Progress in Particle and Nuclear Physics 67 (2012) 337342

Contents lists available at SciVerse ScienceDirect

Progress in Particle and Nuclear Physics
journal homepage: www.elsevier.com/locate/ppnp
Review
Scalar susceptibilities and electromagnetic thermal mass differences in
chiral perturbation theory
R. Torres Andrs

, A. Gmez Nicola
Departamento de Fsica Terica II, Universidad Complutense de Madrid, Spain
a r t i c l e i n f o
Keywords:
Chiral symmetry
Chiral perturbation theory
Isospin breaking
Finite-temperature field theory
a b s t r a c t
We make a thermal analysis of the light scalar susceptibilities using SU(3)-chiral pertur-
bation theory to one-loop order, taking into account the QCD source of isospin breaking
(IB), i.e. corrections coming fromm
u
= m
d
. The value of the connected scalar susceptibility
in the infrared regime, the one relevant when approaching chiral symmetry restoration,
and below the critical temperature is found to be entirely dominated by the
0
mixing,
which leads to model-independent O(
0
) corrections, where m
d
m
u
, in the combi-
nation
uu

ud
of flavour breaking susceptibilities. We also present preliminary results
for the corrections to the real part of the pion self-energy at next-to-leading order in SU(2)-
chiral perturbation theory, taking into account electromagnetic interaction. The results for
zero and finite temperature for the charged and neutral pions are given in terms of the
3-momentumof the external pion, and their difference is calculated to this order, stressing
the fact that, at low and moderate temperature, the mass splitting M

0 grows with
temperature for, at least, non-zero charged pion mass running inside the loops.
2012 Elsevier B.V. All rights reserved.
1. Introduction
The lowenergy sector of QCD has been successfully described over recent years within the chiral Lagrangian framework.
Chiral perturbation theory (ChPT) is based on the spontaneous breaking of the chiral symmetry SU
V
(N
f
) SU
A
(N
f
)
SU
V
(N
f
) with N
f
= 2, 3 light flavours and provides a consistent, systematic and model-independent scheme for calculating
lowenergy observables [13]. The effective ChPT Lagrangian is constructed as an expansion of the formL = L
p
2 +L
p
4 +
where p denotes a meson energy scale compared to the chiral scale

1 GeV. The formalism can also be extended to

finite temperature T, in order to describe meson gases and their evolution towards chiral symmetry restoration for T below
the critical temperature T
c
[4,5], where T
c
180200 MeV from lattice simulations [68]. The use of ChPT in this context
is important for providing model-independent results for the evolution of the different observables with T, supporting the
original predictions for chiral restoration [9], also confirmed by lattice simulations, which are consistent with a crossover-
like transition for N
f
= 3 (2 + 1 flavours in the physical case), a second-order one for N
f
= 2 in the O(4) universality class
and a first-order one in the degenerate case of three equal flavours.
The invariance under the SU
V
(2) vector group is the isospin symmetry, which is a very good approximation to Nature.
However, there are several processes where isospin breaking corrections are phenomenologically relevant, for example
those of sumrules for quark condensates [3], meson masses [10] or pion scattering [11,12]. There are two possible sources of
isospin breaking: the QCDm
d
m
u
light quark mass difference and electromagnetic interactions. Both can be accommodated
within the ChPT framework. From the first source we expect corrections of order (m
d
m
u
)/m
s
, encoded in the quark
mass matrix, which generates also a
0
mixing term in the SU(3) Lagrangian [3]. On the other hand, the electromagnetic

Corresponding author.
E-mail address: [email protected] (R.T. Andrs).
0146-6410/$ see front matter 2012 Elsevier B.V. All rights reserved.
doi:10.1016/j.ppnp.2011.12.040
338 R.T. Andrs, A.G. Nicola / Progress in Particle and Nuclear Physics 67 (2012) 337342
interactions are included in the ChPT effective Lagrangian via the external source method and give rise to newterms [1013]
of order L
e
2 , L
e
2
p
2 and so on, e being the electric charge. It is possible to accommodate these terms into the ChPT power
counting scheme by considering formally e
2
= O(p
2
/F
2
), with F the pion decay constant in the chiral limit.
The purpose of this paper is to calculate the leading thermal contributions to the connected and disconnected scalar
susceptibilities, taking into account isospin breaking, and to showour preliminary results concerning the thermal evolution
of the masses of the charged and neutral pions.
2. Light scalar susceptibilities and the role of the
0
mixing
Calculations for light quark condensates, uu and

dd, at zero and finite temperature have been made to one-loop in

[14,15], respectively, within the framework of SU(3)-ChPT taking into account both sources of IB. The main feature that we
want to stress is that there is a
0
mixing term appearing through the tree level mixing angle defined by tan 2 =

3
2
m
d
m
u
m
s
m
.
Different light quark masses allow us to consider three independent light scalar susceptibilities defined as

ij
=

m
i
q
j
q
j
=

2
m
i
m
j
log Z(m
u
= m
d
). (1)
From now on in this section, we will neglect the electromagnetic corrections because they are small and not relevant
for our present discussion, so we will put e = 0. Then, to leading order in the mixing angle, the contribution of the

0
mixing in the quark condensate sum is of order
2
, whereas for the difference it goes like . The thermal functions
g
i
(T, M
i
), i =
0
, , defined as g
i
(T) =
1
4
2
F
2
_

0
dp
p
2
E
p
1
e
Ep
1
, with E
2
p
= p
2
+ M
2
i
and = T
1
, are suppressed by those
coefficients and the quark condensates do not receive important corrections. The important point is that differentiating with
respect to a light quark mass is essentially the same as differentiating with respect to
m
d
m
u
m
s
, so the suppression of the
thermal functions is smaller in the case of the susceptibilities than in the quark condensate case.
Because of the linearity in in uu dd for a small mixing angle, the combinations
uu

ud
and
dd

du
receive an
O(1) IB correction due to
0
mixing, which would not be found if m
u
= m
d
was taken from the beginning. The analysis
of the -dependence of uu dd shows that, up to O(),
uu

dd
, so combinations like
uu

dd
, which also vanish with
m
u
= m
d
, are less sensitive to IB.
One can also relate these flavour breaking susceptibilities to the connected and disconnected ones [16], often used in
lattice analysis [17,18]:
dis
=
ud
, and
con
=
1
2
(
uu
+
dd
2
ud
). From the previous analysis, we get
con

uu

ud
.
Therefore, our model-independent analysis including IB effects provides the leading non-zero contribution for the
connected susceptibility which arises partially from
0
mixing. This is particularly interesting for the lattice, where
artefacts such as taste breaking mask the behaviour of
con
with the quark mass and T when approaching the continuum
limit [18]. In fact, our ChPT approach is useful for exploring the chiral limit (m
u,d
0) or infrared (IR) regime, which gives
a qualitative picture of the behaviour near chiral symmetry restoration. In this regime M

T M
K
, and therefore we
can neglect thermal heavy particles, which are exponentially suppressed.
The leading order results for the connected and disconnected susceptibilities at zero temperature are the following:

IR
con
(T = 0) = 8B
2
0
[2L
r
8
() + H
r
2
()]
B
2
0
16
2
_
1 + log
M
2
K

2
_

B
2
0
24
2
log
M
2

2
+O(
2
), (2)

IR
dis
(T = 0) = 32B
2
0
L
r
6
()
3B
2
0
32
2
_
1 + log
M
2

2
_
+
B
2
0
288
2
_
5 log
M
2

2
1
_
+O(
2
), (3)
where B
0
is the parameter which relates masses and quark condensates at tree level via the Gell-MannOakesRenner
formula, and L
6
, L
8
, H
2
are low energy constants.
The log term in Eq. (3) is the dominant one at T = 0 and can be found in [16], but the connected IR susceptibility (2) is
not zero at T = 0, because it receives contributions of order O(1) in the mixing angle.
If we consider the pion gas in a thermal bath, then expressions (2)(3) are modified according to
[
con
(T)
con
(0)]
IR
=
B
2
0
18
T
2
M
2

+O
_

2
B
4
0
T
4
M
4

_
+O
_
exp
_

M
,K
T
__
, (4)
[
dis
(T)
dis
(0)]
IR
=
3B
2
0
16
T
M

+O
_

2
B
4
0
T
4
M
4

_
+O
_
exp
_

M
,K
T
__
. (5)
Note that, as we have already mentioned, the eta mass term in Eq. (4) and in the subleading corrections in the mixing
angle comes from the -analysis and the IR expansion of the g
1
(M

), and does not have anything to do with thermal etas.

R.T. Andrs, A.G. Nicola / Progress in Particle and Nuclear Physics 67 (2012) 337342 339
Fig. 1. Connected IR susceptibility normalized to B
2
0
, for fixed tree level eta mass and m
s
= 80 MeV.
Fig. 2. Disconnected IR susceptibility normalized to B
2
0
, for several light quark mass ratios and fixed tree level eta mass (m/m
s
= 0.05 is the physical case).
Figs. 1 and 2 show, respectively, the connected susceptibility (4) for fixed tree level eta mass (proportional to

B
0
m
s
in
the IR regime), and the disconnected one (5) for several values of the light quark mass ratio m/m
s
, and also with fixed tree
level eta mass. The leading scaling with T and the light quark mass in this regime for the disconnected piece goes like
T

m
,
i.e. with the same scaling as was calculated in [16,17], whereas the connected susceptibility grows quadratically in T over
a mass scale much greater than the SU(2) Goldstone bosons one. Therefore, in the continuum limit, we only expect
dis
to
peak near the transition, as the m 0
+
limit in Fig. 2 clearly shows.
3. Charged and neutral thermal pion masses in SU(2)-ChPT at O(p
4
)
If we consider virtual photons in the calculation of the real part of the self-energy in the mass shell, there are four relevant
diagrams that correct the masses at order O(p
4
): there are pion tadpoles, diagram(a) in Fig. 3, and the tree level NLOdiagram
needed for renormalization, (b), where both charged and neutral pions participate; and diagrams with virtual photons, (c)
and (d), which only modify the charged pion mass.
The photontadpole diagram(c) is proportional to the photonmass andtherefore vanishes at zero temperature, while pion
tadpoles (a) and the photon exchange diagram (d) are finite and chiral scale independent once regularized and combined
with diagram (b). The LO corrections to the masses of the SU(2) NGB are calculated as M
2
=

M
2
+ (

M
2
), where

M is the
respective tree level mass. At zero temperature the neutral and charged pion masses are [11]
M
2

0
=

M
2

0
_
1 + 2

0
+ e
2
K

0 + 2l
r
3

M
2

0
F
2
_
2
B
2
F
2
l
r
7
(m
d
m
u
)
2

4
3
Be
2
k
7
(m
d
m
u
), (6)
where
,0
=
M
2

,
0
32
2
F
2
log
M
2

,
0

, and K

0 =
20
9
_
k
r
1
+ k
r
2

9
10
(2k
r
3
k
r
4
) k
r
5
k
r
6

1
5
k
r
7
_
, with k
i
being electromag-
netic low energy constants, Z the parameter that corrects the leading order charged pion mass, and
M
2

=

M
2

0
_
1 +
e
2
4
+
0
+ e
2
K
A

+ 2l
r
3

M
2

0
F
2
_
+2e
2
F
2
_
Z
_
1 +
e
2
4
_
+ e
2
K
B

(3 + 4Z)

_

4
3
Be
2
k
7
(m
d
m
u
), (7)
340 R.T. Andrs, A.G. Nicola / Progress in Particle and Nuclear Physics 67 (2012) 337342
Fig. 3. Diagrams for the pionself-energy at order O(p
4
). (a) and(b) represent piontadpole contributions andthe tree level NLOdiagramwhichrenormalizes
the loops, respectively. (c) (photon tadpole) and (d) (one-photon exchange) only correct the charged pion self-energy.
with the definitions
K
A

=
20
9
_
k
r
1
+ k
r
2
k
5

1
5
(23k
r
6
+ 18k
r
8
+ k
r
7
)
_
,
and
K
B

=
10
9
_
2Z(k
r
1
+ k
r
2
)
1
2
k
13
k
14
_
.
There is a factor 1/2 in the coefficient of k
7
for both masses which does not appear in [11], and that was also noted by [19].
At this point the total mass difference between charged and neutral pions becomes
M
2

M
2

0
= 2

M
2

0
(
0

) + 2
B
2
F
2
l
r
7
(m
d
m
u
)
2
+ 2e
2
F
2
_
Z
_
1 +
e
2
4
_
+ e
2
K
B

_
+

M
2

0
e
2
_
1
4
+K
A

0
_
2e
2
F
2
(3 + 4Z)

, (8)
which has pure strong, pure EM and mixed EMstrong contributions.
If the pions are immersed in a thermal bath, there appear new contributions with no UV divergences, since these only
appear in the zero-temperature part and they have already been renormalized. The pion tadpoles, charged or not, give rise
to g
1
(M, T) thermal functions. For the neutral pion mass we get
M
2

0
=

M
2

0
(T = 0)
_
1 +
1
F
2
_
g
1
(M
2

, T)
1
2
g
1
(M
2

0
, T)
__
, (9)
where it is worth noting that the neutral pion mass decreases with T, contrary to what happens if we do not consider
electromagnetic effects, as can be seen if we put M
2

= M
2

0
in the last expression.
As for the charged pion, we can separate the contributions coming from pion tadpoles, M
2
tadpoles
, and the two different
contributions fromthe virtual photon diagrams: one coming fromthe photon tadpole which is not zero at finite temperature
and gives a typical thermal screening contribution, M
2
Ph. tadpole
; and the other due to the one-photon exchange, M
2
Ph. Exchange
.
For the first two ones we get M
2
Tadpoles
=

M
2

0
2F
2
4Ze
2
g
1
(M
2

, T) and M
2
Ph. Tadpole
=
1
3
e
2
T
2
. The latter has the typical form
of a Debye or screening mass of the electric field in a thermal bath [20], which always grows with T.
The one-photon exchange diagram is more complex and its contributions to the real part depend, in general, on the
3-momentumof the external pion, which is a direct consequence of the Lorentz symmetry breaking in the thermal bath. The
Matsubara sums can be performed in the standard way, before performing the analytic continuation to external continuous
frequencies:
Re
_

Ph. Exchange
_
= e
2
__
d
3
k
(2)
3
n()
2
(2q k)
2
|
k
0
=
(q
0
)
2

2
+
_
d
3
k
(2)
3
n()
2
(2q k)
2
|
k
0
=
(q
0
+ )
2

2
+
_
d
3
k
(2)
3
n(

)
2

(2q k)
2
|
k
0
=q
0

(q
0

)
2

2
+
_
d
3
k
(2)
3
n(

)
2

(2q k)
2
|
k
0
=q
0
+

(q
0
+

)
2

2
_
,
with
2
=

k
2
the photon energy squared inside the loop and
2
= ( q

k)
2
+M
2

the pion energy squared, also inside the

loop; here k is the virtual photon 4-momentum. The above 3-momentum integrals can be written as
Re
_

Ph. Exchange
_
=
e
2
2
2
_

0
dk
_

0
d n(k) sin

M
2

_
E
2
M
2

cos k(E
2
sin
2
+ M
2

cos
2
)
E
2
sin
2
+ M
2

cos
2

+
e
2
4
2
_

0
dk
_

0
d
n(

sin
k
2
(M
2

cos
2
+ E
2
sin
2
) + 2M
2

(E
2
k
_
E
2
M
2

cos )
E
2
sin
2
+ M
2

cos
2

. (10)
R.T. Andrs, A.G. Nicola / Progress in Particle and Nuclear Physics 67 (2012) 337342 341
Fig. 4. Preliminary results for the charged (dashed line) and neutral (solid line) masses and their difference (dotdashed line) at LO in the static limit.
Fig. 5. Different results for the chargedneutral pion mass difference: (a) (solid line) corresponds to our preliminary results in the chiral limit keeping
corrections e = 0 for the tree level charged pion mass inside the loops; (b) (dotdashed line) corresponds to the same preliminary calculation with
m
u
= m
d
= 0 and e = 0 also inside the loops; and the full dashed line is the result given in [21].
It is clear nowthat the charged pion real part of the self-energy depends on the energy of the external pion E
2
= | q|
2
+M
2

,
and, therefore, on the external 3-momentum.
The final result will take the form M
2

=

M
2

(T = 0) + M
2
Tadpoles
+ M
2
Ph. Tadpole
+ Re
_

Ph. Exchange
_
. In Fig. 4 we have
plotted our preliminary results for the neutral and charged pion real masses as a function of the temperature of the thermal
bath, taking physical values for the pion masses and calculating M
2
Ph. Exchange
in the static limit, i.e., for E = M

. Once we
have the thermal and isospin breaking corrections to the masses separately for any value of the external momentum, we can
calculate it in the limit where temperatures are (i) much greater than the masses and the external momenta (which means
that we have to set the masses inside the loops to zero), and (ii) sizable withrespect to the momenta running inside the loops,
T k m, q. With these assumptions, we are led to the HTL result given in [21], M
2

M
2

0
=

M
2

_
1
T
2
6
_
+
1
4
e
2
T
2
,
which serves as a consistency check.
Moreover, our lowtemperature analysis allows us to assume a slightly different chiral limit, in the sense that we can still
assume m
u
= m
d
= 0 and consider e = 0 inside the loops. In Fig. 5 we show the plot of our preliminary calculation both in
this latter limit, and also considering m
u
= m
d
= 0, e = 0, to be compared with those appearing in the HTL result [21] where
the contributions from the screening-like term, always increasing with T and inherent to the thermal bath, are responsible
for the final growth of the mass difference, contrary to what one would expect naively from the sum rule relating the axial
and vector spectral functions [20] applying chiral symmetry restoration arguments.
References
[1] S. Weinberg, Physica A 96 (1979) 327.
[2] J. Gasser, H. Leutwyler, Ann. Phys. 158 (1984) 142.
[3] J. Gasser, H. Leutwyler, Nuclear Phys. B 250 (1985) 465.
[4] J. Gasser, H. Leutwyler, Phys. Lett. B 184 (1987) 83.
[5] P. Gerber, H. Leutwyler, Nuclear Phys. B 321 (1989) 387.
[6] [MILC Collaboration], C. Bernard, et al., Phys. Rev. D 71 (2005) 034504.
[7] Y. Aoki, S. Borsanyi, S. Durr, Z. Fodor, S.D. Katz, S. Krieg, K.K. Szabo, J. High Energy Phys. 0906 (2009) 088.
[8] M. Cheng, et al., Phys. Rev. D 81 (2010) 054504.
342 R.T. Andrs, A.G. Nicola / Progress in Particle and Nuclear Physics 67 (2012) 337342
[9] R.D. Pisarski, F. Wilczek, Phys. Rev. D 29 (1984) 338.
[10] R. Urech, Nuclear Phys. B 433 (1995) 234.
[11] M. Knecht, R. Urech, Nuclear Phys. B 519 (1998) 329.
[12] U.G. Meissner, G. Muller, S. Steininger, Phys. Lett. B 406 (1997) 154. (erratum);
U.G. Meissner, G. Muller, S. Steininger, Phys. Lett. B 407 (1997) 454.
[13] G. Ecker, J. Gasser, A. Pich, E. de Rafael, Nuclear Phys. B 321 (1989) 311.
[14] A.G. Nicola, R.T. Andres, J. Phys. G: Nucl. Part. Phys. 39 (2012) 015004.
[15] A.G. Nicola, R.T. Andres, Phys. Rev. D 83 (2011) 076005.
[16] A.V. Smilga, J.J.M. Verbaarschot, Phys. Rev. D 54 (1996) 1087.
[17] S. Ejiri, et al., Phys. Rev. D 80 (2009) 094505.
[18] C.E. DeTar, PoS (2008) 001. LATTICE2008.
[19] J. Schweizer, J. High Energy Phys. 0302 (2003) 007.
[20] Kraemmer, Rebhan, Ann. Phys. 238 (1995) 286332.
[21] C. Manuel, N. Rius, Phys. Rev. D 59 (1999) 054002. arXiv:hep-ph/9806385.