betaFeedback

WorksPhysics and Astronomy

Dirac leptogenesis with charged leptons

Abstract

Popular frameworks that explain the baryon asymmetry of the Universe (BAU) are electroweak baryogenesis, requiring a first-order phase transition, and Majorana leptogenesis, which relies on lepton number violation in the neutrino sector. Dirac leptogenesis offers an alternative mechanism that avoids both requirements, instead allowing the BAU to follow from previously generated chiral asymmetries that remain out of equilibrium in the early Universe. Conventional realizations of this scenario typically invoke ultra-light Dirac neutrinos. In this work, we generalize the mechanism to include Standard Model charged leptons by assuming that their mass-generating interactions have a UV origin and arise from higher-dimensional operators. For concreteness, the analysis is performed within a Froggatt-Nielsen setup, which allows us to naturally address fermionic mass hierarchies, too, now linked to the BAU, and we show that the proposed mechanism is viable by supercooling the flavon phase transition. The framework predicts a flavon with a (1.7-50) GeV mass, bounded by the non-observation of lepton flavor violation and the requirement for the Universe to not reheat above sphaleron temperatures, which could be indirectly probed by future lepton flavor violation experiments.

Introduction

The observed baryon asymmetry of the Universe (BAU) [1, 2, 3, 4] cannot be explained within the current version of the Standard Model (SM) of particle physics, which motivates the search for new particles and interactions. Since Sakharov’s foundational work [5], numerous mechanisms have been proposed to explain this matter-antimatter asymmetry. Among the most extensively studied are electroweak baryogenesis [6, 7, 8] and Majorana leptogenesis [9, 10, 11]. The former requires a sufficiently strong first-order electroweak phase transition (FOPT), while the latter relies on lepton-number violation (LNV) in the neutrino sector, neither of which is provided by the SM.

Dirac leptogenesis [12, 13] offers an alternative based on Dirac neutrinos that requires neither a FOPT nor LNV in the neutrino sector. In its standard realization, the SM is supplemented with right-handed neutrinos, νR\nu_R, with tiny Yukawa couplings that account for the observed light neutrino masses [14, 15]. As shown in Ref. [12], these feeble interactions prevent left-right equilibration, allowing a primordial chiral asymmetry in the neutrino sector to remain sequestered until after electroweak (EW) sphalerons freeze out. While active, sphalerons partially convert the left-handed lepton asymmetry into baryon number, leaving a nonzero BAU that survives to the present day.

The simplicity and minimalism of Dirac leptogenesis have motivated numerous studies connecting the origin of the BAU via this mechanism to other open questions in particle physics and cosmology. These include Dirac neutrino mass generation through minimal ultraviolet (UV) completions [16, 17, 18], Dirac seesaws [17], and radiative mechanisms [19], as well as connections to the flavor puzzle [20, 21], dark matter [19, 22, 23, 24], inflation and reheating [25, 26, 27, 28], and the strong CP problem [26, 29, 30]. Since the original proposal [12] leaves the source of the primordial chiral asymmetry unspecified, several production mechanisms have also been explored [23, 25, 28, 31].

Despite their diversity, these realizations predominantly retain the central ingredient of the original mechanism, namely small fundamental or effective Dirac-neutrino Yukawa couplings that keep the right-handed neutrino asymmetry sequestered until after sphaleron freeze-out. Even when these couplings originate from a UV flavor symmetry or seesaw-like constructions, the interactions relevant during this epoch are typically already well described by the low-energy effective theory with the feebly coupled νR\nu_R sector serving as the asymmetry reservoir.

A different possibility was explored in Ref. [32], where the chiral reservoir is provided instead by the right-handed electron. In this scenario, the generated electron asymmetry remains successfully sequestered until sphaleron freeze-out because a non-standard cosmological evolution increases the Hubble rate, thus delaying the equilibration of electron Yukawa interactions. This demonstrates that the essential ingredient of Dirac leptogenesis is not necessarily the presence of a right-handed neutrino, but rather the existence of a chiral charge whose equilibration is sufficiently delayed.

In this work, we show that the chiral asymmetry sequestering mechanism of Dirac leptogenesis can also be extended beyond the right-handed neutrino reservoir, without a modified cosmological history above sphaleron freeze-out as a necessary condition. We demonstrate that this is realizable when giving a UV origin to the Yukawa couplings by considering fermion mass-generating interactions that remain higher-dimensional before sphaleron freeze-out, i.e. without yet reducing to effective dimension-four Yukawa couplings. In this case, the corresponding chirality-changing rates can be parametrically different from those naively inferred from the observed low-energy fermion masses. The same exponential suppression that explains fermion hierarchies also leads to a suppression of the equilibration rate. As a consequence, the right-handed components of other SM leptons, which are already present in the theory, can act as asymmetry reservoirs. We demonstrate this mechanism within a Froggatt-Nielsen (FN) framework [33] in which the FN symmetry remains unbroken until after sphaleron freeze-out.

The paper is structured as follows. In Section II we review the working principle of Dirac leptogenesis, and identify the standard requirement to successfully sequester the lepton asymmetry until sphaleron freeze-out. In Section III, we show how higher-order operators that generate fermion masses in an FN setup allow extending the leptogenesis framework beyond neutrinos, as long as the FN symmetry remains unbroken until after sphaleron freeze-out. We review the relevant constraints from flavor-violation observables in § 1.3 and from the supercooling of the FN phase transition in § 1.4. Finally, we conclude in Section VI.

Dirac leptogenesis in a nutshell

The sequestering mechanism underlying Dirac leptogenesis [12] is based on the fact that EW sphalerons act only on fields charged under SU(2)LSU(2)_{L}. While active, sphalerons partially convert a left-handed neutrino asymmetry (ΔLL\Delta L_{L}) into a baryon asymmetry, with a compensating lepton asymmetry (ΔLR\Delta L_{R}) remaining sequestered in right-handed neutrinos. This process can take place with global lepton number remaining conserved (ΔL=ΔLL+ΔLR=0\Delta L = \Delta L_{L} + \Delta L_{R} = 0) during the generation of the neutrino asymmetry. A final non-vanishing baryon asymmetry can then be obtained, provided the right-handed component remains out of equilibrium, while sphalerons are active. More specifically, successful sequestering requires interactions that would equilibrate the two chiral sectors to remain inefficient until after sphaleron freeze-out. This ordering allows a persistent nonzero baryon asymmetry to be generated before sphaleron processes cease. Subsequent left-right equilibration can redistribute the lepton asymmetry but cannot erase the previously generated BAU.

To better understand how the above can be realized in a given model, consider the collision term CC of the Boltzmann equations describing the evolution of the leptonic asymmetry. It can be written schematically as

C=S+WLR+Tsph,(1)C = S + W_{LR} + T_{\mathrm{sph}} , \tag*{(1)}

where SS includes processes that source the chiral lepton asymmetry, WLRW_{LR} those that wash it out, and TsphT_{\mathrm{sph}} represents the lepton-to-baryon asymmetry conversion via EW sphalerons. In the present analysis, we do not concern ourselves with a specific asymmetry production mechanism, namely with concrete interactions behind SS (see e.g. Refs. [12, 23, 25, 28, 31] for different production scenarios), but focus instead on the interplay between WLRW_{LR} and TsphT_{\mathrm{sph}}.

Importantly, each term in Eq. (1) is efficient over a finite temperature range. The weak sphaleron processes in the SM, inducing TsphT_{\mathrm{sph}}, are active within a wide temperature range [34, 35, 36, 37]

130 GeV≲T≲1012 GeV,(2)130\ \mathrm{GeV} \lesssim T \lesssim10^{12}\ \mathrm{GeV} , \tag*{(2)}

while the source terms, contributing to SS, on the other hand, are usually most efficient around a localized temperature, which we denote TST_{S}. Finally, the efficiency of the left-right equilibration processes, sourcing WLRW_{LR}, depends on the available operators communicating between the two chiral components.

Crucially, when the efficiency “windows” of WLRW_{LR} and TsphT_{\mathrm{sph}} overlap, the asymmetry generated by SS will efficiently wash out. In the original Dirac leptogenesis scenario [12], the smallness of the Dirac neutrino Yukawas guarantees that such an overlap does not occur, and therefore a non-vanishing baryon asymmetry can be generated. Indeed, a more general conclusion can be extracted from this work, i.e. given a renormalizable Yukawa term for a chiral fermion ff charged under SU(2)LSU(2)_{L}, with coupling yfy_{f}, the efficiency windows of WLRW_{LR} and TsphT_{\mathrm{sph}} will not overlap during radiation domination provided that4 [12, 38, 39, 40]

yf<few×10−7.(3)y_{f} < \mathrm{few} \times10^{-7} . \tag*{(3)}

If νR\nu_{R} is introduced into the SM, such that a Yukawa interaction with the Higgs boson and the active neutrinos can generate light Dirac neutrino masses, Yukawa couplings of the order yν∼O(10−13)y_{\nu} \sim\mathcal{O}(10^{-13}) are needed to satisfy oscillation data. Since this value comfortably satisfies the bound in Eq. (3), right-handed neutrinos are natural candidates to act as the necessary secluded reservoir needed for successful Dirac leptogenesis [12].

The diagram in Figure 1 summarizes the mechanism described above. Starting from symmetric initial conditions, where the lepton number stored in left- and right-handed fermions, denoted by LLL_{L} and LRL_{R}, respectively, is zero, a source term becomes efficient and leads to a period of asymmetry production (SS) that generates a non-vanishing asymmetry in the chiral components while global lepton number remains conserved (ΔL=ΔLL+ΔLR=0\Delta L = \Delta L_{L} + \Delta L_{R} = 0). Weak sphalerons (sph), while active, partially convert the left-handed lepton asymmetry into a baryon asymmetry. Next, there are two possibilities. If the left-right equilibration processes (equ) become efficient only after sphalerons have shut off, they drive the system to LL=LRL_L = L_R without changing the previously generated baryon asymmetry. This is represented in the diagram with a blue arrow, and can be realized for Dirac neutrinos as shown in Ref. [12]. On the other hand, if there is a temperature range below the asymmetry production scale for which (sph) and (equ) become efficient simultaneously, the generated asymmetry will be fully washed out in both the baryonic and leptonic sectors. This is represented by the red arrow and is realized by any fermion not satisfying the bound in Eq. (3), which is the case for all charged fermions in the SM. The precise washout trajectory depends on the interplay of sphalerons and equilibration processes, and is, for simplicity, depicted as a straight line. Clearly, in the case when the equilibration becomes efficient before sphalerons, there will no longer be a lepton asymmetry to convert, also preventing a successful BAU.

Figure 1. Schematic of the chiral lepton asymmetry evolution and relevant processes in Dirac leptogenesis. When left-right equilibration processes become efficient after sphalerons have frozen out (blue arrow), a BAU transferred from the leptonic sector is preserved until today. When their efficiency range overlaps with that of the sphalerons (red arrow), both lepton and baryon asymmetry are erased. See main text for more details.

Note that Eq. (3) assumes renormalizable Yukawa couplings during radiation domination. Under these assumptions, washout effects become efficient below a certain temperature TLRdim=4T_{\mathrm{LR}}^{\mathrm{dim}=4}. However, if the Yukawa interactions are replaced by higher-dimensional operators involving additional fields, the corresponding washout processes become efficient above a given temperature TLRdim>4T_{\mathrm{LR}}^{\mathrm{dim}>4}.5 The dimension-five Weinberg operator exemplifies this behavior, since the ΔL≠0\Delta L \ne0 washout processes it mediates, which are relevant for Majorana leptogenesis, are efficient only at high-enough temperatures. This temperature dependence suggests a way to extend asymmetry sequestering to SM charged leptons, whose renormalizable Yukawa couplings do not satisfy Eq. (3). If their mass-generating interactions remain higher-dimensional while sphalerons are active, left-right equilibration can become inefficient as the Universe cools, opening a temperature window in which a chiral asymmetry can be generated and partially converted into a baryon asymmetry. The observed Yukawa couplings may then emerge after sphaleron freeze-out, when subsequent left-right equilibration can no longer erase the BAU. In the next section, we describe how this idea can be realized within a Froggatt-Nielsen framework [33] whose flavor symmetry remains unbroken until after sphaleron freeze-out. We examine the conditions under which the higher-dimensional interactions decouple sufficiently early for SM right-handed charged leptons to serve as asymmetry reservoirs, while reproducing their observed masses after FN-symmetry breaking.

Dirac leptogenesis in a Froggatt-Nielsen construction

Setup

In an FN realization of the SM, the fermionic mass-generating operators are, except for the heaviest fermions, higher-dimensional [33]. More specifically, the chiral SM fermions carry different O(1)\mathcal{O}(1) charges under a global U(1)FNU(1)_{\mathrm{FN}} symmetry, and there exists a complex scalar SM singlet Φ\Phi, the flavon, with a non-zero FN charge, e.g. qΦFN=−1q_{\Phi}^{\mathrm{FN}}=-1, while the Higgs is uncharged. After symmetry breaking, once ⟨Φ⟩=vΦ/2\langle\Phi\rangle= v_{\Phi}/\sqrt{2}, the complex flavon field can be written in terms of an axial Nambu-Goldstone mode aa, called Axion-Like Particle (ALP) in the following, and a radial mode ρ\rho as

Φ=vΦ+ρ2eia/vΦ.(4)\Phi= \frac{v_{\Phi}+\rho}{\sqrt{2}}e^{ia/v_{\Phi}} . \tag*{(4)}

As a consequence of the U(1)FNU(1)_{\mathrm{FN}} symmetry, the invariant Yukawa interactions now contain, in addition to the Higgs, nf≥0n_f \ge0 insertions of the new field Φ\Phi. For a generic fermion ff, with left-handed component contained in an SU(2)LSU(2)_L doublet Ψ\Psi, the fermionic masses come from interactions of the type

−LFNeff⊃cfij(ΦΛFN)nfijΨ‾L,iHfR,j+h.c.,(5)-\mathcal{L}_{\mathrm{FN}}^{\mathrm{eff}} \supset c_f^{ij}\left(\frac{\Phi}{\Lambda_{\mathrm{FN}}}\right)^{n_f^{ij}}\overline{\Psi}_{L,i}Hf_{R,j}+\mathrm{h.c.}, \tag*{(5)}

with Hf=HH_f = H for down quarks and charged leptons, and Hf=iσ2H∗H_f = i\sigma_2 H^* for up quarks and neutrinos, where HH is the Higgs doublet. These interactions generate the diagrams shown in the upper panel of Figure 2. The coefficients cfijc_f^{ij} are of O(1)\mathcal{O}(1), since they are a function of the ultra-violet couplings that are assumed to be of O(1)\mathcal{O}(1). The exponent nfijn_f^{ij} depends on the charge assignments of the fermions under the U(1)FNU(1)_{\mathrm{FN}} symmetry, and the indices {i,j}\{i,j\} denote different flavors. Such terms can originate e.g. from integrating out vector-like fermion mediators with masses around the scale ΛFN\Lambda_{\mathrm{FN}}. Finally, the operator with nfn_f flavons has a mass dimension of 4+nf4+n_f, rendering it higher-dimensional for nf>0n_f>0.

Effective diagrams contributing to left-right equilibration of the Dirac fermion $f$

Figure 2. Effective diagrams contributing to left-right equilibration of the Dirac fermion ff. The dots in the upper diagrams signify nfn_f flavons in the final state.

We call TFNT_{\mathrm{FN}} the temperature of the Universe at which the Froggatt-Nielsen phase transition (FNPT) occurs and U(1)FNU(1)_{\mathrm{FN}} is broken spontaneously, such that the flavon gets a vacuum expectation value (VEV), ⟨Φ⟩≡vΦ/2\langle\Phi\rangle\equiv v_\Phi/\sqrt{2}. Once this happens, effective renormalizable Yukawa couplings arise, namely6

−LYukFN=yefffΨ‾LHffR+h.c.forT<TFN,(6)-\mathcal{L}^{\mathrm{FN}}_{\mathrm{Yuk}}=y_{\mathrm{eff}}^f\overline{\Psi}_L H_f f_R+\mathrm{h.c.}\quad\text{for}\quad T<T_{\mathrm{FN}}, \tag*{(6)}

with the effective Yukawa yefffy_{\mathrm{eff}}^f given in terms of the parameter ϵ≡vΦ/(2ΛFN)\epsilon\equiv v_\Phi/(\sqrt{2}\Lambda_{\mathrm{FN}}) as

yefff=cfϵnf.(7)y_{\mathrm{eff}}^f=c_f\epsilon^{n_f}. \tag*{(7)}

These effective interactions give rise to the diagrams shown in the lower panel of Figure 2. A small hierarchy vΦ<ΛFNv_\Phi<\Lambda_{\mathrm{FN}} then naturally leads to hierarchical fermion masses, since below the electroweak phase transition (EWPT) temperature TEWT_{\mathrm{EW}} and below TFNT_{\mathrm{FN}}, Eq. (6) leads to

mf=cfϵnfv2,(8)m_f=c_f\epsilon^{n_f}\frac{v}{\sqrt{2}}, \tag*{(8)}

with vv the Higgs VEV. Lighter fermions require larger values of nfn_f, since it leads to larger powers of the suppression factor ϵ\epsilon. With the Cabibbo angle ∼0.23\sim0.23, we assume the expansion parameter to be of the same order of magnitude, namely ϵ∼O(0.1)\epsilon\sim O(0.1), in the following. If a different ϵ\epsilon were chosen, the cfc_f values would have to be tuned in order to obtain the correct fermion masses and mixings.

Avoiding washout

The effective interactions in Eqs. (5) and (6) generate annihilation and scattering processes that contribute to the chiral-lepton-asymmetry washout WLRW_{\mathrm{LR}}. Representative diagrams are shown in Figure 2, where the squared effective vertex is only valid below TFNT_{\mathrm{FN}}. Further relevant tree-level diagrams can be obtained by attaching, e.g., a top-quark pair or soft gauge bosons to the Higgs leg. Even though these extra diagrams can contribute with the same order of magnitude as the ones considered in Figure 2, we ignore them in our calculations as the qualitative conclusions are expected to remain robust under their inclusion. For more details on these additional effects, we refer the reader to [38, 39, 40], and to Appendices A 2 and A § 4.1.2 where the effects of thermal masses and boson enhancements are estimated, respectively.

To determine whether or not the left-right equilibration processes related to fermion ff are efficient at a given temperature TT, we must compute the corresponding thermally-averaged interaction rate ⟨ΓLRf⟩\langle\Gamma_{\mathrm{LR}}^f\rangle. For the tree-level left-right annihilation rate corresponding to the diagram in the top-left corner of Figure 2 and in the limit where the center-of-mass energy ss is much larger than the masses of the involved particles, s≫{mi}s\gg\{m_i\}, the rate is given by (see Appendix A 1a for more details)

⟨ΓLRf⟩(T)=κfd>4∣cf∣2T2nf+1ΛFN2nf,(9)\langle\Gamma_{\mathrm{LR}}^f\rangle(T)=\kappa_f^{d>4}|c_f|^2\frac{T^{2n_f+1}}{\Lambda_{\mathrm{FN}}^{2n_f}}, \tag*{(9)}

with

κfd>4=nf(nf+1)!8(2π)2nf+1.(10)\kappa_f^{d>4}=\frac{n_f(n_f+1)!}{8(2\pi)^{2n_f+1}}. \tag*{(10)}

When the flavon obtains a non-vanishing VEV, effective 4-dimensional Yukawas, given by Eq. (6) and represented by the square vertices in Figure 2, contribute to the left-right equilibration rate. For the bottom left diagram in Figure 2, one obtains (see Appendix A 1c for more details)

⟨ΓLR⟩(T)≃κfd=4∣yefff∣2T,(11)\langle\Gamma_{\mathrm{LR}}\rangle(T)\simeq\kappa_f^{d=4}|y_{\mathrm{eff}}^f|^2T, \tag*{(11)}

with κfd=4=\kappa_f^{d=4}= few ×10−3\times10^{-3}. In particular, for the annihilation process fL+f‾R→t+t‾f_L+\overline{f}_R\rightarrow t+\overline{t} and fL+f‾R→H+Vμf_L+\overline{f}_R\rightarrow H+V_\mu, as well as their scattering counterparts, one obtains ⟨ΓLR⟩∼yt2∣yefff∣2T\langle\Gamma_{\mathrm{LR}}\rangle\sim y_t^2|y_{\mathrm{eff}}^f|^2T and ⟨ΓLR⟩∼g2∣yefff∣2T\langle\Gamma_{\mathrm{LR}}\rangle\sim g^2|y_{\mathrm{eff}}^f|^2T, respectively, with yty_t the top Yukawa and gg a gauge coupling. Clearly, the result in Eq. (11) also holds for heavy fermions with nf=0n_f=0 with the effective Yukawa replaced by the fundamental one.

Processes mediating left-right equilibration become efficient within temperatures for which

⟨ΓLR⟩(T)≳H(T),(12)\langle\Gamma_{\mathrm{LR}}\rangle(T)\gtrsim H(T), \tag*{(12)}

where H(T)H(T) is the Hubble rate, which in the radiation-dominated7 era reads

H(T)=8π3g∗90T2MPl.(13)H(T) = \sqrt{\frac{8\pi^3 g_*}{90}}\frac{T^2}{M_{\mathrm{Pl}}}. \tag*{(13)}

with g∗g_* the effective number of relativistic degrees of freedom of the thermal plasma and MPlM_{\mathrm{Pl}} the reduced Planck mass.

Based on the rates derived above, we can determine the efficiency windows for the left-right equilibration processes of a fermion ff that couples to nfn_f flavons. To that end, we define TLRT_{LR} as the temperature at which the equilibration rate equals the Hubble rate, which can take different values depending on the FN charge assignments and whether or not the U(1)FNU(1)_{\mathrm{FN}} symmetry is broken.

First, for nf=0n_f = 0, i.e. as for the SM Yukawas, the temperature range at which the equilibration rate surpasses the Hubble rate is given by

T<TLR=908π3g∗κfd=4∣cf∣2MPl.(14)T < T_{LR} = \sqrt{\frac{90}{8\pi^3 g_*}}\kappa_f^{d=4}|c_f|^2 M_{\mathrm{Pl}}. \tag*{(14)}

Second, with nf>0n_f > 0, and for temperatures below ΛFN\Lambda_{\mathrm{FN}} the corresponding range reads

T>TLR=ΛFN(ΛFNMPl1κfd>4∣cf∣28π3g∗90)12nf−1,(15)T > T_{LR} = \Lambda_{\mathrm{FN}}\left(\frac{\Lambda_{\mathrm{FN}}}{M_{\mathrm{Pl}}}\frac{1}{\kappa_f^{d>4}|c_f|^2}\sqrt{\frac{8\pi^3 g_*}{90}}\right)^{\frac{1}{2n_f-1}}, \tag*{(15)}

before the FNPT, and

T<TLR=908π3g∗κfd=4∣yfeff∣2MPl(16)T < T_{LR} = \sqrt{\frac{90}{8\pi^3 g_*}}\kappa_f^{d=4}|y_f^{\mathrm{eff}}|^2 M_{\mathrm{Pl}} \tag*{(16)}

after the FNPT. It becomes evident from the lower bound on TLRT_{LR} in Eq. (15) that the washout rate associated with fermion ff decreases for falling temperatures in the range TFN<TUV<ΛFNT_{\mathrm{FN}} < T_{\mathrm{UV}} < \Lambda_{\mathrm{FN}} when nf>0n_f > 0. This is due to the fact that, in the higher-dimensional regime of the interactions, the ratio Φ/ΛFN\Phi/\Lambda_{\mathrm{FN}}, which is set to the fixed value of vΦ/ΛFN∼0.1v_\Phi/\Lambda_{\mathrm{FN}} \sim0.1 only after the flavon acquires a nonzero VEV, appears as mtherm/ΛFN∼T/ΛFNm_{\mathrm{therm}}/\Lambda_{\mathrm{FN}} \sim T/\Lambda_{\mathrm{FN}}, where mthm_{\mathrm{th}} corresponds to the flavon thermal mass. The flavons have to be present as physical fields in the equilibration processes, and their thermal masses decrease as the Universe cools down. Then, as temperature decreases, the ratio decreases more and more, further suppressing the equilibration rate. After symmetry breaking, the same temperature scaling as in the SM is recovered by setting the flavon to its VEV, which gives a contribution to the equilibration rate that dominates over the processes with dynamical flavons when T≪ΛFNT \ll\Lambda_{\mathrm{FN}}.

The derived scaling of the ratio ⟨ΓLR⟩/H\langle\Gamma_{\mathrm{LR}}\rangle/H for different fermions is shown in Figure 3, for ϵ=0.1\epsilon= 0.1. For fundamental or effective d=4d = 4 Yukawas, the ratio is monotonically increasing with temperature and firmly inside the wash-out region for electron, muon, and tau leptons (see Table I). On the other hand, for nf>0n_f > 0 and temperatures above the FN breaking scale, the ratio decreases with temperature, and for low-enough values of TFNT_{\mathrm{FN}} relative to ΛFN\Lambda_{\mathrm{FN}} it exits the shaded wash-out region where WLR\mathcal{W}_{\mathrm{LR}} and TsphT_{\mathrm{sph}} overlap. In that regime, a generated asymmetry is safe from washout. However, as soon as the flavon acquires a non-vanishing VEV, the renormalizable effective vertices will also contribute to the equilibration of the chiral asymmetry and will generally dominate over the processes with nfn_f flavons in the final state. We note that the number of decreasing lines that are able to exit the washout region before sphaleron freezeout is a function of ΛFN\Lambda_{\mathrm{FN}}, which becomes clear in Figure 4. For this illustration we choose a large-enough FN scale, ΛFN=5×108 GeV\Lambda_{\mathrm{FN}} = 5 \times10^{8}\,\mathrm{GeV}, so that all fermions with nf≥1n_f \geq1 can technically constitute viable reservoirs.

Figure 3. Illustration of the scaling of the ratio ⟨ΓLR⟩/H\langle\Gamma_{LR}\rangle/H (solid lines) for cf∼O(1)c_f \sim O(1), ϵ=0.1\epsilon= 0.1, and for different values of nfn_f, i.e. different numbers of flavons in the operator of Eq. (5), which generates the hierarchical fermion masses. We label the curves reproducing the observed charged lepton Yukawas in a way that is compatible with the FN charge assignments specified in Table I. The shaded region corresponds to temperatures where the efficiency windows of WLR\mathcal{W}_{LR} and TsphT_{\mathrm{sph}} overlap. The dashed lines depict the equilibration via effective four-dimensional interactions, which for nf>0n_f > 0 are only possible below the temperature at which the FN symmetry is broken (dotted vertical line), and represented by the dashed line becoming solid. There, the interaction including external flavons, which has the opposite temperature scaling, becomes subleading, as illustrated by the corresponding lines becoming transparent. Left: When the FNPT happens before sphalerons freeze out, equilibration becomes efficient simultaneously with sphaleron processes, leading to a washout of a previously stored asymmetry, unless nfn_f is large enough. Right: When the FNPT is delayed until after the sphalerons have frozen out, the equilibration of the chiral lepton asymmetry can no longer be communicated to the baryon sector. Thus, a BAU can remain until today, provided also that the initial asymmetry was injected sufficiently late, i.e. at temperatures after the solid lines have left the gray washout region; see Section III C for details.

Figure 4. Allowed values of the asymmetry injection temperature for different values of nfn_f and as a function of the EFT scale ΛFN\Lambda_{\mathrm{FN}}, for TFN<TsphT_{\mathrm{FN}} < T_{\mathrm{sph}}.

Crucially, if TFN>TEWT_{\mathrm{FN}} > T_{\mathrm{EW}} and ϵ=0.1\epsilon= 0.1, the left-right equilibration processes associated with fermions with nf<7n_f < 7, which would include all fermions present in the SM other than neutrinos, become efficient before sphalerons freeze out, occurring at Tsph≈TEWT_{\mathrm{sph}} \approx T_{\mathrm{EW}}. As a consequence, any previously generated asymmetry within these fermions is efficiently washed out, as represented in the left plot of Figure 3 by the corresponding lines entering the shaded region before TsphT_{\mathrm{sph}}. Thus, for a baryon asymmetry originating from a leptonic asymmetry within SM charged leptons to survive until today, the FN symmetry breaking has to be delayed until after the sphalerons have shut off. The right plot of Figure 3 depicts such a scenario.

This requirement together with constraints on the flavon VEV from flavor observables, which will be detailed in the next section, results in the necessity of a hierarchy of scales

vΦ≫TFN,(17)v_{\Phi} \gg T_{\mathrm{FN}}, \tag*{(17)}

that can be realized via a supercooled FN phase transition.

To sum up, these results show that for dimension-four Yukawas there is an upper bound on the Yukawa coupling in order to protect the chiral asymmetry from being washed out, which follows from requiring that

the efficiency window of WLR\mathcal{W}_{\mathrm{LR}} must not overlap with that of TsphT_{\mathrm{sph}}. On the other hand, when the Yukawa interactions remain higher-dimensional until after TsphT_{\mathrm{sph}}, the chiral asymmetry washout dynamics is significantly modified and a completely different scale setting of Dirac leptogenesis emerges. Most importantly, and as will be shown in Section III C, this alternative scenario allows other leptons within or beyond the SM to successfully mediate leptogenesis. In particular, SM charged leptons, namely electrons, muons, and tau leptons, can, under this framework, constitute viable asymmetry-sequestering reservoirs.

Allowed production scales

Even though we remain agnostic to the asymmetry production mechanism, the above results allow us to infer the range of temperatures within which such a mechanism would need to become efficient in order to successfully generate the BAU.

The conditions that must be satisfied by the asymmetry production scale TST_S, the equilibration scale TLRT_{\mathrm{LR}}, and the FN breaking scale TFNT_{\mathrm{FN}} in order to avoid washout before the sphaleron freeze-out scale TsphT_{\mathrm{sph}} can be summarized as

TS>Tsph>TLRif nf=0,TLR>TS>Tsph>TFNif 0<nf<nfc,TLR>TS>Tsphif nf≥nfc,(18)\begin{aligned} &T_S > T_{\mathrm{sph}} > T_{\mathrm{LR}} && \text{if } n_f = 0,\\ &T_{\mathrm{LR}} > T_S > T_{\mathrm{sph}} > T_{\mathrm{FN}} && \text{if } 0 < n_f < n_f^c,\\ &T_{\mathrm{LR}} > T_S > T_{\mathrm{sph}} && \text{if } n_f \geq n_f^c, \tag*{(18)} \end{aligned}

where the TLRT_{\mathrm{LR}} are defined in Eqs. (14) to (16) and nfc=7n_f^c = 7 for ϵ=0.1\epsilon= 0.1 and cf∼O(1)c_f \sim\mathcal{O}(1).

In Figure 4, the colored regions show the temperature scales TST_S at which the asymmetry production would need to take place in order to obtain a non-vanishing BAU today, as a function of the EFT scale ΛFN\Lambda_{\mathrm{FN}}. It is important to note that requiring that the source becomes active before the sphaleron freeze-out temperature corresponds to a necessary but not sufficient condition. Realistically, the process of asymmetry injection plus sphaleron conversion is not an instantaneous one, and therefore the source should become efficient early enough for the overall process to be completed before sphalerons shut off. The region spanned by the allowed values increases with nfn_f. The case nf=0n_f = 0 is not included given that the associated rate is independent of ΛFN\Lambda_{\mathrm{FN}}. Exponents satisfying nf≥nfcn_f \ge n_f^c are not included simply because the corresponding Yukawa couplings can trivially sequester the asymmetry even in their renormalizable form, as in the case of neutrinos, and the injection temperature can be nearly as large as the UV scale. The smaller nfn_f is, the higher ΛFN\Lambda_{\mathrm{FN}} has to be so that TLRT_{\mathrm{LR}} (and therefore TST_S) can be larger than TEWT_{\mathrm{EW}}, after which an injected asymmetry could no longer be transferred to the baryon sector via sphalerons. In Figure 3, this requires that the lines exit the shaded region before crossing TEWT_{\mathrm{EW}}. For instance, if ΛFN\Lambda_{\mathrm{FN}} were brought slightly closer to TEWT_{\mathrm{EW}}, this would no longer be the case for the nf=1n_f = 1 line.

For concreteness, consider an explicit FN realization of the SM by fixing ϵ=0.1\epsilon= 0.1 and cf∼O(1)c_f \sim O(1). Table I shows the values of nfn_f, which are determined by the sum of the charges of the chiral fields, i.e. nf≡nfL+nfˉRn_f \equiv n_{f_L} + n_{\bar f_R},8 that can reproduce the hierarchical SM Yukawas when ϵ∼0.1\epsilon\sim0.1. We note that when varying the O(1)O(1) numbers cfc_f, the nfn_f in Table I can differ slightly.

From Figure 4 it is then possible to read off a lower bound on ΛFN\Lambda_{\mathrm{FN}} for the SM leptons. Specifically, for the tau lepton (nf=2n_f = 2) there is a non-vanishing temperature range in which the asymmetry can be injected and remain successfully sequestered if ΛFN>105\Lambda_{\mathrm{FN}} > 10^5 GeV, while for the muon (nf=3n_f = 3) ΛFN>104\Lambda_{\mathrm{FN}} > 10^4 GeV is required, and for the electron (nf=6n_f = 6), we find that ΛFN\Lambda_{\mathrm{FN}} can be as low as O(103)O(10^3) GeV. Note, however, that nf=6n_f = 6 is still smaller than nfc=7n_f^c = 7, which means that if the FN breaking scale is larger than the sphaleron shut-off scale, the corresponding left-right equilibration rate will become efficient simultaneously with the sphalerons and any previously generated electron asymmetry will be washed out.

fftopcharmuptauelectron
nfn_f02526
ffbottomstrangedownmuonneutrino
nfn_f235312

Table 1. Exemplary values of nf=nfL+nfˉRn_f = n_{f_L} + n_{\bar f_R} for the different fermions in the SM, with ϵ=0.1\epsilon= 0.1.

Finally, for Dirac neutrinos one obtains nf=12>nfcn_f = 12 > n_f^c, and they can comfortably store an asymmetry regardless of whether or not the FN breaking happens before or after the EWPT. Although interesting, this is not surprising given that Dirac neutrinos are good asymmetry reservoirs even with fundamental Yukawas.

Experimental Froggatt-Nielsen Bounds

A universal prediction in Froggatt-Nielsen models are off-diagonal Yukawa couplings between the flavor and fermions, which introduce flavor violation into the model. Due to the non-observation of such processes, the FN scale, identified with the masses of the vector-like fermions mediating the Yukawa interactions, has to be much larger than the EW scale. When both the axial and radial modes are sufficiently heavy, the dominant constraint comes from integrating out the heavy messenger fermions, resulting in a universal bound on ΛFN\Lambda_{\mathrm{FN}}. However, when a light scalar mode exist, the loop diagram involving it can dominate and lead to more stringent constraints on ΛFN\Lambda_{\mathrm{FN}}.

The axial mode can obtain its mass from some explicit symmetry breaking, rendering it a pseudo-Nambu-Goldstone boson. If its potential is exclusively generated by a non-vanishing QCD anomaly of the global U(1)FNU(1)_{\mathrm{FN}} symmetry, it can solve the strong CP problem, connecting its resolution to that of the flavor puzzle [41, 42, 43, 44], and potentially explain the Dark Matter abundance in regions of the parameter space. However, for the sake of this analysis, we treat its mass as a free parameter, so it can be light or heavy. On the other hand, as will be described in detail in § 1.4, the mass of the radial mode mρm_\rho is bounded from above by requiring the delay of the FNPT until after EW sphalerons freeze out. This follows from a supercooled potential needing to be flat enough, i.e. the curvature at the VEV must be small. Therefore, we expect at least the radial mode to give an unavoidable contribution to the flavor-violating processes, and the constraint on ΛFN\Lambda_{\mathrm{FN}} to carry a mass dependence, in the region of interest.

In the following, we focus on the radial mode contribution and highlight the dominant constraints for an FN symmetry realized only in the leptonic sector, which corresponds to the minimum requirement for the proposed leptogenesis mechanism to work. We also comment on the additional constraints that must be included when generalizing the FN to include all fermions, which would be needed to explain all flavor hierarchies. We relegate the corresponding details to Appendix C.

Constraints from lepton flavor violation

There are strong bounds on vΦv_{\Phi} from rare lepton decays [45, 46, 47] when τ→μ+ρ\tau\to\mu+ \rho and μ→e+ρ\mu\to e + \rho are kinematically allowed, i.e. when mρ≤(mτ−mμ)=1.67m_{\rho} \le(m_{\tau} - m_{\mu}) = 1.67 GeV for the former case. We set this as a lower bound for the flavon mass and keep in mind that it might be possible to construct a model in which these constraints are alleviated to some extent. For flavons heavier than the SM leptons, the most stringent constraint comes from charged lepton flavor violation (CLFV), especially from the radiative decay μ→eγ\mu\to e\gamma, which is constrained by the MEG II experiment to be [48]

BR(μ→eγ)<1.5×10−13.(19)\mathrm{BR}(\mu\to e\gamma) < 1.5 \times10^{-13}. \tag*{(19)}

The branching ratio is given by [49]

Br(μ→eγ)=mμ34πΓμ(∣Ceγμe∣2+∣Ceγeμ∣2)(20)\mathrm{Br}(\mu\to e\gamma) = \frac{m_{\mu}^{3}}{4\pi\Gamma_{\mu}}\left(\left|C_{e\gamma}^{\mu e}\right|^{2}+\left|C_{e\gamma}^{e\mu}\right|^{2}\right) \tag*{(20)}

with

Ceγij=e32π2mρ2∑k=1,2,3(16(mljckie∗ckje+mlickieckje∗)+mlkckieckje∗(32+log⁡mlj2mρ2)).(21)C_{e\gamma}^{ij} = \frac{e}{32\pi^{2}m_{\rho}^{2}}\sum_{k=1,2,3}\left(\frac{1}{6}\left(m_{l_j}c_{ki}^{e*}c_{kj}^{e}+m_{l_i}c_{ki}^{e}c_{kj}^{e*}\right)+m_{l_k}c_{ki}^{e}c_{kj}^{e*}\left(\frac{3}{2}+\log\frac{m_{l_j}^{2}}{m_{\rho}^{2}}\right)\right). \tag*{(21)}

The couplings cijc_{ij} are derived in Appendix C. Lepton masses and the muon decay width are taken from Ref. [50].

Figure 5 shows the allowed parameter space of the model. There are two main constraints. Firstly, an upper bound (solid red line) on mρm_{\rho} and vΦv_{\Phi} that results from requiring that the FNPT does not reheat the Universe above TsphT_{\mathrm{sph}}, as it would re-activate the sphalerons while keeping the U(1)FNU(1)_{\mathrm{FN}} symmetry broken,9 which would lead to a washout of the previously generated asymmetry. The colored points result from a numerical scan, described in detail in § 1.4, of the input parameters of the model.

Figure 5. CLFV bounds from μ→eγ\mu\to e\gamma. For a flavon running in the loop, the naive constraint is shown in light orange, and the one with texture suppression in orange. The contribution from the VL messenger fermions constrains vΦ>104v_{\Phi} > 10^{4} GeV (grey lower region), which dominates for large mρm_{\rho}. Masses below 1.67 GeV are excluded due to rare lepton decays (left grey region). The origin of the scatter points and lines are explained in detail in § 1.4.

Secondly, a lower bound on the flavon VEV and mass from lepton flavor violation. Due to the rare lepton decays we consider mρ>1.7m_{\rho} > 1.7 GeV (transparent grey region), while for larger masses the dominant bound comes from the non-observation of μ→eγ\mu\to e\gamma (Eq. (19)), shown in light orange. Even though the constraints are given in terms of the mass of the radial mode, the ALP contribution can be obtained by replacing c→icc \to ic; see Eq. (98). In the case where the ALP is much lighter than the radial mode, it gives the leading constraints on vΦv_{\Phi}. If they have similar masses, their contributions to the branching ratio cancel to some extent. We describe in more detail how the bounds change depending on the ALP mass in Appendix C. Here, we treat the mass of the ALP as a free parameter, and thus focus only on the constraints coming from the radial mode, keeping in mind that in a specific model the constraints may become less or more stringent.

Flavor violation originates from off-diagonal components of the Yukawa matrix, however in our leptogenesis analysis we focus on the diagonal Yukawas, which leaves some freedom in choosing the individual U(1)FNU(1)_{\mathrm{FN}} charge assignment for the left- and right-handed fermion fields. This gives rise to different “textures” of the Yukawa matrices. As shown in Ref. [51], specific textures can decrease the bound on ΛFN\Lambda_{\mathrm{FN}} by half an order of magnitude. This possibility is shown in Figure 5 with a darker orange shading. We note that the precise bounds determined in Refs. [51, 52] cannot be directly applied to our case, since their work is based on SMEFT, while we allow for a light flavon. Therefore, we merely adopt the relative changes determined there to our numerical benchmark. We do, however, take their constraints as a lower bound on vΦv_{\Phi}, which becomes relevant for larger flavon masses, where the contribution from the integrated-out messenger fermions dominates. Taking into account the different normalizations, we set vΦmin⁡=104v_{\Phi_{\min}} = 10^{4} GeV, shown as the lower gray region in Figure 5.

FN in the leptonic and quark sectors

For charged leptons to be able to successfully sequester a chiral asymmetry following the prescription proposed in this work, it is necessary and sufficient to charge only these fermions under U(1)FNU(1)_{\mathrm{FN}}. Thus, we treat the constraints from μ→eγ\mu\to e\gamma as unavoidable. Ideally, however, quarks should also carry U(1)FNU(1)_{\mathrm{FN}} charges, so that all fermion hierarchies can be explained. In this case, additional bounds must be considered and, in particular, regions with ms<mK,mBm_s < m_K,m_B are very constrained due to invisible meson decays [49]. This then translates to the necessity of many orders of magnitude in supercooling, i.e. TFN≲TEW≪vΦT_{\mathrm{FN}} \lesssim T_{\mathrm{EW}} \ll v_{\Phi}, which is incompatible with not reheating the Universe to above TsphT_{\mathrm{sph}}; see § 1.4 for more details.

For larger masses instead, kaon mixing constraints can exclude additional parameter space compared to μ→eγ\mu\to e\gamma. We refer the reader to Appendix C for more details, and here simply emphasize that for favorable charge assignments and ms≳5 GeVm_s \gtrsim5\ \mathrm{GeV}, part of the viable parameter space of Figure 5 can remain viable even when charging all fermions under U(1)FNU(1)_{\mathrm{FN}}. Independently of whether all SM fermions are charged under the U(1)FNU(1)_{\mathrm{FN}}, larger radial mode masses allow for smaller flavon VEVs. This implies less required supercooling, and therefore populating high-mass regions is anyways expected to be more natural than populating the low-mass counterpart, which becomes excluded in the case of an FN symmetry realized within all fermions.

Testability

As can be seen in Figure 5, the viable parameter points live in a narrow band, bounded from below by flavor observables, and bounded from above by requiring the FNPT not to reheat the Universe above TsphT_{\mathrm{sph}}. However, as the branching ratio is proportional to 1/vΦ41/v_{\Phi}^{4}, to probe one additional order of magnitude in the mρ−vΦm_{\rho}-v_{\Phi} plane, the constraint has to be improved by four orders of magnitude. Specifically, to fully test the parameter space via μ→eγ\mu\to e\gamma, the sensitivity will have to be improved to be ∼6×10−17\sim6 \times10^{-17}. The current leading constraints come from the MEGII experiment with a targeted final sensitivity of 6×10−146 \times10^{-14} [53].10

As experimental precision improves, other lepton flavor violating observables will become relevant as well. The Mu3e experiment is designed to probe Br⁡(μ→3e)\operatorname{Br}(\mu\to3e) down to a sensitivity of 10−1610^{-16}, improving previously existing bounds by four orders of magnitude [54, 55]. The BelleII experiment plans to obtain a sensitivity of Br⁡(τ→3μ)=3.3×10−10\operatorname{Br}(\tau\to3\mu) = 3.3 \times10^{-10} [56], competitive with μ→3e\mu\to3e due to the larger tau-flavon couplings. Furthermore, in the presence of flavon-quark couplings, μ→e\mu\to e conversion in nuclei can become important. The Mu2e experiment [57] and the COMET experiment [58] will look for μ→e\mu\to e conversion in aluminum nuclei, improving the current limit by more than four orders of magnitude to ∼10−17\sim10^{-17}. Both plan to start data taking in 2027 [59, 60].

Thus, as constraints on flavor violation improve, the region in which the model remains viable will be probed progressively. However, the experiments currently under construction will not yet probe the full parameter space, only barely surpassing the current μ→eγ\mu\to e\gamma bound.

Phase transition dynamics

For successful sequestering of charged-lepton chiral asymmetries under the prescription described above, without conflicting with flavor observables, a hierarchy between the flavon VEV vΦv_{\Phi} and the FNSB temperature TFNT_{\mathrm{FN}} is required.11 In other words, the FNPT must be supercooled.

To understand the scalar field dynamics, we consider the simple potential

V(Φ)=V0+VT+VCW+VCT+VDaisy.(22)V(\Phi) = V_0 + V_T + V_{\mathrm{CW}} + V_{\mathrm{CT}} + V_{\mathrm{Daisy}} . \tag*{(22)}

V0V_0 the tree-level potential, for which we assume a negligible portal coupling between the flavon and the Higgs boson. VTV_T is the 1-loop finite-temperature potential, containing contributions from all fields with which the flavon is in thermal equilibrium. It is given by

VT=T42π2∑iniJB,F(mi2(Φ)T2),(23)V_T = \frac{T^4}{2\pi^2} \sum_i n_i J_{B,F}\left(\frac{m_i^2(\Phi)}{T^2}\right) , \tag*{(23)}

with the functions mi2m_i^2 denoting the (squared) field-dependent masses, given in Appendix B.1, and the functions JB/FJ_{B/F} for bosons (B) and fermions (F) given by

JB/F(y)=±∫0∞x2log⁡(1∓exp⁡(−x2+y)) dx,(24)J_{B/F}(y) = \pm\int_0^\infty x^2 \log\left(1 \mp\exp\left(-\sqrt{x^2+y}\right)\right) \, dx , \tag*{(24)}

where the lower signs correspond to fermions and the upper ones to bosons.

Due to the supercooled nature of the phase transition, the minimum in field space is much larger than the phase transition temperature, so the high temperature approximation does not accurately capture the relevant dynamics. Therefore, in the following, we include the full thermal functions, which can be well approximated by a sum of Bessel functions of the second kind as [62]

JB,F(y)=∓∑l=15(∓1)lyl2K2(ly).(25)J_{B,F}(y) = \mp\sum_{l=1}^{5} (\mp1)^l \frac{y}{l^2} K_2(\sqrt{ly}) . \tag*{(25)}

There is one intricacy in this FN set-up that requires an additional ingredient in the model. Since the field-dependent masses of the SM fermions are zero in the h=0h=0 direction, before EWSB they do not contribute to VTV_T in the flavon potential. In addition, the flavon self-interaction is bounded to be small due to the supercooling requirement, which can only be satisfied in flat potentials. Thus, to ensure that the flavon potential has a minimum at Φ=0\Phi=0 before TsphT_{\rm sph}, we consider the minimal extension of the field content by an additional Weyl fermion ξ\xi with U(1)FNU(1)_{\rm FN} charge satisfying the condition [ξc‾ξ]FN=1[\overline{\xi^c}\xi]_{\rm FN}=1, such that the interaction

L⊃−yξΦξc‾ξ,(26)\mathcal{L}\supset-y_{\xi}\Phi\overline{\xi^c}\xi, \tag*{(26)}

is added to the Lagrangian. With values of the coupling yξ∼O(1)y_{\xi}\sim\mathcal{O}(1), ξ\xi will give the dominant contribution to the thermal potential before EWSB.12 Compared to the FN fermions appearing in a UV completion, which have large masses of O(ΛFN)\mathcal{O}(\Lambda_{\rm FN}), the fermion ξ\xi will acquire a mass mξ∼yξvΦ<ΛFNm_{\xi}\sim y_{\xi}v_{\Phi}<\Lambda_{\rm FN}. The one-loop Coleman-Weinberg (CW) potential in the Landau gauge and MS‾\overline{\rm MS}-scheme is given by

VCW=∑inimi464π2(ln⁡(mi2Q2)−Ci),(27)V_{\rm CW}=\sum_i n_i\frac{m_i^4}{64\pi^2}\left(\ln\left(\frac{m_i^2}{Q^2}\right)-C_i\right), \tag*{(27)}

where the sum includes all particles in the thermal bath. The renormalisation scale QQ is chosen to be Q=0.1×ΛFN≃vΦQ=0.1\times\Lambda_{\rm FN}\simeq v_{\Phi} and Cf=CF=CΦ=3/2C_f=C_F=C_{\Phi}=3/2. The degrees of freedom are nq=−12n_q=-12, nl=−4n_l=-4, nF=−2n_F=-2 and nΦ=2n_{\Phi}=2. Since the CW potential can shift the minima, we include a counterterm potential

VCT=δmΦ2+δλΦ4,(28)V_{\rm CT}=\delta m\Phi^2+\delta\lambda\Phi^4, \tag*{(28)}

with the coefficients δm\delta m and δλ\delta\lambda determined from the conditions

∂VCW∂Φ=−∂VCT∂Φ,∂2VCW∂Φ2=−∂2VCT∂Φ2.(29)\frac{\partial V_{\rm CW}}{\partial\Phi}=-\frac{\partial V_{\rm CT}}{\partial\Phi},\qquad \frac{\partial^2 V_{\rm CW}}{\partial\Phi^2}=-\frac{\partial^2 V_{\rm CT}}{\partial\Phi^2}. \tag*{(29)}

This ensures that the values of the VEV and masses of the tree-level potential, evaluated at Φ=vΦ\Phi=v_{\Phi} and h=vh=v, are preserved after the CW contributions are accounted for. Lastly, we include the thermal corrections to the scalar potential from the resummation of Daisy diagrams as [63, 64]

VDaisy=−T12πnΦ(MΦ3−mΦ3),(30)V_{\rm Daisy}=-\frac{T}{12\pi}n_{\Phi}\left(M_{\Phi}^3-m_{\Phi}^3\right), \tag*{(30)}

where MΦ2=mΦ2+Π(T)M_{\Phi}^2=m_{\Phi}^2+\Pi(T) is the thermal mass of the scalar field; see Appendix B 1 for details.

Analytical analysis of scalar potential

As the UV model includes heavy fields which are integrated out to obtain the fermion mass operators, we also consider higher-order operators in the scalar potential. For simplicity, we truncate the potential at dimension six

V0(Φ)=−μ2Φ2−λΦ4+4c3ΛFN2Φ6,(31)V_0(\Phi)=-\mu^2\Phi^2-\lambda\Phi^4+\frac{4c}{3\Lambda_{\rm FN}^2}\Phi^6, \tag*{(31)}

but for example dimension-eight operators, which would be further suppressed, could also be straightforwardly added. The signs are chosen for convenience. Since the ALP plays a subdominant role in the present leptogenesis framework, we do not reduce the analysis by assuming a specific term. However, the explicit symmetry breaking (exSB) can also give a contribution to the mass term of the radial mode, imagine e.g. VexSB∼μexSB2(ΦΦ+Φ†Φ†)V_{\rm exSB}\sim\mu_{\rm exSB}^2(\Phi\Phi+\Phi^\dagger\Phi^\dagger). From now on, we assume that the contribution from the exSB is implicitly included into the quadratic term of the radial mode. For the analytical estimate, we neglect μ2\mu^2, which is required to be small in magnitude so that thermal corrections can lead to a positive curvature at the origin, enabling a thermal barrier to survive until TsphT_{\rm sph}. The flavon VEV is given by

vΦ=λcΛFN,(32)v_{\Phi}=\sqrt{\frac{\lambda}{c}}\Lambda_{\rm FN}, \tag*{(32)}

from which we can derive the condition λ∼10−2c\lambda\sim10^{-2}c in order to obtain ϵ∼0.1\epsilon\sim0.1. The radial mode after FNSB has a resulting mass of

mρ=2λ2cΛFN∼c(10−2ΛFN).(33)m_{\rho}=\sqrt{\frac{2\lambda^2}{c}}\Lambda_{\rm FN}\sim\sqrt{c}\left(10^{-2}\Lambda_{\rm FN}\right). \tag*{(33)}

The advantage of considering additional operators13 is the “freeing up” of μ2\mu^2, which here only determines the phase transition temperature, while the flavon’s mass and VEV depend primarily on λ\lambda and cc. Thermal corrections will lead to a positive effective mass parameter μeff2>0\mu_{\rm eff}^2>0, thus introducing a thermal barrier into the scalar potential when ∣μ2∣<yξ2T2|\mu^2|<y_{\xi}^2T^2.

Supercooling

As we require TFN≪vΦT_{\rm FN}\ll v_{\Phi}, the universe might enter a vacuum-dominated era, called thermal inflation (TI) [65, 66]. Since ρrad∼a−4\rho_{\mathrm{rad}} \sim a^{-4}, for sufficiently long supercooling the vacuum energy ΔV∼const.\Delta V \sim\mathrm{const.} will necessarily start dominating. Then, the duration of the TI can be expressed in terms of e-folds NN as

TFNTTI=e−N,(34)\frac{T_{\mathrm{FN}}}{T_{\mathrm{TI}}} = e^{-N}, \tag*{(34)}

where TTIT_{\mathrm{TI}} is the temperature at which the vacuum energy starts dominating over the radiation energy density. Once the PT completes at TFNT_{\mathrm{FN}}, the stored energy will be released into the plasma14 and the reheating temperature can be estimated as

∣ΔV∣∼ρrh=π230g∗Trh4⇒Trh∼(30∣ΔV∣π2g∗)1/4(35)\lvert\Delta V\rvert\sim\rho_{\mathrm{rh}} = \frac{\pi^2}{30}g_*T_{\mathrm{rh}}^4 \Rightarrow T_{\mathrm{rh}} \sim\left(\frac{30\lvert\Delta V\rvert}{\pi^2g_*}\right)^{1/4} \tag*{(35)}

To successfully sequester the asymmetry, sphaleron processes should remain inactive once the PT completes, and therefore the universe should not reheat above the EW temperature.15 To avoid such a problematically high reheating temperature, ΔV\Delta V has to be sufficiently small, which in turn decreases the tunneling probability. At

the same time, if the potential was too shallow, the flavon would be problematically light with respect to lepton flavor violation observables, specifically rare lepton decays. As a consequence the second minimum necessarily becomes global already at T≫TsphT \gg T_{\mathrm{sph}}, and the critical temperature is much larger than the final FNPT temperature TFNT_{\mathrm{FN}}. Figure 6 shows requirements in terms of the effective potential, highlighting the required magnitudes in VEV and ΔV\Delta V.

Effective potential requirements

Figure 6. Illustration of the requirements the effective potential in Eq. (22) has to realize. Above the sphaleron temperature, a small barrier remains (orange), keeping the field trapped at the origin. The barrier disappears at some TFN≲TsphT_{\mathrm{FN}} \lesssim T_{\mathrm{sph}} (blue), allowing the field to obtain a VEV. Although the VEV is much larger than EW scales, the potential has to be sufficiently shallow such that the Universe is not reheated to temperatures above sphaleron freeze-out once the phase transition completes.

For an analytical estimate, we can combine Eq. (35), where the vacuum energy is given by ΔV=V(vφ)−V(0)∼V(vφ)=−124mφ2vφ2\Delta V = V(v_\varphi)-V(0) \sim V(v_\varphi) = -\frac{1}{24}m_\varphi^2v_\varphi^2, with the requirement Trh<TEWT_{\mathrm{rh}} < T_{\mathrm{EW}}, to obtain an upper bound on the mass of the radial mode, namely

mφ2≲4π2g∗TEW45vφ2∼(50 GeV)2(g∗100)(104 GeVvφ)2.(36)\begin{aligned} m_\varphi^2 \lesssim\frac{4\pi^2g_*T_{\mathrm{EW}}^4}{5v_\varphi^2} \\ &\sim(50\,\mathrm{GeV})^2\left(\frac{g_*}{100}\right)\left(\frac{10^4\,\mathrm{GeV}}{v_\varphi}\right)^2 . \tag*{(36)} \end{aligned}

Of course, the mere existence of a barrier does not guarantee delayed FNSB, so as a next step we will calculate tunneling probabilities, using the full thermal potential.

Tunneling probability

Since we require U(1)FNU(1)_{\mathrm{FN}} to be broken at the very latest before BBN, where we already expect to have massive fermions, we have to ensure that tunneling or quantum fluctuations are efficient enough to bring the field to the global minimum, or that the barrier disappears before then. The thermal tunneling rate is given by [67, 68, 69]

Γ=T4(S32πT)3/2e−S3/T,(37)\Gamma= T^4\left(\frac{S_3}{2\pi T}\right)^{3/2}e^{-S_3/T}, \tag*{(37)}

and once this becomes comparable to the Hubble rate, bubbles start nucleating. The nucleation temperature is defined by the moment when one bubble per Hubble volume has formed, namely

∫tctndt Γ(t)H(t)3=1,(38)\int_{t_c}^{t_n} dt\,\frac{\Gamma(t)}{H(t)^3}=1, \tag*{(38)}

where the integral spans from the critical time tct_c, i.e. when the minima are equally favorable, to the nucleation time tnt_n.

To calculate the action, we follow the tunneling potential formalism from Ref. [70], and in the relevant parameter space the action becomes small, i.e. nucleation becomes favorable, just before the barrier disappears completely. In that sense, our setup differs from what is often done in supercooled phase transitions; see e.g. Ref. [71]. The potential is not (quasi)conformal, but contains an explicit negative μ2\mu^2 term which will always start dominating over thermal corrections, thus ensuring that the field cannot get stuck in the unbroken phase. This makes the

ParameterRangeParameterValue
μ2\mu^2[10−8,10] GeV2[10^{-8}, 10]\ \mathrm{GeV}^2ϵ\epsilon0.10.1
ΛFN\Lambda_{\mathrm{FN}}[104,107] GeV[10^4, 10^7]\ \mathrm{GeV}λ\lambda−μ2/(2ϵΛFN)2+2cϵ2-\mu^2/(\sqrt{2}\epsilon\Lambda_{\mathrm{FN}})^2+2c\epsilon^2
cc[10−15,1][10^{-15}, 1]nfn_fTable I
yξy_{\xi}[10−8,1][10^{-8}, 1]cfc_fyfSM/ϵnfy_f^{\mathrm{SM}}/\epsilon^{n_f}

caption: TABLE II. Inputs for the flavon phase-transition scan. Parameters on the left are sampled uniformly in log⁡10\log_{10} of their values. The right block gives fixed and derived parameters.

:::

competition between expanding bubbles and the expanding Universe due to the vacuum domination less crucial. If some bubbles form, there is very little time before the barrier disappears and the phases start mixing. Thus, we consider the temperature at which the barrier disappears to be the temperature at which the phase transition occurs, TFNT_{\mathrm{FN}}, and treat it as a second-order phase transition. Since TFNT_{\mathrm{FN}} is much smaller than the critical temperature, the phase transition is supercooled.

A more detailed analysis of the model, particularly keeping possible gravitational wave signals in mind, goes beyond the scope of this paper.16 In addition, since the here-presented model contains an asymmetry in the charged leptons, primordial magnetic fields may be sourced [73]. Both of these questions are left for future work.

Numerical scan

In order to present numerical results, we scan over the input parameters μ2\mu^2, λ\lambda, cc, ΛFN\Lambda_{\mathrm{FN}}, cfc_f, yξy_{\xi} in the ranges specified in Table II and filter for a phase transition occurring between TsphT_{\mathrm{sph}} and BBN. Here, we continue using the simplifying assumption of diagonal Yukawas, as for the leptogenesis analysis. Working with the full textures is essential only when determining the bounds from flavor violation. The values of cfc_f are then obtained in terms of yfSMy_f^{\mathrm{SM}}, the SM Yukawas, such that the generated fermion masses match observations.

Figure 5 shows the successful points in the mρm_{\rho}-vΦv_{\Phi} plane together with the constraints from flavor violation. The numerically determined reheating temperature is in good agreement with the analytical estimate from requiring a viable reheating temperature per Eq. (36). The red (green) scatter points satisfy the condition Trh<TsphT_{\mathrm{rh}}<T_{\mathrm{sph}} (Trh<Tsph/2T_{\mathrm{rh}}<T_{\mathrm{sph}}/2), and the solid lines are obtained from the analytical estimate.

Dilution

When there is reheating of the thermal bath, the baryon yield will be diluted, independently of sphaleron processes. The dilution will be estimated in the following.

The baryon asymmetry yield can be expressed as

YB=nB−nBˉs,(39)Y_B = \frac{n_B-n_{\bar B}}{s}, \tag*{(39)}

where s∼g⋆T3s\sim g_\star T^3 is the entropy density. By reheating the bath from TFNT_{\mathrm{FN}} to TrhT_{\mathrm{rh}}, the yield will be diluted according to

YBinYBfin=sfinsin∼(TrhTFN)3.(40)\frac{Y_B^{\mathrm{in}}}{Y_B^{\mathrm{fin}}}=\frac{s_{\mathrm{fin}}}{s_{\mathrm{in}}}\sim\left(\frac{T_{\mathrm{rh}}}{T_{\mathrm{FN}}}\right)^3. \tag*{(40)}

As ΔV∼ρrh\Delta V\sim\rho_{\mathrm{rh}}, see Eq. (35), the reheating temperature can be determined directly from the scalar potential. Figure 7 shows the dilution factor (Eq. (40)) for different reheating temperatures, as a function of TFNT_{\mathrm{FN}}, after which the PT completes and thermal inflation ends. We note it is possible that the value of the initially injected asymmetry may have been higher than expected, and diluted by the additional entropy to its nowadays value. We set the maximal initially generated asymmetry yield to order one, in which case the absolute upper bound on the dilution factor is of O(1010)\mathcal{O}(10^{10}), to still be able to match the observed BAU [1, 2, 4].

Figure 7. Dilution factor as a function of the temperature at which the phase transition completes, i.e. when thermal inflation ends. The colored lines correspond to different reheating temperatures, and a dilution >1010>10^{10} is phenomenologically excluded since even an initial O(1)\mathcal{O}(1) asymmetry would not be able to match observations. The dashed lines indicate how many of the points in the numerical scan lie below a certain dilution.

ηB=nB−nBˉnγ∼6×10−10,(41)\eta_B=\frac{n_B-n_{\bar B}}{n_\gamma}\sim6\times10^{-10}, \tag*{(41)}

which can be equivalently written in terms of the baryon asymmetry yield YB=(nB−nˉB)/s∼9×10−11Y_B = (n_B - \bar{n}_B)/s \sim9 \times10^{-11}.

In the numerical scan, the reheating for all but 0.2% of the points led to a dilution factor ≤102\le10^2.17 In this case, an initial asymmetry is required of at most two orders of magnitude larger than expected without additional entropy injection. For consistency reasons, we require the FNSB to happen before BBN, which places a lower bound on TFNT_{\mathrm{FN}}. In general, for large reheating close to the EWPT, the FNPT has to complete before TFN≳0.1T_{\mathrm{FN}} \gtrsim0.1 GeV, while for smaller reheating temperatures there is no problematically high dilution.

potential while also accounting for transition completion, reheating, and the resulting dilution of the primordial asymmetry. Combining these requirements, a light flavon is expected with a mass in the range 1.7−501.7 - 50 GeV, and where the allowed VEV is constrained from both above and below, leading to precise predictions which could be progressively testable via future lepton flavor violation experiments. Finally, a more detailed treatment of the phase transition dynamics could furthermore establish whether bubble formation occurs, results in observable gravitational-wave signatures and whether primordial magnetic fields are sourced.

Conclusion and Outlook

We have shown that Dirac leptogenesis, which does not rely on lepton number violation to source an initial lepton asymmetry, can proceed via all charged leptons, and not only Dirac neutrinos with tiny Yukawa couplings as originally proposed in Ref. [12].

In fact, the desired dynamics can arise in frameworks explaining the fermion mass hierarchies from a UV origin, such as a Froggatt-Nielsen mechanism. There, the low-energy Yukawa couplings, which normally lead to a washout of the asymmetry for the charged leptons before sphaleron freeze-out, are only induced once the scalar field associated with the U(1)FNU(1)_{\mathrm{FN}} symmetry breaking obtains a VEV. Before, the left-right equilibration proceeds via higher-dimensional operators which exit equilibrium as the Universe cools down since they become thermally suppressed with (T/ΛFN)2nf(T/\Lambda_{\mathrm{FN}})^{2n_f} as the temperature drops below the eventual flavon VEV. Thus, the supercooling of the FN phase transition until sphaleron freeze-out opens a temperature range in which a chiral lepton asymmetry can be sourced, successfully remain sequestered by the leptons, and converted to a baryon asymmetry by EW sphalerons.

To assess the viability of this framework, we estimated the thermally averaged chirality-equilibration rates and identified the allowed asymmetry-production temperatures. Explicitly, with the presented benchmark, successful sequestering requires ΛFN>105\Lambda_{\mathrm{FN}} > 10^5 GeV for the tau lepton, ΛFN>104\Lambda_{\mathrm{FN}} > 10^4 GeV for the muon and ΛFN>103\Lambda_{\mathrm{FN}} > 10^3 GeV for the electron for the mechanism to work. However, there are additional, more stringent lower bounds on ΛFN\Lambda_{\mathrm{FN}}, since FN constructions intrinsically lead to flavor violation. Since the described setup requires that the FNSB has to be delayed until sphalerons have shut off, this leads to a hierarchy vΦ≫TFNv_\Phi\gg T_{\mathrm{FN}}.

We studied the dominant lepton flavor violation process μ→eγ\mu\to e\gamma as well as the finite-temperature flavon potential while also accounting for transition completion, reheating, and the resulting dilution of the primordial asymmetry. Combining these requirements, a light flavon is expected with a mass in the range 1.7−501.7 - 50 GeV, and where the allowed VEV is constrained from both above and below, leading to precise predictions which could be progressively testable via future lepton flavor violation experiments. Finally, a more detailed treatment of the phase transition dynamics could furthermore establish whether bubble formation occurs, results in observable gravitational-wave signatures and whether primordial magnetic fields are sourced.

Acknowledgments

We are grateful to Admir Greljo, Yann Gouttenoire, Joerg Jaeckel, Mark Hindmarsh, Manfred Lindner, Maura E. Ramirez-Quezada, Aleks Smolković, Carlo Tasillo and Jorinde van de Vis for useful discussions. J. P. G. acknowledges funding from the International Max Planck Research School for Precision Tests of Fundamental Symmetries (IMPRS-PTFS). MH and FG would like to express special thanks to the Mainz Institute for Theoretical Physics (MITP) of the Cluster of Excellence PRISMA+ (Project ID 390831469) for its hospitality and support during the completion of this work.

Left-right equilibration

Thermally averaged rates: fully unbroken phase

We aim to estimate the contributions of the 2→nf+12 \to n_f + 1 processes in the top row of Figure 2 to the thermally averaged rate of left-right equilibration processes

⟨ΓLRf⟩(T)=γLRf(T)nfeq(T)(42)\langle\Gamma_{\mathrm{LR}}^f \rangle(T) = \frac{\gamma_{\mathrm{LR}}^f(T)}{n_f^{\mathrm{eq}}(T)} \tag*{(42)}

where nfeq(T)n_f^{\mathrm{eq}}(T) is the equilibrium number density of fermion ff at temperature TT and

γLRf(T)=g1g2T32π4∫smin⁡∞dss σ^LRf(s)K1(sT)(43)\gamma_{\mathrm{LR}}^f(T) = \frac{g_1 g_2 T}{32\pi^4}\int_{s_{\min}}^\infty\mathrm{d}s \sqrt{s}\,\hat{\sigma}_{\mathrm{LR}}^f(s) K_1\left(\frac{\sqrt{s}}{T}\right) \tag*{(43)}

assuming Maxwell-Boltzmann (MB) statistics. K1K_1 denotes the modified Bessel function of the first kind, gig_i (i=1,2i=1, 2) denotes the internal degrees of freedom of the initial-state particles, and

smin⁡=max⁡{(m1+m2)2,(∑i=3nf+3mi)2}(44)s_{\min} = \max\left\{(m_1 + m_2)^2,\left(\sum_{i=3}^{n_f+3} m_i\right)^2\right\} \tag*{(44)}

with mim_i denoting masses of initial state particles for i=1,2i = 1, 2 and of final state particles for i≥3i \ge3. The reduced cross-section is defined as

σ^LRf(s)=λ(s,m12,m22)sσLRf(s),(45)\hat{\sigma}^{f}_{\mathrm{LR}}(s)=\frac{\lambda(s,m_1^2,m_2^2)}{s}\sigma^{f}_{\mathrm{LR}}(s), \tag*{(45)}

with

λ(s,m12,m22)=s2+m14+m24−2sm12−2sm22−2m12m22,(46)\lambda(s,m_1^2,m_2^2)=s^2+m_1^4+m_2^4-2sm_1^2-2sm_2^2-2m_1^2m_2^2, \tag*{(46)}

and the cross section σ\sigma is obtained from the phase space integral of the squared matrix element associated with the left-right equilibration processes, namely

σLRf(s)=12s∫dΦ∣MLRf∣2.(47)\sigma^{f}_{\mathrm{LR}}(s)=\frac{1}{2s}\int d\Phi\left|\mathcal{M}^{f}_{\mathrm{LR}}\right|^2. \tag*{(47)}
Left-right annihilation

For the annihilation process fL(p1)+fˉR(p2)→H(k0)+∑i=1nfΦ(ki)f_L(p_1)+\bar{f}_R(p_2)\to H(k_0)+\sum_{i=1}^{n_f}\Phi(k_i) we have

∣MLRf∣2=(nf!)2∣cf∣2ΛFN2nf(s−2mf2),(48)\left|\mathcal{M}^{f}_{\mathrm{LR}}\right|^2=\frac{(n_f!)^2|c_f|^2}{\Lambda_{\mathrm{FN}}^{2n_f}}(s-2m_f^2), \tag*{(48)}

which can be factored out of the phase space integral, namely

σLR,sf(s)=12s1nf!∣MLR,sf∣2Φnf+1,(49)\sigma^{f}_{\mathrm{LR},s}(s)=\frac{1}{2s}\frac{1}{n_f!}\left|\mathcal{M}^{f}_{\mathrm{LR},s}\right|^2\Phi_{n_f+1}, \tag*{(49)}

where we have included the symmetry factor due to identical final-state particles, and defined

Φnf+1=∫(2π)4δ(4)(pin−kf)∏i=0nfd3ki(2π)32Ei,(50)\Phi_{n_f+1}=\int(2\pi)^4\delta^{(4)}(p_{\mathrm{in}}-k_f)\prod_{i=0}^{n_f}\frac{d^3k_i}{(2\pi)^3 2E_i}, \tag*{(50)}

with pin=p1+p2p_{\mathrm{in}}=p_1+p_2, the total initial momentum, and kf=∑i=0nfkik_f=\sum_{i=0}^{n_f}k_i, the total final momentum. In the limit s≫{mi}s\gg\{m_i\} the integral of Eq. (50) can be solved analytically. We start by performing the factorization

dΦnf+1(s;k0,…,knf)=dq22πdΦ2(s;q,k0)×dΦnf(q;k1,…,knf),(51)d\Phi_{n_f+1}(s;k_0,\ldots,k_{n_f})=\frac{dq^2}{2\pi}d\Phi_2(s;q,k_0)\times d\Phi_{n_f}(q;k_1,\ldots,k_{n_f}), \tag*{(51)}

with qμ=∑i=1nfkiμq^\mu=\sum_{i=1}^{n_f}k_i^\mu. Integrating over {ki}\{k_i\} one obtains

Φnf+1(s)=12π∫dq2Φ2(s;q2,0)Φnf(q2),(52)\Phi_{n_f+1}(s)=\frac{1}{2\pi}\int dq^2\Phi_2(s;q^2,0)\Phi_{n_f}(q^2), \tag*{(52)}

where the two-body phase space reads

Φ2(s;q2,0)=λ(s;q2,0)8πs=s−q28πs.(53)\Phi_2(s;q^2,0)=\frac{\sqrt{\lambda(s;q^2,0)}}{8\pi s}=\frac{s-q^2}{8\pi s}. \tag*{(53)}

Next, we take the inductive ansatz

Φnf(q2)=(q2)nf−22(4π)2nf−3Γ(nf)Γ(nf−1),(54)\Phi_{n_f}(q^2)=\frac{(q^2)^{n_f-2}}{2(4\pi)^{2n_f-3}\Gamma(n_f)\Gamma(n_f-1)}, \tag*{(54)}

with Γ(n)=(n−1)!\Gamma(n)=(n-1)!, and for which nf=2n_f=2 gives the correct value Φ2(s)=1/8π\Phi_2(s)=1/8\pi. Substituting Eqs. (53) and (xref)(A13) into Eq. (52) then gives

Φnf+1(s)=12π18πsI1(s)12(4π)2nf−3Γ(nf)Γ(nf−1),(55)\Phi_{n_f+1}(s)=\frac{1}{2\pi}\frac{1}{8\pi s}I_1(s)\frac{1}{2(4\pi)^{2n_f-3}\Gamma(n_f)\Gamma(n_f-1)}, \tag*{(55)}

with

I1(s)=∫0sdq2(s−q2)(q2)nf−2=snf(nf−1)nf.(56)I_1(s)=\int_0^s dq^2(s-q^2)(q^2)^{n_f-2}=\frac{s^{n_f}}{(n_f-1)n_f}. \tag*{(56)}

Thus, we obtain

Φnf+1(s)=s(nf+1)−22(4π)2(nf+1)−3Γ(nf+1)Γ((nf+1)−1),\Phi_{n_f+1}(s)=\frac{s^{(n_f+1)-2}}{2(4\pi)^{2(n_f+1)-3}\Gamma(n_f+1)\Gamma((n_f+1)-1)},

which confirms the inductive hypothesis. Substituting this result in Eq. (49), then in Eq. (45) and finally in Eq. (43) gives

γLR,sf(T)=∣cf∣28(4π)2nf+1Γ(nf)TΛFN2nfI2(T),(57)\gamma^{f}_{\mathrm{LR},s}(T)=\frac{|c_f|^2}{8(4\pi)^{2n_f+1}\Gamma(n_f)}\frac{T}{\Lambda_{\mathrm{FN}}^{2n_f}}I_2(T), \tag*{(57)}

with

I2(T)=∫0∞ds snf+1/2K1(sT).(58)I_2(T)=\int_0^\infty ds\,s^{n_f+1/2}K_1\left(\frac{\sqrt{s}}{T}\right). \tag*{(58)}

With the change of variables x=s/Tx=\sqrt{s}/T the integral can be rewritten as

I2(T)=2T2nf+3∫0∞dx x2nf+2K1(x),(59)I_2(T)=2T^{2n_f+3}\int_0^\infty dx\,x^{2n_f+2}K_1(x), \tag*{(59)}

and using the identity

∫0∞dx xμKν(x)=2μ−1Γ(1+μ+ν2)Γ(1+μ−ν2)(60)\int_0^\infty dx\,x^\mu K_\nu(x)=2^{\mu-1}\Gamma\left(\frac{1+\mu+\nu}{2}\right)\Gamma\left(\frac{1+\mu-\nu}{2}\right) \tag*{(60)}

for μ=2nf+2\mu=2n_f+2 and ν=1\nu=1, we obtain

∫0∞dx x2n+2K1(x)=22nf+1Γ(nf+2)Γ(nf+1).(61)\int_0^\infty dx\,x^{2n+2}K_1(x)=2^{2n_f+1}\Gamma(n_f+2)\Gamma(n_f+1). \tag*{(61)}

Substituting these results in Eq. (57) and then on Eq. (42), and using the MB equilibrium number density

nfeq=T3π2,(62)n_f^{\mathrm{eq}}=\frac{T^3}{\pi^2}, \tag*{(62)}

gives the final result for the thermally-averaged tree-level left-right annihilation rate for fermion ff in the limit s≫{mi}s\gg\{m_i\}, namely

⟨ΓLR,sf⟩(T)=κfd>4∣cf∣2T2nf+1ΛFN2nf,(63)\left\langle\Gamma^{f}_{\mathrm{LR},s}\right\rangle(T)=\kappa_f^{d>4}|c_f|^2\frac{T^{2n_f+1}}{\Lambda_{\mathrm{FN}}^{2n_f}}, \tag*{(63)}

with

κfd>4=nf(nf+1)!4(2π)2nf+1.(64)\kappa_f^{d>4}=\frac{n_f(n_f+1)!}{4(2\pi)^{2n_f+1}}. \tag*{(64)}
Left-right scattering

For the scattering process fL(p1)+H(p2)→fR(k0)+∑i=1nfΦ(ki)f_L(p_1) + H(p_2) \to f_R(k_0) + \sum_{i=1}^{n_f} \Phi(k_i) we have

∣MLR,tf∣2=2(nf!)2∣cf∣2ΛFN2nf(p1⋅k0),(65)\left|\mathcal{M}_{\mathrm{LR},t}^{f}\right|^2 = 2\frac{(n_f!)^2|c_f|^2}{\Lambda_{\mathrm{FN}}^{2n_f}}(p_1 \cdot k_0), \tag*{(65)}

which gives the cross-section

σLR,tf(s)=nf!∣cf∣2sΛFN2nfpμ∫dΦnf+1k0μ.(66)\sigma_{\mathrm{LR},t}^{f}(s) = \frac{n_f!|c_f|^2}{s\Lambda_{\mathrm{FN}}^{2n_f}}p_\mu\int d\Phi_{n_f+1}k_0^\mu. \tag*{(66)}

In order to evaluate the phase-space vector integral, we note that, due to Lorentz covariance, it must be proportional to Pμ≡p1μ+p2μP^\mu\equiv p_1^\mu+ p_2^\mu, namely

∫dΦnf+1k0μ=A(s)Pμ.(67)\int d\Phi_{n_f+1}k_0^\mu= A(s)P^\mu. \tag*{(67)}

Contracting both sides with PμP_\mu gives

∫dΦnf+1(P⋅k0)=A(s)P2=A(s)s,(68)\int d\Phi_{n_f+1}(P \cdot k_0) = A(s)P^2 = A(s)s, \tag*{(68)}

and in the limit s≫{mi}s \gg\{m_i\}, the final-state particles can all be treated on equal footing, allowing us to write

∫dΦnf+1(P⋅ki)=1nf+1∫dΦnf+1(P⋅∑i=0nfki)=snf+1Φnf+1(s),(69)\begin{aligned} \int d\Phi_{n_f+1}(P \cdot k_i) &= \frac{1}{n_f+1}\int d\Phi_{n_f+1}\left(P \cdot\sum_{i=0}^{n_f} k_i\right) \\ &= \frac{s}{n_f+1}\Phi_{n_f+1}(s), \tag*{(69)} \end{aligned}

where in the last equality we have used P⋅∑i=0nfki=P2=sP \cdot\sum_{i=0}^{n_f} k_i = P^2 = s. Thus, the unknown function A(s)A(s) reads

A(s)=1nf+1Φnf+1(s),(70)A(s) = \frac{1}{n_f+1}\Phi_{n_f+1}(s), \tag*{(70)}

and the cross-section is given by

σLR,tf(s)=nf!∣cf∣22(nf+1)Φnf+1(s)ΛFN2nf,(71)\sigma_{\mathrm{LR},t}^{f}(s) = \frac{n_f!|c_f|^2}{2(n_f+1)}\frac{\Phi_{n_f+1}(s)}{\Lambda_{\mathrm{FN}}^{2n_f}}, \tag*{(71)}

where we have used p1⋅P≃s/2p_1 \cdot P \simeq s/2.

Finally, the thermally-averaged tree-level left-right scattering rate for fermion ff in the limit s≫{mi}s \gg\{m_i\} can be written in terms of its annihilation counterpart Eq. (63) as

⟨ΓLR,tf⟩(T)=1nf+1⟨ΓLR,sf⟩(T).(72)\left\langle\Gamma_{\mathrm{LR},t}^{f}\right\rangle(T) = \frac{1}{n_f+1}\left\langle\Gamma_{\mathrm{LR},s}^{f}\right\rangle(T). \tag*{(72)}
Thermally averaged rates: U(1)FN in broken phase

Below TFNT_{\mathrm{FN}}, the flavon obtains a non-vanishing VEV, and the effective square vertices of Figure 2 must be included in the computation of the left-right equilibration rate. These can mediate ss- and tt-channel annihilation processes such as fL+fˉR→Hf_L+\bar{f}_R\to H, fL+fˉR→t+tˉf_L+\bar{f}_R\to t+\bar{t}, and fL+fˉR→H+Vμf_L+\bar{f}_R\to H+V_\mu, where VμV_\mu stands for a vector boson, as well as their scattering counterparts, i.e., fL+t→fR+tˉf_L+t\to f_R+\bar{t}, and fL+H→fR+Vμf_L+H\to f_R+V_\mu.

Consider the process fL+fˉR→Hf_L+\bar{f}_R\to H. By detailed balance, we have

γ(fL+fˉR→H)=γ(H→fL+fˉR)=nHeqΓ(H→fL+fˉR)K1(mH/T)K2(mH/T).(73)\begin{aligned} \gamma(f_L+\bar{f}_R\to H)=\gamma(H\to f_L+\bar{f}_R) \\ &=\frac{n_H^{\mathrm{eq}}\Gamma(H\to f_L+\bar{f}_R)K_1(m_H/T)}{K_2(m_H/T)}. \tag*{(73)} \end{aligned}

where

nHeq=gHmH2T2π2K2(mHT),(74)n_H^{\mathrm{eq}}=g_H\frac{m_H^2T}{2\pi^2}K_2\left(\frac{m_H}{T}\right), \tag*{(74)}

and

Γ(H→fL+fˉR)=∣yefff∣216πmH.(75)\Gamma(H\to f_L+\bar{f}_R)=\frac{|y_{\mathrm{eff}}^f|^2}{16\pi}m_H. \tag*{(75)}

Thus, we have

⟨ΓLR⟩(T)=gH32π∣yefff∣2mH3T2K1(mHT).(76)\left\langle\Gamma_{\mathrm{LR}}\right\rangle(T)=\frac{g_H}{32\pi}|y_{\mathrm{eff}}^f|^2\frac{m_H^3}{T^2}K_1\left(\frac{m_H}{T}\right). \tag*{(76)}

For T≫mHT\gg m_H we can approximate K1(mH/T)≈T/mHK_1(m_H/T)\approx T/m_H, and taking mH(T)∼cHTm_H(T)\sim c_HT we obtain

⟨ΓLR⟩(T)≃κfd=4∣yefff∣2T,(77)\left\langle\Gamma_{\mathrm{LR}}\right\rangle(T)\simeq\kappa_f^{d=4}|y_{\mathrm{eff}}^f|^2T, \tag*{(77)}

with

κfd=4=cH232π.(78)\kappa_f^{d=4}=\frac{c_H^2}{32\pi}. \tag*{(78)}

Other processes modify κfd=4\kappa_f^{d=4} by factors of O(1)\mathcal{O}(1). In particular, for the annihilation process fL+fˉR→t+tˉf_L+\bar{f}_R\to t+\bar{t} and fL+fˉR→H+Vμf_L+\bar{f}_R\to H+V_\mu, as well as their scattering counterparts, one obtains ⟨ΓLR⟩∼yt2∣yefff∣2T\langle\Gamma_{\mathrm{LR}}\rangle\sim y_t^2|y_{\mathrm{eff}}^f|^2T and ⟨ΓLR⟩∼g2∣yefff∣2T\langle\Gamma_{\mathrm{LR}}\rangle\sim g^2|y_{\mathrm{eff}}^f|^2T, respectively, with yty_t the top Yukawa and gg a gauge coupling.

Effects of thermal masses

The results in Section A 1 are obtained assuming massless states, which allowed us to write analytical expressions for the thermally averaged rates of the left-tight equilibration processes. Importantly, even in the fully unbroken phase where ⟨Φ⟩=⟨H⟩=0\langle\Phi\rangle=\langle H\rangle=0, particles still acquire thermal masses due to interactions with the plasma. More precisely, the mass of particle aa in the thermal bath is given by

ma2(T)=ma,02+Πa(T),(79)m_a^2(T)=m_{a,0}^2+\Pi_a(T), \tag*{(79)}

where ma,0m_{a,0} corresponds to the bare mass and Πa(T)\Pi_a(T) to the finite temperature self-energy contribution. For the high temperatures that we are interested in, the bare masses can be safely neglected, and the masses of weakly coupled particles are well described by

ma(T)∼ciT,(80)m_a(T) \sim c_i T, \tag*{(80)}

with cic_i a function of the relevant gauge, Yukawa, or scalar couplings.

An important consequence of non-vanishing thermal masses is their effect on the kinematics of the equilibration processes. For instance, the process

fL+fˉR→H+nfΦ,(81)f_L + \bar{f}_R \to H + n_f \Phi, \tag*{(81)}

requires

s>mH(T)+nfΦ,(82)\sqrt{s} > m_H(T) + n_f \Phi, \tag*{(82)}

which shows that thermal masses reduce the phase space accessibility, suppressing the corresponding rate. From this, we can already conclude that the inclusion of thermal masses, if relevant, will have the effect of allowing for a wider range of temperatures for which the asymmetry production could take place. In this sense, the computations made above are a conservative estimate.

We define the ratio

F(T)≡ΓLRm≠0(T)ΓLRm=0(T),(83)F(T) \equiv\frac{\Gamma_{LR}^{m\ne0}(T)}{\Gamma_{LR}^{m=0}(T)}, \tag*{(83)}

to quantify the difference between the rates in the massless limit and with the inclusion of thermal masses.

For the annihilation process in Eq. (81), we can factor out the squared amplitude from the phase space integral, which in the presence of non-vanishing masses cannot be evaluated analytically. Thus, we evaluate the MB reaction density by Monte Carlo integration over the collision energy and final-state phase space. At each sampled energy, we generate a massless final state with RAMBO [74] and rescale its momenta to accommodate the thermal masses while conserving energy and momentum. We use the same samples for the massive and massless calculations, weighting the massive contribution to account for the modified phase space, incoming kinematics, and squared amplitude. We then include the change in the equilibrium fermion density to obtain the rate ratio FF in Eq. (83).

The resulting values of F(T)F(T) are shown in Figure 8 for ci∈(0.1,1)c_i \in(0.1,1) for fermions and the Higgs boson, where gauge interactions are expected to give sizable contributions to the thermal masses, and cΦ∼5⋅10−2c_\Phi\sim5 \cdot10^{-2} for the flavon, which is conservative since we assume the Higgs portal to be negligibly small when describing the FNPT. These results demonstrate that the effects of thermal masses to our observable of interest are at most of O(1)O(1).

Figure 8. Fraction of thermally averaged rates with and without thermal masses included, for the processes fL+fˉR→H+nfΦf_L+\bar{f}_R\to H+n_f\Phi, as a function of different values of the thermal coefficients of the fermions and the Higgs.

Beyond Maxwell-Boltzmann

We estimate the effect of quantum statistics assuming massless states and based on the s-channel processes

fL+fˉR→H+nfΦf_L+\bar{f}_R\to H+n_f\Phi. For this, we define the ratio RR between the estimated rate using quantum statistics and massless states, γQSm=0\gamma_{QS}^{m=0}, and the rate in the massless MB approximation, γMBm=0\gamma_{MB}^{m=0}, namely

R=γQSm=0γMBm=0.(84)R = \frac{\gamma_{QS}^{m=0}}{\gamma_{MB}^{m=0}}. \tag*{(84)}

At fixed x=s/Tx=\sqrt{s}/T, we use RAMBO [74] to sample the massless (nf+1)(n_f+1)-body phase space uniformly. For each event jj, we define wj(x)w_j(x) as its quantum-statistical weight relative to the MB weight. Writing EijE_{ij} for the outgoing Higgs and flavon energies in the collision centre-of-mass frame, we have

wj(x)=1(1+e−x/2)2∏i=0nf11−e−Eij/T×1[min⁡i(Eij/T)≥ϵIR].(85)w_j(x)=\frac{1}{(1+e^{-x/2})^2}\prod_{i=0}^{n_f}\frac{1}{1-e^{-E_{ij}/T}}\times\mathbb{1}\left[\min_i(E_{ij}/T)\geq\epsilon_{\mathrm{IR}}\right]. \tag*{(85)}

The first factor is the ratio of the incoming Fermi-Dirac and MB occupation products, and the product accounts for Bose enhancement of the outgoing bosons. The infrared cutoff sets the event weight to zero if any outgoing boson has Eij<ϵIRTE_{ij}<\epsilon_{\mathrm{IR}}T. Varying ϵIR\epsilon_{\mathrm{IR}} probes sensitivity to soft bosons within massless kinematics.

Since the squared amplitude in Eq. (48) depends only on ss, we estimate the phase-space average by the arithmetic mean w‾(x)=M−1∑j=1Mwj(x)\overline{w}(x)=M^{-1}\sum_{j=1}^{M}w_j(x) over all MM generated events, including those with zero weight. We approximate the quantum correction for all pair velocities by its value for a pair at rest in the plasma.18

The results in Figure 9 show that, without an infrared cut, quantum statistics enhances the massless MB annihilation rate by an order of magnitude, more specifically factors of about 7 to 20, for nf>3n_f>3. The enhancement decreases as ϵIR\epsilon_{\mathrm{IR}} increases, indicating that soft final-state bosons contribute substantially. This means that the realistic ratio RR is expected to be smaller than the value computed with ϵIR=0\epsilon_{\mathrm{IR}}=0, since e.g. thermal masses are expected to weaken Bose enhancement by raising the energies of such low-momentum modes.

Figure 9. Effects of quantum statistics on the results obtained in the MB approximation with massless states, for different numbers of flavon in the final state of the process fL+fˉR→H+nfΦf_L+\bar{f}_R\to H+n_f\Phi. Different values of the infrared cutoff ϵIR\epsilon_{\mathrm{IR}}, resembling final temperature effects, are included.

Phase transition dynamics

Field-dependent masses

For the potential in Eq. (31) the flavon field dependent mass is given by

mΦ2=40cΛFN2Φ4−12λΦ2−2μ2.(86)m_\Phi^2=40\frac{c}{\Lambda_{\mathrm{FN}}^2}\Phi^4-12\lambda\Phi^2-2\mu^2. \tag*{(86)}

The thermal mass corrections in the Daisy potential in Eq. (30) are

Π(T)=T212(−12λ+40cΦ2ΛFN2+yξ22+∑iyi22h2niΦ2ni−2ΛFN2ni).(87)\Pi(T)=\frac{T^2}{12}\left(-12\lambda+40c\frac{\Phi^2}{\Lambda_{\mathrm{FN}}^2}+\frac{y_\xi^2}{2}+\sum_i\frac{y_i^2}{2}\frac{h^{2n_i}\Phi^{2n_i-2}}{\Lambda_{\mathrm{FN}}^{2n_i}}\right). \tag*{(87)}

The SM fermions have field-dependent masses including both Higgs and flavon

mf2=12yf2h2Φ2nfΛFN2nf.(88)m_f^2=\frac{1}{2}y_f^2h^2\frac{\Phi^{2n_f}}{\Lambda_{\mathrm{FN}}^{2n_f}}. \tag*{(88)}

which, before EWSB, while h=0h=0 do not contribute to the flavon thermal potential, which motivated the inclusion of some additional fermions charged under U(1)FNU(1)_{\mathrm{FN}} that couple to Φ\Phi directly, see Eq. (26) and accompanying description, with field-dependent mass

mξ2=12yξ2Φ2.(89)m_\xi^2=\frac{1}{2}y_\xi^2\Phi^2. \tag*{(89)}

Without Φ6\Phi^6 term

The scalar potential

V0=−μ2Φ2+λΦ4(90)V_0=-\mu^2\Phi^2+\lambda\Phi^4 \tag*{(90)}

can be expressed in terms of vev vΦv_\Phi and radial mode mass after SB m02m_0^2 as

V(Φ)=−m022Φ2+m022vΦ2Φ4.(91)V(\Phi)=-\frac{m_0^2}{2}\Phi^2+\frac{m_0^2}{2v_\Phi^2}\Phi^4. \tag*{(91)}

From this, ΔV=V(0)−V(vΦ)=−m028vΦ2\Delta V=V(0)-V(v_\Phi)=-\frac{m_0^2}{8}v_\Phi^2 follows. The reheating temperature can be estimated as

∣ΔV∣=ρrh=π230g∗Trh4⇒Trh=(30∣ΔV∣π2g∗)1/4.(92)|\Delta V|=\rho_{\mathrm{rh}}=\frac{\pi^2}{30}g_*T_{\mathrm{rh}}^4\quad\Rightarrow\quad T_{\mathrm{rh}}=\left(\frac{30|\Delta V|}{\pi^2g_*}\right)^{1/4}. \tag*{(92)}

and therefore

m02≲π2g∗30TEW4vΦ2∼(10 GeV)2(g∗100)(104 GeVvΦ)2.(93)\begin{aligned} m_0^2\lesssim\frac{\pi^2g_*}{30}\frac{T_{\mathrm{EW}}^4}{v_\Phi^2} \\ &\sim(10\ \mathrm{GeV})^2\left(\frac{g_*}{100}\right)\left(\frac{10^4\ \mathrm{GeV}}{v_\Phi}\right)^2. \tag*{(93)} \end{aligned}

Then, requiring vΦ≳104 GeVv_\Phi\gtrsim10^4\ \mathrm{GeV}, as motivated in § 1.3, we see that m0≲10 GeVm_0\lesssim10\ \mathrm{GeV}. However, a stronger constraint comes from the supercooling. At leading order around Φ=0\Phi=0, the tree level potential in combination with the finite temperature corrections can be expressed as

V0+VT∼(yξ2T2−μ2)Φ2(94)V_0+V_T\sim(y_\xi^2T^2-\mu^2)\Phi^2 \tag*{(94)}

and since the flavon mass is given by m02=2μ2m_0^2=2\mu^2, large flavon masses imply earlier FNSB. Additionally, the CW contribution when yξy_\xi is too large (∼1\sim1) prevents a minimum from forming in the correct range, and, carrying out a numerical scan tracking all contributions, we find feasible points with a vev in the correct range (∼0.1ΛFN\sim0.1\Lambda_{\mathrm{FN}}) and SB delayed until after EWSB only for

m02≲1 GeV2,(95)m_0^2\lesssim1\ \mathrm{GeV}^2, \tag*{(95)}

incompatible with the constraints from lepton flavor violation. Furthermore, the found minima are only metastable. After EWSB, the CW contribution of the SM fermions drives the potential to negative values at large Φ\Phi. Neither restriction applies when including higher dimensional operators in the potential.

Flavor bounds

To determine the contribution to flavor changing observables from the light flavon, the mass matrix

mijf=v2yijfϵnijf(96)m_{ij}^{f} = \frac{v}{\sqrt{2}}y_{ij}^{f}\epsilon^{n_{ij}^{f}} \tag*{(96)}

has to be diagonalised as

(ULf)†mfURf=m^f.(97)\left(U_{L}^{f}\right)^{\dagger}m^{f}U_{R}^{f}=\widehat{m}^{f}. \tag*{(97)}

Although this diagonalises the Higgs couplings, the flavon interactions retain off-diagonal entries

L⊃cijfˉLi(ρ+ia)fRj+h.c.(98)\mathcal{L}\supset c_{ij}\bar{f}_{L i}(\rho+ia)f_{R j}+\mathrm{h.c.} \tag*{(98)}

with coupling matrix

c=1vΦ(ULf)†(mfnf)URf,(99)c=\frac{1}{v_{\Phi}}\left(U_{L}^{f}\right)^{\dagger}\left(m^{f}n^{f}\right)U_{R}^{f}, \tag*{(99)}

and these induce the flavor changing processes characteristic of FN frameworks. The couplings of radial and axial mode differ by a factor of ii.

ALP contribution to μ→eγ\mu\to e\gamma

In the main text, we took the ALP mass as a free parameter and therefore did not explicitly include its contribution to the flavor observables, as it can both increase and decrease the constraints. Figure 10 shows the ratio

Color-coding of the $\mu \to e\gamma$ bound on the flavon vev

Figure 10. The color-coding shows the μ→eγ\mu\to e\gamma bound on the flavon vev as a ratio between the estimate in § 1.3 when only the radial mode is taken into account, compared to the value when both radial and axial mode are explicitly included. See text for details.

r=(vΦmin⁡)ρ+a(vΦmin⁡)ρ(100)r=\frac{\left(v_{\Phi}^{\min}\right)_{\rho+a}}{\left(v_{\Phi}^{\min}\right)_{\rho}} \tag*{(100)}

between the minimal allowed vev vΦmin⁡v_{\Phi}^{\min} when both fields are taken into account and only for the radial mode. Changing the couplings in Eq. (21) by (i)(i), the term in the second line obtains a minus sign and therefore, for similar radial and axial masses, the contributions to μ→eγ\mu\to e\gamma show some cancellations. If one field is lighter than the other, it gives the leading constraints on the flavon vev, which can be understood from the plot as follows: when ma>mρm_a>m_\rho, the ratio remains at 11 while for lighter ALPs the ratio grows larger.

Bounds from the quark sector

We separate the constraints that come from flavor changing processes in the quark sector into two regimes, depending on whether the flavon is heavier or lighter than the kaon and B-meson mass. Only the processes leading to the dominant constraints are listed, see for instance [49] for a more complete overview over all flavor observables.

Invisible meson decays

When ρ\rho is very light, invisible meson decays become kinematically allowed which are tightly constrained, see e.g. Ref [49]. Specifically, when mρ<mK(mB)∼0.5 GeV (5 GeV)m_\rho<m_K(m_B)\sim0.5\,\mathrm{GeV}\ (5\,\mathrm{GeV}), the invisible decays K→πρK\to\pi\rho and B→K(∗)ρB\to K^{(*)}\rho require the flavon vev to be larger than compatible with the reheating constraints determined in § 1.4.19 Therefore, when showing the constraints on the flavon vev in Figure 11 we have narrowed the mass range compared to Figure 5.

Figure 11. FCNC bounds from K−K‾K-\overline{K} oscillations, naively (light orange) or with texture suppression (orange). Low masses are not included due to very stringent invisible meson decay bounds. The excluded lower region comes from μ→eγ\mu\to e\gamma, see § 1.3. The colourful lines and points showcase maximal reheating temperatures for the viable points, see for details § 1.4.

Meson mixing

For heavier scalars, the dominant constraints come from meson mixing P0−Pˉ0P^0-\bar{P}^0, particularly from kaon mixing. As for μ→eγ\mu\to e\gamma, the precise constraint depends on the FN charge assignments which induce the off-diagonal quark couplings. Indeed, doing a naive analysis similar to the one in Ref. [61], and comparing the UTfit constraint [75]

∣Im⁡[C4sd]∣≲4×10−18 GeV−2(101)\left|\operatorname{Im}\left[C_{4}^{sd}\right]\right| \lesssim4 \times10^{-18}\,\mathrm{GeV}^{-2} \tag*{(101)}

with the expected Wilson coefficient

C4sd∼(εnsd−1vh2ΛFN1mρ)2,(102)C_{4}^{sd} \sim\left(\varepsilon^{n_{sd}-1}\frac{v_h}{2\Lambda_{\mathrm{FN}}}\frac{1}{m_\rho}\right)^2, \tag*{(102)}

where nsdn_{sd} is the sum of charges of the down-strange mixing terms, which might e.g. be nsd∼4n_{sd} \sim4 with ε∼0.1\varepsilon\sim0.1, we find

vΦ≳10 TeV×(1 TeVmρ)(103)v_\Phi\gtrsim10\,\mathrm{TeV}\times\left(\frac{1\,\mathrm{TeV}}{m_\rho}\right) \tag*{(103)}

[naive].

However, as shown in [52], the naive bound can be up to two orders of magnitude higher than strictly required. A favorable charge assignment can thus improve the bound by a factor 100 to be roughly, matching what was also found in [49],

vΦ≳10 TeV×(10 GeVmρ)(104)v_\Phi\gtrsim10\,\mathrm{TeV}\times\left(\frac{10\,\mathrm{GeV}}{m_\rho}\right) \tag*{(104)}

[best case].

Figure 11 shows the constraints in the quark sector, with the naive bound in light orange and the stringent bound in orange. The mass range is cut off at 5 GeV, due to the invisible meson decays (transparent gray cut-off in the plot). This, in combination with Trh<TsphT_{\mathrm{rh}}<T_{\mathrm{sph}} (red line), leads to an upper bound on the VEV of ∼105\sim10^5 GeV. With ε∼0.1\varepsilon\sim0.1, this can be translated to an upper bound on ΛFN∼106\Lambda_{\mathrm{FN}} \sim10^6 GeV. Comparing with Figure 4, it becomes clear that with such a low UV scale for a lepton with FN charge 1 (as is expected for the tau) there is no temperature range in which equilibration has become ineffective while sphalerons are active.

References

References

  1. [1]P. A. R. Ade et al. (Planck), “Planck 2015 results. XIII. Cosmological parameters,” Astron. Astrophys. 594, A13 (2016), arXiv:1502.01589 [astro-ph.CO].DOI
  2. [2]Ryan J. Cooke, Max Pettini, and Charles C. Steidel, “One Percent Determination of the Primordial Deuterium Abundance,” Astrophys. J. 855, 102 (2018), arXiv:1710.11129 [astro-ph.CO].DOI
  3. [3]N. Aghanim et al. (Planck), “Planck 2018 results. VI. Cosmological parameters,” Astron. Astrophys. 641, A6 (2020), [Erratum: Astron. Astrophys. 652, C4 (2021)], arXiv:1807.06209 [astro-ph.CO].DOI
  4. [4]Tsung-Han Yeh, Jessie Shelton, Keith A. Olive, and Brian D. Fields, “Probing physics beyond the standard model: limits from BBN and the CMB independently and combined,” JCAP 10, 046 (2022), arXiv:2207.13133 [astro-ph.CO].DOI
  5. [5]A. D. Sakharov, “Violation of CP Invariance, C asymmetry, and baryon asymmetry of the universe,” Pisma Zh. Eksp. Teor. Fiz. 5, 32–35 (1967).
  6. [6]V. A. Kuzmin, V. A. Rubakov, and M. E. Shaposhnikov, “On the Anomalous Electroweak Baryon Number Non-conservation in the Early Universe,” Phys. Lett. B 155, 36 (1985).
  7. [7]M. E. Shaposhnikov, “Possible Appearance of the Baryon Asymmetry of the Universe in an Electroweak Theory,” JETP Lett. 44, 465–468 (1986).
  8. [8]Glennys R. Farrar and M. E. Shaposhnikov, “Baryon asymmetry of the universe in the standard electroweak theory,” Phys. Rev. D 50, 774 (1994), arXiv:hep-ph/9305275.DOI
  9. [9]M. Fukugita and T. Yanagida, “Baryogenesis Without Grand Unification,” Phys. Lett. B 174, 45–47 (1986).DOI
  10. [10]M. A. Luty, “Baryogenesis via leptogenesis,” Phys. Rev. D 45, 455–465 (1992).pubmed.ncbi.nlm.nih.gov/10014395
  11. [11]Laura Covi, Esteban Roulet, and Francesco Vissani, “CP violating decays in leptogenesis scenarios,” Phys. Lett. B 384, 169–174 (1996), arXiv:hep-ph/9605319.DOI
  12. [12]Karin Dick, Manfred Lindner, Michael Ratz, and David Wright, “Leptogenesis with Dirac neutrinos,” Phys. Rev. Lett. 84, 4039–4040 (2000), arXiv:hep-ph/9907562.DOI
  13. [13]Evgeny K. Akhmedov, V. A. Rubakov, and A. Yu. Smirnov, “Baryogenesis via neutrino oscillations,” Phys. Rev. Lett. 81, 1359–1362 (1998), arXiv:hep-ph/9803255.DOI
  14. [14]Y. Fukuda et al. (Super-Kamiokande), “Evidence for oscillation of atmospheric neutrinos,” Phys. Rev. Lett. 81, 1562–1567 (1998), arXiv:hep-ex/9807003.
  15. [15]Q. R. Ahmad et al. (SNO), “Direct evidence for neutrino flavor transformation from neutral current interactions in the Sudbury Neutrino Observatory,” Phys. Rev. Lett. 89, 011301 (2002), arXiv:nucl-ex/0204008.
  16. [16]Hitoshi Murayama and Aaron Pierce, “Realistic Dirac leptogenesis,” Phys. Rev. Lett. 89, 271601 (2002), arXiv:hep-ph/0206177.DOI
  17. [17]Pei-Hong Gu and Hong-Jian He, “Neutrino Mass and Baryon Asymmetry from Dirac Seesaw,” JCAP 12, 010 (2006), arXiv:hep-ph/0610275.DOI
  18. [18]D. G. Cerdeno, A. Dedes, and T. E. J. Underwood, “The Minimal Phantom Sector of the Standard Model: Higgs Phenomenology and Dirac Leptogenesis,” JHEP 09, 067 (2006), arXiv:hep-ph/0607157.
  19. [19]Debasish Borah and Arnab Dasgupta, “Common Origin of Neutrino Mass, Dark Matter and Dirac Leptogenesis,” JCAP 12, 034 (2016), arXiv:1608.03872 [hep-ph].DOI
  20. [20]Mu-Chun Chen, Jinrui Huang, and William Shepherd, “Dirac Leptogenesis with a Non-anomalous U(1)’ Family Symmetry,” JHEP 11, 059 (2012), arXiv:1111.5018 [hep-ph].
  21. [21]Labh Singh, Monal Kashav, and Surender Verma, “Minimal type-I Dirac seesaw and leptogenesis under A4 modular invariance,” Nucl. Phys. B 1007, 116666 (2024), arXiv:2405.07165 [hep-ph].arxiv.org/abs/2405.07165
  22. [22]Manoranjan Dutta, Nimmala Narendra, Narendra Sahu, and Sujay Shil, “Asymmetric self-interacting dark matter via Dirac leptogenesis,” Phys. Rev. D 106, 095017 (2022), arXiv:2202.04704 [hep-ph].DOI
  23. [23]Maximilian Berbig, “Diracogenesis,” JHEP 01, 061 (2024), arXiv:2307.14121 [hep-ph].
  24. [24]G. Arcadi, J. P. Garcés, and M. Lindner, “Baryogenesis and Dark Matter from light Sterile Neutrinos,” (2026), arXiv:2603.19407 [hep-ph].arxiv.org/abs/2603.19407
  25. [25]Steven Abel and Veronique Page, “Affleck-Dine (pseudo)-Dirac neutrinogenesis,” JHEP 05, 024 (2006), arXiv:hep-ph/0601149.DOI
  26. [26]Maximilian Berbig, “S.M.A.S.H.E.D.: Standard Model Axion Seesaw Higgs inflation Extended for Dirac neutrinos,” JCAP 11, 042 (2022), arXiv:2207.08142 [hep-ph].DOI
  27. [27]Neil D. Barrie and Chengcheng Han, “Affleck-Dine Dirac Leptogenesis,” (2024), arXiv:2402.15245 [hep-ph].arxiv.org/abs/2402.15245
  28. [28]Aqeel Ahmed, Juan P. Garcés, and Manfred Lindner, “Primordial Dirac leptogenesis,” Phys. Rev. D 113, L111702 (2026), arXiv:2511.03794 [hep-ph].arxiv.org/abs/2511.03794
  29. [29]Pei-Hong Gu, “Peccei-Quinn symmetry for Dirac seesaw and leptogenesis,” JCAP 07, 004 (2016), arXiv:1603.05070 [hep-ph].DOI
  30. [30]K. S. Babu and Ajay Kaladharan, “Dirac leptogenesis in left-right symmetric models,” Phys. Rev. D 112, 035015 (2025), arXiv:2410.24125 [hep-ph].
  31. [31]Tomáš Blažek, Julian Heeck, Jan Heisig, Peter Maták, and Viktor Zaujeć, “Dirac leptogenesis from asymmetry wash-in via scatterings,” Phys. Rev. D 110, 055042 (2024), arXiv:2404.16934 [hep-ph].DOI
  32. [32]Mu-Chun Chen, Seyda Ipek, and Michael Ratz, “Baryogenesis from Flavon Decays,” Phys. Rev. D 100, 035011 (2019), arXiv:1903.06211 [hep-ph].DOI
  33. [33]C. D. Froggatt and Holger Bech Nielsen, “Hierarchy of Quark Masses, Cabibbo Angles and CP Violation,” Nucl. Phys. B 147, 277–298 (1979).DOI
  34. [34]Guy D. Moore, “The Sphaleron rate: Bodeker’s leading log,” Nucl. Phys. B 568, 367–404 (2000), arXiv:hep-ph/9810313.DOI
  35. [35]D. Bodeker, Guy D. Moore, and K. Rummukainen, “Chern-Simons number diffusion and hard thermal loops on the lattice,” Phys. Rev. D 61, 056003 (2000), arXiv:hep-ph/9907545.DOI
  36. [36]Luis Bento, “Sphaleron relaxation temperatures,” JCAP 11, 002 (2003), arXiv:hep-ph/0304263.DOI
  37. [37]Michela D’Onofrio, Kari Rummukainen, and Anders Tranberg, “Sphaleron Rate in the Minimal Standard Model,” Phys. Rev. Lett. 113, 141602 (2014), arXiv:1404.3565 [hep-ph].DOI
  38. [38]Denis Besak and Dietrich Bodeker, “Thermal production of ultrarelativistic right-handed neutrinos: Complete leading-order results,” JCAP 03, 029 (2012), arXiv:1202.1288 [hep-ph].DOI
  39. [39]I. Ghisoiu and M. Laine, “Right-handed neutrino production rate at T > 160 GeV,” JCAP 12, 032 (2014), arXiv:1411.1765 [hep-ph].DOI
  40. [40]D. Bodeker and M. Laine, “Kubo relations and radiative corrections for lepton number washout,” JCAP 05, 041 (2014), arXiv:1403.2755 [hep-ph].DOI
  41. [41]Lorenzo Calibbi, Florian Goertz, Diego Redigolo, Robert Ziegler, and Jure Zupan, “Minimal axion model from flavor,” Phys. Rev. D 95, 095009 (2017), arXiv:1612.08040 [hep-ph].
  42. [42]Yohei Ema, Koichi Hamaguchi, Takeo Moroi, and Kazunori Nakayama, “Flaxion: a minimal extension to solve puzzles in the standard model,” JHEP 01, 096 (2017), arXiv:1612.05492 [hep-ph].DOI
  43. [43]Yohei Ema, Daisuke Hagiwara, Koichi Hamaguchi, Takeo Moroi, and Kazunori Nakayama, “Supersymmetric Flaxion,” JHEP 04, 094 (2018), arXiv:1802.07739 [hep-ph].DOI
  44. [44]Tommi Alanne, Simone Blasi, and Florian Goertz, “Common source for scalars: Flavored axion-Higgs unification,” Phys. Rev. D 99, 015028 (2019), arXiv:1807.10156 [hep-ph].DOI
  45. [45]A. Jodidio et al., “Search for Right-Handed Currents in Muon Decay,” Phys. Rev. D 34, 1967 (1986), [Erratum: Phys.Rev.D 37, 237 (1988)].
  46. [46]R. Bayes et al. (TWIST), “Search for two body muon decay signals,” Phys. Rev. D 91, 052020 (2015), arXiv:1409.0638 [hep-ex].DOI
  47. [47]I. Adachi et al. (Belle-II), “Search for Lepton-Flavor-Violating τ Decays to a Lepton and an Invisible Boson at Belle II,” Phys. Rev. Lett. 130, 181803 (2023), arXiv:2212.06334 [hep-ex].
  48. [48]K. Afanaciev et al. (MEG II), “New limit on the μ+ → e+γ decay with the MEG II experiment,” Eur. Phys. J. C 85, 1177 (2025), [Erratum: Eur.Phys.J.C 85, 1317 (2025)], arXiv:2504.15711 [hep-ex].arxiv.org/abs/2504.15711
  49. [49]Admir Greljo, Aleks S. Smolkovič, and Alessandro Valenti, “Froggatt-Nielsen ALP,” JHEP 09, 174 (2024), arXiv:2407.02998 [hep-ph].arxiv.org/abs/2407.02998
  50. [50]F. Takahashi et al. (Particle Data Group), “Review of Particle Physics,” Int. J. Mod. Phys. A 41, 2630011 (2026).
  51. [51]Claudia Cornella, David Curtin, Gordon Krnjaic, and Micah Mellors, “Testing the Froggatt-Nielsen mechanism with lepton flavor and number violating processes,” Phys. Rev. D 112, 115010 (2025), arXiv:2501.00629 [hep-ph].DOI
  52. [52]Claudia Cornella, David Curtin, Ethan T. Neil, and Jedidiah O. Thompson, “Mapping and probing Froggatt-Nielsen solutions to the quark flavor puzzle,” Phys. Rev. D 111, 015042 (2025), arXiv:2306.08026 [hep-ph].arxiv.org/abs/2306.08026
  53. [53]K. Afanaciev et al. (MEG II), “Operation and performance of the MEG II detector,” Eur. Phys. J. C 84, 190 (2024), arXiv:2310.11902 [physics.ins-det].arxiv.org/abs/2310.11902
  54. [54]K. Arndt et al. (Mu3e), “Technical design of the phase I Mu3e experiment,” Nucl. Instrum. Meth. A 1014, 165679 (2021), arXiv:2009.11690 [physics.ins-det].DOI
  55. [55]Gavin Hesketh, Sean Hughes, Ann-Kathrin Perrevoort, and Nikolaos Rompotis (Mu3e), “The Mu3e Experiment,” in Snowmass 2021 (2022) arXiv:2204.00001 [hep-ex].arxiv.org/abs/2204.00001
  56. [56]W. Altmannshofer et al. (Belle-II), “The Belle II Physics Book,” PTEP 2019, 123C01 (2019), [Erratum: PTEP 2020, 029201 (2020)], arXiv:1808.10567 [hep-ex].DOI
  57. [57]L. Bartoszek et al. (Mu2e), “Mu2e Technical Design Report,” (2014), 10.2172/1172555, arXiv:1501.05241 [physics.ins-det].arxiv.org/abs/1501.05241
  58. [58]R. Abramishvili et al. (COMET), “COMET Phase-I Technical Design Report,” PTEP 2020, 033C01 (2020), arXiv:1812.09018 [physics.ins-det].DOI
  59. [59]Anna Driutti (Mu2e), “Status of the Mu2e experiment at Fermilab,” PoS WIFIA2025, 020 (2026).DOI
  60. [60]Thomas Clouvel (COMET), “Update on the status of COMET experiment,” in 18th International Workshop on Tau Lepton Physics (2026).DOI
  61. [61]Jason Baldes, Thomas Konstandin, and Geraldine Servant, “Flavor Cosmology: Dynamical Yukawas in the Froggatt-Nielsen Mechanism,” JHEP 12, 073 (2016), arXiv:1608.03254 [hep-ph].DOI
  62. [62]Greg W. Anderson and Lawrence J. Hall, “The Electroweak phase transition and baryogenesis,” Phys. Rev. D 45, 2685–2698 (1992).pubmed.ncbi.nlm.nih.gov/10014659
  63. [63]M. E. Carrington, “The Effective potential at finite temperature in the Standard Model,” Phys. Rev. D 45, 2933–2944 (1992).pubmed.ncbi.nlm.nih.gov/10014687
  64. [64]Peter Brockway Arnold and Olivier Espinosa, “The Effective potential and first order phase transitions: Beyond leading-order,” Phys. Rev. D 47, 3546 (1993), [Erratum: Phys.Rev.D 50, 6662 (1994)], arXiv:hep-ph/9212235.arxiv.org/abs/hep-ph/9212235
  65. [65]David H. Lyth and Ewan D. Stewart, “Cosmology with a TeV mass GUT Higgs,” Phys. Rev. Lett. 75, 201–204 (1995), arXiv:hep-ph/9502417.DOI
  66. [66]David H. Lyth and Ewan D. Stewart, “Thermal inflation and the moduli problem,” Phys. Rev. D 53, 1784–1798 (1996), arXiv:hep-ph/9510204.DOI
  67. [67]Sidney R. Coleman, “The Fate of the False Vacuum. 1. Semiclassical Theory,” Phys. Rev. D 15, 2929–2936 (1977), [Erratum: Phys.Rev.D 16, 1248 (1977)].
  68. [68]Andrei D. Linde, “Fate of the False Vacuum at Finite Temperature: Theory and Applications,” Phys. Lett. B 100, 37–40 (1981).DOI
  69. [69]Andrei D. Linde, “Decay of the False Vacuum at Finite Temperature,” Nucl. Phys. B 216, 421 (1983), [Erratum: Nucl.Phys.B 223, 544 (1983)].DOI
  70. [70]J. R. Espinosa, “A Fresh Look at the Calculation of Tunneling Actions,” JCAP 07, 036 (2018), arXiv:1805.03680 [hep-th].DOI
  71. [71]Marek Lewicki, Oriol Pujolàs, and Ville Vaskonen, “Escape from supercooling with or without bubbles: gravitational wave signatures,” Eur. Phys. J. C 81, 857 (2021), arXiv:2106.09706 [astro-ph.CO].DOI
  72. [72]Tomasz P. Dutka, Tae Hyun Jung, and Chang Sub Shin, “What happens when supercooling is terminated by curvature flipping of the effective potential?” JHEP 05, 182 (2025), arXiv:2412.15864 [hep-ph].DOI
  73. [73]M. Joyce and Mikhail E. Shaposhnikov, “Primordial magnetic fields, right-handed electrons, and the Abelian anomaly,” Phys. Rev. Lett. 79, 1193–1196 (1997), arXiv:astro-ph/9703005.DOI
  74. [74]R. Kleiss, W. J. Stirling, and S. D. Ellis, “A new monte carlo treatment of multiparticle phase space at high energies,” Comput. Phys. Commun. 40, 359–373 (1986).DOI
  75. [75]Ludovico Vittorio et al. (UTfit), “Global fits of the Unitarity Triangle: Summer 2025 update from the UTfit collaboration,” PoS EPS-HEP2025, 302 (2026).DOI

Paper details

vlnc:W_zuznk2gu
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 10

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents