General CP Violation in Minimal Left-Right Symmetric Model

and Constraints on the Right-Handed Scale

Yue Zhang,1, 2 Haipeng An,2 Xiangdong Ji,2, 1 and Rabindra N. Mohapatra2
1
Center for High-Energy Physics and Institute of Theoretical Physics,
Peking University, Beijing 100871, China
2
Maryland Center for Fundamental Physics,
Department of Physics, University of Maryland,
College Park, Maryland 20742, USA
arXiv:0712.4218v1 [hep-ph] 27 Dec 2007

(Dated: October 9, 2018)

Abstract
In minimal left-right symmetric theories, the requirement of parity invariance allows only one
complex phase in the Higgs potential and one in the Yukawa couplings, leading to a two-phase
theory with both spontaneous and explicit CP violations. We present a systematic way to solve
the right-handed quark mixing matrix analytically in this model and find that the leading order
solution has the same hierarchical structure as the left-handed CKM matrix with one more CP-
violating phase coming from the complex Higgs vev. Armed with this explicit right-handed mixing
matrix, we explore its implications for flavor changing and conserving processes in detail, low-
energy CP-violating observables in particular. We report an improved lower bound on the WR
mass of 2.5 TeV from ∆MK and ∆MB , and a somewhat higher bound (4 TeV) from kaon decay
parameters ǫ, ǫ′ , and neutron electric dipole moment. The new bound on the flavor-changing
neutral Higgs mass is 25 TeV.

1
I. INTRODUCTION

The physics beyond the Standard Model (SM) has been the central focus of high-energy
phenomenology for more than three decades. Many proposals, including supersymmetry,
technicolor, little Higgs, and extra dimensions, have been made and studied thoroughly in
the literature, tests are soon to be made at the Large Hadron Collider (LHC). One of the
earliest proposals, the left-right symmetric model (LRSM), was motivated by the hypothesis
that parity is a perfect symmetry at high-energy, and is broken spontaneously at low-energy
due to the asymmetric vacuum [1]. Asymptotic restoration of parity has a definite aesthetic
appeal [2]. The model has a number of additional attractive features, including a natural
explanation of weak hyper-charge in terms of baryon and lepton numbers, existence of
right-handed neutrinos and entailed seesaw mechanism for neutrino masses, possibility of
spontaneous CP (charge-conjugation-parity) violation, and natural solution for the strong
CP problem. The model can be constrained strongly by low-energy physics and predicts
clear signatures at colliders. It so far remains a decent possibility for new physics.
The LRSM is best constrained at low-energy by flavor-violating mixing and decays, par-
ticularly CP violating observables. In making theoretical predictions, the major uncertainty
comes from the unknown right-handed quark mixing matrix, similar in spirit to that of the
left-handed quark Cabibbo-Kobayashi-Maskawa (CKM) mixing. The new mixing is a uni-
tary matrix, depending on 9 real parameters: 6 CP violation phases and 3 rotational angles.
All are physical after the left-handed CKM mixing is rotated into a standard 4-parameter
form.
Historically, two special CP violation scenarios in LRSM have been considered. The first
one, “ the manifest left-right symmetry”, assumes that there is no spontaneous CP violation
i.e. all Higgs vacuum expectation values (vev’s) are real. Then the quark mass matrices are
hermitian, and the left- and right-handed quark mixings become identical, modulo the sign
uncertainty of the elements from negative quark masses. The second scenario, “pseudoman-
ifest left-right symmetry”, assumes that the CP violation comes entirely from spontaneous
symmetry breaking (SSB) of the vacuum and all Yukawa couplings are real [3]. Here the
quark mass matrices are complex and symmetric, implying that the right-handed quark
mixing is related to the complex conjugate of the CKM matrix multiplied by additional CP
phases. There are few studies of the model with general CP violations in the literature [4, 5].
It has been noted that there are problems with both manifest and pseudo-manifest scenar-
ios. The manifest LRSM with real potential and vev’s always provides more minimization
conditions than the number of vev’s, thus has a fine-tuning problem. In any case, in the
non-suspersymmetric LRSM, the assumption of all parameters of the model being real is
not technically natural since loop corrections lead to one of the parameters being complex.
On the other hand, the pseudo-manifest LRSM, where exact CP is assumed before sym-
metry breaking and all couplings in the Higgs potential are real, leads in the decoupling
limit to either a model with light triplet Higgs, which is already excluded by SM precision
test, or a two Higgs doublet model, which is excluded by experiment due to large tree level
flavor-changing neutral current [6]. Therefore, neither of the two special scenarios can be
realistic.
In a recent paper [7], we reported a systematic approach to analytically solving the right-
handed quark mixing in the minimal LRSM where only the requirement of parity invariance
is imposed prior to symmetry breaking, leaving automatically only one CP phase in the
Higgs potential and one in the Yukawa couplings and leading to a theory with both explicit

2
and spontaneous CP violations. This model therefore falls in-between the above two extreme
cases and is free of the problems described above. Our approach is based on the observation
that in the absence of any fine tuning, mt ≫ mb implies that the ratio of the two vev’s of
the Higgs bi-doublet, ξ = κ′ /κ, is small and is of the order of mb /mt . In the leading-order in
ξ, we find a linear equation for the right-handed quark mixing matrix which can be readily
solved. We present an analytical solution of this equation valid to O(λ3 ), where λ = sin θC
is the Cabibbo mixing parameter. The leading right-handed quark mixing is nearly the
same as the left-handed CKM matrix, except for additional phases which are fixed by ξ,
spontaneous CP phase α, and the quark masses.
Our work is similar in spirit to the detailed numerical study of the general CP LRSM
made by Kiers et al. [5], with the two vev’s ratio κ′ /κ fixed exactly to mb /mt . It is
interesting that the gross feature of the right-handed CKM was obtained already in the
appendix of that paper, in particular the hierarchical structure of flavor mixing and the
magnitude of the Dirac CP phase. However, realistic studies were made numerically for
lack of an explicit solution with known precision. In fact, much of the numerical work of
Ref. [5] goes into solving the right-handed CKM in the presence of 11 input parameters,
which must be scanned through using Monte Carlo to obtain the physical quark masses and
the left-handed CKM mixing. Because the extensive nature of numerical study, it is difficult
to see some of the physics in a clear way, in particular, the interplay between the explicit
and spontaneous CP violations in physical observables. Our explicit analytic solution for
the right-handed CKM makes extensive analytical studies simple and straightforward.
In this paper, we first give the detailed method obtaining the analytical solution to
righthanded quark mixing [7]. With this explicit right-handed mixing, we study the neutral
kaon and B-meson systems and the neutron electric dipole moment (EDM) to obtain the
lower bound on righthanded W -boson mass scale. The neutral kaon mass mixing provides
a rigorous lower bound MWR > 2.5 TeV, with the use of the new lattice QCD calculations
of the four-quark matrix elements and the strange quark mass. The indirect CP violation ǫ
receives large contributions from both explicit and spontaneous CP phases, and from both
the gauge-boson box diagram and flavor-changing neutral Higgs (FCNH). There are strong
cancelations among all the contributions, which in turn constrain severely the relation among
the spontaneous CP phase, MWR , and FCNH mass MH . We use the cancelation condition
to fix the spontaneous CP phase, which is then used to predict the neutron EDM in terms
of the model parameters. Using the experimental bound on the EDM, we obtain a strong
lower bound on MWR , which can be improved with better calculations of the hadronic matrix
elements and more precise experimental data. Furthermore, we study implications of direct
CP violation in K 0 and B 0 decays. In the former case, a strong lower bound on MWR is
obtained under the factorization assumption for the four-quark matrix elements. The CP
violating observables in the kaon and B systems and the neutron EDM provide competitive
or even stronger bounds than the well-known kaon mass mixing. We also present a detailed
study of the Higgs sector in the presence of the spontaneous CP phase, including the mass
spectrum, the neutral and charged couplings to the quarks, and the bound on the FCNH
mass.
The paper is organized as follows. We first briefly introduce the minimal LRSM in Sec.
II. In Sec. III, we report our method in solving for the righthanded CKM matrix in the
scenario of generic CP violation, including both spontaneous and explicit phases as is the
case with only parity invariance. In Sec. IV, we study the Higgs sector, including the mass
spectrum and Higgs couplings to quarks. In Sec. V, we explore the well-known neutral

3
kaon mass difference to find out updated constraints on the right-handed W mass and the
FCNH mass. We also consider similar constraints from the neutral B-meson system. The
CP violations in various processes are discussed in Sec. VI, including ǫ, ǫ′ , neutron EDM
and CP asymmetry in B → J/ψKS , to constrain the mass of WR and the spontaneous
CP phase α. We find that they place consistently strong lower bounds on MWR and MH .
We conclude the paper in Sec. VII, with a table summarizing various constraint on the
right-handed scales.

II. THE MINIMAL LEFT-RIGHT SYMMETRIC MODEL

The minimal LRSM is based on the gauge group SU(2)L × SU(2)R × U(1)B−L . Parity is
assumed to be a good symmetry at the Lagrangian density level, and is broken spontaneously
by vev’s of Higgs fields. The electric charge formula can be written as a generalized Gell-
mann-Nishijima formula including the third component of the right-handed isospin, T3L and
T3R , and the difference between baryon and lepton numbers [8],

Q = T3L + T3R + (B − L)/2 . (1)

This in turn gives an explicit explanation for the standard model U(1)-hypercharge in terms
of physical quantum numbers rather than an arbitrarily adjustable quantum number Y .
In the matter sector, the left-handed fermions form fundamental representations of the
SU(2)L gauge group, while right-handed ones form the representations of the SU(2)R gauge
group. The right-handed neutrinos are introduced automatically so that the right-handed
leptons also form doublets under SU(2)R
       
uL 1 uR 1
QL = ∈ 2, 1, , QR = ∈ 1, 2, ,
dL 3 dR 3
   
νL νR
LL = ∈ (2, 1, −1) , LR = ∈ (1, 2, −1) ,
lL lR

where the quantum numbers are those of the above gauge groups. From these, we can easily
write down the gauge-coupled Lagrangian density for fermions,
 
fermion µ gL −
→ g′
L = QLi γ i∂µ − W Lµ · ~τ − Bµ QLi
2 6
 
µ gR −
→ g′
+ QRi γ i∂µ − W Rµ · ~τ − Bµ QRi
2 6
 
µ gL −
→ g′
+ LLi γ i∂µ − W Lµ · ~τ + Bµ LLi
2 2
 
µ gR −
→ g′
+ LRi γ i∂µ − W Rµ · ~τ + Bµ LRi ,
2 2
where the index i = 1, 2, 3 labels fermion generation with all the fields being in flavor
eigenstates. W~ R,Lµ and Bµ are the gauge fields associated with the above gauge groups,
with corresponding couplings gL = gR and g ′ . ~τ are Pauli matrices for isospins. The right-
handed currents couple to the gauge bosons WR in a way symmetric to the left-handed
counterparts.

4
 The Higgs sector contains a bidoublet φ, belonging to the (2, 2, 0) representation of the
gauge group, which is the left-right symmetric version of the SM Higgs doublet and two
triplets ∆L,R belonging to (3, 1, 2) and (1, 3, 2), respectively,
 0 +  + √ 
φ1 φ2 δL / 2 δL++√
φ = , ∆L = ,
φ−1 φ2
0
δL0 −δL+ / 2
 + √ 
δR / 2 δR++√
∆R = . (2)
δR0 −δR+ / 2

The Higgs boson’s kinetic energy and coupling to the gauge fields are canonical,

L Higgs kin
= Tr[(Dµ ∆L )† (D µ ∆L )] + Tr[(Dµ ∆R )† (D µ ∆R )]
+ Tr[(Dµ φ)† (D µ φ)] , (3)

where the covariant derivatives are

gL −→ gR − →
Dµ φ = ∂µ φ + i W Lµ · ~τ φ − i φW Rµ · ~τ ,
2 2
g(L,R) h−
→ i
Dµ ∆(L,R) = ∂µ ∆(L,R) + i W (L,R)µ · ~τ , ∆(L,R) + ig ′ Bµ ∆(L,R) .
2
Again, left-right symmetry is explicit.
Like the SM, we use the vev’s of neutral Higgs fields to break the gauge symmetry. The
symmetry group SU(2)L × SU(2)R × U(1)B−L is first broken to SU(2)L × U(1)Y by the vev
h∆R i = vR at TeV or multi-TeV scale. Then electroweak symmetry breaking is induced by
the vev’s of φ. With a generic Higgs potential, all the neutral components of the Higgs get
vev’s,
 iα 
κe 1 0
hφi = ,
0 κ′ eiα2
   
0 0 0 0
h∆L i = , h∆R i = , (4)
vL eiθ1 0 vR eiθ2 0

where there are 4 neutral complex components and so generically 4 phases. At first it appears
that 3 of them can be eliminated through gauge symmetry because we have 3 generators,
T3L , T3R , and B − L, commuting with the electromagnetic charge operator Q. In reality,
however, we can eliminate only 2. If the transformation parameters associated with the
above 3 operators are θL , θR and θB−L , the Higgs fields transform as,

hφi → eiT3L θL hφie−iT3R θR ,

h∆L i → eiT3L θL h∆L ie−iT3L θL eiθB−L ,
h∆R i → eiT3R θR h∆R ie−iT3R θR eiθB−L . (5)

This leads a transformation of the vev phases,

1 1 1 1
α1 → α1 + θL − θR , α2 → α2 − θL + θR ,
2 2 2 2
θ1,2 → θ1,2 − θL,R + θB−L . (6)

5
It is clear that there are two independent combinations of transformation parameters, al-
lowing removing two phases only. Conventionally the phases of κ and vR are set to zero,
and thus the general form of Higgs vev’s is simplified to
 
κ 0
hφi = ,
0 κ′ eiα
   
0 0 0 0
h∆L i = , h∆R i = . (7)
vL eiθL 0 vR 0

If one chooses vL = 0, the only remaining phase α is physically relevant.

Since the bidoublet φ transforms non-trivially under both SU(2)L and SU(2)R , the gauge
bosons WL and WR are not the mass eigenstates after SSB. From (3) we get the W -boson
mass matrix,
   +µ 
W−Mass − −

1 g 2 (κ2 + κ′2 + 2vL2 ) −g 2 κκ′ e−iα WL
−L = WLµ WRµ 2 , (8)
2
−g κκ e′ iα 2 2
g vR WR+µ
± 1 2
√
where WL,R = (WL,R ∓ iWL,R )/ 2, and g = gL = gR by parity. Since vL is at most on the
√ √
order of the left-handed neutrino masses and κ2 + κ′2 on the SM scale, vL ≪ κ2 + κ′2 ,
and we can neglect vL2 in Eq. (8). The mass matrix can be diagonalized by a unitary
transformation  +   + 
WL cos ζ − sin ζeiλ W1
= , (9)
WR+ sin ζe−iλ cos ζ W2+
where W1+ and W2+ are mass eigenstates (1 and 2 subscripts shall not be√confused with
√ the
Cartesian components of the isospins), with masses MW1 ≃ MWL = g κ2 + κ′2 / 2 and
MW2 ≃ MWR = gvR . The parameters ζ and λ are related to the CP phase α and the masses
of W1 , W2 , κ and κ′
 2
κκ′ MWL
λ = −α , tan ζ = − 2 ≃ −2ξ , (10)
vR MWR

where again ξ = κ′ /κ. If there is no cancelation in generating quark masses, ξ has a natural
size mb /mt , and thus ζ is suppressed by both (MWL /MWR )2 and mb /mt and is smaller than
4 × 10−5 , which is much smaller than the current experimental bound 10−2 . Even so this
tiny mixing will be the dominating contribution to the neutron EDM, as we will discuss
below. In terms of the mass eigenstates, the charged current couplings in the quark sector
are
gL

L W−current = − √ ūLi γ µ cos ζW1µ+
− sin ζeiλ W2µ
+
dLi (11)
2
gR

− √ ūRi γ µ sin ζe−iλW1µ
+
+ cos ζW2µ+
dRi + h.c. .
2
The above expression is in quark flavor basis. In the next section, we will consider the quark
mass basis in which it will be modified by the CKM mixing matrices.

6
III. RIGHT-HANDED QUARK MIXING MATRIX

In this section, we will focus on the quark sector, solving for the right-handed quark
mixing matrix compatible with the observed quark masses and the left-handed mixing. The
solution is valid in the general CP violation scenario in which both explicit and spontaneous
CP breakings are allowed. The only significant assumption we will make is that bidoublet
higgs vev’s and related Yukawa couplings have a hierarchical structure and there is no
large cancelation in generating quark masses, and the solution can be made in a systematic
expansion of the relevant small parameter.
The most general Yukawa coupling of the quark fields with the Higgs bidoublet φ is given
by
LY = QLi (hij φ + h̃ij φ̃)QRj + h.c. , (12)
where φ̃ = −iτ2 φ∗ iτ2 and flavor indices i, j = 1, 2, 3. Parity symmetry, under which φ ↔ φ†
and QLi ↔ QRi , constrains h and h̃ be hermitian matrices. After SSB, the above Lagrangian
density yields the following up- and down-type quark mass matrices,

MU = κh + κ′ e−iα h̃ ,
MD = κ′ eiα h + κh̃ . (13)

There are two terms in each, and we assume that there is no fine-tuned cancelation to
generate a quark mass scale. Since the top quark mass is much larger than that of the
bottom quark, the assumption implies that h and h̃, κ and κ′ should not be on the same
order. Without loss of generality, we take κ′ ≪ κ and h̃ ≪ h. To leading order in κ′ /κ, we
have

MU ≃ κh ,
MD = κ′ eiα h + κh̃ . (14)

The two terms in the down-type quark masses can be on the same order, however. Because
of the flavor independence of the gauge coupling and the hermiticity of h, we can work in
the basis where MU is diagonal,
 
mu
MU = SU  mc  ≡ SU McU , (15)
mt

in which SU = diag{su , sc , st } is the sign of the up-type quark masses. It is present be-
cause the eigenvalues of a hermitian matrix can either be positive or negative, and by
convention we take all mi positive. In this basis, MD is not diagonal and is related to
cD = diag{md , ms , mb } via left- and right-handed CKM rotations. Since the phase factor
M
eiα in MD is generically non-zero, VLCKM 6= VRCKM .

cD V CKM† SU .
MD = VLCKM M (16)
R

For simplicity, we will omit the superscript CKM henceforth. From Eqs. (14), (15) and (16),
we have ′
cD V † SU − κ SU M
κh̃ = VL M cU eiα . (17)
R
κ

7
Two comments are in order. First, through the phase transformations that are chirally
independent but isospin-dependent, uiL,R → eiαi uiL,R and diL,R → eiβi diL,R , one can bring VL
to a standard form with only 4 parameters (3 rotations and 1 CP violation phase) and the
above equation remains the same. The h̃ matrix, however, will be subjected to a unitary
transformation and remains hermitian. Second, after the transformation, all parameters in
the unitary matrix VR must be physical, including 3 rotations and 6 CP-violating phases.
To make further progress, one uses the hermiticity condition for h̃, which leads to the
following equation,
McD Vb † − VbR M cU SU VL ,
cD = 2iξ sin α V † M (18)
R L

where VbR is the quotient between the left and right mixing VR = SU VL VbR . There are a total
of 9 equations in the above expression, which is just enough to solve 9 parameters in VbR .
It is interesting to note that if there is no spontaneous CP violation, α = 0, we recover the
solution VR = SU VL SD , in which SD = diag{sd , ss , sb } is the sign matrix for down-type
quark masses. This is just the manifest LRS case.
The above linear equations can be readily solved utilizing the hierarchy between down-
type quark masses. We find an analytical expression for VR up to order of O(λ3), where λ is
sine of Cabibbo angle. We begin directly from (18) and the left side is anti-hermitian, and
we can write it explicitly
 
−2imd ImVbR11 −ms VbR12 −mb VbR13
 ms VbR12
∗
−2ims ImVbR22 −mb VbR23  , (19)
mb VbR13
∗
mb VbR23
∗
−2imb ImVbR33

where we have used md ≪ ms ≪ mb . The right-hand side of Eq. (18) depends on the phys-
ical quark masses, the standard CKM matrix, and the spontaneous CP violation parameter
ξ sin α. Thus we can solve VbR in terms of these up to O(λ3 )
 
mb mc 2 mt 2 4

b
ImVR11 = −r sin α λ sc + st A λ (1 − ρ) + η2 2
, (20)
md mt mc
 
b mb mc mt 2 4
ImVR22 = −r sin α sc + st A λ , (21)
ms mt mc
ImVbR33 = −r sin αst , (22)
 
mb mc mt
VbR12 = 2ir sin α λ sc + st λ4 A2 (1 − ρ + iη) , (23)
ms mt mc
VbR13 = −2ir sin αAλ3 (1 − ρ + iη)st , (24)
VbR23 = 2ir sin αAλ2 st , (25)

where r ≡ (mt /mb )ξ, and λ, A, ρ, and η are Wolfenstein parameters for VL . The above
solution exists only when |r sin α| ≤ 1, which is an interesting and unexpected constraint.
Since the natural size of ξ is mb /mt , r ∼ 1, allowing angle α ∼ 1. Given the physical values
of various parameters, we find the following power counting: ImVbR11 ∼ λ, ImVbR22 ∼ λ,
VbR12 ∼ λ2 , VbR13 ∼ λ3 and VbR23 ∼ λ2 . Using unitarity condition for VbR , we can solve all other
elements.

8
 Defining new phases sin θi = SDii ImVRii , where i = 1, 2, 3, we have up to O(λ3 ),

VbRii = SDii eiθi ,

VbR21 = −sd ss VbR12
∗
ei(θ1 +θ2 ) ,
VbR31 = −sd sb VbR13
∗
ei(θ1 +θ3 ) ,
VbR32 = −ss sb Vb ∗ ei(θ2 +θ3 ) .
R23 (26)

Therefore, we can write the righthanded CKM matrix in a more compact form

VR = PU VeL PD , (27)

in which PU = diag(su , sc e2iθ2 , st e2iθ3 ), PD = diag(sd eiθ1 , ss e−iθ2 , sb e−iθ3 ), and

 
1 − λ2 /2 λ Aλ3 (ρ − iη)
VeL =  −λ 1 − λ2 /2 Aλ2 e−2iθ2  , (28)
Aλ (1 − ρ − iη) −Aλ2 e2iθ2
3
1

which differs from VL by a small phase in 23 and 32 elements.

The phases θi are functions of parameter r sin α and the signs of the quark masses si and
s̃i . Numerically we have,

θ1 = − sin−1 [0.31(sd sc + 0.18sd st )r sin α] ,

θ2 = − sin−1 [0.32(ss sc + 0.25ss st )r sin α] ,
θ3 = − sin−1 [sb st r sin α] , (29)

where we have taken from the Particle Data Group the central values of the quark masses
mu = 2.7 MeV, md = 5 MeV, ms = 98 MeV, mc = 1.25 GeV, mb = 4.2 GeV, and mt = 174
GeV at scale 2 GeV [9]. It shall be noted that since only the quark mass ratios enter
the mixing matrix and the quark masses run multiplicatively, the result is independent of
quark mass scale. The parameters for the left-hand quark mixing are taken as λ = 0.2272,
A = 0.818, ρ = 0.221, and η = 0.34.
A few remarks about the above result are in order. First, the hierarchical structure of
the right-handed mixing is similar to that of the left-handed CKM, namely 1-2 mixing is of
order λ, 1-3 order λ3 and 2-3 order λ2 . Second, every element now has a substantial CP
phase. When r is of order 1, the elements involving the first two families have CP phases
of order λ, and the phases involving the third family are of order 1. These phases are all
related to the single spontaneous CP-violating phase α, and generate rich phenomenology
for K and B meson systems as well as the neutron EDM. Finally, from (27) and (29), it
is clear that the final solution is a function of sign bi-products si sj . We can always fix one
of them, say su , to be positive, then we are left with 25 = 32 distinct sectors. The actual
physical choice must be determined by phenomenology, as we will illustrate in the following
sections.

IV. HIGGS POTENTIAL, MASS SPECTRUM AND COUPLINGS

In this section, we discuss several issues related to the Higgs sector. In particular, we
consider the possibility of spontaneous CP violation from the Higgs potential, the mass

9
spectrum of the Higgs bosons, and the Higgs couplings to the quark sector. The results are
useful for phenomenological studies in the following sections. Some of the results presented
here have appeared in the literature before, and we include them for completeness.
The most general renormalizable Higgs potential invariant under parity is given by [10]
h i h i
V (φ, ∆L , ∆R ) = −µ21 Tr(φ† φ) − µ22 Tr(φ̃φ† ) + Tr(φ̃† φ) − µ23 Tr(∆L ∆†L ) + Tr(∆R ∆†R )
h i2 h i2 
 †
2 † †
+λ1 Tr(φ φ) + λ2 Tr(φ̃φ ) + Tr(φ̃ φ)
h i
+λ3 Tr(φ̃φ† )Tr(φ̃† φ) + λ4 Tr(φ† φ) Tr(φ̃φ† ) + Tr(φ̃† φ)
h i2 h i2 
† †
+ρ1 Tr(∆L ∆L ) + Tr(∆R ∆R )
h i
+ρ2 Tr(∆L ∆L )Tr(∆†L ∆†L ) + Tr(∆R ∆R )Tr(∆†R ∆†R )
h i
† † † † † †
+ρ3 Tr(∆L ∆L )Tr(∆R ∆R ) + ρ4 Tr(∆L ∆L )Tr(∆R ∆R ) + Tr(∆L ∆L )Tr(∆R ∆R )
h i
+α1 Tr(φ† φ) Tr(∆L ∆†L ) + Tr(∆R ∆†R )
n h i o
iδ2 † † † †
+ α2 e Tr(φ̃φ )Tr(∆L ∆L ) + Tr(φ̃ φ)Tr(∆R ∆R ) + h.c.
h i h i
+α3 Tr(φφ† ∆L ∆†L ) + Tr(φ† φ∆R ∆†R ) + β1 Tr(φ∆R φ† ∆†L ) + Tr(φ† ∆L φ∆†R )
h i h i
+β2 Tr(φ̃∆R φ† ∆†L ) + Tr(φ̃† ∆L φ∆†R ) + β3 Tr(φ∆R φ̃† ∆†L ) + Tr(φ† ∆L φ̃∆†R ) , (30)

where there are a total of 18 parameters, µ21,2,3 , λ1,2,3,4 , ρ1,2,3,4 , α1,2,3 , and β1,2,3 . Due to the
left-right symmetry (LRS), only one of them, α2 can become complex and all other couplings
are real. We have included an explicit phase eiδ2 in α2 , introducing an explicit CP violation
in the Higgs potential.
After SSB, the Higgs fields acquire vev’s, and the potential is minimized with respect to
them. The six minimization conditions are
∂V ∂V ∂V ∂V ∂V ∂V
= = = = = =0, (31)
∂κ ∂κ′ ∂α ∂vL ∂vR ∂θL
which lead to six relations among the vev’s and coefficients in the Higgs potential [10, 11]
   
µ21 α1 vL2 α3 ξ 2 vL2  2
 2
= 1 + − 1 + + λ 1 (1 + ξ ) + 2λ 4 ξ cos α ǫ
vR2 2 vR2 2(1 − ξ 2 ) vR2
  vL /vR
+ β2 cos θL − β3 ξ 2 cos(θL − 2α) , (32)
1 − ξ2
   
µ22 α2 vL2 α3 ξ vL2
= cos(α + δ2 ) + cos(α − δ2 ) 2 + 1+ 2
vR2 2 cos α vR 4(1 − ξ 2 ) cos α vR
  2
1 ǫ
+ 2λ2 ξ cos 2α + λ3 ξ + λ4 (1 + ξ 2 ) cos α
2 cos α
  vL /vR
+ β1 (1 − ξ 2 ) cos(θL − α) − 2β2 ξ cos θL + 2β3 ξ cos(θL − 2α) , (33)
4(1 − ξ 2 ) cos α

10
  
µ23 vL2 1 2 2
 2
= ρ 1 1 + + α 1 (1 + ξ ) + α 3 ξ ǫ
vR2 vR2 2
 
vL2 ξǫ2
+2α2 cos(α + δ2 ) − cos(α − δ2 ) 2 , (34)
vR 1 − vL2 /vR2

 
8α2 ξǫ2 sin α sin δ2 vL
(2ρ1 − ρ3 ) − 2 2
= [β1 ξ cos(θL − α) + β2 cos θL + β3 ξ 2 cos(θL − 2α)]ǫ2
1 − vL /vR vR
(35)

0 = β1 ξ sin(θL − α) + β2 sin θL + β3 ξ 2 sin(θL − 2α) (36)

2α2 (1 − ξ 2 )(1 − vL2 /vR2 ) sin δ2 = {2ξ sin(θL − α)(β2 + β3 )

  vL  
+ sin θL + ξ 2 sin(θL − 2α) β1 + ξ sin α α3 (1 + vL2 /vR2 ) + (4λ3 − 8λ2 )(1 − ξ 2 )ǫ2
vR
(37)

where ǫ = κ/vR represents a hierarchy in symmetry breaking. The above equations can be
solved for the Higgs vev’s in terms of the parameters in the Higgs potential.
Historically, two special cases of the general potential have been studied in the literature,
namely “manifest” and “pseudo-manifest” LRS limits. The manifest LRS assumes real
Higgs potential i.e. δ2 = 0, and in addition, no spontaneous CP violation, α = θL = 0.
The only source of CP asymmetry is from the Yukawa couplings. In this case, the quark
mass matrices are hermitian due to parity invariance, and the left- and right-handed CKM
matrices are identical up to quark mass signs. Most early studies were made based on this
simplification. At the level of Higgs potential, this scenario necessitates fine-tuning: From
the neutrino and quark mass hierarchy, we have vL ≪ κ′ < κ ≪ vR . Taking all the phases
to zero in Eqs. (32)-(37), the following relations are found at leading order in ǫ2 [10]

µ21 α1 α3 ξ 2 µ22 α2 α3 ξ 2 µ23

= − , = + , = ρ1 . (38)
vR2 2 2(1 − ξ 2 ) vR2 2 4(1 − ξ 2 ) vR2

There are three equations for only two vev’s vR and ξ, implying a relation among parameters
in the Higgs potential, which can only be achieved through fine-tuning.
On the other hand, pseudo-manifest LRS requires P and CP invariance of the Lagrangian
(δ2 = 0), with the complex vev phase α alone to explain the source of CP violation in the
quark sector. The Higgs potential is real when δ2 = 0, but the vev could be complex. The
Yukawa couplings are real and symmetric. The right-handed CKM matrix is related to the
complex conjugate of its left-handed counterpart with additional diagonal phase matrices
multiplied on both sides.
 However,
2 when Higgs potential is real, the spontaneous CP phase
mWL
is proportional to ∼ and therefore goes to zero in the vR → ∞ limit [12]. If one
mWR
allowed for fine tuning of parameter, one can generate a large enough phase [13] but at the
price of large flavor changing neutral current. Phenomenology of these models have been
extensively studied in literature [14, 15], and it has been established that this scenario fails
to produce large enough CP asymmetry in B → ψK decay in the decoupling limit even

11
when the maximum spontaneous CP phase is allowed. Away from the decoupling limit, the
scenario has been ruled out by the sign correlation between ǫ and the above B-decay CP
asymmetry [14].
In view of these results, for the minimal LRSMs to be realistic and natural, both explicit
and spontaneous CP phases must be taken into account. Anyway, as noted this is precisely
what happens in the minimal model. In Ref. [11], an approximate relation was derived
between the spontaneous CP phase α and the explicit CP phase δ2 in the Higgs potential,
 
−1 2|α2 | sin δ2
α ∼ sin , (39)
α3 ξ
where small ξ requires a hierarchy between α2 and α3 , and/or small δ2 . Clearly, when δ2 = 0,
one has α ∼ 0. A pioneering numerical study of the general CP scenario has been made
in Ref. [5]. We will consider the Higgs spectrum and coupling in this general case in the
remainder of the section.

A. Higgs Mass Spectrum

With the Higgs vev’s in Eq. (7) and the minimization conditions in Eqs. (32)-(37), the
Higgs mass spectrum can be found in the presence of the CP phase δ2 in the Higgs potential
as well as the spontaneous phase α. We further restrict to the case κ′ ≪ κ, vL = 0. Thus
θL becomes irrelevant and all βi decouple. We will keep only terms linear in ǫ and ξ for
simplicity.
In the minimal LRSM, there are 20 scalar degrees of freedom in the Higgs fields φ, ∆L
and ∆R , including 2 double-charged, 4 single-charged and 4 complex neutral Higgs bosons.
After SSB, the mass eigenstates are linear combinations of those. Two single-charged and
two real neutral Higgs bosons get absorbed and become longitudinal components of WL ,
WR , Z and Z ′ .


G+L = −ξe
−iα +
φ2 + φ+1 ,
 
1
G+R = − √ ǫφ+ +
2 + δR ,
2
√
G0Z ′ = 2 Im δR0 ,
√

G0Z = 2 Im φ0∗ 1 + ξe
−iα 0
φ2 , (40)

where we have neglected terms of order ǫ2 and ξ 2 . Among the remaining 14 fields, only one
real and neutral component h0 acquires mass at the electroweak scale κ, identified as the
SM Higgs boson, while the other Higgs fields have masses of order vR . The physical Higgs
states and their masses are collected in Table I. In the limit α → 0, δ2 → 0 and κ, κ′ ≪ vR ,
our results agree with those in Ref. [16], except for the SM Higgs mass.
The calculation of SM Higgs mass is a bit involved and warrants a little further discussion.
From the (tree-level) Higgs potential, one can write down the mass matrix for 8 neutral Higgs
components in the basis of {Re φ01 , Im φ01 , Re φ02 , Im φ02 , Re δL0 , Im δL0 , Re δR0 , Im δR0 }.
The rows and columns containing Re δL0 , Im δL0 and GZ ′ = Im δR0 are already diagonal
and hence decouple.
The remaining 5 × 5 sub-matrix M 2 is somewhat complicated. Since the major com-
ponents have vR -scale masses, we can work in perturbative expansion with respect to 3

12
 Higgs state Mass2
√
 
α21
h0 = 2 Re φ0∗ 1 + ξe −iα φ0
2 4λ 1 − 2
ρ1 κ + α3 vR ξ
2 2
√
H10 = 2 Re (−ξeiα φ0∗ 0
1 + φ2 )
2
α3 vR
√
H20 = 2 Re δR 0 2
4ρ1 vR
0
√ 0 2
H3 = 2 Re δL (ρ3 − 2ρ1 )vR
√
A01 = 2 Im (−ξeiα φ0∗ 0
1 + φ2 )
2
α3 vR
√
A02 = 2 Im δL0 (ρ3 − 2ρ1 )vR 2

H1+ = δL+ (ρ3 − 2ρ1 )vR 2 + 1 α κ2

2
3
H2+ = φ+
2 + ξe iα φ+ + √1 ǫδ +
1 2 R
α3 v 2 + 1 κ2
R 2
++ 2 + α κ2
δR 4ρ2 vR 3
δL++ 2 + α κ2
(ρ3 − 2ρ1 )vR 3

TABLE I: Physical Higgs states and mass spectrum at leading order in minimal LRSMs. We
assume vL = 0 and keep only linear terms in ǫ = κ/vR and ξ = κ′ /κ. All fields but h0 have masses
on the vR scale. h0 is identified as the SM Higgs boson.

parameters: the spontaneous phase α, ξ = κ′ /κ and ǫ = κ/vR . As we shall see later, in

order to satisfy the CP constraints and to have the right-handed scale at TeV, they must be
approximately of the same order: α, ξ, ǫ ∼ O(10−2 ). Naively, the SM Higgs mass squared
should be of electroweak scale ∼ κ2 = ǫ2 vR2 , and we must work up to the order ǫ2 in di-
agonalization: M 2 (α, ξ, ǫ) = M(0)
2
+ M(1)2 2
+ M(2) + · · · . At zeroth order, in the basis of
0 0 0 0 0
{Re φ1 , Im φ1 , Re φ2 , Im φ2 , Re δR } ,
 
0
 
 0 
2 2  
M(0) = vR  α3  , (41)
 
 α3 
4ρ1
which is already diagonal. Re φ01 and Im φ01 have zero mass and the other three have masses
at scale ∼ vR . This means Re φ01 and Im φ01 are the dominant components of the SM Higgs
h0 and the Goldstone boson GZ . At the first order, only the first two rows of M(1) 2
are
relevant,  
2 2 0 0 0 0 2α1 ǫ
M(1) = vR , (42)
0 0 0 α3 ξ 0
2
and at the second order, only the upper 2 × 2 block of M(2) is needed
 
2 2 α3 ξ 2 + 4ǫ2 λ1 0
M(2) = vR . (43)
0 α3 ξ 2
Using the standard formula in perturbation theory, the SM Higgs boson state and mass up
to the second order in the expansion are
√

h0 ≃ 2 Re φ0∗ 1 + ξe
−iα 0
φ2 ,
 
2 α12
mh0 = 4λ1 − κ2 + α3 ξ 2 vR2 . (44)
ρ1

13
There are two contributions to mh0 . One is the usual electroweak breaking mass 4λ1 κ2 ,
α2
and the other is the mass shift α3 ξ 2 vR2 − 1 κ2 , due to additional couplings in the Higgs
ρ1
potential. Because ξ = mb /mt ∼ ǫ = κ/vR , the two parts are roughly comparable when
λ1 ∼ α3 ∼ O(1).

B. Charged and Neutral Currents

In this subsection, we present the charged and neutral couplings between quarks and
Higgs mass eigenstates. From the Yukawa coupling term (12), we can express h and h̃ in
terms of the vev’s and quark mass matrices
Mu κ − Md κ′ e−iα Md κ − Mu κ′ eiα
h= , h̃ = . (45)
κ2 − κ′ 2 κ2 − κ′ 2
Using  
∗ φ0∗
2 −φ+ 1
φ̃ = τ2 φ τ2 = , (46)
−φ−2 φ 0∗
1
one can write
LY = QLi (hij φ + h̃ij φ̃)QRj + h.c. ≡ LN + LC . (47)
The neutral Higgs-quark coupling part is
LN = ūLi (hij φ01 + h̃ij φ0∗ ¯ 0 0∗
2 )uRj + dLi (hij φ2 + h̃ij φ1 )dRj + h.c.
√ 1/2 n 


= 2GF ūLi M̂U ii h0 − iG0Z − 2ξeiα H10 − iA01 uRi
 0

 o
¯ c
+ dLi MDii h + iGZ − 2ξe 0 −iα 0 0
H1 + iA1 dRi
√ 1/2   
+ 2GF ūLi VL M cD V † (H10 − iA01 )uRj
R
ij
  
¯ † 0
+ dLi VL M̂U VR (H1 + iA1 )dRj + h.c. ,0
(48)
ij

where VL,R are left- and right-handed CKM matrices. h0 has the known SM Higgs couplings
to the quark fields. The second term in Eq. (48) changes the quark flavors through H10 and
A01 bosons. They are called the flavor changing neutral Higgs (FCNH) bosons in LRSMs.
The dominant contribution comes from the intermediate top, bottom and charm quark
masses. Taking into account the CKM hierarchy, we found that the transitions from d to
s with intermediate charm quark mass and from b to s with intermediate top quark masse
are most significant.
The charged Higgs-quark coupling part of Lagrangian density is
LC = ūLi (hij φ+ + ¯ − −
1 − h̃ij φ2 )dRj + dLi (hij φ2 − h̃ij φ1 )uRj + h.c.
√ 1/2   
= 8GF ūLi M̂U VR − 2ξe−iα VL McD dRj H2+
ij
 
− ūRi VR McD − 2ξe−iα M̂U VL dLj H +
2
ij
    
− ūLi VL McD dRj G + ūRi M̂U VL dLj G + h.c. .
+ +
(49)
L L
ij ij

14
Again, the couplings are proportional to quark masses and hence the heavy-quark contri-
butions stand out. With the above couplings, we will study their contributions to various
flavor changing and conserving processes in the minimal LRSM.

0
V. K 0 − K AND NEUTRAL B-MESON MASS MIXING

In this section, we consider the neutral kaon and B-meson mass mixing in the minimal
LRSM, using the righthanded quark mixing matrix obtained in the previous section. We
first study the WL − WR mixing-box contribution to the KL − KS mass difference ∆mK and
derive an improved bound (2.5 TeV) on the mass of right-handed gauge boson WR , using
the updated hadronic matrix element and strange quark mass. Historically, the kaon mass
mixing provided the most stringent constraint upper bound (1.6 TeV) on the mass scale of
the right-handed WR boson [17]. With our new right-handed CKM mixing, the conclusion
does not change significantly, although in the literature, quite different mixings have been
speculated upon and the result did change dramatically, and we rule these possibilities out.
In the past few years, significant progress has been made in hadronic physics through lattice
QCD simulations, helping to tighten the bound. We also consider the contribution from
the FCNHs and constraint on their masses. Because the box and FCNH contributions are
additive, the bounds are valid independently. In the last subsection we explore Bd − B d
and Bs − B s mass mixing. Because the hadronic contributions arise dominantly from short
distance, the bounds on MWR and MH turn out to be significant as well.
It is useful to provide our convention for neutral meson mixing at the beginning. For a pair
0
of neutral mesons, |P 0 i and |P̄ 0 i, we assume under CP transformation CP |P 0i = |P i. If the
0
effective hamiltonian in the basis of |P 0 i and P i is Hij = Mij − iΓij /2, a pair of eigenstates
0 p
are |P1,2i = p|P 0i ± q|P i. The ratio q/p is chosen to be (M12 ∗
− iΓ∗12 /2)/(M12 − iΓ12 /2).
In the CP symmetric limit, it is possible to have q/p = 1, and P1 is then CP-even and P2
is CP-odd. The mass difference is M2 − M1 = −2Re(q/p(M12 − iΓ12 /2)), and the width
difference Γ2 − Γ1 = 4Im(q/p(M12 − iΓ12 /2)).

A. Kaon Mixing and the Boxing Diagram

In SM, the leading-order short-distance ∆S = 2 process comes from the box diagram
with WL boson and up-type quark exchanges. Flavor change happens at the vertices via
the CKM mixing matrix. The short distance contribution comes mainly from internal loop
momentum flow at the scales around c and t quark masses, whereas the momentum re-
gion around WL -boson mass is suppressed due to the celebrated Glashow-Iliopoulos-Maiani
(GIM) mechanism. The long-distance contribution comes from one or two up-quark ex-
changes and must be calculated using non-perturbative methods. It has been generally
accepted that this latter contribution does not dominate over the short distance one. In
fact, a chiral perturbation calculation [18] puts the long distance contribution at about half
of the experimental mass difference.
In the LRSM, there are new box-diagram contributions which turn out to be quite large
[17, 19]. The dominant one comes from WL − WR interference with one internal vector-
boson being the lefthanded WL and the other righthanded WR . As such, the chirality of the
internal as well as external quarks must be flipped, and the contribution is proportional to
the internal quark masses, as shown in Fig. 1. There is no GIM suppression even in the

15
 d u,c,t s d WL s

WL WR u,c,t u,c,t

s u,c,t d s WR d

FIG. 1: Dominant contribution to KL -KS mass difference in LRS model.

manifest LRS scenario: VL = VR . The effective Lagrangian from the WL − WR box-diagram

is
G2F MW 2 X
RL √
LLR = − 2
L
2η λLR
i λj xi xj [(4 + ηxi xj )
4π ij
bLR + h.c. ,
×I1 (xi , xj , η) − (1 + η)I2 (xi , xj , η)] O
(50)
 2  2
MWL mi
in which λLR
i
∗
= VLi2 VRi1 , η = , and xi = , i = u, c, t. The four quark
MWR MWL
operator is
bLR (M 2 ) = (sPL d)(sPR d) ,
O (51)
WR

where PL,R = (1 ∓ γ5 )/2 are chiral projection operators. The loop-related integrals are

η ln(1/η)
I1 (xi , xj , η) =
(1 − η)(1 − xi η)(1 − xj η)
 
xi ln xi
+ + (i ↔ j) ,
(xi − xj )(1 − xi )(1 − xi η)
ln(1/η)
I2 (xi , xj , η) =
(1 − η)(1 − xi η)(1 − xj η)
 
x2i ln xi
+ + (i ↔ j) .
(xi − xj )(1 − xi )(1 − xi η)
(52)

The above contribution is evaluated in ’t Hooft-Feynman gauge. Generally, we have other

contributions from the exchange of single-charged Higgs bosons as well as the Goldstone
bosons if the calculation is not in unitary gauge. In fact, to maintain gauge invariance, we
have also to include the triangle diagrams with the FCNH bosons and the related self-energy
corrections [20]. It turns out that in the gauge we use, the box diagram with WL and WR
interchanges alone gives the dominant contribution.
The Wilson coefficient in Eq. (50) is proportional to the internal quark masses and the
corresponding CKM matrix elements. The up-quark contribution is negligible and the top-
quark contribution is suppressed by the off-diagonal CKM elements. Therefore the charm-

16
quark exchange dominates the effective interaction which simplifies
 
G2F MW 2
LR RL 1
HLR = L
2ηλc λc xc 1 + ln xc + ln η
4π 2 4
 2 2

× (sd) − (sγ5 d) . (53)
The hadronic matrix element of the four-quark operator is expressed in the vacuum satura-
tion form,
 2
2 mK
< K0 |d(1 − γ5 )sd(1 + γ5 )s|K 0 >= −2MK FK B4 (µ) , (54)
ms (µ) + md (µ)
where the kaon decay constant FK = 113 MeV and B4 = 1 corresponds to vacuum saturation
approximation. In this form, the matrix element diverges in the chiral limit, an important
reason for the enhanced contribution of the box diagram. The correction factor, B4 , can be
and has been calculated using lattice QCD. In a recent calculation [21], the domain-wall
fermion was used, and B4 = 0.81 was found at µ = 2 GeV in naive dimensional regularization
scheme. In the same scheme and scale, the strange quark mass is ms = 98(6) MeV, which
is smaller than what one has naively expected in the past.

B. An Improved Lower Bound on MWR

Considering only the box diagram, the new contribution to the mass difference of KL −KS
can be expressed as
∆mK = 2 Re η4 (µ)hK 0 |H ∆S=2(µ)|K̄ 0 i , (55)
where η4 is a factor characterizing the QCD radiative correction in scale running from MWR
to µ ∼ 2 GeV [22, 23, 24]. There are several enhancement factors here comparing to the SM
box diagram. First, due to absence of the GIM mechanism, the Wilson coefficient is about
a factor of 30 larger. Second, the hadronic matrix element is chirally enhanced by a factor
of 20. Finally, the short-distance QCD correction η4 = 1.4 gives another enhancement. The
only suppression comes from 2the difference between the left and right-handed symmetry
MWL
breaking scales, η = . Therefore the new contribution can be approximated by
MWR
 2
3 MWL
∆MK−LR ∼ 10 × × ∆MK−SM . (56)
MWR
The sign can both be positive or negative depending on the product st sc . With the standard
criteria that the new contribution should not exceed the experimental value [17], we find a
lower bound for MWR ,
MWR > 2.5 TeV . (57)
On the other hand, the SM contributions from both long and short distances have the
same sign as the experimental number and account for more than one-half of its value.
Therefore, a less conservative bound is obtained if requiring the new physics contribution
is less than one-half of the experimental data. If this new standard is adopted, the above
bound change to 4 TeV. Giving the long-distance uncertainty in ∆MK , this bound shall
be used at less confidence level. Nonetheless, as we shall see in the next section, the CP
violating observables are providing equally competitive bounds albeit with a significant
hadronic physics uncertainty.

17
 C. Tree-Level FCNH Contribution and A Lower Bound on MH

0
In the LRSM, there is also a new contribution to the K 0 − K mixing mediated by the
FCNH. The FCNH boson is a complex field and can be expressed in terms of the two real
fields H10 and A01 . The effective lagrangian follows from Eq. (48)
 !2 " #
GF  X λRL i + λ LR
i (sd) 2
(sγ 5 d) 2
LF CN H = √ mi 2
− 2
2 i
2 m H10
m A01
!2 " # 
X λRL − λLR (sd) 2
(sγ 5 d) 2
− i i
mi −  . (58)
2 2
i
2 m A 0 m H 0
1 1

The corresponding Feynman diagram is shown in Fig. 2. According to our previous discus-
sion, the two scalar fields H10 and A01 have the same masses, roughly corresponding to the
righthand scale, m2H 0 ≃ m2A0 ≃ α3 vR2 . Therefore, it is convenient to rewrite Eq. (58) in a
1 1
more compact form
GF X  
HF CN H ≃ − √ 2
mi mj λLR RL
i λj (sd)2 − (sγ5 d)2 . (59)
2mH 0 i,j
1

d s d s

H10, A10
H10, A10

s d s d

FIG. 2: ∆S = 2 effective interaction induced by flavor-changing neutral Higgses.

It is easy to check that the FCNH and the box diagram contributions have the same sign
because 4(1+ln xc )+ln η < 0, and thus they cannot cancel each other, even allowing possible
freedom in choosing the quark mass sign. Therefore, the lower bound on the righthanded-W
boson mass remains. One can also derive a lower bound on the masses of H10 and A01 using
∆MK . A straightforward calculation shows that if demanding the FCNH contribution is
less than the experimental data,

MH10 , MA01 > 15 TeV . (60)

which is about twice as large as in [5]. One can obtain this value presumably by a large α3
parameter in the Higgs potential. However, one cannot make α3 arbitrarily large. As we
shall discuss later, large α3 not only causes naturalness problem, but also leads to a large
SM Higgs mass which threatens the perturbative unitarity [25].

18
 D. Constraints From Neutral Bd and Bs Mass Mixing

0
The physics of Bd − B d and Bs − B s mixing in SM is similar to that of K 0 − K mix-
ing, coming from the W -boson box diagram and FCNH. However, in the former case, the
intermediate top quark contribution dominates almost entirely due to its mass and CKM
couplings. Because of this, the SM calculation can be done quite accurately with the help
of the lattice QCD matrix elements. In fact, in many global CKM fits, the mass differences
∆MB and ∆MBs have been used to determine the top CKM couplings Vtd and Vts . However,
there are still appreciable uncertainties in the lattice calculations and global CKM fits, and
the beyond SM physics could contribute as much as 20% of the mass differences without
running into conflict with the present SM calculations and experimental data. We will use
this possible discrepancy as a constraint on the LRSM.
The LRSM contribution to the mixing can be taken directly from (50)

q G2F MW 2
L∗ R R∗ L

HLR (µb ) ≃ 2
L
2η S η(V tb V tq )(V tb V tq )xt (4 + ηx2t )
4π
×I1 (xt , xt , η) − (1 + η)I2 (xt , xt , η)] b̄PL q b̄PR q + h.c. , (61)

where the leading-logarithmic running factor ηS = 2.112 [22] and the functions Ii are defined
in Eq.(52). The hadronic matrix element can be defined using a factorization,
"  2 #
q 1 1 m B
< Bq |qPL bqPR b|B q >= −MBq fB2 q B4 (µ) + q
, (62)
12 2 mb + mq

where the minus arises from our definition of the CP transformation for the meson states.
The decay constants have been calculated in lattice QCD: fBd = 216 MeV, fBs = 1.20fBs [26]
and the non-perturbative B-factors are B4d = 1.16 and B4s = 1.17 [27]. The ratio of the new
contribution to the SM one is ∼ 102 MW2
L
/MW2
R
, which is smaller than the kaon mixing case
in the absence of chiral enhancement.

1 35
30
D M Bs Hps-1 L

0.8
D M Bd Hps-1 L

25
0.6 20
0.4 15
10
0.2 5
1 2 3 4 5 1 2 3 4 5
MR HTeVL MR HTeVL

FIG. 3: Bd -B d and Bs -B s mass differences from the WL − WR box diagram with top-quark inter-
mediate state plus the SM contribution. The shaded regions are within 20% of the experimental
values.

The experimental values for the mass differences of Bd and Bs are (0.507 ± 0.12)ps−1 and
(17.77 ± 0.12)ps−1 , respectively [9], with central values shown as horizontal lines in Fig. 3.

19
The shades around the central values are within 20%, and are considered as the combined
experimental and SM theory error. In the same figure, we also plot the mass differences
as a function of MWR , calculated as a sum of the LRSM contribution and the experimental
central values. The agreement between experiment and theory is reached only when MWR is
larger than 2.5 TeV. The sign of the LRSM contribution is related to that of sd sb and ss sb ,
which we have chosen to be +1. If taking as −1, approaching to the central value comes
from below.
Thus the constraint on the MWR mass from the neutral B-meson mixing is roughly
comparable to the kaon case, due to a better theoretical understanding of the SM physics.
The future improvement can come from a better determination of Vtd from other sources
and a better determination of B-parameter and decay constants.
We have also studied the constraint on the FCNH mass, MH . We find that from ∆MBd ,
the bound is 12 TeV, and for ∆MBs , the bound is 25 TeV.

VI. CP-VIOLATING OBSERVABLES IN LRSM

As mentioned before, in a generic yet minimal LRSM we have both explicit and spon-
taneous CP violations. A prominent feature of right-handed quark mixing VR obtained in
the previous section is its phases, which are entirely determined by the Dirac phase δCP and
the spontaneous CP phase α. These physical phases generate interesting effects in various
CP-violating observables. The phenomenology of CP violation in LRSM is rich, which in
turn constraints the model severely. In this section, we will explore a number of CP violat-
ing observables including ǫ, ǫ′ , neutron EDM and CP asymmetry in B → J/ψKS , to place
constrains on the mass of WR as well as the spontaneous CP phase α. Some results here
have appeared before in a Rapid Communication paper [7] and there are important updates
in this more extensive study.
The indirect CP violation ǫ receives large contributions from both explicit and sponta-
neous CP phases. Unless there is a strong cancelation, the right-handed WR mass must
be larger than 15 TeV. We use the cancelation condition to fix the spontaneous CP phase,
which is then used to predict the neutron EDM. Using the experimental bound on the EDM,
we obtain a strong lower bound on MWR , which can be improved with better calculations of
the hadronic matrix elements and more precise experimental data. We obtain a strong lower
bound on MWR from the direct CP violation parameter ǫ′ , calculated under the factorization
assumption for the four-quark matrix elements. Therefore, we find that the CP violating
observables in the kaon system and neutron EDM provide competitive bounds as the well-
known kaon mass mixing. These bounds can be improved further with better knowledge of
the non-perturbative hadronic physics.

A. Indirect CP violation ǫ in Kaon Decay

We first study the CP violating parameter ǫ in kaon mixng. This indirect CP violation
parameter is related to the flavor mixing interaction by,
0
eiπ/4 ImhK 0 |H ∆S=2 |K i
ǫ=− √ , (63)
2 ∆mK

20
where we have neglected the direct CP contribution ξ0 from kaon decay, which can be
justified posteriori [5]. In LRSM, according to the previous section, both the WL − WR box
diagram in Eq. (53) and the tree level FCNH exchange in Eq. (59) can make significant
contributions. In the present case, their signs can be different due to both charm and top
quark contributions, in contrast to the mass mixing [5]. For simplicity, we ignore that latter
contribution and consider the constraint from the box diagram alone.
There are two sources of CP phases in VR which enter the WL − WR box diagram: the
Dirac phase δCP inherited from VL , and the spontaneous phase α. In the manifest LRS case,
only δCP is present and there is a very tight lower bound on mass of WR which we find no
lighter than 15 − 20 TeV (see below). If the spontaneous CP phase is also present, one can
seek for certain cancelation between the two contributions to lower the bound on MWR . In
fact, one can roughly estimate the size of r sin α for a cancelation. The Dirac phase δCP
appears in the expression ǫ proportional to VtsR∗ VtdR ∼ λ5 sin δCP via top quark exchange in
the box diagram, while the spontaneous CP phase contributes through VcsR∗ VcdR ∼ λ2 r sin α.
Hence r sin α should be of order λ3 sin δCP ∼ 0.01 when cancelation happens.

30 30
25 25
M HTeVL

20 20
15 15
R

10 10
5 5

-0.2 -0.1 0.0 0.1

-0.1 0.0 0.1 0.2

r sin Α
30 30
25 25
20 20
15 15
10 10
5 5

-0.5 0.0 0.5 -0.5 0.0 0.5 1.0

FIG. 4: Typical scenarios from the ǫ constraint: The first and two figures correspond to sd = ss = 1
and sd = ss = −1, respectively (small r sin α solution). The third and fourth correspond to sd =
−ss = 1 and sd = −ss = −1 respectively (large r sin α solution). In all cases, su = sc = st = sb = 1.

The present experiment value is |ǫ|expt = (2.232 ± 0.007) × 10−3 [9]. In SM, ǫ can be
calculated quite accurately because the top quark dominates the box diagram and the main
contribution is due to short-distance QCD physics. The only large uncertainty comes from
0
the CKM matrix element Vtd and the hadronic matrix element related to K 0 − K mixing.
Because of this, we assume that the new contribution accounts for less than 1/4 of the
experimental value.
Using the box-diagram result in the previous section, we find an approximate expression
for ǫLR valid for MWR > 200 GeV ,
 2
1 TeV  
ǫLR = 1.58 ss sd Im g(MWR , θ2 , θ3 )e−i(θ1 +θ2 ) , (64)
MWR

21
where

g(MWR , θ2 , θ3 ) = −2.22 + [0.076 + (0.030 + 0.013i) sc st

 2
MWL
× cos 2(θ2 − θ3 )] ln , (65)
MWR
where θi are given in Eq. (29) and the QCD running correction has been taken into account.
When θi = 0, the dominant CP violating contribution comes from charm-top interference in
the box diagram.
We search for the allowed region in MWR -r sin α plane shown in Fig. 4. The spontaneous
CP violation effect can always cancel the effect of the Dirac phase, thus ǫ itself places no
bound on MWR . The cancelation depends on choices of the quark mass signs. The 32 choices
of the signs (for fixed su = 1) roughly correspond to two scenarios: small r sin α solutions
for ss = sd and large r sin α solutions for ss = −sd . The existence of two scenarios can be
readily seen through the above approximate expression. According to (29), θ1 and θ2 are
similar in size and their relative sign depends on the bi-product sd ss . When ss = sd , θ1 and
θ2 have the same sign and add constructively in the exponential in (64). So α itself must
be small in order for θ1 + θ2 to cancel the phase in g(MWR , θ2 , θ3 ). This generates the small
r sin α solutions. On the other hand, if ss = −sd , θ1 + θ2 cancels and is proportional to
r sin α multiplied with a small coefficient. So r sin α has to be large to cancel the phase in
g(MWR , θ2 , θ3 ). This corresponds to the large r sin α solutions. For the small r sin α case, we
find r sin α ≃ ±0.05, but for large r sin α, it can take several different positive and negative
values. As we will see in the next subsection, if one includes the constraint from neutron
EDM, only small r sin α is phenomenologically viable.

20
M HTeVL

15
10
R

5
0
0.00 0.05 0.10 0.15 0.20

r sin Α

FIG. 5: Constraint on MWR and the spontaneous phase α from kaon decay parameter ǫ, with
sd = ss = −1 and all other si = 1. Red triangles are for MH = ∞, blue squares are for MH = 75
TeV, and green dots for MH = 20 TeV.

Of course, one has to consider the FCNH contribution which has been known to be large
[28]. In fact, with just the Dirac CP phase in the FCNH contribution, ǫ places a limit on
the Higgs mass on the order of 100 TeV. With the new spontaneous CP contribution, there
is a possibility of cancelation. In fact, one can make similar plots as in Fig. 4, in which the ǫ
bound can be satisfied even for very low MH ∼ 1 TeV. However, the required r sin α for the

22
cancelation, ±0.2, is very different from that needed for the box diagram. The conclusion is
that there is no bound on MH coming from ǫ when FCNH contribution is considered alone.
Because of the conflict in the spontaneous CP phase required for individual cancelations
in the box and FCNH contributions, one might expects their combined contribution places a
joint bound on MWR and MH . This, however, is not the case, because the two contributions
again cancel each other, as was first found in [5]. In fact, with the analytical solution for
the right-handed mixing, we arrive at an even stronger conclusion: For any given pair of
MWR and MH , we can always find a r sin α such that the total contribution to ǫ vanishes.
This situation is extremely interesting, because it implies that ǫ itself, unlike the kaon
mass difference, is completely useless in constraining the individual parameters in the new
contribution. It does, however, provide a correlation among different parameters, as shown
in Fig. 5, where for several different values of MH = ∞, 75, 25 TeV, we plotted the allowed
regions in the MR and r sin α plane. Because of the FCNH contribution, the pattern of the
correlation changes considerably as MH changes. The general trend is that the spontaneous
CP parameter r sin α increases toward 0.2 as MH becomes smaller, consistent with the
cancelation pattern in FCNH contribution to ǫ found above.

B. Neutron EDM

The neutron EDM imposes another constraint on the new CP phases in VR and MWR .
Non-zero EDM implies both P and T (or equivalently CP in local quantum field theories)
violations. At the quark level, sources of flavor-neutral CP violation are mainly from the
penguin diagrams in the SM, and from the tree level WL − WR exchange in the LRSM
[29, 30]. Generally, there are several contributions to the neutron EDM, including valence
quark EDM, quark chromomagnetic dipole moment (CDM) induced EDM, dimension-6 pure
gluonic operator contribution, as well as the contributions at hadronic level. The present
experimental upper bound on neutron EDM is 3.1 × 10−26 e cm [9].
In the SM, contributions to the neutron EDM mainly come from the CP violating pen-
guin diagrams. The flavor changing nature of CKM CP violation means that the leading
contribution is at least second order in weak interaction (∼ G2F ). The predicted neutron
EDM is well within the experimental bound—about 10−33 e cm.

u d (s)

WL

WR

d (s) u

FIG. 6: Dominant quark-level effective operators contributing to neutron EDM in LRSM.

In the LRSM, the flavor-conserving CP violating four-quark operator arises from the
tree-level diagram with WL − WR mixing exchange, as shown in Fig. 6
√ L R∗ q q
Luq→uq = −2 2GF sin ζe−iα Vuq Vuq (O− − O+ ) + h.c. , (66)

23
where q = d, s, and
q 2
O+ = ūγ µ PL q q̄γµ PR u − ūPR uq̄PL q ,
3
q 2
O− = ūPR uq̄PL q . (67)
3
q q
At low energy, short-distance QCD effect enhances the operator O− and suppresses O+ . The
effective Lagrangian reduces to [31]
√
2 2GF 


Lud(s)→ud(s) = −i L R∗
η− sin ζ Im e−iα Vud Vud ūγ5 ud̄d − ūudγ ¯ 5d
3

L R∗
+ Im e−iα Vus Vus (ūγ5 us̄s − ūus̄γ5 s) , (68)
  98
αS (µ2 )
where the leading-log QCD factor η− = 2
≃ 3.5. The CP violating πnn
αS (MW L
)
coupling is proportional to the hadronic matrix element ḡπnn = hπn|Lud(s)→ud(s) |ni which,
in the factorization approximation, is
hπn|q̄γ5 q q̄ ′ q ′ |ni ≃ hπ|q̄γ5 q|0ihn|q̄ ′q ′ |ni , (69)
for q = u, d and q ′ = u, d, s. From the SSB of the chiral symmetry, 2mu hπ|ūγ5 u|0i =
¯ 5 d|0i = −iFπ m2 , with Fπ = 93 MeV. hn|ūu|ni ≃ 4 and hn|dd|ni
−2md hπ|dγ ¯ ≃ 5 can be fixed
π
1
from the neutron-proton mass difference and the πN σ-term: σN = (mu + md )hn|ūu +
2
¯
dd|ni ≃ 45 MeV [32], where we use quark masses mu = 2.7 MeV, md = 5.0 MeV at µ = 2
GeV. We neglect hn|s̄s|ni ≪ hn|ūu|ni.
The neutron EDM from the Feynman diagram in Fig. 7 is
 2
e gπnn ḡπnn µN mπ
dn = F , (70)
8π 2 2mN m2N
where µN = −1.91 is the neutron anomalous magnetic dipole moment and the loop function
is p
3 3s − s2 s(5s − s2 ) − 4s s − s2 /4
F (s) = − s − ln s + p arctan . (71)
2 2 2 s − s2 /4 s/2
The contribution is suppressed by the mixing
2 angle ζ between WL and WR , but is enhanced
Λχ
by the chiral logarithmic factor ln . Putting in all the known physical parameters,
mπ
we arrive at an approximate formula


|den | ≃ 3 × 10−19 sin ζ Im e−iα Vud
L R∗
Vud ecm , (72)
which is approximately a function of r sin α for small α.
For a fixed MH , the neutron EDM and ǫ can be used to give a lower bound for MWR
as well as a corresponding solution for r sin α. On the left panel in Fig. 8, we have shown
the neutron EDM constraint as a function of r sin α and MWR , and it is obvious that the
EDM limit prefers small r sin α. The smallest spontaneous CP phase is obtained when MH
is large and decouples, and then
MWR > 8 TeV , (73)

24
 n n n

FIG. 7: A dominant contribution to the neutron EDM through chiral pion exchange. The shaded
blob is CP violating coupling ḡπnn , and the black dot is the strong coupling gπnn . The neutron
couples to photon via its anomalous magnetic moment.

which is a very tight bound. At this point, we also fix the product
r sin α ≃ 0.05 . (74)
For a lower MH , the spontaneous CP phase must be large from the ǫ constraint, and the
corresponding lower bound on MWR increases considerably. For example, when MH = 75
TeV, r sin α is now greater than 0.1, and the lower bound on MWR becomes ∼ 18 TeV.
We note that there is considerable hadronic uncertainty in the evaluation of CP-violating
coupling ḡπnn and, to the less extent, chiral perturbation expansion. However, even one
allows a factor of 5 over-estimate in the hadronic calculation, the combined ǫ and EDM
will still provide a strong constraint on MWR on the order of order 4 TeV, as shown on
the right panel in Fig. 8. If MH = 25 TeV, the lower bound on MWR becomes 8 TeV. A
future improvement on the neutron EDM data and theoretical calculation can strengthen
this bound considerably.
We finally comment on the Higgs boson exchange contributions to the neutron EDM.
According to Sec. IV B, the Higgs bosons H10 , A01 and H2+ have CP violating couplings (Eq.
(48) and Eq. (49)) to the quark fields. The valence quark EDM receives contributions from
virtue H10 , A01 and H2+ exchange as shown in Fig. 9. Potentially large contribution also comes
from neutral Higgs boson exchange and virtual top-quark effect at two loops [33], as well
as two loop pure gluonic operators due to charged/neutral Higgs exchange [34]. A complete
analysis of these contributions has been carried out in a pseudomanifest LRS limit with two
doublets Higgs fields instead of triplets considered here [35]. With the explicit form of right-
handed CKM mixing, we re-evaluate these contributions in the general case of CP violation.
As discussed in Sec. V C, H10 and A01 must be heavy enough to suppress their contribution
to kaon mixing, and we take their masses to be 15 TeV. In this case, the charged-Higgs H2+
exchange dominates, whose contribution to d-quark EDM is approximated by
!
2e mt GF m2t m2H +

dd ≃ √ 2ξ 2 ln 2
2
ηd Im VtdL VtdR∗ e−iα , (75)
3 4 2π 2 mH + mt
2

where the scaling factor ηd ≃ 0.12. The d-quark CDM fd come from a similar diagram with
3 ηf
photon replaced by gluon leg, efd ≃ dd . The contribution to u-quark EDM and CDM
2 ηd

25
 20 12
10

M HTeVL
M HTeVL

15
8
10 6

R
R

4
5
2
0 0
-0.05 0.00 0.05 0.10 0.15 0.20 -0.05 0.00 0.05 0.10 0.15 0.20

r sin Α r sin Α

FIG. 8: Constraints on the mass of WR and the spontaneous CP violating parameter α from kaon
decay parameter ǫ (MH = ∞, red triangle; MH = 75 TeV, blue square; MH = 20 TeV, large green
dots) and neutron EDM (yellow dots). In the right panel, the theoretical EDM result is reduced
by a factor of 5.

mb
is suppressed by a factor and is negligible compared to that of d-quark. Meanwhile, the
mt
two loop diagrams are found to be negligibly small. To a certain level of approximation, the
neutron EDM can be related to the quark EDM and CDM through the SU(6) relation:
 
e 1 1 4 2
dn = (4dd − du ) + fd + fu . (76)
3 3 3 3
A more accurate relation would use the tensor charges of the nucleon.

FIG. 9: Higgs exchange contributions relevant to neutron EDM. The dashed lines include both
charged and neutral Higgs bosons exchanges. The two-loop diagrams contain closed top-quark
loops.

In Fig. 10, we plot the Higgs exchange contributions to the neutron EDM as a function
of r for different values of α. We choose the FCNH mass to be 15 TeV, and the charged

26
 0.25
1.0
0.20

dn e H ´ 10-26 e cmL
0.15
0.5
0.10

0.05 0.0

0.00

0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

r r

FIG. 10: One scenario (all sq = 1) for the higgs exchange contribution to neutron EDM as a
function of r for different values α = 0.05 (solid line), α = 0.5 (long dashed line) and α = 1 (short
dashed line). We choose the FCNH mass to be 15 TeV and the charged Higgs mass equal to 3
TeV.

Higgs mass be 3 TeV. We find the contributions are always smaller than 10−26 e cm, well
within the experimental bound. Therefore one can neglect the Higgs exchange contribution
without altering the W -mass bound for the neutron EDM.

C. Direct CP Violation ǫ′

The direct CP violation in neutral kaon to ππ decay is calculated via

  
i q ImA2 ImA0
′
ǫ =√ ω − ei(δ2 −δ0 ) , (77)
2 p ReA 2 ReA 0

where the decay amplitudes A0 and A2 are defined as the matrix elements of the ∆S = 1
effective Hamiltonian between the neutral-K meson and the isospin I = 0 and 2 ππ states,

h(2π)I |(−i)H∆S=1 |K 0 i = AI eiδI . (78)

δI is the strong phase for ππ scattering, ω ≡ A2 /A0 and p, q are the mixing parameters for
0
K 0 − K . To an excellent approximation, ω can be taken as real and q/p = 1. Therefore,
we focus on calculating the imaginary part of the decay amplitudes.
In the SM, the contributions to ǫ′ come from both QCD and electromagnetic penguin
diagrams [36]. The QCD penguin contributes exclusively to ∆I = 1/2 decay, whereas the
electromagnetic penguin is mainly responsible for ∆I = 3/2 decay. Both contributions
are important but have opposite signs. Therefore, the final result depends on precision
calculation of hadronic matrix elements. The state-of-art chiral perturbation theory [24,
37, 38, 39] and lattice QCD calculations [40, 41] have not yet been sufficiently accurate to
reproduce the experimental result [42]. A review of the standard model calculation can be
found in Ref. [43, 44].
In LRSM, each element in the righthanded CKM matrix has a substantial CP phase.
As a consequence, there are tree level contributions to the phases of A2 and A0 . Following
closely the work by Ecker and Grimus [23], the ∆S = 1 effective Hamiltonian from Eq. (11)
and the tree-level Feynman diagram in Fig. 11 is

27
 " − 2b  4 #
tree
√ αS (µ2 ) αS (µ2 ) b LL
H∆S=1 = 2GF λLL
u
LL
O+ (µ) + O− (µ)
αS (ML2 ) αS (ML2 )
" − 2b  4 #
√ ML2 RR αS (µ2 ) RR αS (µ2 ) b RR
+ 2GF 2 λu O+ (µ) + O− (µ)
MR αS (MR2 ) αS (MR2 )
"  8b  − 1b #
√ α S (µ 2
) α S (µ 2
)
+ 2 2GF sin ζλLR
u e
iα
O−LR
(µ) − O+LR
(µ)
αS (ML2 ) αS (ML2 )
"  8b  − 1b #
√ α S (µ 2
) α S (µ 2
)
+ 2 2GF sin ζλRL
u e
−iα RL
O− (µ) − RL
O+ (µ) , (79)
αS (ML2 ) αS (ML2 )

where we have taken into account the leading-logarithm QCD corrections with renormaliza-
tion scale µ taken to be around 1 GeV, and b = 11 − 2Nf /3 with Nf the number of active
fermion flavors. The mixing coupling λAB
u
CKM∗ CKM
= VAud VBus , A, B are L, R. The four quark
operators are
LL,RR
O± = dγ µ PL,R uuγµ PL,R s ± dγ µ PL,R suγµ PL,R u ,
LR,RL 2
O+ = dγ µ PL,R uuγµ PR,L s + dPR,L suPL,R u ,
3
LR,RL 2
O− = dPR,L suPL,R u , (80)
3
where PL,R are projection operators. There are also new penguin diagrams involving the
right-handed gauge boson contributing to ǫ′ . However, these contributions are suppressed
by loop factors and are neglected here.
u s u s
WL
WR
WR
d u d u

FIG. 11: New tree-level contribution to the ∆S = 1 interaction from LRSM.

The hadronic matrix elements of the four-quark operators are calculated using the fac-

28
torization assumption,
LL,RR 0 X
h(2π)I=0 |O+ |K i = ± √ ,
3 6
LL,RR 0 2X
h(2π)I=2 |O+ |K i = ± √ ,
3 3
LL,RR 0 X
h(2π)I=0 |O− |K i = ± √ ,
2 6
LL,RR 0
h(2π)I=2 |O− |K i = 0 ,
LR,RL 0 4X
h(2π)I=0 |O+ |K i = ± √ ,
9 6
LR,RL 0 2X
h(2π)I=2 |O+ |K i = ± √ ,
9 3
 
LR,RL 0 1 X Y Z
h(2π)I=0 |O− |K i = ∓ √ + + ,
6 18 2 6
 
LR,RL 0 1 X
h(2π)I=2 |O− |K i = ∓ √ −Z , (81)
6 3 6
where the parameters X, Y and Z are
0
¯ µ γ5 u|0ihπ +|ūγ µ s|K i ,
X ≡ −hπ − |dγ
√
= i 2Fπ (m2K − m2π ) ≃ 0.03i GeV3
0
Y ≡ −hπ + π − |ūu|0ih0|d̄γ5 s|K i ,
√
= i 2FK A2 (1 − m2K /m2σ )−2 ≃ 0.273i GeV3
¯ 5 u|0ihπ +|ūs|K 0 i
Z ≡ −hπ − |dγ
√
= i 2Fπ A2 (1 − m2π /m2σ )−2 ≃ 0.18i GeV3 , (82)
where A = m2K /(ms + md ), Fπ = 93 MeV and FK = 1.22 Fπ .
The numerical estimates are taken from Ref. [15]. Y and Z are much bigger than X
due to chiral enhancement. Clearly the factorization approximation must be improved as
indicated by the empirical ∆I = 1/2 rule, which is beyond the scope of this paper. We note,
however, that for our estimation of the bound on MWR , a multiplicative uncertainty factor
on the matrix elements is reduced by a square-root.
To calculate the weak phases of the decay amplitudes A0 and A2 , we use the experimental
value for the real parts of A0 and A2 : ReA0 ≃ 3.33 × 10−7 and ω ≃ 1/22. The dominant new
contribution is from the WL − WR mixing term due to enhanced hadronic matrix elements
and larger CP violation phase α. In fact, because the phase α in the apparent factor eiα is
much larger than the phase in λRL,LR
u which is typically of order θ1 and θ2 given in Eq. (29),
ǫ′ is approximately a function of r sin α, rather than r and sin α independently. Since r sin α
has been fixed by ǫ and den in the previous subsections, ǫ′ is approximately a function of
1
MWR only. In Fig. 12, we plot ǫ′ as a function of MWR for α = arcsin , r = 0.5 and
10
sd ss = 1 which is required by the neutron EDM calculation. [All si = 1.] Requiring that the
new contribution should be no larger than |ǫ′expt | = 3.92 × 10−6 [9], we get a lower bound
MWR > 4.2 TeV , (83)

29
 9. ´ 10-6
8. ´ 10-6
7. ´ 10-6
6. ´ 10-6

Ε'
5. ´ 10-6
4. ´ 10-6
3. ´ 10-6
2. ´ 10-6
3.0 3.5 4.0 4.5 5.0 5.5 6.0
MR HTeVL

1
FIG. 12: ǫ′ as a function of MWR for α = arcsin , r = 0.5 and all sq = 1. The shaded part is
10
allowed by the experimental data

Here we obtain a slightly tighter bound on MWR than that from kaon mixing. However,
because of the ∆I = 1/2 rule, the factorization assumption might have overestimated the
phase of A2 . If we take r sin α = 0.15, as required by low MH , the bound changes to 7.4
TeV.
Finally, there are also tree-level FCNH contributions to H∆S=1 , as one can see from the
lagrangian in Eq. (48) [15, 23, 30]. Since the relevant coupling is suppressed by either the
Cabibbo angle or the quark masses, their contribution is negligible.

D. CP Violation in Bd → J/ψKS Decay: SJ/ψK

The CP violation in B-meson decay was first observed in Bd → J/ψKS . In SM, the decay
proceeds mainly through the tree-level b → cc̄s and the penguin contributions are expected
to be suppressed by CKM and/or loop factors. The tree-level diagram is shown in the left
panel in Fig.13 with an intermediate WL exchange.

b c b c
WL
W L, R
c WR
c

s s

FIG. 13: Tree level Feynman diagrams for Bd → J/ψKs from SM and LRSM.

The relevant CP asymmetry is defined as

2 Imλd
SJ/ψKS = , (84)
1 + |λd |2

30
where
 
q A (B̄d → J/ψKS )
λd = · ,
p B A (Bd → J/ψKS )
A is a decay amplitude and (q/p)B is from the B-meson mixing.
The magnitude of λd is close to 1 and thus SJ/ψKS ∼ Imλd . In the SM, SJ/ψKS predomi-
nantly comes from B − B̄ mixing, and is related to the β angle of the unitary triangle because
the ratio of the decay amplitude is independent of the hadronic matrix element when the
tree operator dominates. Experimentally, sin 2βexpt = 0.673 ± 0.028 [9].
In the LRSM, the effective β angle will receive new contributions from initial and final
neutral meson mixings [3] and from the new tree-level decay operators through WR exchange
and WR −WL mixing, as shown in Fig. 13. We will not consider the kaon mixing contribution
0 0
for the following reason: Since KS is dominantly CP-even |KS i = p|K 0 i+q|K i and B decay
0
involves K , λd is proportional to (q/p)∗K . The imaginary part of (q/p)∗K is proportional to
the imaginary part of ǫ which is known to be on the order of 10−3 , much smaller than the
phase in (q/p)B . The ǫ constraint on the LR symmetric model has already been studied
independently and we decouple the kaon-mixing effect from SJ/ψKS . Thus we write
!
Bd ,LR
M
2β ef f ≈ 2β + arg 1 + 12 Bd ,SM
M12
  −1
hJ/ψK 0 |H LR |B 0 i hJ/ψK0|H LR |B0 i
+ arg 1 + 1+ . (85)
hJ/ψK 0 |H SM |B 0 i hJ/ψK0|H SM |B0 i
The H LR operator is similar to that in Eq. (79), with substitutions s → b, u → c, and
d → s.
Bd ,LR Bd ,SM
The ratio M12 /M12 can be calculated from Eq. (58) and a corresponding ex-
pression from the SM. Its magnitude is around 102 MW 2
L
2
/MW R
and carries a phase factor
R∗ R L∗ L −i(θ1 +θ3 )
Vtd Vtb /Vtd Vtb = sd sb e . The hadronic matrix element for the decay is less known.
In the naive factorization approximation for hJ/ψK 0 |H SM |B 0 i, the decay rate is under
predicted by an order of magnitude [45]. Therefore, the non-factorization contribution must
be significant [46]. In the ratio of matrix elements, we expect the factorization approach
work better. Using this approximation, we find
hJ/ψK 0 |H LR |B 0 i 2
MW λRR
c 2 [αS (mb )/αS (MWL )]−2/b − [αS (mb )/αS (MWL )]4/b
= 2
L
, (86)
hJ/ψK 0 |H SM |B 0 i MW R
λLL
c 2 [αS (mb )/αS (MWR )]−2/b − [αS (mb )/αS (MWR )]4/b
where λAB
c = VcsA∗ VcbB for A, B = L, R, so λRR LL
c /λc = ss sb e−i(θ2 +θ3 ) . The non-perturbative
2
B → K form factors F+ (mJ/ψ ) has been canceled out and the result is independent of
2 2
hadronic parameters. The magnitude of the ratio is O(MW L
/MW R
) and hence is much
smaller than the mixing contribution to 2βeff . This conclusion remains valid even if we
underestimate this ratio by an order of magnitude.
The modified CP asymmetry in LRSM depends on the righthanded scale MWR and r sin α.
We take r sin α ≃ 0.05 as determined from the ǫ constraint. There are two independent
choices for quark mass signs which generate different predictions. We take either st = +1
or st = −1, and the results are shown in Fig. 14. Demanding sin 2βef f to lie within the
experimental error bar, we get a moderate lower bound on MWR ,
MWR > 0.7 TeV . (87)

31
 0.74
0.72
0.70

sin 2Βeff
0.68
0.66
0.64
0.62
0.60
0.0 0.5 1.0 1.5 2.0 2.5 3.0
MR HTeVL

FIG. 14: Predicted CP asummetry in Bd0 → J/ψKS as a function of MWR , for r sin α = 0.05. The
solid line corresponds to the sign choice si = 1 and the dashed line corresponds to si = 1 except
st = −1. The shaded part is allowed by the experimental data.

As we have commented above, this constraint comes predominantly from the Bd −B d mixing
contribution.

VII. CONCLUSION

In this paper, we have made a comprehensive study of CP violating observables in the low-
energy sector of the minimal LRSM with the only assumption of parity invariance imposed on
the theory. This is made possible by an explicit solution for the right-handed quark mixing
matrix with explicit dependence on the spontaneous CP violation phase α. Although the
hadronic physics uncertainty is still large, the CP observables do provide significant and
strong constraint on the right-handed W -boson mass scale WMR and the FCNH mass scale
MH . In fact, a new experiment result and/or improved theoretical calculation on EDM
might provide the strongest bound yet on the right-handed gauge boson mass.
We stress the point that CP must be violated both explicitly and spontaneously in the
minimal model, with one bidoublet and two triplets. Up to O(λ3 ), we can write VRCKM in
a compact form as in Eqs. (27), (28) and (29). We find that VR has the same hierarchical
structure as VL , i.e. elements of the two mixing matrices are suppressed by the same orders
of Cabibbo angle λ = 0.22. And because of the spontaneous CP phase α, each element in
VRCKM acquires a phase angle proportional to r sin α. Therefore, the phenomenology related
to CP violation in the minimal LRSM turns out to be very rich. We explored mass differences
of neutral kaon and B-meson with updated lattice results, and found an updated lower bound
on righthanded W -boson mass: MWR > 2.5 TeV. With the CP violating processes, we find a
combined bound MWR > 4 ∼ 8 TeV from ǫ and neutron EDM constraints when the FCNH
contribution is ignored. And we can also fix r sin α ≃ 0.05 from the combined bound. When
the FCNH contribution is added to ǫ, the bound is stronger when the MH becomes small. We
go on to study ǫ′ and CP asymmetry in Bd → J/ψKS decay. By applying the experimental
constraints and theoretical uncertainty, we conclude that the lower bound on righthanded
W -boson mass surviving from all above experimental constraint is about 4 TeV. If the WR
really have a mass close to this lower bound, it is possible to detect its signal in the up
coming LHC.

32
 MWR (TeV) MH10 , MA01 (TeV) |r sin α|
∆mK 2.5 15 −
∆mBd 2.5 12 −
∆mBs 2.7 25 −
4 (8) ∗ 100 0.05
den & ǫ
8 (20) ∗ 25 0.15
4.2 − ∗ 0.05
ǫ′
7.4 − ∗ 0.15

0.8 − ∗ 0.05
B → Jψ KS
1.3 − ∗ 0.15

TABLE II: A summary of bounds on MWR and MH from different physical observables. Stars on
the items indicate input values. For the neutron EDM case, the numbers in the parenthesis are
direct result from Eq. (72) and those without are obtained by reducing theoretical values by a
factor of 5.

We also find that the lower bound on MH10 > 25 TeV is tighter than the bound previous
bounds [19, 28]. Perhaps, this suggests that one should have two bidoublets so that one
can invoke cancelation between them, as in the spontaneous CP violation model discussed
in Ref. [47]. In that case for our analytic solution to remain valid, the second bi-doublet
should develop vev. All bounds are shown in Table II for easy reading of the results of the
paper.
Finally, we would like to comment on the constraint on r arising from the consideration
of the mass shift of SM Higgs boson in LRSM. According to discussions in Sec. IV, the SM
Higgs mass is  
2 α12
mh = 4λ1 − κ2 + α3 ξ 2 vR2 , (88)
ρ1
α12 2
to second order in α, ξ and ǫ. The shift in mass due to LRS is α3 ξ 2 vR2 − κ and can be
ρ1
expressed in terms of the masses of FCNH

∆mh0 ≃ ξMH10 . (89)

From the discussions below Eq. (59), FCNH masses have to be large enough to suppress the
tree level contribution to kaon mixing. On the other hand, the SM Higgs mass should not
exceed TeV scale in order to preserve perturbative unitarity. A recent analysis [25] yields

m2h0 < 870 GeV . (90)

The lower bound on the FCNH boson mass MH10 > 25 TeV yields an upper bound on
r = ξmt /mb < 1.44. From discussions on CP violating observables, 0.05 < |r sin α| < 0.15,
this translates into a lower bound on the spontaneous phase |α| > 0.035.
This work was partially supported by the U. S. Department of Energy via grant DE-
FG02-93ER-40762. Y. Z. acknowledges the hospitality and support from the TQHN group
at University of Maryland and a partial support from NSFC grants 10421503 and 10625521.

33
X. J. is supported partially by a ChangJiang Scholarship at Peking University. R. N. M. is
supported by NSF grant No. PHY-0652363.

[1] R. N. Mohapatra and J. C. Pati, Phys. Rev. D 11, 566 (1975); R. N. Mohapatra and J. C. Pati,
Phys. Rev. D 11, 2558 (1975); G. Senjanovic and R. N. Mohapatra, Phys. Rev. D 12, 1502
(1975); Phys. Rev. D 23, 165 (1981); For a review, Rabindra N. Mohapatra, CP Violation,
World Scientific Publ. Co., C. Jarlskog, Ed., 1989.
[2] T. D. Lee, talk given at the Center for High-Energy Physics, Peking University, Nov. 2006.
[3] R. N. Mohapatra, F. E. Paige and D. P. Sidhu, Phys. Rev. D 17, 2462 (1978); D. Chang,
Nucl. Phys. B 214, 435 (1983); G. C. Branco, J. M. Frere and J. M. Gerard, Nucl. Phys. B
221, 317 (1983); G. C. Branco and L. Lavoura, Phys. Lett. B 165, 327 (1985).
[4] P. Langacker and S. Uma Sankar, Phys. Rev. D 40, 1569 (1989); G. Barenboim, J. Bernabeu,
J. Prades and M. Raidal, Phys. Rev. D 55, 4213 (1997).
[5] K. Kiers, J. Kolb, J. Lee, A. Soni and G. H. Wu, Phys. Rev. D 66, 095002 (2002).
[6] G. Barenboim, M. Gorbahn, U. Nierste and M. Raidal, Phys. Rev. D 65, 095003 (2002).
[7] Y. Zhang, H. An, X. Ji and R. N. Mohapatra, Phys. Rev. D 76, 091301 (2007).
[8] R. E. Marshak and R. N. Mohapatra, Phys. Lett. B 91, 222 (1980).
[9] W. M. Yao et al. [Particle Data Group], J. Phys. G 33, 1 (2006).
[10] N. G. Deshpande, J. F. Gunion, B. Kayser and F. I. Olness, Phys. Rev. D 44, 837 (1991).
[11] K. Kiers, M. Assis and A. A. Petrov, Phys. Rev. D 71, 115015 (2005).
[12] A. Masiero, R. N. Mohapatra and R. D. Peccei, Nucl. Phys. B 192, 66 (1981).
[13] J. Basecq, J. Liu, J. Milutinovic and L. Wolfenstein, Nucl. Phys. B 272, 145 (1986).
[14] G. Ecker and W. Grimus, Phys. Lett. B 153, 279 (1985); G. Barenboim, J. Bernabeu and
M. Raidal, Nucl. Phys. B 478, 527 (1996); P. Ball, J. M. Frere and J. Matias, Nucl. Phys. B
572, 3 (2000).
[15] J. M. Frere, J. Galand, A. Le Yaouanc, L. Oliver, O. Pene and J. C. Raynal, Phys. Rev. D
46, 337 (1992).
[16] P. Duka, J. Gluza and M. Zralek, Annals Phys. 280, 336 (2000) [arXiv:hep-ph/9910279].
[17] G. Beall, M. Bander and A. Soni, Phys. Rev. Lett. 48, 848 (1982).
[18] M. Neubert, Z. Phys. C 50, 243 (1991).
[19] R. N. Mohapatra, G. Senjanovic and M. D. Tran, Phys. Rev. D 28, 546 (1983); G. Ecker,
W. Grimus and H. Neufeld, Phys. Lett. B 127, 365 (1983) [Erratum-ibid. B 132, 467 (1983)];
F. J. Gilman and M. H. Reno, Phys. Rev. D 29, 937 (1984); S. Sahoo, L. Maharana, A. Roul
and S. Acharya, Int. J. Mod. Phys. A 20, 2625 (2005).
[20] D. Chang, J. Basecq, L. F. Li and P. B. Pal, Phys. Rev. D 30, 1601 (1984); W. S. Hou and
A. Soni, Phys. Rev. D 32, 163 (1985); J. Basecq, L. F. Li and P. B. Pal, Phys. Rev. D 32,
175 (1985).
[21] R. Babich, N. Garron, C. Hoelbling, J. Howard, L. Lellouch and C. Rebbi, Phys. Rev. D 74,
073009 (2006); D. J. Antonio et al., arXiv:hep-lat/0702026.
[22] A. J. Buras, S. Jager and J. Urban, Nucl. Phys. B 605, 600 (2001).
[23] G. Ecker and W. Grimus, Nucl. Phys. B 258, 328 (1985).
[24] A. J. Buras, arXiv:hep-ph/9806471.
[25] A. Djouadi, arXiv:hep-ph/0503172.
[26] N. Tantalo, arXiv:hep-ph/0703241.

34
[27] D. Becirevic et al., Nucl. Phys. B 634, 105 (2002) [arXiv:hep-ph/0112303].
D. Becirevic, V. Gimenez, G. Martinelli, M. Papinutto and J. Reyes, JHEP 0204, 025 (2002)
[arXiv:hep-lat/0110091]; H. Wittig, Eur. Phys. J. C 33, S890 (2004) [arXiv:hep-ph/0310329].
[28] M. E. Pospelov, Phys. Rev. D 56, 259 (1997) [arXiv:hep-ph/9611422].
[29] G. Beall and A. Soni, Phys. Rev. Lett. 47, 552 (1981); G. Ecker, W. Grimus and H. Neufeld,
Nucl. Phys. B 229, 421 (1983); J. M. Frere, J. Galand, A. Le Yaouanc, L. Oliver, O. Pene
and J. C. Raynal, Phys. Rev. D 45, 259 (1992).
[30] X. G. He, B. H. J. McKellar and S. Pakvasa, Phys. Rev. Lett. 61, 1267 (1988).
[31] X. G. He and B. McKellar, Phys. Rev. D 47, 4055 (1993).
[32] J. Donoghue, E. Golowich and B. Holstein, Dynamics of the Standard Model (Cambridge
University Press, Cambridge, 1994).
[33] S. M. Barr and A. Zee, Phys. Rev. Lett. 65, 21 (1990) [Erratum-ibid. 65, 2920 (1990)].
[34] S. Weinberg, Phys. Rev. Lett. 63, 2333 (1989).
[35] D. Chang, X. G. He, W. Y. Keung, B. H. J. McKellar and D. Wyler, Phys. Rev. D 46, 3876
(1992) [arXiv:hep-ph/9209284].
[36] M. A. Shifman, A. I. Vainshtein and V. I. Zakharov, Sov. Phys. JETP 45, 670 (1977) [Zh.
Eksp. Teor. Fiz. 72, 1275 (1977)].
[37] G. Buchalla, A. J. Buras and M. K. Harlander, Nucl. Phys. B 337, 313 (1990).
[38] G. Buchalla, A. J. Buras and M. E. Lautenbacher, Rev. Mod. Phys. 68, 1125 (1996)
[arXiv:hep-ph/9512380].
[39] S. Bosch, A. J. Buras, M. Gorbahn, S. Jager, M. Jamin, M. E. Lautenbacher and L. Silvestrini,
Nucl. Phys. B 565, 3 (2000) [arXiv:hep-ph/9904408].
[40] T. Blum et al. [RBC Collaboration], Phys. Rev. D 68, 114506 (2003) [arXiv:hep-lat/0110075].
[41] D. Pekurovsky and G. Kilcup, Phys. Rev. D 64, 074502 (2001) [arXiv:hep-lat/9812019].
[42] H. Burkhardt et al. [NA31 Collaboration], Phys. Lett. B 206, 169 (1988); V. Fanti et al.
[NA48 Collaboration], Phys. Lett. B 465, 335 (1999) [arXiv:hep-ex/9909022]; A. Alavi-Harati
et al. [KTeV Collaboration], Phys. Rev. Lett. 83, 22 (1999) [arXiv:hep-ex/9905060].
[43] S. Bertolini, J. O. Eeg and M. Fabbrichesi, Phys. Rev. D 63, 056009 (2001)
[arXiv:hep-ph/0002234].
[44] A. J. Buras and M. Jamin, JHEP 0401, 048 (2004) [arXiv:hep-ph/0306217].
[45] J. Chay and C. Kim, [arXiv:hep-ph/0009244]; H. Cheng and K. Yang, Phys. Rev. D 63,
074011 (2001) [arXiv:hep-ph/0011179].
[46] C. Chen and H. Li, Phys. Rev. D 71, 114008 (2005) [arXiv:hep-ph/0504020].
[47] Y. L. Wu and Y. F. Zhou, arXiv:0709.0042 [hep-ph].

35