PHYSICAL REVIEW D 97, 116012 (2018)
Decay rate of electroweak vacuum in the standard model and beyond
So Chigusa,1 Takeo Moroi,1 and Yutaro Shoji2 1Department of Physics, University of Tokyo, Tokyo 113-0033, Japan 2Institute for Cosmic Ray Research, The University of Tokyo, Kashiwa 277-8582, Japan
(Received 26 March 2018; published 14 June 2018)
We perform a precise calculation of the decay rate of the electroweak vacuum in the standard model as well as in models beyond the standard model. We use a recently developed technique to calculate the decay rate of a false vacuum, which provides a gauge invariant calculation of the decay rate at the one- loop level. We give a prescription to take into account the zero modes in association with translational, dilatational, and gauge symmetries. We calculate the decay rate per unit volume, γ, by using an analytic formula. The decay rate of the electroweak vacuum in the standard model is estimated to be 3 þ40þ184þ144þ2 log10γ × Gyr Gpc ¼ −582−45−329−218−1 , where the first, second, third, and fourth errors are due to the uncertainties of the Higgs mass, the top quark mass, the strong coupling constant and the choice of the renormalization scale, respectively. The analytic formula of the decay rate, as well as its fitting formula given in this paper, is also applicable to models that exhibit a classical scale invariance at a high energy scale. As an example, we consider extra fermions that couple to the standard model Higgs boson, and discuss their effects on the decay rate of the electroweak vacuum.
DOI: 10.1103/PhysRevD.97.116012
I. INTRODUCTION vacuum can be evaluated through a rate of bubble nucle- ation in unit volume and unit time as formulated in [26,27]. In the standard model (SM) of particle physics, it has The rate is expressed in the form of been known that the Higgs quartic coupling may become negative at a high scale through quantum corrections, so γ ¼ Ae−B; ð1:1Þ that the Higgs potential develops a deeper vacuum. The detailed shape of the Higgs potential depends on the Higgs where B is the action of a so-called bounce solution, and and the top masses; with the recently observed Higgs mass prefactor A is quantum corrections having mass dimension of ∼125 GeV, it has been known that the electroweak (EW) 4. The bounce solution is an Oð4Þ symmetric solution of vacuum is not absolutely stable if the SM is valid up to ∼1010 1 the Euclidean equations of motion, connecting the two GeV or higher. In such a case, the EW vacuum can vacua. Although the dominant suppression of the decay rate decay into the deeper vacuum through tunneling in quan- comes from B, the prefactor A is also important. This is tum field theory. The lifetime of the EW vacuum has been because of large quantum corrections from the top quarks one of the important topics in particle physics and and the gauge bosons. Thus, it is essential to calculate both cosmology. A and B to determine the decay rate precisely. In the SM, The decay rate of the EW vacuum has been discussed for there are infinite bounce solutions owing to (i) the classical a long time. The calculation of the decay rate at the one- scale invariance at a high energy scale, (ii) the global loop level first appeared in [15] and was also discussed in ð2Þ ð1Þ ð1Þ – symmetries corresponding to SU L × U Y=U EM,as other literature [16 25]. However, there are subtleties in the well as (iii) the translational invariance. For the calculation treatment of zero modes related to the gauge symmetry of the prefactor A, a proper procedure to take account of the breaking which make it difficult to perform a precise and effects of the zero modes related to (i) and (ii) were not well reliable calculation of the decay rate. The lifetime of a understood until recently. In addition, the previous calcu- lations of A were not performed in a gauge-invariant way, 1For the absolute stability of the EW vacuum in the SM, which made the gauge invariance of the result unclear. see [1–14]. Recently, a prescription for the treatments of the gauge zero modes was developed [28,29], based on which a Published by the American Physical Society under the terms of the Creative Commons Attribution 4.0 International license. complete calculation of the decay rate of the EW vacuum Further distribution of this work must maintain attribution to became possible. The calculation has been performed by the author(s) and the published article’s title, journal citation, the present authors in a recent publication [25] and also by and DOI. Funded by SCOAP3. [24]. The purpose of this paper is to give a more complete
2470-0010=2018=97(11)=116012(34) 116012-1 Published by the American Physical Society SO CHIGUSA, TAKEO MOROI, and YUTARO SHOJI PHYS. REV. D 97, 116012 (2018) and detailed discussion about the calculation of the decay the quartic coupling constant negative at a high scale. In the rate. In [25], we have numerically evaluated the functional SM, the instability occurs when the Higgs amplitude determinants of fluctuation operators, which are necessary becomes much larger than the EW scale. Since the typical for the calculation of the decay rate. Here, we perform the field value for the bounce configuration is around that scale, calculation analytically; a part of the analytic results was we can neglect the quadratic term in the Higgs potential. first given in [24]. The effects of the zero modes and the In this section, we use a toy model with Uð1Þ gauge modes with the lowest angular momentum are carefully symmetry to derive relevant formulas. The calculation of taken into account, on which previous works had some the decay rate of the EW vacuum is almost parallel to that in confusion. We give fitting formulas of the functional the case with Uð1Þ gauge symmetry; the application to the determinants based on analytic results, which are useful SM case will be explained in the next section. for the numerical calculation of the decay rate. We also provide a C þþpackage to study the electroweak vacuum A. Setup stability (ELVAS), which is available at [30]. In this paper, we discuss the calculation of the decay rate Let us first summarize the setup of our analysis. We of the EW vacuum in detail. We first present a detailed study the decay rate of a false vacuum whose instability is formulation of the calculation of the decay rate at the one- due to an RG running of the quartic coupling constant of a loop level. We derive a complete set of analytic formulas scalar field, Φ. We assume that Φ is charged under the Uð1Þ that can be used for any models that exhibit classical scale gauge symmetry (with charge þ1); the kinetic term invariance at a high energy scale like the SM. Then, as one includes of the important applications, we calculate the decay rate of L ∋ ½ð∂ − ÞΦ †ð∂ − ÞΦ ð Þ the EW vacuum in the SM. We find that the lifetime of the kin μ igAμ μ igAμ ; 2:1 EW vacuum is much longer than the age of the Universe. There, we see that one-loop corrections from the top quark where Aμ is the gauge field and g is the gauge coupling and the gauge bosons are very large although there is an constant, while we consider the following scalar potential: accidental cancellation. It shows the importance of A for the evaluation of a decay rate. We also evaluate the decay VðΦÞ¼λðΦ†ΦÞ2: ð2:2Þ rates of the EW vacuum for models with extra fermions that couple to the Higgs field. In such models, the EW vacuum The quartic coupling, λ, depends on the renormalization tends to be destabilized compared with that of the SM since scale, μ, and is assumed to become negative at a high scale the quartic coupling of the Higgs field is strongly driven to due to the RG effect. As we have mentioned before, we a negative value. (For discussion about the stability of the neglect the quadratic term assuming that λ becomes negative EW vacuum in models with extra particles, see [22,31– at a much higher scale. In this setup, the scalar potential has 54].) We consider three models that contain, in addition to scale invariance at the classical level. In the application to the the SM particles, (i) vectorlike fermions having the same case of the SM, Φ corresponds to the Higgs doublet and λ SM charges as the left-handed quark and the right-handed corresponds to the Higgs quartic coupling constant. down quark, (ii) vectorlike fermions with the same SM Hereafter, we perform a detailed study of the effects of charges as left-handed lepton and right-handed electron, the fields coupled to Φ on the decay rate of the false and (iii) a right-handed neutrino. We give constraints on vacuum. We consider a Lagrangian that contains the their couplings and masses, requiring that the lifetime of the following interaction terms: EW vacuum be long enough. L ∋ κσ2jΦj2 þð Φψ¯ ψ þ Þþ ðΦÞ ð Þ This paper is organized as follows. In Sec. II,we int y L R H:c: V ; 2:3 summarize the formulation for the decay rate at the one- loop level, where we provide an analytic formula for each where σ is a real scalar field, and ψ L and ψ R are chiral field that couples to the Higgs boson. The detail of the fermions [with relevant Uð1Þ charges].2 We take y real and calculation is given in Appendixes A–D. In Sec. III,we κ > 0. We neglect dimensionful parameters that are evaluate the vacuum decay rate in the SM. Readers who are assumed to be much smaller than the typical scale of the interested only in the results can skip over the former bounce. In addition, gauge fixing is necessary to take into section to this section. In Sec. IV, we analyze decay rates in account the effects of gauge boson loops. Following models with extra fermions. Finally, we conclude in Sec. V. [29,55], we take the gauge fixing function of the following form: II. FORMULATION F ¼ ∂μAμ: ð2:4Þ We first discuss how we calculate the decay rate of the EW vacuum. In the SM, the EW vacuum becomes unstable due to the renormalization group (RG) running of the 2We assume that there exist other chiral fermions that cancel quartic coupling constant of the Higgs boson, which makes out the gauge anomaly.
116012-2 DECAY RATE OF ELECTROWEAK VACUUM IN THE … PHYS. REV. D 97, 116012 (2018)
Then, the gauge fixing term and the Lagrangian of the where h and φ are the physical Higgs mode and the ghosts (denoted as c¯ and c) are given by Nambu-Goldstone (NG) mode, respectively. At the one- loop level, the prefactor can be decomposed as 1 L ¼ F 2 ð Þ ð Þ GF ; 2:5 A h 2ξ A −B ¼ AðσÞAðψÞAðAμ;φÞAðc;c¯Þ −B ð Þ e V e ; 2:13 4D L ¼ −¯∂ ∂ ð Þ FP c μ μc; 2:6 V AðXÞ where 4D is the volume of spacetime, and is the 3 contribution from particle X. Each of the factors has a where ξ is the gauge fixing parameter. form of The bounce solution is an Oð4Þ symmetric object [56,57], and transforms under a global Uð1Þ symmetry. ðXÞ wðXÞ ð Þ DetM Thus, choosing the center of the bounce at r ¼ 0 (with A X ¼ ; ð2:14Þ pffiffiffiffiffiffiffiffiffi McðXÞ r ≡ xμxμ), we can write the bounce solution as Det ð Þ where MðXÞ and Mc X are the fluctuation operators around 1 Φj ¼ pffiffiffi iθϕ¯ ð Þ ð Þ the bounce solution and around the false vacuum, respec- bounce e r ; 2:7 2 tively. Here, wðXÞ ¼ 1 for Dirac fermions and Faddeev- Popov ghosts, and wðXÞ ¼ −1=2 for the other bosonic with our choice of the gauge fixing function, without loss of fields. The fluctuation operator is defined as second θ ϕ¯ generality. Here, is a real parameter. The function, , derivatives of the action: obeys δ2SðXÞ 3 MðXÞδð4Þðx − yÞ¼ ; ð2:15Þ ∂2ϕ¯ ðrÞþ ∂ ϕ¯ ðrÞ − λϕ¯ 3ðrÞ¼0; ð2:8Þ δXðxÞδXðyÞ r r r where the brackets indicate the evaluation around the ∂ ϕ¯ ð0Þ¼0 ϕ¯ ð∞Þ¼0 cðXÞ with boundary conditions r and . For a bounce solution. In addition, M can be obtained from λ negative , we have a series of Fubini-Lipatov instanton MðXÞ with replacing ϕ¯ by the vacuum expectation value at solutions [58,59]: the false vacuum, i.e., ϕ¯ → 0. jλj −1 Since the Faddeev-Popov ghosts do not couple to the ϕ¯ ð Þ¼ϕ¯ 1 þ ϕ¯ 2 2 ð Þ r C Cr ; 2:9 Higgs boson with the present choice of the gauge fixing 8 ð ¯Þ function, Mðc;c¯Þ ¼ Mc c;c . Thus, we have which is parametrized by ϕ¯ (i.e., the field value at the C Aðc;c¯Þ ¼ 1: ð2:16Þ center of the bounce). We also define R, which gives the size of the bounce, as sffiffiffiffiffi B. Functional determinant 8 ≡ ϕ−1 ð Þ For the evaluation of the functional determinants, we first R jλj C : 2:10 decompose fluctuations into partial waves, making use of the Oð4Þ symmetry of the bounce [15]. The basis is The action of the bounce is given by ðΩÞ constructed from YJ;mA;mB , the hyperspherical function on S3 with Ω being a coordinate on S3. The decomposition 8π2 ¼ B ¼ : ð2:11Þ of each fluctuation is given in Appendix A. Here, J 3jλj 0; 1=2; 1; … is a non-negative half integer that labels the total angular momentum in four dimensions, and mA ϕ¯ Notice that the tree level action is independent of C owing and mB are the azimuthal quantum numbers for the A-spin ð4Þ ≃ ð2Þ ð2Þ to the classical scale invariance. and the B-spin of so su A × su B, respectively. Once the bounce solution is obtained, we may integrate The four-dimensional Laplacian operator acts on the hyper- over the fluctuation around it. We expand Φ as spherical function as
1 2 θ ¯ L Φ ¼ pffiffiffi ei ðϕ þ h þ iφÞ; ð2:12Þ −∂2Y ðΩÞ¼ Y ðΩÞ; ð2:17Þ 2 J;mA;mB r2 J;mA;mB with 3 ∂ In the non-Abelian case, one of μ in Eq. (2.6) is replaced by pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi the covariant derivative. The interaction of the ghosts with the ≡ 4 ð þ 1Þ ð Þ gauge field does not affect the following discussion. L J J : 2:18
116012-3 SO CHIGUSA, TAKEO MOROI, and YUTARO SHOJI PHYS. REV. D 97, 116012 (2018)
ðΩÞ In addition, YJ;mA;mB is normalized as Here, Z rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1 κ dΩjY j2 ¼ 1: ð2:19Þ ¼ − 1 − 1 − 8 ð Þ J;mA;mB zκ 2 jλj ; 2:28 With the above expansion, the functional determinants sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1 2 can be expressed as ¼ − 1 − 1 − 8 g ð Þ zg 2 jλj ; 2:29 ðXÞ ð Þ Y∞ ðXÞ n DetM X DetM J ¼ J ; ð2:20Þ y ðXÞ ðXÞ pffiffiffi Mc Mc zy ¼ i ; ð2:30Þ Det J¼0 Det J λ MðXÞ where J is the fluctuation operator for each partial where g is the gauge coupling constant, and ΓðzÞ is the ðψÞ ¼ð2 þ 1Þð2 þ 2Þ wave labeled by J. In addition, nJ J J gamma function. ðXÞ ¼ð2 þ 1Þ2 for fermions, and nJ J for the others. The explicit forms of the fluctuation operators are shown in C. Zero modes Appendix A. Using a theorem [29,60–63], the ratio of the In the calculation of the decay rate of the EW vacuum functional determinants can be calculated as with the present setup, there show up zero modes in association with dilatation, translation, and global trans- ð Þ MðXÞ Ψ Ψ −1 M h Det J Det Det formation of the bounce solution. Consequently, 0 , ¼ lim lim →0 ; ð2:21Þ ð Þ ðAμ;φÞ ðXÞ →∞ ˆ r ˆ h Mc r Ψ Ψ M1 2 and M0 have zero eigenvalues. Their determi- Det J Det Det = nants vanish as shown in Eqs. (2.24) and (2.27) [see also ˆ ˆ ˆ where Ψ ¼ðΨ1; Ψ2; …Þ and Ψ ¼ðΨ1; Ψ2; …Þ are sets of Eq. (A80)]; a naive inclusion of those results gives a independent solutions of divergent behavior of the decay rate, which requires a careful treatment of the zero modes. In the present case, we ðXÞ can consider the effect of each partial wave (labeled by J) M Ψi ¼ 0; ð2:22Þ J separately. Thus, in this subsection, we consider the case ðXÞ cðXÞ ˆ where the fluctuation operator M for a certain value of J M Ψi ¼ 0; ð2:23Þ J J has a zero eigenvalue and discuss how to take account of its ð Þ effect. Because only bosons have zero modes in the and are regular at r ¼ 0. Notice that, when M X is an J MðXÞ n × n object, there are n independent solutions that are calculation of vacuum decay rates, J is considered to be a fluctuation operator of a bosonic field. regular at r ¼ 0. Since Ψ and Ψˆ obey the same linear i i We first decompose the fluctuation of the bosonic field differential equation at r → ∞, the ratio of the two (denoted as X) into eigenstates of the angular momentum: determinants converges for each J. X With a Fubini-Lipatov instanton, we can calculate the X ∋ c G ðrÞY ðΩÞ; ð2:31Þ ratio analytically, as first pointed out in [24]. For the i i J;mA;mB i convenience of readers, we give the details of the calcu- lation in Appendix A. The ratios are given by where ci is the expansion coefficient while Gi obeys ð Þ DetM h 2Jð2J − 1Þ MðXÞG ¼ ω G ð Þ J ¼ ; ð2:24Þ J i i i; 2:32 McðhÞ ð2J þ 3Þð2J þ 2Þ Det J with ωi being the eigenvalue. We normalize them as MðσÞ Γð2 þ 1ÞΓð2 þ 2Þ Det J J J δ ¼ ; ð2:25Þ ij cðσÞ Γð2 þ 1 − ÞΓð2 þ 2 þ Þ hG jG i¼ ; ð : Þ DetM J zκ J zκ i j N 2 2 33 J i ðψÞ DetM ½Γð2J þ 2Þ 2 2 where the inner product is defined by J ¼ ; ð : Þ cðψÞ Γð2 þ 2 − ÞΓð2 þ 2 þ Þ 2 26 Z DetM J zy J zy J † hG jG i¼ drr3G ðrÞG ðrÞ: ð : Þ i j i j 2 34 ðAμ;φÞ DetM J Γð2J þ 1ÞΓð2J þ 2Þ 3 J ¼ : cðAμ;φÞ þ 1 Γð2 þ 1 − ÞΓð2 þ 2 þ Þ N ’ DetM J J zg J zg We leave i s unspecified since, as we see below, the final J result is independent of them. We denote the zero mode as ð : Þ 2 27 G0 (and hence ω0 ¼ 0).
116012-4 DECAY RATE OF ELECTROWEAK VACUUM IN THE … PHYS. REV. D 97, 116012 (2018)
˜ ˜ When there is a zero mode, the saddle point approxi- hG0jρjG0i¼2π: ð2:40Þ mation for the path integral breaks down and we need to go back to the original path integral. The path integral is Then, we have defined as the integration over the coefficient ci, and the Z ð Þ X Y ð Þ functional determinant of M originates from the follow- dci −1c c hG jðM X þνρÞjG i J pffiffiffiffiffiffi e 2 j k j J k ing integral: 2πN i sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffii Z Y ð Þ hG˜ jG˜ i 1 Y 1 dci −1c c hG jM X jG i 0 0 pffiffiffiffiffiffi e 2 j k j J k ; ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ; ð2:41Þ 2πN 2π ν þ Oðν2Þ ω þ OðνÞ i i i≠0 i where the summations over the indices j and k are implicit. which gives Applying the saddle point approximation for the modes sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi other than the zero mode, we get hG˜ jG˜ iY 1 0 0 pffiffiffiffiffi Z Y Z Y 2π ω 1 ðXÞ 1 i dci − c c hG jM jG i dc0 i≠0 pffiffiffiffiffiffi e 2 j k j J k ¼ pffiffiffiffiffiffi pffiffiffiffiffi : Z 2πN 2πN ω pffiffiffi Y ð Þ i i 0 i≠0 i dci −1c c hG jðM X þνρÞjG i ¼ lim ν pffiffiffiffiffiffi e 2 j k j J k : ð2:42Þ ν→0 2πN ð2:35Þ i i
ðXÞ The integration in the above expression is nothing but Notice that M , G and ω may depend on c0. J i i DetðMðXÞ þ νρÞ, and can be evaluated with the use of We are interested in the case where there exists a MðXÞ 1 1 symmetry (at least at the classical level) and the Eq. (2.21).If J is a × object, for example, we obtain Lagrangian is invariant under a transformation, which ð Þ we parametrize by z. The zero mode is in association with Det½M X þ νρ Ψˇ ðrÞ J ¼ ν lim þ Oðν2Þ; ð2:43Þ such a symmetry. Then, the transformation of the bounce ðXÞ →∞ ˆ DetMc r ΨðrÞ solution with z → z þ δz can be seen as a shift of X as J where → þ δ G˜ þ Oðδ 2Þ ð Þ X X z 0YJ;mA;mB z ; 2:36 MðXÞΨˇ ð Þ¼−ρð ÞΨð Þ ð Þ ˜ J r r r ; 2:44 where the function G0 is proportional to G0. The integration over c0 can thus be regarded as the integration over the while the functions Ψ and Ψˆ obey Eqs. (2.22) and (2.23), collective coordinate z: respectively. Then, we interpret the functional determinant sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi of the fluctuation operator with the zero eigenvalue as ˜ ˜ dc0 hG0jG0i pffiffiffiffiffiffi → ð Þ ð Þ−1 2 Z dz; 2:37 M X = Ψˇ ð Þ −1=2 2πN 0 2π Det J r → dz lim →∞ : ð2:45Þ McðXÞ r Ψˆ ð Þ Det J r and hence sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi The above argument can be applied to the zero modes of Z Z our interest. In Appendix B, we obtain the following Y ð Þ hG˜ jG˜ iY 1 dci −1c c hG jM X jG i 0 0 pffiffiffiffiffiffi e 2 j k j J k → dz pffiffiffiffiffi: replacements to take care of the dilatational, translational, 2πN 2π ω i i i≠0 i and gauge zero modes: ð2:38Þ Z sffiffiffiffiffiffiffiffi MðhÞ −1=2 16π Det 0 → d ln R ; ð2:46Þ Next, we discuss how we evaluate the integrand of the cðhÞ jλj DetM0 right-hand side of Eq. (2.38). To omit the zero eigenvalue ! from the functional determinant, we introduce a regulator to ð Þ −2 M h 2 Det 1 2 32π V4 the fluctuation operator: = → D ; ð2:47Þ McðhÞ jλj R4 Det 1=2 MðXÞ → MðXÞ þ νρð Þ ð Þ J J r ; 2:39 ! sffiffiffiffiffiffiffiffi ðAμ;φÞ −1=2 Z DetM0 16π where ν is a small positive number, and ρðrÞ is an arbitrary → dθ : ð2:48Þ McðAμ;φÞ jλj function that satisfies Det 0
116012-5 SO CHIGUSA, TAKEO MOROI, and YUTARO SHOJI PHYS. REV. D 97, 116012 (2018) κ 1 2 κ D. Renormalization ½ AðσÞ ¼ − þ þ ε ln εσ 2 ln σ After taking the product over J, we have a UV diver- jλj εσ εσ 3jλj gence and thus we need to renormalize the result. In this 1 − S ð ÞþOðε Þ ð Þ paper, we use the MS scheme, which is based on the 2 σ zκ σ ; 2:50 dimensional regularization. In this subsection, we explain 2 2 5 1 2 þ 2jλj how the divergences can be subtracted using counterterms ½ AðψÞ ¼ y þ þ y ε ln εψ 2 ln ψ in the MS scheme. jλj εψ εψ 3 jλj Since the dimensional regularization cannot be directly þ Sψ ðz ÞþOðεψ Þ; ð2:51Þ used in this evaluation, we first regularize the result by y using the angular momentum expansion as where the functions, Sσ and Sψ , are given in Eqs. (C13) and X∞ ðXÞ MðXÞ (C16), respectively. In addition, we define ðXÞ ðXÞ nJ Det J ½ln A ε ≡ w ln : ð2:49Þ Z X ð1 þ ε Þ2J McðXÞ 1 J¼0 X Det J ðhÞ 0ðhÞ A ¼ V4 d ln R A ; ð2:52Þ D R4 We call this regularization “angular momentum regulari- Z zation.” Here, ε is a positive number, which will be taken X AðAμ;φÞ ¼ dθA0ðAμ;φÞ: ð2:53Þ to be zero at the end of the calculation. Since the divergence is at most a power of J, the regularized sum converges. In Appendix C, we calculate the sum analytically, and obtain Then, the primed quantities are given by
1 jλj jλj X∞ ð2 þ 1Þ2 2 ð2 − 1Þ 0ðhÞ J J J ½ln A ε ¼ − ln þ 4 ln þ ln h 2 16π 32π 2J ð2 þ 3Þð2 þ 2Þ ð1 þ εhÞ J J J¼1 1 2 3 5 π 5 jλj ¼ 3 þ − ε − − 3γ − 6 A þ − þ Oðε Þ; ð : Þ ε2 ε ln h 4 E ln G 2 ln 3 2 ln 8 h 2 54 h h ∞ 1 jλj X ð2J þ 1Þ2 J Γð2J þ 1ÞΓð2J þ 2Þ 3 0ðAμ;φÞ ½ln A ε ¼ − ln þ ln A 2 16π ð1 þ ε Þ2J J þ 1 Γð2J þ 1 − z ÞΓð2J þ 2 þ z Þ J¼1=2 A g g 3g2 1 2 1 g4 3 1 1 jλj ¼ − − 1 þ − þ ε þ − γ − 2 A − jλj ε2 ε 3 jλj2 ln A 4 3 E ln G 2 ln 8π A A 3 3 − S ð Þ − Γð1 − ÞΓð2 þ ÞþOðε Þ ð Þ 2 σ zg 2 ln zg zg A ; 2:55
where AG ≃ 1.282 is the Glaisher number. where ð Þ Next, we relate the above results with those based on the δMðXÞ ¼ MðXÞ − Mc X ; ð2:58Þ dimensional regularization in D dimension, which contain ε¯ δMðXÞ ½ the regularization parameter, D, defined as and J is defined similarly. Here, OðδÞ indicates 1 2 hthe expansioni up to δ. The most important pointh is thati ð Þ ð Þ ¼ þ ln 4π − γE; ð2:56Þ DetM X DetM X ε¯ 4 − ln ð Þ has the same divergence as ln ð Þ D D b X b X DetM div DetM δMðXÞ with γE being the Euler’s constant. We convert the results does when does not have a derivative operator. based on two different regularizations by calculating the The first line of Eq. (2.57) can be calculated by directly following quantity: evaluating the traces in a momentum space with using ðXÞ −1 the dimensional regularization; the result is denoted as ½ AðXÞ ¼ ðXÞ ½ðMc Þ δMðXÞ ð Þ ln div w Tr ½ln A X . On the other hand, the second line of div;ε¯D ð Þ w X ð Þ −1 ð Þ −1 Eq. (2.57) can be evaluated as − ½ðMc X Þ δMðXÞðMc X Þ δMðXÞ 2 Tr ½McðXÞ þ δMðXÞ X∞ cðXÞ ðXÞ Det J J ˆ −1 ð1Þ ð Þ Det½M þ δM ln ¼ Tr½Ψ Ψ ¼ ðXÞ X J J McðXÞ w nJ ln ðXÞ ; Det J OðδM2Þ Mc J J¼0 Det J OðδM2Þ J 1 þ ½Ψˆ −1Ψð2Þ − ½Ψˆ −1Ψð1ÞΨˆ −1Ψð1Þ ð Þ ð2:57Þ Tr 2 Tr ; 2:59
116012-6 DECAY RATE OF ELECTROWEAK VACUUM IN THE … PHYS. REV. D 97, 116012 (2018)
1 ε¯ where Subtracting = D, we obtain the renormalized prefactor in the MS scheme. McðXÞΨðpÞ ¼ −δMðXÞΨðp−1Þ ð Þ J J ; 2:60 E. Dilatational zero mode for p ¼ 1 and 2, and Ψð0Þ ¼ Ψˆ . Then, we calculate In the calculation of the decay rate of the EW vacuum, ðXÞ we have an integral over R in association with the classical ½ln A ε div; X scale invariance, as we saw in Eq. (2.46). So far, we have ð Þ ð Þ ð Þ X n X Det½Mc X þ δM X performed a one-loop calculation of the decay rate, based ≡ wðXÞ J ln J J : ð1 þ ε Þ2J cðXÞ on which the decay rate is found to behave as J X DetM OðδM2Þ J J Z ð1Þ ð2:61Þ 1 8π2 8π2β γðone-loopÞ ∝ d ln R exp − − λ ln μR ; R4 3jλðμÞj 3jλðμÞj2 We relate the results based on two regularizations by ð Þ replacing 2:67
ð1Þ ½ln AðXÞ → ½ln AðXÞ : ð2:62Þ where β is the one-loop β function of λ. (Here, we only div;εX div;ε¯D λ show the R and μ dependencies of the one-loop correc- The expressions of ½ln AðXÞ and ½ln AðXÞ for each tions.) Thus, using the purely one-loop result, the integral div;εX div;ε¯D field are given in Appendix D, where we further simplify does not converge. the relation so that the left-hand side only includes terms We expect, however, that the integration can converge that are divergent at the limit of εX → 0. We summarize the once higher order effects are properly included. To see the results below: detail of the path integral over the dilatational zero mode, (i) Scalar field: let us denote the decay rate as Z 1 2 κ −B ð Þ e eff þ þ ln εσ γ full ¼ ð Þ ε2 ε 3jλj d ln R 4 ; 2:68 σ σ R κ 1 μR → −1 − þ 1 þ γ þ ln ; ð2:63Þ B 3jλj 2ε¯ E 2 where eff fully takes account of all the effects of higher D order loops. B (ii) Higgs field: In order to discuss how eff should behave, it is instructive to rescale the coordinate variable as 1 2 1 μR rffiffiffiffiffi þ − ln ε → þ γ þ ln ; ð2:64Þ 2 ε h 2ε¯ E 2 jλj xμ ε h D ¯ h x˜μ ≡ ϕ xμ ¼ ; ð2:69Þ 8 C R (iii) Fermion: as well as the fields as 2 5 1 2 þ 2jλj þ þ y ε 2 ln ψ ˜ ¯ −1 εψ εψ 3 jλj Φ ≡ ϕ Φ; ð2:70Þ C y2 1 μR → − þ 1 þ γ þ ln σ˜ ≡ ϕ¯ −1σ ð Þ 3jλj 2ε¯ E 2 C ; 2:71 D 2 1 25 μ R ˜ ¯ −1 − þ þ γ þ ð Þ Aμ ≡ ϕ Aμ; ð : Þ 3 2ε¯ 4 E ln 2 ; 2:65 C 2 72 D ˜ ≡ ϕ¯ −1 ð Þ (iv) Gauge and NG fields: c C c; 2:73 3g2 1 2 1 g4 ¯˜ ≡ ϕ¯ −1 ¯ ð Þ − 1 þ þ þ ε c C c; 2:74 jλj ε2 ε 3 jλj2 ln A A A 2 4 jλj −1=4 1 2 1 μ ¯ −3=2 → − þ g þ g þ 1 þ γ þ R ψ˜ ≡ ϕ ψ: ð2:75Þ 3 jλj jλj2 2ε¯ E ln 2 8 C D 4 31 þ 1 þ g − π2 ð Þ 2 : 2:66 Using the rescaled fields, all the explicit scales disappear jλj 3 from the action as a result of scale invariance:
116012-7 SO CHIGUSA, TAKEO MOROI, and YUTARO SHOJI PHYS. REV. D 97, 116012 (2018) Z Z 1 1 κ 4 4 ˜ pyffiffiffiffiffi pgffiffiffiffiffi If the effects of higher order loops are fully taken into d xL ¼ d x˜ L ; ; ; ð2:76Þ B μ ℏ ℏ˜ jλj jλj jλj account, eff should be independent of because the decay rate is a physical quantity; in such a case, we may choose where L is the total Lagrangian, any value of the renormalization scale μ. In the perturbative calculation, the μ dependence is expected to cancel out jλj order by order; as shown in Eq. (2.67), we can explicitly see ℏ˜ ¼ ℏ ð Þ 8 ; 2:77 the cancellation of the μ dependence at the one-loop level [64]. In our calculation so far, however, we only have the and L˜ includes canonically normalized rescaled fields one-loop result, in which μ dependence remains. As κ pyffiffiffiffi μ and depends only on the combinations jλj, and indicated in Eq. (2.81), the dependence shows up in jλj the form of lnpμR with p ¼ 1; 2; ….Ifj ln μRj ≫ 1, the pgffiffiffiffi. In addition, the rescaled bounce solution is given by jλj logarithmic terms from higher order loops may become comparable to the tree-level bounce action and the pertur- 1 bative calculation breaks down. In order to make our one- ð Þ 2 : 2:78 loop result reliable, we should take μ ∼ Oð1=RÞ, i.e., we 1 þ x˜ use R-dependent renormalization scale μ.4 With such a choice of μ (as well as with the use of coupling constants Based on L˜ and ℏ˜, we expect: κ y g evaluated at the renormalization scale μ), the integration (i) Only positive powers of , pffiffiffiffi and pffiffiffiffi appear in jλj jλj jλj over the size of the bounce is dominated only by the region the decay rate since there is no singularity when any where jλð1=RÞj becomes largest. In the case of the SM, the of these goes to zero. In particular, they cannot be in integration over the size of the bounce converges with this a logarithmic function. prescription as we show in the following section. (ii) When we renormalize divergences using dimen- sional regularization, we introduce a renormaliza- tion scale μ˜. It is always in a logarithmic func- F. Final result tion and is related to the original renormalization Here, we summarize the results obtained in the previous scale as subsections and Appendixes A–E. The decay rate with a resummation of important logarithmic terms is given by μ˜ ¼ μ ð Þ Z R: 2:79 1 γ ¼ dlnR ½A0ðhÞAðσÞAðψÞAðAμ;φÞe−B ; ð2:82Þ R4 MS;μ∼1=R (iii) In subtracting zero modes associated with transformations of Eq. (2.78), the result should be κ pyffiffiffiffi pgffiffiffiffi where again a polynomial of jλj, and . Notice that, jλpj ffiffiffiffiffiffiffiffi jλj ˜ 3 5 π 5 jλj μR for each zero mode, we have 1=ℏ since Eq. (2.37) ½ln A0ðhÞ ¼ − − 6 ln A þ ln − ln þ 3 ln ; MS 4 G 2 3 2 8 2 implies ð2:83Þ sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ˜ ˜ dc0 hG0jG0i 1 κ κ2 μ pffiffiffiffiffiffiffiffi ¼ dz : ð2:80Þ ½ AðσÞ ¼ − S ð Þþ þ 1 þ γ þ R ˜ ˜ ln σ zκ 2 E ln ; 2πℏN 0 2πℏ MS 2 jλj 3jλj 2 ð2:84Þ (iv) Quantum corrections have ℏ˜ l−1 at the lth loop ℏ˜ since the loop expansion is equivalent to the y4 μR ½lnAðψÞ ¼ − 1 þ γ þ ln expansion. MS 3jλj2 E 2 Based on the above arguments, B is expected to be eff 2 expressed as 2y 25 μR − þ γE þ ln þ Sψ ðzyÞ; ð2:85Þ 3jλj 4 2 8π2 jλðμÞj X∞ jλðμÞj l−1 B ¼ þ nzero þ eff ln ½ AðAμ;φÞ ¼ V þ½ A0ðAμ;φÞ ð Þ 3jλðμÞj 2 8 8 ln MS ln G ln MS; 2:86 l¼1 κðμÞ yðμÞ gðμÞ × Pl ; pffiffiffiffiffiffiffiffiffiffiffi ; pffiffiffiffiffiffiffiffiffiffiffi ; ln μR ; ð2:81Þ 4 jλðμÞj jλðμÞj jλðμÞj This is equivalent to summing over large logarithmic terms appearing in higher loop corrections if we work with a fixed μ. Since we have calculated the decay rate at the one-loop level, it is P l where l is the contribution at the -loop level, and nzero is preferable to use, at least, the two-loop β-function to include the the number of zero modes. next-to-leading logarithmic corrections.
116012-8 DECAY RATE OF ELECTROWEAK VACUUM IN THE … PHYS. REV. D 97, 116012 (2018)
L ∋ ¯ c þ ð Þ with Yukawa ytHqLtR H:c:; 3:2 2 4 1 2g g μR where qL is the left-handed third generation quark doublet, ½ A0ðAμ;φÞ ¼ þ þ 1 þ γ þ ln 2 E ln c MS 3 jλj jλj 2 tR is the right-handed antitop quark, and yt is the top Yukawa coupling constant. 4 31 1 1 − g − π2 − − γ − 2 Assuming that λ < 0, the bounce solution for the SM is 2 E ln AG jλj 3 4 3 given by 1 jλj 1 3 − ln þ ln π − SσðzgÞ 1 0 2 8 2 2 θaσa Hj ¼ pffiffiffi ei ; ð3:3Þ 3 bounce 2 ϕ¯ ðrÞ − Γð1 − ÞΓð2 þ Þ ð Þ 2 ln zg zg : 2:87 where σa is the Pauli matrices and function ϕ¯ is given by ϕ¯ Here, VG is the volume of the group space generated by the Eq. (2.9). In particular, remember that contains a free broken generators. The definitions of SσðzÞ, Sψ ðzÞ can be parameter, which we choose R, because of the classical found in Appendix C. Here, we note that the analytic results scale invariance. for the scalar, Higgs, and fermion contributions were first The results given in the previous section can be easily given in [24] with different expression. We emphasize that applied to the case of the SM. Taking account of the effects the final result does not depend on the gauge parameter, ξ, of the (physical) Higgs boson, top quark, and weak bosons 6 and hence our result is gauge invariant. The above result is (as well as NG bosons), the decay rate of the EW vacuum also applicable to the case where the Uð1Þ symmetry is not in the SM can be written in the following form: gauged as explained in Appendix E. Z 1 We have also derived fitting formulas of the functions γ ¼ ½A0ðhÞAðtÞAðW;Z;φÞ −B ð Þ d ln R 4 e : 3:4 necessary for the calculation of the decay rate; the result is R MS given in Appendix F. The fitting formulas are particularly useful for the numerical calculation of the decay rate. In As we have mentioned, the relevant renormalization scale μ ∼ ð1 Þ addition, a C þþ package to study the electroweak of the integrand is O =R ; in the following numerical μ ¼ 1 λðμÞ vacuum stability (ELVAS) is available at [30], which is analysis, we take =R unless otherwise stated. If also applicable to various models with (approximate) is positive, there is no bounce solution; the integrand is classical scale invariance. taken to be zero in such a case. In addition, since we neglect the mass term in the Higgs potential, 1=R should be much larger than the EW scale. This condition is automatically III. DECAY RATE OF THE EW VACUUM satisfied in the present analysis because λ becomes negative IN THE SM at the scale much higher than the electroweak scale. 0ðhÞ A. Decay rate The Higgs contribution A is given in Eq. (2.83), while the top-quark contribution is given by Now, we are in a position to discuss the decay rate of the EW vacuum in the SM. As we have discussed, the decay of ðtÞ ðψÞ ½ln A ¼ 3½ln A j → ; ð3:5Þ the EW vacuum is induced by the bounce configuration MS MS y yt whose energy scale is much higher than the EW scale. Thus, we approximate the Higgs potential as5 where the factor of 3 is the color factor. ð2Þ As for the gauge contributions, we have SU L × Uð1Þ =Uð1Þ broken symmetries, instead of Uð1Þ in ð Þ¼λð † Þ2 ð Þ Y EM V H H H ; 3:1 our previous example. Thus, we have a different volume of the group space, VG. To calculate VG, we first expand H where H is the Higgs doublet in the SM and λ is the Higgs around the bounce solution with θa ¼ 0 as λ quartic coupling constant. Notice that depends on the renormalization scale μ; in the SM, λ becomes negative 1 iðφ1 − iφ2Þ μ ∼1010 H ¼ pffiffiffi : ð3:6Þ when is above GeV with the best-fit values of the 2 ϕ¯ − iφ3 SM parameters. In addition, the relevant part of the Yukawa couplings are given by Here, φ1 and φ2 are NG bosons absorbed by charged W-bosons while φ3 is that by Z-boson. With the change of 5 We assume that the Higgs potential given in Eq. (3.1) is θa, the NG modes are transformed as applicable at a high scale. In particular, we assume that the effect of Planck suppressed operators, which may arise from the effect of quantum gravity, is negligible. For the discussion about the 6We checked that the effect of the bottom quark is numerically effect of Planck suppressed operators, see [65–73]. unimportant.
116012-9 SO CHIGUSA, TAKEO MOROI, and YUTARO SHOJI PHYS. REV. D 97, 116012 (2018)
a a a ˜ 2 φ → φ þ θ GGY0;0;0 þ Oðθ Þ; ð3:7Þ where pffiffiffiffiffiffiffi ˜ 2 ¯ GG ¼ 2π ϕ: ð3:8Þ The integration over the zero modes in association with the gauge transformation of the bounce solution can be θa replaced by the integrationZ over the parameter : 3 2 d θ ¼ 2π ≡ VSUð2Þ; ð3:9Þ with the above definition of θa. Then, following the argument given in Appendix B, the gauge contribution is evaluated as
½ AðW;Z;φÞ ¼ V þ 2½ A0ðAμ;φÞ ln MS ln SUð2Þ ln MS g→gW þ½ln A0ðAμ;φÞ j ; ð3:10Þ MS g→gZ
A0ðAμ;φÞ where is given in Eq. (2.87), and FIG. 1. Top: The RG evolution of the SM coupling constants as pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi μ λ 0 2 þ 2 functions of (in units of GeV). The dashed line indicates < . g2 gY g2 The black dotted line shows the scale where ϕ¯ ¼ M , taking gW ¼ ;gZ ¼ ; ð3:11Þ C Pl 2 2 μ ¼ 1=R. The horizontal axis is common with the bottom panel. Bottom: The integrand of the decay rate with μ ¼ 1=R, taking the with g2 and g being the gauge coupling constants of Y central values for the SM parameters. In the shaded region, λ is SUð2Þ and Uð1Þ , respectively. L Y positive and the integrand is zero. B. Numerical results In order to understand the μ dependence of λ, let us show Now, let us evaluate the decay rate of the EW vacuum in one-loop RG equations of λ and y (although, in our the SM. The decay rate of the EW vacuum is very sensitive t numerical calculation, we use RG equations including to the coupling constants in the SM. In our numerical two- and three-loop effects and the contribution from the analysis, we use [74]: bottom and the tau Yukawa couplings):
mh ¼ 125.09 0.24; ð3:12Þ dλ g2 þ g2 g2 ¼ 173 1 0 6 ð Þ 16π2 ¼ 12λ 2λ þ 2 − Y 2 − 2 mt . . ; 3:13 μ yt 4 2 d ln one-loop α ð Þ¼0 1181 0 0011 ð Þ s mZ . . ; 3:14 2 þ 2 2 2 2 − 6 4 þ 6 gY g2 þ 12 g2 yt 4 4 ; where mh and mt are the Higgs mass and the top mass, α respectively, while s is the strong coupling constant. ð3:15Þ Following [75], the gauge couplings, the top Yukawa coupling, and the Higgs quartic coupling are determined μ ¼ dy 9 9 17 at mt; the calculations are done in the on-shell scheme 16π2 t ¼ 2 − 8 2 − 2 − 2 μ yt 2 yt g3 4 g2 12 gY : at next-to-next-to leading order precision. In addition, we d ln one-loop use three-loop QCD and one-loop QED β functions ð3:16Þ [76–78] together with values in [74] in order to determine the bottom and the tau Yukawa couplings at μ ¼ m . t 4 λ First, we show the RG evolution of the SM coupling At a low energy scale, the term proportional to yt drives to constants in Fig. 1. We use mainly three-loop β functions a negative value. As the scale increases, yt decreases while λ summarized in [75] and the central values for the SM gY increases, which brings back to a positive value. Notice ¯ that λ is bounded from below in the SM. parameters. The black dotted line indicates where ϕC ≃ 2 4 1018 We show the integrand of γ in the bottom panel of Fig. 1, reaches the Planck scale MPl . × GeV. We also show the running above the Planck scale, assuming together with that of there are no significant corrections from gravity. For Z 1 1010 GeV ≲ μ ≲ 1030 GeV, λ becomes negative; for such γ ¼ −B ð Þ tree d ln R 4 e : 3:17 a region, we use a dashed line to indicate λ < 0. R
116012-10 DECAY RATE OF ELECTROWEAK VACUUM IN THE … PHYS. REV. D 97, 116012 (2018)
δBð1Þ for each X, where eff is the one-loop contribu- B ½ AðXÞ tion to eff, and ln MS is a contribution from particle X; the region that does not satisfy these conditions is excluded from the integration.7 By numerically integrating over R, we obtain8
½γ 3 ¼−582þ40þ184þ144þ2 ð Þ log10 Pl × Gyr Gpc −45−329−218−1 ; 3:18
3 þ40þ183þ145þ2 log10½γ∞ × Gyr Gpc ¼−580−44−328−218−1 ; ð3:19Þ
where the first, second, third, and fourth errors are due to the Higgs mass, the top mass, the strong coupling constant, and the renormalization scale, respectively. (In order to estimate the uncertainty due to the choice of the renorm- alization scale, we vary the renormalization scale from 1=2R to 2=R.) Currently, the largest error comes from the uncertainty of the top mass. With a better understanding of the top quark mass at the future LHC experiment [79–85], or even with at future eþe− colliders [86], a more accurate FIG. 2. Top: The integrand of the decay rate with the central determination of the decay rate will become possible. One values of SM parameters. The solid line corresponds to a result at can see that the predicted decay rate per unit volume is −4 the one-loop level and the dashed one corresponds to that at the extremely small, in particular, compared with H0 ≃ 3 3 tree level. The horizontal axis is common with the bottom figure. 10 Gyr Gpc (with H0 being the expansion rate of the ϕ¯ ¼ We show C MPl with the vertical dotted line. Bottom: The present Universe). Such a small decay rate is harmless for size of each quantum correction. The dashed line corresponds to realizing the present Universe observed.9 δBð1Þ eff . In Fig. 3, we show the decay rate in mh vs mt plane. In 4 the red region, γ becomes larger than H0, which we call the They are also shown in a linear scale in the top panel unstable region. In the yellow region, the EW vacuum is 4 of Fig. 2. metastable, meaning that 0 < γ