs-Doublet Model (2HDM) is an extension of the SM and together with the MSSM one important extension of the Higgs sector. As described in the previous section, the MSSM requires at least two Higgs doublets to prevent gauge anomalies. Because of this, two charged Higgs bosons and three neutral Higgs bosons exist in this 2HDM model after ab- sorption of three degrees of freedom to give mass to W +, W − and Z gauge bosons. But also theories with even more than two Higgs doublets exist. For example the NMSSM [21] contains two Higgs doublets and a Higgs singlet, leading to seven Higgs bosons. A problem that arrises when adding too many Higgs fields is that the rela- tion between the W and Z mass, mW 2 mZ = with θW the Weinberg angle [41] (2.9) tanθW without fine-tuning, is not represented correctly [42]. SUSY favours a 2HDM as the minimal necessary extension. In the 2HDM, the Higgs mechanism works similarly to the SM Higgs mechanism, as described in section 2.1.2. The Two-Higgs-Doublet Model One extension of the Higgs sector is to consider two Higgs doublets [43, 44]. By introducing two complex, SU (2) doublet scalar fields Φ1 and Φ2 with hy- percharge Y = 1 the Higgs potential can be written as [45]: 2The Weinberg angle also called weak mixing angle is a parameter in electroweak interaction theory. Spontaneous symmetry breaking rotates the original W 0 and B0 vector boson plane by the Weinberg angle, producing as a result the Z0 boson, and the photon. 20 (Φ ,Φ )= 2 Φ†Φ + 2 Φ†Φ − 2 Φ†Φ + V 1 2 m11 1 1 m22 2 2 m12 1 2 h.c. (2.10) 2 † † 1 † + λ4 Φ Φ2 Φ Φ1 + λ5 Φ Φ2 1 2 2 1 + λ Φ†Φ + λ Φ†Φ Φ†Φ + 6 1 1 7 2 2 1 2 h.c In this potential λi are real parameters and mij are parameters related to mass. The minimum of the potential is given when all λi ≥ 0 by: 1 0 1 0 < Φ1 >= √ ,<Φ2 >= √ (2.11) 2 v1 2 v2 The potential is gauge invariant, which guarantees the correct pattern of electroweak symmetry breaking over a large range of parameters. Φ1 and Φ2 contain the physical Higgs bosons. √ 1 2(G+ cosβ − H+ sinβ) Φ = √ , (2.12) 1 vcosβ − hsinα + H cosα + i(G0 cosβ − Asinβ) 2 √ 1 2(G+ sinβ + H+ cosβ) Φ2 = √ (2.13) 2 vsinβ − hcosα + H sinα + i(G0 sinβ + Acosβ) G+ and G0 are charged and neutral Goldstone bosons, the parameter α is the mixing angle between the mass eigenstates h and H,tanβ is the ratio of the vacuum expectation values v1 and v2, one of the key parameters of this model: v tanβ = 2 (2.14) v1 A detailed description of the derivations leading to equation (2.10)-(2.14)can be found in [45, 46]. There exist four types of 2HDM model (I - IV). They differ in the way the up-type quarks, down-type quarks and leptons couple to (and get mass from) the two Higgs doublets fields Φ1 and Φ2. One example how the physical Higgs bosons in the 2HDM couple to fermions is: 1 ( f ) ( f ) √ F¯ κ sin(β − α)+ρ cos(β − α) fRh, (2.15) 2 the couplings of H, H± and A are left out for simplicity. ¯ ¯ F stands for the isodoublets L and Q of the left-handed√ fermions, fR are the ( ) κ( f ) = 2m f = 2 + 2 right handed isosinglets eR, uR and dR, v with v v1 v2. The different types correspond to the different expressions for ρ( f ) in ta- ble 2.2. 21 Table 2.2. Coupling of the two Higgs doublets Φ1 and Φ2 in the 2HDM to the leptons (eR) and up-type (uR) and down-type quark (dR). The coupling strength is described by the Yukawa-matrices ρ. (u) (d) (e) uR dR eR ρ ρ ρ (u) (d) (e) Type I Φ2 Φ2 Φ2 κ cotβκcotβκcotβ (u) (d) (e) Type II Φ2 Φ1 Φ1 κ cotβ −κ tanβ −κ tanβ (u) (d) (e) Type III (Flipped) Φ2 Φ1 Φ2 κ cotβ −κ tanβκcotβ (u) (d) (e) Type IV (Lepton-specific) Φ2 Φ2 Φ1 κ cotβκcotβ −κ tanβ MSSM Higgs Sector The MSSM is a Type-II 2HDM where the vacuum expectation values of the two Higgs fields are given by: 0 0 < Φ1 >= ,<Φ2 >= (2.16) v1 v2 The phase of the Higgs doublet fields are chosen so that v1 and v2 are real and non-negative. There is no CP violation in the Higgs sector. For v1,v2 > 0 the parameter tanβ is limited by 0 ≤ β ≤ π/2. The 2HDM model has five physical Higgs bosons: • two charged Higgs bosons H±; • 0 0 > two neutral CP-even Higgs boson scalars H and h where mH0 mh0 ; • one neutral Higgs boson A0 which is a CP-odd pseudoscalar. At tree level in the MSSM, the properties of the five Higgs bosons can be expressed by only two parameters. Commonly used are tanβ and one of the Higgs masses. 0 ± β An expression relating the two gauge boson masses (mZ, mW ), tan and the Higgs boson mass mA is given by [47]: 1 m2 = m2 + m2 ∓ (m2 + m2 )2 − 4m2 m2 cos2 2β (2.17) h0,H0 2 A Z0 A Z0 A Z0 2 = 2 + 2 mH± mA0 mW ± (2.18) 2.2.4 Production and decay of charged Higgs bosons Charged Higgs bosons with a mass lighter than the top quark mass mt are dominantly produced at the LHC through gluon-gluon fusion in which top quark pairs are created with at least one top quark decay into a light charged Higgs boson instead of a W gg → tt¯ → H+bW −b¯, H+bH−b¯ and W +bH−b¯. (2.19) 22 For higher charged Higgs boson masses, the charged Higgs boson is directly produced. In the four-flavour scheme (4FS) there are no b quarks in the initial state, they are non-partonic [48], and therefore the lowest-order QCD production processes are gluon-gluon fusion and quark-antiquark annihilation, gg → tbH± and qq¯ → tbH±, (2.20) respectively. In the five-flavour scheme (5FS), the leading order process for the inclusive tbH± cross-section is gluon-bottom fusion, gb → tH±. (2.21) If all orders in perturbation theory were taken into account the 4FS and the 5FS would be identical, but the way of ordering the perturbative expansion is different. At any finite order, the two schemes include different parts of the all-order result which makes the cross-section predictions not fit exactly and leads to ambiguities [49]. The Feynman diagrams of the production processes for the light and heavy charged Higgs bosons are shown in figure 2.7. g b g H+ g b H+ ¯b g t ¯b∗ H+ t¯ W − t g ¯b ¯b t¯ t¯ g Figure 2.7. Production of light charged Higgs bosons in tt¯ pairs via gluon-gluon fusion (left), the production of heavy charged Higgs bosons in gluon-b-quark fusion (center) and gluon-gluon fusion (right). The MSSM parameter space has a large number of free parameters making a compleate scan impractical in experimental and phenomenological studies. Therefore charged Higgs boson searches are often interpreted in benchmark scenarios. In these scenarios only the two parameters that enter the Higgs sec- β tor tree-level predictions, MH± and tan are varied. Other SUSY parameters, entering via radiative corrections, are fixed to particular benchmark values, exhibiting certain features of the MSSM charged Higgs boson phenomenol- max ogy. One of these benchmark scenarios is the mh scenario, described in the next section in detail, which is used for some of the performance plots in this section. Charged Higgs bosons lighter than the top quark mass decay dominantly to τν as can be seen in figures 2.8 and 2.9. The cross-section and branching ratio for a top quark decaying into a charged Higgs boson as a function of the H± mass and tanβ is shown in figure 2.10. 23 Figure 2.8. Branching ratios of the MSSM charged Higgs bosons with tanβ = 3[50]. 0 0 10 10 -1 -1 10 LHC Higgs XS WG 2013 10 LHC Higgs XS WG 2013 ) ) +- +- max max -2 β -2 β mh , tan = 10 mh , tan = 50 10 ± 10 ± BR(H -> tb) BR(H -> tb) BR(H BR(H BR(H -> cs) BR(H -> cs) τ ν τ ν BR(H -> τ) BR(H -> τ) μ ν μ ν BR(H -> μ) BR(H -> μ) -3 -3 10 10 -4 -4 10 10 100 200 300 400 500 600 100 200 300 400 500 600 +- +- MH [GeV] MH [GeV] (a) BR for tanβ = 10 (b) BR for tanβ = 50 max Figure 2.9. Branching ratios of the charged MSSM Higgs boson in the mH scenario as a function of charged Higgs boson mass. The left (right) column shows the results for tanβ = 10(50) [51]. 24 40 ) 0.1 ± max bH 0.09 mh : mt(mt) = 166.8 GeV, mb(mt) = 2.31 GeV FeynHiggs → tanβ = 5 0.08 tanβ = 10 b) * 2 [pb] HDECAY LHC HIGGS XS WG 2011 WG XS HIGGS LHC BR(t 30 LHC Higgs XS WG 2011 + tanβ = 30 0.07 tanβ = 50 μ 0.06 = 200 GeV 0.05 20 b) BR(t -> W + 0.04 = 100 GeV ± 0.03 H m ± = 120 GeV H 10 0.02 m ± = 140 GeV m H 0.01 ± = 160 GeV mH (pp -> tt) BR(t H σ 0 0 1020304050 0 β 100 110 120 130 140 150 160 tan ± MH [GeV] ± ± ± (a) σtt · BR(t → bH ) · BR(t → bW ) · 2 (b) BR(t → bH ) ± ± Figure 2.10. Left: The cross-section σtt · BR(t → bH ) · BR(t → bW ) · 2 including uncertainties, where different tanβ regions are indicated with different colours. Right: ± The branching fraction BR(t → bH ) calculated with FEYNHIGGS and HDECAY as a function of tanβ [52]. A charged Higgs boson that is heavier than the top quark, will in the MSSM decay dominantly into tb. The cross-section for direct heavy charged Higgs boson production is shown in figure 2.11. In this plot the 4FS and the 5FS are combined according to the Santander matching [51, 49]. The 4FS and 5FS calculations provide the unique description of the cross-section in the / → / → ∞ asymptotic limits mH± mb 1andmH± mb , respectively. For charged Higgs boson masses away from these asymptotic regions both schemes can be used but include different types of higher-order contributions. The Santander matching interpolates between the asymptotic limits of very light and very heavy charged Higgs bosons. Benchmark scenarios in charged Higgs boson searches max The mh scenario defines a benchmark point optimized to maximize the the- oretical upper bound on mh for a given tanβ and fixed mt and the soft SUSY breaking parameter MSUSY . This benchmark point provides the largest parameter space in the mh direc- tion and conservative exclusion limits for tanβ. max The mh scenario used in the ATLAS papers I and II is defined in [53]. max Other versions of the mh scenario exists with small variations a sign change max max in the description of mh between [54]and[53] and the (constrained) mh max scenario [55]. Further an (updated) mh scenario [56] exists, taking the dis- covered Higgs boson into account. Details about the different benchmark points should be taken from the pub- lications dedicated to this scenario [53, 54, 55, 56]. 25 [pb] + s=8 TeV tH β → -1 tan =30 pp 10 LHC HIGGS XS WG 2013 σ NLO, matched 10-2 matched 4FS 5FS 200 250 300 350 400 450 500 550 600 MH+ [GeV] Figure 2.11. Next to leading order (NLO) cross-section prediction for pp → tH− + X at the LHC with 8 TeV for a 2HDM with tanβ = 30: The Santander matching is used for these predictions. Shown is the central prediction together with an estimate of the theoretical uncertainties [51]. max The allowed mass of the light CP-even Higgs boson in the mh scenario is only in a small region of the mA − tanβ plane when taking into account the discovery of the SM-like Higgs boson at around 125 GeV [56], this is shown max in figure 2.12.Themh scenario was designed to maximize the value of mh, in the decoupling region this scenario yields mh values that are higher than the ones of the observed boson. max Due to this, a modification of the mh scenario was suggested by theo- mod reticians, called the mh scenario [56]. Two variants of this scenario exist mod+ mod− mh and mh , with a sign and absolute value difference on the parameter Xt /MSUSY ,whereXt is the stop mixing parameter. This scenario is now used in the latest ATLAS charged Higgs boson search [58]. A detailed description mod on the mh scenario, as well as other suggested benchmark scenarios can be found in [56]. Experimental constraints - direct searches For decades charged Higgs bosons have been searched for, at the colliders LEP [59], Tevatron [60] and LHC using the ATLAS [61, 62, 63, 58]and CMS [64, 65, 66] experiments. At tree level, the Higgs sector in the MSSM has only two free parameters, tanβ and one of the Higgs masses, see section 2.2.3. At loop level, due to a large number of MSSM parameters, there are additional contributions. In order to be able to present results in a two dimensional histogram, benchmark 26 β 80 -1 tan ATLAS s=8 TeV, ∫ L dt = 19.5 - 20.3 fb 70 MSSM mmax scenario, M = 1 TeV, h/H/A ττ→ h SUSY 60 Obs 95% CL limit Exp 95% CL limit 1 σ σ 50 2 =700 GeV H Obs 95% CL limit m ± σ 1 theory =500 GeV H 40 m mh = 130.2 GeV =300 GeV H 30 m =170 GeV m = 130 GeV H h 20 m = 122 GeV h m 10 m = 128 GeV h mh = 125 GeV 100 200 300 400 500 600 700 800 900 1000 mA [GeV] Figure 2.12. Expected (dashed line) and observed (solid line with markers) 95% CL β max upper limits on tan as a function of mA for the mh scenario of the MSSM. Values of tanβ above the lines are excluded. The vertical dashed line at 200 GeV indicates the transition point between low- and high-mass categories. Lines of constant mh and mH are also shown in red and blue colour, respectively [57]. scenarios are defined. These scenarios fix all parameters except tanβ and one of the Higgs boson masses. One often used scenario is the mh-max scenario, described before. Theoretical limits for the lowest possible mass of the charged Higgs boson are given by equation (2.18). Experimental exclusion limits are set by the LEP > . experiments at mH± 79 3 GeV at 95% confidence level independently of the branching ratios [67], assuming BR(cs¯)+BR(τν)=1. For BR(τν)=1theset limits are 87.8 GeV. For heavier charged Higgs bosons the best current limits are set by the ATLAS and CMS experiments. • ATLAS: < ± < For the mass range 80 GeV mH 160 GeV, 95% confidence level upper limits on B(t → H+b) are set in the range 0.23 − 1.3%, and for < ± < the mass range 180 GeV mH 1000 GeV, 95% confidence level up- per limits are set on the production cross-section of a charged Higgs boson in the range 0.0045 − 0.76 pb, both with the assumption that B(H± → τν)=1[58]. • CMS: In the mass range 80 GeV to 160 GeV,95% confidence level upper limits on B(t → H+b) are set in the range 0.16 − 1.2% and for the mass range < < 180 GeV mH± 600 GeV, 95% confidence level upper limits are set on the production cross-section of a charged Higgs boson in the range 0.026 − 0.38 pb, both with the assumption that B(H± → τν)=1[66]. 27 Figure 2.13 presents the limits on the branching ratio measured by ATLAS for light charged Higgs boson (a) and the limit on the cross-section for the heavy charged Higgs boson (b). Similar plots are shown for CMS in figure 2.14. max mod± In figure 2.15 the interpretation of the limits in the mh and mh scenar- ios from ATLAS are shown, figure 2.16 shows the similar plots for CMS. 1 ν + ATLAS Data 2012 ATLAS Data 2012 τ→ Observed CLs ) [pb] Observed CLs -1 ν 10 -1 ∫Ldt = 19.5 fb + ∫Ldt = 19.5 fb + Expected τ→ Expected H ± σ ± σ 10-1 1 s = 8 TeV 1 s = 8 TeV + B × ± 2σ 1 ± 2σ + B(H × + bH H -1 → -2 σ 10 t 10 B 10-2 10-3 -3 80 90 100 110 120 130 140 150 160 10 200 300 400 500 600 700 800 900 1000 mH+ [GeV] mH+ [GeV] (a) Light charged Higgs (b) Heavy charged Higgs Figure 2.13. The expected and observed 95% CL upper limits measured in ATLAS for the light (left) and heavy (right) charged Higgs boson searches [58]. The vertical axis shows the branching ratio (a) and cross-section (b) as a function of the charged Higgs boson mass. The dashed line represents the expected average limit with absence of a charged Higgs boson. The green and yellow bands indicate the corresponding 68% and 95% uncertainty of those values. The solid line represents the observed limit from data, the black dots indicate the mass points used for the measurement. 19.7 fb-1 (8 TeV) 19.7 fb-1 (8 TeV) 0.025 1 ντ→ + + + + + + CMS t → H b, H τ→ ντ CMS pp → (b)Ht , H τ→ ντ + Preliminary τ +jets final state (pb) 0.9 Preliminary τ +jets final state H h h ντ→ B 0.02 Observed 0.8 Observed × + b Expected median ± 1σ Expected median ± 1σ + H H Expected median ± 2σ 0.7 Expected median ± 2σ B → t × + B 0.015 H 0.6 σ 0.5 0.01 0.4 0.3 0.005 0.2 95% CL limit on 0.1 95% CL limit for 0 0 80 90 100 110 120 130 140 150 160 200 250 300 350 400 450 500 550 600 mH+ (GeV) mH+ (GeV) (a) Light charged Higgs (b) Heavy charged Higgs Figure 2.14. The expected and observed 95% CL upper limits measured in CMS for the light (left) and heavy (right) charged Higgs boson searches [66]. The vertical axis shows the branching ratio (a) and cross-section (b) as a function of the charged Higgs boson mass. The dashed line represents the expected average limit with absence of a charged Higgs boson. The green and yellow bands indicate the corresponding 68% and 95% uncertainty of those values. The solid line represents the observed limit from data, the black dots indicate the mass points used for the measurement. 28 β 60 β 60 Median expected exclusion tan tan 50 Observed exclusion 95% CL 55 ATLAS Observed +1σ theory σ -1 40 Observed -1 theory 50 ∫Ldt = 19.5 fb ATLAS s=8 TeV -1 30 ∫Ldt = 19.5 fb 45 Data 2012 max MSSM mh scenario 20 s=8 TeV 40 Median expected exclusion Data 2012 Observed exclusion 95% CL MSSM mmax scenario 10 h 35 Observed +1σ theory Observed -1σ theory 0 30 80 90 100 110 120 130 140 150 160 200 220 240 260 280 300 + + mH [GeV] mH [GeV] max max (a) Light charged Higgs mh (b) Heavy charged Higgs mh β 60 β 60 Median expected exclusion tan tan 50 Observed exclusion 95% CL 55 ATLAS Observed +1σ theory σ -1 40 Observed -1 theory 50 ∫Ldt = 19.5 fb ATLAS s=8 TeV -1 30 ∫Ldt = 19.5 fb 45 Data 2012 mod+ MSSM mh scenario 20 s=8 TeV 40 Median expected exclusion Data 2012 Observed exclusion 95% CL MSSM mmod+ scenario 10 h 35 Observed +1σ theory Observed -1σ theory 0 30 80 90 100 110 120 130 140 150 160 200 220 240 260 280 300 + + mH [GeV] mH [GeV] mod+ mod+ (c) Light charged Higgs mh (d) Heavy charged Higgs mh β 60 β 60 Median expected exclusion tan tan 50 Observed exclusion 95% CL 55 ATLAS Observed +1σ theory σ -1 40 Observed -1 theory 50 ∫Ldt = 19.5 fb ATLAS s=8 TeV -1 30 ∫Ldt = 19.5 fb 45 Data 2012 mod- MSSM mh scenario 20 s=8 TeV 40 Median expected exclusion Data 2012 Observed exclusion 95% CL MSSM mmod- scenario 10 h 35 Observed +1σ theory Observed -1σ theory 0 30 80 90 100 110 120 130 140 150 160 200 220 240 260 280 300 + + mH [GeV] mH [GeV] mod− mod− (e) Light charged Higgs mh (f) Heavy charged Higgs mh Figure 2.15. Interpretation of the limits on the branching fractions measured in ATLAS of the light charged Higgs boson (left) and the heavy charged Higgs boson max mod+ mod− (right), in the context of the MSSM mh (a, b), mh (c, d) and mh (e, f) scenar- ios [58]. 29 19.7 fb-1 (8 TeV) 19.7 fb-1 (8 TeV) β β 60 CMS 60 CMS Preliminary Preliminary tan tan + + + 50 t → H b, H τ→ ντ 50 τ +jets final state + + + h pp → (b)Ht , H τ→ ντ MSSM updated mmax τ +jets final state h h Observed MSSM updated mmax 40 40 h Observed ±1σ (th.) Observed 30 Excluded 30 Observed ±1σ (th.) Expected median ± 1σ Excluded Expected median ± 2σ Expected median ± 1σ mMSSM ≠ 125±3 GeV Expected median ± 2σ 20 h 20 MSSM mh ≠ 125±3 GeV 10 10 0 0 90 100 110 120 130 140 150 160 200 250 300 350 400 450 500 550 600 mH+ (GeV) mH+ (GeV) max max (a) Light charged Higgs mh (b) Heavy charged Higgs mh 19.7 fb-1 (8 TeV) 19.7 fb-1 (8 TeV) β β 60 CMS 60 CMS Preliminary Preliminary tan tan + + + 50 t → H b, H τ→ ντ 50 τ +jets final state h mod+ mMSSM mMSSM + + h + pp → (b)Ht , H τ→ ντ 40 Observed 40 τ +jets final state h mMSSM mMSSM mod+ Observed ±1σ (th.) h 30 Excluded 30 Observed Expected median ± 1σ Observed ±1σ (th.) Expected median ± 2σ Excluded mMSSM ≠ 125±3 GeV Expected median ± 1σ 20 h 20 Expected median ± 2σ mMSSM ≠ 125±3 GeV 10 10 h 0 0 90 100 110 120 130 140 150 160 200 250 300 350 400 450 500 550 600 mH+ (GeV) mH+ (GeV) mod+ mod+ (c) Light charged Higgs mh (d) Heavy charged Higgs mh 19.7 fb-1 (8 TeV) 19.7 fb-1 (8 TeV) β β 60 CMS 60 CMS Preliminary Preliminary tan tan + + + 50 t → H b, H τ→ ντ 50 τ +jets final state h mod- mMSSM mMSSM + + h + pp → (b)Ht , H τ→ ντ 40 Observed 40 τ +jets final state h mod- Observed ±1σ (th.) mMSSM h 30 Excluded 30 Observed Expected median ± 1σ Observed ±1σ (th.) Expected median ± 2σ Excluded MSSM ≠ 125±3 GeV 20 mh 20 Expected median ± 1σ Expected median ± 2σ mMSSM ≠ 125±3 GeV 10 10 h 0 0 90 100 110 120 130 140 150 160 200 250 300 350 400 450 500 550 600 mH+ (GeV) mH+ (GeV) mod− mod− (e) Light charged Higgs mh (f) Heavy charged Higgs mh Figure 2.16. Interpretation of the limits on the branching fractions measured in CMS of the light charged Higgs boson (left) and the heavy charged Higgs boson (right), in max mod+ mod− the context of the MSSM mh (a, b), mh (c, d) and mh (e, f) scenarios [66]. 30 Experimental constraints - indirect searches The charged Higgs boson enters measurements of some parameters at tree level or through loops. Small deviations from the SM predictions in precision measurements can be indications of charged Higgs bosons or BSM physics. Similarly precission measurements can set constraints on the charged Higgs boson. Exclusion limits can be obtained indirectly by studying flavour physics, measurement of the electron electric dipole moment or other precision mea- surements [68]. The set limits, though, are highly model dependent and cannot replace direct searches. What is excluded by one model can be allowed by an- other model. 3 Taking B → Xsγ as an example, which decays via b → sγ, a charged Higgs boson or SUSY particles could contribute via the penguin loop as shown in figure 2.17. This would change the branching fraction with respect to the SM-only scenario and can therefore be used to probe physics beyond the SM. Comparing the observed branching ratio with the calculated branching ratio from a certain model can give predictions on the reliability of the model. The combined average experimental limit determined by the Heavy Flavour Averaging Group [69]is: −6 BR(B → Xsγ)=(355 ± 24 ± 9) × 10 (2.22) In the paper [70], indirect mass constraints at 95% C.L. are set via B → Xsγ, excluding charged Higgs bosons in the 2HDM type II up to 295 GeV. q˜ t t ± − bsW bsH± bsχ˜ (a) Standard Model (b) charged Higgs (c) charginos, squarks Figure 2.17. Contribution to b → sγ transition in the SM (a), from a charged Higgs boson (b) and from charginos and squarks (c). The loose photon is to be attached in all possible positions. Further constraints which can be used in indirect searches are described in the following: • Bu → τντ : This process can be mediated at tree level by a charged Higgs boson. It is helicity suppressed in the SM. Measurements have been performed by the Belle and BaBar experiments: ( + → τ+ν )=( . +0.53 ± . ) × −4 [ ] BR B τ 1 83−0.49(stat.) 0 24(sys.) 10 71 (2.23) ( − → τ−ν )=( . +0.38 +0.29 ) × −4 [ ] BR B ¯τ 1 54−0.37(stat.)−0.31syst. 10 72 (2.24) 3 + = ¯ 0 = ¯ 0 = ¯ + = ¯ ∗ B-mesons: B ub, B db, Bs sb, Bc cb similarly for B ’s [1], a possible subscript indicates the quark content. 31 • B → Dτντ 4: The ratio between B → D ν ,when is either an electron or a muon, and B → Dτντ is sensitive to new physics. For example contributions from the charged Higgs boson in models with BSM physics. The semileptonic decay B → D ν depends on the CKM matrix element |Vcb| which is known to a good precision [68]. The ratio was measured by the BaBar collaboration [73]. BR(B → Dτντ) R(D)= =(0.440 ± 0.058 ± 0.042) (2.25) BR(B → D ν ) The ratio for R(D∗) was measured to be 0.332 ± 0.024 ± 0.018. The combination of R(D) and R(D∗) excludes the type II 2HDM charged β/ Higgs boson with a 99.8% confidence level for any value of tan mH+ > < above mH+ 10 GeV [73]. Values for mH+ 10 GeV are excluded by B → Xsγ. • Ds → τντ : The combined average experimental limit determined by the Heavy Flavour Averaging Group [69]is: −2 BR(Ds → τντ)=5.44 ± 0.22 × 10 (2.26) + − • Bd,s → μ μ : These decays can be enhanced or supressed from Higgs- mediated contributions, they are helicity suppressed in the SM. Limits on the branching fraction were set by a combination of the ATLAS, CMS and LHCb experiment [74]: + − −9 BR(Bs → μ μ ) < 4.2 × 10 , (2.27) + − −10 BR(Bd → μ μ ) < 8.1 × 10 . (2.28) In the CMSSM (Constrained Minimal Supersymmetric Model) [75, 76], as an example, certain areas in the parameter space can be excluded by the previously mentioned constraints [77]. Figure 2.18 shows constrained areas − β in the mH+ tan plane. ∗ The 2HDM type-II model is excluded by B → Dτντ and B → D τντ mea- surements because they can not be explained in this model simultaneously [78]. Charged Higgs bosons though, still can exist in other models that have more parameters. 4 + = ¯ 0 = + = ∗+ = ¯ ∗0 = D-mesons: D cd, D du¯, Ds cs¯, D cd (different parity), D du¯ (different parity) [1], a possible subscript indicates the quark content. 32 Allowed Direct b→ s γ ντ→ Bu μ→ + μ- Bs B → D ντ K νμ→ Figure 2.18. Combined exclusion in CMSSM models by different constraints. The constraints are applied in the order they appear in the legend, and the colour coding corresponds to the first constraint by which a point is excluded [77]. 33 3. The CERN laboratory The world largest laboratory for particle physics CERN1 (Conseil Européen pour la Recherche Nucléaire), is located to the north-west of Geneva on the French-Swiss border. It was founded on the 29th of September 1954. Today, CERN has twenty-one member states and several hundred universities world- wide participating in the research conducted there. At the beginning CERN was a laboratory for nuclear research but the energy frontier quickly moved on to higher energies to study more fundamental particles. CERN is the provider of the particle accelerators and other main infrastructure needed to push the high energy frontier. The experiments at the accelerators are constructed by international collaborations. Although mostly associated with particle physics, CERN also contributes to achievements in computing science and engineering science. For example, the World Wide Web was invented at CERN. 3.1 Scientific achievements made by CERN CERN has played an important role for our understanding of fundamental forces and basic constituents of matter. Its research has been awarded with several Nobel prices [79]. Some of the major achievements are: • 1973: The discovery of neutral currents [80, 81, 82]; • 1983: The discovery of the W and Z bosons [12, 13, 14, 15]; • 1989: The determination of the number of light neutrino families [5]; • 1995: The first creation of anti-hydrogen atoms [83]; • 1999: The discovery of direct CP-violation [84]; • 2012: The observation of a Higgs boson at 125 GeV [6, 7]; 3.2 The accelerator infrastructure CERN is the provider of the acceleration and de-acceleration infrastructure, producing particles at different energies for the experiments. The interplay between the different accelerator stages at CERN is shown in figure 3.1. The various accelerators in the LHC complex are listed and described in the following: 1the European Organization for Nuclear Research 34 Figure 3.1. CERN accelerator complex [85]. The accelerators relevant for physics at the LHC are described in the text. 35 3.2.1 Linac-2 The Linac-2 [86] linear accelerator, accelerates protons after the ionization of hydrogen to an energy of 50 MeV with a pulse length of 1 Hz and a current of 175 mA. Its main purpose is to be a pre-accelerator for the Proton Synchrotron Booster (PSB). 3.2.2 Linac-3 The Linac-3 [87] linear accelerator is a second linear accelerator at CERN, for heavy ions instead of protons. After striping off all electrons, it accelerates ions to an energy of 4.2 MeV/nucleon. Subsequently, the ions are injected into the Low Energy Ion Ring (LEIR). 3.2.3 Linac-4 The Linac-4 [88] is a future linear proton accelerator, that will replace Linac-2 in 2017/2018. Its construction was approved in 2007 and it is needed in order to provide enough protons to be able to reach the design luminosity of the LHC. Protons will be accelerated to an energy of 160 MeV. 3.2.4 Proton Synchrotron Booster (PSB) The Proton Synchrotron Booster [89] is one of the pre-accelerators at CERN. It receives the protons from Linac-2 and accelerates them to an energy of 1.4 GeV. The protons are subsequently injected into the Proton Synchrotron or delivered to Isolde [90] to produce radioactive ion beams. 3.2.5 Proton Synchrotron (PS) The Proton Synchrotron [91] is a key accelerator at CERN. It gets protons from the PSB or ions from the LEIR and accelerates them to higher energies. Protons can reach energies up to 28 GeV and ions up to 72 MeV/nucleon. The Proton Synchrotron acts as a pre-accelerator for the Super Proton Synchrotron. Its protons can also be used for anti-proton production for the Anti-proton De- accelerator (AD), production of neutrons and beam delivery to the East Area of CERN. 3.2.6 Super Proton Synchrotron (SPS) The Super Proton Synchrotron is the second largest accelerator at CERN. It is 6.9 km in circumference and can accelerate protons to energies up to 450 GeV. Before the LHC era, it was used as the proton-anti-proton collider for the UA1 36 and UA2 experiments. Today it acts as a pre-accelerator for the LHC, but it also has delivered protons for the neutrino beam to Gran Sasso (CNGS) and it provides a proton beam for the fixed target experiments in the North Area of CERN. There are plans to further upgrade the SPS to a Super-SPS in order to reach the design luminosity in LHC. 3.3 The Large Hadron Collider (LHC) The Large Hadron Collider [92] provides the highest collision energy in the world, required for precision studies of the Standard Model and to probe new physics beyond it. The LHC is built in the former Large Electron Positron collider (LEP) tunnel with a circumference of 27 km, which allows operation at a centre-of-mass energy of up to 14 TeV for proton-proton collisions and a luminosity of 1034 cm−2s−1. In addition to proton-proton collisions, the LHC provides heavy ion collisions for studies of quark-gluon plasma. More than 10 000 scientists and engineers from hundreds of universities are using the beams of the LHC for their research. This work is done by four collabora- tions ALICE, ATLAS, CMS, LHCb, each with an experimental facility at four points along the LHC ring, the locations are shown in figure 3.2. Figure 3.2. Schematic of the LHC ring, with the positions of the experiments [92, 93]. The start-up of the LHC was on the 10th of September 2008, but was stopped after a short time due to a major accident with the magnet system. On the 20th of November 2009, the LHC restarted and successfully collided 37 protons after a few days. The first collisions were at an energy per beam of 450 GeV, the injection energy of the SPS. The beam energy was increased in steps and after the winter shut-down in 2009/2010, the energy per beam reached 3.5 TeV, half of the design energy of the LHC. In 2012 a further increase to 4 TeV per beam was achieved. More than 5 fb−1 of integrated luminosity were recorded at a centre-of-mass energy of 7 TeV and more than 21 fb−1 of data with a centre-of-mass energy of 8 TeV. At the end of 2012, the LHC was shut down for maintenance and is ready to restart in 2015 with a collision energy of 13 TeV. 3.4 The four experiments at the LHC ALICE (A Large Ion Collider Experiment) [94] is a heavy ion experiment designed to study the conditions of the universe directly after the Big Bang. Heavy ion runs are used to produce temperatures 100 times higher than inside the sun to produce quark-gluon plasma. The expansion and cooling down of this plasma is studied in order to understand how the universe was formed. CMS (Compact Muon Solenoid) [95] is in size the second biggest experi- ment at the LHC. It is a general purpose detector, designed for searching the Higgs boson, for supersymmetry, dark matter and for precision measurements of the Standard Model. The detector is built inside a big 4 T solenoid magnet. CMS was the only detector at the LHC built over ground and installed in 15 slices in the CMS cavern. The detector system has a length of 21 m, a width of 15 m and a height of 15 m. It weighs 12500 t. LHCb (Large Hadron Collider beauty) [96] is a detector system designed to study the B-mesons in detail. The focus is the study of the matter-antimatter symmetry and rare decays of B-mesons, which are sensitive to new physics beyond the Standard Model. It is the smallest detector at the LHC build as a single arm spectrometer detecting particles propagating in the forward direc- tion. The experiment does not use the full luminosity of the LHC to avoid pile-up2 events. ATLAS (A Toroidal LHC ApperatuS) [98] is the largest detector at the LHC. It is like CMS a general purpose detector with the same physics goals. More details about the ATLAS detector system, which was used to collect the data used in this thesis, are described in the next chapter. 2pile-up: In high-luminosity colliders, there is a non-negligible probability that one single bunch crossing may produce several separate events, so-called pile-up events. [97] 38 4. The ATLAS detector system ATLAS [98] is a forward-backward symmetric detector with respect to the interaction point, as shown in figure 4.1. It is designed to handle the high interaction rates, radiation doses, particle multiplicities and energies at the LHC. The detector system is the result of over 15 year long collaboration between several thousands of physicists, en- gineers, technicians and students. ATLAS offers a rich set of opportunities for new physics, high precision measurements of the Standard Model and the possibility to study strongly interacting matter at high energy densities. For example top quarks [99, 100] will be produced at large rates making study of its spin and couplings possible. The search for the Standard Model Higgs boson was one of the strongest physics cases driving the requirements of the sub-systems of ATLAS. These are described in [101, 102] and summerized in table 4.1. Table 4.1. General performance goals of the ATLAS detector [101, 102]. The units for E and pT are in GeV. Detector component Required resolution η covrage Measurement (Trigger) σ / = . ⊕ ± . Tracking pT pT 0 05%√ pT 1% 2 5 EM calorimetry σE /E = 10%/ E ⊕ 0.7% ±3.2 (±2.5) Hadronic calorimetry (jets) √ σ / = / ⊕ ± . (± . ) barrel and end-cap E E 50% √E 3% 3 2 3 2 forward σE /E = 100%/ E ⊕ 10% 3.1 < |η| < 4.9 (3.1 < |η| < 4.9) σ = = ± . (± . ) Muon spectrometer pT 10% @pT 1TeV 2 7 2 4 4.1 Conventions Conventions used in ATLAS are: • The nominal interaction point of ATLAS is defined as the origin of the coordinate system. • The z-axis is defined by the beam direction, the xy-plane is transverse to the z-axis. • The positive x-axis is pointing from the interaction point to the centre of the LHC ring. • The positive y-axis is pointing upwards. • The azimuthal angle φ is measured around the beam axis. 39 Figure 4.1. View of the ATLAS experiment with its sub-detectors [98]. 40 • The polar angle θ is measured from the beam axis. • η η = ( θ ) The pseudo-rapidity is defined as lntan 2 . • Transverse quantities like transverse momentum pT , transverse energy miss ET and missing transverse energy ET are defined in the xy-plane. • The distance ΔR in pseudo-rapidity-azimuthal angle space is defined as ΔR = Δη2 + Δφ 2. • The 4-momentum vector P =(E, px, py, pz) can be expressed in terms of E,η,φ and pT by P =[E, pt cos(φ), pt sin(φ), pt sinh(η)] for massless particles. 4.2 The inner detector Physics measurements in ATLAS require high momentum and vertex resolu- tions, hence a high granularity to handle the large track density. The highest granularity around the vertex region is achieved using semiconductor pixel detectors, followed by a SemiConductor Tracking (SCT) detector. The to- tal number of precision layers is limited by the material and power dissipa- tion they introduce, and the cost factor. A large number of tracking points is required for reliable pattern recognition and is achieved by a straw tube tracker/Transition Radiation Tracker (TRT) which provides the possibility of continuous track-following with much less material per point and lower costs. Additionally the electron identification and charged pion rejection capabili- ties are enhanced by the detection of transition radiation photons in the straw tubes. Further jet rejection of a factor ∼ 40 at an efficiency of better than 90% for pT > 20 GeV electrons were design parameters [103]. The inner detector plays a major role for electron identification, for γ/π0 separation in case of photon conversions, important to reduce the amount of misidentified leptons. The stand-alone performance of the inner detector is of importance for the identification of jets from b-quarks (b-tagging) used for example in the charged Higgs boson searches. All discrimination and pattern recognition need to work under the pile-up conditions occurring when running at LHC design luminosity. An overview of the inner detector is shown in figure 4.2. The inner detector is surrounded by a 2 T solenoid magnet with a length of 5.3 m and a diameter of 2.5 m. The pixel detector and the SCT, have concentric cylindrical layers in the central part and disc layers perpendicular to the beam in the end-cap, which offer a precision tracking up to |η| < 2.5. With over 80 million readout channels, the pixel detector is the most granular detector in ATLAS, capable to reconstruct the secondary vertex of b-quark decays. The SCT has around 6.3 million readout channels. The TRT system covers a pseudo-rapidity of up to |η| < 2.0 and has approximately 351000 readout channels. The combination of theses three systems provides a very robust pattern recognition and high precision in both R − φ and z coordinates. The σ / = . ⊕ design momentum resolution is pT pT 0 05%pT 1% [102]. 41 Figure 4.2. Overview of the inner detector with subsystems indicated [98]. 4.3 The calorimeters A calorimeter is an instrument which measures the energy of particles and their position by absorbing these particles. Characteristic interactions with matter, e.g. atomic excitation or ionization, are used to generate detectable effects. In this way, electrically charged particles but also electrically neutral particles can be detected as opposed to the inner detector, which only detects charged objects. Calorimeters are made from cell structures, aligned in form of towers, along the direction of interaction. The energy of the particles or jets is determined by integrating over the whole volume of interaction. Calorime- ters as a whole usually contain an electromagnetic system to absorb electrons and photons, followed by a hadronic system to absorb hadrons. The hadrons start their showering in the electromagnetic calorimeter, but only the hadronic calorimeters can fully stop them. Critical performance parameter for the design of the calorimeter in ATLAS are the energy resolution, the accuracy of the position and angular measure- ments, the ability to reconstruct e.g. the missing transverse energy with high precision, and the reconstruction of τ-leptons [104]. The capability of separa- tion γ/π0, γ/jet and e/jet are important [103] in order to reduce misidentifica- tion rates and the resulting backgrounds. Different technologies are used in ATLAS to build a hermetic calorimeter covering the region |η| < 4.9, as shown in figure 4.3.Intheη region of the inner detector the electromagnetic calorimeter has a fine granularity (segmen- tation: Δη ∼ 0.025 − 0.1andΔΦ = 0.025 − 0.1[102]), allowing to measure electrons and photons with high precision. Outside the inner detector region, 42 Figure 4.3. Overview of the calorimeter with differnetsegments indicated [98]. the granularity of the calorimeters gets coarser, but still fine enough to recon- struct jets and missing transverse energy, important for many physics signa- tures, for example SUSY. Calorimeters must have sufficient stopping power for protecting the muon system from showers punching through. The Electromagnetic Calorimeter: The electromagnetic calorimeter (EM) is a lead-liquid-argon sampling calorime- ter, made of a barrel part (|η| < 1.475) and two end-cap parts (1.375 < |η| < 3.2). The EM calorimeter has an accordion geometry in order to cover the complete azimuthal angle φ without any azimuthal crack. 1 The total thickness of the EM calorimeter is > 22 radiation lengths (X0)inthe√ > 2 ΔE ≤ / barrel and 24 X0 in the end-caps. The energy resolution is: E 10% E (stochastic term) with a constant term < 0.7%. The Hadronic Calorimeters: Three subsystems are used for hadronic calorimetry, based on different tech- nologies in order to provide the best performance. TheTilecalorimeteris a sampling calorimeter situated directly outside the 1Radiation lengths is a material characteristic, describing the energy loss of high energy, electromagnetic-interacting particles in the material. 2Stochastic term: statistical fluctuations, e.g. from the shower, light yield, sampling. Con- stant term, dominant at high-energy: e.g. cell-to-cell calibration inaccuracies, intrinsic non- uniformities, radiation damage. The units for E are in GeV. 43 EM calorimeter. It uses steel as an energy absorber and scintillating tiles as active material. The barrel covers |η| < 1.0 and is completed by two extended barrels covering√ a range from 0.8 < |η| < 1.7 each. The energy resolution is: ΔE ≤ / ⊕ E 50% E 3%. The Hadronic end-cap calorimeter is a sampling calorimeter made of two wheels per end-cap, situated directly behind the end-cap of the EM calorime- ter. For the absorbing material, 25 mm parallel copper plates are used in the wheels which are closest to the interaction point, the other wheels are made from 50 mm copper plates. The copper plates have small gaps, providing space for the√ liquid argon as an active material. The energy resolution is: ΔE ≤ / ⊕ E 50% E 3%. The Forward calorimeter is another sampling liquid argon calorimeter. It is built with three modules per end-cap, the first uses copper for absorbing and is sensitive to electromagnetic showers. The other two layers use tung- sten and measure√ the energy of hadronic showers. The energy resolution is: σE /E = 100%/ E ⊕ 10%. 4.4 The muon spectrometer In the ATLAS muon system, the momentum of muons is determined in a large superconducting air-core toroid magnet system. It contains detector technolo- gies for fast triggers and high precision tracking with ∼ 60 μm intrinsic reso- lution. Emphasis is put on reliable, high resolution, stand-alone performance over a pT range from 5 GeV to 1000 GeV [103]. Good momentum resolu- tion is needed, above large backgrounds, for the detection of decays involving muons. The resolution can be futher improved by combining the stand alone results of the muon spectrometer with these of the inner detector. An example are light charged Higgs boson seaches, where the decay product has one or two leptons. A schematic view of the muon system is shown in figure 4.4. A large central toroid magnet bends the muons in the region |η| < 1.4. In the high-η regions (1.6 < η < 2.7), two smaller toroidal superconducting magnets are inserted. Each of the three magnets is made from eight coils which are placed symmetrically around the beam axis. A magnetic bending power of 1.4 to 5.5 Tm is provided by the barrel toroid in the range |η| < 1.4. In the end-cap region, a bending power of 1 to 7.5 Tm is provided by the end-cap magnets. The region 1.4 < |η| < 1.6 is called the transition region, where the magnetic fields overlap. Muon chambers arranged in three cylindri- cal layers around the beam pipe are used in the barrel region to measure muon tracks. The transition and end-cap regions are covered by three layers of muon chambers arranged along φ, meaning the centre point of the tubes are oriented tangential to the circle around the beam axis. 44 Figure 4.4. Overview of the ATLAS muon system [98]. The muon chambers The region with |η| < 2.7iscoveredbyMonitored Drift Tubes (MDT). In the high-η region the rates and background conditions are expected to be up to 30 kHz per tube at full LHC luminosity. The MDTs, which are filled with Ar/CO2 gas, have a diameter of 29.97 mm providing an average resolution of 60-80 μm. The maximal count rate of the MDT is 150 Hz/cm2. Cathode Strip Chambers (CSC) are used in the region 2 < |η| < 2.7where the particle flux and the density of tracks are highest. The CSC can handle counting rates up to 1000 Hz/cm2. The CSC system is made of two disks of eight chambers, which enable the measurement of η and φ coordinates. The CSC is filled with an Ar/CO2 mixture, the resolution reached is 60 μm. Trigger chambers provide fast information to the Level 1 trigger logic about muon transverse momentum and bunch-crossing identification. The trigger chambers measure the φ coordinate to give a full space trajectory in combina- tion with the η coordinate of the MDTs. The region |η| < 2.4iscoveredusing two different types of technologies. In the barrel region |η| < 1.05, Resistive Plate Chambers (RPC) are used and, in the region 1.05 < |η| < 2.4, Thin Gap Chambers (TGC) are used. 45 4.5 The trigger system The ATLAS detector has a three-level trigger system: Level 1 (L1), Level 2 (L2) and event filter (EF) level. The levels are staged, meaning that the next level refines the event selection from the previous level by applying more pre- cise or additional requirements. The Level 1 trigger uses high transverse momentum muons, electrons, pho- tons, jets, τ-leptons, large missing and total transverse energies for triggering. Only reduced detector information is used in order to make a decision with a latency shorter than 2.5 μs. The rate is reduced from the Beam Cross Over (BCO) frequency of 40 MHz to 75 kHz (to be increased to 100 kHz in Run 2). The L1 trigger defines Regions Of Interest (ROI) that seed the L2 trigger. At L2, the full precision of all detector data in the ROI is used. The L2 trigger is configured to reduce the data rate to approximately 3.5 kHz with a latency less than 40 ms per event. At the EF level the data rate is further reduced to about 200 Hz for storage and offline analysis. The EF uses advanced near offline quality algorithms for processing and requires around 4 seconds per event. The stored data size is approximately 1.3 Mbyte per event [102]. A schematic layout of the trigger system is shown in figure 4.5. Interaction rate ~1 GHz CALO MUON TRACKING Bunch crossing rate 40 MHz Pipeline LEVEL 1 memories TRIGGER < 75 (100) kHz Derandomizers Regions of Interest Readout drivers (RODs) LEVEL 2 Readout buffers TRIGGER (ROBs) ~ 1 kHz Event builder EVENT FILTER Full-event buffers and ~ 100 Hz processor sub-farms Data recording Figure 4.5. Schematics layout of the trigger system [105]. 46 5. Physics analysis in ATLAS 5.1 Charged Higgs boson searches in ATLAS One major task of the LHC and the ATLAS experiment is the exploration of the Higgs sector and physics beyond the SM. Several theories beyond the SM predict the existence of more than one Higgs boson (see chapter 2.2.3). The observation of a charged Higgs boson would clearly indicate physics beyond the SM. At the start up of the LHC, the upper limits on a light charged Higgs + ( + < ) ( → ) boson mH mtop , in terms of the branching fraction B t bH ,were of the order of 15-20%, as set by the Tevatron [60, 106]. The LEP experi- > . ments had excluded a charged Higgs boson below mH+ 79 3 GeV at 95% confidence level [67]. The light charged Higgs boson can be produced in several extensions of the SM through the decay of a top quark into a charged Higgs boson and a bottom quark. For tanβ > 2, where tanβ is the ratio of the vacuum expectation values of the two Higgs doublets, the charged Higgs boson is assumed to predomi- nantly decay into τν. In the region 1 < tanβ < 2, this contribution remains sizeable. 5.1.1 Measurement of discriminating variables for charged Higgs boson searches The first publication related to charged Higgs boson searches√ in ATLAS, Pa- per VI, was released after analysing 35 pb−1 of data at s = 7TeVandit summarized results of a first measurement of discriminating variables [107]. θ ∗ One of the variables was cos l defined as [108]: 2m2 4pb · pl cosθ ∗ = bl − 1 (5.1) l 2 − 2 2 − 2 mtop mW mtop mW b l 2 b l with p · p = 2EbEl(1−cosθbl)=4EbEl sin (θbl/2),wherep and p are the 4-momenta of the lepton and the b-quark coming from the same top quark. The variable mbl is the invariant mass of the b-quark and the lepton, and θbl is the angle between them. If the top quark decays into a charged Higgs boson, instead of a W boson, the b-quark gets a smaller momentum. Also, a charged lepton arising from a leptonically decaying τ is likely to have a smaller momentum, compared to 47 a lepton coming directly from a W boson. The presence of a charged Higgs boson in a leptonic top quark decay therefore strongly reduces the invariant · θ ∗ − product pb pl and has values of cos l close to 1 as a consequence. This variable is able to discriminate between the lepton coming from τ → l + νs and a lepton coming directly from the W boson. In this first analysis, two channels were considered: • One top quark decays hadronically and the other leptonically (single- lepton analysis), • both top quarks decay leptonically (di-lepton analysis). θ ∗ Since cos l does not give any information on the charged Higgs boson H H mass, two transverse masses mT (single-lepton analysis) and mT2 (di-lepton analysis) were defined [109]. These three variables were measured in the first ATLAS data from 2010 and a good overall agreement between data and Monte Carlo simulations was observed. The amount of data at this time was not enough to draw any con- clusions about whether or not there is a charged Higgs boson. 5.1.2 Charged Higgs boson searches in the single- and di-lepton channels The search in the single- and di-lepton channels was pursued in Paper VII and the first limits were set with 1.03 fb−1 of data in 2011, when the LHC was θ ∗ running at 7 TeV [110]. With the help of the discriminating variables cos l , H H ( → +) mT and mT2, upper limits on the branching fraction B t bH between 5.2% and 14.1% were set in the H+ mass range from 90 GeV to 160 GeV. The branching fraction of a charged Higgs boson decaying into τν was assumed β max to be 100%. Limits on tan were set in the mh scenario of the MSSM. Values of tanβ larger than 30 - 40 were excluded in the H+ mass range of 90 GeV to 140 GeV. This analysis used the excellent lepton identification of the ATLAS detector, by triggering and selecting at least one isolated lepton. In order to estimate the amount of non-isolated leptons, which appear as a background in this analysis, a data-driven method was used. Non-isolated leptons arise from the semileptonic decay of b-andc-quarks and from decay in flight of π± or K-mesons. In case of misidentified electrons, they also arise from the reconstruction of a π0, photon conversion or shower fluctuations. All leptons coming from such mechanisms are referred to as misidentified leptons, as opposed to true isolated real leptons coming from e.g. W and Z decays. A data driven method, the Matrix Method, was used to estimate the contribution of this background, see section 5.2 for details. Other background contributions were modelled using Monte Carlo simulations. With more and more data collected by ATLAS it became clear that the di- lepton channel would not give enough sensitivity for further charged Higgs θ ∗ boson searches and this analysis was discontinued. The cos l distribution for 48 the H+ side in di-lepton events as well as the combined limit from the single- and di-lepton channels are shown in figure 5.1 600 -1 ATLAS Preliminary ∫Ldt = 1.03 fb 500 + mH+ = 130 GeV tt (with H ) + Br(t → bH ) = 10% tt (W+W-) Events / 0.2 400 Single top 0.4 + ATLAS Preliminary Z + jets bH 0.35 Data 2011 s = 7 TeV → Observed CLs 300 Diboson t 0.3 Expected -1 ± 1σ ∫Ldt = 1.03 fb QCD & W ± 2σ 0.25 200 Data 2011 0.2 0.15 100 0.1 0.05 0 -1 -0.5 0 0.5 1 95% C.L. upper bound on Br 0 cosθ*, H+ side 90 100 110 120 130 140 150 160 mH+ [GeV] θ ∗ (a) cos l distribution. (b) Combined limit from the single- and di- lepton channels. θ ∗ + Figure 5.1. a) Reconstruction of cos l on the H side of the di-lepton events, in ATLAS data and Monte Carlo simulations. b) Upper limits on B(t → bH+) for the combined single-lepton and di-lepton channels, as a function of the charged Higgs boson mass, obtained for an integrated luminosity of 1.03/fb and with the assumption that B(H+ → τν)=100%. This paper [110] is not printed in the thesis, which is why the plots are presented here. 5.1.3 Combination of the most sensitive channels for charged Higgs boson searches In 2012 a paper on charged Higgs boson searches was published using the full 2011 dataset of 4.6fb−1 collected by ATLAS at 7 TeV [63]. This paper combines searches based on the following decay channels1: + • lepton + jets: tt¯ → bbW¯ H → bb¯(qq¯ )(τlepν) + • τ+lepton: tt¯ → bbW¯ H → bb¯(lν)(τhadν) (with l = e,μ) + • τ+jets: tt¯ → bbW¯ H → bb¯(qq¯ )(τhadν) + θ ∗ H In the lepton jets, channel the discriminating variables cos l and mT were used to discriminate between charged Higgs bosons and W bosons. In the τ+lepton and the τ+jets channels, the discriminating variables are the missing miss 2 transverse energy ET and the transverse mass mT . Limits on the branching fraction B(t → bH+) were set between 5% and 1% in the mass range from 90 GeV to 160 GeV. This was a significant improve- ment over the whole charged Higgs boson mass range, compared to existing 1τ τ lep: leptonically decaying tau, had: hadronically decaying tau 2 = l miss( − φ ) mT pT ET 1 cos l,miss 49 limits provided by the Tevatron experiments, in particular for H+ masses close to the top quark mass. These combined exclusion limits were dominated by the exclusion power of the τ+jets channels, especially in the upper H+ mass range. The τ+lepton channel has a similar exclusion power compared to the τ+jets channel for low charged Higgs boson mass but a much worse exclu- sion power for charged Higgs boson masses close to the top quark mass. The lepton+jets channel lies somewhere in between with better exclusion limits than τ+lepton in the upper mass region, but lower compared to τ+jets and, for masses close to the W boson mass, lower limits in general [111, 112, 113]. max An interpretation of these limits was done, in the mh scenario, thereby excluding tanβ above 12-26 as well as between 1 and 2-6 for charged Higgs boson masses between 90 GeV and 150 GeV. The Matrix Method was again used to estimate the amount of misidentified leptons in the lepton + jets and τ+lepton channels. Other background contributions were determined from simulation, data driven methods or a combination of both. 5.1.4 Charged Higgs boson searches through the violation of lepton universality The following paper published by ATLAS reported on a new search for charged Higgs bosons in the τ+lepton channel through the violation of lepton univer- sality in tt¯ events [62]. Here the 4.6 fb−1 of 7 TeV data were re-analysed. As- suming that a charged Higgs boson is produced in top quark decays, and that H+ predominantly decays into a τ−lepton and a neutrino, more final states with taus would be found in tt¯ decays compared to the absence of a charged Higgs boson (in the SM, the top quark decays into a W boson which, with equal probability, decays into the three lepton generations). The search was performed in events with a final state consisting of e/μ + τhad.Ratiosoftt¯ event yields between e+τhad and e+ μ, as well as between μ +τhad and μ +e events, were calculated and compared with simulation. BR(tt¯ → bb¯ + lτ + Nν) R = had (5.2) l BR(tt¯ → bb¯ + ll + Nν) This ratio-based method reduced the impact of common systematic uncertain- ties between the numerator and the denominator. With this method, upper limits on the branching fraction B(t → bH+) between 3.2% to 4.4% could be set for charged Higgs boson masses in the range from 90 GeV to 140 GeV. A combination of the results from the published τ+jets analysis and the ratio method yields upper limits on B(t → bH+) from 0.8% to 3.4% in the H+ mass max range of 90 GeV to 160 GeV. Also, the exclusion in the mh scenario could be further improved. The Matrix Method was used in order to estimate the contribution of misidentified leptons to the backgrounds. 50 5.1.5 Charged Higgs boson searches with the 2012 dataset Charged Higgs boson searches were then pursued with the 2012 ATLAS data at 8 TeV and extended to charged Higgs boson masses beyond the top quark mass. The latest ATLAS publication on charged Higgs boson searches at the time of writing is the τ+jets analysis [58]. The corresponding limits were presented in section 2.2.4, in figure 2.13. 5.2 The Matrix Method A significant background in charged Higgs boson searches, involving one or two isolated leptons, consists of events with reconstructed electrons and muons arising from the semileptonic decay of hadrons with b-orc-quarks, from the decay-in-flight of π± or K-mesons and, in the case of electrons, from π0 mesons, photon conversions or shower fluctuations. Such events are dif- ficult to estimate from simulation. The probability that the leptons detected in such events are actually misidentified by the reconstruction software is very small and therefore huge amounts of events would have to be simulated, which in turn would use large amounts of computing resources. Also the modelling of lepton isolation in simulation is very difficult to match with reality. A better approach is therefore to estimate misidentified electrons and muons by data driven methods. One of them is the so-called Matrix Method, a method already used at the Tevatron [114]. This data driven method exploits the dif- ference in the lepton identification between real, prompt, and misidentified or non-prompt electrons and muons. It is based on the selection of two cat- egories of events using loose and tight lepton selection requirements while keeping the same kinematic selection as the baseline analysis. The tight selec- tion is the one used in the baseline analysis and contains mainly real leptons. The loose selection has more relaxed isolation requirements and is enriched with misidentified and non-prompt leptons. The tight sample is a subset of the loose sample. In a data sample containing events with a single lepton, the number of events with one tight lepton (Nt ) and the number of events with one loose lepton (Nl) can be expressed as linear combinations of the number of events with a real (r) or a non-prompt, misidentified lepton ( f ): l = l + l , N Nr Nf (5.3) t = t + t . N Nr Nf (5.4) t l ε The ratio between Nr and Nr defines the efficiency r that a real lepton in the loose sample would pass the tight selection criteria. In the same way, the efficiency that a misidentified lepton passes the tight selection criteria is 51 Nt ε = f defined as f l . This leads to the following equations: Nf t = ε l, Nr rNr (5.5) t = ε l , Nf f Nf (5.6) which can be combined together with equation (5.4): t = ε l + ε l . N rNr f Nf (5.7) By solving the equation system given by (5.3)and(5.7), and by using equation (5.6), the number of loose leptons in the tight sample can be estimated by: ε t = f (ε l − t). Nf rN N (5.8) εr − ε f The efficiencies εr and ε f need to be known, as well as the total number of loose and tight events. The efficiencies are estimated in regions enriched with misidentified or real leptons. Not only the total number of misidentified lep- tons needs to be estimated, but also the shapes of various distributions need to be described correctly. For this purpose, equation (5.8) can be generalized into a weight: ε f ωi = (εr − δi) (5.9) εr − ε f where δi is equal to 1 if the event i passes the tight event selection and 0 otherwise, and εr and ε f may depend on the properties of the event i.The weights are built in such a way that: ω = t . ∑ i Nf (5.10) i∈loose data The Matrix Method can also be used for di-lepton events. For further details, see Paper III. 5.2.1 The Matrix Method related to top quark and charged Higgs boson physics in ATLAS Even though the Matrix Method has already been used for some time, it needed further development through out the LHC Run 1. The main reasons were the change of conditions, e.g. more pile-up, but also different require- √ments at the trigger level. The data collected during 2011 had an energy of s = 7 TeV. The pileup conditions during that time were significantly differ- ent in the beginning and the end of the data taking. Also, a better understand- ing of the detector changed the recommendations regarding object definitions. Therefore, the efficiencies εr and ε f had to be recalculated and monitored at 52 all times. But also the parametrization, which was used to calculate the event weight, had to be adjusted at all times in order to provide the best possible description of this background. At the beginning of the 7 TeV run, the trigger menues contained no isolation requirements and the threshold was 20 GeV for electrons and 18 GeV for muons. Later in the 7 TeV run, the threshold of the electrons had to be increased to 22 GeV. When moving to 8 TeV in 2012, isolation criteria were included at the trig- ger level for electrons and muons. Furthermore, a combination of two triggers, with a threshold of 24 GeV and 60 GeV for electrons, and 24 GeV and 36 GeV for muons, were used. This required major changes in the parametrization of the Matrix Method, but also in the estimation of the efficiencies. At this time, the Matrix Method, which was developed for charged Higgs boson searches, was combined with studies performed in the group working on misidentified leptons in top physics, in order to handle the increasingly challenges of the Matrix Method. The outcome of this combination was Paper III, which also gave the opportunity to compare the Matrix Method with the Fitting Method (also known as jet-lepton and anti-lepton method), an alternative method for misidentified lepton estimation. The Fitting Method defines a model for the non-prompt and misidentified leptons background shape for different distribu- tions. A maximum likelihood fit of a discriminating variable is performed on data to obtain its total normalisation. Monte Carlo simulation is used to build the templates for the shape. The two methods give comparable results within systematic uncertainties, which are between 10% and 50% for the Matrix Method and 50% for the Fitting Method. 5.2.2 Physics analyses The first charged Higgs boson searches in LHC Run 1 were cut-based analy- ses. With the search for heavy charged Higgs bosons, multivariate techniques also found their way into the analyses. Paper I, II and VI - IX describe light charged Higgs boson searches where cut-based physics analyses were used. Therefore, a general description of cut- based analyses for charged Higgs boson searches is given in the following. When searching for physics processes, the overall goal is to select a sub- set of the total data enriched in the physics process of interest. This can be achieved by selection requirements (“cuts”) which favour a certain type of events. When the physics process of interest is identified, its signature in the detector has to be analysed: • Does the process contain leptons, or only hadronically decaying objects? • How many jets are expected in the decay, how many are b-tagged? • Does the process involve neutrinos, hence missing transverse energy? • What is the expected energy range of all decay products? 53 Once the signature is understood the cut-based analysis can be set up. In or- der to estimate the efficiency of these “cuts” and their ability to reduce back- grounds, Monte Carlo simulations are used. In charged Higgs boson searches, and other analyses, event generators like HERWIG [115], ALPGEN [116], PYTHIA [117], MC@NLO [118], POWHEG [119], AcerMC [120], SHERPA [121] are commonly used. Simulations based on different generators may re- sult in systematic uncertainties in the modelling of physics processes. In order to get the best background rejection, together with the best signal selection, a cut optimization is performed. Usually the first step in the analysis is event cleaning, removing those which have uncertain conditions or poorly reconstructed objects. Such events could arise from the fact that parts of the detector system were not fully operational. Jets could have wrong properties due to problems with the LAr (Liquid Ar- gon) calorimeter, or be arising from noise spikes in the calorimeter. Such requirements are usually given to the analyser from specialised working and performance groups. The next step is to choose an appropriate trigger menu and select only events that fired such triggers. If the event contains leptons, a lepton trigger could be used. For events with hadronically decaying taus, special triggers exist, usually in combination with another object (lepton or missing transverse energy). For the charged Higgs boson searches in ATLAS involving electrons or muons, a single-lepton trigger was used. Leptons are easy to trigger on and ATLAS has a very good lepton identification. Single- + lepton triggers have low pT thresholds, and hence are not pre-scaled. The H analysis team was working very closely with the top physics group, who are also using the same triggers, hence the efficiencies of the triggers were well understood. Also, systematic effects were much better understood thanks to this collaboration. Even for the selection of di-lepton events, the same single- lepton triggers were used for the charged Higgs boson searches. With different triggers, the Matrix Method would have had to be developed for each trigger separately, with subsequent problems in finding control regions for measuring the real and misidentification efficiencies. The charged Higgs boson searches did not only share the trigger studies with the top group, but also the object definitions, and performed the same or compatible overlap removals. The overlap removal is important to avoid counting the same object multiple times in the analysis. An electron can easily be reconstructed as an electron and a jet, muons can overlap with electrons, especially in the tracker, and tau objects usually need to be selected from the jet collection. The overlap removal is essentially a priority list: if objects like electrons, taus and jets overlap, which one to count as an electron, which one to count as a jet and which one as a tau? By having triggers, object selections and overlap removals in common with the top group, a lot of analysis tools could also be shared. 54 Following the trigger requirements, every selected event is required to have a certain amount of jets and b-jets, in order to account for the hadronically decaying objects in the physics signature of interest. Light charged Higgs bosons, produced through top pair production are associated with two b-jets. Additional jets could come from the decay of the W boson or other underlying events. Such a requirement removes a large amount of other physics processes which do not contain b-quarks. In physics analyses containing hadronically decaying taus, requirements that a tau object is found in every event are imposed. Since the charged Higgs boson dominantly decays to taus, this is a natural choice. Tau candi- dates are selected from jets with one or three associated tracks (one- or three- prong taus)3 reconstructed in the inner detector [122]. Special algorithm re- ject electrons and muons. Hadronic taus are discriminated against quark- and gluon-initiated jets by their shower shape and tracking variables. Misidentified taus are a major background in the charged Higgs boson analyses involving hadronic taus. The majority of misidentified taus in the final event selection originate from jets, with different misidentification probabilities, depending on the initial parton (light quark, heavy quark or gluon). Other sources for misidentification are electrons and muons, in which case mainly one-prong taus are misidentified. Data driven methods are used in order to estimate the amount of misidentified taus. Correcting the simulation to match the misiden- tification probability measured in contrrol regions of the data is one method. A Matrix Method can also be used for misidentified taus. In the decay of the W boson arising from e.g. top quarks, but also in the decay of the charged Higgs bosons, neutrinos are involved. These neutrinos miss cannot be detected by ATLAS and result in missing transverse energy (ET ). miss Therefore, in the cut-based analysis a certain amount of ET is required. This also reduces other physics processes which do not involve as much missing transverse energy as charged Higgs boson production in tt¯ events. The main background surviving the event selection for charged Higgs boson searches are mostly tt¯ events, because the final state resembles the H+ signal, as can be seen in the Feynman diagram of figure 5.2. After the final selection, an excess (or deficit) of events can indicate new physics if statistically significant. In order to enhance the signal-background separation, discriminating variables are used, where charged Higgs bosons end up dominantly in a certain region. Multi-variate analyses combining several discriminating variables into one, may also be used. However, such analyses were not performed for this thesis. After the final selection, distributions of discriminating variables are used to determine limits. These limits set constraints for example on the branching 3"prong" refers to the number of charged particles in the final state of the tau decay. Generally, the tau decays into a single π± (1-prong) and any number of π0s, or three π± (3-prong) and any number of π0s. 55 g b g b t W + t H+ g g tW¯ − tW¯ − g g ¯b ¯b (a) tt¯ background (b) charged Higgs signal Figure 5.2. The production and decay of tt¯ events (a) and the production of a light charged Higgs boson from a tt¯ decay (b). One W boson in the tt¯ decay is replaced by a charged Higgs boson. ratio of a top quark decaying into a charged Higgs boson. The limit is obtained by comparing the predictions from the SM-only hypothesis with the observed and selected events. A profile likelihood statistical analysis is performed with the test parame- ters of interest. The statistical treatment usually used in ATLAS is described in [123], the limit itself is derived using the CLs criterion [124]andtheasymp- totic approximation [125]. The results are presented in limit plots. An example for a neutral Higgs boson is given in figure 5.3. On the horizontal and vertical axis are the parameters of the theory under test, e.g. the mass of the Higgs boson and the cross-section ratio (σ/σSM), respectively. The solid black line shows observed limits at 95% confidence level, which represents the certainty that a Higgs boson with the given mass does not exist. The dotted black line shows the average expected limit (derived from simula- tion) in the absence of a Higgs boson. The green and yellow bands indicate the corresponding 68% and 95% certainty of the average expected limit. If the solid black line lies below the dotted black line, there is a deficit, indicating that less data than the expected background is observed. If it lies above, there is an excess, i.e. more data than the expected background is observed. If the solid black line lies below the red line (exclusion line at 95% CL), then the SM Higgs boson is excluded with a 95% certainty, since it is not produced with the expected cross section for that mass. If the solid black line lies above the red line and at the same time above the dotted black line, an excess is observed, indicating that there could be a Higgs boson around that mass. 56 Figure 5.3. Simplified exclusion plot made by ATLAS, for the illustration of excess, deficit and exclusion in limit computations [126]. 57 6. Future data readout challenges at LHC 6.1 Introduction The High Luminosity LHC (HL-LHC) will produce proton-proton collisions in 2024 with an expected instantaneous luminosity of 5×1034 cm−2s−1,which is an order of magnitude higher than at Run 1. The ATLASdetector at the LHC is expected to collect around 300 fb−1 of physics data until 2022, whereas the HL-LHC will deliver data of the order of 250 fb−1 per year [127]. The higher luminosity will increase the mean number of interactions per bunch crossing to around 140, resulting in large particle fluxes. In order to keep the good performance of vertex and track reconstruction, lepton identification and heavy flavour tagging in ATLAS under HL-LHC operation, the current tracking system must be replaced by a more granular (silicon) tracker. To take advantage of the luminosity increase for new discoveries and precision measurements in ATLAS, it is essential to improve the trigger system and the capability to reconstruct physics objects over a large acceptance range [128]. Single-lepton trigger pT -thresholds need to be kept at around 20 GeV in order to detect key-signatures like W /Z bosons and tt¯ pairs. The effect of trigger threshold on signal yields is shown in figure 6.1. At the same time, the trigger rates need to be kept low in order to be able to bring all data out of the detector. Therefore, new trigger concepts need to be implemented when upgrading ATLAS for HL-LHC. A Level-0/Level-1 trigger, i.e. a two-step first level hardware trigger architecture, is considered for that purpose. The use of calorimeter-only information for triggering will not reduce the trigger rates enough to maintain low pT -thresholds. Further information from the inner detector is needed already at L1. Sufficient trigger rate reduction can be achieved by implementing a track trigger at L1. Such a self-seeded L1Track trigger is designed as follows. A fast track reconstruction of all high momentum tracks (pT > 10 GeV) can be performed in the full coverage of the tracker without the need for external seeds. The idea is to send the positions of the hit coincidences in the inner detector layers to track finder units, which will perform a hit matching. The expected data rate at HL-LHC is around 1 Gbit/s per silicon module. The concept of the L1Track trigger is not yet fully developed and one of the main challenges is to transfer the data from the silicon modules to the central trigger processor of the detector, for which one possible approach will be presented in the following sections. Details about the Phase II upgrade plan can be found in the Letter of In- tent [128]. 58 %& ' "# && $&"%& #" &%'!'#"$ Figure 6.1. Acceptance of muons from tt¯, WH and SUSY processes as a function of the true muon transverse momentum [128]. 6.2 Wireless technology in future detector systems In the L1Track trigger for the Phase II ATLAS upgrade data transfer is a ma- jor challenge. Tracks propagate in the radial direction of the detector, however the readout in the current detector systems is mainly routed longitudinally, i.e. parallel to the beam. This makes layer to layer communication with short con- nections complicated and a different concept has to be found. Simply connect- ing the different detector layers by cables in the radial direction is not feasible since that will add non-sensitive material, reducing the detector performance significantly. Figure 6.2 shows degradation of the radiation and interaction lengths in the current ATLAS detector for η > 0.7 due to the routing of ser- vices. Data transfer methods for future detectors with less material than today would indeed result in benefits. One technology solution to bring data out from the tracker is using wireless links instead of cables. Doing so, the signal can be routed more freely, en- abling to spread data links over the whole detector volume, hence avoiding the congestion currently in the transition region between the barrel and end-caps. This is especially important for the first 50 cm in the inner detector volume, where the transition between parallel data transfer and high bandwidth se- rial transfer is done. Wireless links open up new possibilities for the detector readout. The data can be transferred in the radial direction to the outside of the detector, but also local communication between detector layers which opens for putting track finding intelligence inside the tracker. This would make it 59 ) ) 0 2.5 λ 0.7 Services Services TRT 0.6 TRT 2 SCT SCT Pixel 0.5 Pixel Beam-pipe Beam-pipe 1.5 0.4 Radiation length (X Interaction length ( 1 0.3 0.2 0.5 0.1 0 0 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 |η| |η| (a) Radiation lengths (X0) (b) Interaction lengths (λ) Figure 6.2. Material distribution (X0, λ) at the exit of the ID envelope, including the services and thermal enclosures. The distribution is shown as a function of |η| and averaged over φ. The breakdown indicates the contributions of external services in their active volume and of individual sub-detectors [129, 130]. possible to reconstruct track primitives locally in real time [131]. Ideas on radial data transfer and inter-layer communication are shown in figure 6.3. (a) Radial data transfer (b) Inter-layer communication Figure 6.3. Radial data transfer (left): The data is transferred via wireless links from the inside of the detector to the outside. Inter-layer communication (right): The inter- esting data from the four layers is collected on one layer for further processing. In this scenario extra hardware is placed on the second layer in order to decide if the event is of interest or not, and whether it should be sent outside the detector. 6.2.1 Choice of the frequency band for wireless data links In the consumer market, wireless technology can be found in many places, such as radio, terrestrial TV, satellite TV and wireless data networks. With the “internet of things” [132] it is becoming even more frequent in our everyday life. These systems operate mainly at frequencies up to 5 GHz and are not 60 suitable for wireless communication in a track trigger system, because they do not provide high enough bandwidth for all data to be transferred. Their modulation techniques are complicated and too power hungry for a tracker. Furthermore, wavelengths at 5 GHz are around 6 cm, which gives a feature size not compatible with trackers. A much more promising candidate for de- tector systems is the 60 GHz frequency band technology. With wavelengths of the order of 5 mm, feature sizes are significantly smaller. Also, up to 7 GHz of bandwidth is available [133]. This opens for small components [134]and high data rates with simple modulation schemes and low power consumption. The industry has discovered the advantages of 60 GHz frequency, for the increasing demand of data transfer capacity. One application that the industry is in particular interested in is the wireless transfer of high definition video between consumer products [135] over short distances, which needs a lot of bandwidth. With the 60 GHz technology, transfer over long ranges is chal- lenging due to the high absorption of 60 GHz waves in air and most materials. This is why this technology is interesting for short distance communication in future trackers. The absorption in materials prevents signal from penetrat- ing through the different layers of the detector system. This makes it possible to re-use the same frequency at different places without interference. Fur- thermore it is of particular importance that the connection antenna to antenna should be in the far-field1, in order to benefit from directivity and gain of the antenna. Transmission distances are of the order of 10 cm in a tracker, which requires high frequencies. Of course, even higher frequencies could be used, e.g. the 80 GHz frequency band, but this would put tighter requirements on the fabrication tolerances, component costs would be higher and it would result in larger power losses in the materials. Taking this into consideration, 60 GHz seems to be the best compromise. 6.2.2 Connection concept The replacement of cables by wireless links can be achieved in a straightfor- ward way. The data of a detector layer is collected in the same way as it is now with optical links. The optical link, with a transceiver and a optical fi- bre is replaced with a wireless transceiver. The transceiver chip generates a carrier wave at 60 GHz from its built-in local oscillator. One challenge is that compared to optical links, chip designs for wireless must be done for very high frequency. In the transmitter part of the chip, the detector data is modu- lated on the 60 GHz carrier and transmitted via an antenna through free space. Obviously less material is needed when using a wireless link. 1The far-field region is where the angular field distribution is independent on the distance to the antenna. The minimum distance where the far-field starts depends on the size of the antenna and the wavelength of the electromagnetic field [136]. 61 The receiver chip gets the signal from free space through the antenna, the data is then demodulated and is digitally available for further processing units. Depending on the scenario in which the wireless link is operated, the required bandwidth is different and so is also the type of antenna. 6.2.3 Requirements on the antenna system When considerations were made about which material and antenna to use for demonstrating a wireless link suitable for trackers, it was clear that a low den- sity design of the antenna would be optimal. Antennas of high density ma- terials would add much non sensitive material to the detector thus should be avoided. Furthermore, the antenna should be cheap and easy to produce in large quantities. It should provide polarization and should radiate most power in one direction. The decision was to use patch antennas. A detailed descrip- tion of this antenna type is given in section 7.6. The patch antenna is radiating its energy mainly in the region perpendicular to the patch and not, like for example a dipole antenna, circularly around radiating elements. Combining several patches in an array gives the possibility to increase the focussing and the gain compared to a single patch antenna. Also, for some use scenarios, it is of importance that the patch antenna has a limited bandwidth of around a few percent [137]. This makes it possible to place several data links, operating at different frequencies, close to each other without causing interference. In scenarios where fewer links are needed, but with a larger bandwidth requirement than the few percent of the patch antenna, wideband antennas are clearly better. Horn antennas [138] provide both a large gain and bandwidth, and they can be produced from Kapton foils [139], giving them a very low density. A disadvantage is their voluminous 3D structure, which may make the positioning of the components complicated. An 3D antenna made from a Kapton foil also is very sensitive to shocks, but this is a minor concern when used in a detector. Slot antennas for 60 GHz can be produced with a bandwidth of more than 7 GHz [140], on a printed circuit board (PCB) substrate. The antennas are produced in the same way as patch antennas, hence have similar advantages in terms of low material density and low fabrication cost. Vivaldi antennas are types of wideband antennas that can be used for 60 GHz. These have models that also can be produced on PCB substrate. Depending on the design, Vivaldi antennas can be constructed both with a radiation direction parallel to the substrate [141], and perpendicular to the substrate [142] like the patch and slot antennas. Other types of antennas considered are Yagi antennas [143], inverted-F an- tennas [144] and many more which are described in the literature [137]. 62 Choice of material for antennas The Roger (ULTRALAM 3850) and DuPont (Pyralux AP9161R) materials were chosen for production. They are copper laminates with two layers of copper and a special (proprietary) substrate in between. The laminates are thin and light, hence do not add much material to the detector. Furthermore they are flexible and can easily be bent, which can be of importance for placing the antennas in the detector. Their electrical properties are good in terms of small dielectric constants (εr = 2.9 at 10 GHz for Rogers, εr = 3.4at1MHz for DuPont), hence the materials suffer small energy losses when radiating at high frequency. The materials are expected to be radiation hard, which is important for working in the conditions of a particle detector. 6.3 Radiation damage to antenna substrate In particle detectors, the radiation from collision products [145]isextremely high, which requires electronics and materials to be very radiation hard. This is also required from antennas to be used in trackers. The effect of radiation damage on the antenna material must be studied. For patch antennas, it is crucial to understand how the electric properties of the PCB substrate changes, when radiated with high doses. Changes to the dielectric constant of the material will shift the resonance frequency of the antenna, which in the worst case can stop a wireless link from working. If several thousands of wireless links are placed in a detector system, a change in frequency could have tremendous effects and could break down the whole system. 6.3.1 Effect of changes in the dielectric constant Studies were performed using the ANSYS HFSS (High Frequency Structure Simulation) program [146] in order to simulate the effect of dielectric changes. A designed single patch antenna, which was optimized for the Rogers PCB substrate, was used for this study. The antenna is resonant at 60 GHz. The dielectric constant was changed from its nominal value 2.9 in the simulation 2 from 2.5 to 3.3. In figure 6.4 the effect on the S-parameter S11 is shown. For smaller values of the dielectric constant, the resonant frequency of the antenna moves to higher values, and vice versa. From the plot, a change of approximately 0.1 in the dielectric constant leads to a 1 GHz frequency shift. 2Scattering parameter, described in section 7.2 in detail. 63 0 -5 S11 [dB] -10 -15 -20 -25 ε=2.9 Nominal ε=2.5 -30 ε=2.6 ε=2.7 -35 ε=2.8 ε=3.0 ε=3.1 -40 ε=3.2 ε=3.3 -45 54 56 58 60 62 64 f [GHz] Figure 6.4. The effect of a change in the dielectric constant on the resonant frequency of a single patch antenna optimized for 60 GHz on Rogers PCB substrate. 6.3.2 Radiation damage studies In order to study the effect of radiation in the two different PCB substrates, two studies were performed. In the first study, small capacitors were produced on both substrates used for antennas. These capacitors were measured for different frequencies ranging from 100 kHz to 1 MHz with an LCR meter. The frequency range is limited by the measurement device. The capacitors were radiated with electrons and re-measured after irradiation. A second study was done with single patch antennas produced on the two substrates, which were measured before and after radiation with the Vector Network Analyser (VNA) at 60 GHz. Radiation with electrons was done at the university hospital in Uppsala reaching a dose of 100 kRad (with a few % precision), which is a dose relevant for space applications but only a small fraction of what is required for LHC. The two studies showed no measurable effect on capacitance and resonance frequency, hence no effect was observed on the dielectric constant for these substrates up to the delivered dose. In order to increase the radiation dose, the test structures were radiated at CERN with orders of magnitude higher doses. Unfortunately the test struc- tures have not been returned for post-radiation measurements and no results are available at the time of writing. 64 6.4 Outlook The next step in this project is to demonstrate the transfer of data at 60 GHz with the required data bandwidth. This project is performed in collaboration with the University of Heidelberg in Germany. There, a 60 GHz transmitter- receiver chip is being designed [147, 148]. This chip will be connected with the antennas designed in Uppsala, and feasibility studies of data transfer will be performed. Possible disturbances from the surrounding detector system will be studied as well. A future possible development would be the integration of the antenna on the transceiver chip. This would have the benefit of a smaller antenna on top of the chip, saving space and making the coupling from the chip to the antenna shorter, possibly resulting in smaller losses. Disadvantages are of course the loss of flexibility in terms of antenna choice, but also in terms of operation at different frequencies of the 60 GHz band. 65 7. Wireless technology 7.1 Electromagnetic waves In radio communications, the propagation of electromagnetic waves in free space is the basis for sending and receiving information between two points which are not connected physically with each other [149]. Inside the receiv- ing and transmitting components, the radio signals exists in form of alternat- ing currents. Outside these conductors, radio signals exist as electromagnetic waves. On the transmitter side, the alternating current in the antenna gener- ates an electromagnetic field and the energy coming from the transmitter is converted into real electromagnetic waves. The antenna on the receiver side captures the electromagnetic waves, and the fields are converted back into al- ternating currents. The concept of electromagnetic waves can be described by Maxwell’s equa- tions [150]: ρ ∇ · E = (7.1) ε0 ∇ · B = 0 (7.2) ∂B ∇ × E = − (7.3) ∂ t ∂E ∇ × B = μ J + ε (7.4) 0 0 ∂t where ε0 is the permittivity of free space, μ0 is the permeability of free space, ρ is the electric charge density, J is the electric current density, B is the mag- netic field and E is the electric field. From these equations it can be seen that whenever an electric current flows, a magnetic field is induced in the surrounding space. Any variation of this magnetic field results in the creation of an electric field. With an oscillating current an oscillating magnetic field is created as well as an oscillating elec- tric field. In turn, an oscillating electric field creates a magnetic field again. Oscillating magnetic and electric fields cannot exist alone. Therefore, the ex- pression “electromagnetic field“ is used. Electromagnetic waves are periodic, oscillating, i.e. with a sinusoidal form and can be characterized by a frequency ( f ). The frequency is expressed in Hertz (Hz). The wavelength λ is in turn defined by: ν λ = (7.5) f 66 where ν is the wave velocity, which can have different values depending on the medium in which the wave propagates. In vacuum ν is equal to the speed of light c = 3 × 108 m/s. Plane waves can in general be expressed as: π ( , )= π − 2 z + Φ , Ex z t E0 sin 2 ft λ (7.6) π ( , )= π − 2 z + Φ Hy z t H0 sin 2 ft λ (7.7) where E0 is the amplitude of the electric field, H0 is the amplitude of the magnetic field and Φ is a phase. The propagation is along the z direction, see figure 7.1. A nomenclature was established for different frequency bands, see table 7.1. Figure 7.1. A plane electromagnetic wave in free space. 7.2 The scattering parameter An electrical network can be seen as a black box containing basic electrical elements, which are connected to each other and behave linearly. The inputs to the black box are called ports and they are interacting with other electrical circuits. 67 Table 7.1. Radio frequency portion of the electromagnetic spectrum [151]. Frequency designation Frequency range Ultra low frequency (ULF) < 3 Hz Extremely low frequency (ELF) 3 Hz to 3 kHz Very low frequency (VLF) 3-30 kHz Low frequency (LF) 30-300 kHz Medium frequency (MF) 300 kHz to 3 MHz High frequency (HF) 3-30 MHz Very high frequency (VHF) 30-300 MHz Ultra high frequency (UHF) 300 MHz to 3 GHz Super high frequency (SHF) 3-30 GHz Extremely high frequency (EHF) 30-300 GHz Submillimeter 300 GHz to 3 THz The scattering matrix describes the relation between the voltage wave ap- plied on the ports to the voltage waves reflected from or transmitted to the ports in an electrical network. The electrical circuit in the black box can be characterized by a square matrix of dimension N,foraN-port circuit, having N2 elements. Each element of the matrix is a complex number, the S-parameter. These elements can be used to calculate the response to a signal which is applied to a port. For a multi- port electrical network, the port numbering goes from 1 to the total number of ports N. The S-parameter√ is defined in terms of the normalized complex√ + = in/ − = re f / incoming Vn Vn Z0 and normalized complex reflected Vn Vn Z0 wave voltage1 [152]. The relation between normalized incoming and reflected waves in matrix form is given by: ⎛ ⎞ ⎛ ⎞⎛ ⎞ − + S11 S12 ... S1N V1 V1 ⎜ − ⎟ ⎜ . ⎟⎜ + ⎟ ⎜ V2 ⎟ ⎜ S21 ...... ⎟⎜ V2 ⎟ ⎜ . ⎟ = ⎜ ⎟⎜ . ⎟ (7.8) ⎝ . ⎠ ⎝ . . . . ⎠⎝ . ⎠ − . . . . + VN SN1 ...... SNN VN S-parameters are defined under certain conditions: • The characteristic impedance of the circuit, Z0; • The applied frequency, f of the voltage wave; • The port numbers, n. Figure 7.2 shows the analogy between an incident light-wave and a RF-wave, which is partly reflected and transmitted. What an optic lens is for a light wave, is ”the device under test“ for the RF wave. The left side of the schematic can be interpretet as port one and the right side as port two. 1 ± Vn are defined in terms of the terminal voltage Vn, the terminal current In, and the characteristic in re f + Vn+√InZ0 √Vn − Vn−√InZ0 V√n impedance Z0: V = = and V = = . n 2 Z0 Z0 n 2 Z0 Z0 68 Figure 7.2. Analogy between an incident light-wave and a RF-wave, which is partly reflected and transmitted. The two port network For an electric circuit with only two ports, the equation (7.8) simplifies to: − + V S11 S12 V 1− = 1+ (7.9) V2 S21 S22 V2 + An incoming wave V1 in port one will result in an exiting wave in port one V − and port two V −. Assuming that port two is terminated by a load identical 1 2 − to the characteristic impedance Z0, all reflected power on this port V will be + 2 absorbed and V2 will be zero. In such a two port system the S-parameters simplify to: − − V1 V2 S11 = + and S21 = + (7.10) V += V += 1 V2 0 1 V2 0 S11 is the reflected power on port one and S21 is the power transferred from port one to port two. A similar relation can be found in the case where port one is terminated + = with a load identical to the characteristic impedance Z0,i.e.whenV1 0: − − V2 V1 S22 = + and S12 = + (7.11) V += V += 2 V1 0 2 V1 0 69 It should be noted that the parameter S11 also can be expressed as S11 = Z1−Z0 where Z1 is the input impedance of port one. In the case of an antenna, Z1+Z0 the electrical input of the antenna is defined as port one and the radiating part of the antenna as port two. In this case, the S11 parameter describes how much power is reflected by the antenna back to the transmitter, often called “reflection coefficient” Γ. It is used to characterize the input impedance of the antenna, i.e. the matching. The antenna is designed to be resonant at a certain frequency at which also a good matching is achieved. If the antenna is resonant, it will radiate power and only a smaller fraction will be reflected. In case of no resonance, the input power will mostly be reflected. The S- parameters are commonly given in deciBels2 (dB). The S-parameters can be converted into dB by: ( )= | | Sxx dB 20log10 Sxx dB (7.12) when the two measurement ports use the same reference impedance. The S-parameters can be measured with a Vector Network Analyser (VNA). Figure 7.3 shows the simulated S11 parameter of a patch antenna resonant at 58 GHz. In the non resonant region, where almost all power is reflected, S11 is close to 0 dB. In the region around 58 GHz, the antenna reflects almost no power and transmits most of it as radio waves, the S11 drops to very small dB values: . 10 S11 [dB] 0 -10 -20 -30 -40 -50 54 56 58 60 62 64 66 f [GHz] Figure 7.3. The S11 parameter of a simulated antenna resonant at 58 GHz.