Restoring minimal modular neutrino masses via Kähler $R$-parity violation

Modular symmetry arising in string compactification has played an important role in the bottom-up approach [1] to understanding the structure of the observed fermion masses and mixing angles; for recent reviews, see [2, 3, 4]. The modular symmetric framework is described in the context of supersymmetry by three elements: (i) the underlying symmetry group $\Gamma_{N}$, which is a finite discrete modular group; (ii) the modulus $\tau$, which is a chiral superfield; and (iii) modular forms $Y_{r}^{(k_{y})}(\tau)$, which transform as an $r$-dimensional representation of the underlying $\Gamma_{N}$ and possess modular weight $k_{y}$. Given the group $\Gamma_{N}$, the modular forms are completely determined as functions of $\tau$ alone. The scalar components of $Y_{r}^{(k_{y})}(\tau)$ play the role of the Yukawa couplings. This aspect distinguishes modular symmetry from conventional discrete symmetries, in which the Yukawa couplings are functions of the unknown vacuum expectation values of the flavon fields that break the underlying symmetry group. However, in the modular symmetric framework, the arbitrariness enters through the choice of the representations and modular weights of the various matter fields, and it is these choices that determine the structure of the effective Yukawa couplings and, ultimately, the spectrum of the matter fields and their interactions. These choices are not unique and can lead to a large number of possibilities in determining the underlying model [5].

The minimality has always been the guiding principle in making such a choice. For the leptonic sector, the simplest possibility among all the analysed models is the very first model discussed by Feruglio in [1]. It is based on the group $\Gamma_{3}\simeq A_{4}$, which happens to be the smallest modular group with a 3-dimensional irreducible representation to which the three generations of leptonic doublets $L_{i}$ can be assigned. The neutrino masses are attributed in this model to some source of lepton number violation at a high scale $\Lambda_{L}$ and are described in terms of a dimension-5 Weinberg operator. The charged leptons obtain masses through an $A_{4}$ triplet flavon field $\phi$, while a triplet $Y_{3}^{(2)}$ with weight 2 completely determines the neutrino masses and leptonic mixing in terms of a single parameter $\tau$. This simple model correctly reproduces the solar angle and the ratio of the solar and atmospheric mass scales in accordance with the inverted ordering of neutrino masses for a specific choice of $\tau$. It, however, fails to obtain viable values of the two other angles, $\theta_{23}$ and $\theta_{13}$, consistent with the data. This led to investigations of more complex models involving different choices of representation assignments, field content, and/or symmetry group, leading to different classes of effective Yukawa couplings [6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16]. For example, an extensive analysis of possibilities within the $A_{4}$ group was performed in [13], considering both the Weinberg operator and seesaw mechanisms for neutrino mass generation. While none of the Weinberg-operator models can describe the data, several seesaw models can accommodate it, albeit with a more complicated Yukawa sector than originally envisaged in [1].

Our aim in this paper is to reanalyse the aforementioned simplest model, listed as Example-1 by Feruglio in [1]. We find that the original model in its simplest form contains a hitherto unnoticed source of neutrino masses, whose inclusion allows for a very good fit to all leptonic observables. This new source arises from $R$-parity violation [17, 18, 19, 20, 21, 22, 23], see [24] for a review and references. If lepton number conservation is not insisted upon, then this term is allowed in the Kähler potential of the model without altering the field content or symmetry of the original model. This lepton-number-violating term generates an effective bilinear $R$-parity-violating (RPV) term $\epsilon_{i}L_{i}H_{u}$ in the superpotential once supersymmetry is broken in theories with local supersymmetry. The mechanism is identical to the Giudice–Masiero mechanism for generating the $\mu$ term [25]. The $\epsilon_{i}$ are determined by the modular form $Y_{3}^{(2)}(\tau)$ and its derivative and introduce two additional parameters compared to the original model. Including this contribution allows for both normal and inverted neutrino mass orderings and describes the entire neutrino spectrum in terms of only three or four parameters, depending on the modulus contribution to supersymmetry breaking.

The same RPV term in the Kähler potential also allows the fermionic submultiplets of the flavon triplet chiral superfield $\phi$ to couple to $L_{i}$, generating a Dirac mass term. Thus, if massive, the $\phi$ fermions automatically play the role of right-handed neutrinos. This opens the possibility of a seesaw model within the same framework, leading to a setup similar to the model discussed as Example-2 by Feruglio in [1]. Unlike the latter, we show that the seesaw version in our framework, together with the RPV contribution, can precisely describe the leptonic spectrum and yield a phenomenologically consistent neutrino mass spectrum.

We first reformulate Feruglio’s original model in supergravity rather than global supersymmetry to implement the Giudice–Masiero (GM) mechanism for $R$-parity violation. This requires a modification of the relation between the modular weights of the matter fields relative to that adopted in [1]. Supersymmetry (SUSY) breaking is necessary for implementing the GM mechanism. We assume that one source of SUSY breaking can be the $F$-term of the modulus $\tau$, but do not consider it to be the sole source. An additional source of SUSY breaking is also present and contributes to the gravitino mass, $m_{3/2}$. It may arise, for example, from a hidden-sector superpotential, gaugino condensation or dilaton dynamics. The SUSY-breaking scales associated with these two sources need not be the same, and one can dominate over the other in realistic cases. As we will argue, this is required for the consistency of the model.

Recall [1] that the complex modulus $\tau$, with ${\rm Im}(\tau)>0$, and a chiral superfield $\Phi_{I}$ with weight $k_{\Phi_{i}}$ transforms as

under the transformations of the homogeneous modular group with $a,b,c,d\in{\mathbb{Z}}$ and $ad-bc=1$. Here, $\rho(\gamma)$ denotes the representation of the corresponding finite modular group $\Gamma_{N}$ according to which $\Phi_{I}$ transforms. For the model of our interest, $\Gamma_{N}\simeq A_{4}$, under which the leptonic superfield $L$ and a flavon $\phi$ transform as $3$-dimensional irreducible representations, while the three generations of the charged-lepton superfields $E^{c}_{i}$ transform as $1$, $1^{\prime\prime}$ and $1^{\prime}$, respectively. The Higgs superfields $H_{u,d}$ are ordinary singlets under $A_{4}$. The modular weights of $L$, $H_{u}$, $H_{d}$, $\phi$, $E_{i}^{c}$ are denoted by $k_{L}$, $k_{u}$, $k_{d}$, $k_{\phi}$ and $k_{E^{c}_{i}}$, respectively.

Generalisation of the original model to local SUSY implies that Kähler potential $K\equiv K(\tau,\overline{\tau};\Phi_{I},\bar{\Phi}_{I})$ and superpotential $W\equiv W(\tau,\Phi_{I})$ cannot be treated as independent functions. The model is instead characterised by a combination,

which is required to be invariant under the modular transformations. For a usual Kähler potential of the form

where

The transformation of $K$ is

This can be compensated by the second term in Eq. (2) if $W$ transforms as

Therefore, $W$ must carry total weight $+1$ in this generalisation to supergravity, provided $K$ has the simplest aforementioned form.

Let us now revisit Feruglio’s model in light of this generalisation. The charged lepton masses come from

where $[...]_{r}$ denotes the representation $r$ of $A_{4}$ constructed from the tensor product of superfields within it. The vacuum expectation value (VEV) of the flavon is assumed to have the form

This breaks $A_{4}$ to a $Z_{3}$ subgroup and leads to a diagonal mass matrix for the charged leptons. The neutrino mass originates from the Weinberg operator

where $Y$ is a triplet containing three linearly independent weight 2 modular forms, $Y_{i}\equiv Y_{i}(\tau)$. This completely determines the structure of the neutrino mass matrix,

leading to a very predictive framework for the neutrino masses and mixing. Along with this, the model also contains a usual $\mu$-term,

which does not play a direct role in determining the lepton masses and mixing in the $R$-parity-conserving case. However, it will become relevant in the case we describe next.

Generalisation to local SUSY and the resulting requirement, Eq. (6), for the above $W_{e}$, $W_{\nu}$ and $W_{\mu}$ lead to the following constraints on the combinations of the weights:

The second condition uniquely fixes the modular weight of $LH_{u}$ to $3/2$. As argued in [1], other operators involving $\phi$, or combinations of $\phi$ and $Y$, must be forbidden in $W_{\nu}$ to maintain the predictivity of the simplest structure of $M_{\nu}$. This is ensured by requiring $k_{\phi}<-2$ in the present case. The negative modular weight of $\phi$ also guarantees that no higher-dimensional operators involving $\phi$ and $Y$ can appear in $W_{e}$.

The bilinear $R$-parity violation, if enabled, can provide an additional source of lepton number violation and neutrino masses [24] within this setup. This would be desirable since the minimal model fails to reproduce the observed data. The usual bilinear RPV term in the superpotential is forbidden by holomorphicity if $k_{\phi}<-2$ within the present setup. However, one can introduce a term in the Kähler potential,

provided $k_{\phi}=-7/2$. The combination of the superfields and $Y^{\dagger}(\overline{\tau})$ above transforms with a factor of $(c\tau+d)^{2}\,(c\overline{\tau}+d)^{2}$ which is compensated by the transformation of $Z^{-2}$, leaving the entire term invariant under the action of modular transformation.

Once the SUSY is broken by the $F$-terms of the moduli and/or hidden sector fields, it induces an effective trilinear term in the superpotential. Computation of this for sub-Planck scale $\Lambda$, following the generalisation of [25, 26], leads to (see Appendix A)

Further simplification allows one to express

in terms of the canonically normalised fields, where

and $g=g_{2}/g_{1}$. Here, $g_{1,2}$ result from the fact that there are two independent contractions, i.e. $[Y^{\dagger}(\overline{\tau})(\phi L)_{3_{S}}]_{1}$ and $[Y^{\dagger}(\overline{\tau})(\phi L)_{3_{A}}]_{1}$ allowed in $K_{\rm GM}$. They are suitably rescaled to absorb the factors arising from the normalisation of kinetic terms. Also,

Note that $(Y_{D})_{ij}$ are no longer restricted to be holomorphic functions of modular forms, as they effectively originate from the Kähler potential through SUSY breaking.

Once the scalar component of $\phi_{i}$ acquires non-vanishing vacuum, see Eq. (8), $W_{\rm eff}$ leads to an effective bilinear RPV term parametrised as

As is well known [27], the generation of $\epsilon_{i}$ is not sufficient to obtain neutrino masses, since the $\epsilon_{i}$ term can be rotated away from the superpotential by redefining $L_{i}$ and $H_{d}$. The $\epsilon_{i}$ cannot, however, be simultaneously removed from the soft SUSY-breaking terms if they are non-universal. This leads to sneutrino VEVs, and in turn, masses for the neutrinos.

Following the convention used in [28], we parameterize the relevant soft terms as,

Computation of the above soft parameters for the Kähler potential given in Eq. (3), following [26], leads to

Consequently, a non-zero $F^{\tau}$ gives rise to non-universal soft SUSY-breaking parameters at the mediation scale itself in the present framework. The structure of the $F^{\tau}$-induced soft SUSY breaking and some of its phenomenological aspects have been discussed in [29, 30, 31, 32].

The sneutrino VEVs in the basis in which the bilinear terms are rotated away from $W$ can be expressed as [27] $\omega_{i}=\kappa_{i}\epsilon_{i}$, where

is obtained by minimizing the full scalar potential [33, 28, 34]. The sneutrino VEVs mix neutrinos with neutralinos and generate a contribution to the neutrino masses when the neutralinos are integrated out. In the seesaw limit ($\omega_{i}\ll\mu,M_{1,2}$, where $M_{1,2}$ are the gaugino soft masses), this contribution is given by¹¹ 1 In addition to this, $\epsilon_{i}$ also generates a loop-induced [24] contribution to neutrino masses. This is sub-dominant compared to the tree-level contribution considered here, and will be neglected.

where,

Before we reassess the phenomenological viability of the neutrino spectrum in light of this new source of neutrino masses, let us point out some features which crucially depend on whether $F^{\overline{\tau}}$ is the only source of SUSY breaking. For convenience, we parametrize at the vacuum value of $\tau$,

where $|\xi|\leq\sqrt{3}$ in general, and the highest value is achieved if $F^{\overline{\tau}}$ is the sole source of SUSY breaking²² 2 Assuming vanishing vacuum energy density for the supergravity scalar potential leads to a relation $m_{3/2}^{2}=\frac{1}{3}K_{I\bar{J}}F^{I}F^{\bar{J}}$, where $I,J$ stand for moduli and hidden sector fields. If only $F^{\overline{\tau}}$ is nonvanishing, then $K$ specified in Eq. (3) leads to $m_{3/2}^{2}=\frac{1}{3}|F^{\overline{\tau}}|^{2}Z^{2}$ at the minimum of the modulus.. On the other hand, if $F^{\overline{\tau}}$ is the only source of SUSY breaking, then the soft masses determined in Eq. (20), along with the relations in Eq. (12) and $k_{\phi}=-7/2$, lead to the fact that $(2m_{L}^{2}+m_{E^{c}}^{2})/m_{3/2}^{2}=3-5\xi^{2}$. Requiring this quantity to be positive, to prevent the charge-breaking minima, leads to an even more stringent limit,

Hence, $F^{\overline{\tau}}$ cannot be the only source of SUSY breaking within the present setup.

Computation of $\kappa_{i}$ by substituting Eqs. (12,20,24) in Eq. (21) leads to

where we ignore ${\cal O}(M_{Z}^{2}/m_{3/2}^{2})$ terms. Apparently, $\kappa_{i}$ do not vanish at high scale even in the $\xi\to 0$ limit. However, they are flavour universal in this limit. When $\xi\neq 0$, small but calculable non-universality among them gets induced through the second term in Eq. (26). Note that $\mu$ is an independent parameter in the present setup, and one expects $\mu\sim{\cal O}(m_{3/2})$ for low-energy SUSY. Alternatively, the $\mu$-term itself can be generated using the GM mechanism with a slight modification within this setup; see Appendix B. With $\mu\sim{\cal O}(m_{3/2})$ and neglecting ${\cal O}(\xi^{2})$ terms, $\kappa_{i}$ at low energy can be written as $\kappa_{i}=\kappa_{0}x_{i}$ with

An additional source of SUSY breaking in the conventional approach would introduce flavour-universal terms $m_{0}^{2}$, $m_{3/2}A_{0}$ and $m_{3/2}B_{0}$ in $m_{\Phi_{I}}^{2}$, $A_{i}$ and $B_{\mu}$, respectively. This replaces $1\to 1+B_{0}-A_{0}$ in the first term of $\kappa_{i}$. For $F^{\overline{\tau}}=0$, this coefficient can anyway be absorbed into the definition of $\kappa_{0}$. Otherwise, it constitutes an additional parameter. For definiteness, we set $A_{0}=B_{0}$ and neglect their contribution to $\kappa_{i}$ in the subsequent analysis.

Combining both the contributions, the neutrino mass matrix can be parameterised as

where

with $\tilde{X}_{i}=X_{i}\,x_{i}$ and $g=g_{2}/g_{1}$. $X_{i}$ involves terms anti-holomorphic in $Y_{i}$ as well as their derivatives. They are expressed more explicitly in Appendix C. The parameter

quantifies the relative strength of two contributions. The leptonic mixing angles, CP phases and the ratio of solar to atmospheric squared-mass differences are completely determined by complex $\tau$, $g$ and real $r$, $\xi$.

Numerically, the non-vanishing RPV contribution significantly improves the fit to the current best-fit neutrino oscillation data, as shown in Table 1. We also compare with Feruglio’s original model, which was found to be closer to the data only for the inverted ordering. Remarkably, even for $|F^{\bar{\tau}}|=0$, the RPV contribution renders the model viable, although the resulting $\sum m_{\nu}$ is slightly above the current limit, $\sum m_{\nu}<0.12$ eV, from cosmological simulations [37]. This can be brought within the bound by allowing a small $|F^{\bar{\tau}}|$, in which case the $\kappa_{i}$ become non-universal. Interestingly, viable normal ordering is also obtained for $|F^{\bar{\tau}}|\neq 0$. Although the RPV contribution introduces additional parameters that may seemingly improve the fit trivially, they arise intrinsically within the framework, rather than from any model extension or enlarged symmetry.

Using the fitted value of $r$ in Table 1 and Eqs. (30,27) along with $A_{0}\sim m_{3/2}^{-1}$, $u\sim 10^{-2}$, we find $g_{1}^{2}\sim m_{3/2}/\Lambda_{L}$. Note that $g_{1}$ sets the overall strength of the term in $K_{\rm GM}$, and for values close to the above, the RPV contribution plays a non-trivial and decisive role in making the minimal model consistent with the data. It is to be noted that the required strength of the $R$-violating contribution compared to the Weinberg operator as measured by $r$ is less than $0.3\%$.

The $W_{\rm eff}$ in Eq. (15) induced by the non-minimal Kähler term also gives rise to a Dirac-type Yukawa interaction term between the $L_{i}$ and the fermionic components of $\phi_{i}$. If the latter acquire large but finite masses, then one naturally obtains a seesaw contribution to the neutrino masses within this setup. Essentially, the fermionic partners in $\phi_{i}$ play the role of right-handed neutrinos. Moreover, the VEVs of their scalar counterparts break $R$-parity spontaneously, leading to a framework like that in [38, 39], studied earlier but without modular symmetry.

To generate masses for $\phi_{i}$ fermions, we consider the simplest possibility of introducing a bare mass term, $\frac{M_{T}}{2}(\phi\phi)_{1}$. This leads to the following mass matrix for the fermion triplet:

The bare mass term sets $k_{\phi}=1/2$. This along with the $K_{\rm GM}$, $W_{e}$ and $W_{\mu}$ implies $k_{u}=-5/2-k_{L}$, $k_{d}=7/2+k_{L}$ and $k_{E^{c}_{i}}=-3-2k_{L}$. Consequently, $(LH_{u})^{2}$ carries a weight $-5$ and $\phi$ being the only flavon with positive weight, the lowest-order Weinberg operator arises only at $(\phi/\Lambda_{L})^{12}$, making it vanishingly small for $\Lambda_{L}\geq\Lambda$. Similarly, the lowest order correction to $W_{e}$ arises at $(\phi^{4}Y/\Lambda^{4})$ and it is also negligible for $u\sim 10^{-2}$.

For $M_{T}\gg v_{u}$, the seesaw-induced light neutrino mass matrix is given by

where

This, along with the RPV contribution of Eq. (22), can be combined to write the full neutrino mass matrix as

where $r$ in the present case is given by

Comparing with the previous case, this essentially implies that $M_{T}$ is a source of lepton-number violation, with $M_{T}\sim(g_{1}m_{3/2}/\Lambda)^{2}\Lambda_{L}$. However, the flavour structures of the mass matrices are very different.

Numerically, this case provides an excellent fit to the data, as shown in Table 2. Even in the limit of vanishing RPV contribution, a good fit is obtained for normal ordering, with only $\sin^{2}\theta_{23}$ deviating by more than $3\sigma$ from its experimental value. Switching on the RPV contribution, even with $|F^{\overline{\tau}}|=0$, yields an excellent fit to the data. An equally good fit is also obtained for inverted ordering. The comprehensive predictions for both orderings are given in Appendix D.

The values of $r$ extracted from the numerical solutions in Table 2, along with $A_{0}\sim m_{3/2}^{-1}$, $u\sim 10^{-2}$ in Eq. (35) leads to

Hence, $\Lambda$ cannot be made arbitrarily large in the present case. Also, $M_{T}$ cannot be much larger than $m_{3/2}$ without lowering $\Lambda$. One finds $\Lambda\sim{\cal O}(10\,m_{3/2})$ for $M_{T}\sim m_{3/2}$. Recall that this constraint follows from the requirement that both the contributions to neutrino masses are comparable in size. Moreover, reproducing the scale of the atmospheric squared-mass difference from the fits requires the overall coefficient in Eq. (34) to be of ${\cal O}(10^{-13})$ GeV. This, along with Eq. (36), implies

Nevertheless, the small value of $g_{1}$ is technically natural.

We have reanalysed the basic idea of describing neutrino masses solely in terms of modular forms. Taking the minimal modular $A_{4}$ model proposed by Feruglio as an example, we have demonstrated that supersymmetric models based on modular symmetry admit an additional source of neutrino masses through $R$-parity violation. The latter is inherently present in these models once lepton number conservation is not insisted upon. RPV terms may be forbidden in the superpotential by the choice of symmetry and field content, but they can arise in the Kähler potential and, within the framework of local supersymmetry, reappear in the effective superpotential after SUSY breaking. The resulting RPV contribution to the neutrino mass matrix is also described by the modular forms, but this is not necessarily restricted to holomorphic functions. We have shown that this additional contribution can provide an important, and in some cases crucial, correction to the neutrino mass matrix, qualitatively altering the viability of the model. In particular, we considered examples of the simplest models proposed by Feruglio based on $A_{4}$ modular symmetry, which are not in precise agreement with the neutrino oscillation data. Including the RPV contribution restores their viability and yields an excellent fit to the data.