Arxiv:1811.08034V1 [Hep-Ph] 20 Nov 2018 & Tion Seems to Suggest a Sizeable Hierarchy in the Sfermion Model the Strong Dynamics with a Slice of Ads5

UMN-TH-3804/18

Predicting the Sparticle Spectrum from Partially-Composite Supersymmetry

Yusuf Buyukdag, Tony Gherghetta, Andrew S. Miller School of Physics and Astronomy, University of Minnesota, Minneapolis, Minnesota 55455, USA Email: [email protected], [email protected], [email protected]

We use the idea of partial compositeness in a minimal supersymmetric model to relate the fermion and sfermion masses. By assuming that the Higgs and third-generation matter is (mostly) elementary, while the first- and second-generation matter is (mostly) composite, the Yukawa coupling hierarchy can be explained by a linear mixing between elementary states and composite operators with large anomalous dimensions. If the composite sector also breaks supersymmetry, then composite sfermions such as selectrons are predicted to be much heavier than the lighter elementary stops. This inverted sfermion mass hierarchy is consistent with current experimental limits that prefer light stops (O(10) TeV) to accommodate the 125 GeV Higgs boson, while predicting heavy first- and second-generation sfermions (& 100 TeV) as indicated by flavor physics experiments. The underlying dynamics can be modelled by a dual 5D gravity theory that also predicts a gravitino dark matter candidate (& keV), together with gauginos and Higgsinos, ranging from 10–90 TeV, that are split from the heavier first- and second-generation sfermion spectrum. This intricate connection between the fermion and sfermion mass spectrum can be tested at future experiments.

I. INTRODUCTION then depend on the relative compositeness of the Stan- Supersymmetry provides a compelling theoretical dard Model fermions. To obtain an order-one Yukawa framework for addressing some of the shortcomings of coupling with the Higgs, the top quark must be mostly the Standard Model of particle physics. These include elementary, while, since the elementary and composite dark matter, gauge coupling unification, and the sta- sectors mix with an irrelevant coupling, the smallness bilization of the hierarchy between the electroweak and of the electron Yukawa coupling follows from assuming Planck scales. Because supersymmetry must be broken, that the electron is mostly composite. The remainder of the stability of the electroweak scale requires that the the Standard Model Yukawa couplings are generated by sparticle spectrum should not be too heavy. A vital varying degrees of compositeness. If one now further sup- clue for determining the superpartner mass scale comes poses that the composite sector is responsible for break- from the recent discovery of the 125 GeV Higgs bo- ing supersymmetry, then an inverted hierarchy immedi- son [1, 2]. To obtain this mass in minimal supersym- ately follows. Selectrons, which are mostly composite, re- metry, the Higgs quartic coupling must receive sizeable ceive large supersymmetry-breaking masses, while stops, radiative corrections. These can arise from the top quark which are elementary, obtain suppressed supersymmetry- superpartners (or stops), provided that the lightest stop breaking contributions. In this way the fermion mass has mass of O(10) TeV. In the minimal framework there hierarchy determines the sfermion mass hierarchy and are no other sizeable contributions to the Higgs quar- predicts an inverted mass spectrum. tic coupling, and consequently the rest of the sparticle The underlying strong dynamics that would be re- spectrum is not determined. The spectrum must only sponsible for such a mechanism is similar to single-sector be compatible with the current LHC limits that require models of supersymmetry breaking that were originally proposed in [6, 7], with related work in [8–12]. Even stop masses to be & 1120 GeV [3] and gluino masses if the underlying gauge theory were completely known, & 1970 GeV [4]. Other indirect constraints such as the absence of flavor-changing processes, prefer the first- however, predictions for the spectrum would be difficult and second-generation scalar masses to be much heav- to obtain due to the nonperturbative dynamics. There- ier, 100 TeV. Thus, the current experimental situa- fore, we will instead use the AdS/CFT correspondence to arXiv:1811.08034v1 [hep-ph] 20 Nov 2018 & tion seems to suggest a sizeable hierarchy in the sfermion model the strong dynamics with a slice of AdS5. In light mass spectrum, which is inverted compared to the well- of the Higgs boson discovery, this enables us to obtain known fermion mass hierarchy. For instance, the electron specific quantitative predictions for the sparticle spec- (top quark) is the lightest (heaviest) charged fermion, trum that can then be used to help guide future experi- while the selectron (stop) may be the heaviest (lightest) mental searches. charged sfermion. In this Letter, we provide a mechanism that explains II. PARTIALLY-COMPOSITE the origin of the inverted sfermion mass hierarchy and SUPERSYMMETRY predicts the sparticle spectrum. The mechanism re- To illustrate the mechanism of partially-composite su- lies on partial compositeness [5], whereby the Stan- persymmetry,√ consider the elementary chiral superfield dard Model fields are admixtures arising from the lin- Φ = φ + 2 θψ + θθF , where φ is a complex scalar, ψ ear mixing of elementary states with composite oper- is a Weyl fermion, and F is an auxiliary field. In addi- ators. Assuming that the Higgs fields are elementary, tion, we introduce a√ corresponding supersymmetric chiral the magnitudes of the corresponding Yukawa couplings operator O = Oφ + 2 θOψ + θθOF . The scaling dimen- 2 sions of the component operators are dim Oφ = 1 + δO, fermion), the Yukawa coupling is of order one for suffi- 3 dim Oψ = 2 + δO, and dim OF = 2 + δO, where δO ≥ 0 ciently large N. Conversely, when 0 ≤ δ < 1 (correspond- is the anomalous dimension [13]. ing to a sizeable composite admixture), the Yukawa cou- The supersymmetric Lagrangian contains separate ele- pling has a power-law suppression that depends on the mentary and composite sectors, together with linear mix- degree of compositeness. This explains why composite c ing terms of the form [Φ O ]F for each chiral superfield fermions (identified with the first- and second-generation Φ and charge-conjugate composite operator Oc. At the Standard Model fermions) have small Yukawa couplings, UV scale ΛUV, it is given by while elementary fermions (such as the top quark) have order-one Yukawa couplings. † 1 c LΦ = Φ Φ D + δ−1 ([Φ O ]F + h.c.) , (1) The composite sector is also responsible for supersym- ΛUV metry breaking. Soft scalar masses are generated only where δ is the anomalous dimension of Oc. We have taken for the composite sector fields since there is no direct order-one UV coefficients for the higher-dimension terms, coupling of the supersymmetry breaking to elementary (1) and omitted a kinetic mixing between the elementary fields. For example, the massive scalar field, φ obtains and composite sectors in our minimal setup. The com- a soft mass posite sector is assumed to confine at an infrared scale, g(1)2 h i |F |2 Λ . In the limit of large-N strong dynamics, the two- Φ † (1)† (1) (1)2 X (1)† (1) IR ξ4 2 X X Φ Φ = ξ4gΦ 2 φ φ , (4) point function for the composite operator components ΛIR D ΛIR can be written as a sum over one-particle states. In par- where X = θθFX is a composite-sector spurion and ξ4 is ticular, for the scalar component, the two-point function a dimensionless parameter. Given the scalar admixture hO (p)O (−p)i = P a2 /(p2 + m2 ) to leading order in φ φ n n √ n (2), the corresponding sfermion mass-squared becomes: 1/N, where an = h0|Oφ|ni ∝ N/(4π) is the matrix el- th  2 2 2(δ−1) ement for Oφ to create the n state with mass mn from (δ−1) 16π |FX | ΛIR  ζ N Λ2 Λ δ ≥ 1 , the vacuum [14]. m2 ' Φ IR UV (5) e 2 2 The elementary-composite mixing in the Lagrangian (1−δ) 16π |FX |  ζ N Λ2 0 ≤ δ < 1 , (1) mixes the elementary fields (φ, ψ) with the compos- Φ IR 2 ite resonance states. Assuming for simplicity just the where, for a large-N gauge theory, ξ4 ' 16π /N [14]. lowest-lying composite state Φ(1) = (φ(1), ψ(1)) with mass When the sfermion is mostly elementary (δ ≥ 1), the (1) gΦ ΛIR, the two-state system can be diagonalized to ob- soft mass is power-law suppressed since the supersymme- tain the massless eigenstate [15] try breaking is transmitted via the elementary-composite   mixing. (Note, however, that for sufficiently large δ, ra- v diative corrections will become increasingly important.)  1 u δ − 1  |Φ i ' N |Φi − u |Φ(1)i , This contrasts with the case 0 ≤ δ < 1, where the mass 0 Φ (1)√ t 2(1−δ)  gΦ ζΦ ΛIR  eigenstate is mostly composite and there is no power-  Λ − 1  UV law suppression. Thus, elementary sfermions (identi- (2) (1) fied with the stops) are much lighter than the com- where Φ0 = (φ0, ψ0), gΦ and ζΦ are order-one constants, posite sfermions (identified with the first- and second- and NΦ is a normalization constant. Given that ΛIR generation sfermions), predicting an inverted mass hier- ΛUV, this expression shows that the massless eigenstates archy! are mostly elementary for δ > 1, whereas for 0 ≤ δ < The fermion and sfermion mass hierarchies critically 1 they are an admixture of elementary and composite depend on the anomalous dimensions δ. To illustrate states. this, we consider the δe,t values required to obtain the This elementary-composite admixture of the massless electron top-quark Yukawa coupling ratio ye/yt at the eigenstate can now be used to explain the fermion mass IR scale. These are plotted in Fig. 1 for various values of hierarchy [16], and then predict the sfermion mass spec- −5 ΛIR/ΛUV, where the ratio ye/yt (∼10 ) at the IR scale trum. Consider elementary chiral fermions, ψL,R, that is determined via two-loop renormalization group evolu- are coupled to the elementary Higgs field, H, via the † tion (assuming a universal soft mass threshold). Using Yukawa interaction λ ψLψRH +h.c., where λ is an order- (5), this then predicts the sfermion mass ratio m /m e et one proto-Yukawa coupling (for simplicity, we assume one at the IR scale. As shown in Fig. 1, the allowed region fermion generation and ignore the distinction between H u is 0 . δe . 0.9 and 1 . δt . 1.8, depending on the and Hd). Diagonalizing the fermion Lagrangian with the value of ΛIR/ΛUV. The largest value of the ratio me/mt Higgs contribution gives the Yukawa coupling expression −3 e−16 e is approximately 140 (390) for ΛIR/ΛUV ' 10 (10 ).  λ 16π2 Note that the slanted horizontal contours in Fig. 1 end on  ζ (δ − 1) N δ ≥ 1 , y ' Φ (3) the right, at the δt value for which radiative corrections ψ 2 2(1−δ)  λ (1 − δ) 16π ΛIR 0 ≤ δ < 1 , to the soft mass (5) begin to dominate. These corrections ζΦ N ΛUV are calculated in Ref. [15]. where we have assumed that δ ≡ δL = δR. We clearly see A partially-composite analysis can also be done for that when δ ≥ 1 (corresponding to a mostly elementary the vector and gravity supermultiplets. They lead to a 3

standard model in the AdS5 bulk. The N = 1 chiral mat- 10-16 0.8 ter and vector superfields are embedded into 5D N = 1 10-10 hypermultiplets and vector supermultiplets, respectively. The 4D superfields are then identified with the mass- 10-7 less zero modes, while the massive Kaluza-Klein states 0.6 form 4D N = 2 supermultiplets with masses of order 10-5 ΛIR. The N = 1 Higgs supermultiplets, meanwhile, are

e 4D fields confined to the UV brane. In this setup, each δ 0.4 10-4 fermion zero mode obtains a mass from a UV boundary- bulk Yukawa interaction with 5D Yukawa coupling, Y (5). 350 100 200 40 The fermion mass hierarchy then arises from the overlap 0.2 10 of the UV-localized Higgs fields with the left- and right- 10-3 ( 1 ∓c)ky handed bulk fermion fields with profiles ψL,R ∝ e 2 in the fifth dimension, where the c are dimensionless bulk fermion mass parameters [20, 21]. Once the c parameters 0.0 1.0 1.2 1.4 1.6 1.8 are determined for each fermion flavor, the sparticle mass spectrum can then be predicted. δ t Supersymmetry is only broken on the IR brane and can be parameterized by introducing a boundary interaction FIG. 1. The estimated range of anomalous dimensions δe, δt 2 with a spurion X = θ FX for each 5D hypermultiplet that gives rise to the observed hierarchy between the electron µ 18 Φ(x , y): and top quark Yukawa couplings, assuming ΛUV = 10 GeV, tan β = 3, and a soft mass threshold at 50 TeV. The slanted Z √ Z X†X horizontal and vertical lines are contours of the ratio ΛIR/ΛUV d5x −g d4θ Φ†Φ δ(y − πR) . (7) and the sfermion mass ratio m /m , respectively. Λ2 k e te UV This interaction leads to the sfermion soft mass mostly elementary gauge boson and gaugino, and an ele-  1 1/2 F ( 1 ∓c)πkR 1 mentary graviton and gravitino [17]. Since supersymme-  ±c − e 2 ±c > , m ' 2 ΛIR 2 (8) try breaking occurs in the composite sector, this implies e L,R 1/2  1 ∓ c F ±c < 1 , that the gauginos are lighter than the mostly compos- 2 ΛIR 2 √ √ ite first- and second-generation sfermions and compara- −πkR ble in mass to the mostly elementary third-generation where F ≡ FX e , and the back-reaction on the sfermions. On the other hand, since the gravitino has a √sfermion profile by the boundary mass is negligible (i.e., tiny composite admixture, it is almost always the light- F/ΛIR . 1). Furthermore, note that possible flavor est supersymmetric particle (LSP). These are the quali- off-diagonal mass terms in (7) have been neglected since tative features of the partially-composite sparticle spec- the sfermion mass scale is assumed to be O(100) TeV. Us- −πkR trum. Further details are presented in Ref. [15]. ing the AdS/CFT dictionary relations ΛIR/ΛUV = e 1 and δ = |c ± 2 |, the expressions (8) are seen to be consis- III. A 5D GRAVITY MODEL tent with the masses (5) obtained in the 4D holographic theory. Quantum corrections to the tree-level scalar The partially-composite supersymmetric framework masses (8) arising from loops of bulk hypermultiplets generically relates the fermion and sfermion mass spec- and vector supermultiplets are important for suppressed tra that result from some (unknown) strong dynam- 1 masses (±c & 2 ). These are computed in Ref. [15], and ics. In order to model the underlying dynamics and their effect is typically to reduce the sfermion mass hier- obtain quantitative predictions, we now consider a five- archy. dimensional (5D) dual gravity model that is motivated by Similarly, introducing an IR-boundary gaugino inter- the AdS/CFT correspondence [18]. The 5D spacetime, a aα a action term XWα W , where Wα is the gauge field (xµ, y), where µ = 0, 1, 2, 3 labels the four-dimensional strength superfield, gives rise to gaugino masses Mλa ' (4D) coordinates and the fifth coordinate, y, is compact- 2 gaF/ΛIR [22], with ga (a = 1, 2, 3) the corresponding 1 ified on an orbifold (S /Z2). The anti-de Sitter (AdS) Standard Model gauge couplings. The gaugino masses metric is given by are suppressed relative to the heavier sfermions (with ±c < 1 ). Alternatively, if the supersymmetry-breaking ds2 = e−2kydx2 + dy2 , (6) 2 sector does not contain any gauge singlets, the gaug- where k is the AdS curvature scale. The 5D spacetime is ino masses may instead be generated by a boundary in- † a aα a slice of AdS5 bounded by two 3-branes located at the teraction X XWα W . This leads to gaugino masses, 2 2 3 orbifold fixed points: a UV brane at y = 0 and an IR Mλa ' gaF /ΛIR that are further suppressed. brane at y = πR, where R is the orbifold radius [19]. When supersymmetry is spontaneously broken on the Besides gravity, we introduce the full matter and IR boundary, the effective 4D cosmological constant re- gauge-sector content of the minimal supersymmetric ceives a positive contribution from FX . In the 5D warped 4

16 √FIG. 2. The sparticle mass spectrum for two benchmark scenarios: singlet spurion case (hatched) with ΛIR = 2 × 10 √GeV, 10 (5) 6 F = 4.75 × 10 GeV, tan β ∼ 3 and Y k = 1; and non-singlet spurion case (solid) with ΛIR = 6.5 × 10 GeV, F = 2 × 106 GeV, tan β ∼ 5 and Y (5)k = 1. geometry, this contribution can be canceled by the addi- • Gravitino dark matter: Assuming R-parity conser- tion of a constant superpotential W on the UV brane√ [22– vation, the gravitino LSP makes an excellent dark mat- 28], giving rise to a gravitino mass m3/2 ' F/( 3MP ). ter candidate, provided m3/2 & 1 keV [30, 31]. Since the gravitational coupling is Planck-scale sup- • Higgs mass and electroweak symmetry break- pressed, the gravitino mass is lower than the character- ing: The observed 125 GeV Higgs boson can be accom- istic soft-mass scale F/ΛIR by a warp factor. modated if the mass of the lightest stop is O(10) TeV. The Higgs sector does not couple directly to the IR Since the Higgs-sector soft terms are generated radia- brane, and therefore the Higgs soft terms m2 , m2 , tively, the requirement that the Higgs scalar potential Hu Hd and b as well as the trilinear soft scalar couplings (a- correctly breaks electroweak symmetry leads to further terms) at the IR-brane scale are zero at tree level. How- indirect constraints on the soft masses of the sfermions. ever, these soft terms are generated via radiative correc- • Supersymmetric flavor problem: To avoid gener- tions from their interactions with bulk hypermultiplets ating excessive flavor-changing processes, the first- and and vector supermultiplets [15]. The resulting values for second-generation sfermions must be at least 100 TeV. the Higgs soft masses, obtained at the IR-brane scale, • Gauge coupling unification: To preserve the suc- must be run down to near the electroweak scale in or- cessful supersymmetric prediction of gauge coupling der to check that electroweak symmetry is broken. The unification (assuming any underlying dynamics is Higgs µ-term is assumed to arise on the UV brane from a SU(5) symmetric), the gaugino and Higgsino masses higher-dimensional superpotential term allowed by an ex- must be lighter than 300 TeV. tra U(1) symmetry, as in the Kim-Nilles mechanism [29]. Its value, along with tan β (the ratio of the Higgs vacuum • Charge- and color-breaking minima: Since expectation values), is determined by the conditions for the predicted sfermion mass spectrum at ΛIR is electroweak symmetry breaking. flavor-dependent and the first- and second-generation sfermions are typically hierarchically larger than the The parameters of the 5D model therefore consist of third-generation sfermions, there are both one-loop D- the IR√ brane scale, ΛIR, and the supersymmetry breaking term and two-loop gauge contributions to scalar masses scale, F . In addition, there is a universal 5D Yukawa that can lead to charge and color-breaking minima. coupling, Y (5), and the six bulk fermion mass parameters cLi,Qi (one for each generation of leptons and quarks). Subject to the above constraints, we choose two bench- These parameters can be used to determine both the mark scenarios corresponding to the singlet and non- fermion and sparticle mass spectra. However, there are a singlet spurion cases. The√ singlet case has parameter val- 16 10 (5) number of phenomenological and theoretical constraints ues ΛIR = 2×10 GeV, F = 4.75×10 GeV, Y k = which restrict the possible parameter values. These in- 1, and tan β ' 3 at the IR-brane scale, whereas the non- 6 clude: singlet case has parameter values ΛIR = 6.5 × 10 GeV, 5 √ F = 2 × 106 GeV, Y (5)k = 1, and tan β ' 5 at the 90 TeV. These masses are split from the heavier first- IR-brane scale. The sfermion pole mass predictions are and second-generation sfermions, thereby preserving the presented in Fig. 2, where the spread in the masses re- successful supersymmetric prediction of gauge coupling sults from a scan over the c-parameters in order to fit the unification. A more detailed analysis of this model is Yukawa coupling hierarchy. The Higgs mass lies in the given in Ref. [15]. range 124–126 GeV, with sign µ = −1 . Furthermore, the mass of the LSP gravitino is 535 GeV (1 keV) for the The partially-composite supersymmetric model intri- singlet (non-singlet) spurion case. The sfermion masses cately connects the generation of the fermion mass hier- obtained directly result from explaining the fermion mass archy with the sfermion masses. It is striking that the hierarchy. They reveal a distinctive, flavor-dependent in- predicted sparticle spectrum seems to provide an appeal- verted mass hierarchy, in contrast to usual supersymmet- ing fit to the current experimental constraints. While ric models where scalar and gaugino masses are uncon- not directly accessible at the 13 TeV LHC, the signa- strained by the fermion mass spectrum. tures of this sparticle spectrum, such as distinctive long- lived NLSP decays, may be within the reach of a future high-energy collider. Alternatively, the heavy first- and IV. CONCLUSION second-generation sfermions could be indirectly probed In this Letter, we have presented a partially-composite at flavor-violation experiments such as the Mu2e experi- supersymmetric model that assumes the first two gener- ment [32] or at experiments aiming to measure the elec- ations of matter are (mostly) composite, while the Higgs tric dipole moment of the electron [33]. Thus, partial and third generation matter are (mostly) elementary. compositeness and supersymmetry are intriguing possi- This feature can then be used to explain the fermion bilities that could together play a central role in address- mass hierarchy, predicting, as a consequence, a distinct ing some of the shortcomings of the Standard Model. sparticle mass spectrum with an inverted sfermion mass hierarchy: light stops and staus and heavy first-and second-generation sfermions. The underlying dynamics Acknowledgments We thank Jason Evans, Ben Har- responsible for the compositeness can be modelled with ling, and Alex Pomarol for helpful discussions. This work a dual 5D gravity theory that further predicts a grav- is supported in part by the U.S. Department of Energy itino LSP, together with gauginos and Higgsinos rang- under Grant DE-SC0011842 at the University of Min- ing from the lightest neutralino at 10 TeV to gluinos at nesota.

[1] G. Aad et al. [ATLAS Collaboration], Phys. Lett. B 716, [15] Y. Buyukdag, T. Gherghetta and A. Miller, to appear. 1 (2012) [arXiv:1207.7214 [hep-ex]]. [16] R. Contino and A. Pomarol, JHEP 0411, 058 (2004) [2] S. Chatrchyan et al. [CMS Collaboration], Phys. Lett. B [hep-th/0406257]. 716, 30 (2012) [arXiv:1207.7235 [hep-ex]]. [17] B. Batell and T. Gherghetta, Phys. Rev. D 76, 045017 [3] A. M. Sirunyan et al. [CMS Collaboration], JHEP 1710, (2007) [arXiv:0706.0890 [hep-th]]. 019 (2017) [arXiv:1706.04402 [hep-ex]]. [18] J. M. Maldacena, Int. J. Theor. Phys. 38, 1113 [4] M. Aaboud et al. [ATLAS Collaboration], JHEP 1806, (1999) [Adv. Theor. Math. Phys. 2, 231 (1998)] [hep- 107 (2018) [arXiv:1711.01901 [hep-ex]]. th/9711200]. [5] D. B. Kaplan, Nucl. Phys. B 365, 259 (1991). [19] L. Randall and R. Sundrum, Phys. Rev. Lett. 83, 3370 [6] N. Arkani-Hamed, M. A. Luty and J. Terning, Phys. Rev. (1999) [hep-ph/9905221]. D 58, 015004 (1998) [hep-ph/9712389]. [20] Y. Grossman and M. Neubert, Phys. Lett. B 474, 361 [7] M. A. Luty and J. Terning, Phys. Rev. D 62, 075006 (2000) [hep-ph/9912408]. (2000) [hep-ph/9812290]. [21] T. Gherghetta and A. Pomarol, Nucl. Phys. B 586, 141 [8] T. Gherghetta and A. Pomarol, Nucl. Phys. B 602, 3 (2000) [hep-ph/0003129]. (2001) [hep-ph/0012378]. [22] Z. Chacko and E. Ponton, JHEP 0311, 024 (2003) [hep- [9] M. Gabella, T. Gherghetta and J. Giedt, Phys. Rev. D ph/0301171]. 76, 055001 (2007) [arXiv:0704.3571 [hep-ph]]. [23] L. Randall and R. Sundrum, Nucl. Phys. B 557, 79 [10] S. Franco and S. Kachru, Phys. Rev. D 81, 095020 (2010) (1999) [hep-th/9810155]. [arXiv:0907.2689 [hep-th]]. [24] M. A. Luty and R. Sundrum, Phys. Rev. D 64, 065012 [11] N. Craig, R. Essig, S. Franco, S. Kachru and G. Torroba, (2001) [hep-th/0012158]. Phys. Rev. D 81, 075015 (2010) [arXiv:0911.2467 [hep- [25] M. A. Luty, Phys. Rev. Lett. 89, 141801 (2002) [hep- ph]]. th/0205077]. [12] O. Aharony, L. Berdichevsky, M. Berkooz, Y. Hochberg [26] T. Gherghetta and A. Pomarol, Phys. Lett. B 536, 277 and D. Robles-Llana, Phys. Rev. D 81, 085006 (2010) (2002) [hep-th/0203120]. [arXiv:1001.0637 [hep-ph]]. [27] H. Itoh, N. Okada and T. Yamashita, Phys. Rev. D 74, [13] G. Cacciapaglia, G. Marandella and J. Terning, JHEP 055005 (2007) [hep-ph/0606156]. 0906, 027 (2009) [arXiv:0802.2946 [hep-th]]. [28] T. Gherghetta, B. von Harling and N. Setzer, JHEP [14] E. Witten, Nucl. Phys. B 160, 57 (1979). 1107, 011 (2011) [arXiv:1104.3171 [hep-ph]]. 6

[29] J. E. Kim and H. P. Nilles, Phys. Lett. 138B, 150 (1984). [32] L. Bartoszek et al. [Mu2e Collaboration], [30] M. Bolz, A. Brandenburg and W. Buchmuller, Nucl. [arXiv:1501.05241 [physics.ins-det]]. Phys. B 606, 518 (2001) Erratum: [Nucl. Phys. B 790, [33] V. Andreev et al. [ACME Collaboration], Nature 562, 336 (2008)] [hep-ph/0012052]. no. 7727, 355 (2018). [31] J. Pradler and F. D. Steﬀen, Phys. Lett. B 648, 224 (2007) [hep-ph/0612291].