Lepton flavor-violating transitions in effective field theory and gluonic operators

Alexey A. Petrov1, 2 and Dmitry V. Zhuridov1

1
Department of Physics and Astronomy, Wayne State University, Detroit, MI 48201, USA
2
Michigan Center for Theoretical Physics, University of Michigan, Ann Arbor, MI 48109, USA
(Dated: August 9, 2017)
Lepton flavor-violating processes offer interesting possibilities to probe new physics at multi-TeV
scale. We discuss those in the framework of effective field theory, emphasizing the role of gluonic
operators. Those operators are obtained by integrating out heavy quarks that are kinematically
inaccessible at the scale where low-energy experiments take place and make those experiments
sensitive to the couplings of lepton flavor changing neutral currents to heavy quarks. We discuss
constraints on the Wilson coefficients of those operators from the muon conversion + (A, Z)
e + (A, Z) and from lepton flavor-violating tau decays with one or two hadrons in the final state,
arXiv:1308.6561v3 [hep-ph] 7 Aug 2017

e.g. () and + with = , e. To illustrate the results we discuss explicit examples

of constraining parameters of leptoquark models.

I. INTRODUCTION The most important scales for this study are the scales
associated with heavy quark masses, mt , mb , and mc .
In the framework of EFT one must integrate out heavy
As follows from observations of neutrino oscillations,
quarks that are not kinematically accessible at the scale
there is a good evidence that the individual lepton flavor
where the experiment takes place, resulting in changes
is not conserved. Explicit calculations of the standard
of Wilson coefficients of quark and gluon operators.
model (SM) rates for the charged lepton-flavor violating
The relation between all those scales can be done with
(LFV) transitions indicate that those are tiny [1, 2], well
the help of renormalization group, keeping track of which
beyond the capabilities of current and currently planned
degrees of freedom are kept and which are integrated out.
experiments. Yet, many models of beyond the stan-
We shall list the most important operators for our anal-
dard model (BSM) physics do not exclude relatively large
ysis below.
rates for such transitions, so experimental and theoreti-
cal studies of LFV processes like e, () or
+ (A, Z) e + (A, Z) could provide a sensitive test A. Quark operators
of those BSM schemes.
The language of effective field theory (EFT) is very The lowest dimensional local operators that contribute
useful in the studies of flavor violating processes for sev- to lepton-flavor violating transitions without photons in
eral reasons. First, it allows to probe the new physics the final state [5] have operator dimension 6. There are,
(NP) scale generically, without specifying a particular in general, twelve types of operators that can be con-
model of NP. Studies of specific models in this frame- structed,
work are equivalent to specifying Wilson coefficients of
effective operators. Second, EFT allows for studies of 12
(6) 1 X X q1 2 q1 2
relative contributions of various operators and may even L1 2 = C Qi + H.c., (1)
2 i=1 q i
provide clues as to what experiments need to be done
to discriminate among different possible models of new
physics [3]. where is a high scale of new physics, and Ciq1 2 are
dimensionless Wilson coefficients. The four fermion op-
Interactions of flavor-changing neutral currents erators can be split into three classes which we define
(FCNC) of leptons with hadrons, either in muon con- according to their Dirac structure,
version or in tau or meson decays can be described in (i) scalar operators,
terms of effective operators of increasing dimension [3].
In order to set up an EFT calculation, however, one Qq
1
1 2
= (1R 2L ) (qR qL ),
must first discuss the multitude of scales present in
Qq
2
1 2
= (1R 2L ) (qL qR ),
lepton FCNC transitions. The highest scale, which we
denote as , is the scale associated with new physics Qq
3
1 2
= (1L 2R ) (qR qL ), (2)
that generates the FCNC interaction. There could Qq 1 2
= (1L 2R ) (qL qR ),
4
be many ways to generate the flavor-changing neutral
current of leptons, yet, below the scale any heavy new where (q) is the SM charged lepton (quark).
physics particles are integrated out resulting only in a The scalar operators above are defined below the scale
few effective operators [4]. We shall keep track of the of electroweak symmetry breaking (EWSB) in the stan-
leading contribution due to NP which, below the scale , dard model as they are not invariant under electroweak
is proportional to 1/2 . The second highest scale is the SU (2)L -symmetry. The proper definition of those oper-
one associated with electroweak symmetry breaking, v. ators above EWSB scale should include Higgs doublet
 2

fields H. The operators of Eq. (2) result from the sub- 2 1 2 1

stitution H v and redefinition of proper Wilson coef-
ficients [6] to scale out quark or lepton Yukawa coupling,
q q q q
which would result in (dimensionless) factor of GF m mq
in front of the scalar operators.
q
These mass factors properly suppress flavor-violating q
transitions of the first generation of quarks and lep-
tons that are well constrained experimentally. Notice,
however, that they are not model-universal. For ex- g g g g
ample, models with FCNC Higgs boson interactions of-

ten employ factors of m1 m2 /v (so called Cheng-Sher FIG. 1: Feynman graphs for the calculation of matching co-
ansatz [7]) to suppress flavor-changing lepton currents, efficients of gluonic operators. The large dots indicate the
while generic leptoquark or R-parity violative supersym- effective vertices described by Eq. (1).
metric models do not have any factors of mass, relying on
the smallness of coupling constants to suppress those ef-
fective operators of dimension 7:
fects [8]. In the following we shall absorb all mass factors
into the definition of Wilson coefficients Ciq1 2 . L a a
O11 2 = 1R 2L G G ,
(ii) vector operators, 4s
L a e a
O21 2 = 1R 2L G G ,
Qq 1 2
= (1L 2L )(qL qL ), 4s
5
L a a
Qq
6
1 2
= (1L 2L )(qR qR ), O31 2 = 1L 2R G G , (6)
4s
Qq
7
1 2
= (1R 2R )(qL qL ), (3) L a e a
O41 2 = 1L 2R G G ,
Qq
8
1 2
= (1R 2R )(qR qR ), 4s
where a = 1, . . . , 8 is gluon color index, L = b2s /(2)
and (iii) tensor operators, is the one-loop beta function of three flavor QCD with
b = 11 2nL /3 (nL = 3 is the number of light quarks)
Qq
9
1 2
= (1R 2L )(qR qL ), and s = gs2 /(4);
Qq
10
1 2
= (1R 2L )(qL qR ), Ga = Aa Aa + gs f abc Ab Ac (7)
Qq
11
1 2
= (1L L
2R )(qR qL ), (4) is the gluon strength tensor, and
Qq
12
1 2
= (1L
2R )(qL qR ), ea = 1 Ga
G (8)
2
All quark flavors need to be considered, but the operator is the dual one. In Eq. (5) we do not include the operators
basis needed to describe a particular experiment could with dimension higher than 7. It can be easily seen that
include a smaller number of operators. there are no other possibilities besides the four operators
in Eqs. (6).
By calculating the loop diagrams in Fig. 1, using the
B. Gluonic operators standard methods [10], the coefficients ci1 2 can be ex-
pressed through Ciq1 2 in Eq. (1) as
The low energy experiments such as muon conversion 2 X I1 (mq ) q1 2
+ N e + N or tau decay () have a naturally c11 2 = (C1 + C2q1 2 ), (9)
9 mq
q=c,b,t
defined scale of the order of the mass of heavier lepton.
In order to write appropriate set of effective operators 2i X I2 (mq ) q1 2
c21 2 = (C1 C2q1 2 ), (10)
at that scale one must integrate out quarks with masses 9 mq
q=c,b,t
above that scale [9].
2 X I1 (mq ) q1 2
The flavor changing Lagrangian for the effective ver- c31 2 = (C3 + C4q1 2 ), (11)
tices with 1 , 2 and two gluon external legs at the en- 9 mq
q=c,b,t
ergies lower than heavy quarks masses can be written 2i X I2 (mq ) q1 2
as c41 2 = (C3 C4q1 2 ), (12)
9 mq
q=c,b,t
4
(7) 1 X 1 2 1 2 where the coefficients (see Ref. [11] for I1 and Appendix
L1 2 = 2 c Oi + H.c., (5)
i=1 i A) at the leading order are
1 1
where ci are the Wilson coefficients, and Oi are the ef- I1 = , I2 = (13)
3 2
 3

Note, as previously discussed, that while the Wilson coef- the energy m Eb (neglecting the atomic recoil energy
ficients in Eq. (9)-(12) explicitly contain factors of 1/mq , of a muonic atom, see [19]). Following Ref. [20] the -e
in many models the coefficients Ciq1 2 contain factors of conversion amplitude can be written as
mq , which we absorbed as part of their definition. Also, Z h
e 1 3 (e) ()
we do not explicitly write out contributions to Wilson MN N = d x c1 ,W PL 1s
coefficients due to possible colored heavy states that are 2
 L a a
not SM quarks; those contributions would result in ad- (e) ()
+ c3 ,W PR 1s hN | G G |N i
ditive modifications of Eqs. (9)-(12). Also, in this paper, 4s
 
we ignored running of ci in between different scales. (e) () (e)
+ c2 ,W PL 1s + c4 ,W PR 1s
()
Integrating out heavy particles could also result in 
higher dimensional gluonic operators, as would happen L a e a
hN | G G |N i , (16)
for vector-like dimension-6 operators. For instance, a set 4s
of operators of dimension 8 can be written as
where hN | and |N i are the final and initial states of the
(8) L a a nucleus, respectively; the 1s initial muon wave function
O1 = (1L 2L ) G G
4s 1/2
!
(8) L a a () g 1
O2 = (1L 2L ) G G (14) 1s = 1/2 (17)
4s if 1
(8) L ea
O3 = (1L 2L ) G G is normalized to 1 and corresponds to the quantum num-
4s
bers = 1 and = 1/2 of the operators
(8) (8) !
Another three operators, O4 O6 could be obtained
l +1 0
by substituting left-handed lepton fields with the right- K= (18)
handed ones. Here we shall concentrate on the operators 0 ( l + 1)
of dimension 7, leaving analysis of higher-dimensional op-
erator for future work. and jz , respectively, where l is the orbital angular mo-
This paper is organized as follows. In section II we mentum; and = 1 final electron wave functions
reexamine constraints on the Wilson coefficients of oper- !
1/2
ge 1
ators O1e and O3e from -e conversion data. We consider 1/2(e)
1,W = 1/2 (19)
constraints on Wilson coefficients of operators O1 O4 ife 1
from tau decays in section III. As an example, in sec-
tion IV we consider how our constraints translate into and
constraints on couplings of LFV lepton currents with !
1/2
1/2(e) ge+ 1
heavy quarks in leptoquark models. We conclude in sec- 1,W = 1/2 (20)
tion V . ife+ 1

are normalized as
Z
II. CONSTRAINTS FROM -e CONVERSION (e) (e)
d3 x ,W (x) ,W (x) = 2 (W W ),(21)
Muon conversion on a nucleus [2, 1217] offers a sensi-
where W is the energy. The electron mass was neglected
tive probe of new physics and a nice possibility to study it
in Eqs. (19) and (20) so that ge+ = ife and ife+ = ge .
experimentally providing an interesting interplay of par-
Using the normalization
ticle and nuclear physics effects. The number of relevant
operators in Eqs. (1) and (5) is reduced if one only con- Z 1 Z 2

siders coherent + N e + N transitions1 [3]. d cos d
= (22)
The initial state in the -e conversion process 1 0

of the eigenfunctions of ( l + 1) and jz , we have

+ (A, Z) e + (A, Z) (15)
(e) () 1

is the 1s state of the muonic atom with the binding en- 1,W P 1s = g g fe f
2 e
ergy Eb , and the final electron state is the eigenstate with (e) ()
= 1,W PR 1s , (23)
(e) () 1

1,W PL 1s = ge g fe f (24)
2
1 There are also important non-local contributions from the oper-
ators governing e transitions with the photon attached to with = L, R.
a nucleus. Those contributions are well known [18] and will not The pseudoscalar nucleon current couples to the nu-
be discussed here. clear spin leading to incoherent contribution [21]. The
 4

matrix element in Eq. (16) relevant to the coherent con-

TABLE I: Nucleon densities model parameters, and the over-
version process (N = N ) can be expressed by the proton 5/2
lap integrals in the unit of m for several nuclei
(p) and neutron (n) densities in a nucleus as

L a a Nucleus Model c, fm z, fm S (p) S (n)

hN | G G |N i
4s 48
22 Ti FB 0.0368 0.0435
9h i 197
79 Au 2pF 6.38 0.535 0.0614 0.0918
= ZG(g,p) (p) + (A Z)G(g,n) (n) , (25)
2
where A and Z represent mass and atomic number of
the nucleus, and the matrix element of gluon operator
expansion (FB)
between the nucleon states is defined as
P
s a a av sin(vrR1 )/(vrR1 ) for r R
G(g,N ) = hN | G G |N i (26)
4 (r) = v (34)
0 for r > R
with N = n, p. This scalar matrix element can be calcu-
lated by using the trace of the energy-momentum tensor are given in Ref. [24, 25].
[22] and applying the flavor SU (3) symmetry. Since The branching ratio for -e conversion on a nucleus N
is flavor symmetric, the proton and neutron scalar is defined as
matrix elements are equal. For the strange-quark sigma
term s ms hp|ss|pi = 50 MeV the numerical result is N
Be conv ( N e Ng.s. )/capt ( N ), (35)
G(g,N ) = 189 MeV [23].
The nucleon densities are assumed spherically symmet- where g.s. stands for ground state. The SM muon cap-
N
ric and normalized as ture rates [20, 26], the upper bounds on Be , and the
Z 48 197
binding energies of 22 Ti and 79 Au are given in Table II
dr4r2 (N ) (r) = 1. (27) with the respective references. The upper bounds on the
0

As usual for spherical nuclei, the two-parameter Fermi N

TABLE II: Muon capture rates, bounds on Be , and binding
(2pF) charge distribution is used energies for several nuclei N
0
(r) = , (28)
1 + exp[(r c)/z] Nucleus capt ( N ), s1 N
Be (90% C.L.) Eb , MeV
48 6
where c is the half-density radius. 22 Ti 2.59 10 4.3 1012 Ref. [28] 1.25 Ref. [29]
197
The formula for the coherent conversion rate can be 79 Au 13.07 106 7 1013 Ref. [19] 10.08 Ref. [19]
written as
4

conv (N eN ) = 4
|c1 |2 + |c3 |2 a2N , (29) parameters of the Lagrangian in Eq. (5) for one nonzero

coefficient c at a time are given in Table III. It shows that
where the bound on -e conversion rate on gold gives the best
limit (see also Ref. [27]).
aN = G(g,p) S (p) + G(g,n) S (n) . (30)

The overlap integrals are defined as [20] III. CONSTRAINTS FROM DECAYS.
Z
1 The analysis presented above only deals with
S (p) = drr2 Z(p) (ge g fe f ), (31)
2 2 0 experimentally-interesting coherent e conversion. As
Z
1 a result, no parity-violating operator give any contribu-
S (n) = drr2 (A Z)(n) (ge g fe f ).(32) tion to the experimental transition rates. Moreover, we
2 2 0
had to resort to models to describe nuclear effects af-
The parameters of model 2pF of nucleon densities in fecting conversion rates. It might be advantageous to
Eq. (28) [24], and the overlap integrals in Eqs. (31) and use other experimental observables to study LFV new
(32) [20] for the same distributions physics couplings to heavy quarks via gluonic operators.
LFV tau decays offer such opportunity. Besides, anal-
(p) (r) = (n) (r) (r) (33) yses of tau decays have different theoretical uncertain-
ties than muon conversion calculations; in fact, one can
of neutrons and protons in the nuclei 48 197
22 Ti and 79 Au are use chiral symmetry and low energy theorems to pro-
shown in Table I. The parameters of the Fourier-Bessel vide model-independent evaluations of operator matrix
 5

where Q = p p is the momentum transfer to the

TABLE III: Upper bounds on the parameters of the La-
hadronic system and p are the 4-momenta of M . Note
grangian in Eq. (5) from muon conversion experiments. (q)
that GM (0) = 1 [32]. Just like before, flavor content of
the operators should match that of the final state mesons.
Coefficient Bound on |ce 2
i |/ , GeV
3

48 197
The matrix elements of gluonic operators Eq. (6) are
conversion on 22 Ti conversion on 79 Au easiest estimated in the chiral limit, where mu = md =
c1 2.5 1011 1.2 1011 ms = mM = 0. In that limit, a low-energy theorem
c2 states that [33]
11
c3 2.5 10 1.2 1011 s a a 2
c4 hM + M | G G |0i = q 2 , (39)
4 9
We do not expect the results to change much away from
the chiral limit, so shall use this estimate in what follows.
elements. While the tau decays have been studied in It is these operators that we are most interested in the
a variety of models [8], to the best of our knowledge paper. Finally, parity invariance of strong interactions
gluonic operator contribution (and thus constraints on implies that
heavy quark couplings from those decays) has not been s a ea
previously considered. In what follows we shall use tau hM + M |q5 q|0i = hM + M | G G |0i
4
decays to constrain matrix elements of gluonic operators.
= hM + M |q 5 q|0i = 0. (40)

With the definitions above one can calculate the differ-

A. Parity-conserving gluonic operators ential decay rate for the decay M + M . For the
scalar and gluonic operators one has
Complimentary to muon conversion experiments con-
d( M + M ) m h i
sidered in section II, parity-conserving operators can also 2 2
2
= 3 4
|AMM | + |BMM |
be probed in lepton-flavor-violating tau decays dq 32(2)
M + M , where = , e and M = , () , K [30, 31]. s  2
In what follows, let us concentrate on the case of three- 4m2M q2
1 2 1 2 , (41)
body decays + and K + K , which are q m
the most interesting experimentally since all the final
particles are charged. Transitions to other states (like where we set m = 0 and defined
) can be obtained by employing flavor SU (3)
2c 1 X  q 
symmetry relations. AMM = 1 2
q + C1 + C2q qM B0 ,
In order to constrain the Wilson coefficients of the op- 9 2
q=u,d,s
erators in Eq. (2), Eq. (3), and Eq. (6) one needs to con- 2c 1 X  q 
strain hadronic matrix elements. For the scalar operators BMM = 3
q2 + C3 + C4q qM B0 .(42)
9 2
we shall follow [30] to state q=u,d,s

+ +
h |qq|0i = hK K |qq|0i = qM B0 , (36) Integrating Eq. (41) we obtain constraints on c1 and c3

listed in Table IV. Finally, for completion, we present the

where for charged final states qM = 1 if the flavor contribution to the differential decay rate due to vector
of the quark field q in the operator matches the fla- operators,
vor content of the meson and zero otherwise. For ex-
ample, hK + K |ss|0i = hK + K |uu|0i = B0 , while dV ( M + M ) m3 h i
2 2
= |C | + |D |
hK + K |dd|0i = 0. Matrix elements for other light fi- dq 2 768 3 4
MM MM

nal states can be related to Eq. (36) via flavor SU (3)  3/2  2  
relations [30], e.g., 4m2M q2 q2
1 2 1 2 1 2 2 , (43)
q m m
3
3h8 8 |uu|0i = h8 8 |qq|0i = B0 . (37) where we set m = 0 and defined
4
Note that B0 = 1.96 GeV can be estimated from the chi- 1 X  q 
CMM = C5 + C6q qM GM
ral Lagrangian relations, m2 = (mu + md ) B0 assuming 2
q=u,d,s
that mu = md = 5 MeV.
1 X  q 
For the vector operators Eq. (3) one can use the defi- DMM = C7 + C8q qM GM . (44)
nition of the pion (kaon) form factor and crossing sym- 2
q=u,d,s
metry,
This result could be used to constrain Wilson coefficients
(q)
hM + M |q q|0i = qM GM (Q2 ) (p+ p ) , (38) of vector operators.
 6

TABLE IV: Upper bounds on the parameters of the Lagrangian in Eq. (5) from tau decay experiments.

Bound on |c 2
i |/ , GeV
3

Coef B( + ) B( e + ) B( K + K ) B( eK + K ) B( ) B( e ) B( ) B( e)
8 8 8 8 7
< 2.1 10 < 2.3 10 < 4.4 10 < 3.3 10 < 1.3 10 < 1.6 107 < 1.3 107 < 1.6 107
c1 6.8 108 6.5 108 9.4 108 8.2 108
c2 2.3 107 2.5 107 1.6 107 1.5 107
c3 6.8 108 6.5 108 9.4 108 8.2 108
c4 2.3 107 2.5 107 1.6 107 1.5 107

B. Parity-violating gluonic operators where AM and BM are defined as

Constraints on the parity-violating contributions can 2i X   b hq

q M
AM = c aM + C2q C1q
be obtained from the lepton-flavor-violating meson and 9 2 4mq
q=u,d,s
tau decays, M and M e, where = , e, and X  q 
M = , , . The analysis of decays involving an 1
+ m C5 q
C6q bq fM (48)
is especially interesting, as meson contains consider- 2
q=u,d,s
able amount of glue, which makes it possible to con- 1 X  q 
strain gluonic LFV operators resulting from integrating m q
C7 C8q bq fM ,
2
out heavy quarks. q=u,d,s
To calculate the decay rates one needs to parameterize 2i X   b hq
q M
the hadronic matrix elements, BM = c aM + C4q C3q
9 4 4mq
q=u,d,s
q 1 X  q 
hM (p)|q 5 q|0i = ibq fM p , m C5 q
C6q bq fM (49)
hM (p)|q5 q|0i = q
ibq hM , (45) 2
q=u,d,s
s 1 X  q 
hM (p)| Ga G ea |0i = aM , + m q
C7 C8q bq fM .
4 2
q=u,d,s

where q = u, d, s, and bu,d = 1/ 2, bs = 1. The form
factors defined above in the Feldmann-Kroll-Stech (FKS) The current experimental bounds on flavor-violating tau
mixing scheme [34] are constrained for and mesons decays allow to put stringent constraints on Wilson co-
to be [35] efficients c
i [37]. We display them in Table IV. These
results, along with the ones displayed in Table III, can be
m2 m2 translated into bounds on flavor-changing interactions of
a = sin 2 (fq bq sin + fs cos ) , leptons with heavy quarks in particular models. As an
2 example on how this can be done, we consider a generic
m2 m2 leptoquark model.
a = sin 2 (fq bq sin + fs cos ) , (46)
2

where = 39.3o 1.0o is a mixing angle of and

in the flavor basis [35]. Numerically, anomaly matrix IV. MODEL EXAMPLE: LEPTOQUARKS
elements are a = 0.022 0.002 GeV3 , a = 0.057
0.002 GeV3 . The decays constants in Eq. (45) used in
numerical work are fq = 108 3 MeV, fq = 89 3 MeV, The Wilson coefficients of effective gluonic operators
that were constrained in the previous sections can be
fs = 111 6 MeV, and fs = 136 6 MeV [35, 36].
used to put bounds on parameters of operators describ-
Neglecting terms of the order O(m /m ), which would ing lepton interactions with heavy quarks in particular
change our answer to at most 5% for = , we can write models of NP. Let us provide an example of how this can
for the decay rate, be done using generic leptoquark model.
The general renormalizable, B and L conserving, and
m h i m2M
2
2 2 SU (3) SU (2) U (1) invariant LQ-lepton-quark inter-
( M ) = |AM | + |B M | 1 ,
84 m2 actions are given in Refs. [3840]. The relevant for our
(47) considertion scalar LQs (S) and vector LQs (V ) interac-
 7

tions are Finally, for the common scales MS and MV of scalar and
vector LQ masses, respectively, we get
c
LS = (LS0 qL i2 2L + RS0 ucR 1R ) S0  2

t t MS
2
+ LS1/2 uR 2L + RS1/2 qL i2 1R S1/2 + H.c.,(50) |t
RS0 LS0 | = | t

RS1/2 LS1/2 | < 1.2 10 (56)
,

1 TeV
LV = LV0 qL 2L + RV0 dR 1R V0  2

b b b b 4 MV
+ LV dc 2L + RV q c 1R V + H.c., |LV0 RV0 | = |LV1/2 RV1/2 | < 1.6 10 (57)
,
1/2 R 1/2 L 1/2 1 TeV
(51) where 6= = e, .
In leptoquark models the couplings of heavy quarks
where q, u and d are doublet, singlet up and singlet
can also be constrained from the photon dipole-like op-
down quarks, respectively; we omit flavor indexes, the
erators that also contribute to e. Those have been
subindexes 0 and 1/2 indicate SU (2) singlet and doublet
recently constrained in Ref. [41]. Assuming the domi-
LQ, respectively; and couplings are assumed to be real.
nance of the dipole operator over all other contributions,
Consider -e conversion on 19779 Au induced by lepto- one obtains comparable bounds on heavy quark couplings
quark exchange. For the values of the loop integral in
to leptoquarks which are of order 103 (MLQ /1 TeV)2 for
Eq. (A1) we simply have I1 (mt ) = I1 (mb ) = I1 (mc ) =
0.333 since the muon mass and the binding energy of the the products of couplings with same chiralities |i ei
LQ LQ |.
muonic gold are much lower than c, b and t quark masses. Here we only considered quarks of the last two genera-
The expressions for relevant Wilson coefficients in Eq. (1) tions i = 2, 3 (either c, t or s, b), LQ = LS0 , RS0 , LS1/2 ,
are given in Table V, where the quark flavor indexes are RS1/2 , LV0 , RV0 , LV1/2 and RV1/2 , with MLQ being the
u = u, c, t and d = d, s, b. mass of the correspondent scalar or vector LQ [41].
Similar constraints are also available from tau decays.
For = 1 and = 2
TABLE V: The Wilson coefficients for the model with LQs
|t t
RS0 LS0 |
|t t
RS1/2 LS1/2 |
= < 2.3 104 GeV2 ,(58)
Ciu /2 Expression Cid /2 Expression MS20 MS21/2
1 u u b b
u RS 2 d
2
LV 1
C1
2
1/2 LS1/2
2
C1
2
1/2 RV1/2
2 |LV
b
b | |LV
b
b |
1/2 RV1/2
2MS MV 0 RV0
1/2 1/2 = < 4.4 106 GeV2 (, 59)
u
C2
1 u 2 u
RS
0
LS
0
d
C2
2 b
LV RV
0
1 b
0
MV20 MV21/2
2 2MS 2 2 MV 2
0 0
u
C3
2 u 1 u
RS LS d
C3
1 b
LV RV
2 b and,
0 0 0 0
2 2MS 2 2 MV 2
2 u
0
u 1 b
0
b |RS
t
t | |RS
t
t |
1/2 LS1/2
1 2 0 LS0
u
C4
RS
1/2 LS1/2
d
C4
LV
1/2 RV1/2 = < 2.3 104 GeV2 ,(60)
2 2
2MS 2 2
MV MS20 MS21/2
1/2 1/2

|b b
LV0 RV0 |
|b b
LV1/2 RV1/2 |
= < 4.4 106 GeV2 (. 61)
We assume that only the couplings for a single quark MV20 MV21/2
flavor are nonzero at a time. From Eqs. (9) and (11) it
While for e = 1 and = 2 ,
follows that the best bounds for the scalar LQs (left half
of the Table V) are relaxed by the factor 2mt /mb 75 |et t
RS0 LS0 |
|et t
RS1/2 LS1/2 |
with respect to the ones for the vector LQs (right half of = < 2.2 104 GeV2 ,(62)
MS20 MS21/2
the Table V). Using the bound on |c1 |, we have for e = 1
and = 2 |LV
b
eb | |LV
b
eb
1/2 RV1/2
|
0 RV0
2 = 2 < 4.2 106 GeV2 (, 63)
t MV0 MV1/2
|et t
RS0 LS0 |
|et
RS1/2 LS1/2 |
= < 1.2 108 GeV2 ,(52)
MS20 MS21/2 and,

|b |b eb |RS
t
et | |RS
t
et |
LV1/2 RV1/2 |
eb
LV0 RV0 | 0 LS0
=
1/2 LS1/2
< 2.2 104 GeV2 ,(64)
= < 1.6 1010 GeV2(,53) 2
M S0 2
MS1/2
MV20 MV21/2
|eb b
LV0 RV0 |
|eb b
LV1/2 RV1/2 |
and, using the bound on |c3 |, we have = < 4.2 106 GeV2 (. 65)
MV20 MV21/2
|t et
RS0 LS0 |
|t et
RS1/2 LS1/2 |
= < 1.2 108 GeV2 ,(54) Clearly, constraints on the coefficients of operators con-
MS20 MS21/2 taining tau-lepton fields are much weaker than the ones
b containing muon fields. We expect those constraints to
|eb b
LV0 RV0 |
|eb
LV1/2 RV1/2 |
= < 1.6 1010 GeV2(.55) improve with new data coming from Belle II collabora-
MV20 MV21/2 tion.
 8

V. CONCLUSIONS where q = m2q /E 2 with the process energy scale E. For

q > 1/4 they take the form
We considered contributions of heavy quark-induced
operators to leptonic FCNC transitions. We constrained
Wilson coefficients of effective gluonic operators in e,
, and e transitions. These bounds can be used to
study interactions of leptonic FCNCs with heavy quarks " !
p
that are kinematically inaccessible in the described ex- 1 + i 1 + 4q
periments. We provided an explicit example of con- I1 = 2q q (4q 1) Li2 (A3)
2q
straints on the parameters of a generic leptoquark model. !#
2i
+ Li2 p , (A4)
1 + 4q i
Acknowledgments " !
p
1 + i 1 + 4q
We thank Will Shepherd for useful conversations, and I2 = q Li2 (A5)
2q
German Valencia for pointing out a computational error !#
in Table III. A.A.P. would like to thank Kavli Institute 2i
+ Li2 p , (A6)
for Theoretical Physics at the University of California 1 + 4q i
Santa Barbara for hospitality where part of this work
was performed. This work was supported in part by the
U.S. Department of Energy under contract DE-FG02-
12ER41825 and by the National Science Foundation un-
der Grant No. NSF PHY11-25915.
for q 1 they are
Appendix A: Integrals

In this appendix we present the results of computation

of the integrals needed to obtain the matching coefficients
of gluonic operators of dimension 7. The matching coef-  
1 7 1 2
ficients can be computed to be I1 = 1+ + O(q ) , (A7)
3 120 q
Z 1 Z 1x  
1 4xy 1 1
I1 (mq ) = q dx dy , (A1) I2 = 1 + 1 + O(2
q ) . (A8)
0 0 q xy 2 12 q
Z 1 Z 1x In this paper we use the leading order result in the q -
1
I2 (mq ) = q dx dy , (A2) expansion.
0 0 q xy

[1] See, e.g., W. J. Marciano and A. I. Sanda, Phys. Lett. B [7] T. P. Cheng and M. Sher, Phys. Rev. D 35, 3484 (1987).
67, 303 (1977). [8] W. -j. Li, Y. -d. Yang and X. -d. Zhang, Phys. Rev. D
[2] J. D. Vergados, Phys. Rept. 133, 1 (1986). 73, 073005 (2006) [hep-ph/0511273].
[3] V. Cirigliano, R. Kitano, Y. Okada and P. Tuzon, Phys. [9] For lepton flavor-conserving operators a similar approach
Rev. D 80, 013002 (2009) [arXiv:0904.0957 [hep-ph]]. is followed in H. Potter and G. Valencia, Phys. Lett. B
[4] A good example of this approach can be seen in discus- 713, 95 (2012) [arXiv:1202.1780 [hep-ph]]. In that paper
sion of charm mixing, see, e.g., E. Golowich, J. Hewett, the scale denotes a combination of a new physics scale,
S. Pakvasa and A. A. Petrov, Phys. Rev. D 76, 095009 electroweak scale, and scales associates with heavy quark
(2007) [arXiv:0705.3650 [hep-ph]]. masses.
[5] For similar approach to generate lepton flavor-violating [10] M. E. Peskin and D. V. Schroeder, An Introduction to
transitions with photons in the final state see, e.g. quantum field theory, Reading, USA: Addison-Wesley
M. Raidal and A. Santamaria, Phys. Lett. B 421, 250 (1995) 842 p.; R. N. Cahn, A Higgs primer, LBL-29789.
(1998); K. J. Healey, A. A. Petrov and D. Zhuridov, Phys. [11] T. G. Rizzo, Phys. Rev. D 22, 178 (1980) [Addendum-
Rev. D 87, 117301 (2013). ibid. D 22, 1824 (1980)].
[6] For example, Q1 operator would result from an operator [12] W. J. Marciano and A. I. Sanda, Phys. Rev. Lett. 38,
of the form (1R HL2L ) (qR HQL ) (where L2L and QL are 1512 (1977).
the correspondingly electroweak doublets of leptons and [13] O. U. Shanker, Phys. Rev. D 20, 1608 (1979).
quarks). This is a dimension 8 operator that is suppressed [14] J. Bernabeu, E. Nardi and D. Tommasini, Nucl. Phys. B
by two powers of the NP scale and two powers of the scale 409, 69 (1993) [hep-ph/9306251].
v associated with electroweak symmetry breaking. [15] T. S. Kosmas, G. K. Leontaris and J. D. Vergados, Prog.
 9

Part. Nucl. Phys. 33, 397 (1994) [hep-ph/9312217]. dos, Phys. Rev. C 56, 526 (1997) [nucl-th/9704021].
[16] R. Barbieri and L. J. Hall, Phys. Lett. B 338, 212 (1994) [30] D. Black, T. Han, H. -J. He and M. Sher, Phys. Rev. D
[hep-ph/9408406]. 66, 053002 (2002).
[17] R. Barbieri, L. J. Hall and A. Strumia, Nucl. Phys. B [31] C. -H. Chen and C. -Q. Geng, Phys. Rev. D 74, 035010
445, 219 (1995) [hep-ph/9501334]. (2006).
[18] For a recent review see, for example, A. Czarnecki, [32] For more elaborate studies of hadronic effects, see
W. J. Marciano and K. Melnikov, AIP Conf. Proc. 435, J. T. Daub, H. K. Dreiner, C. Hanhart, B. Kubis and
409 (1998) [hep-ph/9801218]. U. G. Meissner, JHEP 1301, 179 (2013).
[19] W. H. Bertl et al. [SINDRUM II Collaboration], Eur. [33] M. B. Voloshin, Sov. J. Nucl. Phys. 44, 478 (1986) [Yad.
Phys. J. C 47, 337 (2006). Fiz. 44, 738 (1986)].
[20] R. Kitano, M. Koike and Y. Okada, Phys. Rev. D [34] T. Feldmann, P. Kroll and B. Stech, Phys. Rev. D 58,
66, 096002 (2002) [Erratum-ibid. D 76, 059902 (2007)] 114006 (1998).
[hep-ph/0203110]. [35] M. Beneke and M. Neubert, Nucl. Phys. B 651, 225
[21] T. S. Kosmas, S. Kovalenko and I. Schmidt, Phys. Lett. (2003) [hep-ph/0210085].
B 511, 203 (2001) [hep-ph/0102101]. [36] A. A. Petrov, Phys. Rev. D 58, 054004 (1998)
[22] M. A. Shifman, A. I. Vainshtein and V. I. Zakharov, [37] J. Beringer et al. [Particle Data Group Collaboration],
Phys. Lett. B 78, 443 (1978). Phys. Rev. D 86, 010001 (2012), see also Y. Amhis
[23] H.-Y. Cheng and C.-W. Chiang, JHEP 1207, 009 (2012) et al. [Heavy Flavor Averaging Group Collaboration],
[arXiv:1202.1292 [hep-ph]]. arXiv:1207.1158 [hep-ex]; Y. Miyazaki et al. [Belle Col-
[24] H. De Vries, C. W. De Jager and C. De Vries, Atom. laboration], Phys. Lett. B 719, 346 (2013).
Data Nucl. Data Tabl. 36, 495 (1987). [38] W. Buchmuller, R. Ruckl and D. Wyler, Phys. Lett. B
[25] G. Fricke et al., Atom. Data Nucl. Data Tabl. 60, 177 191, 442 (1987) [Erratum-ibid. B 448, 320 (1999)].
(1995). [39] S. Davidson, D. C. Bailey and B. A. Campbell, Z. Phys.
[26] T. Suzuki, D. F. Measday and J. P. Roalsvig, Phys. Rev. C 61, 613 (1994) [hep-ph/9309310].
C 35, 2212 (1987). [40] M. Gonderinger and M. J. Ramsey-Musolf, JHEP
[27] R. Harnik, J. Kopp and J. Zupan, JHEP 1303, 026 1011, 045 (2010) [Erratum-ibid. 1205, 047 (2012)]
(2013) [arXiv:1209.1397 [hep-ph]]. [arXiv:1006.5063 [hep-ph]].
[28] C. Dohmen et al. [SINDRUM II. Collaboration], Phys. [41] E. Gabrielli, Phys. Rev. D 62, 055009 (2000)
Lett. B 317, 631 (1993). [hep-ph/9911539].
[29] T. S. Kosmas, A. Faessler, F. Simkovic and J. D. Verga-