Intriguing aspects of meson condensation
Abstract
We analyze various aspects of pion and kaon condensation in the framework of chiral perturbation theory. Considering a system at vanishing temperature and varying the isospin chemical potential and the strange quark chemical potential we reproduce known results about the phase transition to the pion condensation phase and to the kaon condensation phase. However, we obtain mesonic mixings and masses in the condensed phases that are in disagreement with the results reported in previous works. Our findings are obtained both by a theory group analysis and by direct calculation by means of the same lowenergy effective Lagrangian used in previous works. We also study the leptonic decay channels in the normal phase and in the pion condensed phase, finding that some of these channels have a peculiar nonmonotonic behavior as a function of the isospin chemical potential. Regarding the semileptonic decays, we find that that they are feeding processes for the stable charged pion state.
I Introduction
The properties of strongly interacting matter in an isospin and/or strangeness rich medium are relevant in a wide range of phenomena including the astrophysics of compact stars and heavyion collisions. It is known that depending on the value of the isospin chemical potential, , and on the value of the strangeness chemical potential, , three different phase can be realized: the normal phase, the pion condensed () phase and the kaon condensed () phase Migdal:1990vm (); Son:2000xc (); Kogut:2001id (). The realization of a mesonic condensate can drastically change the low energy properties of matter, including the mass spectrum and the lifetime of mesons.
Previous analyses of the meson condensed phases by QCDlike theories were developed in Kogut:1999iv (); Kogut:2000ek (). Pion condensation in twoflavor quark matter was studied in Son:2000xc (); Son:2000by () and in threeflavor quark matter in Kogut:2001id (). In particular, the phase diagram as a function of and was presented in Kogut:2001id (). Finite temperature effects in chiral perturbation theory (PT) have been studied in Loewe:2002tw (); Loewe:2004mu (); He:2005nk (); Xia:2014bla (). One remarkable property of quark matter with nonvanishing isospin chemical potential is that it is characterized by a real measure, thus the lattice realization can be performed with standard numerical algorithms Alford:1998sd (); Kogut:2002zg (). The phase and the phase have been studied by NJL models in Toublan:2003tt (); Barducci:2004tt (); Barducci:2004nc (); Ebert:2005wr (); Ebert:2005cs () and by random matrix models in Klein:2004hv (). All these models find results in qualitative and quantitative agreement, and in particular, the phase diagram of matter has been firmly established. However, regarding the low energy mass spectrum in threeflavor quark matter, we found that it was only studied in Kogut:2001id (). Our results are in disagreement with those of Kogut:2001id (), the most relevant difference is in the mixing between mesonic states. Regarding the pion decay, previous works focused on density and temperature effects in standard decay channels Barducci:1990sv (); Dominguez:1993kr (); Loewe:2011tm (), but not all the decay channels have been considered.
In the present paper we analyze the c phase and the c phase in a realization of PT Gasser:1983yg (); Leutwyler:1993iq (); Ecker:1994gg (); Scherer:2002tk (); Scherer:2005ri () that includes only the pseudoscalar mesons. Therefore, the considered chiral Lagrangian approach is valid for MeV, MeV, and MeV. These bounds come from the masses of the proton, the rho meson and the omega baryon, respectively. Moreover, PT is valid in the energy range GeV, corresponding to the breaking scale of the theory. For definiteness we take the following values of the mesonic masses in vacuum: MeV, MeV and MeV. Unless explicitly stated, we will assume that in vacuum all the pion masses and all the kaon masses are equal. By this model we discuss the mixing and the masses of the pseudoscalar mesonic octet and the most relevant pion decay channels in the normal phase, in the c phase and in the c phase. Regarding the mesonic mixing, we discuss the disagreement with the results of Kogut:2001id () by theory group analysis and by explicit calculation using the PT Lagrangian. Regarding the decay channels, since the masses of the mesons strongly depend on and , by changing these chemical potentials some decay channels can become kinematically forbidden and/or other channels that are not allowed in vacuum can be opened.
As we shall formally see, the presence of a baryonic chemical potential is immaterial for the chiral Lagrangian, because mesons have no baryonic charge. However, it is clear that at large values of we expect a transition between hadronic matter and a different phase, presumably a color superconducting phase Rajagopal:2000wf (); Alford:2007xm (); Anglani:2013gfu (). In principle we should limit ourselves to considering , however since the effective Lagrangian is blind to the baryonic chemical potential, we can assume that such inequality is always satisfied. Although we will consider the range of values of MeV, it is worth emphasizing that at asymptotic the system can be studied by perturbative QCD and the ground state is a Fermi liquid with Cooper pairing of quarks Son:2000xc (); Son:2000by ().
One interesting topic that to the best of our knowledge has not been previously discussed in the pion and kaon condensed phases is the screening mass of the photon. By the NishijimaNakanoGell–Mann (NNG) formula
(1) 
it is possible to relate the electric charge, , to the third component of isospin, , and hypercharge, . In particular, if the vacuum carries isospin and/or strangeness charges, then it will be a superconductor because the gauge group will be broken. Thus, by the HiggsAnderson mechanism the photon will acquire a Meissner mass. We evaluate the treelevel screening masses finding that in the two meson condesed phases they have the same formal expression. Moreover, the Debye and Meissner masses are equal. In principle, any quark chemical potential breaks the Lorentz symmetry, therefore the Debye and Meissner masses of the photon can be different. However, we will show that the treelevel Lagrangian has to lead to equal Debye and Meissner masses.
The present paper is organized as follows. In Sec. II we briefly review the aspects of PT that are relevant for our work. In Sec. III we consider twoflavor quark matter. We discuss pion condensation driven by an isospin chemical potential reviewing known results and generalizing the study of the lowenergy Lagrangian. In Sec. IV we consider threeflavor quark matter, determining the mixing angles and the masses of the pseudoscalar octet. In Sec. V we discuss the pion decay channels in the normal phase and in the phase. In Sec. VI we summarize our results. In the Appendix A we discuss some details about the  vertex factor relevant for pion decays.
Ii General setting
In this section we briefly review the aspects of chiral symmetry that are relevant for meson condensation. The general Lorentz invariant Lagrangian density describing the pseudoscalar mesons can be written as
(2) 
where corresponds to the meson fields, describes scalar and pseudoscalar external fields and the covariant derivative is defined as
(3) 
with and the external vectorial and axial currents, respectively. The Lagrangian has two free parameters and , related to the pion decay and to the quarkantiquark condensate, respectively, see for example Gasser:1983yg (); Leutwyler:1993iq (); Ecker:1994gg (); Scherer:2002tk (); Scherer:2005ri ().
The Lagrangian density is invariant under provided the meson field transform as
(4) 
and the chiral symmetry breaking corresponds to the spontaneous global symmetry breaking . The combination of the NambuGoldstone bosons (NGBs), with , corresponding to mass eigenstates can be identified with the pseudoscalar mesons fields. In standard PT, the mass eigenstates are charge eigenstates as well. Thus mesons are particles with a well defined mass and charge. The presence of a medium can change this picture. In particular, if the vacuum carries an electric charge, then the mass eigenstates will not typically be charge eigenstates. The presence of a medium can be taken into account by considering appropriate external currents in Eq. (2).
At vanishing temperature the vacuum is determined by maximizing the Lagrangian density with respect to the external currents. The pseudoscalar mesons are then described as oscillations around the vacuum. We use the same nonlinear representation of Kogut:2001id () corresponding to
(5) 
where are the generators and is a generic matrix to be determined by maximizing the static Lagrangian. The reasoning behind the above expression is that under mesons can be identified as the fluctuations of the vacuum as in Eq. (4) with .
In the following we will assume that , , , where is the diagonal quark mass matrix and is a constant, that with these conventions is equal to . Moreover, we will assume that , meaning that the vectorial current consists of the electromagnetic field and a quark chemical potential, with a matrix in flavor space. We first study the case and then the case. Since the twoflavor case is simpler to treat mathematically, it will allow us to establish a number of results that are useful for the description of the threeflavor case.
Iii Twoflavor case
In twoflavor quark matter the vacuum expectation value of the fields can be expressed as
(6) 
where corresponds to the energetically favored direction in space. Assuming equal light quark masses, , the Lagrangian can be written as
(7) 
where is the pion mass for vanishing isospin chemical potential. Expanding the covariant derivative we obtain
(8) 
and considering the quark chemical potential
(9) 
we can write
(10) 
with
(11)  
(12) 
Given that in Eq. (III) the interaction terms between and are proportional to commutators of these two fields, the only relevant term in is the one proportional to , and this is consistent with the fact that mesons have no baryonic charge. Note that both and explicitly break chiral symmetry giving mass to the (pseudo) NGBs. Equal light quark masses leave invariant ensuring that pions have equal masses. The isospin chemical potential induces a further symmetry breaking, such that with the effect of removing the pion mass degeneracy with a contribution proportional to the isospin charge. Since pions are an isotriplet, it follows that the contribution of the isospin chemical potential to the mass vanishes and the contributions to the is a Zeemanlike splitting, thus
(13)  
(14) 
clearly the condensation of charged pions happens at . The only symmetry of the Lagrangian in Eq. (7) is ; when it is spontaneously broken it leads to a massless NGB, corresponding to one of the two charged pions depending on the sign of the isospin chemical potential.
At the microscopic level, the breaking pattern induced by the isospin chemical potential and the light quark masses is
(15) 
where corresponds to the electromagnetic gauge symmetry. In the broken phase one of the two charged pions condense, spontaneously breaking the symmetry, meaning that the system becomes an electromagnetic superconductor. Formally, can be expressed as a combination of the generator of and of , thus the breaking of leads to a screening mass for the photon by the HiggsAnderson mechanism.
Regarding the Lorentz symmetry, the isospin chemical potential explicitly breaks boost symmetry, however by expressing the isospin chemical potential as the expectation value of the field we can formally consider a Lorentz invariant Lagrangian. To formally preserve Lorentz symmetry we will as well employ the Lorenz gauge .
iii.1 Ground state
For vanishing mesonic fluctuations the Lagrangian is a functional of and ; upon substituting Eq. (6) in Eq. (7) we obtain
(16) 
which is a function of the parameters and and a functional of . For vanishing external electromagnetic field and for , the global maximum is at and is independent of , meaning that the ground state has an global symmetry. In this case the custodial is still present and only the curvature of the potential (the pion masses) are affected by the isospin chemical potential. In other words, the isospin chemical potential is not sufficient to tilt the vacuum in one direction, thus the vacuum is the same obtained with .
The stationary point of corresponds to and , which is a global maximum for . In this case the vacuum is tilted by an angle and the ground state has only a residual symmetry (isomorphic to ) for rotations and ; the angle cannot be determined maximizing the ground state Lagrangian and is signaling the existence of a massless NGB.
The ground state Lagrangian can be easily determined and is given by
(17) 
Regarding the screening masses of the electromagnetic field, they can be inferred from Eq. (16). The electromagnetic field has both a Debye mass and a Meissner mass, which are equal and given by
(18) 
The screening masses vanish in the unbroken phase and are equal to in the broken phase, signaling the breaking of . In principle, the Debye and Meissner masses could be different, because the Lorentz symmetries is explicitly broken by . However, from the fact that the isospin chemical potential can be introduced as in Eq. (12) it is clear that both treelevel screening masses must be equal.
Generic chemical potential
To properly understand the previous results regarding the ground state configuration we consider a more general setting with
(19) 
corresponding to a quark chemical potential pointing to an arbitrary direction in isospin space. The ground state Lagrangian is obtained maximizing
(20) 
as a function of and . It is clear that , thus is in the plane perpendicular to . This leads to the residual symmetry for rotations around . For the ground state is tilted by an angle . The ground state Lagrangian is the same reported in Eq. (17), but with .
iii.2 Quadratic Lagrangian
The leading order Lagrangian describing the in medium pions can be obtained expanding Eq. (III) at the second order in the fields. For definiteness we consider and in the vev in Eq. (6). We decompose the Lagrangian at the second order in the fields as follows
(21) 
where is the kinetic term, is the mass term and is the term linear in the derivatives.
The kinetic part of the Lagrangian can be written as
(22) 
that manifestly shows meson mixing. Since is a symmetric matrix, it can be diagonalized. By the transformation
(23)  
(24)  
(25) 
we obtain the canonical kinetic term
(26) 
One of the peculiar aspects of the field redefinition above is that in the phase for the terms proportional to diverge. In other words, for vanishing light quark masses the above field renormalization does not seem to work. The correct prescription for handling this issue seems to be to consider the limit only in the physical results.
Regarding the electric charge eigenstates, we find that
(27) 
where . Note that the standard definition of the charge eigenstates is obtained for , as expected.
For the mass term we find
(28) 
that in the rotated basis turns out to be
(29)  
The term with a linear dependence on the derivative is given by
(30) 
that by the rotated basis redefinition turns in
(31) 
An interesting aspect is that in the phase this is the only mixing term between the fields. Since it scales as , it vanishes for .
Note that no term of the quadratic Lagrangian depends on , thus the symmetry has been absorbed in the redefinition of the fields. Assuming that no electromagnetic field is present, we can replace in all the Lagrangian terms , obtaining in momentum space
(32) 
where and is the inverse propagator. The energy spectrum is obtained from the poles of the propagator and in the phase we find
(33)  
(34)  
(35) 
where are the two mass eigenstates (note that in this case the subscript does not indicate the electric charge). The dispersion law of the massless mode is linear in momentum with a velocity that tends to the speed of light for and that vanishes for . The charge eigenstates can be expressed as a linear combination of the fields as follows:
(36) 
where
(37) 
Given the particular expression of the coefficients, the propagating particles oscillate between the two electric charge eigenstates with a mixing angle depending on the energy. This is a rather peculiar behavior because in the Standard Model one typically has mixing angles that are not energy/momentum dependent. Note that this oscillation also means that the propagating particles oscillate between isospin eigenstates. This is possible because in the condensed phase the vacuum carries isospin charge which is related to the electric charge by the NNG formula (1), thus nor the electric charge nor the isospin charge are conserved.
This result is important for the determination of the width of the pion decays discussed in Sec. V.
The mixing between the charged pion states can be simply understood in twoflavor quark matter. The states are the only states having a non vanishing value of the third component of isospin, and since the vacuum has a nonvanishing , these states can mix. In the threeflavor case things become a little more involved.
Iv Threeflavor case
In threeflavor quark matter besides the isospin chemical potential one has to consider the strange quark chemical potential. Microscopically, strange quark states can be occupied by electroweak processes if the light quark chemical potential exceeds the strange quark mass. The formal expression of the in medium effective chiral Lagrangian is given by Eq. (7) in which the mesonic octet is introduced by replacing
(42) 
in Eq. (5), where are the GellMann matrices.
The isospin and strange quark chemical potential can be introduced by considering
(43) 
where is the socalled strange quark chemical potential. Note that the actual strange quark chemical potential is , however the diagonal contribution of the baryonic chemical potential is immaterial for mesons.
For threeflavor quark matter the spontaneous symmetry breaking pattern is the following
(44) 
and the corresponding NGBs are identified with the mesonic pseudoscalar octet. The quark masses explicitly break the chiral symmetry, giving mass to the pseudo NGBs. A similar effect is produced by the isospin chemical potential and the strange quark chemical potential, with the additional fact of breaking Lorentz symmetry. The symmetry of the Lagrangian is thus reduced to
(45) 
The breaking of this symmetry leads, at most, to the appearance of one NGB and of the screening masses for the electromagnetic field.
For the external vector current we can write
(46) 
where
(47)  
(48) 
are the relevant components of the electromagnetic field.
iv.1 Ground state
In the threeflavor case the most general vev, , depends on parameters, corresponding to the possible orientations in space. However, in the twoflavor case we have found that rotations around the direction of the chemical potential leave the vacuum invariant. We assume that the same is true in the threeflavor case and therefore the vacuum Lagrangian only depends on two angles, and , corresponding to the angles between the vacuum and third component of isospin, and between the isospin and the hypercharge, respectively. This is exactly the same assumption used in Kogut:2001id (), in which it was found that there are three different vacua:

Normal phase:
(49) (50) characterized by
(51) 
Pion condensation phase:
(52) (53) characterized by
(54) (55) 
Kaon condensation phase:
(56) (57) characterized by
(58) (59) Note that the kaon condensation can only happen for
(60) and MeV for our parameter choice.
Screening masses
As in the twoflavor case we can determine the screening masses of the electromagnetic field. Remarkably, we find that the screening masses are independent of the angle and have the same expression obtained in the twoflavor case
(61) 
The nonvanishing value of the Meissner mass implies that in both mesonic condensed phase the system is an electromagnetic superconductor. Note that across the first order phase transition between the two condensed phases the screening masses are discontinuous, because is discontinuous.
iv.2 Mixing
In the presence of background isospinrich matter or strangenessrich matter, the Hamiltonian carries the third component of the isospin charge and of the strangeness charge (or hypercharge). The corresponding charges are explicitly broken, meaning that states with different third component of isospin and different hypercharge can mix. Indeed, the effect of a nonvanishing and is not only to produce a Zeemanlike splitting of the masses but also to tilt the vacuum in a certain direction in the isospin space corresponding to one of the nondiagonal generators of . Let us discuss this issue more in detail. Given that the Hamiltonian has terms proportional to and , the symmetry is explicitly broken. However, for labeling the mesonic states we can use spin, spin and spin quantum numbers (actually only two of them are independent, indeed is one of the Casimir operators), because , and commute with and . In Fig. 1 we report the weight diagram of the pseudoscalar mesonic octet. In the top panel the axes correspond to , , and and the values of the spin, spin and spin multiplets are reported. In the lower panel, mesonic states that can mix are marked with a different symbol. These diagrams are valid both in the phase and in the phase. For example, from the top panel we see that charged pions can mix because they both have , however charged pions cannot mix with kaons, because they all have . In Table 1 we report the and quantum numbers of the mesons with well defined spin and spin. In turn, the only allowed mixings are the followings: , , , corresponding to , , mixing. Regarding the and the , they have no welldefined spin or spin, thus a different reasoning must be used for understanding whether they mix or not. We will see that their mixing will depend on the particular spontaneously induced charge of the vacuum.
Mixing states  

One of the important aspects is that the third component of isospin and the hypercharge form the Cartan subalgebra of , thus the associated charges cannot directy induce mixing between different states. In other words, the and charges can induce Zeemanlike mass splittings, but whether mixing between states will happen or not depends on the spontaneously induced charge of the vacuum. Note that the operator associated to this induced charge can be described in terms of lowering and raising operator of one of the subgroups of .
Let us first consider the normal phase. In the normal phase there is no operator that can induce the mixing of the mesonic states, thus the mesonic states remain unchanged but the and charges will induce Zeemanlike mass splittings.
In any of the condensed phases, there is an additional charge that is spontaneously induced, and the corresponding operator will lead to mixing.
Let us first focus on isospin (or spin). We have to consider two cases. Suppose that the vacuum has a charge that commutes with , as in the phase, say the charge corresponding to , see Eq. (IV.1). The operators can induce mixing among the charged pions and among the kaons. On the other hand, spin conservation does not allow the to mix with the .
Suppose that the vacuum has a charge that does not commute with as in the phase, see Eq. (59). Any operator that does not commute with isospin will commute with spin or with spin. In the phase , then the vacuum is not invariant under this charge. However, since it follows that spin is conserved. The lowering and raising operator inducing the mixings will be . Regarding the and the , in this case we have that and do not mix. Since and , these will be the mass eigenstates.
In Kogut:2001id () it was found a different mixing in both condensed phases, with just two blocks and , meaning that states with different spin, spin and spin mix. As we will see in the next section, using the treelevel Lagrangian we find agreement with the mixing reported in Table 1.
iv.3 Mesonic mass spectrum
For vanishing chemical potentials and for equal light quark masses, the treelevel values of the mesonic octet masses in PT are known to be given by
(62)  
(63)  
(64) 
where is the strange quark mass. In the normal phase the effect of the isospin and strange quark chemical potential is a Zeemanlike mass splitting by contribution proportional to the isospin charge and strangeness,
(65)  
(66)  
(67)  
(68)  
(69) 
Below we will discuss the masses of the scalar mesons in the condensed phases. To obtain the eigenstates we follow the same procedure used in the twoflavor case, thus we first expand in order to obtain the quadratic terms in the fields then, if necessary, we rescale and rotate them to have canonical kinetic terms. We will denote the mass eigenstates with a tilde.
In Fig. 2 we report the obtained results for the pseudoscalar mesonic octet masses as a function of for three different values of the strange quark chemical potential. In the top panel we take MeV that is smaller than MeV, in the middle panel we take MeV and in the bottom panel MeV, which is the largest possible value of the strange quark chemical potential that can be considered in the present realization of PT. The solid vertical lines correspond to the second order phase transitions between the normal phase and a condensed phase. The dashed vertical lines correspond to first order phase transitions between the phase and the phase. The top panel and the middle panel should be compared with the corresponding results reported in Kogut:2001id () in Fig. 4 and in Fig. 5, respectively. In the normal phase our results agree with those of Kogut:2001id (), but in both condensed phases we disagree with the results reported in Kogut:2001id (). As already discussed, this is due to the fact that we find a different mixing pattern, even if we use the same model of PT of Kogut:2001id (). For this reason we discuss our results in detail.
Pion condensation phase
We find mixing within the following pairs of states: and , while the and fields do not mix. Thus, in agreement with the discussion in Sec. IV.2, we do not find mixing between the and the . The reason is that in the phase spin is conserved and therefore the state and the state cannot mix.
Since the Lagrangian can be organized in a block diagonal form, we can treat separately the various sectors. By the field rescaling
(70)  
(71) 
we obtain canonical kinetic terms. As in Sec. III.2 it is useful to turn to momentum space, so that one can absorb the terms linear in energy in the propagator. In this way we obtain the canonical Lagrangian
(72) 
where