arXiv:1110.3210v1 [hep-ph] 14 Oct 2011 r tfn Bertolini Stefano Dr. Supervisor: set fSmer raigin Breaking Symmetry of Aspects coaItrainl ueir iSuiAvanzati Studi di Superiore Internazionale Scuola rn nfidTheories Unified Grand hsssbitdfrtedge of degree the for submitted Thesis ret,Spebr2011 September Trieste, otrPhilosophiae Doctor SISSA uaD Luzio Di Luca Candidate: 2 Abstract
SO
We reconsider the issue of spontaneous symmetry breaking in (10) grand unified theories. The emphasis is put on the quest for the minimal Higgs sector leading to a phenomenologically viable breaking to the standard model gauge group. Longstanding results claimed that nonsupersymmetric SO(10) models with just the adjoint representation triggering the first stage of the breaking cannot provide a successful gauge unification. The main result of this thesis is the observation that this no-go is an artifact of the tree level potential and that quantumSO corrections opens in a natural way the vacuum patterns favoured by gauge coupling unification. An analogous no-go, preventing the breaking of (10) at the renormalizable level with representationsSO up to the adjoint, holds in the supersymmetric case as well. In this respect we show that a possibleSO way-out is provided by considering the flipped (10) embedding of the hypercharge. Finally, the case is made for the hunting of the minimal (10) theory. 4 Contents
Foreword 9
1 From the standard model to SO 19
(10) 1.1 Thestandardmodelchiralstructure ...... 20 1.2 TheGeorgi-Glashowroute...... 21 1.2.1 Charge quantization and anomaly cancellation ...... 23 1.2.2 Gaugecouplingunification ...... 24 1.2.3 Symmetrybreaking ...... 25 1.2.4 Doublet-Tripletsplitting ...... 28 1.2.5 Protondecay...... 29 1.2.6 Yukawasectorandneutrinomasses ...... 30 1.3 ThePati-Salamroute ...... 32 1.3.1 Left-Rightsymmetry ...... 33 SO1.3.2 Leptonnumberasafourthcolor ...... 38 1.3.3 Onefamilyunified ...... 40 1.4 (10)grouptheory ...... 40 1.4.1 Tensorrepresentations ...... 41 1.4.2 Spinorrepresentations...... 42 1.4.3 Anomalycancellation ...... 48 1.4.4 Thestandardmodelembedding.SO ...... 48 1.4.5 TheHiggssectorH ⊕ H ...... 49 1.5 Yukawa sectorH ⊕ inH renormalizable (10)...... 52 1.5.1 10 126 withsupersymmetry ...... 55 1.5.2 10 126 withoutsupersymmetry ...... 58 1.5.3d Type-Ivstype-IIseesaw ...... 60 1.6 Protondecayd ...... 61 1.6.1 d =6(gauge)...... 62 1.6.2 d =6(scalar)...... 64 1.6.3 =5...... 66 1.6.4 =4...... 68 Contents 26 Intermediate scales in nonsupersymmetric SO unification 71 SO (10) 2.1 Three-step (10)breakingchains ...... 72 2.1.1 Theextendedsurvivalhypothesis...... 73 2.2 Two-loop gauge renormalization groupU equations ...... 75 2.2.1 Thenon-abeliansector ...... 75 2.2.2 The abelian couplings and (1)mixing ...... 76 2.2.3 Somenotation...... 78 2.2.4U One-loopmatching...... R ⊗ U B−L 79 2.3 Numericalresults ...... 81 2.3.1 (1) φ (1) mixing ...... 82 2.3.2 Two-loopresults(purelygauge)126 ...... 82 2.3.3 The Higgsmultiplets...... 88 2.3.4 Yukawaterms ...... 91 3 The quantum vacuum of the minimal GUT 95 2.3.5 TheprivilegeofbeingminimalSO ...... 92
(10) 3.1 TheminimalSO(10)Higgssector ...... 95 3.1.1 Thetree-levelHiggspotential ...... 96 3.1.2 Thesymmetrybreakingpatterns ...... 97 3.2 Theclassicalvacuum ...... 100 3.2.1 Thestationarityconditions ...... 100 3.2.2 Thetree-levelspectrum ...... 101 3.2.3 Constraints ona the potential parameters ...... 101 3.3 Understandingthescalarspectrumλ ...... 102 2 3.3.1 45 only with =0...... 102a λ β τ 2 3.3.2 16 only with =0...... 102β τ 2 2 3.3.3 A trivial 45-16 potential ( = = = =0)...... 103 3.3.4 A trivialχR 45-16 interaction ( = =0) ...... 103 3.3.5 Atree-levelaccident ...... 103 3.3.6 The =0limit...... 105 3.4 Thequantumvacuum...... 105 3.4.1 Theone-loopeffectivepotential ...... 105 3.4.2 Theone-loopstationaryequations ...... 106 3.4.3 Theone-loopscalarmass...... 107 3.4.4 One-loopPGBmasses...... 108 4 SUSY- breaking with small representations 113 3.4.5SO Theone-loopvacuumstructure ...... 110
(10) SO 4.1 Whatdoneutrinostellus? ...... 113 4.2 SUSY alignment: a case for flipped (10) ...... 116 4.3 TheGUT-scalelittlehierarchy ...... 118 Contents 7
4.3.1 GUT-scale thresholdsand protondecay...... 119 4.3.2 GUT-scaleSO thresholds and one-step unification ...... 120 4.3.3 GUT-scale thresholds and neutrino masses ...... 120 4.4 Minimal flipped (10)Higgsmodel ...... 120 4.4.1 Introducingthemodel...... 122 4.4.2 SupersymmetricvacuumE ...... 127 4.4.3Y Doublet-TripletB−L splittingE in flipped models ...... 129 6 4.5 Minimal Eembedding...... 131 6 4.5.1 and into ...... 134SU 6 4.5.2 The vacuummanifoldE ...... 134 4.5.3 Breaking the residual (5) via effective interactions ...... 138 6 4.5.4 A unified scenario ...... SO 139 4.6 Towardsarealisticflavor...... 140 Outlook:4.6.1 the quest Yukawa for sector the minimal of the flipped nonsupersymmetric(10)modelSO . .. ..theory ...... 147140
A One- and Two-loop beta coefficients (10) 153 U
A.1 Beta-functions with (1)mixing ...... 159 B A.2SO Yukawacontributionsalgebra representations ...... 165 161
(10) B.1 TensorialrepresentationsC ...... 165 B.2 Spinorialrepresentations...... 166 B.3 The charge conjugation ...... 167 CB.4 Vacuum TheCartangenerators stability ...... 171 168 D Tree level mass spectra 173
D.1 Gaugebosons...... 173 D.1.1 Gaugebosonsmassesfrom45 ...... 174 D.1.2 Gaugebosonsmassesfrom16 ...... 174 D.2 Anatomyofthescalarspectrum ...... 175 D.2.1 45only...... χR 6 175 D.2.2 16only...... a λ β τ 175 D.2.3 Mixed 45-16 spectrum ( =0)β τ ...... 176 2 2 D.2.4 A trivial 45-16 potential ( = χR= = =0)...... 178 D.2.5 A trivial 45-16 interaction ( = =0) ...... 178 D.2.6 The 45-16 scalar spectrum for =0 ...... 179 Contents E8 One-loop mass spectra 183
E.1 GaugecontributionstothePGBmass ...... 183 F Flipped SO vacuum 189 E.2 ScalarcontributionstothePGBmass...... 184 SO (10) F.1 Flipped (10)notation...... 189 F.2 Supersymmetricvacuummanifold...... 191 F.3 Gaugebosonspectrum ...... 198 F.3.1 Spinorialcontribution ...... 199 F.3.2 Adjointcontribution ...... 200 G E vacuum 203 F.3.3 Vacuumlittlegroup...... 201 SU 6 E 3 G.1 The (3) formalism...... 203 6 G.2 vacuummanifold...... 206 G.3 Vacuumlittlegroup ...... 208 Foreword
This thesis deals with the physics of the 80’s. Almost all of the results obtained here could have been achieved by the end of that decade. This also means that the field of grand unification is becoming quite old. It dates back in 1974 with the seminal papers of Georgi-Glashow [1] and Pati-Salam [2]. Those were the years just after the foundation of the standard model (SM) of Glashow-Weinberg-Salam [3, 4, 5] when simple ideas (at least simple from our future perspective) seemed to receive an immediate confirmation from the experimental data. Grand unified theories (GUTs) assume thatSU allC the⊗U fundamentaQ l interactions of the SM (strong and electroweak) have a common origin. The current wisdom is that we live in a broken phase in which the world looks (3) (1) invariant to us and the low- energy phenomena are governed bySU strongL ⊗ interactions U Y and electrodynamics. Growing withU the energyQ we start to see the degreesU Q Ï of freedom SU L ⊗ of U a newY dynamics which can be interpreted1 as a renormalizable (2) SU(1) gaugeC ⊗ SU theoryL ⊗ spontaneously U Y broken into (1) . Thus,G in analogy to the (1) (2) (1) case, one can imagine that at higher energies the SM gauge group (3) (2) (1) is embedded in ′ Ma simpleU ≫ M groupW . G g g g SU TheC ⊗ first SU implicationL ⊗ U Y of the grand unification ansatz is thatgU at some mass scale ′ 3 2 g theg relevantg symmetry is and the , and coupling constants of (3) (2) (1) merge into a single gauge coupling . The rather different 3 2 values for , and at low-energy are then due to renormalization effects. Actually one of the most solid hints in favor of grand unification is the fact that the running within15 the SM shows an approximate convergence of the gauge couplings around 10 GeV (see e.g. Fig. 1). This simple idea, though a bit speculative, may have a deep impact on the under- standing of our low-energy world. Consider for instance som2e unexplained features of the SM like e.g. charge quantization or anomaly cancellation . They appear just as the 1 At the time of writing this thesis one of the main ingredients of this theory, the Higgs boson, is still missing experimentally. On the other hand a lot of indirect tests suggest that the SM works amazingly well and it is exciting that the mechanism of electroweak symmetry breaking is being tested right now at the2 Large Hadron Collider (LHC). In the SM anomaly cancellation implies charge quantization, after taking into account the gauge invariance of the Yukawa couplings [6, 7, 8, 9]. This feature is lost as soon as one adds a right-handed Foreword -1 10 Αi 60 50 40 30 20 10
log10HΜGeVL 5 10 15 18
U Y G
One-loop running of the SM gauge couplings assuming the (1) embedding into . Figure 1: SO natural consequence of starting with an anomaly-free simple group such as (10). Most importantly grand unification is not just a mere interpretation of our low- energy world, but it predicts new phenomena which are correlated with the existing ÷ ones. The most prominentτp of& these is the instability of matter. The current lower bound on the proton lifetime is something33 34 like 23 orders of magnitude bigger than the age of the Universe, namely 10 yr depending on the decay channel [11]. This number is so huge that people started to consider baryon number as an exact symmetry of Nature [12, 13,3 14]. Nowadays we interpret it as an accidental global symmetry of the standard model . This also means that as soon as we extend the SM there is the chance to introduce baryon violating interactions. Gravity itself could be responsible for the breaking of baryon number [17]. However among all the possible frameworks there is only one of them which predicts a proton lifetime close to its experimental limitMU and this theory is grand unification. Indeed we can roughly estimate it by dimensional arguments. The exchange of a baryon-number-violating− MU vector boson of mass τp ∼αU ; yields something like mp4 1 − 5 αU ∼ τp & (1) MU & 1 33 and by putting in numbers15 (we take 40, cf. Fig. 1) one discovers that 10 yr corresponds to 10 GeV, which is consistent with the picture emerging in Fig. 1. ÷ NoticeMU that the gauge running is sensitive to the log of the scale. This means that a 10% variation on the gauge couplings at the electroweak scale induces a15 100%18 one on . Were the apparent unification of gauge couplings in the window 10 GeV just νR νR an accident, then Nature would have played a bad trick on us. neutrino3 , unless is a MajoranaSU particleL [10]. B L − π/α − In the SM the baryonic current is anomalous and baryone number∼ e violation can arise from instanton 2 transitions between degenerate (2) vacua which lead to ∆2 = ∆ =173 3 interactions for three flavor families [15, 16]. The rate is estimated to be proportional to and thus phenomenologically irrelevant. Foreword 11 G U Another firm prediction of GUTs are magnetic monopoles [18, 19]. Each time a simple gauge group is broken to a subgroup with a (1)SU factor there are topologically − nontrivial configurationsQm of theπ/e Higgs field whichMm leadsαU M toU stable monopole solutions of the gauge potential. For instance the breaking of 1 (5) generates a monopole with magnetic charge = 2 and mass = [20]. The central core of a GUT monopole contains the fields of the superheavy gauge bosons which mediate proton decay, so one expects that baryon number can be violated in baryon-monopole scattering. Quite surprisingly it was found [21, 22, 23] that these processes are not suppressed by powers of the unification mass, but have a cross section typical of the strong interactions. Though GUT monopoles are too massive to be produced at accelerators, they could have been produced in the early universe as topological defects arising via the Kibble mechanism [24] during a symmetry breaking phase transition. Experimentally one − − − − tries< to measure their interactions as they pass through matter. The strongest bounds − ÷ − − − on the flux16 < of monopoles2 1 1 come from their interactions with the galactic magnetic field (Φ 10 cm 18sr 29 sec 2) and1 the1 catalysis of proton decay in compact astrophysical objects (Φ 10 cm sr sec ) [11]. Summarizing the model independent predictions of grand unification are proton decay, magnetic monopoles and charge quantizationM (andU /M theW ir∼ deep connection). How- ever once we have a specific model we can do even more. For insta13 nce the huge ratio ÷ between themf unification/mν ∼ and the electroweak scale, 10 , reminds us about the well established hierarchy7 13 among the masses of charged fermions and those of neutrinos, 10 . This analogy hints to a possible connection between GUTs and neutrino masses. The issue of neutrino masses caught the attention of particle physicists since a long time ago. The model independentd way to parametrize them is to consider the SM as an effective field theory by writingY all the possible operators compatible with gauge ν ℓT H C HT ℓ : invariance. Remarkably at the =L 5 level there is only one operator [25] hHi2 v 2 ( ) ( ) (2) Λ v After electroweak symmetry breakingMν =Yν and: neutrinos pick up a Majorana mass term 2L √ m : = atm ∼ (3) Λ The lower bound on the highest neutrino. Y eigenvalue÷ inferred: from ∆ 0 05 eV L ν O tells us that the scale at which the lepton number14 15 is violated is Λ (10 GeV) (4) Actually there are only three renormalizable ultra-violet (UV) completion of the SM which can give rise to the operator in Eq. (2). They go under the name of type- I [26, 27, 28, 29, 30], type-II [31, 32, 33, 34] and type-III [35] seesaw and are respectively Foreword 12 ; ; F ; ; H ; ; F obtainedL by introducing a fermionic singletℓ (1 1 0)H , a scalar triplet (1 3 +1) and a fermionic triplet (1 3 0) . These vector-like fields, whose mass can be identified with Λ , couple at the renormalizable level with and so that the operator in Eq. (2) is generated after integrating them out. Since their mass is not protected by the chiral symmetry it can be super-heavy, thus providing a rationale for the smallness of neutrino masses. Yν L Notice that this is still an effective field theoryYν language and we cannot tell atL this level if neutrinos are light because is small or because ΛYν is large.L It is clear that4 SOwithout a theory that fixes the structure of we don’tSO have much to say about Λ . As an example of a predictive theory which can fix both and Λ we can mention (10) unification. The most prominentYν feature of (10) is that a SM fermion family B L SO plus a right-handedL neutrino fit into a single− 16-dimensional spinorial representation. In turnM thisB−L . readilyMU implies that is correlated to the charged fermion Yukawas. At the same time Λ can be identifiedSO with the generator of (10), and its breaking scale, , is subject to the constraints of gauge coupling unification. Hence we can say that (10) is also a theory of neutrinos, whose self-consistency can be tested against complementary observablesSO such as the proton lifetime and neu- trino masses. The subject of this thesis will be mainly (10) unification. In the arduousSO attempt of describing the state of the art it is crucial to understand what has been done so far. In this respect we are facilitated by Fig. 2, which shows the number of (10) papers per year from 1974 to 2010. •By looking÷ at this plot it is possible to reconstruct the following historical phases: τ 1974 1986: Golden age of grand unification.p ∼ These are the years of the foun- dation in which the fundamentalτp & aspects of the31 theory are worked out. The first estimate of the proton lifetime30 yields 10 yr [37], amazingly close to the experimental bound 10 yr [38]. Hence the great hope that proton decay is • behind÷ the corner. 1987 1990: Great depression. Neither proton decay nor magnetic monopoles are observed so far. Emblematically the last workshop on grand unification is & • held in 1989 [39]. 1991: SUSY-GUTs. The new data of the Large Electron-Positron collider (LEP) seem to favor low-energy supersymmetry as a candidate for gauge coupling uni- fication. From now on almost all the attention is caught by supersymmetry. L 4 The other possibility is that we mayL probe experimentallyV A the new degrees of freedom at the scale Λ in such a way to reconstruct the theory of neutrino masses. This could be the case for left-right symmetric theories [30, 34] where Λ is the scale of the + interactions. For a recent study of the interplay between LHC signals and neutrinoless double beta decay in the context of left-right scenarios see e.g. [36]. Foreword SOH10L 13
50
40
30
20
10 yr 1980 1985 1990 1995 2000 2005 2010
SO
Figure 2: Blue: number of papers per year with the keyword " (10)" in the title as a function of the years. Red: subset of papers with the keyword "supersymmetry" either in the title or in the abstract. Source:& inSPIRE. • SO 1998: Neutrino revolution. Starting from 1998 experiments begin to show that atmospheric [40] and solar [41] neutrinos change flavor. (10) comes back with & • a rationale for the origin of the sub-eV neutrino mass scale. 2010: LHC era. Has supersymmetry something to do with the electroweak scale? The lack of evidence for supersymmetry at the LHC would undermine & τ ÷ • SUSY-GUT scenarios. Back to nonsupersymmetric GUTs? p ∼ 34 35 2019: Next generation of proton decay experiments sensitive to 10 yr [42]. The future of grand unification relies heavily on that.
Despite the huge amount of work done so far, the situation does not seem very clear at the moment. Especially from a theoretical point of view no model of grand unification emerged as "the" theory. The reason can be clearly attributed to the lack of experimental evidence on proton decay. In such a situation a good guiding principle in order to discriminate among models and eventually falsify them is given by minimality, where minimality deals interchange- ably with simplicity, tractability and predictivity. It goes without saying that minimality could have nothing to do with our world, but it is anyway the best we can do at the moment. It is enough to say that if one wants to have under control all the aspects of the theory the degree of complexity of some minimal GUT is already at the edge of the tractability. Quite surprisingly after 37 years there is still no consensus on which is the minimal theory. Maybe the reason is also that minimality is not a universal and uniquely defined concept, admitting a number of interpretations. For instance it can be understood as a mere simplicity related to the minimum rank of the gauge group. This was indeed Foreword 14 SU the remarkable observation of Georgi and Glashow: (5) is the unique rank-4 simple group which contains the SM and has complex representations. However nowadays we can say for sure that the Georgi-Glashow model in its5 original formulation is ruled out because it does not unify andSO neutrinos are massive . SU From a more pragmatic point of view one could instead use predictivity as a measure of minimality. This singlesSO out (10) as the best candidate. At variance with (5), the fact that all the SM fermionsSO of one family fit into6 the same representation makes the Yukawa sector of (10) much more constrained . Actually, if we stick to the (10) case, minimality is closely related toSO the complexity of the symmetry breaking sector. Usually this is the most challenging and arbitrary aspect of grand unified models. While the SM matter nicely fit in three (10) spinorial families, this synthetic feature has no counterpart in the Higgs sector where higher- dimensional representations are usually needed in order to spontaneously break the enhanced gauge symmetry down to the SM. Establishing the minimal Higgs content needed for the GUT breaking is a7 basic question which has been addressed since the early days of the GUT program . Let us stress that the quest for the simplest Higgs sector is driven not only by aesthetic criteria but it is also a phenomenologically relevant issue related to the tractability and the predictivity of the models. Indeed, the details of the symmetry breaking pattern, sometimes overlooked in the phenomenological analysis, give further constraints on the low-energy observables such as the proton decay and the effective SM flavor struc- ture. For instance in order to assess quantitatively the constraints imposed by gauge coupling unification on the mass of the lepto-quarks resposible for proton decay it is crucial to have the scalar spectrum under control. Even in that case some degree of arbitrariness can still persist due to the fact that the spectrum can never be fixed completely but lives on a manifold defined by the vacuum conditions. This also means that if we aim to a falsifiable8 (predictive) GUT scenario, better we start by considering a minimal Higgs sector . 5 Moved by this double issue of the Georgi-Glashow model, two minimal extensions which can cure at the6 same time both unification and neutrino masses have been recently proposed [43, 44]. Notice that here we do not have in mind flavor symmetries, indeed the GUT symmetry itself already constrains the flavor structure just because some particles live together in the same multiplet. Certainly one could improve the predictivity by adding additional ingredients like local/global/continuous/discrete symmetries on top of the GUT symmetry. However, though there is nothing wrong with that, we feel that it would be a no-ending process based on assumptions which are difficult to disentangle from the unification idea. That is why we prefer to stick as much as possible to the gauge principle without further7 ingredients. Remarkably the general patterns of symmetry breaking in gauge theories with orthogonal and unitary groups were already analyzed in 1973/1974 by Li [45], contemporarily with the work of Georgi and8 Glashow. SO As an example of the importance of taking into account the vacuum dynamics we can mention the minimal supersymmetric model based on (10) [46, 47, 48]. In that case the precise calculation of the mass spectrum [49, 50, 51] was crucial in order to obtain a detailed fitting of fermion mass parameters Foreword 15 SO The work done in this thesis can be understood as a general reappraisal of the issue of symmetry breaking in (10) GUTs, both in their ordinary and supersymmetric realizations. We can already anticipateSO that, before considering9 any symmetry breaking dynam- ics, at least two Higgs representations are required by the group theory in order to SU achieve• H a full breakingH of (10) to the SM: SU U • 16H or 126H : they reduceH the rank but leave an (5) little group unbroken.⊗ SU 45 or 54 or 210 : they admit for little groups different from (5) (1), yielding the SM when intersectedH H with (5). H H H While theH choice between 16 or 126 is a model dependent issue related to the details of the Yukawa sector, the simplest option among 45 , 54 and 210 is given by the SU U adjoint 45 . ⊗ H H However,H since the early 80’s, it has been observed that the vacuum dynamics aligns the adjoint along an (5) (1) direction, making the choice of 16 (or 126 ) and 45 alone not phenomenologically viable.SU In the nonsupersymmetric case the alignment is only approximate [56, 57, 58, 59], but it is such to clash with unification constraints which do not allow for any (5)-like intermediate stage, while in the supersymmetric limit the alignment is exact due to F-flatness [60, 61, 62],H thus never landing to a supersymmetric SM vacuum. The focus of the thesis consists in the critical reexamination of these two longstanding no-go for the settings with a 45 driving the GUT breaking. SO Let us first consider the nonsupersymmetric case. We start by reconsidering the issue of gauge coupling unification in ordinary (10) scenarios with up to two inter- mediate mass scales, a needed preliminary step before entering the details of a specific Umodel. After complementing the existing studies in several aspects, as the inclusion of the (1) gauge mixing renormalization at the one- and two-loop level and the reassessment of the two-loop beta coefficients, a peculiar symmetry breaking pattern with just the adjoint representation governing the first stage of the GUT breaking emerges as a potentially viable scenarioSO [63], contrary to what claimed in the literatureH [64]. HThis brings us to reexamine the vacuum of the minimal conceivable Higgs potential responsibleH ⊕ H for the (10) breaking to the SM, containing an adjoint 45 plus a spinor SU U SU 16 . As already remarked, a⊗ series of studies inh theH i early 80’s [56, 57, 58, 59] ofh theH i 45 16 model indicated that the only intermediate stages allowed by the scalar sector dynamics were (5) (1)SO for leadingÏ 45 or (5) for dominant 16 H. and9 show a tension between unification constraints and neutrino masses [52, 53]. It shouldF be mentionedF that a one-step (10) SM breaking can be achieved via only one 144 irreducible Higgs representation [54]. However, such a setting requires an extended matter sector, including 45 and 120 multiplets, in order to accommodate realistic fermion masses [55]. Foreword 16 SU SO Since an intermediate (5)-symmetric stage is phenomenologically not allowed, this observation excluded the simplest (10) Higgs sector from realistic consideration. One of the main results of this thesis is the observation that this no-go "theorem" SU SU U is actually an artifact of the tree-level potential and, as we haveC ⊗ shownL ⊗ in Ref.R [65] SU SU SU U (see alsoC ⊗ Ref.L [66]⊗ forR a⊗ briefB−L overview), the minimization of the one-loop effective potential opens in a natural way also the intermediate stageSOs (4) (2) (1) and (3) h H i(2) (2) (1) , which are the options favoured by gauge unification. This result is quite general, since it applies whenever the (10) breaking is triggered bySO the 45 (while other Higgs representations control the intermediate and weak scale stages) and brings back from oblivion the simplestH scenario⊕ H ⊕of nonsupersymmetricH H (10) unification. It is thenSO natural to consider the Higgs system 10 16 45 (where the 10 is needed to give mass to the SM fermions at the renormalizable level) as the potentially minimal (10) theory, as advocated long ago by Witten [67]. However, apart from issues related to fermion mixings, the mainB obstacleL with such a model is given by M −α /π M /M M M neutrino masses. They can be generatedR radiatively∼ U atB−L the twUo-loop level, butB−L turn≪ outU to be too heavy. The reasonMR being that the breaking2 2 is communicated to right- ÷ handed neutrinos at the effective level ( ) and since by13 unification14 constraints, undershoots by several orders of magnitude the value 10 GeV naturally suggested by the type-I seesaw. At these point one can consider two possible routes. Sticking toH the request ofH Higgs representations with dimensions up to the adjoint one can invoke TeV scale supersymmetry, or we can relax this requirement and exchange the 16 with the 126 in the nonsupersymmetricMB−L case. MU In the formerM caseR the gauge running within the minimalF superF HsymmetricH /MP SM (MSSM) prefers in the proximity of so that one can naturally reproduce the desired range for , emerging from the effective operator 16 16 16 16SO . Motivated by this argument, we investigate under which conditions an Higgs sector containing only representations up to the adjoint allows supersymmetric (10) GUTs to breakSO spontaneously to the SM. Actually it is well known [60, 61, 62] that the relevant superpotential does not support, at the renormalizable level, a supersymmetric breaking of the (10) gauge group to the SM. Though the issue can be addressed by giving up renormalizability [61, 62], this option may be rather problematic due to the active role of Planck induced operators in the breaking of the gauge symmetry. They introduce an hierarchy in the mass spectrum at the GUT scale which may be an issue for gauge SO U unification, proton decay and neutrino masses. ⊗ In this respectH we pointedH ⊕ out [68]H that the minimal Higgs scenario that allows for a SO U renormalizable breaking to the SM⊗ is obtained considering flipped (10) (1) with one adjoint 45 and two 16 16 Higgs representations. Within the extended (10) (1) gauge algebra one finds in general three in- equivalent embeddings of the SM hypercharge. In addition to the two solutions with Foreword SU SU U SO 17 SU ⊗ SO SO U the hypercharge stretching over the (5) or the (5) (1) subgroups⊗ of (10) (respectively dubbed as the “standard” and “flipped” (5)U embeddings [69, 70]), there is a third, “flipped” (10) [71, 72, 73], solution inherent to the SU(10) (1) case, with a non-trivial projectionSO of the SM hypercharge ontoSO the (1) factor. Whilst the difference between the standard and theSO flipped (5) embedding is se- mantical from the (10) point of view, the flipped SO (10) case is qualitatively different. In particular, the symmetry-breaking “power” of the (10) spinor and adjoint repre- sentations is boosted with respect to the standard (10) case, increasing the number of SM singlet fields that may acquire non-vanishing vacuum expectation values (VEVs). This is at the root of the possibility of implementing the gauge symmetry breaking by E SO U means of a simple renormalizable Higgs sector. ⊗ The model is rather peculiar in the flavor sector and can be naturally embedded 6 in a perturbative grand unified scenario above the flipped (10) (1)H partial- unificationH scale. On the otherMB−L hand,/MU sticking to the nonsupersymmetricd case with a 126 in place of a 16MR, neutrinoMB−L masses are generated at the renormalizable level. This lifts the ∼ SO problematic suppression factor inherentH ⊕ toH ⊕ the H= 5 effective mass and yields , that might be, at least in principle, acceptable. As a matter of fact a nonsupersymmetric (10) model including 10 45 126 in the Higgs sector has all the ingredients to be the minimal realistic version of the theory. This option at the timeH ⊕ of writingH the thesis is subject of ongoing research [74]. Some preliminary results are reported in the last part of the thesis. We have performedMB−L the minimization of the 45 126 potential and checked that the vacuum constraints allow for threshold corrections leading to phenomenologically reasonable values of . If the model turned out to lead to a realistic fermionic spectrum it would be important then to perform an accurate estimate of the proton decay branching ratios.SO The outline of the thesis is the following:SU the first Chapter is an introduction to the field of grand unification. The emphasis is put on the construction of (10) starting from theSO SM and passing through (5) and the left-right symmetric groups. The second Chapter is devoted to the issue of gauge couplingsU unification in nonsupersym- metric (10). A set of tools for a general two-loop analysis of gauge coupling unifi- SO cation, like for instance the systematizationH ⊕ H of the (1) mixing running and matching, is also collected. Then in the third Chapter we consider the simplest and paradigmatic (10) Higgs sector made by 45 16 . After reviewing the old tree level no-go argument we show, by means of an explicit calculation, that the effective potential al- SO lowsH ⊕ forH those⊕ H patterns which were accidentally excluded at tree level. In the fourth Chapter we undertake the analysis of the similar no-go present in supersymmetry with 45 16 16 in the Higgs sector. The flipped (10) embedding of the hypercharge is proposed as a way out in order to obtain a renormalizable breaking with only rep- resentations up to the adjoint. We conclude with an Outlook in which we suggest the possible lines of development of the ideas proposed in this thesis. The case is made Foreword 18 SO for the hunting of the minimal realistic nonsupersymmetric (10) unification. Much of the technical details are deferred in a set of Appendices. Chapter 1
From the standard model to O S
SO (10)
InSU this chapter we give the physical foundations of (10) as a grand unified group, starting from the SM and browsing in a constructive way through the Georgi-Glashow (5) [1] and the left-right symmetric groups such as the Pati-Salam one [2]. This will offer us the opportunity to introduce the fundamental concepts of GUTs, as charge quantization, gauge unification, proton decay and the connection with neutrino masses in a simplifiedSO and pedagogical way.
The (10) gauge groupSO as a candidate forSU the unification of the elementary in- teractions was proposed long ago by Georgi [75] and Fritzsch and Minkowski [76]. The main advantage of (10) with respectSO to (5) grand unification is that all the known SM fermions plus three right handed neutrinos fit into three copies of the 16-dimensional spinorial representation of (10).SU In recent years the field received an extra boost due to the discovery of non-zero neutrino massSOes in the sub-eV re- B L gion. Indeed, while in the SM (and similarly− in (5)) there is no rationale for the origin of the extremely small neutrino mass scale, the appeal of (10) consists in the predictive connection between the local16 breaking scale (constrained by gauge coupling unification somewhat below 10 GeV) and neutrino massesSO around 25 orders of magnitude below. Through the implementation of some variant of the seesaw mech- anism [26, 27, 28, 29, 30, 31, 32, 33, 34] the inner structure of (10) and its breakingSO makes very natural the appearance of such a small neutrino mass scale. This striking connection with neutrino masses is one of the strongest motivations behind (10) andV it canA be traced back to the left-right symmetric theories [2, 77, 78] which provide a direct connection of the smallness of neutrino masses with the non-observation of the + interactions [30, 34]. SO
Chapter 1. From the standard model to 1.120 The standard model chiral structure (10)
SU C ⊗ U Q The representations of the unbroken gauge symmetry of the world, namely (3) (1) , are real. In other words, for each colored fermion field1 of a given electric charge we have a fermiong field of opposite colorG and charge . If notD g so we would ∗ observe for instance a massless charged fermion field and thisD isg not the case. More formally, being an element of a group , a representation ( ) is said to be real (pseudo-real) if it is equalSD g toS− its conjugateD∗ g representg ationG; ( ) up to a similarity transformation, namely ∈ S 1 ( ) = ( ) forall (1.2) whit symmetric (antisymmetric).S A complex representation is neither real norTa pseudo-real. G D g igaTa It’s easy to prove that must be either symmetric or antisymmetric. Suppose ST S− T∗ : generates a real (pseudo-real) irreduciblea unitary− a representation of , ( ) = exp , 1 so that Ta − T = − T T T (1.3) STaS T S T S Ta ; Because the are hermitian,− wea can write a − 1 1 = or− T ( − ) T = (1.4) Ta S STaS S which implies 1 1 − T =Ta (; S )S : (1.5) or equivalently 1 =0 (1.6) But if a matrix commutesS with− ST allλI the generatorsST ofλS: an irreducible representation, Schur’s Lemma tells us that it is a multiple of the identity, and thus 1 ψL iσω 1 = ψL or = ψL Ï e ψL (1.7) µν i T Asσω is usual ≡ω inσµν grandσµν ≡ unificationγµ; γν we useγµ the; γν Weylg notationµν in which all fermion fields areψL Cψ left-L T { } − T handedC (LH) four-componentσµνC spinors.Cσµν Given a fieldC transformingγµC γµ as under the Lorentz 2− − group (γ , and = 21 ) an invariant mass term is given by σ where is such that = γ or (up to a sign); γi = i. Using; the following representation for the matrices σi 0 − 0 1 0 − † T σi = C = C iγ γ C C C C(1.1) 1 0 0 − iθ − − U ψL e ψ1L 2 0 Ï where are the Pauli matrices, an expression for reads = ψL , with = = = . ′T ′ Notice that the mass termψL isCψ notL invariant under the (1) transformation ψL ψLand in order to ′ avoid the breaking of any abelianψL quantum number carried by (such as lepton numberψL or electric charge) we can construct where for every additive quantum number andψR have opposite ′T ψ ψ C ψR ψ ψL charges.R ≡ L This just means that if is associated with a certain fundamental particle, is associatedR with its antiparticle. In order to recast a more familiar notation let us define a field by the equation . In therms of the right-handed (RH) spinor , the mass term can be rewritten as . 1.2. The Georgi-Glashow route λ 21 λ S ± 2 By transposing twice we get back to where we started and thus we must have = 1 and so = 1, i.e. must beψL either symmetric or antisymmetric. ψ D g ψ The relevanceL Ï of thisL fact for the SM is encoded in the following observation: given a left-handed fermion field transforming under some representation, reducible or T irreducible,D g ( ) , oneψL canCSψ constructL C a gauge invariant mass term only if the representation is real. Indeed, it is easy to verify (by using Eq. (1.2) and the unitarity of SU( )) that the mass term , where denotes the Dirac chargeS conjugation matrix, is invariant. Notice that if the representation were pseudo-real2 (e.g. a doublet of (2)) the mass term vanishes because of the antisymmetry of . SU U The SM is built in such a wayL ⊗ that thereY are no bare mass terms and all the masses stem from the Higgs mechanism. Its representations are said to be chiral because they are chargedG under the (2) (1) chiral symmetry in such a way thatG fermions SU SU U SU U are massless as long as the chiralC symmetry⊗ L ⊗ is preserved.Y Ï AC co⊗mplexQ representation of a group may of course become real when restricted to a subgroup of . This is exactly what happens in the (3) (2) (1) (3) (1) case. When looking for a unified UV completion of the SM we would like to keep this feature. Otherwise we should also explain why, according to the Georgi’s survival hypothesis [79], all the fermions do not acquire a super-heavy bare mass of the order of the scale at which the unified gauge symmetry is broken. 1.2 The Georgi-Glashow route
SU SU U The bottom line of the last sectionC ⊗ was thatL a⊗ realisticY grand unified theory is such that the LH fermions are embedded in a complex representation of the unified group (in SU SU U particular complex under (3) (2) (1) ). If we furtherC require⊗ minimalityL ⊗ Y (i.e. rank 4 as inSU the SM) one reaches the remarkable conclusion [1] that the only simple group with complex representations (which containsSU (3) (2) (1) i i ;:::; SU C SU L as a subgroup) is (5). ⊗ Let us consider the fundamental representation of (5) andSU denoteC it as a 5- dimensional vector 5 ( = 1 SU5).L It is usual to embed (3) (2) in such a way that the first three components of 5 transform as a triplet of (3) and the last ; ; : two components as a doublet of (2) ⊕ 5=(3 1) (1 2) (1.8) q ; ; ℓ ; ; uc ; ; dc ; ; ec ; ; : In the∼ SM we have∼ 15 Weyl− fermions∼ per family− with∼ quantum numb∼ ers 1 1 2 1 T 6C C 2 3 3 (3 2 + ) − (1 2 ) (3 1 ) (3 1 + ) (1 1 +1) (1.9) 2 The relation = and the anticommuting property of the fermion fields must be also taken into account. SO
Chapter 1. From the standard model to 22 SU SU(10)
How to embed these into (5)? One would be tempted to try with a 15 of (5). A S ; Actually from the tensor product ⊗ ⊕
A S ⊗ ⊕ 5 5=10 15 (1.10) ; ⊕ and the fact that 3⊕ 3 = 3 6 one concludes that some of the known quarks should SU C SU L U Y belong to color sextects, which is⊗ not the case.⊗ So the next step is to try with 5 10 or better with 5 10 since there is no (3 1) in the set of fields in Eq. (1.9). The ; ; ; ; ; decomposition of 5 under (3) (2) ⊕ (1) −is simply 1 1 5 = (3 1 + 3 ) (1 2 2) SU(1.11) Y dc Y ℓ Y dc Y ℓ where we have exploited the fact that− the hypercharge is a traceless generator of (5), which implies the condition1 3 ( )+21 ( ) = 0. So, up to a normalization factor, one 3 2 A ; ; ; ; ; ; : may choose ( ) = and⊗ ( ) = . Then− ⊕ from Eqs. (1.10)–(1.11)⊕ we get 2 1 3 6 10 = (5 5) = (3 1 ) (3 2 +⊕) (1 1 +1) (1.12) dc uc uc u d Thus the embedding of a SM fermion family into 5 −10 reads dc uc uc u d 1c − c 3c 2 1 1 d ; u0 u u d ; 2 3 − 1 2 c2 e u 0u u e 3 − 2 − 1 − 3c 3 5 = ν 10 = d d 0d e (1.13) − − 1 − 2 − 3 − 0 1 2 3 SU L q u d 0ℓ ν e iσ ℓ ℓ∗ ∼ where we have expressed the (2) doubletsSU as =3 ( ) and = ( ). Notice in 2 particular that the doublet embedded in 5 is . It may be useful to know how the (5) generators act of 5 and 10. From the i † i k k l transformation properties U k ; ij Ui Uj kl ; Ï Ï † U iT T5 T( ) 5 10 10 (1.14)
i i k where = exp and δ= , weTk deduce; that δ theij actionT; ofij the: generators is − { } 5 = 5 10 = 10 SU (1.15)
Alreadyσ at this elementary level we can list a set of important featuresiσ ℓ of ℓ∗(5) which are typical of any GUT. ∼ 3 SU L 2 2 Here is the second Pauli matrix and the symbol " " stands for the fact that and transform in the same way under (2) . 1.2. The Georgi-Glashow route 1.2.1 Charge quantization and anomaly cancellation 23
SU C SU The charges of quarks and leptonsQ area;a;a;b; related. Leta usb write; the most general electric charge generator compatible with the (3) invariance− − and the (5) embedding Q = diag ( 3 ) (1.16) Q dc a Q e b Q ν a b where Tr = 0. Then by applying− Eq. (1.15)− we find Q uc a Q u a b Q d a b Q ec a; ( ) = ( ) = − ( )=3 + − (1.17) ( )=2 ( ) = + ( ) = (2 + ) ( ) = 3 Q(1.18)ν so that apart for a global normalization factor the charges do depend just on one parameter,4 whichQ ec mustQ bee fixedQ byu some extraQ uc assumption.Q d Let’Qs saydc web; require ( ) = 0 , that readily implies− − − 3 3 b ( ) = ( ) = 2 ( ) = 2 ( ) = 3 ( )=3 ( ) = (1.19) i.e. the electric charge of the SM fermions is a multiple of 2 . SU Let us consider now the issue of anomalies. We already know that in the SM SU all the gauge anomalies vanish. This property is preserved in (5) since 5 and 10⊃ SU U A SU U A U B have equal⊗ and⊃ opposite⊗ anomalies,⊗ so that the theory is still anomaly free. In order to see this explicitly let us decompose; 5 and; 10 under; the; branching chain (5) (4) (1) (3)⊕ − (1) (1)⊕ − ⊕ − − ; ; ; ; ; 5 = 1(4) ⊕4( −1) = 1(4 0) ⊕1( 1 −3) ⊕3( −1 −1) ⊕ − − (1.20) U R 10 = 4(3) 6( 2)R = 1(3 3) 3(3 1) 3( 2 2) 3( 2 2) (1.21)A where the (1) charges are givena upb to ac normalizationabc factor. The anomaly ( ) TR;TR TR R d ; relative to a representation is{ defined} by A dabc Tr = ( ) (1.22) R R R R R R ; where is aA completely⊕ symmetricA A tensor. Then, givenA the propertie−A s 1 1 1 2 SU SU ( ) = ( ) + ( ) and ( ) = ( ) (1.23) SU SU ; SU SU SU SU ; it is enoughA to computeA the anomalyA of theA (3) subalgebraA of A(5), (3) (3) (3) (3) (3) (3) (5) = (3) A ⊕ (10) = (3) + (3) + (3) (1.24) in order to conclude that (5 10) = 0. We close this section by noticing that anomaly cancellation and charge quantization are closely related. Actually it is not a chance that in the SM anomaly cancellation im- plies charge quantization, after taking into account the gauge invariance of the Yukawa couplings4 [6, 7, 8, 9, 10]. SU L H H U Q Q H ⊂ That ish neededi in order to give mass to the SM fermions with the Higgs mechanism. The simplest possibility is given by using an (2) doublet 5 (cf. Sect. 1.2.6) and in order to preserve (1) it must be ( )=0. SO
Chapter 1. From the standard model to 1.2.224 Gauge coupling unification (10)
MU SU ′ g g g SU C SU L U Y ⊗ ⊗ ′ At someg grandU unification mass scale theg relevantg g symmetry is (5) and the 3 2 , , coupling constants of (3) (2) (1) merge into one single gauge 3 2 coupling . The rather different values for , , at low-energy are then due to ′ renormalization effects.g g Before considering the running of the gauge couplings we need to fix the relative 2 g T g ′Y: normalization between and , which enter the weak interactions 2 3 + Y (1.25) ζ ; We define T2 Y ζ− / Y T Tr 2 = 3 (1.26) ≡ Tr′ 1 2 g Y g Y 1 3 so that is normalized as . In a unified theory based on a simple group, 1g 1 ζg ′ : the coupling which unifies is then ( ≡= ) 1 SUp (1.27)
Evaluating the normalization overζ a 5 of (5)− one finds; 2 2 1 1 3 − 2 3 2 +2 2 5 = 1 1 = (1.28) 2 + 2 3
gU g MU g MU g MU : and thus one obtains the tree≡ level matching condition
µ log10HΜGeVL 5 10 15 18 U Y SU Figure 1.1: One-loop running of the SM gauge couplings assuming the (1) embedding into (5). As weMU can see, the gauge couplings do not unify in the minimal framework, although a small perturbation may suffice to restore unification. In particular, thresholds effects at the scale5 (or below) may do the job, however depending on the details of the UV completion . ÷ By now Fig. 1.1 remains one of the most solid hints in favor of the grand unification idea. Indeed, being the gauge coupling evolution sensitive to the log15 of18 the scale, it is intriguing that they almost unify in a relatively narrow window, 10 GeV, which is still allowed by the experimental lower bound on the proton lifetime and a consistent 1.2.3effective Symmetry quantum field breaking theory description without gravity. H H ⊕SU SU C SU L U Y SU C U Q The Higgs sector of the Georgi-Glashow⊗ ⊗ model spans over the reducible⊗ 5 24 representation. These two fields are minimally needed in order to break the (5) H gauge symmetryh downi to (3) (2) (1) and further to (3) (1) . Let us concentrate on the first stage of the breaking which is controlled by the rank- conserving VEV 24 . The fact thati the adjoint preservesi the rank is easily seen by δ H T ; H ; considering the action of the Cartanh ij generatorsh on thei j adjoint vacuum Cartan 24 = [ 24 ] (1.32) i U† i Ul k : derived from the transformation propertiesj Ï k ofj thel adjoint H SU h i 24 ( ) 24 (1.33) Since 24 can be diagonalized by an (5) transformation and the Cartan genera- tors5 are diagonalSU by definition, one concludes that the adjoint preserves the Cartan It turns out that threshold corrections are not enough in order to restore unification in the minimal Georgi-Glashow (5) (see e.g. Ref. [81]). SO Chapter 1. From the standard model to 26 (10) V H m λ λ ; subalgebra. The scalar potential− is givenH by H H 2 2 2 2 4 1 2 H H (24 ) = Tr24 + Tr24 + Tr24 Ï − (1.34) H h i where just for simplicity we have imposed theSU discrete symmetry 24 24 . The minimization of the potential goes as follows. First of all 24 is transformed into a H h ; h ; h ; h ; h ; real diagonal traceless matrixh byi means of an (5) transformation 1 2 3 4 5 h h h h h 24 = diag( H ) (1.35) 1 2 3 4 5 where + + + + = 0. With 24 in the diagonal form, the scalar potential V H m h λ h λ h : reads − i i 2 i 2 i 2 i 2! i 4 X 1 X 2 X hi (24 ) = + + (1.36)µ i hi ′ V H V H µ H Since the ’s are− not all independent,P we need to use the lagrangian multiplier in order to account for the constraint = 0. The minimization of the potential ′ (24 ) = (24 ∂V) Tr24H yields m hi λ hj hi λ hi µ : ∂hi − − 2 j 2 3 (24 ) 1 X 2 = 2 +4 +4 =0 (1.37) hi λ x λ a m x µ a h : Thus at the minimum all the − ’s satisfy− the same cubic equation j 3 2 j 2 2 1 X 4 + 4 2 = 0 with = (1.38) hi φ φ φ x 1 2 3 This means that the ’s can take at most three different values, , and2 , which are the three roots of the cubic equation.φ φ Noteφ that: the absence of the term in the cubic equation implies that n n n 1 φ 2 φ 3 φ H + + =0 h i (1.39) 1 2 3 1 2 3 H φ ;:::;φ ;:::;φ n φ n φ n φ : Let ,h andi the number of times , and appear in 24 , 1 2 3 1 1 2 2 3 3 24H = diag( SU n )SU withn SU n + + =0 (1.40) h i ⊗ ⊗ SU n SU n SU n SU n 1 2 3 Ï ⊗ ⊗ Thus 24 is invariantU under ( ) ( ) ( ) transformations.H This implies h i1 2 3 that the most general form of symmetry breaking is ( ) ( ) ( ) ( ) as well as possible (1) factors (total rank is 4) which leave 24 invariant. To find the absolute minimumn φ wen φ haven toφ use the relations φ φ φ 1 1 2 2 3 3 1 2 3 + + = 0 and + + =0 (1.41) 1.2. The Georgi-Glashow route n ; n ; n 27 { } V H 1 2 3 to compare different choices of in order to get the one with the smallest (24 ). It turns out (see e.g. Ref. [45]) that for the case of interest there are two possible SU SU SU U SU SU U ; patterns for the symmetryÏ ⊗ breaking⊗ Ï ⊗ λ λ (5) (3) (2) (1) or (5) (4) (1) (1.42) λ > λ > 1 2 depending on the relative magnitudes of the parameters and . In particular for 1 2 H V ; ; ; ; : 0 and 0 the absoluteh i minimum is given− by− the SM vacuum [45] and the adjoint VEV reads 24 = diag(2 2 2 3 3) (1.43) Then the stabilityλ of theH vacuumλ requiresH > λ > λ h i 2 h i ÍÑ − 2 4 1 2 1 7 2 Tr 24 + Tr 24 0 (1.44) 30 ∂V H h i V m V λ λ and the minimum∂V condition ÍÑ − 2 2 ( 24 ) 1 2 =0 60 +2 (30 +7 ) =0 (1.45) m V : yields λ 2 λ 2 = 1 2 (1.46) 2(30 +7 ) D ∂ ig A ; ; Let us now write the covariantµ derivativeH µ H µ H Aµ H × 24 = 24 + 24 (1.47) µ † µ Dµ H D H g Aµ; H H ; A where and 24 areh 5 i5 tracelessh i hermitian matrices.h i h Theni from the canonical kinetic term, 2 i i Tr 24 24 =H Trhj δ ; 24 [ 24 ] (1.48) h ij j and the shape of the vacuum 24 = (1.49) where repeated indices are noti summed,µ wej can easilyi extractµ j the gauge bosons mass g Aµ; H H ; A g Aµ A hi hj : matrix from the expressionh i j h i i j i − 2 2 2 i A24µ j [ 24 i ;] =; ( j) ( ;) ( ) MX (1.50)g V i;j ; ; i;j ; 2 2 2 The gauge boson fields ( ) havingEq:=1 2 3 and =4 5 are massive, = 25 , while =1 2 3 and =4 5 are still massless. Notice that the hypercharge− genera- SU /SU C SU L U Y tor commutes with the vacuum in (1.43) and hence the associated⊗ gauge boson⊗ is massless asM well.X The number of massive gauge bosons is then 24 M(8+3+1)U = 12 and their quantum numbers correspond to the coset (5) (3) (2) (1) . Their mass is usually identified with the grand unification scale, . SO Chapter 1. From the standard model to 1.2.428 Doublet-Triplet splitting (10) SU C SU L U Y SU C U Q ⊗ ⊗ Ï ⊗ H T The second breaking step, (3) (2) (1); (3) (1) , is driven by a H H 5 where T 5 = SU L H (1.51) H decomposesH into a color triplet and an (2) doublet . The latter plays the same role of the Higgs doublet of the SM. The most general potential containing both 24 V V V V ; ; and 5 can be written as H H H H V H = (24 ) + (5 ) + (24 5 ) (1.52) † † where (24 ) is defined inV Eq.H (1.34), µ H H λ H H ; − 2 2 (5 ) = † 5 5 + 5 5 † (1.53) V H ; H α H H H β H H H : 2 2 and H H (24 5 ) = 5 5 Tr24H + 5 24 5 T Ï − (1.54) Again we have imposed for simplicity the discrete symmetry 24 24 . It is in- structive to compute the mass of the doublet and the triplet in the SM vacuum M µ α β V ; M µ α β V : just after the firstH stage− of the breaking T − 2 2 2 2 2 2 M M H = X+ (30 W+9 ) = + (30 +4 ) (1.55) ≫ W Z TheMW gauge hierarchy requires that the doublet , containing the would-be M /M M GoldstoneO bosonsX W eaten∼ by the and the and theH physical Higgs boson, live at the scale. This2 2 is unnatural26 and can be achieved at2 the prize of a fine-tuning ofM oneT part in ( ) 10 in the expression for . If we follow the principle that only the minimal fine-tuning6 needed for the gauge hierarchy is allowed then is automatically kept heavy . This goes under the name of doublet-triplet (DT) splitting. Usually, but not always [84, 85], a light triplet is very dangerous for the proton stability since it can couple to the SM fermions in such a way that baryon7 number is not anymore an accidental global symmetry of the low-energyMX lagrangianMW . A final comment about the radiative stability of the fine-tuning is in order. While su- persymmetry6 helps in stabilizing the hierarchy between and against radiative In some way this is an extension of the Georgi’s survival hypothesis for fermions [79], according to which the particles do not survive to low energies unless a symmetry forbids their large mass terms. This hypothesis is obviously wrong for scalars and must be extended. The extended survival hypothesis (ESH) reads: Higgs scalars (unless protected by some symmetry) acquire the maximum mass ∗ compatible with the pattern of symmetry breakingqqT [82].qℓT In practice this corresponds to the requirement ofT the7 minimal numberU of fine-tuningsB to be imposed onto the scalar potential [83]. Let us consider for instance the invariants and . There’s no way to assign a baryon charge to in such a way that (1) is preserved. 1.2. The Georgi-Glashow route 29 corrections, it does not say much about the origin of this hierarchy. Other mecha- nisms have to be devised to render the hierarchy natural (for a short discussion of the solutions proposed so far cf. Sect. 4.4.3). In a nonsupersymmetric scenario one needs to compute the mass of the doublet in Eq. (1.55) within a 13-loop accuracy in order to 1.2.5stabilize Proton the hierarchy. decay SU /SU C SU L U Y The theory predicts that⊗ protons⊗ eventually decay. The most emblematic contribution SUto proton decay is due to the exchange of super-heavy gauge bosons which belong to αβ αi ij the coset (5) (3) (2)ψα; ψi ;(1) . Let usψ denote; ψ ; ψ the matter; representations of (5) as SU C SU L 5 = ( ) 10 = (1.56) where the greek and latin indices runα respectivelyi α i fromα i 1 to 3 ( (3) space) and 1 to Xβ ;Xj ;Xα Xi ;Xi ;Xα ; 2 ( (2) space). Analogously the adjoint 24− can be represented as 3 2 ; ; 24; = (1.57) ; ; ⊕ ⊕ Xα from which⊕ we can readily recognize the gauge bosons associai ted to the SM unbroken generatorsα ((8 1) c(3 1) (1 1))βi and theβα two super-heavyc ij leptoquarkαj c gauge bosons Xi ψα ψi d ν;e ; ψ ψ d;u u ; ψ ψ e u;d : ((3 2) (3 2)).Ï Let usÏ consider nowÏ the gauge actionÏ of onÏ the matterÏ fields α Xi : ( ) ( ) ( ) (1.58) ud ucec ; Thus diagrams involving the exchange ofÏ a boson generate processes like (1.59) p π e whoseÏ amplitude is proportional to the gauge boson propagator. After dressing the operator0 + with a spectator quark u, we can have for instance the low-energy process αU mp , whose decay rate can bep estimatedπ e by simple: dimensional analysis Ï ∼ M2 X 5 0 + 4 − τ p π e > : Γ( ) αU (1.60) → × 0 + 33 1 Using ( ) 8 2 10 MyearsX > [11]: we extract (for = 40) the naive lower bound on the super-heavy gauge boson mass× 15 2 3 10 GeV (1.61) B L p π e which points directly− to the grand unification scale extrapoÏ lated by the gauge running (see e.g. Fig. 1.1). 0 + Notice that Bis conservedL in the process . This selection rule is B L Xα −/ a general feature− ofi the gauge induced proton decay and can be traced back to the presence of a global accidental symmetry in the transitions of Eq. (1.58) after assigning ( )=2 3. SO Chapter 1. From the standard model to 1.2.630 Yukawa sector and neutrino masses (10) SU 8 ∗ The (5) Yukawa lagrangianY F Y canF be written schematicallyF Y F H as ; L H 5 1 5 10 = 5 10 5 + 10 10 5 + h.c. SU (1.62) 8 5 where is the 5-index Levi-Civita tensor. After denoting the (5) representations dc uc q T synthetically as ; F ℓ F qT ec H H −3 5 = 2 10 = 2 iσ 5 = (1.63) 3 2 2 where is the 3-index Levi-Civitac tensor and ∗ = , we project Eq. (1.62) over the ∗ c T u q T c ∗ c ∗ SM components.F Y F Thisd yieldsℓ d Y qH ℓY e H ; H qT ec H∗ Ï −3 5 2 5 5 2 5 10 5 = c T + (1.64) F Y F H u Y Y qH : Ï 1 5 10 SU 1 10 10 10 10 5 L + (1.65) ℓY ecH∗8 ecY T ℓH∗ 2 After rearranging the order of the (2) doublet and singlet fields in the second term 5 Y 5 Y T Y Y T ; of Eq. (1.64), i.e. = d e , one gets u u = and = (1.66) mb MU mτ MU whichms MU showsm aµ M deepU connectionmd MU betweenme M flavorU and the GUT symmetry (which is not related to a flavor symmetry). The first relation in Eq. (1.66) predicts ( ) = ( ), mb MU mτ MU ( ) = ( ) and ( ) = ( ) at the GUT scale. So in order− to test this relation one has to run the SM fermion masses starting from their low-energy values. While ( )m =d/me ( )m iss/m obtainedµ in the MSSM with a typical 20 30% uncertainty [86], the other two relations are evidently wrong. By exploiting the fact that the ratio between and is essentially independent of renormalization m /m m /m ; effects [87], we get the scale free relationd s e µ = (1.67) md me H SU h i which isSU off by one order of magnitude. Notice that = comes∗ from theαx fact thatmn the fundamentaly ∗ β 5 breaks (5) F Y F H F m Cxy Y F αβn H F Y F H downαβγδ to x (4) whichmn remainsy an≡ accidental symmetry of the Yukawa sector. So one≡ 8 F αβm Cxy Y F γδn H α;β;γ;δ; m; n x;y SU 5 5 5 10 More precisely 5 10 5 5 ( ) (10 ) (5 ) and 10 10 5 10 (10 ) ( ) (10 ) (5 ) , where ( ), ( ) and ( ) are respectively (5), family and Lorentz indices. 1.2. The Georgi-Glashow route 31 SU H expects that considering higher dimensional representations makes it possible to fur- ther break the remnant (4). This∗ is indeed what happens: by introducing a 45 which couples to the fermionsF inF theH followingF F wayH [88] Yd Ye H H 5 10 −45 + 10 10 45 + h.c. (1.68) The first operator leads to = 3 , so that if both 5 and 45 are present more ∗ freedom is available to fit all fermionF masses.F H AlternativelH ; y one can built an effective coupling [89] h i 1 H 45 5 10 ( 24 5 ) (1.69) MP b τ − Λ which mimics the behavior of the 45 . If we take the cut-off to be the planckMd scale Me , this nicely keeps unification while corrects the relations among the first two families. However in both cases we loose predictivity since we are just fitting and U G G F in the extended Yukawa structure. − G FinallyF what aboutG H neutrinos? It turnsH out [90] that the Georgi-Glashow model3 h i 5 has an accidental1 global (1)2 symmetryG with the charge assignmentSU (5 ) = , (10 ) = + 5 and (5 ) = +G 5 . TheY VEV 5 breaks this global symmetry but leaves invariant a linear combinationG Y of4 and a Cartan generator of (5). It easyB toL see that 5 − any linear combination of 4+ , Q, and any color generators is left invariant. The B L 5 extra conserved charge −+ when acting on the fermion fields is just . Thus neutrinos cannot acquire neither a Dirac (because of the field content) nor a Majorana (because of the global symmetry) mass term and they remainU exactlyG massless even at the quantum level. Going at the non-renormalizable level we can break the accidental (1) symmetry. For instance global charges are expected to be violated by gravity and the simplest effective operator one can think of is [91]F F H H : MP 1 5 5 5 5 mν MW(1.70)/MP − ∼O ∼ 2 However5 its contribution to neutrino masses is too much suppressed ( ( ) 10 eV). Thus we have to extendSU the field content of the theoryF in order to generate phenomenologicallyF viableF H neutrino masses.F Actually, the possibilities are many. Minimally one may add an (5) singletSU fermion field 1 . Then, through its renor- malizable couplingMP 5 1 5 , one integrates 1 out and generates an operator similar to that in Eq. (1.70), but suppressed by the (5)-singlet mass term whichU G can be taken well below . H A slightly different approach could be breaking the accidental (1) symmetry by adding additional scalar representations. Let us take for instance a 10 and consider f F F H M H H H : then the new couplings [90]L ⊃ 10 5 5 10 + 10 10 5 (1.71) SO Chapter 1. From the standard model to 32 (10) G F G H G H − U G B L f 3 2 − SinceM (5 ) = 5 and (5 ) = + 5 there’s no way to assign a -charge to 10 in order to preserve (1) . Thus we expect that loops containing the breaking sources and can generate neutrino masses. So what is wrong with the two approaches above? In principle nothing. But maybe we should try to do more than getting out what weSU put in. Indeed we are just solving the issue of neutrino masses "ad hoc", without correlations9 to other phenomena. In addition we do not improve unification of minimal (5) . Guided by this double issue of the Georgi-Glashow model, two minimal extensions which can cureH at the; H same; timeH both; H neutrino masses; H and unification have been • ⊕ ⊕ recently proposed ; H Add a 15 = (1 3) (6 1) (3 2) [43]. Here (1 3) is an Higgs triplet respon- sible for type-II seesaw. The model predicts generically light leptoquarks (3 2) F ; F ; F ; F ; F ; F ; F ; F • and fast proton decay⊕ [92].⊕ ⊕ ⊕ Add a 24 = (1 1) (1 3); F(8 1) (3 2) (3 2) [44]. Here (1 1) and (1 3) are fields responsible respectively for type-I and type-III seesaw. The model predicts a light fermionSU triplet (1 3) and fast proton decay [93]. Another well motivated and studied extension of the Georgi-Glashow model is given by supersymmetricR (5) [94]. In this case the supersymmetrization of the spectrum is enough in order to fix both unification and neutrino masses. Indeed, if we do not impose by hand -parity conservation Majorana neutrino masses are automatically 1.3generated The by lepton Pati-Salam number violating route interactions [95]. V A In the SM there is an intrinsic− lack of left-right symmetry without any explanation of the phenomenological facts that neutrino masses are very small and the weak interactions are predominantlyu u u . The situationν can be schematicallydc dc dc depicteddc inec the following q ℓ way d d d e uc uc uc uc 1 2 3 1 2 3 c = (c ) c q= ; 1 ; 2 ℓ 3 ; ; = d ; ; u 1 2; ; 3 e ; (1.72); − = ( −) ? SU C SU L U Y ⊗ 1 ⊗ 1 1 2 where = (3 2 + 6 ), = (1 2 2), = (3 1 + 3 ), = (3 1 3) and = (1 1 +1) underd (3) (2) (1) . Considering the SM as an effectiveY theory, neutrino masses can be generated by a ν ℓT H C HT ℓ ; = 5 operator [25] of the type L 2 2 ( ) ( ) (1.73)H 9 Λ An analysis of the thresholds corrections in the Georgi-Glashow model with the addition of the 10 indicates that unification cannot be restored. 1.3. The Pati-Salam route iσ C 33 T H v Mνν Cν 2 h i 2 where = and is the charge-conjugation matrix. After electroweak symmetry v breaking, = , neutrinos pick upM aν MajoranaYν : mass term with 2L = √ matm :(1.74) Λ ∼ The lower bound on the highest neutrino eigenvalue÷ inferred from ∆ 0 05 eV L . Yν : tells us that the scale at which the leptonO number is violated is 14 15 Y Λ (10 GeV) ν (1.75) L σi ab σi cd δadδcb δabδcd Notice that without a theory which fixes the structure of we don’t− have much to say about Λ . Actually, by exploiting the Fierz identity ( ) ( ) =2 , one finds that T T T T T T the operatorℓ H inC Eq.H (1.73) ℓ canℓ beC equivalentlyσiℓ H σiH written inℓ thrσieeH C differentH σiℓ ways: − 2 2 1 2 2 2 2 ( ) ( ) = ( )( ) = ( ) ( ) (1.76) 2 Each operator in Eq. (1.76) hints to a different renormalizable UV completion of the SM. Indeed one can think those effective operators as the result of the integration of ℓT H Cνc ℓT C σ ℓ ℓT σ H CT ; an heavy state with a renormalizable couplingi ofi the type i i c 2 2 2 ν i Ti Y Y ( ) ( )∆c ( ∗ ) (1.77) Y ν i Ti ⊕ i where , ∆ and are a fermionic singlet ( = 0), a scalar triplet ( = +1) and a fermionicL triplet ( = 0). Notice that being , ∆ ∆ and vector-like states their mass is not protected by the electroweak symmetry and it can be identified with the scale Λ , thus providing a rationale for the smallness of neutrino masses. This goes under the name of seesaw mechanism and the three options in Eq. (1.77) are classified 1.3.1respectively Left-Right as type-I [26,symmetry 27, 28, 29, 30], type-II [31, 32, 33, 34] and type-III [35] seesaw. c Guidedν by the previous discussion on the renormalizable origin of neutrino masses, it is then very natural to to fill the gap in the SM by introducing a SM-singlet fermion field . In such a way the spectrum looks more "symmetric" and one can imagine that at higher energies10 the left-right symmetry is restored, in the sense that left and right chirality fermions are assumed to play an identical role prior to some kind of spontaneous10 symmetry breaking. ψ ψcT C ψc Cγ ψ∗ As alreadyR ≡ L stressed we work inL a≡ formalismR in which all the fermions are left-handed four com- ponents Weyl spinors. The right chirality components are obtained by means of charge conjugation, 0 namely or equivalently . SO Chapter 1. From the standard model to 34 (10) SU C SU L SU R ⊗ ⊗ ⊗ U B L Z Z SU L − ⊗ ↔ SU TheR smallest gauge group that implement this idea is (3) (2) (2) 2 2 (1) [2, 77, 78], where is a discrete symmetry which exchange (2) u u u ν dc dc dc ec q(2) . The field contentℓ of the theoryqc can be schematically depicteℓc d as d d d e uc uc uc νc 1 2 3 − 1 − 2 − 3 − = q 1 ;2 ; ;3 ℓ= ; ; ; = qc 1 ; ; 2 ∗; 3 ℓc =; ; ∗; (1.78) − − SU C SU L SU R U B L ⊗ ⊗ 1 ⊗ − 1 where = (3 2 1 + 3), = (1 2 1 1), = (3 1 2 3 ), = (1 1 2 +1), under (3) (2) (2) (1) . Given this embedding of the fermion fields one B L readily verifies that the electricQ chargeTL formulaTR takes− : the expression 3 3 = + + (1.79) 2 ∗ Next we have to state theδL Higgs; sector.; ; Inδ theR early; ; days; of the development; ; ; of left-right theories the breaking to the SM was minimally achieved by employing the following set of representations: = (1 2 1L +1), ; ; =; (1 1 2 +1)R and Φ; =; (1; 2 2 0) [2, 77, 78].δL However,δR as pointed out in [30, 34], in order to understand the smallness of neutrino i masses it is better to consider ∆ = (1 3 L;R1 +2) andL;Rσi ∆/ = (1 1 3SU+2)L;R in place of and . σ ∗σ ≡ ChoosingSU the matrixL SU representationR ∆ = ∆ 2 for the (2) adjoint and ˜ 2 2 defining the conjugate† doublet Φ Φ† , the transformation† properties for† the Higgs L UL L U ; R UR R U ; UL U ; UL U ; fields underÏ (2)L and Ï(2) read R Ï R Ï R ∆ ∆ ∆ ∆ Φ Φ Φ˜ Φ˜ (1.80) and consequentlyδL L TL; weL haveδL R δL TL δL TL 3 3 3 δR∆L = ∆ δR∆R =0TR; R δRΦ = ΦTR δRΦ˜ = Φ˜TR − − 3 3 3 δB−∆L =0L L δB−∆L =R ∆ R δB−ΦL = Φ δB−Φ˜L = Φ˜ : (1.81) ∆ = +2∆ ∆ = +2∆ Φ=0 Φ=0˜ Then, given the√ expression for the electric charge operator in Eq.∗ (1.79), we can de- compose these fields in the charge eigenstates L;R / √ φ− φ φ − −φ∗ + ++ L;R ; 0 + ; 0 + : − / φ1 φ1 −φ2 φ 2 ∆ 0 2 ∆+ 0 0 ∆ = Φ= 2 2 Φ=˜ 1 1 (1.82) ∆ ∆ 2 Z 2 In order to fix completely the theory one has to specify the action of the sym- metry on the field content. There are two phenomenologically viable left-right discrete 1.3. The Pati-Salam route ZP ZC 35 c 2 ψL2 ψR ψL ψL symmetries: and .ÎÏ They are defined as ÎÏ ∗ P L R C L R Z ÎÏ † Z ÎÏ T : µ ÎÏ µ µ ÎÏ µ∗ 2 W∆L W∆ R 2 W∆L W∆R : ÎÏ and : ÎÏ (1.83) Φ Φ Φ Φ CP ZP ZC c ∗ PTheψ implicationsL ψR C of thisψL twoψL casesCγ differψR by the tiny amount of violation. Indeed Ï Ï ≡ 2 2 when restricted to the fermion fields we can identifyCP and respectively with P C 0 Z : Z and : . In the former case the Yukawa matrices are C hermitian whileZ in the latter they are symmetric. So if is conserved (real couplings) 2 2 C and lead to the same predictions. Z 2 Notice that involves an exchange between spinors with the same chirality. In 2 principle this would allow the embedding of into a gauge symmetry which com- C mutes with the Lorentz group. TheZ gauging is conceptually important since it protects theSO symmetry from unknown UV effects. 2 Remarkably itSO turns out that can be identified with a finite gauge transformation C of (10) which, historically,Z goes under the name of D-parity [96, 97, 98, 99, 100]. The connection with (10) motivates our notation in terms of left-handed fermion fields 2 which fits better for the case. Let us consider now the symmetry breakingv sector. From Eq.v (1∗.82) we deduce that the SM-preservingL;R vacuum directions are ; ∗ : h i vL;R h i v v 1 2 0 0 0 D E 0 ∆ = Φ = 2 Φ˜ = 1 (1.84) 0 0 0 vL vR The minimization of the scalar potential (see e.g. Appendix B of Ref. [34]) shows that be- vL vR ; vLvR γv ; v ; side the expected6 left-right symmetric minimum = , we have also the asymmetric one γ 2 1 2 ∗ P = = (in theL approximationR L R = 0) (1.85)Z C ↔ ↔ whereZ is a combination of parameters of the Higgs potential. Since the discrete 2 left-rightvR symmetryv vL is defined to transform ∆ ∆ (∆ ∆ ) in the case of 2 ≫ ≫ ( ), the VEVs in Eq. (1.85) breaks it spontaneously.v Phenomenologically we have to 1 R SU C SU L SU R U B L Z SU C SU L U Y require ⊗ ⊗ which⊗ leads to− the⊗ followingÊÏ breaking⊗ pattern⊗ v ≫vL 2 SU C U Q ; (3) (2) (2) (1) (3) ÊÏ1 (2) ⊗(1) MW ;MZ MW ;MZ (3)R R ≫(1) L (1.86)L MWR MZR where the gauge hierarchy is set by the gauge boson masses . i i B L Let us verifyD thisµ R by computing∂µ R igR TR; andR AR .µ Weig startB−L from− theR A covariantB−L µ ; derivative ∆ = ∆ + ∆ + ∆ ( ) (1.87) 2 SO Chapter 1. From the standard model to 36 (10) † µ Dµ R D R ; and the canonically normalized kinetich termi h i Tr ( ∆ ) ∆ (1.88) M g v ; M g g v ; M ; which leads to WR R R ZR R B−L R Y 2 2 2 2 2 2 2 = = 2( + ) =0 (1.89) ± AR iAR gRAR gB−LAB−L gB−LAR gRAB−L whereWR ∓ ; ZR ; Y − : 1 √ 2 3 3 gR gB−L gR gB−L + = = q 2 2 = q 2 2 (1.90) − − − 2 gY gR gB−L + Z + gR gL g 2 2 2 11 ≡ 2 Given the relation = + gand the symmetry in Eq. (1.83) which implies = , we obtain MZR MWR : MWR : g 2gY ∼ 2 − 2 2 22 2 = 2 6 L (1.91) v h i 6 h i 6 At the next stage of symmetry breaking ( Φ = 0 and ∆ = 0) an analogous calculation M g v v ;2 M g g v v ; M ; yields (inWL the approximationL = 0)ZL Y L A 2 2 2 2 2 2 2 2 2 2 1 1 1 1 = +2 = + +4 =0 (1.92) 2 2 ± AL iAL gLAL gY AY gY AL gLAY where WL ∓ ; ZL − ; A : 1 √ 2 g3 L gY g3 L gY + = ρ = p 2 2 = p 2 2 (1.93) 2 + + M g g Notice that in order to preserve ρ= 1 atW treeL level,Y ; where 2 g 2 ≡ MZL 2 2 +2 vL v (1.94) ≪ MWL MWR SU L U Y 1 ⊗ one has to require . On the other hand at energy scales between and , (2) (1) is still preserved and Eq. (1.79) implies T B L : R − − 3 1 R SU R T∆R = ∆( ) B L (1.95) 2 − B L 3 Since ∆ is an (2) triplet ∆ = 1 andB−L we get a violation of by two units. Y TR Then11 two classes of and violating3 processes can arise: This relation comes directly from = + 2 (cf. Eq. (1.79)). For a formal proof see Sect. 2.2.4. 1.3. The Pati-Salam route B L 37 • B L • ∆ = 0 and ∆ = 2 which imply Majorana neutrinos. ∆ = 2 and ∆ = 0 which lead to neutron-antineutron oscillations. Let us describe the origin of neutrino masses while postponing the discussion of neutron-antineutronT oscillationsc T to thec next section.T cT ∗ c ν Y ℓ C ℓ Y ℓ C ℓ Y ℓ C Lℓ ℓ C R ℓ ; TheL piece⊃ of lagrangian relevant for neutrinos is Φ 2 Φ 2 SU ∆ L SU2 R 2 Φ + ˜ Φ˜ + ⊗ ∆ + ∆ + h.c. c (1.96)c ℓ UL ℓ ℓ UR ℓ T † Ï Ï The invarianceUL;R ofU Eq.L;R (1.96) under the (2) (2) might not be obvious. So let us recall that, on top of the transformation properties in Eq. (1.80), , , 2 2 and = T. Afterc projectingT c Eq. (1.96)T on the SMcT vacuumc ∗ directions and ν Y ν Cν v Y ν Cν v Y ν Cν vL ν Cν vR : taking onlyL the⊃ pieces relevant to neutrinos we get Φ 2v Φ 1 ∆ ˜ + ν νc+ + + h.c. (1.97) 2 Let us take for simplicity = 0 and consider real parameters. Then the neutrino Y vL Y v mass matrix in the symmetric basis ( T ) reads ; Y v Y vR ∆ Φ 1 ˜ vR v vΦL 1 ∆ (1.98) ≫ ≫˜ 1 and, given the hierarchy ρρ,T the matrixρ in Eq. (1.98) is block-diagonalized by a similarity transformation involving− T the orthogonalT ; matrix ρ1 ρ ρ −2 − − 1 1 ρ Y Y v /vR 2 ρ (1.99) 1 O 1 2 Φ ∆ 1 − T v where = ˜ . The diagonalizationmν Y vL Y isY validY up to: ( ) and yields − vR2 1 1 ∆ Φ ∆ Φ = ˜ ˜ (1.100) vL γv /vR The two contributions go under the name12 of type-II and type-I seesaw respectively.2 vL 1 From12 the minimization of the potential (see Eq. (1.85)) one gets = and Even without performing the complete minimization† we can† estimate† the induced VEV by looking V M L λ R : at the following piece of potential ⊃− L L L 2 ∆ Tr ∆ ∆ + Tr ∆ Φ∆˜ Φ (1.101) V M vL λvLvR v ; On the SM-vacuum Eq. (1.101) reads h i⊃− L | | 2 2 2 vL 1 ∆ + (1.102) vR v and from the extremizing condition with respectvL λ to | we| : get M 2 1L 2 = ∆ (1.103) SO Chapter 1. From the standard model to 38 (10) − T v hence the effective neutrinom massν matrixY γ readsY Y Y : − vR2 1 1 ∆ Φ ∆ Φ = ˜ ˜ (1.104) V A vR V A mν ThisÏ∞ equation is crucial since− it shows a deep connection between the smallness of neutrino masses and the non-observation of + currents [30, 34]. Indeed in the limit we recover thevR structure and vanish. Nowadays we know that neutrino are massive, but this information is not enough in order to fix the scale because theV detailedA Yukawa structures are unknown. In this respect one can adopt two complementary approaches. From a pure phenomenological P C point of view one can hope that the + interactions are just behind the13Z cornerZ and experiments such usM theWR LHC are probing rightKL nowKS the TeV region . Depending P − C 2 2 MonW theR & choice of the discreteZ left-rightMWR symmetry& : which can be eitherZ or , the strongest bounds on are given by the mass difference which yields 2 2 4 TeV in the case of and 2 5 TeV in the case of [101, 102]. Alternatively one can imagineSO some well motivated UV completion in which the Yukawa structure of the neutrino mass matrix is correlated to thatv ofR the charged fermions. For instance in (10) GUTs it usually not easy to disentangle the highest eigenvalue in Eq. (1.104) from14 the top mass. This implies that the scaleSO must be very heavy, somewhere close to 10 GeV. As we will see in Chapter 2 this is compatible with unification constraints and strengthen the connection between (10) and neutrino 1.3.2masses. Lepton number as a fourth color One can go a little step further and imagine a partial unification scenario in which quarks and leptons belong to the same representations. The simplest implementation u u u ν dc dc dc ec is obtainedQ by collapsing the multipletsQc in Eq. (1.78) in the following way d d d e uc uc uc νc 1 2 3 − 1 − 2 − 3 − 1 2 3 SU =C U B−L SU C = 1 2 3 Q (1.105); ; c ⊗ ∗ ⊂ Q ; ; SU C SU L SU R ⊗ ⊗ so that (3) (1) (4) and the fermion multiplets transform as = (4 2 1) SU L SU R and ↔= (4 1 2 ) under (4) (2) (2) , which is known as the Pati-Salam group [2]. Even in this case one can attach an extra discrete symmetry which exchange (2) (2) . L ; ; R ; ; The Higgs sector of the model is essentially an extension of that of the left-right symmetric model presented in Sect. 1.3.1. IndeedO we have ∆ = (10 3 1), ∆ = (10 1 3) 13 It has been pointed out recently [36] that a low (TeV) left-right symmetry scale could be welcome in view of a possible tension between neutrinoless double beta decay signals and the upper limit on the sum of neutrino masses coming from cosmology. 1.3. The Pati-Salam route ; ; ∗ / / 39 ⊕ − ⊕ − SU C SU C U B L ⊃ ⊗ − R and Φ = (1 2 2 ). From the decompositionh i 10 = 6(+2 3) 3( 2 3) 1( 2) under (4) (3) (1) and the expression for the electric charge operator in Eq. (1.79), we can readily see that ∆ contains a SM-single direction and so the first h Ri stage of the breakingSU C isSU givenL bySU R SU C SU L U Y ; ⊗ ⊗ ÊÏ∆ ⊗ ⊗ SU C U Q (4) (2) (2)⊗ (3) (2) (1) (1.106) h i L whileh i≪h the finali breaking to (3) (1) is obtained by means of the bi-doublet VEV Φ . Analogously to the left-right symmetric case an electroweak triplet VEV ∆ Φ is induced by the Higgs potential and the conclusions about neutrino masses are the same. SU C SU C / / A peculiar feature of the Pati-Salam model is that the proton⊕ is stable⊕ − in spite⊕ of SU C U B L the quark-lepton⊗ transitions− due to the (4) interactions. Let us consider first gauge XPS XPS interactions.≡ The adjoint≡ of − (4) decomposes as 15 = 1(0) 3(+4 3) 3( 4 3) 8(0) 4 4 under (3) (1) . In particular the transitionsPS between quark and leptons due 3 g PS µ 3 µ c µ c c µ c to 3(+PS ) and Xµ uγ3( ν ) comedγ e fromXµ theu currentγ ν d interactionsγ e L ⊃ √ + + + G +h.c.G XPS (1.107) 2 c c c− G u G d G ν G e G XPS G u G d G ν 2 c − 3 GIt turnse out thatG Eq.1 (1.107) hasB an accidentalL global symmet2 ry , where ( 1 ) = , −B L 3 3 B L 3 ( ) = ( −) = + , ( ) = ( ) = +1, ( ) = + , ( ) = ( ) = , ( ) = ( ) = 1. is nothing but + when evaluated on the standard fermions. Thus, given that is also a (gauge) symmetry, we conclude that both and are conserved by the gauge interactions. B The situation regardingqqqℓ the scalar interactions is more subtle. Actually in the min- qqqℓ QQQQ Q imal model there is an hidden⊂ discrete symmetry which14 forbids all the ∆ = 1 tran- sitions,ijkl like for instance (seeSU e.g.C Ref. [103] )4 . A simple way to see it is that any operator of the type and the term must be contractedSU C with an tensor in orderQ to form an (4) singlet. However, since the Higgs fields in the minimal model are4 eitherB singlets or completely symmetric in the (4) space, they cannot mediate operators. d On the other hand ∆ = 2 transitions like neutron-antineutron oscillations are R allowed and they proceed throughh =i 9udd operatorsudd ; of the type [103] M R ∆6 ∆ ( )( ) R R(1.108) h i h i which are generated by the Pati-Salam breaking VEV ∆ .SU The fact that ∆ B can be14 pushed down relatively close to the TeV scale without making the proton to decay Notice that this is just the reverse of the situation with the minimal (5) model where ∆ = 2 transitions are forbidden. SO Chapter 1. From the standard model to 40 (10) is phenomenologicallyτN > interesting, since one can hope in testable neutron-antineutronτn−n > oscillations (for32 a recent review see Ref. [104]). Present bounds on nuclear instability ¯ give8 10 yr, which translatesB into a bound on the neutron oscillationd time 10 sec. Analogous limits come from direct reactor oscillations experiments. This sets a lower bound on the scale of ∆ = 2 nonsupersymmetric ( = 9) operators that varies from 10 to 300 TeV depending on model couplings. Thus neutron-antineutron 1.3.3oscillations One probe family scales unified far below the unification scale. The embedding of the left-right symmetric models of the previous sections into a grand unified structureSU requiresSO the presence of a rank-5 group. Actually there are only two candidates which have complex representations and can contain the SM asF a subgroup.F SU U ⊕ These are15 ⊗(6) and (10). The former group even though smaller it is somehow redundant since the SM fermions would be minimally embedded into 6 15 which ; under (5) (1) decompose⊕ − as − ⊕ 6 = 1(+5) 5( 1) and 15=5( 4) 10(+2) (1.109) SO yielding an exotic 5 on top of the SM fermions. SU U SO SO SU SU SU Thus⊗ we are left with⊗ (10). There are essentially two ways of⊗ looking⊗ at this unified theory, accordingSO to the two maximal subalgebras which contain the SM: (5) (1) and (6) (4). The latter is locally isomorphicSO to (4) (2) (2). SU U ; ; ; ; The group theory of − (10)⊕ will be⊕ the− subject of the next⊗ section, but let us alread⊕ y an- SU C SU L SU R ticipate that the⊗ spinorial⊗ 16-dimensional representation of (10) decomposes in the following way 16 = 1( 5) 5(+3) 10( 1) under (5) (1) and 16 = (4 2 1) (4 1 2) under (4) (2) (2) , thus providing a synthesis of both the ideas of Georgi- Glashow and the Pati-Salam. 1.4 O group theory S SO O (10) T (10) is the special orthogonal groupφ of rotationsO inφ a 10-dimOφ ensional vectorφ φ space. Its T Ï Odefining representationOO is given by the group of matrices Owhich leave invariant the normSU of a 10-dimentional real vector . Under , and since is invariant must be orthogonal, = 1. Here special means det = +1 which selects the 15 (6) as a grand unified group deserves anyway attention especially in its supersymmetric version. SU SU The reason is that it has an in-built mechanism in which the doublet-triplet splitting⊗ can be achieved in a very natural way [105, 106]. The mechanism is based on the fact that the light Higgs doublets arise as pseudo-Goldstone modes of a spontaneously broken accidental global (6) (6) symmetry of the Higgs superpotential. SO 1.4. group theory (10) 41O Tij Tji i;j ;::: − group of transformations continuously connected with the identity. The matrices O T ; may be written in terms of 45 imaginary generatorsij ij = , for =1 10, as 1 2 ij = exp (1.110) Tij ab i δa iδbj ; where are the parameters of the transformation.− A convenient basis for the gener- ators is a;b;i;j ;::; [ ] SO ( ) = ( ) (1.111) where = 1 10 and the square bracket16 stands for anti-symmetrization. They Tij ;Tkl i δikTjl δjlTik δilTjk δjkTil : satisfy the (10) commutation relations − − SO = ( + ) (1.112) OOT O In oder to study the group theoryδij of (10) it is crucial to identify the invariant tensors. T TheOO conditions = 1 and det = +1 give rise to two of them. The first one is δij OikOjlδkl OikOjk δij ; simply the Kronecker tensor Ï which is easily proven to be invariant because of = 1, namely ijklmnopqr = = (1.113) while the second one is the 10-index Levi-Civita tensor . Indeed, from the O O O O O O O O O O O definition ofi′j determinant′k′l′m′n′o′p′q′r′ i′i j′j k′k l′l m′m n′n o′o p′p q′q r′r ijklmnopqr O det = ijklmnopqr (1.114) SO and the fact that det = +1, we conclude that is also invariant. The irreducible representations of (10) can be classified into two categories, single-valued and double-valued representations. The single valued representations have the same transformations properties as the ordinary vectors in the real 10- dimensional space and their symmetrized or antisymmetrized tensor products. The doube-valued representations, called also spinor representations, trasform like spinors 1.4.1in a 10-dimentional Tensor representations coordinate space. n SO The generaln -index irreducible representations of (10) are built by means of theφi antisymmetrization or symmetrization (including trace subtraction) of the tensor prod- φi Oij φj ; uct of -fundamental vectors. Starting fromÏ the 10-dimentional fundamental vector , whose transformation rule is SO J ;J iJ J T J T J T (1.115) 16 ≡ ≡ ≡ These are an higher dimensionalTij generalizationTkl of the well known (3) commutation relations 1 2 3 1 23 2 31 3 12 [ ]= , where , and . Then the right hand side of Eq. (1.112) takes just into account the antisymmetric nature of and . SO Chapter 1. From the standard model to 42 (10) δij δij we can decomposeφi φ thej φ tensorj φi productφi ofφj twoφj of themφi inφ thek folloφk wingφk wayφk φi φj ⊗ − ⊗ ⊗ ⊗ − ⊗ ⊗ : ⊗ 1 φA 1 φS Sδ ( ij )+ ( + ij ) + ij = |2 {z } |2 {z 10 } |10 {z } (1.116) A φij S Since theφij symmetry properties of tensors under permutation of the indices are not changed by the group transformations, the antisymmetricO tensor and the symmetric A S tensor clearly do not transform into each other. In general one canφ aijlsoφij separateSδij a tensor in a traceless part and a trace. Because is orthogonal also the traceless/ / − property is preserved− by the group transformations. So we conclude that , and form irreducible representations whose dimensions are respectively 10(10 1) 2 = 45, 10(10 + 1) 2 1 = 54 and 1. One can continue in this way by considering higher order representations and separating each time the symmetric/antisymmetric pieces and subtracting traces. ijklmnopqr However something special happens for 5-index tensorsφ andnopqr the reason has to do with the existence of the invariant i which induces the following duality map when applied to a 5-indexφijklm completelyφijklm antisymmetricijklmnopqr tensorφnopqr : Ï ≡ − φ ˜ ijklm (1.117) 5! This allows us to define the self-dual and the antiself-dual components of in the ijklm φijklm φijklm ; following way ≡ √ 1 Σijklm φijklm + φ˜ijklm : (1.118) ≡ √2 − 1 ˜ ijklm Σijklm ijklm ijklm (1.119) 2 − ijklm ijklm ˜ One verifies that Σ˜ = Σ (self-dual) and Σ = −Σ (antiself-dual). Since the duality property is not changed by the group transformat1 10! ions Σ and Σ do 2 5!(10 5)! 1.4.2form irreducible Spinor representations whose dimension is = 126. SO x ; x ;:::;x x x ::: x We have defined the (10) group by those linear2 transformations2 xi 2 on the coordinates 1 2 10 1 2 10 , suchx thatx the::: quadraticx formγ x γ+ x +::: +γ xis left; invariant. If we write this quadratic form as the square of a linear form of ’s, 2 2 2 2 1 2 10 1 1 2 2 10 10 + + + = ( + + + ) (1.120) γi; γj δij : { } we have to require =2 (1.121) SO 1.4. group theory (10) γ 43 Eq. (1.121) goes under the name of Clifford17 algebra and the ’s haveγ to be matrices in orderSO N to anticommute with each other .N For definiteness let18 us build an explicit representation of the ’s which is valid for (2 ) groups [107] . We start with = 1. Since the Pauli matrices satisfy the σi;σj δij ; Clifford algebra { } =2 (1.125) i we can chooseγ σ γ σ : i − (1) (1) 1 1 0 1 2 2 0 = = and = = (1.126) N > 1 0 0 N N Then the case 1 is constructedN by recursion. The iteration from to +1 is N γi defined by γi N i ; ;:::; N; ( ) γi ! ( +1) − 0( ) = for =1 2 2 (1.127) 0 i γ N γ N : N N i − ( +1) ( +1) 2 +1 0 1 2 +2 0 =N and = (1.128) γi 1 0 0 ( ) Given the fact that the matrices satisfy the Cliffordγ algebra let us check explicitly 17 In particular it can be shown that the dimension of the matrices must be even. Indeed from Eq. (1.121) we obtain γj γiγj γj γi γj γj γiγj γi ; j ( + ) = 2 or = (1.122) with no sum over . Taking the trace we get γj γiγj γi : i j Tr = Tr (1.123) 6 But for the case = this implies γj γiγj γiγj γj γi ; − − Tr = Tr = γiTr γi (1.124) γi γi − 2 γi and− hence, putting together Eqs. (1.123)–(1.124), we have Tr = 0. On the other hand, = 1 implies that the eigenvalues of are either +1 or 1. This means that to get Tr = 0, the number of +1 and 118 eigenvalues must be the same, i.e. must be even dimensional. For an alternative approach to the construction of spinor representations by means of creation and annihilation operators see e.g. Ref. [108]. SO Chapter 1. From the standard model to 44 N (10) γi ( +1) γ N ; γ N that the N onesN satisfy iti as well,j δij γi ; γj N N δij ; n ( ) ( )o γ ; γ δij n ( +1) ( +1)o j i ( )0 ( ) 2 0 = n No = N =2 (1.129) N N γi 0 2 γi γi ; γ N 0 N N − ; γi ( ) ! γi ( ) ! n ( +1) ( +1)o − 2 +1 0( ) 0( ) = + N =0 (1.130) γ N : 0 0 2 ( +1) 2 +1 =1 (1.131) N N N N N Analogouslyγi one; γ N finds δij ; γ N ; γ N ; γ N : 2 n ( +1) ( +1)o n ( +1) ( +1)o ( +1) 2 +2 2 +1 2 +2 ′ 2 +2 =2 x=0i Oikxk O=1 (1.132) γi Now consider a rotation in the coordinate space, = , where is an orthogonal γ′ O γ : matrix. This rotation induces a transformationi ik k on the matrix = (1.133) ′ ′ γ ; γ OikOjl γk; γl δij : Notice that the anticommutation{ i relationsj } remain{ } unchanged, i.e. γ = =2 (1.134) γ Because the original set of matrices form a complete matrix algebra, the new set of γ′ S O γ S− O O γ S O γ S− O : matrices must bei related toi the original set byik ak similarityi transformation, 1 1 O S O N = (Ï) ( ) or = ( ) ( ) (1.135) ψi The correspondence ( ) serves as a 2 -dimensional representation of the rota- ′ tion group which is called spinor representation.ψi S O ij ψj ; The quantities , which transform like Oik S O = ( ) (1.136) O δ S O iS ; are called spinors. For anik infinitesimalik ik rotation we can paraijmetrizeij and ( ) by 1 ik ki 2 − = + and ( )=1+ (1.137) i Skl; γi γlδik γkδil ; with = . Then Eq. (1.135) implies − ikγk lkγkδil γkδil γkδjl [ ] = ( − ) (1.138) Skl 1 2 i where we have used = Skl= ( γk; γl : ). One can verify that a solution for in Eq. (1.138) is = [ ] (1.139) 4 SO 1.4. group theory (10) 45 kl S O π S O By19 expressing theSO parameterN in terms of rotationsS angle,O one can see that ( (4 )) = 1 , i.e. ( ) is a double-valued representation.γχ However for (2 ) groups the representationN ( ) is not irreducible. To see this γχ i γ γ γ N : we construct the chiral projector defined− by··· 1 2 2 γχ γi N= ( ) γχ; Skl(1.141) ψ ψ′ S O ψ i 20 ij j anticommutes with since 2 is even and consequently we get = 0 − (cf. Eq. (1.139)). Thusψ if transformsγχ ψ as = ( ψ) , the positiveγχ ψ and negative chiral ≡ ≡ − components + 1 ψ ψ− 1 N−(1 + ) and (1 ) (1.142) 2 + 2 − transform separately. In other1 wordsψ andψ form two irreducible spinor repre- sentations of dimension 2 . + ψ T SOWhichN is the relation betweenψ Cψ SOand N ? In order to address this issue it is necessary to introduce the conceptT of conjugation. Let us consider a spinor of S C CSij : (2 ). The combination is anij (2− ) invariant provided that C = C iσ(1.143) N (1) N 2 N C The conjugation matrix Ccan be constructedN iteratively.N We: start from = for C ( ) = 1 and define ( +1) − ( +1)0 ( ) = (1.144) N − ( T) N N 0 C γ C γi : i − One can verify that ( ) 1 ( ) T γi ( ) = ( ) (1.145) N T − N By transposing Eq. (1.145) and substitutingγi; C backC we: get h ( ) 1 ( )i (( ) ) =0 (1.146) N T − N N N T Then the Shur’s LemmaC impliesC λ I C λ C ; ( ) 1 ( ) ( ) ( ) SO (( ) ) = or = ( ) (1.147) 19 This is easily seen for (3). In this case the Clifford algebra is simply given by the three Pauli i ' σi'i ' matrices and a finite transformationS O ' lookse likeσi'i | | i | | ; ' 2 | | ( ( )) = = cos + sin (1.140) ' ' '2 ' '2 ' ' ≡− ≡− ≡− | | q SO N 2 2 2 23 1 13 2 12 3 1 2 3 where20 we have defined , , and = + + . Notice that this would not be the case for (2 + 1) groups. SO Chapter 1. From the standard model to 46 λ λ λ (10) − 2 CT N N / C: which yields = 1. In order to choose− between = +1 and = 1 one has to apply Eq. (1.144), obtaining ( +1) 2 = ( ) (1.148) N − T N N C γ C γχ ; On the other hand Eq. (1.141) and Eq. (1.145)χ lead− to ( ) 1 ( ) γT γ χ χ ( ) = ( ) (1.149) N − N N C γχC γχ : which by exploiting = (cf. again Eq. (1.141))− yields ( ) 1 ( ) ( ) = ( ) (1.150) N − ∗ N N − ∗ N N C Sij γχ C C S γχ C Sij γχ : This allows us to write ij − − ( ) 1 ( ) ( ) 1 ( ) γ ( ) ( (1 + )) = ( ) (1 + ) = 1 + ( ) (1.151) SO N N ψ ψ− where we have also exploited the hermicityC of the matrices.+ Eq. (1.151) can be in- − terpretedSO in theN followingN way:ψ for (2 ) with evenψ and SOarek self-conjugate i.e. real or pseudo-real depending+ on whetherSO is symmetric or antisymmetric (cf. Eq. (1.148)), while for (2 ) with odd is the conjugate of . Thus only (4 +2) can have Spinorscomplex will representations be spinors and remarkably (10) belong to this class. SO N SO N′ SO N ⊂ We close this section by pointingγ outχ a distinctive feature of spinorial representations: spinors of (2 ) decompose into the directN sum of spinors of (2 ) (2 ) [107]. Indeed, since the construction ofN in Eq.γχ (1.141) is such that γχ N ; ( ) γχ ! ( +1) − 0( ) = M− (1.152) ψ SO N 0 M M− + SO N 1 the positive-chirality1 spinor of (2 +2 ) contains 2 positive-chirality spinors ψ ψ ψ− andSO 2 N Mnegative-chiralityÏ SO N M− spinors⊕ SO ofN M−(2 ). More explicitly + + ψ ψ− (2 +2 ) (2 Ï+2 ×2) SO N (2M−+2 ⊕2) × SO N M− Ï··· + M− ψ M− ψ− : (2 +2 4) Ï ×(2 SO+2 N 4) ⊕ × SO N 2 2 1 + 1 (2 ) (2 ) 2 2 (1.153) SO N ;M N ;M Let us exemplify this important concept in the case of the 16-dimensional positive- chirality spinor of (10).− By takingSO respectivelySO ( = 3 = 2) and ( = 2 = 3) we• obtain × ⊕ × ⊃ + 16=2 4 2 4 under (10) (6), SO 1.4. group theory (10) − SO SO 47 • × ⊕ × ⊃ − + − 16=4SO 2 SO4 2 under (10) (4), SO SO SO + + ⊃ ⊗ where 4 (4 ) and 2 (2 ) are respectively the positive (negative) chiral components ; −; − : of the (6) and (4) reducible spinors. Thus⊕ under (10) (6) (4) the 16 decomposes as + + SO SO SU SU SU 16 = (4 2 ) (4 2 ) (1.154) ⊗ As weSO will showSU in Sect. 1.4.4 the Lie algebras (6) and (4) are locally isomorphic to (4) and (2) (2). This allows us to make− the; following identifications between the (6) and (4) representations∼ ∼ SO SU SU + ⊗ 4 4 4 4 (1.155) ; − ; ; and the (4) and (2) (2)∼ ones ∼ + SO 2 (2 1) 2 (1 2) (1.156) SU C SU L SU R ⊗ ⊗ which justify the decomposition of the; ; (10); spinor; : under the Pati-Salam algebra (4) (2) (2) as anticipated in Sect.⊕ 1.3.3, namely 16 = (4 2 1) (4 1 2) (1.157) This striking group-theoretic feature of spinors, which under the natural restriction to an orthogonal subgroup decompose into several copies of identical spinors of the subgroup,SO hints to a suggestive connection with the repetitive structure of the SM families [107] and motivatesSO theSO study of unification in higher orthogonal groups than (10) [27, 107, 109, 110]. To accommodate at least the three observed matter families we must use either (16) or (18). Following− the decomposition in Eq. (1.153) we SO ψSO ψSO ψSO get• Ï × ⊕ × + + − SO ψ (16) ψ (10) ψ (10) • (16): SO Ï 4 × SO ⊕ 4 × SO , + + (18) (10) (10) (18): 8 8 SO k. However there is a fundamental difference between the two cases above. AccordingψSO to the discussion below Eq. (1.151) only (4 + 2) groups have complex spinor+ rep- (16) resentations. This meansψSO that one can write a super-heavy bare mass term for and it is difficult to explain+ why it should be light. On the other hand no bare mass (18) term can be written for , making the last group a more natural choice. The obviousψSO difficulty one encounters inψ thisSO class of models is the overabundance − ψofSO sequential+ or mirror families. If we decide+ to embed the SM fermions into three (10) (10) copies of , the21 remaining families in are called sequential, while those in (10) are mirror families. 21 Mirror fermions have the identical quantum numbers of ordinary fermions under the SM gauge group, except that they have opposite handedness. They imply parity restoration at high-energies as proposed long ago by Lee and Yang [111]. SO Chapter 1. From the standard model to 48 (10) It has been pointed out recently [112] that the existence of three (mirror or sequen- tial) familiesSO is still inSO accord with the SM, as long as an additional Higgs doublet is also present. This however is not enough to allow large orthogonal unification scenarios based1.4.3 on Anomaly(16) or cancellation(18). SO SO Tij (10) is an anomaly-free group.T Thisij important property can be understood from a simple groupi j theoretical argument [113]. Let us consider the (10) generators in a given arbitrary representation. transforms like an antisymmetric tensor in the Tij ;Tkl Tmn ; indices and . Then the anomaly, which{ is proportional} to the invariant tensor δ Tri j k l m n (1.158) ↔ ↔ ↔ must be a linear combinationij kl kl of amn productij of Kroneckermn ’s. Furthermore it must be ↔ ↔ i j k ↔ l m n antisymmetric under the exchanges ↔ , ↔ , ↔ and symmetric under the exchange of pairs , and . However the most general form δjkδlmδni δikδlmδnj δjlδkmδni δilδkmδnj δjkδlnδmi δikδlnδmj δjlδknδmi δilδknδmj ; consistent− with the antisymmetry− in ,− , − ij kl SO ↔ + + + ijklmn is antisymmetric in SOas well and so it mustSU vanish. The proof fails for (6) where the anomaly can beSO proportionalN to theN >invariant tensor . Actually this is consistent with the fact that (6) is isomorphic to (4) which is clearly an anomalous 1.4.4group. On The the standard other hand model( ) is embedding safe for 6. SO From the (10) commutationT relations; T ; T in; Eq. T (1.112); T : we find that a complete set of simultaneously commuting generators can be chosen as 12 34 56 78 90 (1.159) This is also known as the CartanTC; subalgebra TC; TL; TR and; TB can−L : be spanned over the left-right group Cartan generatorsSO SO 3 8 3 3 SO ⊗ SO Tij i;j ; ; ; SO T(1.160)ij i;jLet us; consider; ; ; ; the (4) SO (6) maximal subalgebra of (10). We can span the SO SO (4) generators over with = 1 2 3 4 and the (6) generators⊗ over with =5 6 7 8 9 0. From the (10) commutation relations in Eq. (1.112) one can verify SO SU SU SO SU SO that these∼ two sets⊗ commute (hence∼ the direct product (4) (6)). The next information we need is the notion of local isomorphism for the algebras TL;R T T ; TL;R T T ; TL;R T T ; (4) ≡ (2) ± (2) and (6) ≡ (4).± In the (4) case≡ we define± 1 1 2 1 3 1 2 23 14 2 31 24 2 12 34 ( ) ( ) ( ) (1.161) SO 1.4. group theory (10) 49 i j ijk k i j ijk k i j and checkT byL;T anL expliciti T calculationL ; TR that;TR i TR ; TL;TR : h i h i h i i i TL TR i= ; ; = SU L =0SU R (1.162) SO Thus and ( = 1 2 3) span respectively the (2) and the (2) algebra. On T T T ; T T T ; T T T ; the otherC ≡ hand for the (6) sectorC ≡ we define C ≡ 1 1 2 1 3 1 T 2 T89 T70 ; T 2 T97 T80 ; T 2 T09 T87 ; C ≡ ( + ) C ≡ ( + ) C ≡ ( + ) 4 1 5 1 6 1 T 2 T96 T05 ; T 2T 59 T 06 T ; T 2 T67 T85 ; C ≡ ( + ) C ≡ √ ( + ) C ≡ ( + ) 7 1 8 1 9 1 2 75 86 2 3 65 78 09 2 67 58 TC ( T + T ) ; TC (2 T + T + ;) TC (T + T ) ; ≡ ≡ ≡ 10 1 11 1 12 1 T 2 T75 T68 ; T 2 T69 T05 ; T T2 95 T 06 T ; C ≡ ( + ) C ≡ ( + ) C ≡ √ ( + ) 13 1 14 1 15 1 2 89 07 2 97 08 65 87 90 ( + ) ( + ) 6 ( + + ) i j ijk k and verify after a tedious calculationTC;T thatC i f TC ; h i ijk i f SU= TC i ;:::;(1.163) i SU C SU C TC i ;:::; TC TB L where are the structure constants of (4) (see e.g. [114]). Thus− ( =1 15) spans the (4) algebra15 and, in particular, the (3) subalgebra is spanned by ( = 1 8) while can be identified with the (normalized) generator. Then the hypercharge and electric charge operators read respectively Y TR TB L T T T T T − − 3 r 2 1 12 34 1 65 87 90 = + = ( ) + ( + + ) (1.164) 3 2 3 Q TL Y T T T T : and 3 12 1 65 87 90 = + = + ( + + ) (1.165) 1.4.5 The Higgs sector 3 SO As we have seen in the previous sections (10) offers a powerful organizing principle for the SM matter content whose quantum numbers nicely fit in a 16-dimensional spinorial representation. However there is an obvious prize to pay: the more one unifies the more one has to work in order to break the enhanced symmetry. The symmetry breaking sector can be regarded as the most arbitrary and chal- lenging aspect of GUT models. The standard approach is based on the spontaneous SO Chapter 1. From the standard model to 50 (10) symmetry breaking through22 elementary scalars. Though other ways to face the prob- lem may be conceived the Higgs mechanism remains the most solid one in termsSO of SU C U Q computability⊗ and predictivity. The breaking chart in Fig. 1.2 shows the possible symmetry stagesSO between (10) and (3) (1) with the corresponding scalar representations responsible for the breaking. That gives an idea of the complexity of the Higgs sector in (10) GUTs. SO SU U X ⊗ (10) breakingSU chart with representationsSU up toC theSU 210. L (5)U Y (1) can be understood Figure 1.2: ⊗ ⊗ either in the standard or in the flipped realization (cf. the discussion in Sect. 3.1.2 ). In the former case 16 or 126 breaks it into (5), while in the latter into (3) (2) (1) . For simplicity we are neglecting the distinctions due to the discrete left-right symmetry (cf. Sect. 2.1 for the discussion on the D-parity and Table 2.1 for an exhaustive account of the intermediate stages). In view of such a degree of complexity, better weSO start by considering a minimal Higgs sector. Let us stress that the quest for the simplest Higgs sector is driven not 22 For an early attempt of dynamical symmetry breaking in (10) see e.g. [115]. SO 1.4. group theory (10) 51 only by aesthetic criteria but it is also a phenomenologically relevant issue related to tractability and predictivity of the models. Indeed, the details of the symmetry break- ing pattern, sometimes overlooked in the phenomenological analysis, give further con- straints on the low-energy observables such as the proton decay and the effective SM flavor structure. For instance in order to assess quantitatively the constraints imposed by gauge coupling unification on the mass of the lepto-quarks23 resposible for proton decay it is crucial to have the scalar spectrum under control . From the breaking chart in Fig. 1.2 we concludeSO that, before before considering any symmetry breaking dynamics, the following representations are required by the group theoryH in orderH to achieve a full breaking of (10) down toSU the SM: • 16 or 126 : they reduce the rank by one unit but leave an (5) little group unbroken.H H H SU U • SU ⊗ 45 or 54 or 210 : they admit for little groups different from (5) (1), yielding the SM when intersected withSO (5). Ï H It should be also mentioned that a one-stepF (10)F SM breaking can be achieved via only one 144 irreducible Higgs representation [54]. However, such a setting requires an extended matter sector, including 45 and 120 multiplets, in order to accommodate realistic fermion masses [55]. As we will see in the next Chapters the dynamics of the spontaneous symmetry breaking imposes further constraints on the viability of the options showed in Fig. 1.2. On top of that one has to take into account also other phenomenologicalH constraintsH due to the unification pattern, the proton decay and the SM fermion spectrum. We can already anticipate atH thisH level that whileH the choice between 16 or 126 His a model dependent issue related to the details of the Yukawa sector (see e.g. Sect. 1.5), SU U H H H the simplest option among⊗ 45 , 54 and 210 is certainly given by the adjoint 45 . However, since the early 80’s, it has been observed that the vacuum dynamics aligns the adjoint along an (5) (1) direction, making the choice of 16 (or 126 ) and 45 alone not phenomenologically viable. In theSU nonsupersymmetric case the alignment is only approximate [56, 57, 58, 59], but it is such to clash with unification constraints (cf. Chapter 2) which do not allow for any (5)-like intermediate stage, while in the supersymmetric limit the alignment is exact due to F-flatness [60, 61, 62], thus never landingH to a supersymmetric SM vacuum. The critical reexamination of these two longstanding no-go for the setting with the 45 driving the GUT breaking will be the subject of Chapters 3 and 4. 23 Even in that case some degree of arbitrariness can still persist due to the fact that the spectrum can never be fixed completely but lives on a manifold defined by the vacuum conditions. SO Chapter 1. From the standard model to 1.552 Yukawa sector in renormalizable SO (10) SO (10) S A S ; In order to study the (10) Yukawa⊗ sector,⊕ we decompose⊕ the spinor bilinear S A 16 16 = 10 120 126 (1.166) where Hand H denote theH symmetric (S) and antisymmetricSO (A) nature of the bilinear couplings in the family space. At the renormalizable level we have only three possibil- Y F Y H Y H Y H F ; ities: 10 , 120 andL 126 . Thus the most general (10) Yukawa lagrangian is given by Y Y 10 120 126 Y = 16 10 + 120 + 126 16 + h.c. (1.167) 10 126 120 where 24 and are complex symmetricH H matrices while is complex antisym-SO metric . H H It should be25 mentioned that 10 and 120 are real representation from the (10) point24 of view . In spite of that the components of 10 and 120 can be chosen either For completeness we report a concise proof of these statements based of the formalism used in Sect. 1.4.2 and borrowed from Ref. [107]. In a schematic notation we can write a Yukawa invari- ψT C C ψ ; ant term such as those in Eq. (1.167) as D k k ψ SO 5 C C ( Γ )Φ D (1.168) SO k k γ 5 wherek is both a Lorentz and an (10) spinor (hence the need for kand which areSO respectively T ψ γχ γχ ; γi C γχ γχC γ γχ the Dirac and the (10) conjugation matrix).{ } Then Γ denotes− an antisymmetric productχ of matrices and Φk is a scalar field transforming like an antisymmetric tensor with indices under (10).k Using 5 5 thek facts; ; that is an eigenstate of , = 0, = − (cf. Eq. (1.150)) and− = , we deduce that must be odd (otherwise Eq. (1.168) is zero). This singles10! out the antisymmetric10! tensors Φ with = 1 3 5, corresponding respectively to dimensions 10, 3!(10 3)! = 120 and 5!(10 5)! = 252 (actually the duality map defined in Eq. (1.117) is such that only half of these 252 components couples to the spinor T T C CD C C bilinear). D − − Next we consider the constraintsT imposedT T byT theT symmetryT proper− tiesT of the conjugation matrices, ψ 5CDC 5kψ ψ C C ψ ψ CDC C C ψ ; namely = and = (cf. Eq.− (1.148)).D k These− yields k 1 5 5 5 5 5 − T Γ = Γ = ( Γ ) (1.169) C γ C γi i − where in the second− step we1 T have used− T the anti-commutationT k propertiesk ofk thek− / fermion fields. Then, by C γ γ5 k C 5 C γ γ C γk γ γ γk ; exploiting the relation· · · = k(cf.· · ·Eq. (1.145)),− we obtain· · · − − · · · 1 1 ( 1) 2 1 5 5 1 1 5 ( ) = 5 1 = ( ) = ( ) ( ) (1.170) T k k− / k T ψ CDC kψ ψ CDC kψ : which plugged into Eq. (1.169) implies − ( 1) 2+ +1 k ; 5 5 ψ k Γ = ( ) Γ (1.171) Hence for = 1 3 the invariant in Eq. (1.168) is symmetricSO in the flavor space of , while for = 2 is ∗ Ta T antisymmetric.25 − a This can be easily seen from the fact that theS (10) generators in the fundamental representation are both imaginary and antisymmetric (cf. Eq. (1.111)). This implies = which corresponds to the definition of real representation in Eq. (1.3) with = 1. SO 1.5. Yukawa sector in renormalizable (10) 53 H ∗ H ∗ 6 H 6 H real or complex. In the latter case we have 10 = 10 and 120 = 120 , which means that the complex conjugate fieldsSO differ from the original ones by some extra charge. Actually both the components are allowed26 in the Yukawa lagrangian, since Y F Y H Y ∗ Y H Y ∗ Y H F : theyL transform in the same wayH under (10) , andH thus we have 10 10 120 120 126 = 16 10 + ˜ 10 + 120 + ˜ 120 + 126 16 + h.c. (1.172) For instance complex scalars are a must in supersymmetryY Y where the fundamental objects are chiral superfields made of Weyl fermions and complex scalars. However 10 120 in supersymmetry we never see the couplings ˜ and ˜ because of the holomorphicH properties of theH superpotential. Even without supersymmetry there could be the phenomenological need, as we are going to see soon, of having either a complex 10 or a complex 120 . In this case the new structures in Eq. (1.172) areH still there, unless SU C SU L SU R an extra symmetry which forbids⊗ them⊗ is imposed. In order to understand the implications; ; of having; ; a: complex 10 , let us decompose it under the subgroup (4) (2) ⊕(2) SU C SU L U Y 10 = (1 2 2) (6 1 1) ⊗ ⊗(1.173) ; ; ; ; Hu ; ; Hd H ∗ H∗ Hd ≡ ⊕ − ≡ H u H ∗ H∗ Hd In particular the bi-doublet1 can6 beH furtheru1 decomposed6 under (3) (2) (1) , yielding (1 2 2) = (1 2 + 2 ) (1 2 2 ) . Now if 10 = 10 we have = as in the SM, while if 10 = 10 then = as much as in the MSSM or in the two-higgs doublet model (2HDM). Y ToY simplify a bit the discussion let us assume that we are eithHu∗er inHd the supersymmet- 6 10 ric case or in the nonsupersymmetric one with an extra symmetry which forbids ˜ ˜120 and , so that; Eq.; (1.167); ; applies; with complex bi-doublets ( = ). The remaining representations in Eq.⊕ (1.167) decompose as ; ; ; ; ; ; ; ; ; ; ; ; ; ⊕ ⊕ ⊕ ⊕ ⊕ 16 = (4 2; 1); (4 1;2); ; ; ; ; ; (1.174) 120 = (1 2 2) ⊕ (10 1 1) ⊕ (10 1 1) ⊕ (6 3 1) (6 1 3) (15 2 2) (1.175) 126 =; (6; 1 1) ; (10; 3 1); ; (10 1 3) ; (15; 2 2) (1.176) under the Pati-Salam group and thus the fields which can develop a SM-invariant VEV are (10 3 1), (10 1 3), (1 2 2) and (15 2 2). With the exception of the last one we already encountered these representations in the context of the Pati-Salam model vL ; ; ; vR ; ; ; (cf. Sect. 1.3.2). Let us also≡ fix the following notation≡h for the SM-invarianti VEVs vu;d ; ; u;d126 ; vu;d ; ; 126u;d ; ≡ (10 3 1) ≡ (10 1 3) (1.177) D E D E 10 10 126 126 (1 2 2) (15 2 2) (1.178)≡ √ 26 i 1 Alternatively one can imagine a complex 10 as the linear combination of two real 10’s, i.e. 10 1 2 2 (10 + 10 ). This should make clearer the origin of the new structures in Eq. (1.172). SO Chapter 1. From the standard model to 54 vu;d ; ; u;d ; vu;d ; ; u;d : (10) ≡ ≡ 1 D E 15 D E 120 120 120 ; ; ;120; (1 2 2) (15 ⊕2 2) (1.179) Given the embedding of a SM fermion family into (4 2 1) (4 1 2) (c.f. Eq. (1.105)) one u u u u finds the following fermionMu Y massv sumY rulev afterY thev electrowev ak symmetry breaking d d d d Md Y v Y v Y v 1 v 15 10 10 126 126 120 120 120 = d + d + ( d + d ) (1.180) Me Y v Y v Y v 1 v15 10 10 126 126 120 120 120 = u −+ u + ( u +− u) (1.181) MD Y v Y v Y v v 10 10 − 126 126 120 1201 − 12015 MR = Y vR 3 + ( 3 ) (1.182) 10 10 126 126 120 1201 12015 ML = Y vL 3 + ( 3 ) (1.183) 126 = (1.184) 126 MD MR ML = (1.185) ν;νc where , and enter the neutrinoML massMD matrix defined on the symmetric basis T : ( ) MD MR (1.186) ; ; Me MD − h i 27 Eqs. (1.180)–(1.185) follow from the SM decomposition , but it is maybe worth of a comment the 3 factor in front of (15 2 2) for the leptonic components and . ; ; ; ; ; ; : That is understood by looking at the Pati-Salamh i invariant SU C SU C (4 2 1) (15 2 2) (4 1 2) (1.187)⊗ U Q ; ; h i The adjoint of (4) is a traceless hermitian matrix, so the requirement of an (3) v (1) preserving vacuum; implies; the following; ; ; shape for (15u 2; 2) h i ∝ − ⊗ vd 0 (15 2 2) diag(1 1 1 3) (1.188) 0 ; ; − h i which leads to an extra 3 factor for leptons with respect to quarks. Conversely (1 2 2) preserves the symmetry between quarks and leptons.vR In order to understandSO the implications of the sum-rule in Eqs. (1.180)–(1.185) it is useful to estimate the magnitudeMU of the VEVs appearing there: is responsible for the rank reductionv of (10) and gauge unification constrains its value to be aroundvL (or just M /MU below)O W the unification scale , then all the bi-doublets can develop a VEV (collectively denoted2 as ) which is at most of the order of the28 electroweak scale, while is a small ( ) VEV induced by the scalar potential in analogy to what happens in the left-right symmetric models (cf. Sect. 1.3.1). 27 SO 28For a formal proof see e.g. [108]. In the contest of (10) this was pointed out for the first time in Ref. [33]. SO 1.5. Yukawa sector in renormalizable (10) 55 vR v vL ≫ ≫ Thus, given the hierarchy , Eq. (1.186) can be block-diagonalized T Mν ML MDM− M ; (cf. Eq. (1.99)) and the light neutrino mass− matrixR is veryD well approximated by 1 = (1.189) where the first and the second term are the type-II and type-I seesaw contributions already encountered in Sect. 1.3.1. Which is the minimum number of Higgs representations needed in the Yukawa sector in order toF have a realistic theory? With only one Higgs representation at play there is no fermion mixing, since one Yukawa matrix can be always diagonalized by rotating the 16 fields, so at least two of them must be present. Out of the six H H combinations⊕ (see e.g. [116]): H H 1. 10 ⊕126 H H 2. 120 ⊕ 126 H H 3. 10 ⊕ 120 H H 4. 10 ⊕10 H H 5. 120 ⊕ 120 6. 126 126 Md Me Md Me Y − them last three can be readily discardedm sincem they predict wrong mass relations, namely − 120 = (case 4), H = 3 (case 6), while in case 5 the antisymmetry of implies 1 2 3 = 0 (first generation) and = (second and third generation). Notice that in absenceH of H126 (case 3) neutrinos are Dirac andH their mass is related to that of charged leptons which is clearly wrong. In order to cure this one has to introduce the bilinear 16 16 which plays effectively the role of 126 (cf. Sect. 4.1 for a discussion of this case in the context of the Witten mechanism [67, 117, 118]). Though all the cases 1, 2 and 3 give rise to well defined Yukawa sectors, for definiteness we are going to analyze in more detail just the first one. 1.5.1 H H with supersymmetry ⊕ 10 126 H H H H ⊕ ⊕ ⊕ This case has been the most studied especially in the context of the minimal super- symmetric version, featuring 210 126 126 10 in the Higgs sector [46, 47, 48]. The effective mass sum-rule in Eqs. (1.180)–(1.185) can be rewritten in the following SO Chapter 1. From the standard model to 56 (10) M Y v Y v ; way u u u M Y v10 Y v126 ; d 10 d 126 d = + Me Y v10 Y v126 ; 10 d − 126 d M = Y v10 + Y v126 ; D 10 u 126 u = − 3 (1.190) MR Y v10R ; 126 10 126 ML = Y vL ; 3 126 = Y 126Y = Mν 10 ML 126MDM− MD : and, exploiting the symmetry of and − , theR neutrino mass matrix reads 1 = (1.191) 29 In the recent years this model received a lot of attention due to the observation [134] b τ that the dominance of type-II seesaw leads to− a nice correlation between the large atmospheric mixing in the leptonic− sector and the convergence of the bottom-quark θ ◦ and tau-lepton masses at the unification scale ( unification)∼ which is a phenomenon occurring in the MSSM up to 20 30% corrections [86]. b τ 13 Another interesting prediction− of the model is 10 [120], in agreement with the recent data released by the T2K collaboration [135]. Mν ML The correlation between unification and large atmospheric mixing can be Mν Md Me : understood with a simple two generations∝ − argument. Let us assume = in Eq. (1.191), then we get (1.192) In the the basis in which charged leptons are diagonal and for small down quark ms mµ mixing , Eq. (1.192) is approximatedMν by− ; ∝ mb mτ − (1.193) mb mτ and, being the 22 entry the largest one, maximal atmospheric mixingY requiresY a can- cellation between Mande . Md Mu;MD 10 126 ForMν a more accurate analysis [53] it is convenient to express the and Yukawa matrices in terms of and , and substitute them in the expressions for Mu fu r Md r Me ; and : − MD fu r Md r M e ; = (3 +− ) + (1 ) (1.194) = 3(1 ) +(1+3 ) (1.195) 29 For a set of references on the subject see [119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133]. SO 1.5. Yukawa sector in renormalizable (10) 57 v v v f u ; r d u : u v where vd10 u10 vd126 1 = 10 = 10 126 (1.196) 4 MD MD The neutrino mass matrixMν fν is obtainedMd Me as ξ Md Me − ; − fu − fu 1 = ( ) + ( ) (1.197) vL fuvd fν ; ξ : v v126 2 with vd − L R 1 4 = 126 = mx x u;d;e;ν (1.198) 4 In what follows we denote diagonal mass matrices byMd ˆ ,md= , withM eigen-e values corresponding to the particle masses, i.e. being reaMl ande W positive.e†meWe∗ We chooseWe a basis where the down-quark matrix is diagonal: fu = ˆfν. In this basis is a general complex symmetric matrix, that can be written as = ˆ , where is a general unitary matrix. Without loss of generalityWe andfu; fν can be taken to be real and positive. Hence,r theξ independent parameters are given by 3 down-quark masses, 3 charged lepton masses, 3 angles and 6 phases in , , together with two com- plex parameters and : 21 real parameters in total, among which 8 phases. Using Eqs. (1.194), (1.195), and (1.197) all observables (6 quark masses, 3 CKM angles, 1 CKM phase, 3 charged lepton masses, 2 neutrino mass-squared differences, the mass of the lightest neutrino, and 3 PMNS angles, 19 quantities altogether) can be calculated in terms of these input parameters. Since we work in a basis where the down-quark mass matrix is diagonal the CKM T matrix is given by the unitary matrixmu diagonalizingWuMuWu the up-quark mass matrix up to diagonal phase matrices: iβ ˆiβ = iβ iα iα (1.199) Wu e ;e ;e V e ;e ; ; with 1 2 3 1 2 αi;βi CKM = diag( ) T diag( 1) (1.200) mν WνMνWν T whereWe∗Wν areD V unobservableD phasesD atD low energy. The neutrino mass matrix given in Eq. (1.197) is diagonalized by ˆ = , and the PMNS matrix is determined 1 PMNS 2 1 2 by = ˆ ˆ , where ˆ and ˆ are diagonal phase matrices similar to those in Eq. (1.200). Allowing an arbitrary Higgs sector it is possible to obtain a good fit of the SM flavor structure [53]. However, after including the constraints of the vacuum in the minimal supersymmetric version of the theory [49, 50, 51], one finds [52, 53] an irreducible incompatibility between the fermion mass spectrum and the unification constraints. The reason can be traced back in the proximity between the unification scale and the seesaw scale, at odds with the lower bound on the neutrino mass scale implied by the oscillation phenomena. SO Chapter 1. From the standard model to 58 (10) The proposed ways out consist in invoking a split supersymmetry spectrum [136] or resorting to a non-minimal Higgs sector [137, 138, 139, 140], but they hardly pair the appeal of the minimal setting. In this respect it is interesting to notice that without H H supersymmetry⊕ gauge unification exhibits naturally the required splitting between the seesaw and the GUT scales. This is one of the motivations behind the study of the 10 126 system in the absence of supersymmetry. 1.5.2 H H without supersymmetry ⊕ H 10 126 mt mb In the nonsupersymmetric case it would be natural to start with a∼ real 10 . However, as pointed out in Ref. [141] (see also [142] for an earlier reference), this option is not phenomenologicallyθq V viable.cb The reason is that one predicts , at list when working in the two heaviest generations limit with real parameter and in the sensible T approximation = = 0. It is instructiveWu to reproduce this statementWu with the help of the parametrization given in Sect. 1.5.1. Md md Let us start from Eq. (1.194) and apply T (from the left) andT (from the right). mu fu r mdWuW r WuMℓW : Then, taking into account Eq. (1.199) and theu choice− of basis u = ˆ , we get ˆ = (3 + ) ˆ + (1 ) (1.201) T WuW Next• we makeu ∼ the following approximations: Wu V • ∼ 1 (real approximation) V CKM Vcb Vts Aλ • ∼ (real approximation) ∼ ∼ ∼ T 2 WCKMuMℓW mℓ • 1u (for∼ the 2nd and 3th generation and in the limit 0) ˆ (for the self-consistency of Eq. (1.201) in the limits above) mc fu r ms r mµ ; which lead to the system ∼ − mt fu r mb r mτ : ∼ (3 + ) + (1 − ) (1.202) r (3 + ) + (1 ) (1.203) mc mτ mb mt mµ ms mt It is then a simple algebrafu to substitute− back− and− find the relation: ∼ msmτ mµmb ∼ mb − 1 ( ) ( ) 1 u d H v v (1.204)fu 4 | | | | 4 10 10 On1 the other hand a real 10 predicts fu = : and hence from Eq. (1.196) = 4 × ∼ O . More quantitatively, considering16 the nonsupersymmetric running for the fermion masses evaluated at 2 10 [143], one gets 22 4, which is off by a factor of (100). SO 1.5. Yukawa sector in renormalizable (10) 59 H This brief excursus shows that the 10 must be complex. In such a case the fermion M Y v Y v Y v ; mass sum-rule reads u u d ∗ u 10 10 126 Md Y vd Y vu ∗ Y vd ; 10 ˜10 126 = 10 + 10 + 126 Me Y vd Y vu ∗ Y vd ; 10 ˜10 − 126 = 10 + 10 + 126 MD Y10vu Y10vd ∗ Y126vu ; = + ˜ − 3 (1.205) MR Y v10R ; 10 126 10 ˜10 126 ML = Y vL :+ 3 126 = 126 = The three different Yukawa sourcesY would certainly weaken the predictive power of the model. So the proposal in Ref. [141] was to impose a Peccei-QuiPQ Fnn (PQ)α PQ symmetryH [144,α 10 − PQ H α ˜ 145] which forbids− the coupling , thus mimicking a supersymmetric Yukawa sector (see also Ref. [142]).H The following charge assignment:U R U B (16L ) =U ,Y (10 ) =U 2PQ ⊗ − Ï SO U PQ and (126 ) = 2 would suffice. ⊗ In this case 126 is responsibleU PQ bothU Y for (1) Y (1) (1) and the (1) ⊗ ⊥ ⊥ breaking.Y However, since it cannot break the rank of (10) (1) by two units, a global linear combination of (1) (1) (where is the generator orthogonal to ) survives at the electroweak scale. This remnant global symmetry is subsequently broken by the VEV of the electroweak doublets, that is phenomenological unacceptable ÷ B L since it would give rise to a visible axion [146, 147] which is experimentally− excluded. Actually9 astrophysical12 and cosmological limits prefers the PQ breaking scale in the window 10 GeV (see e.g. [148]). It is therefore intriguing to link the breaking scale responsible for neutrino masses and the PQ breaking one in the same model. This has been proposed long ago in [149] and advocatedU againPQ in [141]. What is needed is another representation charged under the PQ symmetry in such a way that it is decoupled from the SM fermions and which breaks (1) completely at very high scales. SOIn summary the PQ approach is very physical and well motivated since it does not just forbid a coupling in the Yukawa sector making it moreθ "predictive", but correlates (10) with other two relevant questions: it offers the axionSO30 as a dark matter candidate and it solves the strong CP problem predicting a zero . However one should neither discard pure minimal (10) solutions with the SM as the effective low-energy theory. Notice that in the PQ case we are in the presence of a 2HDM which is more than what required by the extended survival hypothesis (cf. the discussion in Sect. 1.2.4) in order to set the gauge hierarch31 y. Indeed two different fine-tunings30 are needed in order to get two light doublets . 31This is trueHu as longHd as we ignore gravity [150]. The situation is different in supersymmetry where the minimal fine-tuning in the doublet sector makes both and light. SO Chapter 1. From the standard model to 60 (10) vd vd vu vu 10 M126d ThusMe we could10 minimally126 consider the sum-rule in Eq. (1.205) with either = = 0 or = = 0. The first option leads to a clearly wrong conclusion, i.e. M Y v ; = . So we are left with theu secondd one∗ which implies M Y v10 Y v ; d 10 d d = ˜ Me Y v10 Y v126 ; 10 d − 126 d = 10 + 126 MD Y10vd ∗ ; 126 = 3 (1.206) MR Y v10R ; ˜10 ML = Y vL ; 126 = 126 = Mν ML MDM− MD : − R 1 and MD Mu MR = Md Me (1.207) − Notice that in the case of type-I seesaw the strong hierarchy due to = must by undone by which remains proportional to . More explicitly, in the case of vd type-I seesaw, one finds Mν Mu Md Me − Mu : − v126R 1 =4 ( ) (1.208) Though a simple two generations argument with real parameters shows that Eq. (1.208) could lead to an incompatibility with the data, a full preliminary three generations study 1.5.3indicates Type-I that this vs is not type-II the case seesaw [151]. Here we would like to comment about the interplay between type-I and type-II seesaw in b τ Eq. (1.189). In a supersymmetric context one generally expects these two contributions− to be comparable. As we have previously seen (see Sect. 1.5.1) the dominance of type-II b τ seesaw leads to a nice− connection between the large32 atmospheric mixing and mb : : mτ : : unification and one would like to keep± this feature . On the other± hand without supersymmetry× the convergence is far from being obtained. For instance the running within the16 SM yields = 1 00 0 04 GeV and = 1685 58 0 19 MeV at the scale 2 10 GeV [143]. Thus in the nonsupersymmetric case the dominance of type-II seesaw would represent a serious issue. SO In this respect it is interesting to note that the type-II seesaw contribution can be naturally subdominant in nonsupersymmetricSO (10). The reason has to do with the32 left-right asymmetrization of the scalar spectrum in the presence of intermediate See e.g. Ref. [152] for a supersymmetric (10) model in which the type-II seesaw dominance can be realized. 1.6. Proton decay 61 SU R R H B L ⊂ 33 − symmetryMU breaking stages .M UsuallyR the unification pattern is such that the mass of the (2) vtripletR ∆R 126 responsible for the breaking is well below the GUT ≡h i ∆ M M SU L L H MU scaleR . TheL reason is that must be fine-tuned⊂ at the level of the34 intermediate scale VEV ∆ . Then, unless therevL is aL discrete left-right symmetry which locks ∆ ∆ ≡h i = , the mass of the (2) tripletv ∆ 126 , remains automatically at . v λ R v : On the other hand the induce VEV L ∆M is given by (cf. e.g. Eq. (1.103)) L 2 2 λ v = ∆ (1.209) where and denote a set of parameters of the scalar potential and an electroweak MB L v VEV respectively. Sov weR canMB writeL ; vL λ − ; ∼ − ∼ MU 2 MB2 L − MB L/MU (1.210) − 2 which shows that type-II seesaw is suppressed by a factor ( ) with respect to 1.6type-I. Proton decay The contributions to the proton decay can be classified according to the dimension of the baryond violating operators appearing in the lagrangian. Since the externald states ared fermions and because of the color structure the proton decay operators arise first at the = 6 level. Sometimes the source of the baryon violation isd hidden in a =5or a = 4 operatord involving also scalar fields. These operators are successively dressed with the exchange of other states in order to get effectively the = 6 ones. The so-called = 6 gauge contribution is the most important in nonsupersymmetric GUTs. In particular if the mass of thed lepto-quarks which mediate these operators is constrained by the running then the major uncertainty comes only from fermion mixing. There is also another class of = 6 operators coming fromd the Higgsd sector but they are less important and more model dependent. The supersymmetrization of the scalar spectrum gives rise to = 5 and = 4 baryon and lepton number violating operators which usually lead to a strong enhance- ment of the proton decay amplitudes, though they are very model dependent. In the next subsections we will analyze in more detail just the gauge contribution while we will briefly pass through all the other ones. We refer the reader to the reviews [154, 155, 156] for a more accurate account of the subject. 33 SO 34ForH a similarH phenomenon occurring in the context of left-right symmetric theories see Ref. [153]. As we will see in Chapter 2 this can be the case if the (10) symmetry breaking is due to either a 54 or a 210 . SO Chapter 1. From the standard model to 1.6.162 (gauge) (10) d d = 6 SU C SU L U Y Following⊗ the approach⊗ of Ref. [157], we start by listing all the possible = 6 baryon number violating operators due to the exchange of a vector boson and invariant under B L c µ c (3) (2) (1)OI[25,− 158,k 159]ijk αβ uia γ qjαa eb γµ qkβb ; B L c µ c O − k2 u γ q d γ ℓ ; II 1 ijk αβ ia jαa kb µ βb B L = c µ c (1.211) O − k2 d γ q u γ ℓ ; III 1 ijk αβ ia jβa kb µ αb B L = c µ c (1.212) O − k2 d γ q ν γ q : IV 2 ijk αβ ia jβa b µ kαb = 2 (1.213) 2 k = gU /√ MX k gU /√ MY MX MY (1.214)MU ∼ gU 1 2 In the above expressions i;j;k= 2; ; and = 2SU , whereC α;β ;, SU L and a areb the masses of the superheavy gaugeq u;d bosons andℓ theν;e gaugeSU couplingL at the B L B L unification scale. The indicesOI − =1OII−2 3 are referred to (3) , =1 2 to (2) and and Xare family; ; indices. The fields = ( ) and = ( ) are (2) doublets.SU − B L B L The effectiveY operators; 5 ; and appear whenO weIII− integrateOIV− out the superheavy 6 gauge fieldSU = (3 2 ). This1 is the case inSO theories based on the gauge group X (5). IntegratingY out = (3 2 + 6) we obtain the operators and . This is the case of flipped (5) theories [69, 70], while in (10) models both the lepto-quarks and are present. Using the operators listed above, we can write the effective operators in the physical c c c µ c basis for each decayO e channelα;dβ c [157]eα;dβ ijk ui γ uj eα γµ dkβ ; c c c µ c O eα;dβ c eα;dβ ijk ui γ uj dkβ γµ eα ; ( ) =c ( ) c c µ c (1.215) O νl;dα;dβ c νl;dα;dβ ijk ui γ djα dkβ γµ νl ; ( c ) =c ( c ) c c µ c (1.216) O νl ;dα;dβ c νl ;dα;dβ ijk diβ γ uj νl γµ dkα ; ( ) = ( ) (1.217) ( ) = ( ) (1.218) c αβ β α where c eα;dβ k V V V VUD V VUD† ; c βα β α c e ;d k2 V11V k V V1† V V 1 V †V ; α β 1 1 2 1 UD 2 UD ( ) =c [ + ( α ) (βl ) ] βα l (1.219) 2 11 2 1 1 † c νl;dα;dβ 1 k 1 V3 VUD 2 V V4 EN k1 V 4V3VUDV V VEN ; ( c ) =c + β ( lα) ( βα ) l (1.220) 2 † 1 † 2 † † 1 c νl ;dα;dβ k1 V1 VUD U3EN V V2 U4 EN V1 VUD 4 3; ( ) = 2 ( ) 1( ) + ( 1 ) (1.221) 2 4 2 4 2 α β( ) = [( ) ( ) + ( ) ] (1.222) 6 C C V Uc†U V Ec†D V Dc†E V Dc†D VUD U†D VEN E†N UEN E †N with = = 2. In the equations above we have defined the fermion mixing matrices as: 1 2 3 4 = , = , = , = , = , = and = , 1.6. Proton decay U;D;E 63 UT Y U Y diag ; where define the Yukawa couplingc U diagonalizationU so that T diag Dc YD D YD ; T = diag (1.223) Ec YE E YE ; T = diag (1.224) N YN N YN : = (1.225) VUD U†D K VCKM K K K = D (1.226) VEN K VPMNS K M1 2 1 2 Further, one may write =VEN K=VPMNS , where and are diagonal 3 4 matrices containing respectively three and two phases. Similarly, = in 3 the case of Dirac neutrinos,d or = in the Majorana case. k k b V FromV V thisV briefU excursusEN we can see that the theoretical predictions of the proton 1 2 lifetime1 from the gauge = 6 operators require the knowledge of the quantities , , 1 2 3 4 d B L , , , and . In addition we have three− (four) diagonal matrices containing phases in the case of Majorana (Dirac) neutrino. Since the gauge = 6 operators conserve the nucleon decays into a meson and an antilepton. Let us write the decay rates for the different channels. We assume that in the proton decay experiments one can not distinguish the flavor of the neutrino andd the chirality of charged leptons in the exit channel. Using the Chiral Lagrangian techniques (see e.g. [160]), the decay rates of the different channels due to the gauge p K ν = 6→ operators are [157] + mp mK mp c mp c Γ( ) − AL α D c νi;d;s D F c νi;s;d ; πm f2 | | 3 mB mB 2 2 p π 2 i ( ) 2 2 X 2 = 3 2 ( )+[1+ ( + 3 )] ( ) (1.227) 8 =1 3 3 mp c p π ν AL α D F c νi;d;d ; → πf | | 3 | | π i + 2 2 2 X 2 2 Γ( )= mp mη (1 + + ) ( ) (1.228) =1 c c p η eβ 8 − AL α D F c eβ;d c eβ;d ; → 2πfπm2p 2 | | − | | + 2 2 2 n 2 2o ( m m) Γ( )= p 2 3K (1 + mp3 ) ( ) +c ( ) c (1.229) p K eβ 48 − AL α D F c eβ;s c eβ;s ; → 2πfπ m2p 2 | | mB − | | 0 + ( m ) 2 2 2 n 2 2o p 2 3 c c Γ(p π eβ )= AL α D[1+F (c eβ;d)] (c eβ;d) + ; ( ) (1.230) Ï 8πfπ | | | | 0 + 2 2 2 n 2 2o 2 Γ( νi νe)=;νµ;ντ eβ (1 +e; µ+ ) ( ) + ( ) mB : (1.231) 16 ∼ mB m m fπ ∼ ∼ ∼ Dwhere: = F :and = . In the equations above 1 15 MeV is the ∼ ∼ Σ Λ average baryon mass , 131 MeV is the pion decay constant,α : ∼ − 0 80 and 0 47 are low-energy constants of the Chiral Lagrangian which can be obtained3 from the analysis of semileptonic hyperon decays [161] and 0 0112 GeV is a proton-to-vacuum matrix element parameter extracted via Lattice SO Chapter 1. From the standard model to 64 (10) AL : ∼ MZ QCD techniques [162]. Finally 1 4 takes into account the renormalization from to 1 GeV. SO H H In spite of the complexity and the model-dependency of the⊕ branching ratios in Eqs. (1.227)– (1.231) the situation becomes much more constrained in the presence ofU symmetricc UKu Yukawas,Dc DKd relevantEc forEK realistice K(10)u Kd modelsKe based on 10 126 in the Yukawa sec- tor. In that case we get the following relations for the mixing matrices: = , = and = , where , and / are diagonal matrices involving phases. These relations lead to the remarkable predictionQ [157] k / ; A V 1 A4 V CKM 1 CKM 1 | | | | | | | | 2 11 2 2 12 2 1 4 = 1 2 (1.232) + πm f p K ν m m whereQ p π Ï ; A p D ; A p D F : m 3 2m A α+ mB mB p − K L| | 1 8 Γ( ) 1 2 2 = 2 2 2 2 2 = = 1 + ( +3 ) (1.233) ( ) k gU /√3 MX 3 gU MX 1 Notice that the expression for = 2 is independent from unknown mixing matrices andSO CP violating phases, whileH theH values of and are subject to gauge p K ν ⊕ coupling unificationÏ constraints. This is a clear example of how to test a (nonsuper- symmetric) (10)+ model with 10 126 in the Yukawa sector through the decay MX MY MU channel Γ( ). ∼ ∼ We close this subsection with a naive model-independent estimate for the mass of the superheavy gauge bosons . Approximating the inverse lifetime of m the proton in the following way (cf. the real computationp in Eqs. (1.227)–(1.231)) p αU ∼ MU5 2 4 τ p π e > : Γ (1.234) → × 0 + 33 MU > : ; and using ( ) 8 2 10 yr [11],× one finds the naive lower bound 15 αU− 2 3MU10 GeV αU− (1.235) 1 1 1.6.2where we fixed (scalar)= 40. The bound on as a function of is plotted in Fig. 1.3. d = 6 d T ; ; In nonsupersymmetric scenarios the next-to-leading− contribution to the decay of the proton comesSU from the Higgs induced = 61 operators. In this case the proton decay is mediated by scalar leptoquarks = (3 1 3). For definiteness let us illustrate the case of minimal (5) with just one scalar leptoquark. In this model the scalar leptoquark 1.6. Proton decay MU 65 4.0 ´ 1015 3.5 ´ 1015 3.0 ´ 1015 2.5 ´ 1015 2.0 ´ 1015 1.5 ´ 1015 Α -1 10 20 30 40 50 60 70 80 U MU αU− 1 Figure 1.3: Naive lower bound on the superheavy gauge boson mass as a function of . H lives in the 5 representation together with the SM Higgs. The relevant interactions for proton decay can be written in the following way [155] T cT c T ijk αβ q C A qjβ Tk u C B e Ti L iα i T cT c αβ qiα C C ℓβ Ti∗ ijk ui C D dj Ti∗ h:c: = + + + + (1.236) A B C D In the above equation we have used the same notation as in the previous subsection. T The matricesSU, , and are linear combinationsA ofB theY YukawaU C couplingsD YD inY theE theory and the possible contributions coming from higher-dd imensional operators. In the minimal (5) we have the following relations: = = , and = = = . Now, using the above interactions we can write the Higgs = 6 effective operators for proton decay [155] T T OH dα;eβ a dα;eβ u LCdα u L Ceβ ; c c T c c OH dα;eβ a dα;eβ u LCdα eβ† L Cu ∗ ; ( c ) = ( c ) c c T (1.237) OH dα;eβ a dα;eβ dα† LCu ∗ u L Ceβ ; ( c c) = ( c c) c c c c (1.238) OH dα;eβ a dα;eβ dα† LCu ∗ eβ† L C− u ∗ ; ( ) = ( ) T T 1 (1.239) OH dα;dβ;νi a dα;dβ;νi u LCdα dβ LCνi ; ( c) = ( ) c c c T (1.240) OH dα;dβ;νi a dα;dβ;νi dβ† LCu ∗ dα L C− νi ; ( ) = ( ) 1 (1.241) ( ) = ( ) (1.242) SO Chapter 1. From the standard model to 66 (10) T T T a dα;eβ U A A D α U CE β ; where MT c 12 T T 1 1 a(dα;eβ) = (U (A + A )D) α (Ec†B†U)c∗ β ; (1.243) MT c 12 T1 1 a(dα;eβ) = (Dc†D( †U+c∗ α )U) C(E β ; ) (1.244) MT c c 12 1 1 a(dα;eβ) = (Dc†D†Uc∗)α (Ec†B†U)c∗ β ; (1.245) MT 12 T 1 T 1T a(dα;dβ);ν =i ( U A ) A( D α )D CN βi ; (1.246) MT c 12 1T a(dα;dβ;νi) = ( Dc†(D†+Uc∗ β ) D) C(N αi : ) (1.247) MT 1 1 L γ / MT 2 α β ; SU L i ; ; − ( ) = ( ) ( ) (1.248) SU C 5 Here = (1 ) 2, is the triplet mass, = =1d 2 are (2) indices and =1 2 3 are (3) indices. The above analysis exhibits that the Higgs = 6 contributions are quite model A A D i j dependent,ij − ji and becauseij of this it is possible to suppress them in specific models of fermion masses. For instance, we can set to zero these contributions by the constraints = and = 0, except for = = 3. Also in this case we can make a naive model-independent estimation for the mass of the scalar leptoquark using the experimentalm lowerp bound on the proton lifetime. Approximating the inverse lifetime ofp theYu protonYd in the following way ∼| | MT5 2 τ p π e > : 4 → × Γ (1.249) 0 + 33 MT > : : × and taking ( ) 8 2 10 yr [11], we11 find the naive lower bound 4 5 10 GeV (1.250) This bound tells us that the triplet Higgs has to be heavy, unless some special condition on the matrices in Eq. (1.236) is fulfilled (see e.g. [84, 85]). Therefore since the triplet Higgs lives with the SM35 Higgs in the same multiplet we have to look for a doublet-triplet 1.6.3splitting mechanism . d d = 5 q q q ℓ uc uc dc ec In the presence of supersymmetryMT new = 5 operatorsMT of the type 1 1 ˜ ˜ and ˜ ˜ (1.251) 35 Cf. Sect. 4.4.3 for a short overview of the mechanisms proposed so far. 1.6. Proton decay 67 MT c c c c are generated via colored triplet Higgsino exchangeqqqℓ withu u madsse [159, 163]. These operators can be subsequently dressed at one-loop with an elMectroweakT m gauginom (gluino or wino) or higgsino leading to the standard and operators. Since the amplituded turns out to be suppressed just by the product ˜ , where ˜ is the soft d p scale, this implies a generic enhancement of the proton decay rate with respect to theÏ Kordinaryνµ = 6 operators. p π e Ï c c c c + Another peculiarity of = 5 operators is that the dominantqiqj qkℓl decayui u0 modej d+kel is i;j;k;lwhich; differs; from the standard nonsupersymmetric mode . A simple symmetry argumentSU C showsSU the reason:L the operators ˆ ˆ ˆ ˆ and ˆ ˆ ˆ ˆ (where = 1 2 3 are family indices and color and weak indices are implicit) must be invariant under (3) and (2) . This means that their color and weak indices must be antisymmetrized.i j Howeverk since this operators arei givenj by bosonic superfields, they must be totally symmetric under interchange of all indices. Thus the first operator vanishes for = = and theSU second vanishes for = . Hence a second or third generation memberqqqℓ must exist in the final state. In minimal supersymmetric (5) [94] the coefficient of the baryon number violat- α Y Y mg ing operator can be schematically written as; (see e.g. [164]) π MT m 3 10 5 q˜ 2 ˜ (1.252) 4 mq mg Y Y where we have assumed the dominance of the gluino exchange and that the sfermion ˜ ˜ 10 5 masses ( ) are bigger than the gluino one ( ), while and areSU couplings of the Yukawa superpotential. Though there could be a huge enhancement of the proton decay rate which brought to the claim that minimal supersymmetric (5) was ruled out [165, 166], a closer look to the uncertainties at play makes this claim much more weaker• [167]: SU T The Yukawa couplings in Eq. (1.252) areMd not directlyMe related to those of the SM, since in minimal (5) one needs to take into account non-renormalizable op- erators in order to break the relation = , and thus they can conspire to • suppress the decay mode [168]. A similar suppression could also originate from the soft sector even after includ- MT • ing the constraints coming from flavor violating effects [169]. Last but not least the mass of the triplet is constrained by the running only in the renormalizable version of the theory [166]. As soon as non-renormalizable operators (which are anyway needed for fermion mass relations) are included this is not true anymore [167]. In this respect it is remarkable that even in the worseMT case scenario of the renormalizable theory the recent accurate three-loop analysis in Ref. [170] increases by about one order of magnitude the upper bound on due to the running constraints. SO Chapter 1. From the standard model to 68 SU (10) Thus the bottom-line is that minimal supsersymmetric (5) is still a viable theory and more input on the experimental side is needed in order to say something accurate on 1.6.4proton decay. d R = 4 c ′ c ′′ c c c This last class of operatorsWRPV µi originatesℓihu λijk fromℓiℓiei theλijk -parityqiℓj dk violatingλijk ui dj superpotentialdk : of the MSSM λ′′ µ λ λ ˆ ˆ ˆ ˆ ˆ ˆ ′ ˆ ˆ = + ˆ + ˆR + ˆ (1.253) Notice that violates baryon number′ whilec ′′ , candc c violate lepton number. So for RPV λ qiℓj d λ u d d : instance we have the followingL ⊃ interactionsijk k inijk thei j -parityk violating lagrangian c c c d ˜ + ˜ + h.c. qℓu †d † (1.254) The tree-level exchange of ˜ generates the baryon violating operator with a λ′ λ′′ /m : coefficient which can be written schematicallydc as 2 ˜ (1.255) Barring cancellations in the family structure of this coefficient and assuming a TeV λ′ λ′′ . − : scale soft spectrum, the proton lifetime implies the generic bound [171] 26 R 10 (1.256) SU It’s easy to see that the -parity violating operators are generated in supersymmetric GUTs unless special conditions are fulfilled. For instance in (5) the effective trilinear couplings originate from the operator ijk i j k ; λ λ′ λ′′ SO R Λ 5ˆ 5ˆ 10ˆ (1.257) 1 2 ijk which leads to = = = Λ. Analogouslyi j k in H (10): the -parity violating trilinears stem from the operator MP Λ D E 16ˆ 16ˆ 16ˆ 16ˆ (1.258) Z If one doesn’t like small numbers such asSO in Eq. (1.256) the standard approach is 2 to impose a matter parity whichSO forbids the baryon andR lepton number violating operators [94]. A more physical option in (10) is instead suggested by Eq.R (1.258). Actually it seems that as soon as (10) is preserved the -parity violating trilinears are not generated. In order to better understandB L thisS point let us rephrase the -parity RP − ; in the following language [172] − 3( )+2 = ( ) (1.259) 1.6. Proton decay S 69 B L SO − where the spin quantum number is irrelevantB L as long;::: as the Lorentz symmetry is B L ; − ; ; ;::: preserved. Then, since is− a local generator of (10), it is enough to embed the SMR fermions in representationsSO with odd (e.g. 16 )R and the Higgs doublets B L H H in representations with even (e.g.− 10 120 126 210 ) in order to ensure exact -parity conservation. AfterSO the (10) breaking theRP fate of -parity depends on the order parameter responsible for the breaking. Employing either a 16 or a 126 RP H H RP H H : for the rank reductionh ofi −h(10) thei action of the h operatori h oni their VEV is R 16 = 16 or 126 = 126 (1.260) SO H In the latter case the -parity is preserved by theH vacuum and becomes an exact symmetry of the MSSM. This feature makes supersymmetricMB L/MP (10) modelsH withM 126B L − h i ∼ − very appealing [62]. On the other hand with a 16 atSO play the amount of R-parity B L violation− is dynamically controlled by the parameter , where 16 . Though conceptually interesting it is fair to say that in (10) it is unnatural to have the breaking scale much below the unification36 scale both from the point of view of unification constraints and neutrino masses . H H MB L 36 − MB L As we will see in Chapter 2 when the GUT breaking is driven either by a 45 or a 210− there are vacuum configurations such that can be pulled down till to the TeV scale without conflicting withSO unification constraints. On the other hand the issue of neutrino masses with a low is more serious. One has either to invoke a strong fine-tuning in the Yukawa sector or extend the theory with an (10) singlet (see e.g. [173]). SO Chapter 1. From the standard model to 70 (10) Chapter 2 Intermediate scales in nonsupersymmetric SO unification (10) SO The purpose of this chapter is to review the constraints enforced by gauge unification on the intermediate mass scales in the nonsupersymmetric (10) GUTs, a needed preliminary step for assessing the structure of the multitude of the different breaking patterns before entering the details of a specific model. Eventually, our goal is to envis- age and examine scenarios potentially relevant for the understandingSO of the low energy matter spectrum. In particular those setups that, albeit nonsupersymmetric, may exhibit a predictivity comparable to that of the minimal supersymmetric (10) [46, 47, 48], BscrutinizedL at length in the last few years. MU − The constraints imposed by the absolute neutrinoSO mass scale on the position of the threshold, together with the proton decay bound on the unification scale , provide a discriminating tool among theSO many (10) scenarios and the correspond- ing breaking patterns. These were studied at length in the 80’s and early 90’s, and detailed surveys of two- and three-stepSO (10) breaking chains (one and two interme- diate thresholds respectively) are found in Refs. [174, 100, 175, 64]. We perform a systematicU survey of (10) unification with two intermediate stages. In addition to updating the analysis to present day data, this reappraisal is motivated by (a) the absence of (1) mixing in previous studies, both at one- and two-loops in the gauge coupling renormalization, (b) the need for additional Higgs multiplets at some intermediate stages, and (c) a reassessment of the two-loop beta coefficients reported in the literature. The outcome of our study is the emergence of sizeably different features in some of the breaking patterns as compared to the existing results. This allows us to rescue previously excluded scenarios. All that before considering the effects of threshold corrections [176, 177, 178], that are unambiguously assessed only when the details of a specific model are worked out. Eventually we will comment on the impact of threshold effects in the Outlook of the thesis. SO Chapter 2. Intermediate scales in nonsupersymmetric unification 72 SO(10) H H H H It is remarkable that the chains corresponding to the minimal (10) setup with the smallest Higgs representations (10 , 45 and 16 , or 126 in the renormalizable case) and the smallestSO number of parameters in the Higgs potential, are still viable. The complexity of this nonsupersymmetric scenario is comparable to that of the minimal supersymmetric (10) model, what makes it worth of detailed consideration. In Sect. 2.1 we set the framework of the analysis. Sect. 2.2 provides a collection of theβ tools needed for a two-loop study of grand unification. The results of the numerical study are reported and scrutinized in Sect. 2.3. Finally, the relevant one- and two-loop -coefficients are detailed in Appendix A. 2.1 Three-step SO breaking chains SO G G SM →G →(10)G → The relevant (10) 2 1 symmetry breaking chains with two inter- mediate gauge groups 2 and 1 are listed in Table 2.1. Effective two-step chains are obtained by joining two of the high-energy scales, paying attention to the possible deviations from minimality of the scalar content in the remaining intermediate stageP (this we shall discuss in Sect. 2.3.2). For the purpose of comparison we follow closely the notation of Ref. [64], where denotes the unbroken D-parity [96, 97, 98, 99, 100]. For eachG step the Higgs represen- tation responsible for the breaking is given. H H The breakdown of the lower intermediateH symmetry 1 to the SM gauge group is drivenSO either by the 16- or 126-dimensional Higgs multiplets 16 or 126 . An important feature of the scenariosH with 126 is the fact that in such a case a potentially realistic (10) Yukawa sector can be constructed already at the renormalizable level (cf. Sect. 1.5).F TogetherF withH 10 all theH effectiveH / Dirac Yukawa couplings as well asM theU Majorana mass matrices at the SM level emerge from the contractions of the matter bilinears 16 16 with 126 or with 16 16 Λ, where Λ denotes the scale (above ) at which the effective dimension five Yukawa couplings arise. D-parity is a discrete symmetry acting as chargeSO conjugation in a left-right symmet- ric context [96, 97], and as that it playsSO the role of a left-right symmetry (it enforces for instance equal left and right gauge couplings). (10) invariance then implies exact D- parity (because D belongs to the (10) Lie algebra). D-paritySU R may be spontaneously broken by D-odd Pati-Salam (PS) singlets contained in 210 or 45 Higgs representa- tions. Its breaking can therefore be decoupled from the (2) breaking, allowing for different left and right gauge couplings [98, 99]. The possibility of decoupling the D-paritygL g breakingR from the scale of right-handed interactions is a cosmologically relevant issue. On the one hand baryon asymmetry cannot arise in a left-right symmetric ( = ) universe [96]. On the other hand, the spontaneous breaking of a discrete symmetry, such as D-parity, creates domain walls that, if massive enough (i.e. for intermediate mass scales) do not disappear, overclosing SO 2.1. Three-step breaking chains (10) 73 C L R C L R B L ÊÏ { } ÊÏ { − } Chain G2 G1 C L RP 45 C L R B LP I: ÊÏ210 {4 2 2 } ÊÏΛ {3 2 2 1 − } C L RP 210 C L R B L II: ÊÏ54 {4 2 2 } ÊÏΛ {3 2 2 1 − } C L R B LP 45 C L R B L III: ÊÏ54 {4 2 2 − } ÊÏΛ {3 2 2 1 − } C L R 45 C L R IV: ÊÏ210 {3 2 2 }1 ÊÏΛR {3 2 2 }1 C L RP 45 C L R 210 Σ V: ÊÏ {4 2 2 } ÊÏR {4 2 1 } C L RP 45 C L R VI: ÊÏ54 {4 2 2 } ÊÏλΣ {4 2 1 } C L R B L 210 C L R B L − − VII: ÊÏ54 {4 2 2 } ÊÏR {4 2 2 } C L R B LP 45 C L R B L 45 − Σ − VIII: ÊÏ {3 2 2 1 } ÊÏR {3 2 1 1 } IX: C L R 45 C L R B L 210 Σ − ÊÏ {3 2 2 }1 ÊÏσR {3 2 1 1 } X: C L RP 210 C L R B L 210 − ÊÏ {4 2 2 } ÊÏσR {3 2 1 1 } XI: C L R 210 C L R B L ÊÏ54 {4 2 2 } ÊÏ {3 2 1 1 − } XII: 45 SO 45 4 2 1 Λ 3 2 1 1 T Relevant (10) symmetry breaking chains via two intermediate gauge groups G1 and SOable 2.1: C L Y G2. For each step the representation of the Higgs multiplet responsible for the breaking is given in (10) or intermediate symmetry group notation (cf. Table 2.2). The breaking to the SM group 3 2 1 is obtained via a 16 or 126 Higgs representation. the universe [97]. These potential problems may be overcome either by confining D- parity at the GUT scale or by invoking inflation.SO The latter solution implies that domain walls are formed above the12 reheating temperature, enforcing a lower bound on the D-parity breaking scale of 10 GeV. Realistic (10) breaking patterns must therefore 2.1.1include this The constraint. extended survival hypothesis every stage of at theThroughout symmetry all breaking three stages chain of onlyrunning those we scalars assume are that present the sca tlarhat spectrum develop a obeys vacuum the expectationso called extended value (VEV)survival at hypothesis the current (ESH) or the [82] subsequent which requ leviresels that of the spontaneous symmetry breaking . ESH is equivalent to the requirement of the minimal number of fine-tunings to be imposed onto the scalar potential [83] so that all the symmetry breaking steps are performed at the desired scales. On the technical side one should identify all the Higgs multiplets needed by the SO Chapter 2. Intermediate scales in nonsupersymmetric unification 74 SO (10) SO C L R C L R C L R B L C L R B L { } { } { − } { − } ;Surviving; Higgs; ; multiplets; in; ; (10) subgroups; ; ; φ (10) 4; 2; 1 1 4 ;2 ;2 3 ;2 ;2 ;1 3; 2; 1 11; NotationδR10 2 − 2 − 10 (1 2 + 1) (1; 2; 2) (1; 2; 2; 0)1 (1 2 +1 0)1 δL16 2 2 2 2 16 (4 ;1 ;+ ) (4 ;1 ;2) (1 ;1 ;2 ; 1) (1 ;1 ;+ ; ) φ 16 2 16; ; 1 (4 ;2 ;1) (1; 2; 1; + ) ; ; ;1 126R 2 − 2− 126 (15 2 + ) (15; 2; 2) (1; 2; 2; 0) (1 2 + 0) L126 126 (10; 1; 1) (10; 1; 3) (1 1 3 1) (1 1 1 1) ∆126 126 (10; 3; 1) (1 3 1 1) ∆ 45 45 (15 1 0) (15; 1; 1) ; ; ; Λ210R 210 (15; 1; 1) ; ; ; Λ L45 45 (1 ;1 ;3) (1 1 3 0)σ ΣR45 45 (1 ;3 ;1) (1 3 1 0)σ ΣL210 210 (15; 1; 3) λ210 210 (15 3 1) 210 SO 210 (1 1 1) φ Scalar multiplets contributing to the running of126 the gaugeC L couplingsR B L forC aL givenR B L (10) Table 2.2: − − subgroup according to minimalφ fine tuning.φ The survival ofU B (notL required by minimality) is needed√ / − by aB realisticL / leptonic mass spectrum,10 as discussed126 in the text (in the 3 2 2 1 and 3 2 1 1 stages − only one linear combination of and φ remains). The (1) charge is given, up to a factor 3 2, by ( ) 2 (the latter is reported in the table).126 For the naming of the Higgs multiplets weH followH the notation of Ref. [64] with the addition of . When the D-parity (P) is unbroken the particle content must be left-right symmetric. D-parity may be broken via P-odd Pati-Salam singlets in 45 or 210 . H B L breaking pattern under consideration and− keep them according to the gauge symmetry down to the scale of their VEVs. This typically pulls down a large numberH of scalars in scenarios where 126 providesG the breakdown. On the other hand, one must take into account that the role of 126 is twofold: in addition to triggering the 1 breaking it plays a relevant role in the Yukawa sector where it providesH the necessary breaking of the down-quark/charged-lepton; ; mass de- generacy; (cf.; Eq. (1.190)). For this to work one needs a reasonably large admixture 10 of the 126 component in the effective electroweak doublets. Since (1 2 2) can mix 126 with (15 2 2) only below the Pati-Salam breaking scale, both fields must be present at the Pati-Salam level (otherwise the scalar doublet massC mL atrixR does not provide large enough components of both these multiplets inH the light Higg; ;s fields). Note that the same argument; ; applies also to the 4 2 1 intermediate1 stage when 2 126 oneC L mustR B retainL theC L doubletR B L component1 of 126 , namely (15 2 + ) , in order for it to − − 10 eventuallyφ admixφ with (1 2 + 2) in the light Higgs sector. On the other hand, at the 3 2 2101 and126 3 2 1 1 stages, the (minimal) survival of only one combination of the and scalar doublets (see Table 2.2) is compatible with the Yukawa sector constraints because the degeneracy between the quark and lepton spectra has already equations 2.2. Two-loop gauge renormalization group 75 been smeared-outSO by the Pati-Salam breakdown. In summary, potentially realistic renormalizableH YukawaH textures; in; settings with well-separated (10) and Pati-Salam breaking scalesC L callR for an additional fine; tuning; 10 in the Higgs sector. In; the; scenarios with; ; 126 , the 10 CbidoubletL R (1 2 2) , included 126 in Refs [174, 100, 175, 64], must1 be paired at the1 4 2 2 scale withX, XIan extraXII. (15 2 2) 10 126 scalarF bidoublet (or (1 2 + 2) with (15 2 + 2 ) at the 4 2 1 stage). This can affect the runningφ of the gauge couplings in chains I, II, III, V, VI, VII, and or the sake126 of comparison with previous studies [174, 100, 175, 64] we shall not include the multiplets in the first part of the analysis. Rather, we shall comment 2.2on their Two-loop relevance for gauge gauge unification renormalization in Sect. 2.3.3. group equations U ::: U N G ::: GN Gi In this section we⊗ report,⊗ in order⊗ to⊗ fix⊗ a consistent′ notation, the two-loop renormal- ization group equations (RGEs) for the gauge couplings. We consider a gauge group 1 1 of2.2.1 the form The non-abelian(1) (1) sector , where are simple groups. G ::: GN′ U ⊗ ⊗ 1 tLet us focusµ/µ first on the non-abelian sector corresponding to and defer dg the full treatment of the effects due to thep extra (1) factors to section 2.2.2. Defining 0 gp βp = log( ) we write dt p ;:::;N ′ = (2.1) β G ::: GN ⊗ ⊗ ′ where =1 is the gauge group label. Neglecting for the time being the abelian 1 components, the -functions for the gauge couplings read at two-loop gp level [179, 180,βp 181, 182, 183, 184]C Gp κS Fp ηS Sp π2 − 2 2 2 gp 2 11 4 1 = C (Gp) + C( F) +p C( )Gp κS Fp (4π2) − 3 3 3 2 34 2 2 20 2 2 + C2 S ( C( G)) +ηS 4S ( ) + ( ) ( ) (4 ) p 3 p p 3 g 2 2 2 2 + 4q ( ) + ( ) ( ) κ κC 3Fq S Fp ηC Sq S Sp Y Fp ; π2 − π ) h i 2 2 2 2 2 4 + 2 4 ( ) ( ) + ( ) ( ) 2 ( ) (2.2) κ ; (4 ) (4 ) η ; q p 1 6 1 where = 1 2 for Dirac and Weyl fermions respectively. Correspondingly, = 1 2 for complex and real scalar fields. The sum over = corresponding to contributions Chapter 2. Intermediate scales in nonsupersymmetric SO unification 76 (10) βp q F S R R ::: RN ⊗ ⊗ ′ to fromRp the other gauge sectors labelled by is understood.Gp Given a fermiond Rp 1 or a scalarS Rp field that transforms according to the representation = , where is an irreducible representation of thed R group of dimension ( ), the 2 S Rp T Rp ; factor ( ) is defined by ≡ d Rp 2 ( ) T Rp ( ) ( ) Rp (2.3) ( ) where ( ) is the Dynkin index of the representation . The corresponding Casimir eigenvalue is then given by C Rp d Rp T Rp d Gp ; 2 d G ( ) ( ) = ( ) ( ) (2.4) where ( ) is the dimension of the group. In Eq. (2.2) the first row represents the one- loop contribution while the other terms stand for the two-loop corrections, including that induced by Yukawa interactions.Y Fp The latterC isF accountep Y Y † ; d for in terms of a factor d Gp 4 1 2 ( ) = Tr ( ) (2.5) ( ) abc where the “general” Yukawa couplingY ψaψb hc h:c: + (2.6) ψa;b hc includes family as well as group indices. The coupling in Eq. (2.6) is written in terms of four-component Weyl spinors and a scalar field (be complex or real). The trace includes the sum over all relevant fermion and scalar fields. 2.2.2 The abelian couplings and mixing U U (1) N µ In order to include theA abelianb contributions to Eq. (2.2)ψf at two loops and the one- and two-loop effects of (1) mixing [185], letr us writeµ the most general interaction of abelian gauge bosons and a setψ off γ WeylµQf ψf fermionsgrbAb : as µ grb r;b ;:::;N Ab r r (2.7) Jµ ψf γµQf ψf N N grb × r U Qf The gauge couplingµ constants , = 1 , couple to the fermionic current = Ab. The gauge coupling matrixgrb can be diagonalized by two independentgrb r>b rotations: one acting on the (1) charges and theN N other/ on the gauge boson fields . For a given choice of the charges, can be set in a triangular form ( = 0 for ) by the gauge boson rotation. The resulting ( + 1) 2 entries are observable couplings. 2.2. Two-loop gauge renormalization group equations a 77 Fµν Since in the abelian case is itselfa aµν gauge invariant,a bµν the most general kinetic part F F ξabF F ; of the lagrangian reads at the− renormalizableµν − levelµν µ a b ξab < 1 1 A 6 | | a (2.8) 4 4 where = and 1. A non-orthogonal rotation of the fields may be per- formed to set the gauge kinetic term in a canonical diagonal form. Any further or- thogonal rotation of the gauge fields will preserve this form. Then, the renormalization prescriptiongrb may be conveniently chosengrb to maintain at each scale the kinetic terms canonical and1 diagonal on-shell while renormalizing accordingly the gauge coupling matrix . Thus, even if at one scale is diagonal, in generaland non-zero off-diagonal Uentries are generated by renormalization effects. One shows [187] that in the case the abelian gauge couplings are at a given scale diagonal equal (i.e. there is a (1) unification), there may exist a (scale independent) gauge field basis such2 that the abelian interactions remain to all orders diagonal along the RGE trajectory . dgrb In general, the renormalization of theg abelianraβab ; part of the gauge interactions is determined by dt d = (2.9) β Z / : where, as a consequence of gaugeab invariance,dt ab 1 2 Z 3 = log (2.10) 3 r with denoting the gauge-boson wave-functiongkb Qkgrb ; renormalization matrix. In order to further simplify the notation it is convenient≡ to introduce the “reduced” couplings [187] dgkb (2.11) gkaβab : that evolve according to dt k U =β U (2.12) The index labels the fields (fermions and scalars) that carry (1) charges. In terms of the reduced couplings the -function that governs the (1) running up β κ g g η g g to twoab loops isπ given byfa [179,fb 180, 181]sa sb 1 2 4 1 = κ gfagfbg + gfagfbg C Fq η gsagsbg gsagsbg C Sq (2.13) (4π) 3 fc 3 q sc q h 2 2 2 2 i 4κ 2 2 2 + gfagfb Y Y+† ; ( ) + + ( ) − (4π) 2 2 Tr 1 (4 ) Alternatively one may work with off-diagonalβ kinetic terms while keeping the gauge interactions diagonal2 [186]. Vanishing of the commutator of the -functions and their derivatives is needed [188]. Chapter 2. Intermediate scales in nonsupersymmetric SO unification 78 f (10) s U c C Rp where repeated indices are summed over, labelling fermionsGq( ), scalars ( ) and (1) 2 gaugeU groups ( ). The terms proportional to the quadratic Casimir ( ) represent the two-loop contributions of the non abelian components of the gauge group to the (1) gauge coupling renormalization. Correspondingly,U using the notation of Eq. (2.11), an additional two-loop term that represents the renormalization of the non abelian gauge couplings induced at two loops g by the (1) gauge fields is to bep added to Eq. (2.2), namely βp κ gfcS Fp η gscS Sp : π2 2 2 h 2 2 i ∆ = 4 4 ( f ) + F (s ) S (2.14)κ (4 ) p p ; η ≡; ≡ In1 Eqs. (2.13)–(2.14), we use the abbreviation 1 and and, as before, = 1 2 for Dirac and Weyl fermions, while = 1 2 for complex and real scalar fields 2.2.3respectively. Some notation U When at most one (1) factor is present, and neglecting the Yukawa contributions, the dαi− ai bij two-loop RGEs can be conveniently written as αj ; dt 1 − π − π 2 αi g / π β = ai bij SO (2.15) i 2 8 2 where = 4 . The -coefficients αandj for the relevant (10) chains are given in Appendix A. Substituting the one-loop solution forai intob theij right-hand side of Eq. (2.15) one αi− t αi− t ωjt ; obtains − − π π − 1 1 ˜ ωj aj αj / π ( ) bij (0)bij =/aj + log (1 ) (2.16) ωj t < 2 4 β bwhereij = (0) (2 ) and ˜ = . The analytic solution in (2.16) holds at two loops (for 1) up to higher orderβ effects. A sample of the rescaled -coefficients ˜ is given, for the purpose of comparison with previous results, in Appendix A. We shall conveniently writeβab the -functiongsa γ insr Eq.grb ; (2.13), that governs the abelian mixing, as π 1 γsr = 2 (2.17) (4 U) where include both one- and two-loop contributions. Analogously, the non-abelian gp beta function in Eq. (2.2), includingβ thep (1) contributionγp : in Eq. (2.14), is conveniently written as π2 = 2 (2.18) (4 ) 2.2. Two-loop gauge renormalization group equations 79 γp SO The functions for the (10) breaking chains considered in this work are reported in Appendix A.1. Finally, the Yukawa term inY Eq.Fp (2.5),ypk andY correspondinglykYk† ; in Eq. (2.13), can be written as Yk 4 k ×( ) = Tr k (2.19) where are the “standard” 3 3 Yukawa matricesSO in the family space labelledy bypk the flavour index . The trace is taken over family indices and is summed over the different Yukawa terms present at each stage of (10) breaking. The coefficients 2.2.4are given One-loop explicitly in matching Appendix A.2 The matchingG conditions between effective theoriesGp in the framework of dimensional regularization have been derived in [189, 190]. Let us consider first a simple gauge group spontaneously broken into subgroups . Neglecting terms involving loga- rithms of mass ratios which are assumed to be subleading (massive states clustered C Gp C G near the threshold), the one-loopαp− matching forαG− the gauge: couplings can be written as − π − π 1 2 1 2 ( ) ( ) = Gp (2.20) U X 12 U12Y U Let us turnTY to the case when several non-abelian simple groups (and at most one (1) ) spontaneously break whilst preserving aTX (1) charge. The conserved (1) generator can be written in terms of the relevant generators of the various Cartan TY piTi ; subalgebras (and of the consistently normalized ) as p i p X i = (2.21) 2 P C G where =3 1, and runs over the relevant (and )i indices. The matching condition αY− pi αi− ; is then give by − π 1 i 2 1 2 X ( ) = (2.28) 12 φ TY φ 3 h i h i This is easily proven at tree level [191]. Let us imagine that the gauge symmetry is spontaneously broken by the VEV of an arbitrary setDµ ofφ scalar∂µφ fieldsigiTi Aµ,i suchφ ; that = 0. Starting from the covariant derivative = + (2.22) µ gigj φ † TiTj φ ; we derive the gauge boson mass matrix ij h i h i 2 AY q A = µ i µ i (2.23) which has a zero eigenvector correspondingµij qj to the masslessq gaugej :field = , where j 2 X 2 = 0 with = 1 (2.24) Chapter 2. Intermediate scales in nonsupersymmetric SO unification 80 i X C (10) N U M 2 where forU = , if present,M