betaFeedback

WorksPhysics and Astronomy

Improved indirect limits on charm and bottom quark EDMs

Abstract

We derive indirect limits on the charm and bottom quark electric dipole moments (EDMs) from paramagnetic AMO and neutron EDM experiments. The charm and bottom quark EDMs generate CPCP-odd photon-gluon operators and light quark EDMs at the cc- and bb-quark mass thresholds. These CPCP-odd operators induce the CPCP-odd semi-leptonic operator CSC_S and the neutron EDM below the QCD scale that are probed by the paramagnetic and neutron EDM experiments, respectively. The bound from CSC_S is ∣dc∣<1.3×10−20 e cm\vert d_c \vert < 1.3\times 10^{-20}\,e\,\mathrm{cm} for the charm quark and ∣db∣<7.6×10−19 e cm\vert d_b \vert < 7.6\times 10^{-19}\,e\,\mathrm{cm} for the bottom quark, with its uncertainty estimated as 10%. The neutron EDM provides a stronger bound, ∣dc∣<6×10−22 e cm\vert d_c \vert < 6\times 10^{-22}\,e\,\mathrm{cm} and ∣db∣<2×10−20 e cm\vert d_b \vert < 2\times 10^{-20}\,e\,\mathrm{cm}, though with a larger hadronic uncertainty.

Introduction

Searches for electric dipole moments (EDMs) play a crucial role in probing physics beyond the Standard Model (SM). Given the smallness of the SM contribution from the Cabbibo-Kobayashi-Maskawa phase, EDMs of elementary particles are ideal observables to look for new physics that violates CPCP symmetry (see [1] for a review).

EDMs of elementary particles may be classified into three categories: lepton EDMs, light quark EDMs and heavy quark EDMs, where “light” and “heavy” are in comparison with the QCD scale. Among the lepton EDMs, the electron EDM is tested by paramagnetic EDM experiments such as the ACME experiments [2], while the muon and tau EDMs can be probed indirectly through paramagnetic and diamagnetic EDM experiments [3, 4, 5]. Motivated by the muon g−2g-2 anomaly [6, 7], the muon EDM has been and will be measured by the storage ring experiments as well [8, 9, 10, 11, 12]. Hadronic EDMs can be induced by the EDMs of light quarks, as well as by other operators such as chromo-EDMs (CEDMs) and three-gluon CPCP-odd operators. The light quark EDMs, in particular the up and down quark EDMs, contribute to the neutron EDM and are constrained by the neutron EDM experiments [13, 14]. Discussion of the strange quark contribution was typically focused on the color EDM contributions [15, 16], while no reliable estimate of the neutron EDM induced by the strange quark exists at this point. Among the heavy quark EDMs, the top quark EDM is arguably most deeply investigated. In particular, due to its large Yukawa coupling, the top quark EDM generates light fermion EDMs through diagrams with an intermediate Higgs, which allows one to indirectly constrain the top quark EDM [17, 18]. The light quark CEDMs directly contribute to the neutron EDM, while the heavy quark CEDMs generate, e.g., the three-gluon Weinberg operator at the threshold, which induce the neutron EDM after the confinement [19, 20, 21, 22, 23, 24].

Direct measurements of the heavy quark EDMs are difficult due to their short lifetimes. The current strongest direct bound is from e+e−→qqˉe^{+}e^{-} \to q\bar{q} at LEP and is only of the order of 10−17 e cm10^{-17}\,e\,\mathrm{cm} [25].

Recently, an LHC based experiment is proposed that aims at directly measuring charm baryon dipole moments [26, 27, 28, 29, 30], potentially improving the direct bounds on the heavy quark EDMs. Given this situation, in this paper, we address indirect limits on the heavy quark EDMs. The heavy quark EDMs induce several CPCP-odd operators below the heavy quark mass scale. These operators generate observable signals in neutron/atomic/molecule EDM experiments, which allows us to impose indirect limits on the heavy quark EDMs. Thus, our goal of this paper is to understand the current status of indirect limits on the heavy quark EDMs.

Indirect limits on the charm and bottom quark EDMs were previously considered in [31] and the CEDM case was also analyzed in [32]. There the authors derived limits based on the constraints on the heavy quark CEDMs. Indeed, the charm and bottom quark EDMs well above the heavy quark mass scales, say 1 TeV, generate the CEDM operators through the renormalization group (RG) running at a lower energy scale. These CEDMs in turn source the three-gluon Weinberg operator after integrating out the charm and bottom quark. The Weinberg operator then generates the neutron EDM, and this allows one to derive limits on the charm and bottom quark EDMs. We may phrase it a re-interpretation of bounds on the heavy quark CEDMs at the heavy quark mass threshold, with an assumption that there is no cancellation between the EDM and CEDM contributions (see also discussion at the end of § 3). In contrast, in this paper, we study CPCP-odd operators generated from the EDM operators at the heavy quark mass threshold. Therefore our consideration directly applies to the heavy quark EDMs at the quark mass threshold and is independent of [31]. Even though the CPCP-odd operators induced by the heavy quark EDMs are formally higher dimensional, suppressions from the charm and bottom quark masses are not severe as there is only a little hierarchy between the quark masses and the QCD scale.

In order to make our limits robust, we derive indirect limits on charm and bottom quark EDMs based on multiple observables, paramagnetic and neutron EDMs. Indirect limits always have a potential of having a cancellation among different operators. For instance, paramagnetic EDM experiments are sensitive to only a particular linear combination of the electron EDM ded_e and the CPCP-odd semi-leptonic operator CSC_S (see § 3). Therefore a new physics contribution to ded_e and CSC_S can be such that the linear contribution almost vanishes even though each term is sizable. Deriving constraints from two completely different observables, paramagnetic and neutron EDMs, makes the probability of having such a cancellation less likely. We also minimize the QCD uncertainty as much as possible. For this purpose the paramagnetic EDM plays an essential role as it is sensitive to the heavy quark EDM through the semi-leptonic operator CSC_S. The estimation of CSC_S is not polluted by hadronic uncertainties, and its uncertainty is theoretically well under control.

The structure of this paper is as follows. In § 2, we derive CPCP-odd operators generated by a heavy quark EDM after integrating out the heavy quark. In particular, we see that the heavy quark EDM generates CPCP-odd photon-gluon operators and light quark EDMs. We restrict ourselves above the QCD scale in this section, and thus the operators are written in terms of light quarks and gluons. In § 3, we calculate the semi-leptonic CPCP-odd operator CSC_S induced from the CPCP-odd photon-gluon operator below the QCD scale. This allows us to derive a clean constraint on the heavy quark EDMs from the paramagnetic EDM experiments. § 4 is devoted to an estimation of the neutron EDM induced by the CPCP-odd photon-gluon operators and the light quark EDMs. There we derive limits on the heavy quark EDMs from the neutron EDM experiments, which are stronger than the paramagnetic EDM experiments but have a larger hadronic uncertainties. The conventions of this paper is summarized in App. § 6, while some technical details of our computation can be found in App. § 7.

CP-odd operators from heavy quark EDMs

In this section, we compute CPCP-odd operators induced by integrating out charm and bottom quarks with EDMs. The Lagrangian is given by

L=Q‾[iD̸−mQ−idQ2σμνγ5Fμν]Q−14FμνFμν−14GaμνGμνa(1)\mathcal{L}=\overline{Q}\left[i\not{D}-m_Q-\frac{i d_Q}{2}\sigma_{\mu\nu}\gamma_5 F^{\mu\nu}\right]Q-\frac{1}{4}F^{\mu\nu}F_{\mu\nu}-\frac{1}{4}G^{a\mu\nu}G^a_{\mu\nu} \tag*{(1)}

where QQ is the heavy quark with its mass mQm_Q and its EDM dQd_Q. The covariant derivative is defined by

iDμ=i∂μ+gsGμ−eQQAμ,Gμ=TaGμa(2)iD_\mu=i\partial_\mu+g_sG_\mu-eQ_QA_\mu,\qquad G_\mu=T^aG^a_\mu \tag*{(2)}

where GμaG^a_\mu is the gluon field with its field strength GμνaG^a_{\mu\nu} and coupling gsg_s, and AμA_\mu is the photon field with its field strength FμνF_{\mu\nu} and coupling ee. The SU(3) generator is denoted by TaT^a while the U(1) charge by QQQ_Q, with Qc=Qt=+2/3Q_c=Q_t=+2/3 and Qb=−1/3Q_b=-1/3. See App. § 6 for more details on our conventions.

Throughout this paper, “heavy” means that the mass is larger than the QCD scale. Although our discussion applies equally to the top quark, it has another contribution that puts a stronger constraint on the EDM [17, 18]. Therefore our main focus is on the charm and bottom quarks, i.e., Q=c,bQ=c,b.

Once we integrate out the heavy quark, dQd_Q generates several CPCP-odd operators. For our purpose, there are two types of operators that are important: CPCP-odd photon-gluon operators and light quark EDMs. Below the QCD scale, the former generates the neutron EDM and the semi-leptonic CPCP-odd interaction CSC_S that induces paramagnetic EDMs, while the latter contributes to the neutron EDM. In this section, we keep ourselves above the QCD scale. The operators are then written in terms of the quarks and gluons. We will deal with nonperturbative effects at the QCD scale in the subsequent sections.

CP-odd photon-gluon operators

The heavy quark EDM induces CPCP-odd photon-gluon operators through the diagrams in Figure 1. The effective action after integrating out the heavy quark, to linear order in dQd_Q, is given by

Image not extracted

Figure 1. The effective action to linear order in dQd_Q after integrating out the heavy quark. The thick line on the left hand side indicates the full heavy quark propagator with the covariant derivative DμD_\mu, and the cross dot indicates the EDM operator insertion. The full propagator is expanded with respect to the field strengths, which results in the CPCP-odd photon-gluon operators as shown on the right hand side.

Seff=−idQ2Tr⁡[1iD̸−mQσμνF~μν](3)S_{\mathrm{eff}}=-\frac{i d_Q}{2}\operatorname{Tr}\left[\frac{1}{i\not{D}-m_Q}\sigma_{\mu\nu}\widetilde{F}^{\mu\nu}\right] \tag*{(3)}

where the trace is taken over the spinor, the color and the spacetime. We further expand this expression with respect to the field strength HμνH_{\mu\nu}, given by

Hμν=gsGμν−eQQFμν(4)H_{\mu\nu}=g_sG_{\mu\nu}-eQ_QF_{\mu\nu} \tag*{(4)}

which satisfies [iDμ,iDν]=iHμν[iD_\mu,iD_\nu]=iH_{\mu\nu}. To the lowest order, we obtain (see App. § 7.§ 1 for derivation)

Leff=dQ48π2mQ3tr⁡c[−HμνHμνHρσF~ρσ+2HμσHνρHσμF~ρν](5)\mathcal{L}_{\mathrm{eff}}=\frac{d_Q}{48\pi^2m_Q^3}\operatorname{tr}_c\left[-H_{\mu\nu}H^{\mu\nu}H_{\rho\sigma}\widetilde{F}^{\rho\sigma}+2H^\mu{}_\sigma H^{\nu\rho}H^\sigma{}_\mu\widetilde{F}_{\rho\nu}\right] \tag*{(5)}

where tr⁡c\operatorname{tr}_c is the trace only over the color. This contains CPCP-odd photon-gluon operators as well as a CPCP-odd pure photon operator. The operator quadratic in the photon field is given by

LG2FF~=eQQgs2dQ24π2mQ3tr⁡c[FμνGμνF~ρσGρσ−F~μνGνρFρσGσμ](6)\mathcal{L}_{G^2F\widetilde{F}}=\frac{eQ_Qg_s^2d_Q}{24\pi^2m_Q^3}\operatorname{tr}_c\left[F_{\mu\nu}G^{\mu\nu}\widetilde{F}_{\rho\sigma}G^{\rho\sigma}-\widetilde{F}_{\mu\nu}G^{\nu\rho}F_{\rho\sigma}G^{\sigma\mu}\right] \tag*{(6)}

while the one linear in the photon field is given by

LG3F~=gs3dQ48π2mQ3tr⁡c[−GμνGμνGρσF~ρσ+2GμνGνρGσρF~σμ](7)\mathcal{L}_{G^3\widetilde{F}}=\frac{g_s^3d_Q}{48\pi^2m_Q^3}\operatorname{tr}_c\left[-G_{\mu\nu}G^{\mu\nu}G_{\rho\sigma}\widetilde{F}^{\rho\sigma}+2G^\mu{}_\nu G^{\nu\rho}G^\sigma{}_\rho\widetilde{F}_{\sigma\mu}\right] \tag*{(7)}

Carrying out color traces explicitly, one finds that (6) contains δab\delta^{ab}, while (7) has dabcd^{abc} structure. In that sense, (6) would exist for any choice of the gauge group for GμνG_{\mu\nu} including a U(1), while (7) requires N≥3N\geq3 for SU(NN), which includes of course the color group of the Standard Model.

As we will see, below the QCD scale, operator (6) contributes to paramagnetic EDMs through the semi-leptonic CPCP-odd operator CSC_S while (7) contributes to the neutron EDM. The implication of the CPCP-odd pure photon operator is studied in detail in [5] in the context of the muon EDM. This operator puts only a subdominant constraint on the heavy quark EDM, and thus we do not discuss it any further in this paper.

Light quark EDMs

The heavy quark EDM induces the up and down quark EDMs at the three-loop level through the diagram in Figure 2 (and its permutations), where the closed loop is the heavy quark and the cross dot indicates the heavy quark EDM insertion. In our case the heavy quark EDM operator already contains a derivative acting on the external photon, and thus we can ignore the momentum flow of the photon. The mass and the external momentum of the light quark are small compared to the heavy quark mass scale, and this allows us to expand the diagram with respect to the light quark mass and momentum. To linear order, the amplitude is schematically given by

Image not extracted

Figure 2. An example of the three-loop diagrams that generate the light quark EDMs. The cross dot indicates the heavy quark EDM dQd_Q insertion, the wavy line is the external photon, the closed solid line is the heavy quark and the upper solid line is the light quark, respectively. There are five additional diagrams that are permutations of the gluon lines attached to the light quark line.

iM3-loop=iF~μνpρq‾Sμνρq(8)i\mathcal{M}_{3\text{-loop}}=i\widetilde{F}^{\mu\nu}p^\rho\overline{q}S_{\mu\nu\rho}q \tag*{(8)}

where q=u,dq = u, d is the light quark field and pp is the light quark momentum. Here we used the equation of motion to express quantities that explicitly depend on the light quark mass mqm_q in terms of pp. After projecting SμνρS_{\mu\nu\rho} onto the EDM part, we obtain the light quark EDM dqd_q as

dq=mq96tr⁡[Sμνρ{σμν,γρ}],(9)d_q = \frac{m_q}{96}\operatorname{tr}\left[S_{\mu\nu\rho}\left\{\sigma^{\mu\nu},\gamma^\rho\right\}\right], \tag*{(9)}

where the trace is taken only over the spinor index, and {…,…}\{\ldots,\ldots\} represents anticommutator. We evaluate this expression with the dimensional regularization. Since we expanded the amplitude with respect to the external momentum and the light quark mass, SμνρS_{\mu\nu\rho} depends only on the heavy quark mass and hence is given by the vacuum integrals. In our case, among six propagators, three are massive and three are massless. In this case, we can reduce the vacuum integrals to two irreducible master integrals [33]. We use FIRE6 [34] to reduce the vacuum integrals by integration by parts, and use analytical expressions of the master integrals in [35]. We then obtain

dq=5(8ζ(3)−7)72×(αsπ)3mqmQdQ,(10)d_q = \frac{5\left(8\zeta(3)-7\right)}{72}\times\left(\frac{\alpha_s}{\pi}\right)^3\frac{m_q}{m_Q}d_Q, \tag*{(10)}

where αs=gs2/4π\alpha_s = g_s^2/4\pi and ζ(3)≃1.202\zeta(3) \simeq1.202 with ζ(z)\zeta(z) the Riemann zeta function. We have checked that all the gauge-dependent part of the gluon propagator cancels in our computation. Moreover, although each master integral is divergent, all the divergences cancel and the final result is finite. If we drop the color factors, this is equivalent to one contribution to the electron EDM from the muon/tau EDM, the diagram (a) in [3]. Our result differs from [3] and we will come back to the origin of this difference in a separate publication [36]. Notice that dqd_q is suppressed only by 1/mQ1/m_Q since the loop integral is dominated by the loop momentum of order mQm_Q, while the CPCP-odd photon-gluon operators that we computed in § 2.1 are suppressed by 1/mQ31/m_Q^3. Therefore the light quark EDM is expected to be more important for heavier quarks, although there is another (and a dominant) contribution in the case of the top quark as we mentioned at the beginning of this section.

Paramagnetic EDM

In this section we study a constraint on the heavy quark EDM from paramagnetic atomic/molecule EDM experiments, in particular the ACME experiment [2]. The paramagnetic EDM experiments are sensitive only to a linear combination of two operators, the electron EDM ded_e and the semi-leptonic CPCP-odd operator CSC_S given by

L=CSGF2eiγ5e(pˉp+nˉn),(11)\mathcal{L} = C_S\frac{G_F}{\sqrt{2}}e i\gamma_5 e\left(\bar{p}p+\bar{n}n\right), \tag*{(11)}

where GFG_F is the Fermi constant, ee is the electron (one should not confuse it with the electromagnetic coupling), pp is the proton and nn is the neutron, respectively. It is convenient to define an equivalent electron EDM as [37]

de(equiv)=de+CS×1.5×10−20 e cm,(12)d_e^{(\mathrm{equiv})} = d_e+C_S\times1.5\times10^{-20}\ e\,\mathrm{cm}, \tag*{(12)}

where we focus on ThO molecule. The current bound on the equivalent electron EDM is given by ∣de(equiv)∣<1.1×10−29 e cm|d_e^{(\mathrm{equiv})}| < 1.1\times10^{-29}\ e\,\mathrm{cm} [2].

We now compute CSC_S induced by the CPCP-odd photon-gluon operator, in particular the two-photon two-gluon operator, below the QCD scale. The effective Lagrangian is given by

LG2FF~=eQQgs2dQ48π2mQ3GμνaGαβa(FμνF~αβ−FναF~βμ).(13)\mathcal{L}_{G^2F\tilde{F}} = \frac{eQ_Qg_s^2d_Q}{48\pi^2m_Q^3}G_{\mu\nu}^{a}G_{\alpha\beta}^{a}\left(F^{\mu\nu}\widetilde{F}^{\alpha\beta}-F^{\nu\alpha}\widetilde{F}^{\beta\mu}\right). \tag*{(13)}

Because our starting point here is explicitly isospin symmetric, it will result in the same CSC_S coupling for neutrons and protons. The nucleon matrix element of the gluon part is given by

⟨N∣gs232π2GμνaGαβa∣N⟩=−mN108(ηαμηβν−ηανηβμ)NˉN+⋯ ,(14)\langle N|\frac{g_s^2}{32\pi^2}G_{\mu\nu}^aG_{\alpha\beta}^a|N\rangle=-\frac{m_N}{108}(\eta_{\alpha\mu}\eta_{\beta\nu}-\eta_{\alpha\nu}\eta_{\beta\mu})\bar{N}N+\cdots, \tag*{(14)}

where NN is the nucleon field (either pp or nn) with mNm_N its mass, and ⋯\cdots denotes the traceless tensor part that is irrelevant for our purpose. Here we used that, in the chiral limit, the one-loop trace anomaly dominantly contributes to the nucleon mass,

⟨N∣gs232π2GμνaGaμν∣N⟩=−mN9NˉN,(15)\langle N|\frac{g_s^2}{32\pi^2}G_{\mu\nu}^aG^{a\mu\nu}|N\rangle=-\frac{m_N}{9}\bar{N}N, \tag*{(15)}

where the coefficient in the right-hand-side is related to the beta function [38]. The photons are attached to the electron and induce the operator eˉiγ5e\bar{e}i\gamma_5e at one-loop [39] as shown in Figure 3. This integral is logarithmically divergent, which is regulated by the heavy quark mass scale. Therefore, to the leading logarithmic accuracy, we obtain

Image not extracted

Figure 3. The diagram that generates the CPCP-odd semi-leptonic operator CSC_S. The photons are attached to the electron line and generate the structure eˉiγ5e\bar{e}i\gamma_5e, while the gluons feed into the nucleon NN.

CSGF2=−4QQα227mNmemQ3log⁡(mQme)dQe,(16)C_S\frac{G_F}{\sqrt{2}}=-\frac{4Q_Q\alpha^2}{27}\frac{m_Nm_e}{m_Q^3}\log\left(\frac{m_Q}{m_e}\right)\frac{d_Q}{e}, \tag*{(16)}

where α=e2/4π\alpha=e^2/4\pi and mem_e is the electron mass. By requiring ∣de(equiv)∣<1.1×10−29 e cm|d_e^{(\mathrm{equiv})}|<1.1\times10^{-29}\,e\,\mathrm{cm}, we obtain

∣dc∣<1.3×10−20 e cm,(17)|d_c|<1.3\times10^{-20}\,e\,\mathrm{cm}, \tag*{(17)}

for the charm quark and

∣db∣<7.6×10−19 e cm,(18)|d_b|<7.6\times10^{-19}\,e\,\mathrm{cm}, \tag*{(18)}

for the bottom quark, respectively. The heavy quark EDM also induces the electron EDM at three-loop [3, 40], but its contribution to de(equiv)d_e^{(\mathrm{equiv})} is negligible compared to CSC_S.

We now estimate the precision of our calculation. There is an uncertainty associated with Eq. (15), which is valid only in the chiral limit. However, the quark contribution to the nucleon mass is less than 10 % [41], and hence the uncertainty of Eq. (15) is also less than 10 %. Thanks to the lattice computations, these contributions are rather precisely known and the uncertainty in the non-perturbative matrix element can be further reduced by including the light quark contributions into our computation. Another uncertainty originates from our photon-loop computation, where we include only the leading logarithmic term. The uncertainty associated with this treatment may be estimated by changing the cut-off scale from mQm_Q to 2mQ2m_Q, which results in ≲10%\lesssim10\% change of the result for both the charm and bottom quarks. This can be further improved by performing the full two-loop computation, with the heavy quark and photon loops at the same time. Therefore, the precision of our CSC_S calculation is estimated to be 10%10\% and this can be further improved by properly including the quark contribution to the nucleon mass and performing the full two-loop computation of the photon-loop.

Our constraint is stronger than the one derived from e+e−→qqˉe^+e^- \to q\bar q at LEP by several orders of magnitude [25]. Although our constraint is weaker than [31] at its face value, there are two caveats. First, as we mentioned in the introduction, the constraint in [31] is a re-interpretation of the bounds on the CEDMs d~Q\tilde d_Q at the heavy quark mass scale, i.e., d~Q(mQ)\tilde d_Q(m_Q). Both the quark EDM and CEDM at high energy scale ΛNP\Lambda_{\rm NP} contribute to the CEDM at lower energy due to the RG running, and hence d~Q(mQ)\tilde d_Q(m_Q) is given as a linear combination of dQ(ΛNP)d_Q(\Lambda_{\rm NP}) and d~Q(ΛNP)\tilde d_Q(\Lambda_{\rm NP}). With only this information, one can put a constraint on only this particular linear combination of dQ(ΛNP)d_Q(\Lambda_{\rm NP}) and d~Q(ΛNP)\tilde d_Q(\Lambda_{\rm NP}). Therefore the authors assume that there is no huge cancellation between dQ(ΛNP)d_Q(\Lambda_{\rm NP}) and d~Q(ΛNP)\tilde d_Q(\Lambda_{\rm NP}), which allow them to derive a constraint solely on dQ(ΛNP)d_Q(\Lambda_{\rm NP}) and thus on dQ(mQ)d_Q(m_Q). Since our constraint directly applies to dQ(mQ)d_Q(m_Q), it is an independent constraint, without any assumptions on the relative size of dQ(ΛNP)d_Q(\Lambda_{\rm NP}) and d~Q(ΛNP)\tilde d_Q(\Lambda_{\rm NP}). Second, it is a complicated task to estimate the size of the neutron EDM induced by the Weinberg operator. Since [31] relies on the constraints on d~Q(mQ)\tilde d_Q(m_Q) from the Weinberg operator, it has a large hadronic uncertainty. In contrast, as we have just emphasized, our computation of CSC_S is precise and its uncertainty is well under control. Given the rapid progress of the paramagnetic EDM experiments, CSC_S will continue providing important and clean constraints on the heavy quark EDMs. In the next section, we will see that the neutron EDM provides a stronger constraint than CSC_S, comparable to [31], though with a larger hadronic uncertainty.

Neutron EDM

In Sec. 2 we have seen that the heavy quark EDM generates the CPCP-odd photon-gluon operator and the light quark EDM. Below the QCD scale, these CPCP-odd operators in turn generate the neutron EDM dnd_n, whose experimental upper bound is [14].1

∣dn∣<1.8×10−26 e cm.(19)\lvert d_n\rvert< 1.8 \times10^{-26}\,e\,\mathrm{cm}. \tag*{(19)}

In this section we translate this upper bound to the bound on the heavy quark EDM by estimating the size of the induced neutron EDM, with effects of the nonperturbative QCD taken into account (to the extent it is possible).

Light quark EDM contribution

We first study the neutron EDM induced by the light quark EDM dn(dq)d_n^{(d_q)}. Both the QCD sum rule and the quark model suggest that [1, 42, 43]

dn(dq)=13(4dd−du).(20)d_n^{(d_q)} = \frac{1}{3}(4d_d-d_u). \tag*{(20)}

QCD sum rule calculations [42, 43] and more recently lattice calculations [44] give support to this simple formula within ∼30%\sim30\% accuracy. We thus obtain

dn(dq)=5(8ζ(3)−7)72(αsπ)34md−mu3mQdQ.(21)d_n^{(d_q)} = \frac{5(8\zeta(3)-7)}{72}\left(\frac{\alpha_s}{\pi}\right)^3\frac{4m_d-m_u}{3m_Q}d_Q. \tag*{(21)}

As we will see, there is another contribution to the neutron EDM from the CPCP-odd photon-gluon operator that leads to the QCD condensate power corrections in the light quark propagator. The rest of this section is devoted to estimate the size of this contribution.

CP-odd photon-gluon operator contribution

In § 2.1 we saw that the heavy quark EDM generates the CPCP-odd photon-gluon operator

LG3F~=gs3dQ48π2mQ2tr⁡c[−GμνGμνGρσF~ρσ+2G νμF~ ρνG σρF~ μσ].(22)\mathcal{L}_{G^{3}\widetilde{F}} = \frac{g_{s}^{3}d_{Q}}{48\pi^{2}m_{Q}^{2}}\operatorname{tr}_{c}\left[-G_{\mu\nu}G^{\mu\nu}G_{\rho\sigma}\widetilde{F}^{\rho\sigma}+2G^{\mu}_{\ \nu}\widetilde{F}^{\nu}_{\ \rho}G^{\rho}_{\ \sigma}\widetilde{F}^{\sigma}_{\ \mu}\right]. \tag*{(22)}

Below the QCD scale the gluons confine and condense, and this operator feeds into the neutron EDM. We use the QCD sum rule technique [45] to estimate the size of the neutron EDM induced by this operator.

The starting point of the QCD sum rule is to define the two-point correlator

Π(p)≡i∫d4xeip⋅x⟨0∣T{η(x),ηˉ(0)}∣0⟩.(23)\Pi(p) \equiv i\int d^{4}x e^{ip\cdot x}\langle0|T\{\eta(x),\bar{\eta}(0)\}|0\rangle. \tag*{(23)}

The interpolating function η\eta is typically chosen as

η(x)=j1(x)+βj2(x),(24)\eta(x)=j_{1}(x)+\beta j_{2}(x), \tag*{(24)}

and has an overlap with the neutron one-particle state. The functions j1j_{1} and j2j_{2} have the same quantum numbers as the neutron and are given by

j1=2ϵijk(diTCγ5uj)dk,j2=2ϵijk(diTCuj)γ5dk,(25)j_{1}=2\epsilon_{ijk}(d_{i}^{T}C\gamma_{5}u_{j})d_{k},\qquad j_{2}=2\epsilon_{ijk}(d_{i}^{T}Cu_{j})\gamma_{5}d_{k}, \tag*{(25)}

where i,j,ki,j,k are the color indices with ϵijk\epsilon_{ijk} the totally anti-symmetric tensor, and CC is the charge conjugate matrix (see App. § 6 for its definition). Finally β\beta is a redundant parameter that should disappear in an exact computation. In practice, one can choose β\beta, e.g., to optimize the convergence of the operator production expansion (OPE). The QCD sum rule relies on the quark-hadron duality and evaluate this two point correlator Π(p)\Pi(p) in the hadronic (phenomenological) side and the quark (OPE) side. On the phenomenological side, Π(p)\Pi(p) is expressed in terms of hadronic quantities such as the neutron mass and dipole moments. On the OPE side, Π(p)\Pi(p) is evaluated in terms of perturbative quarks and gluons with non-perturbative effects included in the form of the QCD condensates. One then obtains an estimation of the hadronic quantities by equating these two expressions, after the Borel transformation to reduce effects of excited states. In our case, we have external electromagnetic field FμνF_{\mu\nu} multiplying three Lorentz structures that in principle we can use to derive the sum rules: p̸σμνp̸\not{p}\sigma_{\mu\nu}\not{p}, {p̸,σμν}\{\not{p},\sigma_{\mu\nu}\} and σμν\sigma_{\mu\nu}. In this paper, we focus on the sum rule that follows from p̸σμνp̸\not{p}\sigma_{\mu\nu}\not{p} since it depends most strongly on the momentum and the vacuum susceptibilities are less relevant [46]. Moreover, it is computationally simple to derive the QCD sum rule based on this structure. In the following we evaluate Π(p)\Pi(p) in the OPE and phenomenological sides, respectively.

OPE side. We denote the light quark propagator as

⟨0∣T{qi(x),qˉj(y)}∣0⟩≡δijS(x,y),(26)\langle0|T\{q_{i}(x),\bar{q}_{j}(y)\}|0\rangle\equiv\delta_{ij}S(x,y), \tag*{(26)}

where only the color-diagonal contribution is relevant for the QCD sum rule based on p̸σμνp̸\not p\sigma_{\mu\nu}\not p in our case. The CPCP-odd photon-gluon operator does not distinguish the up and down quarks, and thus we collectively denote uu and dd as qq. With this expression, we obtain

ΠOPE(p)=−24i∫d4xeip⋅x{tr⁡[γ5Scγ5S]S+Sγ5Scγ5S+β2(γ5SScγ5+tr⁡[ScS]γ5Sγ5)+β(SScγ5Sγ5+γ5Sγ5ScS+tr⁡[SScγ5]Sγ5+tr⁡[ScSγ5]γ5S)}.(27)\Pi_{\mathrm{OPE}}(p)=-24i\int d^{4}x e^{ip\cdot x}\left\{\operatorname{tr}\left[\gamma_{5}S^{c}\gamma_{5}S\right]S+S\gamma_{5}S^{c}\gamma_{5}S+\beta^{2}\left(\gamma_{5}SS^{c}\gamma_{5}+\operatorname{tr}\left[S^{c}S\right]\gamma_{5}S\gamma_{5}\right)+\beta\left(SS^{c}\gamma_{5}S\gamma_{5}+\gamma_{5}S\gamma_{5}S^{c}S+\operatorname{tr}\left[SS^{c}\gamma_{5}\right]S\gamma_{5}+\operatorname{tr}\left[S^{c}S\gamma_{5}\right]\gamma_{5}S\right)\right\}. \tag*{(27)}

where Sαβc≡(STC)αβS^{c}_{\alpha\beta}\equiv(S^{T}C)_{\alpha\beta} and SαβT≡SβαS^{T}_{\alpha\beta}\equiv S_{\beta\alpha} with α\alpha and β\beta being the spinor indices, and the trace is taken only over the spinor index. In our case the light quark propagator has two contributions

S=S(0)+S(CP),(28)S=S^{(0)}+S^{(CP)}, \tag*{(28)}

where S(0)S^{(0)} corresponds to the free quark propagator and is given by

S(0)(x)=ix̸2π2x4,(29)S^{(0)}(x)=\frac{i\not x}{2\pi^{2}x^{4}}, \tag*{(29)}

while S(CP)S^{(CP)} is the CPCP-odd part which we compute from now.

The source inserted into the correlator has three gluon fields, and we follow the general idea of [23] where the Weinberg operator contribution to dnd_n was first evaluated, and where two gluons are treated perturbatively, while the third gluon field contributes to the quark-gluon condensate. With the QCD condensate background, we can write down the diagram

Sij(CP)(p)=(30)S_{ij}^{(CP)}(p)= \tag*{(30)}

where the crosses indicate the background fields, either the external photon or the QCD condensation of the quarks and gluons, and the cross dot is the insertion of the CPCP-odd photon-gluon operator (22). Since the QCD vacuum does not violate the Lorentz and color symmetries, we have

⟨qiqˉjGμνa⟩=−1192Tijaσμν⟨qˉσ⋅Gq⟩,(31)\left\langle q_i\bar q_jG_{\mu\nu}^{a}\right\rangle=-\frac{1}{192}T_{ij}^{a}\sigma_{\mu\nu}\left\langle\bar q\sigma\cdot Gq\right\rangle, \tag*{(31)}

where σ⋅G=σμνGμν\sigma\cdot G=\sigma_{\mu\nu}G^{\mu\nu}, and ⟨⋯ ⟩\langle\cdots\rangle corresponds to the vacuum expectation value. We thus obtain

Sij(CP)(p)=δijS(CP)(p)=iδij5αs2dQ864mQ3gs⟨qˉσ⋅Gq⟩p2ip̸[σ⋅F~−2pμpνp2F~μασνα]ip̸.(32)S_{ij}^{(CP)}(p)=\delta_{ij}S^{(CP)}(p)=i\delta_{ij}\frac{5\alpha_{s}^{2}d_{Q}}{864m_{Q}^{3}}\frac{g_{s}\langle\bar q\sigma\cdot Gq\rangle}{p^{2}}\frac{i}{\not p}\left[\sigma\cdot\widetilde F-\frac{2p^{\mu}p^{\nu}}{p^{2}}\widetilde F_{\mu\alpha}\sigma_{\nu}{}^{\alpha}\right]\frac{i}{\not p}. \tag*{(32)}

Its Fourier transformation has an IR divergence, which in the dimensional regularization leads

S(CP)(x)=5αs2dQ27648π2mQ3gs⟨qˉσ⋅Gq⟩Γ(ϵIR)(−ΛIR2x2)−ϵIRF~⋅σ,(33)S^{(CP)}(x)=\frac{5\alpha_{s}^{2}d_{Q}}{27648\pi^{2}m_{Q}^{3}}g_{s}\langle\bar q\sigma\cdot Gq\rangle\Gamma(\epsilon_{\mathrm{IR}})\left(-\Lambda_{\mathrm{IR}}^{2}x^{2}\right)^{-\epsilon_{\mathrm{IR}}}\widetilde F\cdot\sigma, \tag*{(33)}

where we take dIR=4+2ϵIRd_{\mathrm{IR}}=4+2\epsilon_{\mathrm{IR}} and keep only the logarithmic terms, with ΛIR\Lambda_{\mathrm{IR}} the IR cut-off scale. Due to the sensitivity to IR scale, calculations of further terms in the OPE are not possible. As a consequence, this evaluation should be viewed as an estimate, which cannot be systematically improved within QCD sum rule method. With the explicit forms of S(0)S^{(0)} and S(CP)S^{(\mathcal{CP})}, we obtain

\Pi_{\mathrm{OPE}}(p)=-\frac{5\alpha_s^2d_Q}{2^{12}3^3\pi^4m_Q^3}\langle\bar q\sigma\cdot g_sGq\rangle\left[(1-\beta)^2\slashed{p}\tilde F\cdot\sigma\slashed{p}-18(1-\beta^2)\slashed{p}\tilde F\cdot\sigma\right]I(p^2;\varepsilon_{\mathrm{IR}},\varepsilon_{\mathrm{UV}}). \tag*{(34)}

where we perform the dimensional regularization to tame the UV divergence with dUV=4−2εUVd_{\mathrm{UV}}=4-2\varepsilon_{\mathrm{UV}}, and

I(p2;εIR,εUV)≡Γ(εIR)(−ΛIR2p2)−εIRΓ(−εUV−εIR)(−μ2p2)−εUV,(35)I(p^2;\varepsilon_{\mathrm{IR}},\varepsilon_{\mathrm{UV}})\equiv\Gamma(\varepsilon_{\mathrm{IR}})\left(-\frac{\Lambda_{\mathrm{IR}}^2}{p^2}\right)^{-\varepsilon_{\mathrm{IR}}}\Gamma(-\varepsilon_{\mathrm{UV}}-\varepsilon_{\mathrm{IR}})\left(-\frac{\mu^2}{p^2}\right)^{-\varepsilon_{\mathrm{UV}}}, \tag*{(35)}

with μ\mu the renormalization scale. We then perform the Borel transformation, defined as [47]

B[Π(p2=−P2)]=1π∫0∞dP2M2e−P2/M2Im⁡[Π(p)]p2=−P2.(36)\mathcal{B}\left[\Pi(p^2=-P^2)\right]=\frac{1}{\pi}\int_0^\infty\frac{dP^2}{M^2}e^{-P^2/M^2}\operatorname{Im}\left[\Pi(p)\right]_{p^2=-P^2}. \tag*{(36)}

We thus obtain the \slashed{p}\tilde F\cdot\sigma\slashed{p} part as

\mathcal{B}\left[\Pi_{\mathrm{OPE}}(p^2=-P^2)\right]_{\slashed{p}\tilde F\cdot\sigma\slashed{p}}=-\frac{5\alpha_s^2d_Q}{2^{12}3^3\pi^4m_Q^3}(1-\beta)^2\langle\bar q\sigma\cdot g_sGq\rangle\log\left(\frac{M^2}{\Lambda_{\mathrm{IR}}^2}\right). \tag*{(37)}

Taking the imaginary part of I(p2;εIR,εUV)I(p^2;\varepsilon_{\mathrm{IR}},\varepsilon_{\mathrm{UV}}) is somewhat subtle, and we provide details in App. § 7.§ 2. There we also clarify the physical meaning of the scale ΛIR\Lambda_{\mathrm{IR}}; it should be identified with the mass of the constituent quarks. This expression is to be compared with the phenomenological expression.

Phenomenological side. On the phenomenological side, the two-point correlator with the neutron EDM insertion is expressed as

\Pi_{\mathrm{pheno}}(p)=-\lambda_n^2\left[\frac{\slashed{p}+m_n}{p^2-m_n^2}-\frac{d_n}{2(p^2-m_n^2)^2}\slashed{p}\tilde F\cdot\sigma\slashed{p}+\cdots\right]. \tag*{(38)}

where mnm_n is the neutron mass and λn\lambda_n parametrizes the overlap between the interpolating function η\eta and the neutron one-particle state. After the Borel transformation, we obtain

\mathcal{B}\left[\Pi_{\mathrm{pheno}}(p^2=-P^2)\right]_{\slashed{p}\tilde F\cdot\sigma\slashed{p}}=\frac{\lambda_n^2d_n}{2M^4}e^{-m_n^2/M^2}+\cdots, \tag*{(39)}

where ⋯\cdots expresses contributions from excited states which we ignore in the following.

Sum rule. The QCD sum rule of the neutron EDM is obtained by equating Eqs. (37) and (39). We thus obtain

λn2dn(G3F~)2M4e−mn2/M2=−5αs2dQ21233π4mQ3(1−β)2⟨ψˉσ⋅gsGψ⟩log⁡(M2ΛIR2).(40)\frac{\lambda_n^2d_n^{(G^3\tilde F)}}{2M^4}e^{-m_n^2/M^2}=-\frac{5\alpha_s^2d_Q}{2^{12}3^3\pi^4m_Q^3}(1-\beta)^2\langle\bar\psi\sigma\cdot g_sG\psi\rangle\log\left(\frac{M^2}{\Lambda_{\mathrm{IR}}^2}\right). \tag*{(40)}

where dn(F~G3)d_n^{(\tilde F G^3)} is the neutron EDM induced by the CPCP-odd photon-gluon operator (to distinguish it from the one induced by the light quark EDM). We may use the Ioffe formula for the nucleon mass [48, 49]2

λn2mnM4e−mn2/M2=−7−2β−5β216π2⟨qˉq⟩,(41)\frac{\lambda_n^2m_n}{M^4}e^{-m_n^2/M^2}=-\frac{7-2\beta-5\beta^2}{16\pi^2}\langle\bar qq\rangle, \tag*{(41)}

to eliminate λn\lambda_n. Then we obtain the QCD sum rule estimation of the neutron EDM

dn(G3F~)=dQ×5αs2273π2mnm02mQ31−β7+5βlog⁡(M2ΛIR2)(42)d_n^{(G^3\widetilde{F})}=d_Q\times\frac{5\alpha_s^2}{27^3\pi^2}\frac{m_nm_0^2}{m_Q^3}\frac{1-\beta}{7+5\beta}\log\left(\frac{M^2}{\Lambda_{\rm IR}^2}\right) \tag*{(42)}

where we used ⟨qσ⋅gsGq⟩=m02⟨qˉq⟩\langle q\sigma\cdot g_sGq\rangle=m_0^2\langle\bar{q}q\rangle with m02=0.8 GeV2m_0^2=0.8\,\mathrm{GeV}^2 [50].

Constraint on heavy quark EDM

The neutron EDM has two contributions, induced by the light quark EDM and the CPCP-odd photon-gluon operators, and is given by

dn=dn(dq)+dn(G3F~),(43)d_n=d_n^{(d_q)}+d_n^{(G^3\widetilde{F})}, \tag*{(43)}

where

dn(dq)=dQ×5(8ζ(3)−7)72(αsπ)34md−mu3mQ,dn(G3F~)=dQ×5αs2273π2mnm02mQ3log⁡(M2ΛIR2),(44)d_n^{(d_q)}=d_Q\times\frac{5(8\zeta(3)-7)}{72}\left(\frac{\alpha_s}{\pi}\right)^3\frac{4m_d-m_u}{3m_Q},\qquad d_n^{(G^3\widetilde{F})}=d_Q\times\frac{5\alpha_s^2}{27^3\pi^2}\frac{m_nm_0^2}{m_Q^3}\log\left(\frac{M^2}{\Lambda_{\rm IR}^2}\right), \tag*{(44)}

and we take β=−1\beta=-1 following [48, 51, 52]. The parameters should be evaluated at mQm_Q for the former contribution, and at the scale close to ΛQCD\Lambda_{\rm QCD} for the latter contribution. We use mc=1.27 GeVm_c=1.27\,\mathrm{GeV}, mb=4.18 GeVm_b=4.18\,\mathrm{GeV}, αs(mc)=0.38\alpha_s(m_c)=0.38 and αs(mb)=0.223\alpha_s(m_b)=0.223 [53]. The light quark masses (in the MS scheme) also depend on the energy scale, and we take mu(mc)=2.5 MeVm_u(m_c)=2.5\,\mathrm{MeV}, md(mc)=5.4 MeVm_d(m_c)=5.4\,\mathrm{MeV}, mu(mb)=1.8 MeVm_u(m_b)=1.8\,\mathrm{MeV} and md(mb)=4.0 MeVm_d(m_b)=4.0\,\mathrm{MeV} that we obtain by running the light quark masses at 2 GeV2\,\mathrm{GeV} following [53]. Finally we take αs=0.5\alpha_s=0.5, ΛIR=300 MeV\Lambda_{\rm IR}=300\,\mathrm{MeV} and M=800 MeVM=800\,\mathrm{MeV} for the QCD sum rule estimation for definiteness. For these values, the contribution from the CPCP-odd photon-gluon operator is larger by a factor of 16 and 10 for the charm and bottom quarks, respectively. In the bottom quark case, the explicit quark mass suppression is compensated by the running of the strong coupling, and the CPCP-odd photon-gluon operator is still larger than the light quark EDM even with the relative suppression factor 1/mb21/m_b^2. By requiring that ∣dn∣<1.8×10−26 e cm|d_n|<1.8\times10^{-26}\,e\,\mathrm{cm}, we obtain

∣dc∣<6×10−22 e cm,(45)|d_c|<6\times10^{-22}\,e\,\mathrm{cm}, \tag*{(45)}

for the charm quark, and

∣db∣<2×10−20 e cm,(46)|d_b|<2\times10^{-20}\,e\,\mathrm{cm}, \tag*{(46)}

for the bottom quark, respectively.

The constraint from dnd_n is stronger than that from CSC_S by more than an order of magnitude. However, one should note that the estimation of dnd_n has a large hadronic uncertainty. Indeed, the final result is affected by a factor two if we use the sum rule of the nucleon kinetic term instead of the mass term for λn\lambda_n. There are also uncertainties related to the choice of MM and β\beta which can again result in a factor of a few difference in the final result. Therefore our constraint here should be understood as an order-of-magnitude estimation and the numerical factor should be taken with care. In contrast, our calculation of CSC_S is far more precise. Its uncertainty is estimated to be ∼10%\sim10\% and can be further reduced if needed (see the end of § 3). In this sense, the constraints from dnd_n and CSC_S are complementary to each other; dnd_n puts a stronger constraint on the heavy quark EDM, while the calculation of CSC_S is cleaner and its uncertainty is well under control.

Our constraint on dcd_c is stronger than [31] by a factor two, while the one on dbd_b is weaker by a factor two. However, as we noted in the introduction and the end of § 3, our constraint directly applies to dc(mc)d_c(m_c) and db(mb)d_b(m_b), and is independent of the one in [31].

Conclusion

In this paper, we have derived indirect limits on the charm and bottom quark EDMs. The charm and bottom quark EDMs generate the CPCP-odd photon-gluon operators and the light quark EDMs after integrating out the charm and bottom quarks. Photon-gluon operators contribute to the semi-leptonic CPCP-odd operator CSC_S (and ultimately to paramagnetic AMO EDMs) as well as to the neutron EDM at a non-perturbative level. Quark EDM dominantly contributes to the neutron and nuclear EDMs. Performing our evaluation and using the current limits, we obtain

∣dc∣<1.3×10−20 e cm,∣db∣<7.6×10−19 e cm,(47)|d_c| < 1.3 \times10^{-20}\,e\,\mathrm{cm}, \qquad|d_b| < 7.6 \times10^{-19}\,e\,\mathrm{cm}, \tag*{(47)}

from the paramagnetic EDM experiments, and

∣dc∣<6×10−22 e cm,∣db∣<2×10−20 e cm,(48)|d_c| < 6 \times10^{-22}\,e\,\mathrm{cm}, \qquad|d_b| < 2 \times10^{-20}\,e\,\mathrm{cm}, \tag*{(48)}

from the neutron EDM experiment, respectively. Although the constraint from the neutron EDM is stronger, it has a larger hadronic uncertainty. The uncertainty of the constraint from the paramagnetic EDM is estimated as 10 % and can be improved if needed, while the uncertainty from the neutron EDM can be a factor of a few. Our constraint is independent of the one given in [31] in the sense that our constraint directly applies to the EDM operators at the quark mass threshold. By assuming a simple scaling of dQ/e∝(α/π)mQ/ΛQ2d_Q/e \propto(\alpha/\pi)m_Q/\Lambda_Q^2, we may translate our constraint as a lower bound on CPCP-odd new physics scale: Λc>70 GeV\Lambda_c > 70\ \mathrm{GeV} and Λb>20 GeV\Lambda_b > 20\ \mathrm{GeV} from the paramagnetic EDM experiment, and Λc>300 GeV\Lambda_c > 300\ \mathrm{GeV} and Λb>100 GeV\Lambda_b > 100\ \mathrm{GeV} from the neutron EDM experiment.

Our result provides an important benchmark to overcome for the LHC based measurements of the charmed baryon EDMs [26, 27, 28, 29, 30]. The idea of using the bent crystal technique for studying electromagnetic properties of baryons containing a heavy quark is very appealing. However, given the strength of the bounds derived in our work, and the necessity to satisfy independent constraints from dnd_n and CSC_S (hence removing a chance of accidentally large dc(b)d_{c(b)} due to cancellations), one may want to re-evaluate the main goal of the charmed baryon experiment. While it will be difficult to match the indirect sensitivity to dc(b)d_{c(b)}, the planned measurement may achieve sufficient accuracy to extract the values of the magnetic moments μc(b)\mu_{c(b)} and compare it with the QCD predictions.

Acknowledgements Y.E. and M.P. are supported in part by U.S. Department of Energy Grant No. de-sc0011842. M.P. would like to thank Dr. K. Melnikov for the advice in evaluating loop contributions. The Feynman diagrams in this paper are generated by TikZ-Feynman [54].

Convention

Here we summarize our conventions used in this paper. The field strengths are given by

Fμν=∂μAν−∂νAμ,Gμνa=∂μGνa−∂νGμa+gsfabcGμbGνc,Gμν=GμνaTa,(49)F_{\mu\nu} = \partial_\mu A_\nu- \partial_\nu A_\mu,\qquad G_{\mu\nu}^{a} = \partial_\mu G_\nu^{a} - \partial_\nu G_\mu^{a} + g_s f^{abc}G_\mu^{b}G_\nu^{c},\qquad G_{\mu\nu} = G_{\mu\nu}^{a}T^{a}, \tag*{(49)}

where fabcf^{abc} is the SU(3) structure constant. The SU(3) generator satisfies

[Ta,Tb]=ifabcTc,tr⁡c[TaTb]=δab2.(50)\left[T^{a},T^{b}\right] = if^{abc}T^{c},\qquad\operatorname{tr}_{c}\left[T^{a}T^{b}\right] = \frac{\delta^{ab}}{2}. \tag*{(50)}

The dual field strengths are defined as

F~μν=12ϵμνρσFρσ,G~μν=12ϵμνρσGρσ,(51)\widetilde{F}^{\mu\nu} = \frac{1}{2}\epsilon^{\mu\nu\rho\sigma}F_{\rho\sigma},\qquad\widetilde{G}^{\mu\nu} = \frac{1}{2}\epsilon^{\mu\nu\rho\sigma}G_{\rho\sigma}, \tag*{(51)}

with ϵ0123=+1\epsilon^{0123}=+1. We take the gamma matrix as

{γμ,γν}=2ημν,ημν=diag⁡(+,−,−,−),γ5=iγ0γ1γ2γ3=−i4!ϵμνρσγμγνγργσ,(52)\{\gamma^\mu,\gamma^\nu\}=2\eta^{\mu\nu},\qquad\eta_{\mu\nu}=\operatorname{diag}(+,-,-,-),\qquad\gamma_5=i\gamma^0\gamma^1\gamma^2\gamma^3=-\frac{i}{4!}\epsilon^{\mu\nu\rho\sigma}\gamma_\mu\gamma_\nu\gamma_\rho\gamma_\sigma, \tag*{(52)}

and σμν\sigma_{\mu\nu} as

σμν=i2(γμγν−γνγμ)=i2[γμ,γν].(53)\sigma_{\mu\nu}=\frac{i}{2}(\gamma_\mu\gamma_\nu-\gamma_\nu\gamma_\mu)=\frac{i}{2}[\gamma_\mu,\gamma_\nu]. \tag*{(53)}

The charge conjugation matrix CC satisfies

(γμ)TC=−Cγμ,(54)(\gamma^\mu)^T C=-C\gamma^\mu, \tag*{(54)}

such that ψTC\psi^T C has the same Lorentz transformation property as ψˉ\bar{\psi}. In the Weyl representation it is given by

C=iγ0γ2.(55)C=i\gamma^0\gamma^2. \tag*{(55)}

Technical details

In this appendix we provide some technical details that we omit in the main text.

Derivation of CPCP-odd photon-gluon operators

In this subsection we provide derivation of the CPCP-odd photon-gluon operators (5) after integrating out the heavy quarks (see [55] for an extensive review on this procedure). Our starting point is Eq. (3). We can simplify it as

S_{\mathrm{eff}}=-\frac{i d_Q}{2}\operatorname{Tr}\left[\frac{i\slashed{D}+m_Q}{i\slashed{D}+m_Q}\frac{1}{i\slashed{D}-m_Q}\sigma_{\mu\nu}\widetilde{F}^{\mu\nu}\right]=-\frac{i d_Qm_Q}{2}\operatorname{Tr}\left[\frac{1}{(iD)^2-m_Q^2+\frac{1}{2}\sigma\cdot H}(\sigma\cdot\widetilde{F})\right], \tag*{(56)}

where (iD)2=iDμiDμ(iD)^2=iD_\mu iD^\mu and we used that the traces of odd γ\gamma's vanish in the second equality. We further expand the denominator with respect to HH, and obtain up to fourth-order in the gauge fields

Seff=−idQmQ2(T2+T3+T4),(57)S_{\mathrm{eff}}=-\frac{i d_Qm_Q}{2}(T_2+T_3+T_4), \tag*{(57)}

where

T2=−12Tr⁡[1(iD)2−mQ2(σ⋅H)1(iD)2−mQ2(σ⋅F~)],(58)T_2=-\frac{1}{2}\operatorname{Tr}\left[\frac{1}{(iD)^2-m_Q^2}(\sigma\cdot H)\frac{1}{(iD)^2-m_Q^2}(\sigma\cdot\widetilde{F})\right], \tag*{(58)}
T3=+14Tr⁡[1(iD)2−mQ2(σ⋅H)1(iD)2−mQ2(σ⋅H)1(iD)2−mQ2(σ⋅F~)],(59)T_3=+\frac{1}{4}\operatorname{Tr}\left[\frac{1}{(iD)^2-m_Q^2}(\sigma\cdot H)\frac{1}{(iD)^2-m_Q^2}(\sigma\cdot H)\frac{1}{(iD)^2-m_Q^2}(\sigma\cdot\widetilde{F})\right], \tag*{(59)}
T4=−18Tr⁡[1(iD)2−mQ2(σ⋅H)1(iD)2−mQ2(σ⋅H)1(iD)2−mQ2(σ⋅H)1(iD)2−mQ2(σ⋅F~)].(60)T_4=-\frac{1}{8}\operatorname{Tr}\left[\frac{1}{(iD)^2-m_Q^2}(\sigma\cdot H)\frac{1}{(iD)^2-m_Q^2}(\sigma\cdot H)\frac{1}{(iD)^2-m_Q^2}(\sigma\cdot H)\frac{1}{(iD)^2-m_Q^2}(\sigma\cdot\widetilde{F})\right]. \tag*{(60)}

The term linear in σμν\sigma_{\mu\nu} vanishes after taking the trace of the spinor index. In order to compute T2T_2, we use the following identity [55]:

Tr⁡[1((iD)2−mQ2)3f(H,F)]=Tr⁡[1((i∂)2−mQ2)3f(H,F)]−Tr⁡[1((iD)2−mQ2)5HμνHμνf(H,F)],(61)\operatorname{Tr}\left[\frac{1}{\left((iD)^2-m_Q^2\right)^3}f(H,F)\right]=\operatorname{Tr}\left[\frac{1}{\left((i\partial)^2-m_Q^2\right)^3}f(H,F)\right]-\operatorname{Tr}\left[\frac{1}{\left((iD)^2-m_Q^2\right)^5}H_{\mu\nu}H^{\mu\nu}f(H,F)\right], \tag*{(61)}

where f(H,F)f(H,F) is an arbitrary function of HμνH_{\mu\nu} and FμνF_{\mu\nu}. We thus obtain

ddmQ2T2=−Tr⁡[1((i∂)2−mQ2)3(σ⋅H) σ⋅F~]+Tr⁡[1((iD)2−mQ2)5HμνHμν(σ⋅H) σ⋅F~].(62)\frac{d}{dm_Q^2}T_2=-\operatorname{Tr}\left[\frac{1}{\left((i\partial)^2-m_Q^2\right)^3}(\sigma\cdot H)\,\sigma\cdot\widetilde{F}\right]+\operatorname{Tr}\left[\frac{1}{\left((iD)^2-m_Q^2\right)^5}H_{\mu\nu}H^{\mu\nu}(\sigma\cdot H)\,\sigma\cdot\widetilde{F}\right]. \tag*{(62)}

where we ignore the derivatives acting on FF. We can now replace iDiD by i∂i\partial in the last term to the order of our interest, and obtain

ddmQ2T2=−i∫d4x[3eQQQ4π2mQ2FμνF~μν+124π2mQ2tr⁡c[HμνHμνHρσF~ρσ]].(63)\frac{d}{dm_Q^2}T_2=-i\int d^4x\left[\frac{3e_QQ_Q}{4\pi^2m_Q^2}F_{\mu\nu}\widetilde{F}^{\mu\nu}+\frac{1}{24\pi^2m_Q^2}\operatorname{tr}_c\left[H_{\mu\nu}H^{\mu\nu}H_{\rho\sigma}\widetilde{F}^{\rho\sigma}\right]\right]. \tag*{(63)}

By integrating this we obtain

T2=i∫d4x[3eQQQ4π2ln⁡(MR2mQ2)FμνF~μν+148π2mQ4tr⁡c[HμνHμνHρσF~ρσ]],(64)T_2=i\int d^4x\left[\frac{3e_QQ_Q}{4\pi^2}\ln\left(\frac{M_R^2}{m_Q^2}\right)F_{\mu\nu}\widetilde{F}^{\mu\nu}+\frac{1}{48\pi^2m_Q^4}\operatorname{tr}_c\left[H_{\mu\nu}H^{\mu\nu}H_{\rho\sigma}\widetilde{F}^{\rho\sigma}\right]\right], \tag*{(64)}

where the additional mass scale MRM_R comes from the regularization which we did not write down explicitly. Next we compute T3T_3. By ignoring the derivatives acting on F~μν\widetilde{F}_{\mu\nu}, we obtain

T3=14Tr⁡[1((iD)2−mQ2)3(σ⋅H)2(σ⋅F~)+1((iD)2−mQ2)4[(iD)2,σ⋅H](σ⋅H)(σ⋅F~)−1((iD)2−mQ2)5[(iD)2,σ⋅H][(iD)2,σ⋅H]σ⋅F~+⋯ ],(65)T_3=\frac{1}{4}\operatorname{Tr}\left[\frac{1}{\left((iD)^2-m_Q^2\right)^3}(\sigma\cdot H)^2(\sigma\cdot\widetilde{F})+\frac{1}{\left((iD)^2-m_Q^2\right)^4}[(iD)^2,\sigma\cdot H](\sigma\cdot H)(\sigma\cdot\widetilde{F})-\frac{1}{\left((iD)^2-m_Q^2\right)^5}[(iD)^2,\sigma\cdot H][(iD)^2,\sigma\cdot H]\sigma\cdot\widetilde{F}+\cdots\right], \tag*{(65)}

where the dots indicate terms that contain more than two commutators and thus correspond to higher dimensional operators that are out of our interest. The first term is easily seen to generate only O(H4F~)O(H^4\widetilde{F}) operators with the help of the identity Eq. (B.6). The second term is expanded as

Tr⁡[1((iD)2−mQ2)4[(iD)2,σ⋅H](σ⋅H)(σ⋅F~)]=Tr⁡[1((iD)2−mQ2)4iDα[iDα,σ⋅H](σ⋅H)σ⋅F~+iDα1((iD)2−mQ2)4[iDα,σ⋅H](σ⋅H)σ⋅F~+1((iD)2−mQ2)4[iDα,σ⋅H][iDα,σ⋅H]σ⋅F~].(66)\operatorname{Tr}\left[\frac{1}{\left((iD)^2-m_Q^2\right)^4}[(iD)^2,\sigma\cdot H](\sigma\cdot H)(\sigma\cdot\widetilde{F})\right]=\operatorname{Tr}\left[\frac{1}{\left((iD)^2-m_Q^2\right)^4}iD^\alpha[iD_\alpha,\sigma\cdot H](\sigma\cdot H)\sigma\cdot\widetilde{F} +iD^\alpha\frac{1}{\left((iD)^2-m_Q^2\right)^4}[iD_\alpha,\sigma\cdot H](\sigma\cdot H)\sigma\cdot\widetilde{F} +\frac{1}{\left((iD)^2-m_Q^2\right)^4}[iD^\alpha,\sigma\cdot H][iD_\alpha,\sigma\cdot H]\sigma\cdot\widetilde{F}\right]. \tag*{(66)}

The first two terms induce only higher order terms [55], and thus we ignore them. The third term in Eq. (B.10) already contains three field strengths and two commutators, and hence we can ignore any further commutators between DD and HH. By taking the angular average, we obtain

Tr⁡[1((iD)2−mQ2)5[(iD)2,σ⋅H][(iD)2,σ⋅H]σ⋅F~]=Tr⁡[(iD)2((iD)2−mQ2)5[iDα,σ⋅H][iDα,σ⋅H]σ⋅F~].(67)\operatorname{Tr}\left[\frac{1}{((iD)^2-m_Q^2)^5}[(iD)^2,\sigma\cdot H][(iD)^2,\sigma\cdot H]\sigma\cdot\widetilde{F}\right]=\operatorname{Tr}\left[\frac{(iD)^2}{((iD)^2-m_Q^2)^5}[iD^\alpha,\sigma\cdot H][iD_\alpha,\sigma\cdot H]\sigma\cdot\widetilde{F}\right]. \tag*{(67)}

By combining them, we obtain

T3=−mQ24Tr⁡[1((iD)2−mQ2)5[iDα,σ⋅H][iDα,σ⋅H]σ⋅F~]=−gs224π2mQ4∫d4x tr⁡c[(DαGμρ)(DαGνρ)]F~μν=0.(68)\begin{aligned} T_3=-\frac{m_Q^2}{4}\operatorname{Tr}\left[\frac{1}{((iD)^2-m_Q^2)^5}[iD^\alpha,\sigma\cdot H][iD_\alpha,\sigma\cdot H]\sigma\cdot\widetilde{F}\right] \\ &=-\frac{g_s^2}{24\pi^2m_Q^4}\int d^4x\,\operatorname{tr}_c\left[(D^\alpha G_{\mu\rho})(D_\alpha G_\nu{}^\rho)\right]\widetilde{F}^{\mu\nu}=0. \tag*{(68)} \end{aligned}

Finally we compute T4T_4. To the order of our interest, we can simply replace iDiD by i∂i\partial in the denominator. It is then easy to see that

T4=−i48π2mQ4∫d4x tr⁡c[3HμνHμνHρσF~ρσ−4HμνHνρHρσF~σμ].(69)T_4=-\frac{i}{48\pi^2m_Q^4}\int d^4x\,\operatorname{tr}_c\left[3H_{\mu\nu}H^{\mu\nu}H_{\rho\sigma}\widetilde{F}^{\rho\sigma}-4H^\mu{}_\nu H^\nu{}_\rho H^\rho{}_\sigma\widetilde{F}^\sigma{}_\mu\right]. \tag*{(69)}

Therefore we obtain

Seff=dQ48π2mQ3∫d4x tr⁡c[−HμνHμνHρσF~ρσ+2HμνHνρHρσF~σμ],(70)S_{\mathrm{eff}}=\frac{d_Q}{48\pi^2m_Q^3}\int d^4x\,\operatorname{tr}_c\left[-H_{\mu\nu}H^{\mu\nu}H_{\rho\sigma}\widetilde{F}^{\rho\sigma}+2H^\mu{}_\nu H^\nu{}_\rho H^\rho{}_\sigma\widetilde{F}^\sigma{}_\mu\right], \tag*{(70)}

where we ignored the quadratic term that is irrelevant for our purpose. If we drop the gluons, it correctly reduces to the result in [5].

Borel transformation and IR divergence

In this subsection we discuss the Borel transformation of Eq. (34) that contains both the UV and IR divergences. As the Borel transformation is related to the imaginary part, it is equivalent to taking the imaginary part of

I(p2;ϵIR,ϵUV)=Γ(ϵIR)(−ΛIR2p2)−ϵIRΓ(−ϵUV−ϵIR)(−ΛUV2p2)−ϵUV.(71)I(p^2;\epsilon_{\mathrm{IR}},\epsilon_{\mathrm{UV}})=\Gamma(\epsilon_{\mathrm{IR}})\left(-\frac{\Lambda_{\mathrm{IR}}^2}{p^2}\right)^{-\epsilon_{\mathrm{IR}}}\Gamma(-\epsilon_{\mathrm{UV}}-\epsilon_{\mathrm{IR}})\left(-\frac{\Lambda_{\mathrm{UV}}^2}{p^2}\right)^{-\epsilon_{\mathrm{UV}}}. \tag*{(71)}

The limit of this function at ϵUV→0\epsilon_{\mathrm{UV}}\to0, ϵIR→0\epsilon_{\mathrm{IR}}\to0 is not well defined. For example, if the limit is taken along the line ϵUV=aϵIR\epsilon_{\mathrm{UV}}=a\epsilon_{\mathrm{IR}} with fixed aa, one would get an aa-dependent result:

lim⁡ϵIR→0I(p2;ϵIR,aϵIR)=−12(1+a)[log⁡2(−ΛIR2p2)+2alog⁡(−ΛIR2p2)log⁡(−ΛUV2p2)+a2log⁡2(−ΛUV2p2)].(72)\lim_{\epsilon_{\mathrm{IR}}\to0}I(p^2;\epsilon_{\mathrm{IR}},a\epsilon_{\mathrm{IR}})=-\frac{1}{2(1+a)}\left[\log^2\left(-\frac{\Lambda_{\mathrm{IR}}^2}{p^2}\right)+2a\log\left(-\frac{\Lambda_{\mathrm{IR}}^2}{p^2}\right)\log\left(-\frac{\Lambda_{\mathrm{UV}}^2}{p^2}\right)+a^2\log^2\left(-\frac{\Lambda_{\mathrm{UV}}^2}{p^2}\right)\right]. \tag*{(72)}

where we only kept the double logarithmic terms. The purpose of this subsection is to understand the correct prescription of evaluating the imaginary part of this function.

In order to understand the correct prescription, it is helpful to consider a simpler example: a scalar three-body decay. Indeed, the two point correlator Π(p)\Pi(p) contains three quark propagators, and thus the imaginary part of this function is related a three-body decay phase space integral. Therefore we consider the following Lagrangian

L=12(∂ϕ)2−mϕ22ϕ2+12(∂χ)2−mχ22χ2−λ6ϕχ3,(73)\mathcal{L}=\frac{1}{2}(\partial\phi)^2-\frac{m_\phi^2}{2}\phi^2+\frac{1}{2}(\partial\chi)^2-\frac{m_\chi^2}{2}\chi^2-\frac{\lambda}{6}\phi\chi^3, \tag*{(73)}

and study the three-body decay ϕ→3χ\phi\to3\chi. The two-loop diagram is evaluated as

iM2=iλ26∫d4l1(2π)4∫d4l2(2π)4DF(l1)DF(l2)DF(l1+l2−p),(74)i\mathcal{M}_2 = \frac{i\lambda^2}{6} \int\frac{\mathrm{d}^4 l_1}{(2\pi)^4} \int\frac{\mathrm{d}^4 l_2}{(2\pi)^4} D_F(l_1)D_F(l_2)D_F(l_1+l_2-p), \tag*{(74)}

where DF(p)D_F(p) is the Feynman propagator,

iDF(p)=ip2−mχ2+i0,(75)iD_F(p) = \frac{i}{p^2-m_\chi^2+i0}, \tag*{(75)}

and the solid line is χ\chi while the dashed line is ϕ\phi. The cutting rule tells us that

Im⁡M2=12λ26∫dΠLIPS,(76)\operatorname{Im}\mathcal{M}_2 = \frac{1}{2}\frac{\lambda^2}{6}\int\mathrm{d}\Pi_{\mathrm{LIPS}}, \tag*{(76)}

where dΠLIPS\mathrm{d}\Pi_{\mathrm{LIPS}} is the three-body Lorentz invariant phase space integral. This is expanded with respect to mχ2/mϕ2m_\chi^2/m_\phi^2 as

∫dΠLIPS=132π3[mϕ28+3mχ24(log⁡(4mχ2mϕ2)−1)+⋯ ],(77)\int\mathrm{d}\Pi_{\mathrm{LIPS}} = \frac{1}{32\pi^3}\left[\frac{m_\phi^2}{8}+\frac{3m_\chi^2}{4}\left(\log\left(\frac{4m_\chi^2}{m_\phi^2}\right)-1\right)+\cdots\right], \tag*{(77)}

where we used p2=mϕ2p^2=m_\phi^2. We now evaluate the left hand side of Eq. (B.21) in an analogous way as the main text. We may expand the propagator as

iDF(p)=ip2−mχ2=i(1p2+1p2mχ21p2+⋯ ),(78)iD_F(p) = \frac{i}{p^2-m_\chi^2}=i\left(\frac{1}{p^2}+\frac{1}{p^2}m_\chi^2\frac{1}{p^2}+\cdots\right), \tag*{(78)}

where we omit iεi\varepsilon for notational simplicity. In the coordinate space this is given by

iDF(x)=−14π2x2−mχ216π2Γ(εIR)(−ΛIR2x2)−εIR+⋯ .(79)iD_F(x)=-\frac{1}{4\pi^2x^2}-\frac{m_\chi^2}{16\pi^2}\Gamma(\varepsilon_{\mathrm{IR}})(-\Lambda_{\mathrm{IR}}^2x^2)^{-\varepsilon_{\mathrm{IR}}}+\cdots. \tag*{(79)}

The first order term in mχ2m_\chi^2 is IR divergent and this is analogous to our propagator with the CPCP-odd operator insertion in § 4.2. The two-loop amplitude is given by

iM2(p)=(iλ)26μUVdUV−4∫ddUVx eip⋅xiDF(x)iDF(x)iDF(x).(80)i\mathcal{M}_2(p)=\frac{(i\lambda)^2}{6}\mu_{\mathrm{UV}}^{d_{\mathrm{UV}}-4}\int\mathrm{d}^{d_{\mathrm{UV}}}x\,e^{ip\cdot x}iD_F(x)iD_F(x)iD_F(x). \tag*{(80)}

To the first order in mχ2m_\chi^2 we obtain

M2∣O(mχ2)=−λ2mχ2512π4I(p2;εIR,εUV).(81)\left.\mathcal{M}_2\right|_{\mathcal{O}(m_\chi^2)}=-\frac{\lambda^2m_\chi^2}{512\pi^4}I(p^2;\varepsilon_{\mathrm{IR}},\varepsilon_{\mathrm{UV}}). \tag*{(81)}

Note that we get exactly the same function I(p2;εIR,εUV)I(p^2;\varepsilon_{\mathrm{IR}},\varepsilon_{\mathrm{UV}}) here. Eq. (B.21) tells us that

Im⁡M2∣O(mχ2)=[12λ26∫dΠLIPS]O(mχ2)=λ2mχ2512π3log⁡(mχ2mϕ2),(82)\left.\operatorname{Im}\mathcal{M}_2\right|_{\mathcal{O}(m_\chi^2)}=\left[\frac{1}{2}\frac{\lambda^2}{6}\int\mathrm{d}\Pi_{\mathrm{LIPS}}\right]_{\mathcal{O}(m_\chi^2)}=\frac{\lambda^2m_\chi^2}{512\pi^3}\log\left(\frac{m_\chi^2}{m_\phi^2}\right), \tag*{(82)}

where we focus on the leading logarithmic term on the right hand side. Thus the correct prescription of evaluating the imaginary part of II is

Im⁡[I(p2;ϵIR,ϵUV)]=−πlog⁡(ΛIR2p2)(83)\operatorname{Im}\left[I(p^{2};\epsilon_{\mathrm{IR}},\epsilon_{\mathrm{UV}})\right] = -\pi\log\left(\frac{\Lambda_{\mathrm{IR}}^{2}}{p^{2}}\right) \tag*{(83)}

where we identify ΛIR2=mχ2\Lambda_{\mathrm{IR}}^{2}=m_{\chi}^{2} in the present case. This agrees with the Borel transformation formula in [52]. For the example in Eq. (B.17), this prescription is equivalent to neglecting the real part of the UV logarithm.

Our discussion clarifies the physical meaning of ΛIR\Lambda_{\mathrm{IR}} that appears in Eq. (37). This IR divergence originates from the phase space integral and is regulated by the mass of the daughter particles. In the neutron EDM case, the daughter particles are the constituent quarks. Therefore ΛIR\Lambda_{\mathrm{IR}} is identified with the mass of the constituent up and down quarks that we take ΛIR=300\Lambda_{\mathrm{IR}}=300 MeV in the main text.

References

  1. [1]M. Pospelov and A. Ritz, “Electric dipole moments as probes of new physics,” Annals Phys. 318 (2005) 119–169, arXiv:hep-ph/0504231.DOI
  2. [2]ACME Collaboration, V. Andreev et al., “Improved limit on the electric dipole moment of the electron,” Nature 562 no. 7727, (2018) 355–360.DOI
  3. [3]A. G. Grozin, I. B. Khriplovich, and A. S. Rudenko, “Electric dipole moments, from e to tau,” Phys. Atom. Nucl. 72 (2009) 1203–1205, arXiv:0811.1641 [hep-ph].DOI
  4. [4]B. Graner, Y. Chen, E. G. Lindahl, and B. R. Heckel, “Reduced Limit on the Permanent Electric Dipole Moment of Hg199,” Phys. Rev. Lett. 116 no. 16, (2016) 161001, arXiv:1601.04339 [physics.atom-ph]. [Erratum: Phys. Rev. Lett.119,no.11,119901(2017).]DOI
  5. [5]Y. Ema, T. Gao, and M. Pospelov, “Improved Indirect Limits on Muon Electric Dipole Moment,” Phys. Rev. Lett. 128 no. 13, (2022) 131803, arXiv:2108.05938 [hep-ph].
  6. [6]T. Aoyama et al., “The anomalous magnetic moment of the muon in the Standard Model,” Phys. Rept. 887 (2020) 1–166, arXiv:2006.04822 [hep-ph].DOI
  7. [7]Muon g-2 Collaboration, B. Abi et al., “Measurement of the Positive Muon Anomalous Magnetic Moment to 0.46 ppm,” Phys. Rev. Lett. 126 no. 14, (2021) 141801, arXiv:2104.03281 [hep-ex].
  8. [8]Muon (g-2) Collaboration, G. W. Bennett et al., “An Improved Limit on the Muon Electric Dipole Moment,” Phys. Rev. D 80 (2009) 052008, arXiv:0811.1207 [hep-ex].
  9. [9]Y. K. Semertzidis et al., “Sensitive search for a permanent muon electric dipole moment,” in KEK International Workshop on High Intensity Muon Sources (HIMUS 99), pp. 81–96. 12, 1999. arXiv:hep-ph/0012087.DOI
  10. [10]H. Iinuma, H. Nakayama, K. Oide, K.-i. Sasaki, N. Saito, T. Mibe, and M. Abe, “Three-dimensional spiral injection scheme for the g-2/EDM experiment at J-PARC,” Nucl. Instrum. Meth. A 832 (2016) 51–62.DOI
  11. [11]M. Abe et al., “A New Approach for Measuring the Muon Anomalous Magnetic Moment and Electric Dipole Moment,” PTEP 2019 no. 5, (2019) 053C02, arXiv:1901.03047 [physics.ins-det].arxiv.org/abs/1901.03047
  12. [12]A. Adelmann et al., “Search for a muon EDM using the frozen-spin technique,” arXiv:2102.08838 [hep-ex].arxiv.org/abs/2102.08838
  13. [13]E. M. Purcell and N. F. Ramsey, “On the Possibility of Electric Dipole Moments for Elementary Particles and Nuclei,” Phys. Rev. 78 (1950) 807–807.DOI
  14. [14]C. Abel et al., “Measurement of the Permanent Electric Dipole Moment of the Neutron,” Phys. Rev. Lett. 124 no. 8, (2020) 081803, arXiv:2001.11966 [hep-ex].DOI
  15. [15]V. M. Khatsimovsky, I. B. Khriplovich, and A. R. Zhitnitsky, “Strange Quarks in the Nucleon and Neutron Electric Dipole Moment,” Z. Phys. C 36 (1987) 455.
  16. [16]K. Fuyuto, J. Hisano, N. Nagata, and K. Tsumura, “QCD Corrections to Quark (Chromo)Electric Dipole Moments in High-scale Supersymmetry,” JHEP 12 (2013) 010, arXiv:1308.6493 [hep-ph].DOI
  17. [17]V. Cirigliano, W. Dekens, J. de Vries, and E. Mereghetti, “Is there room for CP violation in the top-Higgs sector?,” Phys. Rev. D 94 no. 1, (2016) 016002, arXiv:1603.03049 [hep-ph].DOI
  18. [18]V. Cirigliano, W. Dekens, J. de Vries, and E. Mereghetti, “Constraining the top-Higgs sector of the Standard Model Effective Field Theory,” Phys. Rev. D 94 no. 3, (2016) 034031, arXiv:1605.04311 [hep-ph].DOI
  19. [19]S. Weinberg, “Larger Higgs Exchange Terms in the Neutron Electric Dipole Moment,” Phys. Rev. Lett. 63 (1989) 2333.pubmed.ncbi.nlm.nih.gov/10040862
  20. [20]D. Chang, W.-Y. Keung, C. S. Li, and T. C. Yuan, “QCD Corrections to CP Violation From Color Electric Dipole Moment of b Quark,” Phys. Lett. B 241 (1990) 589–592.DOI
  21. [21]G. Boyd, A. K. Gupta, S. P. Trivedi, and M. B. Wise, “Effective Hamiltonian for the Electric Dipole Moment of the Neutron,” Phys. Lett. B 241 (1990) 584–588.DOI
  22. [22]M. Dine and W. Fischler, “Constraints on New Physics From Weinberg’s Analysis of the Neutron Electric Dipole Moment,” Phys. Lett. B 242 (1990) 239–244.
  23. [23]D. A. Demir, M. Pospelov, and A. Ritz, “Hadronic EDMs, the Weinberg operator, and light gluinos,” Phys. Rev. D 67 (2003) 015007, arXiv:hep-ph/0208257.DOI
  24. [24]F. Sala, “A bound on the charm chromo-EDM and its implications,” JHEP 03 (2014) 061, arXiv:1312.2589 [hep-ph].DOI
  25. [25]A. E. Blinov and A. S. Rudenko, “Upper Limits on Electric and Weak Dipole Moments of tau-Lepton and Heavy Quarks from e+ e− Annihilation,” Nucl. Phys. B Proc. Suppl. 189 (2009) 257–259, arXiv:0811.2380 [hep-ph].DOI
  26. [26]V. G. Baryshevsky, “The possibility to measure the magnetic moments of short-lived particles (charm and beauty baryons) at LHC and FCC energies using the phenomenon of spin rotation in crystals,” Phys. Lett. B 757 (2016) 426–429.
  27. [27]F. J. Botella, L. M. Garcia Martin, D. Marangotto, F. M. Vidal, A. Merli, N. Neri, A. Oyanguren, and J. R. Vidal, “On the search for the electric dipole moment of strange and charm baryons at LHC,” Eur. Phys. J. C 77 no. 3, (2017) 181, arXiv:1612.06769 [hep-ex].DOI
  28. [28]A. S. Fomin et al., “Feasibility of measuring the magnetic dipole moments of the charm baryons at the LHC using bent crystals,” JHEP 08 (2017) 120, arXiv:1705.03382 [hep-ph].DOI
  29. [29]E. Bagli et al., “Electromagnetic dipole moments of charged baryons with bent crystals at the LHC,” Eur. Phys. J. C 77 no. 12, (2017) 828, arXiv:1708.08483 [hep-ex]. [Erratum: Eur.Phys.J.C 80, 680 (2020)].DOI
  30. [30]S. Aiola et al., “Progress towards the first measurement of charm baryon dipole moments,” Phys. Rev. D 103 no. 7, (2021) 072003, arXiv:2010.11902 [hep-ex].DOI
  31. [31]H. Gisbert and J. Ruiz Vidal, “Improved bounds on heavy quark electric dipole moments,” Phys. Rev. D 101 no. 11, (2020) 115010, arXiv:1905.02513 [hep-ph].DOI
  32. [32]U. Haisch and G. Koole, “Beautiful and charming chromodipole moments,” JHEP 09 (2021) 133, arXiv:2106.01289 [hep-ph].DOI
  33. [33]D. J. Broadhurst, “Three loop on-shell charge renormalization without integration: Lambda-MS (QED) to four loops,” Z. Phys. C 54 (1992) 599–606.DOI
  34. [34]A. V. Smirnov and F. S. Chuharev, “FIRE6: Feynman Integral REduction with Modular Arithmetic,” Comput. Phys. Commun. 247 (2020) 106877, arXiv:1901.07808 [hep-ph].DOI
  35. [35]S. P. Martin and D. G. Robertson, “Evaluation of the general 3-loop vacuum Feynman integral,” Phys. Rev. D 95 no. 1, (2017) 016008, arXiv:1610.07720 [hep-ph].DOI
  36. [36]Y. Ema, T. Gao and M. Pospelov, in preparation.
  37. [37]M. Pospelov and A. Ritz, “CKM benchmarks for electron electric dipole moment experiments,” Phys. Rev. D 89 no. 5, (2014) 056006, arXiv:1311.5537 [hep-ph].
  38. [38]M. A. Shifman, A. I. Vainshtein, and V. I. Zakharov, “Remarks on Higgs Boson Interactions with Nucleons,” Phys. Lett. B 78 (1978) 443–446.
  39. [39]V. V. Flambaum, M. Pospelov, A. Ritz, and Y. V. Stadnik, “Sensitivity of EDM experiments in paramagnetic atoms and molecules to hadronic CP violation,” Phys. Rev. D 102 no. 3, (2020) 035001, arXiv:1912.13129 [hep-ph].DOI
  40. [40]A. G. Grozin, I. B. Khriplovich, and A. S. Rudenko, “Upper limits on electric dipole moments of tau-lepton, heavy quarks, and W-boson,” Nucl. Phys. B 821 (2009) 285–290, arXiv:0902.3059 [hep-ph].DOI
  41. [41]J. Ellis, N. Nagata, and K. A. Olive, “Uncertainties in WIMP Dark Matter Scattering Revisited,” Eur. Phys. J. C 78 no. 7, (2018) 569, arXiv:1805.09795 [hep-ph].DOI
  42. [42]M. Pospelov and A. Ritz, “Neutron EDM from electric and chromoelectric dipole moments of quarks,” Phys. Rev. D 63 (2001) 073015, arXiv:hep-ph/0010037.DOI
  43. [43]J. Hisano, J. Y. Lee, N. Nagata, and Y. Shimizu, “Reevaluation of Neutron Electric Dipole Moment with QCD Sum Rules,” Phys. Rev. D 85 (2012) 114044, arXiv:1204.2653 [hep-ph].DOI
  44. [44]T. Bhattacharya, V. Cirigliano, S. Cohen, R. Gupta, H.-W. Lin, and B. Yoon, “Axial, Scalar and Tensor Charges of the Nucleon from 2+1+1-flavor Lattice QCD,” Phys. Rev. D 94 no. 5, (2016) 054508, arXiv:1606.07049 [hep-lat].DOI
  45. [45]M. A. Shifman, A. I. Vainshtein, and V. I. Zakharov, “QCD and Resonance Physics. Theoretical Foundations,” Nucl. Phys. B 147 (1979) 385–447.
  46. [46]H.-x. He and X.-D. Ji, “QCD sum rule calculation for the tensor charge of the nucleon,” Phys. Rev. D 54 (1996) 6897–6902, arXiv:hep-ph/9607408.DOI
  47. [47]B. L. Ioffe and A. V. Smilga, “Nucleon Magnetic Moments and Magnetic Properties of Vacuum in QCD,” Nucl. Phys. B 232 (1984) 109–142.
  48. [48]B. L. Ioffe, “Calculation of Baryon Masses in Quantum Chromodynamics,” Nucl. Phys. B 188 (1981) 317–341. [Erratum: Nucl.Phys.B 191, 591–592 (1981)].
  49. [49]D. B. Leinweber, “QCD sum rules for skeptics,” Annals Phys. 254 (1997) 328–396, arXiv:nucl-th/9510051.DOI
  50. [50]V. M. Belyaev and B. L. Ioffe, “Determination of Baryon and Baryonic Resonance Masses from QCD Sum Rules. 1. Nonstrange Baryons,” Sov. Phys. JETP 56 (1982) 493–501.
  51. [51]B. L. Ioffe, “ON THE CHOICE OF QUARK CURRENTS IN THE QCD SUM RULES FOR BARYON MASSES,” Z. Phys. C 18 (1983) 67.
  52. [52]U. Haisch and A. Hala, “Sum rules for CP-violating operators of Weinberg type,” JHEP 11 (2019) 154, arXiv:1909.08955 [hep-ph].DOI
  53. [53]Particle Data Group Collaboration, P. A. Zyla et al., “Review of Particle Physics,” PTEP 2020 no. 8, (2020) 083C01.
  54. [54]J. Ellis, “TikZ-Feynman: Feynman diagrams with TikZ,” Comput. Phys. Commun. 210 (2017) 103–123, arXiv:1601.05437 [hep-ph].DOI
  55. [55]V. A. Novikov, M. A. Shifman, A. I. Vainshtein, and V. I. Zakharov, “Calculations in External Fields in Quantum Chromodynamics. Technical Review,” Fortsch. Phys. 32 (1984) 585.

Paper details

vlnc:W_hvmr25vy
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

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