ULB-TH-98/18 Leptogenesis with virtual Majorana neutrinos 9 9 9 1 n a J.-M. Fr`ere, F.-S. Ling1, M. H.G. Tytgat2 and V. Van Elewyck3 J 8 1 Service de Physique Th´eorique, CP225 1 Universit´e Libre de Bruxelles v 7 Bvd du Triomphe, 1050 Brussels, Belgium 3 3 1 Abstract 0 9 9 / h p We present a mechanism of leptogenesis based on the out-of-equilibrium - p decay of a scalar particle into heavy virtual Majorana neutrinos. This e h scheme presents many conceptual advantages over the conventional scenario : v of Fukugita and Yanagida. In particular, the standard techniques of quantum i X field theory can be used to compute the lepton asymmetry, without resorting r a to the phenomenological approximations usually made to describe unstable particles. This simplification allows us to address in a well-defined framework some issues raised in the recent literature. We also show, in a toy model, that a successful leptogenesis scenario is possible and requires a rather light scalar particle, 106GeV < m < 1013GeV. A natural embedding of this scheme in a gauged unified theory encompassing the Majorana fermions seems however difficult. 1Aspirant FNRS. 2Collaborateur Scientifique FNRS. 3Chercheur FRIA. 1 Introduction Leptogenesis is an attractive scenario for the origin of the baryon number of the Universe. It rests on the idea that if an antilepton excess is created at a scale well above the electroweak phase transition, T 100GeV, it can ≫ be very efficiently converted into a net baryon asymmetry by spaleron-like processes, that violate B +L but preserve B L. In the simplest scenarios, − the initial lepton asymmetry is created in the out-of-equilibrium decay of heavy Majorana neutrinos [1]. Among other things, such a scheme has the advantage of separating the step of CP and L violation from the step of B violation, which occurs later through sphaleron-like processes, and thus to avoid the pitfalls of maintaining an out-of-equilibrium situation around the electroweak scale. (See i.e. [2] for a review of electroweak baryogenesis scenarios.) Also, as the conversion of L into B takes place at equilibrium and is essentially complete, these mechanisms are largely insensitive to the details of the non-perturbative baryon number violating processes. As is well known, CP violation is a crucial ingredient of leptogenesis [3] and it here arises from the interference between tree-level diagrams and the absorptive part of one-loop diagrams. Traditionally, only the one-loop vertex corrections were taken into account in most calculations [4], even though it was known [5] that the self-energy corrections, through which the different Majorana neutrinos can mix, do also contribute to the CP asymmetry. In particular, intheframework ofthe wave-function formalismof Weisskopf and Wigner [6], it has been argued that, in the limit of nearly degenerate Majo- rana neutrinos, the self-energy contribution could be significantly larger than the vertex term, thus giving an enhanced lepton asymmetry [7, 8, 9]. Using the exact solution of the wave equation with a complex matrix, this effect has been verified for the case of two scalar flavours [10]. The wave-function approach is only a phenomenological approximation however, and one might 2 wish for a more rigorous and systematic formulation. The problem, as is well-known, is that unstable particles are outside the realm of conventional quantum field theory, as they cannot be asymptotic states of the S-matrix. In particular, the self-energy corrections cannot be absorbed into the field wave-function renormalization constant, without destroying the hermiticity of the lagrangian.4 Another, but related, issue is that the Majorana propa- gation eigenstates are not well-defined, an effect that leads to an ambiguity in the initial conditions for leptogenesis.5 Toaddresssomeoftheseproblems, ithasbeenproposedin[12]toconsider lepton number violating scattering processes in which the Majorana particles appear only in intermediate states, like in l φ N∗ lc φ† (1.1) L → → L where l are the left-handed Standard Model (SM) leptons, φ is the Higgs L doublet and N are off-shell (*) Majorana neutrinos. Compared to the Ma- jorana decay, the main advantage of considering processes like (1.1) is that the rules of quantum field theory can be applied straightforwardly. It is for instance manifest that the self-energy corrections to the propagator of the Majorana must be included at one-loop. What is less obvious is how much these corrections contribute to the lepton asymmetry. Actually, as has been shownin[13],unitarityimpliesthatwhenallthescatteringchannelslike(1.1) are taken into account, the resulting lepton excess is precisely zero. This is actually just the requirement of departure from equilibrium: as the initial and final states in the processes (1.1) are the same, at equilibrium no lepton asymmetry can be created. Departure from equilibrium can be provided by the expansion of the Universe, that effectively selects a subset of the pro- cesses (1.1) and can thus lead to a non-vanishing lepton asymmetry [13, 9]. 4A more satisfying approach, that only slightly departs from the canonical rules of quantum field theory, has been advocated in [11]. 5Thishasbeenemphasizedin[10]forinstance,butpresumablyisawell-knownproblem. 3 In the present paper, we study a different mechanism, that is a variation onthe scattering scenario of [12]. In section 2, we will consider a heavy scalar particle, χ, that is allowed to decay into light (unspecified but sterile) right- handedfermionsandheavyright-handedMajorananeutrinos. TheMajorana neutrinos decay into the left-handed SM leptons andHiggs scalar. (See figure 1.) If we impose the mass hierarchy, m m m ,m = 0, M χ φ l ≫ ≫ the χ can be viewed as a source of Majorana neutrinos. This scenario en- compasses the conceptual advantages of the scattering processes (1.1) but furthemore leads to the production of a net lepton asymmetry. Among other things, we will verify that the self-energy corrections do indeed give a non- negligible contribution to the asymmetry. We will study the limit of degen- erate Majorana neutrino masses, and show that the asymmetry has a finite, well-defined expression at one-loop. (Phase counting shows that even in this case, CP violation effects are possible.) Also, the asymmetry so obtained is directly related to the initial abundance of the χ, independently of the basis chosen to define the Majorana states. Weprovide some numerical calculation performed for two flavours that make the various contributions (vertex and self-energy) to the asymmetry more explicit and give some useful estimates. Finally, we discuss the out-of-equilibrium conditions in the Early Universe, that puts limits on the Majorana and χ masses. These constraints compel the leptogenesis scenario to involve a neutral scalar χ at a scale between 106 and 1013GeV. In Section 3, we try to embed our scheme in a more physically motivated framework. For definiteness, we have in mind a natural gauge extension of the model, for instance SO(10). For simplicity, we have confined our argument to its subgroup SU(2) SU(2) U(1), which already imposes L R × × strong constraints. As we will show, adding further gauge degrees of freedom 4 has non-trivial consequences. In this framework, the χ+ is the singly charged component of a triplet of SU(2) while the vacuum expectation value of the R neutral component χ0 breaks SU(2) and gives a Majorana mass to the R right-handed neutrinos, χ+/√2 χ++ ∆ . R ≡  χ0 χ+/√2  −   The decay of the χ+ as the source of the lepton asymmetry is however im- mediately ruled out: the annihilation of χ+χ− pairs into photons is much too fast, so that the charged χ stay in thermal equilibrium at the epoch of interest, T m . The next possibility is to consider the decay of the neutral χ ∼ χ0 (figure 8). Because the lepton number violating decay rate is relatively slow in this case, we have to consider other competing rare decay processes. As almost no dilution is allowed, an analysis of the dominant decay channels reveals that coupling of scalar particle to the SU(2) gauge bosons is suffi- R cient to totally damp out the lepton asymmetry. Adding more fields could resolve this problem, but at the price of simplicity. 2 χ decay leptogenesis: self-energy and ver- tex corrections Wefirst concentrate onthevarioussources ofCP violationinthedecay ofthe scalarparticleχ. Thefinalstateconsideredisaright-handed(sterile)fermion accompaniedbyaleft-handedleptonandaHiggsboson, andreachedthrough the exchange of a virtual heavy Majorana particle χ l N∗ l l φ (see R L R → → diagram 1, figure 1). The Majorana neutrinos are labelled according to their mass; the following mass hierarchy guarantees that the intermediate Majorana neutrino is off mass-shell, M > M > M > m m ,m = 0 3 2 1 χ l φ ≫ 5 We want to study the particular consequences of this mass hierarchy on CP violation, as a theoretical framework and as a possible realistic scenario for leptogenesis. The decay of the χ is expected to produce a lepton asymmetry, since the intermediate Majorana can couple to both lepton-antiboson and antilepton- boson, with final total lepton number 2, 0 or -2 (figure 1).6 The most general Yukawa lagrangian for the particles involved in our scheme is of the form = g L ΦRN +G χlc RN +h.c. (2.1) yuk ij Li j kl Rk l L where L stands for the left-handed SM leptons, Φ is the Higgs doublet, L N are the heavy Majorana neutrinos, l are the light, sterile right-handed R fermions, and R = 1+γ5. Conventionnally, the Majorana states N are chosen 2 so that the Majorana mass matrix M is real and diagonal, which is always possible. The two Yukawa coupling matrices g and G, however, cannot be diagonalisedsimultaneouslywithM, ingeneral. Thelagrangian(2.1)leadsto the tree level decays for the χ of figure 1. The decay channels form hermitian conjugate pairs, e.g. II and III, and CP violation becomes possible only at the one loop level, where tree level and one-loop diagrams can interfere. We expect that both loops including χ and φ scalars will contribute to the global asymmetry, since CP violating phases appear in the coupling matrices g and G in the most general case. We can split the one-loop diagrams into self-energy loops and vertex corrections. While a calculation including the vertex correctionaloneiswell known toyieldanon-vanishing asymmetry, the self-energy correction has been often neglected. However, it has been argued in [8] and [9] that its contribution is far from negligible, and is even enhanced 6This assignment corresponds to total lepton number, left-handed plus right-handed. If the right-handed fermions are sterile, only the left-handed lepton number matters for leptogenesis. Our conclusions are essentially independent of the charge assignement cho- sen. 6 channel I channel IV lc lc R R c c l l N L N L f f Lf = 0 Lf = 0 channel II channel III lc lc R R c c l l N L N L f f Lf = -2 Lf = +2 Figure 1: Decay channels of χ for nearly degenerate Majorana masses. In our scheme, the inclusion of self- energyloopsisautomaticsincetheMajorananeutrinosonlyappearasvirtual intermediate states. According to the Cutkosky rules, the absorptive part of the one-loop di- agrams, which provides the imaginary part needed for the CP asymmetry, are given by the cut diagrams. Diagrams 1,2,5 of figure 2 are purely dis- persive (no unitarity cut operates) and therefore don’t contribute to the CP asymmetry. In particular, the self-energy correction of diagram 5 can be ab- sorbed in the wave-function renormalisation of the χ. Consequently, in this scheme, thesources ofCPviolationareprecisely thesameasthoserelevantin the conventional Majorana neutrinos decay scenario, namely, the asymmetry comes from loops involving the Higgs scalar. At tree level, the χ decay rate is m M M Γ [χ l l φ] = χ (g†g) (G†G) f(0)( i, l) (2.2) 0 → Rj Lk 64(2π)3 li li m m i,l χ χ X 7 lc R Diagram 1 Diagram 2 lc R N c c N lc c R lc l l R N L c N L f f Diagram 3 Diagram 4 lRc l lRc L c c f N N l L N f N lL l L f f Diagram 5 lc N R c c l N L lc R f Figure 2: One-loop corrections. where (x2 1)2 1 (y2 1)2 1 f(0)(x,y) = xy 1+ − ln(1 ) − ln(1 ) x2 y2 − x2 − x2 y2 − y2 ! − − This function has a well-defined limit when the Majorana are degenerate, 1 limf(0)(x,y) = 2x2 1+2x2(x2 1)ln(1 ). y→x − − − x2 while in the limit of very heavy Majorana neutrino m M, χ ≪ m3 (g†g) (G†G) Γ [χ l l φ] χ li li (2.3) 0 Rj Lk → → 3 26(2π)3 M M · Xi,l i l 8 At this order, the CP asymmetry splits into two parts that can be calculated separately. The vertex correction gives the following contribution to the asymmetry, Γas. [χ l l φ] ǫ = 01,v → Rj Lk (2.4) v Γ [χ l l φ] 0 Rj Lk → with m M M M Γas. [χ l l φ] = χ m(g†g) (g†g) (G†G) f ( i, l, n) 01,v → Rj Lk 32(2π)4 ℑ nl ni il v m m m i,l,n χ χ χ X (2.5) Thefunctionf (x,y,z)isagainwell-defined fordegenerate Majoranamasses. v In the limit m M as in Eq. (2.3) above, χ ≪ m5 m((g†g) (g†g) (G†G) ) Γas. [χ l l φ] χ ℑ nl ni il (2.6) 01,v → Rj Lk → 3 210(2π)4 M2M M · iX,l,n i l n The self-energy correction gives Γas. [χ l l φ] ǫ = 01,w → Rj Lk (2.7) w Γ [χ l l φ] 0 Rj Lk → with m M M M Γas. [χ l l φ] = χ m(g†g) (g†g) (G†G) f ( i, l, n) 01,w → Rj Lk 16(2π)4 ℑ nl ni il w m m m i,l,n χ χ χ X (2.8) The complete expression for f is cumbersome, but it can be expressed in w terms of simple functions. Again, the limit of degenerate Majorana masses gives a well defined value, M M M 1 3M2 M2 3M4 M2 m2 lim f ( i, l, n) = +( + )ln − χ Mi,Ml,Mn→M w mχ mχ mχ 16 − 8m2χ 4m2χ 8m4χ M2 (2.9) Also, in the limit of a light χ, m M, χ ≪ m5 m((g†g) (g†g) (G†G) ) Γas. (χ l l φ) χ ℑ ni nl il (2.10) 01,w → Rj Lk → 3 29(2π)4 M2M M · iX,l,n n l i 9 In the conventional scenario, with Majorana neutrinos in the initial state, there are unphysical singularities 1/(M2 M2) in the expression of the ∝ i − j self-energy term (see for instance [10] or [12]). They signal a breakdown of the perturbative expansion for ∆M Γ, where M and Γ are respectively the ∼ mass and width of the Majorana neutrinos, and are responsible for the en- hancement of the lepton asymmetry from the self-energy contribution with respect to one from the vertex. The present scheme offers no such singu- larities, and the limit of degenerate Majorana neutrinos is finite, Eq. (2.9). This holds of course as long as the Majorana are off mass shell, which is guaranteed by our choice of mass hierarchy, because the mass of the χ par- ticle limits the value of q2 flowing through the Majorana propagator to be at most m2 M2. In the case of 2-body scattering of light particles, the χ ≪ i self-energy correction presents the same pole, because the centre of mass en- ergy in a scattering process is not limited in principle, even if this scattering is supposed to occur in the thermal bath at the temperature T2 M2, i.e. ≪ i at the epoch when the lightest heavy Majorana neutrino decays. This pole enhances the contribution of the self-energy compared to the one from the vertex in the degenerate limit, but necessitates one to use either a quite elab- orate resummation scheme (for instance as in [12] which is limited to weak mixing) or a wave-functional approach [10]. The absence of such singularities is another non-negligeable simplification offered by the present approach. On the other hand, there is no large en- hancement M/Γ either (provided m < M). However, the asymmetry is χ ∼ well defined for all Majorana mass patterns, including the limit of degenerate Majorana masses, for which the asymmetry does not necessarily vanish. Letusalsore-emphasisethat,contrarilytothescatteringprocessesof(1.1) studied in [12, 13], a true lepton asymmetry can be produced here. This is simply because the χ is unstable below T < m while the inverse decay, or χ recombinations is Boltzmann suppressed, which is another way to state that 10