Implication of ACME and muon g-2 on supersymmetry
KIAS-NCTS Joint Workshop (High1 resort)! 2014. 2. 10
Department of Physics and Astronomy! Seoul National University! Hyung Do Kim work in collaboration with Sang-Hui Im and Min-Seok Seo October 19, 2009 18:8 World Scientific Review Volume - 9in x 6in lepton
2 B. Lee Roberts
was di±cult to explain by any other means.a Stern proposed an experiment to study space quantization [7] to test the Sommerfeld quantum theory, where he presented the details of what we now call the Stern–Gerlach ex- periment. An atomic beam of silver atoms was to be projected through a gradient magnetic field where the net force on the magnetic dipole would separate the diÆerent magnetic quantum states. For a classical dipole the deflection would be continuous, since the direction of the dipole moment could have any value.b Over the next two years the famous experiments were carried out [8], and the two-band structure observed. By 1924, Stern and Gerlach con- cluded that to within 10%, the magnetic moment of the silver atom in its ground state was one Bohr magneton [9]. Their papers made no reference to the developments in spectroscopy, and in their 1924 review article, no conclusions beyond the magnetic moment were drawn from the two-band structure. Independently, in 1925 Uhlenbeck and Goudsmit [10] proposed the
October 19,“spinning 2009 18:8 electron” to explain World the Scientific fine-structure Review Volume observed - 9in x 6in in the anomalous lepton Zeeman eÆect in atomic spectra.c The fine-structure splitting can be un- derstood as thephysics interaction of dipole of the moments magnetic dipole moment of the electron with the magnetic field produced by the nuclear motion, which in the elec- tron’s rest frame appears to be orbiting about the electron. The electron’s Historical Introduction 7 magnetic dipole moment is along its spin and is given by q µ~ = g ~s, (1.1) moment mustmagnetic be along the spin. We2m can write an expression similar to whereEq. (1.1),q = e is the charge of the≥ particle¥ in terms of the magnitude of ± q the electron chargeelectrice, and thed~ = proportionality¥ ~s, constant g is the g(1.4)-factor for spin (which is sometimes written2mc as g ). In their second paper [11], ≥ ¥ s Uhlenbeckwhere ¥ andis a Goudsmit dimensionless conclude constant that that the isg-factor analogous for spin to g isin twice Eq. that (1.1). for orbitalWhile angular magnetic momentum, dipole moments however (MDMs) the calculated are a natural fine-structure property of charged splitting wasparticles then twice with as spin, large electric as the dipole observed moments splitting. (EDMs) Only are later forbidden in 1926, both when by Thomasparity and showed by time that reversal the factor symmetry. of 2 discrepancy between experiment and calculationThe search was a for kinematic an EDM eÆ datesect [12], back did to spin the start suggestion to become of Purcell an accepted and Ramsey [35] in 1950, well in advance of the paper by Lee and Yang [36], aIn his paper Compton acknowledges A.L. Parson (Smithsonian Misc. Collections, 1915) as firstthat proposing a measurement that the electronof the neutron was a spinning EDM ring would of charge. be a good Compton way modified to search this proposalfor parity to be violation a much smaller in the distribution nuclear force. “concentrated An experiment principally was near mounted its center.” at Compton’sOak Ridge paper [37] is almost soon thereafter unknown. which placed a limit on the neutron EDM bSee Allan Franklin,20 http://plato.stanford.edu/entries/physics-experiment/app5.html of dn < 5 10° e-cm, although the result was not published until after Stanford Encyclopedia£ of Philosophy,Appendix5,foranicediscussionputtingthe Stern–Gerlachthe discovery experiment of parity into violation. historical context. cIn theirOnce Nature parity paper violation [11] of 1926, was they established, acknowledge Landau Compton’s [38] independent and Ramsey suggestion [39] of spin.pointed out that an EDM would violate both P and T symmetries. This can be seen by examining the Hamiltonian for a spin one-half particle in the presence of both an electric and magnetic field,
= µ~ B~ d~ E.~ (1.5) H ° · ° · The transformation properties of E~ , B~ , µ~ and d~ are given in Table 1.1, and we see that while µ~ B~ is even under all three symmetries, d~ E~ is odd under · · both P and T. Thus the existence of an EDM implies that both P and T are not good symmetries of the interaction Hamiltonian, Eq. (1.5). In the context of CPT symmetry, an EDM implies CP violation.
Table 1.1. Transformation properties of the magnetic and electric fields and dipole mo- ments.
E~ B~ µ~ or d~ P -+ + C -- - T +- -
38 The Standard Model value for the electron (muon) EDM is 10° 36 ∑ e-cm ( 2 10° e-cm), well beyond the reach of experiments (which are ∑ £ 27 19 at the 1.6 10° (1.8 10° ) e-cm level). Likewise, the Standard-Model £ £ October 19, 2009 18:8 World Scientific Review Volume - 9in x 6in lepton
Historical Introduction 7 moment must be along the spin. We can write an expression similar to October 19, 2009Eq. 18:8 (1.1), World Scientific Review Volume - 9in x 6in lepton q d~ = ¥ ~s, (1.4) 2mc ≥ ¥ where ¥ is a dimensionless constant that is analogous to g in Eq. (1.1). While magnetic dipole momentsHistorical (MDMs) Introduction are a natural property of charged 7 particles with spin, electric dipole moments (EDMs) are forbidden both by parity and by time reversal symmetry. moment must be along the spin. We can write an expression similar to The search for an EDM dates back to the suggestion of Purcell and Eq. (1.1), Ramsey [35] in 1950, well in advance of the paper by Lee and Yang [36], that a measurement of the neutron EDMq would be a good way to search d~ = ¥ ~s, (1.4) for parity violation in the nuclear force.2mc An experiment was mounted at ≥ ¥ whereOak Ridge¥ is a [37] dimensionless soon thereafter constant which placed that is a analogous limit on the to neutrong in Eq. EDM (1.1). 20 of d < 5 10° e-cm, although the result was not published until after Whilen magnetic£ dipole moments (MDMs) are a natural property of charged the discovery of parity violation. particles with spin, electric dipole moments (EDMs) are forbidden both by Once parity violation was established, Landau [38] and Ramsey [39] parity and by time reversal symmetry. pointed out that an EDM would violate both P and T symmetries. This The search for an EDM dates back to the suggestion of Purcell and can be seen by examining the Hamiltonian for a spin one-half particle in Ramseythe presence [35] in of 1950, both an well electric in advance and magnetic of the field,paper by Lee and Yang [36], that a measurement of the neutron EDM would be a good way to search for parity violation in the nuclear= µ~ force.B~ d~ AnE.~ experiment was mounted(1.5) at H ° · ° · Oak Ridge [37] soon thereafter which~ ~ placed~ a limit on the neutron EDM The transformation20 properties of E, B, µ~ and d are given in Table 1.1, and of d < 5 10° e-cm, although the result was not published until after wen see that£ while µ~ B~ is even under all three symmetries, d~ E~ is odd under the discovery of parity· violation. · both P and T. Thus the existence of an EDM implies that both P and T areOnce not good parity symmetries violation of was the established, interaction Hamiltonian, Landau [38] Eq. and (1.5). Ramsey In the [39] pointedcontext out of CPT thatsymmetry, an EDM would an EDM violate implies bothCP Pviolation.and T symmetries. This can be seen by examining the Hamiltonian for a spin one-half particle in the presence of both an electricTabledipole 1.1. and moments Transformation magnetic field, properties of the magnetic and electric= fieldsµ~ andB~ dipoled~ E.~ mo- (1.5) ments.H ° · ° · The transformation propertiesE~ of E~ , BB~~ , µ~ andµ~ ord~d~are given in Table 1.1, and we see that while µ~ B~ isP even under-+ all three symmetries, + d~ E~ is odd under · C -- - · both P and T. Thus theT existence+- of an EDM - implies that both P and T are not good symmetries of the interaction Hamiltonian, Eq. (1.5). In the context of CPT symmetry,EDM (d) an implies EDM both implies P and T violationCP violation. 38 The Standard Modeland also value CP violation for the if CPT electron symmetry (muon)holds EDM is 10° 36 ∑ e-cm ( 2 10° e-cm),Table well beyond 1.1. Transformation the reach of experiments (which are ∑ £ 27 properties19 of the magnetic and at the 1.6 10° (1.8 10° ) e-cm level). Likewise, the Standard-Model £ £electric fields and dipole mo- ments.
E~ B~ µ~ or d~ P -+ + C -- - T +- -
38 The Standard Model value for the electron (muon) EDM is 10° 36 ∑ e-cm ( 2 10° e-cm), well beyond the reach of experiments (which are ∑ £ 27 19 at the 1.6 10° (1.8 10° ) e-cm level). Likewise, the Standard-Model £ £ October 19, 2009 18:8 World Scientific Review Volume - 9in x 6in lepton
Historical Introduction 5
to break the magnetic moment into two pieces: q~ g 2 µ =(1+a) where a = ° . (1.3) 2m 2 The first piece, predicted by the Dirac equation and called the Dirac mo- ment, is 1 in units of the appropriate magneton, q~/2m. The second piece is the anomalous (Pauli) moment [23], where the dimensionless quantity a is sometimes referred to as the anomaly. The development of radio frequency engineering and microwave tech- Octobernology 19, 2009 18:8 during the Second World World Scientific WarReview Volume was - 9in quickly x 6in put to use afterward lepton in the laboratory. In 1947, motivated by measurements of the hyperfine structure in hydrogen that obtained splittings larger than expected from the Dirac theory [24–26], Schwinger [27] showed that from a theoretical Historical Introduction 5 viewpoint these “discrepancies can be accounted for by a small additional electronto break spin the magnetic magnetic moment moment” into two that pieces: arises from the lowest-order radiative i correction to the Dirac moment.q~ In his paper,g 2 Schwinger points out three µ =(1+a) where a = ° . (1.3) important features of his new2m theory. 2 The first piece, predicted by the Dirac equation and called the Dirac mo- ment,The is 1 in new units Hamiltonian of the appropriate is superior magneton, toq the~/2m original. The second one piece in essen- is thetially anomalous three (Pauli) ways: moment it involves [23], where the the experimental dimensionless electron quantity a mass, is sometimesrather referred than the to unobservable as the anomaly. mechanical mass; an electron now Theinteracts development with of the radio radiation frequency field engineering only in and the microwave presence tech- of an ex- nologyternal during field the Second . . . the World interaction War was of quickly an electron put to with use afterward an external in thefield laboratory. is now In subject 1947, motivated to a finite by radiative measurements correction. of the hyperfine structure in hydrogen that obtained splittings larger than expected from In today’sthe Dirac language, theory [24–26], Schwinger Schwinger pointed [27] showed out that that from one a replaces theoretical the bare mass viewpoint these “discrepancies can be accounted for by a small additional andelectron charge spin with magnetic the physical moment” that (dressed) arises from mass the andlowest-order charge radiative (see Chapter 3 for additionalcorrection details). to the Dirac moment.i In his paper, Schwinger points out three Theimportant one-loop features contribution of his new theory. to a is shown diagrammatically in Fig. 1.1(b) October 19, 2009 18:8 World Scientific Review Volume - 9in x 6in lepton and has theThe value new Hamiltonianae = Æ/(2 isº superior) 0. to00116 the original, which one in essen- is independent of mass tially three ways: it involves' the experimental··· electron mass, and is therather same than for theaµ unobservableand aø . mechanical mass; an electron now October 19, 2009 18:8 World Scientific Review Volume - 9in x 6in lepton In the sameinteracts year, with the Kusch radiation and field Foley only in [29] the presence measured of an ex-ae with 4% precision, ternal field . . . the interaction of an electron with an external and found that the measured electron anomaly agreed well with Schwinger’s magnetic dipole momentfield is now of subject the to electron a finite radiative correction. prediction. They state that: “... the results can be described by g` = 1 and HistoricalIn today’s Introduction language, Schwinger pointed out that one replaces5 the bare mass g = 2(1.00119 0.00005).”j s and charge with± the physical (dressed) mass and charge (see Chapter 3 for iIn responseadditional to details). Nafe, et al. [24], Breit [28] conjectured that this discrepancy could be to break the magneticThomas moment precession into two pieces:Schwinger term explainedThe by one-loop the presence contribution of a to smalla is shown Pauli diagrammatically moment. It’s not in clearFig. 1.1(b) whether this paper Historical Introduction and has the value ae = Æ/(2º5) 0.00116 , which is independent of mass influencedq~ Schwinger’s work,g but2' in a footnote··· Schwinger states: “However, Breit has µ =(1+nota) correctlyand is thewhere drawn same for theaaµ= consequencesand°aø . . of his empirical(1.3) hypothesis.” j 2mIn the same year, Kusch2 and Foley [29] measured ae with 4% precision, The choice that g` =1andgs > 2wasguidedbytheoreticalprejudice.Themodern and found that the measured electron anomaly agreed well with Schwinger’s to break the magnetic momentThe into first two piece, pieces: predictedanomalous by magneticexperiments, the Dirac moment which equation a confine and a single called electron the in Dirac a Penning mo- trap, measure gs directly and fullyprediction. justify this They assumption. state that: “... the results can be described by g` = 1 and j ment, is 1 in units of the appropriategs = 2(1.00119 magneton,0.00005).”q~/2m. The second piece q~ g 2 ± µ =(1+ais) the anomalouswhere (Pauli)a = moment° .iIn response [23], whereto Nafe, et the al.(1.3)[24], dimensionless Breit [28] conjectured quantity that thisa discrepancy could be 2m 2 explained by the presence of a small Pauli moment. It’s not clear whether this paper is sometimes referred to as theinfluencedanomaly Schwinger’s. work, but in a footnote Schwinger states: “However, Breit has not correctly drawn the consequences of his empirical hypothesis.” The first piece, predicted by theThe Dirac development equation of and radio called frequencyj the engineering Dirac mo- and microwave tech- The choice that g` =1andgs > 2wasguidedbytheoreticalprejudice.Themodern ment, is 1 in units of the appropriatenology during magneton, the Secondq / World2mexperiments,. The War second which was confine quickly piece a single put electron to in use a Penning afterward trap, measure gs directly and ~ fully justify this assumption. is the anomalous (Pauli) momentin the [23], laboratory. where In the 1947, dimensionless motivated by quantity measurementsa of the hyperfine structure in hydrogen that obtained splittings larger than expected from is sometimes referred to as the anomaly. the Dirac theory [24–26], Schwinger [27] showed that from a theoretical The development of radioviewpoint frequency these engineering “discrepancies and can microwave be accounted tech- for by a small additional nology during the Second Worldelectron War spin was magnetic quickly moment” put that to use arises afterward from the lowest-order radiative in the laboratory. In 1947, motivatedcorrection to by the measurements Dirac moment.i In of his the paper, hyperfine Schwinger points out three structure in hydrogen that obtainedimportant splittings features of hislarger new than theory. expected from the Dirac theory [24–26], SchwingerThe [27] new showed Hamiltonian that is superiorfrom a to theoretical the original one in essen- viewpoint these “discrepancies can betially accounted three ways: for it involves by a small the experimental additional electron mass, rather than the unobservable mechanical mass; an electron now electron spin magnetic moment” thatinteracts arises with from the the radiation lowest-order field only radiative in the presence of an ex- correction to the Dirac moment.i Internal his paper, field . . . Schwinger the interaction points of an electron out three with an external important features of his new theory.field is now subject to a finite radiative correction. In today’s language, Schwinger pointed out that one replaces the bare mass The new Hamiltonian is superior to the original one in essen- and charge with the physical (dressed) mass and charge (see Chapter 3 for tially three ways: it involves the experimental electron mass, additional details). rather than the unobservable mechanical mass; an electron now The one-loop contribution to a is shown diagrammatically in Fig. 1.1(b) interacts with the radiation field only in the presence of an ex- and has the value a = Æ/(2º) 0.00116 , which is independent of mass ternal field . . . the interaction of an electrone with' an external··· and is the same for a and a . field is now subject to a finite radiative correction.µ ø In the same year, Kusch and Foley [29] measured ae with 4% precision, and found that the measured electron anomaly agreed well with Schwinger’s In today’s language, Schwinger pointed out that one replaces the bare mass prediction. They state that: “... the results can be described by g` = 1 and and charge with the physical (dressed) mass and chargej (see Chapter 3 for gs = 2(1.00119 0.00005).” additional details). ± iIn response to Nafe, et al. [24], Breit [28] conjectured that this discrepancy could be The one-loop contributionexplained to a is shownby the presence diagrammatically of a small Pauli moment. in Fig. It’s 1.1(b) not clear whether this paper and has the value a = Æ/(2ºinfluenced) 0.00116 Schwinger’s, which work, but is in independent a footnote Schwinger of mass states: “However, Breit has e not correctly drawn the consequences of his empirical hypothesis.” j ' ··· and is the same for aµ and aøThe. choice that g` =1andgs > 2wasguidedbytheoreticalprejudice.Themodern experiments, which confine a single electron in a Penning trap, measure gs directly and In the same year, Kusch andfully Foley justify this [29] assumption. measured ae with 4% precision, and found that the measured electron anomaly agreed well with Schwinger’s prediction. They state that: “... the results can be described by g` = 1 and g = 2(1.00119 0.00005).”j s ± iIn response to Nafe, et al. [24], Breit [28] conjectured that this discrepancy could be explained by the presence of a small Pauli moment. It’s not clear whether this paper influenced Schwinger’s work, but in a footnote Schwinger states: “However, Breit has not correctly drawn the consequences of his empirical hypothesis.” j The choice that g` =1andgs > 2wasguidedbytheoreticalprejudice.Themodern experiments, which confine a single electron in a Penning trap, measure gs directly and fully justify this assumption. I. INTRODUCTION
II. EXPERIMENTAL STATUS
For muon g 2[1], exp 10 aµ =11659208.9(5.4)(3.3) 10 ⇥ (1) SM 11 a =116591802(2)(42)(26) 10 µ ⇥ where the SM errors are due to the electroweak, lowest order hadronic, and higher order hadronic contributions so that
exp SM 11 a = a a =287(63)(49) 10 (2) µ µ µ ⇥ . For eEDM, recent ACME experiment reported[2]
29 d < 8.7 10 ecm (3) | e| ⇥ with 90% confidence whereas the SM from CKM matrix predicts[3] 2
(2) has the dimensions of a dipole, ecm, and is determined38 by a ratio of the atomic matrix elementsde of10 the CSP ecm.and (4) de operators. Over the years, significant⇠ theoretical e↵ort has gone into computing the fd and r coe↵ecients for di↵erent molecular and atomic species; see e.g. [9–13]. W q W W equivalentfor experimental electronIf status only one species of EDM lepton is used for flavorfrom an EDM measurement,violation, refer to [4]. q the e↵ects of CSP and de cannot be separated (see e.g. W W electron-nucleon[14, 15] for recent operator discussions). Since the experimental sensitivity is usually reported as an inferred limit on the e e electron EDM, it is convenient to parametrize the e↵ect Pospelov,of CRitzSP as a contribution1311.5537 from an equivalent EDM: III. BASIS PROPERTIES OF ELECTROMAGNETICFIG. 1. Electron DIPOLE EDM de induced OPERATOR by the CKM phase via a 3 equiv closed quark loop. The contributions shown are: O(↵W ↵s) d rC . (3) 2 3 e ⌘ SP (left panel, Fig. 1a), and O(↵ ↵ ) (right panel, Fig. 1b). Taking the three leading experimental limits on the elec- In the presencetron of EDM,the supersymmetry, we list the relevant r coe cients TeV [9–13], scale physicsto the conjugated gives weak an one-loop vertices. Nonzero contribution contributions to 2 20 only start at second order G , and are necessarily pro- r =1.2 10 ecm, F Tl ⇥ portional to the reduced Jarlskog/ invariant the electromagnetic dipole operator 20f such that the anomalous magnetic moment, electric rYbF =0.88 10 ecm, (4) ⇥ 2 5 D20 = s s2s3c1c2c3 sin 2.9 10 , (7) r =1.33 10 ecm. 1 dipole moment or charged leptonThO flavor⇥ violation arising fromJ can be enhanced' ⇥ compared where s and c aref the sines and cosines of the CKM an- Notice that although the f coe cients for these systems i i d gles in the Kobayashi-MaskawaD basis and is the complex actually di↵er widely, the r coe cients are approximately phase. The antisymmetric flavour structure of also to the Standard Modelthe same, prediction, reflecting the very similar and dynamical even can nature be of observed in upcoming experiments. leads to additional loop-level suppression of the EDMsJ of the P, T-odd perturbations to the electron Hamiltonian quarks and leptons in perturbation theory [16, 17] generated by both terms in (1). This leaves only mild As summarized in [5], we beginequiv with the general chirality-flipping electromagnetic dipole species-dependence in de . In this paper, we find that in the Standard Model A. Fundamental fermion EDMs operator, the CKM-induced CSP contribution dominates the di- rect contribution from de, and estimate it as In addition to the general constraints above, the EDM 18 C ( ) 10 , (5) operator for quarks and leptons breaks chiral symmetry, 1 SP J ⇠ µ⌫ 1 µ⌫ [ f ]ijfL µ⌫fR F [ † ]andijf thusR theµ⌫ coefL cientF must be at least linear in a chi- (5) where is the reduced Jarlskogi invariant.j Using the r f rality breakingi parameter.j In perturbation theory, this is J 2 D 2 D coe cients in (4), we can translate this into a character- generically the fermion mass mf itself. It turns out that istic CKM background to searches for the electron EDM, the antisymmetric flavour struture of actually ensures J that all 2-loop contributions to dq vanish [16], and the and let ' be the relative phaseequiv between38 Df operator and fermion mass d ( ) 10 ecm. (6) second-order weak exchanges need to be dressed with a e J ⇠ further gluonic loop. Thus, the d-quark EDM for ex- This is roughly nine orders of magnitude below the best ample arises only at 3-loop order [18, 19], and takes the current sensitivity to de, from ThO [4]. general form (6) The rest of this paper is organized' = as follows.Arg(m Inf⇤ theDf ) next section, we briefly review the CKM contributions ↵ ↵2 m m2 d(est)( ) e s W d c < 10 34ecm. (8) to fundamental fermions and other observable EDMs. In d 3 2 2 J ⇠ J (4⇡) mW mW Section 3, we turn to the CKM contribution to param- agnetic EDMs, and obtain the result (5). We2 finish with This estimates assigns ↵i/(4⇡) per corresponding loop, some concluding remarks in Section 4. and ignores additional numerical suppression or modest numerical enhancement by logarithms of quark mass ra- 2 tios. The factor of mc enters due to the flavour struc- 2. OVERVIEW OF EDMS FROM THE CKM ture of . The corresponding contribution to du is in- PHASE J 2 stead proportional to mums, and somewhat further sup- pressed. The most precise calculation of dq( ) can be In this section, we briefly review existing computations found in Ref. [19]. J of EDMs induced by the CKM phase. We will orga- EDMs of leptons are even further suppressed. A nize the discussion around a simple counting scheme, us- generic de diagram involves a quark loop with a minimum ing the basic symmetries to estimate the largest viable of four W -boson vertices. Such a loop can be attached to contribution to di↵erent classes of EDMs. In particular, the electron line either by two W -boson lines (Fig. 1a), CKM contributions to flavor-diagonal observables neces- at third order in the weak interaction, or via three virtual sarily vanish at first order in the weak interaction, due photons (Fig. 1b), at even higher loop order. discrepancy in anomalous magnetic moment of muon is the only evidence of BSM with a long history
from FNAL
BNL E821 from Hagiwara 2011 Advanced Cold Molecule Experiment (ACME) Harvard-Yale collaboration
Buffer Gas Beam Source Rotational Cooling Interaction Region
B E +V Electric Field Plates Magnetic Field Coils I Adjustable 298 K ThO2 Target Collimators I 50 K 4 K 16 K 690 nm +V Neon Buffer Gas PMT
-V Light Collection /4 and mirrors 20 GHz Pulsed Nd:YAG Microwave remixing
Vacuum 943 nm 1090 nm -V 1090 nm
Atmospheric Pressure Magnetic Shielding I. INTRODUCTION
II. EXPERIMENTAL STATUS
For muon g 2[1], exp 10 aµ =11659208.9(5.4)(3.3) 10 ⇥ (1) SM 11 a =116591802(2)(42)(26) 10 µ ⇥ where the SM errors are due to the electroweak, lowest order hadronic, and higher order hadronic contributions so that
exp SM 11 a = a a =287(63)(49) 10 (2) µ µ µ ⇥ . For eEDM, recent ACME experiment reported[2]
29 d < 8.7 10 ecm (3) | e| ⇥ with 90% confidence whereas the SM from CKM matrix predicts[3]
38 d 10 ecm. (4) e ⇠ for experimental status of lepton flavor violation, refer to [4].
III. BASIS PROPERTIES OF ELECTROMAGNETIC DIPOLE OPERATOR
In the presence of the supersymmetry, TeV scale physics gives an one-loop contribution to the electromagnetic dipole operator such that the anomalous magnetic moment, electric Df dipole moment or charged lepton flavor violation arising from can be enhanced compared Df to the Standard Model prediction, and even can be observed in upcoming experiments. BasicsAs summarized of the in anomalous [5], we begin with magnetic the general moment chirality-flipping electromagnetic dipole Electrostatic properties of a charged particle: operator,Lepton electromagnetic dipole moments Charge Q,Magnetic(dipole)momentµ~,Electricdipolemoment~d 1 µ⌫ 1 µ⌫ For a spin 1/2 particle:[ f ]ijfLi µ⌫fRjF [ f† ]ijfRi µ⌫fLjF (5) e 2 D 2 D 1 µ~ = g ~s, g =2(1 + a), a = (g 2) : anomalous magnetic moment and let ' be the2m relative phase between Df operator and2 fermion mass Dirac ' = Arg(m⇤ D ) (6) Long interplay between| {z experiment} andf theory:f structure of fundamental interactions In Quantum Field Theory (with C,P2 invariance):
γ(k) µ⌫ µ 2 i k⌫ 2 = ( ie)¯u(p 0) F1(k ) + F2(k ) u(p) 2 2m 3 p p’ 6 Dirac Pauli 7 4 5 | {z } | {z } F1(0) = 1 and F2(0) = a if F2 is real 2 ae : Most precise determination of ↵ = e /4⇡. aµ: Less precisely measured than ae , but all sectors of Standard Model (SM), i.e. QED, Weak and QCD (hadronic), contribute significantly. Sensitive to possible contributions from New Physics: 2 2 m` mµ a` 43000 more sensitive than ae [exp. precision factor 19] ⇠ mNP ) me ⇠ ! „ « „ « 3 Lepton electromagnetic dipole moments
The anomalous magnetic moment a (g 2)/2iscomingfromitsrealpart f ⌘
2 mf af = | | f cos ' (7) eQf |D | such that
e µ⌫ MDM = F2 µ⌫ F (8) L 4mf whereas the electric dipole moment is given by its imaginary part
d = sin ' (9) f |Df | such that
i = d F µ⌫. (10) LEDM 2 f µ⌫ 5 This operator is closely related to the lepton flavor violation. The branching ratio of l l i ! j is directly given by
m3 (l l )= li ( [ ] 2 + [ ] 2). (11) i ! j 16⇡ | Df ji| | Df ij|
+ Moreover, l l l l branching ratio can have a simple relation to l l provided the i ! j j j i ! j + photon panguin diagram which is made up of subdiagram and l l pair attached to the Df j i end of the photon line. For supersymmetry case, we have
2 Br(li 3lj) ↵ mli 11 ! = ln 2 . (12) Br(li lj ) 3⇡ ml 4 ! h j i Various one loop diagrams contributing to . In [5], it was assume that Df - Small mixing between gauginos and higgsinos : From Eq. (A1), we impose m2 Z 1 (13) M 2 µ2 ⌧ 1,2 - Sizable tan and then consider dominant diagram under such conditions.
1. The Chirality Conserving Diagrams In this case, the chirality flipping occurs in the external lepton mass insertion by Gordon identity.
3 Lepton electromagnetic dipole moments from supersymmetry Chirality conserving diagrams
(A) (B) ˜ ˜ W ± W3
g g g ˜ g lL 2 ⌫˜lL 2 lL lR lL 2 lL 2 lL lR
(C) (D) B˜ B˜
g ˜ g g ˜ g lL 1 lL 1 lL lR lL lR 1 lR 1 lR right-handed slepton in the loop Figure 1: Chirality conserving diagrams wherem ˜ corresponds to the heaviest particle’s mass in the loop. An important thing to note is that Dl by these one-loop chirality conserving diagrams cannot have a CP violating phase, in other words,
Im(Dl)=0, which means that EDM contribution from these diagrams vanishes. This is because gaug- ino mass insertion flips the lepton chirality so that the CP phase of gaugino mass cannot enter the diagrams. Remember that physical phases of the minimal supersymmetric stan- dard model (MSSM) in addition to the CKM phase of the standard model (SM) are given by
2 (9) Arg(MaµBµ⇤), Arg(au,d,lµBµ⇤), Arg(mfij˜ ), where m2 (i = j)denoteo↵-diagonal elements of sfermion soft supersymmetry breaking fij˜ 6 mass matrices. Then we can always choose a field basis that additional CP phases beyond the SM occur at only M ,a ,andm2 . Therefore, if we neglect the CP phases a u,d,l fij˜ arising from o↵-diagonal elements of a and m2 , gaugino masses are the only source to l fij˜
4 The anomalous magnetic moment a (g 2)/2iscomingfromitsrealpart f ⌘
2 mf af = | | f cos ' (7) eQf |D | such that
e µ⌫ MDM = F2 µ⌫ F (8) L 4mf
whereas the electric dipole moment is given by its imaginary part
d = sin ' (9) f |Df | such that
i = d F µ⌫. (10) LEDM 2 f µ⌫ 5 This operator is closely related to the lepton flavor violation. The branching ratio of l l i ! j is directly given by
m3 (l l )= li ( [ ] 2 + [ ] 2). (11) i ! j 16⇡ | Df ji| | Df ij|
+ Moreover, l l l l branching ratio can have a simple relation to l l provided the i ! j j j i ! j + photon panguin diagram which is made up of subdiagram and l l pair attached to the Df j i end of the photon line. For supersymmetry case, we have
2 Br(li 3lj) ↵ mli 11 ! = ln 2 . (12) Br(li lj ) 3⇡ ml 4 ! h j i Various one loop diagrams contributing to . In [5], it was assume that Df - Small mixing between gauginos and higgsinos : From Eq. (A1), we impose m2 Z 1 (13) M 2 µ2 ⌧ 1,2 -LeptonSizable tan electromagnetic dipole moments and then consider dominantfrom diagram supersymmetry under such conditions.
1. The Chirality Conserving Diagrams In this case, the chirality flipping occurs in the external lepton mass insertion by Gordon identity.
B˜⇤B˜ (LL, RR) : • h i 3 g2m 1 f (14) Df ⇠ 16⇡2m˜ 2
+,0 +,0 W˜ ⇤W˜ (LL) • h i g2m 2 f (15) Df ⇠ 16⇡2m˜ 2
2 2 2 Since g1/g2 =tan ✓W , This case is dominant over all the chirality conserving diagram
0 B˜⇤W˜ (LL) • h i Since bino and wino mixing is suppressed (no direct mixing; mixing through higgsinos only: see Eq. (A1)) by [m /(M 2 µ2)]2 so negligible. Z 1,2
2. The Chirality VIolating Diagrams In this case, the chirality flipping takes place through the a. Slepton LR mixing and b. Higgsino-lepton-slepton coupling.
a. Slepton LR mixing
B˜B˜ (LR) • h i Since the slepton LR mixing is proportional to m (A µ tan ), for small A term or f sizable tan , it is given by
g2M µm tan 1 1 f . (16) Df ⇠ 16⇡2m˜ 4
W˜ 0B˜ • h i Since bino and wino mixing is suppressed by [m /(M 2 µ2)]2 so negligible. Z 1,2 W˜ 0W˜ 0 (LR) • h i Forbidden since wino does not couple to right handed slepton.
b. Higgsino-lepton-slepton coupling
B˜H˜ 0 • h d i 4 contribute non-vanishing CP phases to the diagrams in Fig. (1) 1, and in turn they cannot appear in the final results because of the mentioned reason. However, there can be EDM contribution from chirality conserving diagrams at two-loop order as we shall see later. Lepton electromagnetic dipole moments B. Chirality violatingfrom diagrams supersymmetry
(C) (N1) ˜ ˜ H± B ˜ g v W ± 2 u,d
ylvuµ + alvd g y g ˜ ˜ g lL 2 ⌫˜lL l lR lL 1 lL lR 1 lR typically dominant diagram (N2) (N3) B˜ H˜ 0
H˜ 0 g1vu,d B˜ g1vu,d
y ˜ g g ˜ y lL l lR 1 lR lL 1 lL l lR
(N4) H˜ 0
W˜ 3 g2vu,d
g ˜ y lL 2 lL l lR
Chirality violating diagrams Figure 2: Chirality violating diagrams
In chirality violating diagrams, the lepton chirality is changed in internal propagators in the loop by Higgsino interactions or mixings by Higgs vevs. The dominant diagrams
1 Even if we introduce the right-handed neutrino sector, a CP phase in the PMNS matrix cancels in the diagram (A) in Fig. (1) because each vertex gives opposite sign of phase to each other.
5 B˜⇤B˜ (LL, RR) : • h i g2m 1 f (14) Df ⇠ 16⇡2m˜ 2
+,0 +,0 W˜ ⇤W˜ (LL) • h i g2m 2 f (15) Df ⇠ 16⇡2m˜ 2
2 2 2 Since g1/g2 =tan ✓W , This case is dominant over all the chirality conserving diagram
0 B˜⇤W˜ (LL) • h i SinceLepton bino and electromagnetic wino mixing is suppressed (no directdipole mixing; moments mixing through higgsinos only: see Eq. (A1)) by [m /(M 2 µ2)]2 so negligible. fromZ supersymmetry1,2
2. The Chirality VIolating Diagrams In this case, the chirality flipping takes place through the a. Slepton LR mixing and b. Higgsino-lepton-slepton coupling.
a. Slepton LR mixing
B˜B˜ (LR) • h i Since the slepton LR mixing is proportional to m (A µ tan ), for small A term or f sizable tan , it is given by
g2M µm tan 1 1 f . (16) Df ⇠ 16⇡2m˜ 4
W˜ 0B˜ • h i Since bino and wino mixing is suppressed by [m /(M 2 µ2)]2 so negligible. Z 1,2 W˜ 0W˜ 0 (LR) • h i Forbidden since wino does not couple to right handed slepton.
b. Higgsino-lepton-slepton coupling
B˜H˜ 0 • h d i 4 B˜⇤B˜ (LL, RR) : • h i g2m 1 f (14) Df ⇠ 16⇡2m˜ 2
+,0 +,0 W˜ ⇤W˜ (LL) • h i g2m 2 f (15) Df ⇠ 16⇡2m˜ 2
2 2 2 Since g1/g2 =tan ✓W , This case is dominant over all the chirality conserving diagram
0 B˜⇤W˜ (LL) • h i Since bino and wino mixing is suppressed (no direct mixing; mixing through higgsinos only: see Eq. (A1)) by [m /(M 2 µ2)]2 so negligible. Z 1,2
2. The Chirality VIolating Diagrams In this case, the chirality flipping takes place through the a. Slepton LR mixing and b. Higgsino-lepton-slepton coupling.
a. Slepton LR mixing
B˜B˜ (LR) • h i Since the slepton LR mixing is proportional to m (A µ tan ), for small A term or f sizable tan , it is given by
g2M µm tan 1 1 f . (16) Df ⇠ 16⇡2m˜ 4
W˜ 0B˜ • h i Since bino and wino mixing is suppressed by [m /(M 2 µ2)]2 so negligible. Z 1,2 W˜ 0W˜ 0 (LR) • h Leptoni electromagnetic dipole moments Forbidden since wino doesfrom not couple tosupersymmetry right handed slepton.
b. Higgsino-lepton-slepton coupling
B˜H˜ 0 • h d i There can be a simple (B˜ H˜ )mixingtermproportionalto4 m c s or g v and a d Z W 1 d (B˜ M insertion B˜ H˜ H˜ )proportionaltoM ( m s s )( µ)orM g v sin µ. 1 u d 1 Z W 1 2 For sizable tan , letter is dominant so that
g2M µm tan 1 1 f . (17) Df ⇠ 16⇡2m˜ 4
+,0 ,0 W˜ H˜ • h d i As the previous case, there can be two types of contribution. First, simple (W˜ H˜ ) d mixing term proportional to m c c or g v cos . Second, (W˜ M insertion W˜ Z W 2 2 H˜ H˜ )proportionaltoM (m s c )( µ)orM g v sin µ. For sizable tan the u d 2 Z W 2 2 latter is dominant:
g2M µm tan 2 2 f . (18) Df ⇠ 16⇡2m˜ 4
Relative Size
+,0 ,0 0 2 2 Typically, W˜ H˜ is dominant over B˜H˜ and B˜B˜ by a factor (g /g )(M /M ) • h d i h d i h i ⇠ 2 1 2 1 if M M . For instance, if M and M is the same at GUT scale, (M /M ) g2/g2. 2 & 1 1 2 2 1 ' 2 1
+ 0 0 Loop calculation shows that W˜ H˜ is larger than W˜ H˜ . Especially, for small µ • h d i h d i and large m2 , i.e. for x µ2/M 2 and y m2 /M 2 x 1,y > 1orx<1,y 1, lL ⌘ 2 ⌘ lL 2 ' ' + 0 0 the ratio (r W˜ H˜ / W˜ H˜ )betweentwographsbecomeslarger.Evenfor c ⌘h d i h d i x 1,y <1wefindr > 1: for y =0.1, r 3, y =0.01, r 2. For x =100,y =1, ' c c ' c ' r 4. c '
+ 2 2 2 For M M , W˜ H˜ / B˜B˜ 3(g /g )=3/ tan ✓ 10. But • 1 ' 2 h d i h i⇠ 2 1 W ' + a. For large µ, since W˜ H˜ 1/µ whereas B˜B˜ µ, B˜B˜ can be dominant. h d i/ h i/ h i 2 2 2 2 2 b. For mlL ,M2 ,µ >mlR ,M1 ,wehave
+ 2 W˜ H˜ eg mf h d i 2 tan f 64⇡2 m2 D ⇠ lL 2 (19) B˜B˜ eg1 mf µM1 h i tan f 96⇡2 m2 m2 D ⇠ lL lR
2 ˜ ˜ and so long as µM1/ml & 3/2, BB can be dominant. R h i 5 There can be a simple (B˜ H˜ )mixingtermproportionaltom c s or g v and a d Z W 1 d (B˜ M insertion B˜ H˜ H˜ )proportionaltoM ( m s s )( µ)orM g v sin µ. 1 u d 1 Z W 1 2 For sizable tan , letter is dominant so that
g2M µm tan 1 1 f . (17) Df ⇠ 16⇡2m˜ 4
+,0 ,0 W˜ H˜ • h d i As the previous case, there can be two types of contribution. First, simple (W˜ H˜ ) d mixing term proportional to m c c or g v cos . Second, (W˜ M insertion W˜ Z W 2 2 H˜ H˜ )proportionaltoM (m s c )( µ)orM g v sin µ. For sizable tan the u d 2 Z W 2 2 latter is dominant:
Lepton electromagneticg2M µm dipoletan moments 2 2 f . (18) from supersymmetryDf ⇠ 16⇡2m˜ 4
Relative Size
+,0 ,0 0 2 2 Typically, W˜ H˜ is dominant over B˜H˜ and B˜B˜ by a factor (g /g )(M /M ) • h d i h d i h i ⇠ 2 1 2 1 if M M . For instance, if M and M is the same at GUT scale, (M /M ) g2/g2. 2 & 1 1 2 2 1 ' 2 1
+ 0 0 Loop calculation shows that W˜ H˜ is larger than W˜ H˜ . Especially, for small µ • h d i h d i and large m2 , i.e. for x µ2/M 2 and y m2 /M 2 x 1,y > 1orx<1,y 1, lL ⌘ 2 ⌘ lL 2 ' ' + 0 0 the ratio (r W˜ H˜ / W˜ H˜ )betweentwographsbecomeslarger.Evenfor c ⌘h d i h d i x 1,y <1wefindr > 1: for y =0.1, r 3, y =0.01, r 2. For x =100,y =1, ' c c ' c ' r 4. c '
+ 2 2 2 For M M , W˜ H˜ / B˜B˜ 3(g /g )=3/ tan ✓ 10. But • 1 ' 2 h d i h i⇠ 2 1 W ' In general chargino(wino-higgsino)+ diagram dominates a. For large µ, since W˜ H˜ 1/µ whereas B˜B˜ µ, B˜B˜ can be dominant. h d i/ h i/ h i 2 2 2 2 2 b. For mlL ,M2 ,µ >mlR ,M1 ,wehave
+ 2 W˜ H˜ eg mf h d i 2 tan f 64⇡2 m2 D ⇠ lL 2 (19) B˜B˜ eg1 mf µM1 h i tan f 96⇡2 m2 m2 D ⇠ lL lR
2 ˜ ˜ and so long as µM1/ml & 3/2, BB can be dominant. R h i 5 There can be a simple (B˜ H˜ )mixingtermproportionaltom c s or g v and a d Z W 1 d (B˜ M insertion B˜ H˜ H˜ )proportionaltoM ( m s s )( µ)orM g v sin µ. 1 u d 1 Z W 1 2 For sizable tan , letter is dominant so that
g2M µm tan 1 1 f . (17) Df ⇠ 16⇡2m˜ 4
+,0 ,0 W˜ H˜ • h d i As the previous case, there can be two types of contribution. First, simple (W˜ H˜ ) d mixing term proportional to m c c or g v cos . Second, (W˜ M insertion W˜ Z W 2 2 H˜ H˜ )proportionaltoM (m s c )( µ)orM g v sin µ. For sizable tan the u d 2 Z W 2 2 latter is dominant:
g2M µm tan 2 2 f . (18) Df ⇠ 16⇡2m˜ 4
Relative Size
+,0 ,0 0 2 2 Typically, W˜ H˜ is dominant over B˜H˜ and B˜B˜ by a factor (g /g )(M /M ) • h d i h d i h i ⇠ 2 1 2 1 if M M . For instance, if M and M is the same at GUT scale, (M /M ) g2/g2. 2 & 1 1 2 2 1 ' 2 1
+ 0 0 Loop calculation shows that W˜ H˜ is larger than W˜ H˜ . Especially, for small µ • h d i h d i and large m2 , i.e. for x µ2/M 2 and y m2 /M 2 x 1,y > 1orx<1,y 1, lL ⌘ 2 ⌘ lL 2 ' ' Lepton electromagnetic+ 0 0 dipole moments the ratio (r W˜ H˜ / W˜ H˜ )betweentwographsbecomeslarger.Evenfor c ⌘h d i h d i x 1,y <1wefindrfrom> 1: for supersymmetryy =0.1, r 3, y =0.01, r 2. For x =100,y =1, ' c c ' c ' r 4. c '
+ 2 2 2 For M M , W˜ H˜ / B˜B˜ 3(g /g )=3/ tan ✓ 10. But • 1 ' 2 h d i h i⇠ 2 1 W ' + a. For large µ, since W˜ H˜ 1/µ whereas B˜B˜ µ, B˜B˜ can be dominant. h d i/ h i/ h i 2 2 2 2 2 b. For mlL ,M2 ,µ >mlR ,M1 ,wehave
+ 2 W˜ H˜ eg mf h d i 2 tan f 64⇡2 m2 D ⇠ lL 2 (19) B˜B˜ eg1 mf µM1 h i tan f 96⇡2 m2 m2 D ⇠ lL lR
2 ˜ ˜ and so long as µM1/ml & 3/2, BB can be dominant. R h i 5 For large mu or A, bino diagram dominates Lepton electromagnetic dipole moments from supersymmetry
2 2 2. Slepton universality : At leading order, mlL and mlR are proportional to identity, 2 2 2 2 respectively, say, mlL = m`LI and mlL = m`RI. The small violation from sleptoon universality can be parametrized by
2 2 (mlL )ij (mlR )ij ( ij)LL 2 , ( ij)RR 2 (23) ⌘ m`L ⌘ m`R
3. Proportionality : A term is universal such that LR mixing in the slepton is proportional to the lepton mass : m (A µ tan ). f
Under the assumption 2 and 3,
2 me me e µ,ae 2 aµ. (24) D ' mµ D ' mµ
Phase of Df What importantmuon in a (gis 43,0002) and eEDM,times the larger phase 'thanArg( electronm⇤ )isimportant.Such a µ ⌘ f Df phase should be expressed in tersm of basis independent one to have a physical meaning.
+ For W˜ H˜ which is the most dominant in many cases, we have h d i
g2 g2 ip i vd ip + iM2⇤ i vu⇤ iµ⇤ + p2 p2 (25) W˜ H˜ y h d i' f ⇣ (p2 ⌘ M 2)(p2⇣ µ 2) ⌘ | 2| | | where p p µ and p p µ. The first term comes from a simple (W˜ H˜ )mixingandhas ⌘ µ ⌘ µ d a phase Arg(y v ) = Arg(m )whereasthesecondtermcomesfrom(W˜ M insertion f d f 2 W˜ H˜ H˜ ) mixing and has a phase Arg(y M ⇤µ⇤v⇤). Therefore, Arg(y⇤v⇤ )iszero u d f 2 u f dDf for the former and is given by Arg(y⇤v⇤y M ⇤µ⇤v⇤)= Arg(M µv v )= Arg(M µ(Bµ)⇤). f d f 2 u 2 u d 2 0 B˜H˜ , the phase is just the sum of real and Arg(M µ(Bµ)⇤) and assuming GUT relation h d i 1 Arg(M ) = Arg(M ), the latter is the same as Arg(M µ(Bµ)⇤). 1 2 2 For B˜B˜ (LR), we have h i
˜ ˜ i[iyf (vdA µ⇤vu⇤)]i iM1⇤ BB (LR)= 2 2 2 2 2 2 . (26) h i (p m )(p m ) p M1 lL lR | | While the second term gives the same phase as the previous example, the first term gives
Arg(m⇤ ) = Arg(AM ⇤). f Df 1 The two loop phase coming from Barr-Zee type can be also considered[6].
7 2 2 2. Slepton universality : At leading order, mlL and mlR are proportional to identity, 2 2 2 2 respectively, say, mlL = m`LI and mlL = m`RI. The small violation from sleptoon universality can be parametrized by
2 2 (mlL )ij (mlR )ij ( ij)LL 2 , ( ij)RR 2 (23) ⌘ m`L ⌘ m`R
3. Proportionality : A term is universal such that LR mixing in the slepton is proportional to the lepton mass : m (A µ tan ). f LeptonUnder the assumption electromagnetic 2 and 3, dipole moments 2 me me e µ,ae 2 aµ. (24) from Dsupersymmetry' mµ D ' mµ
Phase of Df What important in (g 2) and eEDM, the phase ' Arg(m⇤ )isimportant.Such µ ⌘ f Df phase should be expressed in tersm of basis independent one to have a physical meaning.
+ For W˜ H˜ which is the most dominant in many cases, we have h d i
g2 g2 ip i vd ip + iM2⇤ i vu⇤ iµ⇤ + p2 p2 (25) W˜ H˜ y h d i' f ⇣ (p2 ⌘ M 2)(p2⇣ µ 2) ⌘ | 2| | | where p p µ and p p µ. The first term comes from a simple (W˜ H˜ )mixingandhas ⌘ µ ⌘ µ d a phase Arg(y v ) = Arg(m )whereasthesecondtermcomesfrom(W˜ M insertion f d f 2 W˜ H˜ H˜ ) mixing and has a phase Arg(y M ⇤µ⇤v⇤). Therefore, Arg(y⇤v⇤ )iszero u d f 2 u f dDf for the former and is given by Arg(y⇤v⇤y M ⇤µ⇤v⇤)= Arg(M µv v )= Arg(M µ(Bµ)⇤). f d f 2 u 2 u d 2 0 B˜H˜ , the phase is just the sum of real and Arg(M µ(Bµ)⇤) and assuming GUT relation h d i 1 Arg(M ) = Arg(M ), the latter is the same as Arg(M µ(Bµ)⇤). 1 2 2 For B˜B˜ (LR), we have h i
˜ ˜ i[iyf (vdA µ⇤vu⇤)]i iM1⇤ BB (LR)= 2 2 2 2 2 2 . (26) h i (p m )(p m ) p M1 lL lR | | While the second term gives the same phase as the previous example, the first term gives
Arg(m⇤ ) = Arg(AM ⇤). f Df 1 The two loop phase coming from Barr-Zee type can be also considered[6].
7 selectron is heavy for electron EDM. This is what we can think of immediately if we take an order one CP violation in mind. As we found, similar spectrum was studies already in [8, 9], even though consider such spectrum for 125GeV Higgs, not electron EDM. We may add Leptonsuch attempts electromagnetic to discussion on electron EDM, but instead,dipole it would moments be also good if we are more systematic. Therefore,from we supersymmetry need to discuss more on various parameters in (N)MSSM satisfying muon g 2, electron EDM, and 125GeV Higgs. For the Barr-Zee type 2-loop diagram to be sizeable in electron EDM, too heavy selectron suppressing 1-loop diagram was suggested. To see this, we may need to more quantitative. For convenience, we saaume that the gaugino and the Higgsino are in the similar mass scale, say, M M µ M. As we discussed so far, 1-loop contribution of MSSM to 1 ⇠ 2 ⇠ ⌘ electromagnetic dipole moment for selectron is roughly given by
↵ tan e,1loop 2 me (33) D ⇠ 4⇡ sin ✓W me˜
whereas 2-loop Barr-Zee diagram gives (e.g. Eq. (12a) in [6] : Eq. (21a) says that 2 2 2 2 F H (M2 /µ ,M2µ/mH )doesnotgiveenhancemente↵ect from M2µ/mH > 1duetoits log dependence)
↵ 2 sin cos e,2loop 2 me (34) D ⇠ 4⇡ sin ✓W M ⇣ ⌘ hence we can estimate
↵ m2 De,2loop e˜ cos2 . (35) ⇠ 4⇡ sin ✓ M 2 De,1loop W Hence, 2-loop Barr-Zee dominates over 1-loop diagram for
4⇡ sin ✓ 1/2 M M m > W 30 . (36) e˜ ↵ cos ⇠ cos ⇣ ⌘ To account for muon g 2, m is roughly 100 300GeV for tan 10. In this case, selectron µ˜ ⇠ ⇠ mass need to be heavier than 300M. For small tan , say, tan =3,thesmuonmassfor muon g 2maybeslightlylargeraboutafactorof1.5-1.7.Ontheotherhand,selectron need to be larger than 90M for Barr-Zee diagram could be dominant. In a renormalization group mixing point of view, such mass hierarchy is slightly awkward. The argument below can be found in [10]. Even though it considers the possoble problem in the natural SUSY, the same can be applied here since gauge interaction rather than Yukawa
9 Lepton electromagnetic dipole moments from supersymmetry interaction is important. Neglecting Yukawa couplings (this may be apparent for lepton sector) running of the smuon soft mass is roughly given by
2 d 2 ↵a 2 ↵a 2 m = C2(µ)M + C(e)C2(µ)m (37) dt µ˜ 4⇡ a 16⇡2 e˜ a a X X and two-loop contribution becomes dominant provided
4⇡ sin ✓ 1/2 m > W 2M 60M (38) e˜ ↵ 2 ⇠ 2 ⇣ ⌘ so surely when Barr-Zee diagram is dominant over 1-loop contribution, two loop RG e↵ect becomes dominant provided such large hierarchy was maintained from high energy scale. Then smuon mass can be, not just heavier, tachyonic as the given two loop RG e↵ect makes smuon mass negative.
V. VIABLE MODELS
In this section, we discuss how electron EDM, muon g 2andleptonflavorviolationsuch as µ e restrict the slepton flavor structure. To do this, we need to simplify expressions ! for each phenomena.
A. Slepton Mass Universality
As a simplest case, we consider the universal slepton case where m m m e˜L,R ' µ˜L,R ' ⌧˜L,R ⌘ m˜ . This can be found in many vanilla models such as mSUGRA, and from such models we can further assume thatm ˜ M . ⇠ 1,2 As discussed in Appendix B, muon g 2canbeestimatedas 2 9 tan 300GeV 10 m˜ 1 µ a =2.8 10 + µ ⇥ 20 m˜ 8 µ 10 m˜ ⇣ ⌘⇣ 2 ⌘ h i (39) 9 70GeV 10 m˜ 1 µ =2.8 10 + ⇥ m/˜ ptan 8 µ 10 m˜ ⇣ ⌘ h i + where the first term comes from W˜ H˜ whereas the second comes from B˜B˜ (LR). So h d i h i roughly, we conclude that slepton mass scale around 100GeV explains muon g 2: tan mµ˜ 100 GeV . (40) ' r 2 10 2 2 2 2 m , µ, M1,2 because the scale of (N2) is determined by 1/ max m , min(µ ,M1 ) while lR lR (C, N3, 4) are suppressed by 1/m2 in this limit and (N1) also becomes⇣ negligble by⌘ small f lL f mixing. 2 2 2 e 2 2 C where x3 = M2 /m and x4 = µ /m . These two relation is used to simplify a . In summary, we have III. EXPERIMENTAL STATUS 2 2 2 2 2 ↵m µM2 tan f (M /m ) f (µ /m ) a = µ 2 L L µ 4⇡ sin2 ✓ m2 M 2 µ2 In this section, we estimate theW magnitudeL of2 observable quantities related(B11) to the elec- 2 ⇣ 2 2 ⌘ 2 2 ↵mµµM1 tan fN (M1 /mR) fN (M1 /mL) + 2 2 2 2 2 tromagnetic dipole operators4⇡ cos in supersymmetric✓W (mR mL) modelsmR and comparemL them with current ⇣ ⌘ experimentalwhere limits. For example, at large tan & 10, dominant contributions to the Anomalous magneticx2 4x +3+2ln momentx of muon2 electromagnetic dipole operatorsf (x)= come from the,f chirality(1) = violating diagrams as we have (1 x)3 3 from supersymmetry (B12) seen. Also as explained in the previousx2 1 section,2x ln x (C)and(N11) will give dominant contri- f (x)= ,f(1) = . N (1 x)3 N 3 butions at large µ m˜,M limit. More specifically, from eq. (3), the muon anomalous l 1,2 In a special limit of mL,R M1,2 m˜ , we have a following simplified formula, magnetic moment aµ from supersymmetric' ⌘ contributions in this case becomes [? ] 2 9 tan 300GeV2 10 m˜ 1 µ aµ =2.8 10 9 70 GeV 10 m˜ + 1 µ (B13) a 2.8 ⇥10 20 m˜ 8 µ +10 m˜ , (13) µ ' ⇥ ⇣ m/˜ p⌘⇣tan ⌘ h 8 µ 10 im˜ which is used in [21]. ✓ Giudice◆ ✓ et al 2012 ◆ where we assume m˜ = M1,2 m˜ . In this expression, we can see the required scale lL,R ⌘ 70 GeV ofm/ ˜ ptan to explain the current deviation of aµ from the SM. From eq. (4), ⇠ Appendix C: Momentum Routing for J5. we can directly estimate theThe supersymmetric same diagram gives contribution muon EDM to the muon EDM,
Consider the one loop diagram for (g 2)µ is the one with boson and fermion internal dµ aµ 22 states and an external photon= is attachedtan to the10 bosoncm propagator.tan , For a moment, mass (14) e 2mµ ⇠ ⇥ insertion is neglected. Then we have to calculate 19 which is much smaller than the current bound on the muon EDM 10 cm [? ]. 4 d k i µ ⇠ 0 4 i( v + i a 5) 2 2 2 2 ieQs(p2+ p 2k2) Assuming slepton mass(2 universality⇡) (m(p0˜ k=) mmL ij, m ˜ = mR ij)anda-term propor- Z Lij s Rij (C1) i i(◆k mf ) tionality (a = A y ), the electron anomalous magnetici( ⇤ + i moment⇤ 5). and EDM are related lij l lij ⇥ (p k)2 m2 (k2 m2 ) v a s f to the muon’s by simple lepton mass scaling relations. Consider the part proportional to ( v v⇤ a a)mµ part where we are interested in. Then 2 we need to consider momentum integralme of 13 ae = 2 aµ 10 , mµ ⇠ (p + p0 2k)µ (15) [(p k)2 m2][(p k)2de m2][km2 e dmµ2 ] 24 0 s =s f 10 cm tan . (p0 ke)µ mµ e ⇠ ⇥ (p k)µ = 2 2 2 2 2 2 + 2 2 2 2 2 2 . [(p0 k) ms][(p k) ms][k mf ] [(p013 k) ms][(p k) ms][k mf ] The current uncertainty related to ae is about 10 both experimentally and theoretically (C2) [? ], so the estimated value is not yet verifiable. On the other hand, the ACME collabo- ration recently announced an order of magnitude18 smaller limit for the electron EDM than
7 2 2 2 2 m , µ, M1,2 because the scale of (N2) is determined by 1/ max m , min(µ ,M1 ) while lR lR (C, N3, 4) are suppressed by 1/m2 in this limit and (N1) also becomes⇣ negligble by⌘ small f lL f mixing. e
III. EXPERIMENTAL STATUS
In this section, we estimate the magnitude of observable quantities related to the elec- tromagnetic dipole operators in supersymmetric models and compare them with current experimental limits. For example, at large tan & 10, dominant contributions to the electromagnetic dipole operators come from the chirality violating diagrams as we have seen. Also as explained in the previous section, (C)and(N1) will give dominant contri- butions at large µ m˜,M limit. More specifically, from eq. (3), the muon anomalous l 1,2 magnetic moment aµ from supersymmetric contributions in this case becomes [? ] 2 9 70 GeV 10 m˜ 1 µ a 2.8 10 + , (13) µ ' ⇥ m/˜ ptan 8 µ 10 m˜ ✓ ◆ ✓ ◆ where we assume m˜ = M1,2 m˜ . In this expression, we can see the required scale lL,R ⌘ 70 GeV ofm/ ˜ ptan to explain the current deviation of a from the SM. From eq. (4), ⇠ µ we can directly estimate the supersymmetric contribution to the muon EDM,
dµ aµ 22 = tan 10 cm tan , (14) e 2mµ ⇠ ⇥ 19 which is much smaller than the current bound on the muon EDM 10 cm [? ]. ⇠ 2 2 2 2 Assuming slepton mass universality (mLij˜ = mL ij, mRij˜ = mR ij)anda-term propor- tionality (alij = Alylij), the electronSlepton anomalous universality magnetic moment and EDM are related to the muon’s by simple lepton mass scaling relations. 2 me 13 a = a 10 , e m2 µ ⇠ µ (15) d m d e e µ 29 24 the previous limit with de/e <= 8.7 10 10 cmcm [? ],tan which . is far smaller than the above e mµ ⇥e ⇠ ⇥ estimation. Thus assuming the simple universality13 and proportionality relations, we need The current uncertainty related to ae is about 10 both experimentally and theoretically 29 [the? ], supersymmetric so the estimatedACME value CP the phasebound is not previous on of yet electron verifiable. limit EDM with Onis d thee/e other < 8.7 hand,10 thecm ACME [? ], collabo- which is far smaller than the above ⇥ ration recently announcedestimation. an order of magnitude Thus assuming smaller the limit simple for the universality electron EDM and than proportionality relations, we need 4 We need , (16) the supersymmetric. 10 CP phase of for a_mu 7and ACME to be compatible to be compatible with the ACME result. 4 . 10 , (16) Breaking the slepton massSUSY CP universality, problem is more we can important get di now↵erent scaling relations for the electron electromagneticto dipole be compatible operators with with the the ACME muon’s. result. If we allow selectron mass much larger than smuon mass,Breaking the slepton mass universality, we can get di↵erent scaling relations for the
2electron2 electromagnetic2 dipole operators with the muon’s. If we allow selectron mass m˜ me m˜ 13 ae 2 2 aµ 2 10 ⇠ me˜muchmµ larger⇠ m thane˜ ⇥ smuon mass, 2 2 (17) de m˜ me dµ m˜ 24 2 2 2 29 10 mcm˜ metan