Novemb er 1998 IASSNS-HEP 98/100
hep-ph/9811473
Continuity of Quark and
Hadron Matter
1
Thomas Schafer
and
2
Frank Wilczek
School of Natural Sciences
Institute for Advanced Study
Princeton, NJ 08540
Abstract
We review, clarify, and extend the notion of color- avor lo cking.
We present evidence that for three degenerate avors the qualitative
features of the color- avor lo cked state, reliably predicted for high
density,match the exp ected features of hadronic matter at low density.
This provides, in particular, a controlled, weak-coupling realization of
con nement and chiral symmetry breaking in this (slight) idealization
of QCD.
1
Research supp orted in part by NSF PHY-9513835.
e-mail: [email protected]
2
Research supp orted in part by DOE grant DE-FG02-90ER40542.
e-mail: [email protected] 1
In a recent study [1] of QCD with three degenerate avors at high density,
a new form of ordering was predicted, wherein the color and avor degrees of
freedom b ecome rigidly correlated in the groundstate: color- avor lo cking.
This prediction is based on a weak coupling analysis using a four-fermion
interaction with quantum numb ers abstracted from one gluon exchange. One
exp ects that sucha weak coupling analysis is appropriate at high density,for
the following reason [2,3,4].Tentatively assuming that the quarks start out
in a state close their free quark state, i.e. with large Fermi surfaces, one
nds that the relevantinteractions, which are scatterings the states near the
Fermi surface, for the most part involve large momentum transfers. Thus,
by asymptotic freedom, the e ective coupling governing them is small, and
the starting assumption is con rmed.
Of course, as one learns from the theory of sup erconductivity [5], even
weak couplings near the Fermi surface can have dramatic qualitative e ects,
fundamentally b ecause there are manylow-energy states, and therefore one
is inevitably doing highly degenerate p erturbation theory. Indeed, the au-
thors of [1] already p ointed out that their color- avor lo cked state, which
is constructed by adapting the metho ds of sup erconductivity theory to the
problem of high-density quark matter, displays a gap in all channels ex-
cept for those asso ciated with derivatively coupled spin zero excitations, i.e.
Nambu-Goldstone mo des. This is con nement. For massless quarks, they
also demonstrated sp ontaneous chiral symmetry breaking.
In very recentwork we [6], and others [7], have reinforced this circle
of ideas by analyzing renormalization of the e ectiveinteractions as one
integrates out mo des far from the Fermi surface. A fully rigorous treatment
will have to deal with the extremely near-forward scatterings, which are
singular due to the absence of magnetic mass for the gluons, at least in
straightforward p erturbation theory. This problem, which is presumably
technical, is any case ameliorated self-consistently for states of the color- avor
lo cking typ e, wherein all the gluons acquire mass through the Anderson-Higgs
mechanism.
In the earlier work [1], several striking analogies b etween the calculated
prop erties of the color- avor lo cked state and the exp ected prop erties of
hadronic matter at low or zero density, based on standard lore and observed
phenomenology,were noted. In addition to con nementandchiral symmetry
breaking, the authors observed that the dressed elementary excitations in the
color- avor lo cked state have the spin quantum numb ers of low-lying hadron 2
states and for the most part carry the exp ected avor quantum numb ers,
including integral electric charge (in units of the electron charge). Thus, as
we shall sp ell out immediately b elow, the gluons match the o ctet of vector
mesons, the quark o ctet matches the baryon o ctet, and an o ctet of collective
mo des asso ciated with chiral symmetry breaking matches the pseudoscalar
o ctet. However there are also a few apparent discrepancies: there is an extra
massless singlet scalar, asso ciated with the sp ontaneous breaking of baryon
numb er (sup er uidity);thereareeight rather than nine vector mesons (no
singlet); and there are nine rather than eightbaryons (extra singlet). We
will argue that these \discrepancies" are sup er cial { or rather that they are
features, not bugs.
Let us rst brie y recall the fundamental concepts of color- avor lo cking.
The case of three massless avors is the richest due to its chiral symme-
try (and adding a common mass do es not change anything essential) so we
shall concentrate on it. The primary condensate, which one calculates using
the metho ds of sup erconductivity theory near the Fermi surface, involves
diquarks. It takes the form [1]
_ _
j
k l
i
(1) + i = q i = hq q hq
2 1 ij
_
_
a b b a Lb La
Rka
Rlb
Here L, R lab el the helicity, i; j; k ; l are two-comp onent spinor indices, a; b
are avor indices, and ; are color indices. A common space-time argu-
ment is suppressed. ; are parameters (dep ending on chemical p otential,
1 2
coupling, :::) whose non-zero values emerge from a dynamical calculation.
This equation must b e interpreted carefully. The value of any lo cal quan-
tity which is not gauge invariant, taken literally, is meaningless, since lo cal
gauge invariance parametrizes the redundantvariables in the theory,and
cannot b e broken [8]. But as weknow from the usual treatment of the elec-
troweak sector in the Standard Mo del, it can b e very convenient to use such
quantities. The p oint is that we are allowed to x a gauge during interme-
diate stages in the calculation of meaningful, gauge invariantquantities {
indeed, in the context of weak coupling p erturbation theory,wemust do so.
For our present purp oses however it is imp ortant to extract non-p erturbative
results, esp ecially symmetry breaking order parameters, that we can match
to our exp ectations for the hadronic side. To do this, we can take suitable
pro ducts of the memb ers of (1) and their complex conjugates, and contract
the color indices. In this waywe can pro duce the square of the standard chiral 3
symmetry breaking order parameter of typ e hq q i and a baryon numb er vio-
L R
2
lating order parameter of typ e h(qqq) i, b oth scalars and singlets under color
and avor. Atthislevel only the square of the usual chiral order parameter
app ears, fundamentally b ecause our condensates preserve left-handed quark
number modulo two. This conservation law is violated by the six-quark
vertex asso ciated with instantons, and byconvolving that vertex with our
four-quark condensate we can obtain the usual two-quark chiral symmetry
breaking order parameter [9].
By demanding invariance of the diquark condensate directly,we infer the
c L R
symmetry breaking pattern SU (3) SU (3) SU (3) U (1) ! SU (3) .
c
Here among the initial microscopic symmetries SU (3) is lo cal color symme-
try, while the remaining factors are chiral family and baryon numb er symme-
tries. The nal residual unbroken symmetry is a global diagonal symmetry.
Indeed, the Kronecker deltas in the nal term of (1) are invariantonlyunder
simultaneous color and avor rotations, so the color and avor are \lo cked".
This lo cking o ccurs separately for the left and right handed quarks, but
since color symmetry itself is vectorial, the e ect is also to lo ckleftand
righthanded avor rotations, breaking chiral symmetry. The global baryon
numb er symmetry is, of course, manifestly broken, but quark number is con-
served mo dulo two. Pro jecting onto the gauge invariant, color singlet, sector
this implies that baryon numb er is violated only mo dulo two. The same
conclusions would emerge from analysis of the gauge invariant symmetry
generators only, up on consideration of the gauge invariant order parameters
we constructed ab ove.
Ordinary electromagnetic gauge invariance, like color symmetry, is vio-
lated by (1), but a linear combination of hyp ercharge (diagonal matrix -2/3,
1/3, 1/3) and electromagnetic charge (diagonal matrix 2/3, -1/3, -1/3) anni-
hilates the combinations correlated by color- avor lo cking, and generates a
true symmetry.Thephysical result is that there is a massless gauge degree
of freedom, representing the photon as mo di ed byitsinteraction with the
condensate. As seen by this mo di ed photon, all the elementary excitations
have appropriate charges to match the corresp onding hadronic degrees of
freedom. In particular, their charges are all integral multiples of the electron
charge [1]. This is, of course, another classic asp ect of con nement.
It was essential, in this construction, that the charges of the quarks add
up to zero. If that were not so, wewould not have b een able to nd a color
generator capable of comp ensating the violation of naive electromagnetic 4
gauge invariance. Yet it seems somewhat accidental that these charges do
add up to zero, and one would b e quite worried if any qualitative asp ect of
con nement dep ended on this accident. This worry touches the form rather
than the substance of our argument. If the quark charges did not up to zero,
it would not b e valid to ignore Coulomb repulsion. One would havetoadda
comp ensating charge background as a mathematical device, or contemplate
inhomogeneous states. Insofar as wewant to use external gauge elds as
a prob e of pure QCD, wemust restrict ourselves to those which preserve
the overall neutrality of the QCD groundstate. Fortunately,inourslightly
idealized version of QCD no awkwardness arises for the physically imp ortant
gauge eld, i.e. the physical photon.
The elementary excitations are of three typ es. The color gluons b e-
come massivevector mesons through the Anderson-Higgs mechanism. Due
to color- avor lo cking, they acquire avor quantum numb ers, which makes
them an o ctet under the residual SU (3) . The quark elds give single-
particle spin 1/2 excitations whose stability is guaranteed by the residual Z
2
quark (or baryon) numb er symmetry. These excitations are massive, due to
the color- avor sup erconducting gap. They form an o ctet with the quantum
numb ers of the nucleon o ctet, plus a singlet. It might seem p eculiar on rst
hearing that a single quark can b ehaveasa baryon, but rememb er that there
is a condensate of diquarks p ervading this phase. In addition there are col-
lective Nambu-Goldstone mo des, asso ciated with the sp ontaneously broken
global symmetries. These are a massless pseudoscalar o ctet asso ciated with
chiral symmetry breaking, and a scalar singlet asso ciated with baryon num-
b er violation. A common quark mass lifts the pseudoscalar o ctet, but not
the singlet, b ecause it sp oils microscopic chiral symmetry but not microscopic
Clearly, there are striking resemblances b etween the elementary excita-
tions of color- avor lo cked quark matter and the low-energy hadron sp ec-
trum. One is tempted to ask whether they mightbeidenti ed. More pre-
cisely, one might ask whether strongly coupled hadronic matter at low density
go es over into the calculable, weak-coupling form of quark matter just de-
scrib ed without a phase transition. If so, then the con nement and chiral
symmetry breaking calculated for the weak coupling phase not only resemble
these central prop erties of low-density QCD, but are rigorously indistinguish-
able from them. This sort of p ossibility, that Higgs and con ned phases are
rigorously indistinguishable, has long b een known to o ccur in simple abstract 5
mo dels [10 ].
As mentioned ab ove, however, at rst sight there app ear to b e several
diculties with this identi cation. Wenow debunk them in turn.
The most profound of the apparent diculties is the existence of an ex-
tra scalar Nambu-Goldstone mo de, and the related phenomenon that baryon
4
number is spontaneously violated (indicating, as in liquid He , sup er uid-
ity). The answer to this comes through prop er recognition of an imp ortant
though somewhat exotic phenomenon for three degenerate avorsonthe
hadron side. Several years ago R. Ja e discovered [11], in the context of
the MIT bag mo del, that a particular dibaryon state, the H, a spin 0 SU(3)
singlet with quark content (udsuds), is surprisingly light. This arose, in his
calculations, b ecause of a particularly favorable contribution from color mag-
netism. Roughly sp eaking, in the H con guration the color elds asso ciated
with the quark sources are minimized, together with the energy they would
otherwise store, by arranging b oth the colors and spins to cancel pairwise
to the greatest extent p ossible. It has b een debated, for QCD with realistic
quark masses, whether H mightbeonlyslightly ab ove the nn or n thresh-
olds. Though at this level the outcome for realistic QCD is unclear, b oth
theoretically [12] and exp erimentally [13 ], it has come to seem quite likely
that in QCD with three degenerate quarks the H will b e the particle with
smallest energy p er unit baryon numb er. Thus at any nite baryon number
density,however small, at zero temp erature one should exp ect, in this con-
text, to nd a Bose condensate of H dibaryons. This condensate gives us
precisely { i.e. with the appropriate quantum numb ers { the sup er uid we
were led to exp ect from our sup er cially very di erent considerations on the
quark matter side.
If the H is ab ove dibaryon threshold, one will have a narrow range of
chemical p otentials where baryon numb er is built up by single baryons. Based
on the same calculations [11, 12 ], it is extremely plausible that in this case
there will b e attraction in the H channel at the Fermi surface, and hence
sup er uidity of the required typ e, now through a BCS mechanism.
This sup er uidity, whatever its source, supplies us with the key to the
riddle of the missing vector meson. For once there is a massless singlet
scalar, the putative singlet vector b ecomes radically unstable, and should
not app ear in the e ective theory.Itmight b e ob jected here that the o ctet
of vector mesons is also unstable { for massless quarks { against decayinto
massless scalar and pseudoscalar mesons. A quick answer is that this is not 6
really an ob jection at all, b ecause there is no harm in having redundant states
(whose instability will app ear immediately up on more accurate calculation).
There is a much prettier and more satisfying answer, however. If weturnon
non-zero masses for the quarks the pseudoscalar o ctet (but not the singlet)
will b ecome massive. Eventually the decayofthevector o ctet (but not the
singlet) will b e blo cked, and then we will b e grateful for the prescience of the
theory in providing the appropriate degrees of freedom.
Finally, there is the question of the \extra" singlet baryon. This is the
most straightforward. In the original calculations [1], it was found that the
singlet gap is much larger than the o ctet gap. Thus the singlet baryon is
predicted to b e considerably heavier than the o ctet. This is not problematic:
a particle of this sort is exp ected in the quark mo del, it could well exist
in reality, and in any case it is radically unstable against decayinto o ctet
baryon and o ctet pseudoscalar, at least for massless or light quarks.
So all the ob jections havebeenanswered. Continuity of quark and hadron
matter, far from b eing paradoxical, now app ears as the default option.
Clearly, sup er uidity of quark/hadron matter has b een essential for the
argument. There is considerable evidence for pairing in nuclei [14]. Its full
realization is limited by the nite size of nuclei, which in turn arises from the
non-negligible strange quark mass and the Coulomb energy that arises in the
most favorable (for QCD), symmetric arrangement of neutrons and protons.
These limitations might b e relieved to some extentinheavy ion collisions ac-
companied by creation of many strange-antistrange pairs, followed bycharge
segregation. An imp ortant signature for this, emphasized by the consider-
ations ab ove, is broadening of vector mesons, esp ecially the singlet. This
e ect might b e observable in the dimuon sp ectrum.
Our considerations here are clearly relevanttoany attempt to mo del the
deep interior of neutron stars, or conditions during sup ernova and hyp ernova
explosions. To do justice to these questions, it will b e very imp ortantto
include the e ects of unequal quark masses and of electromagnetism.That
is an imp ortant task for the future.
In the remainder of this pap er we shall consider a related but simpler prob-
lem, that of extending the analysis to larger numb ers of degenerate quarks.
An imp ortant foundational result, which emerges clearly from this analysis,
is that the color- avor lo cked state for three avors, whichwas rst guessed
to b e favorable b ecause of its large residual symmetry and by analogy to the
3
B phase of sup er uid He , is in fact the global minimum for three avors. It 7
also reapp ears as a building blo ck for larger numb ers of avors.
The renormalization group analysis in [6, 7 ] allows one to classify p ossible
instabilities, and to assess their relative imp ortance, for small but otherwise
arbitrary couplings near the Fermi surface. It was found that the dominant
instability corresp onds to scalar diquark condensation. The analysis do es
not x the color and avor channel of this instability uniquely, indep endent
of initial conditions for the couplings, since there are two equally enhanced
marginal interactions. One gluon exchange, which dominates for weak cou-
pling, is attractive in the color anti-symmetric 3channel, and favors one of
these interactions. During the evolution this interaction will grow, while the
repulsiveinteraction in the color symmetric 6 channel is suppressed. Thus
the instability is driven by a leading interaction of the form
ac bd ad bc
L = K ( )
o n
C C (C $ C ) ; (2)
5 5 5
b c d a
where as b efore ; ;::: are color indices and a;b;::: are avor indices. The
Dirac structure of the interaction b ecomes more transparent when written in
achiral basis. Wehave
j
l k i
+(L $ R): (3)
ij kl
L L L L
The renormalization group analysis only provides the form of the domi-
nantinteraction, not the structure of the order parameter. In particular, it
do es not tell us whether color- avor lo cking is the preferred state in three
avor QCD. In order to answer this question, wehave to p erform a varia-
tional analysis. Since the interaction is attractiveins-wave states, it seems
clear that the dominant order parameter is an s-wave, to o. We then only
have to study the color- avor structure of the primary condensate. For this
purp ose, we calculate the e ective p otential for the order parameter
j
ij i
: (4) i =
h
ab Lb La
is a N N matrix in avor space and a N N matrix in color
f f c c
ab
space. Overall symmetry requires that is symmetric under the combined
ab
exchange (a ) $ (b ). Also, since the interaction only involves color and
avor anti-symmetric terms, the e ectivepotential do es not dep end on color
and avor symmetric comp onents of . This means that the e ective
ab 8
p otential has at least N (N +1)N (N +1)=4 at directions. These trivial
c c f f
at directions will b e lifted by subleading interactions not included in our
analysis. We will comment on the imp ortance of subleading terms b elow.
We calculate the e ectivepotential in the mean eld approximation. This
approximation corresp onds to resumming all \cactus" diagrams. These di-
agrams are exp ected to b e dominant b oth in the limit of large chemical
p otential and in the large N ;N limit. In the mean eld approximation, the
c f
quadratic part of the action b ecomes
a b a b
: (5) + = K M
ba L L ba ab ab ab L L
Integrating over the fermion elds we obtain the familiar tr log term in the
e ective p otential. In order to evaluate the logarithm, wehave to diagonalize
the mass matrix M. Let us denote the corresp onding eigenvalues by ( =
1;:::;N N ). These are the physical gaps for the N N fermion sp ecies.
c f f c
Adding the mean eld part of the e ective p otential, we nally obtain
X
ab
V () = ( )+M : (6)
ef f
ab
Here, ( ) is the kinetic term in the e ective action for one fermion sp ecies,
Z
q q
3
d p
2 2 2 2
( )= (p ) + + (p + ) + : (7)
3
(2 )
This integral has an ultra-violet divergence. This divergence can b e removed
by expressing in terms of the renormalized interaction [15]. In this work
we are not really interested in the exact numerical value of the gap, but
only in the symmetries of the order parameter. For simplicity,we therefore
regularize the integral byintro ducing a sharp three-momentum cuto .
The e ectivepotential (6) dep ends on N (N 1)N (N 1)=4 parame-
c c f f
ters. We minimize this function numerically. In order to make sure that the
minimization routine do es not b ecome trapp ed in a lo cal minimum we start
the minimization from several di erent initial conditions. For the numerical
analysis wealsohaveto xthevalue of the chemical p otential , the coupling
constant K , and the cuto . Wehavechecked that the symmetry breaking
pattern do es not dep end on the values of these parameters. Wehaveused
2
=0:5 GeV, = 0:6GeVandK =3:33= , similar to what was considered
in [3, 4]. 9
After we determine the matrix that minimizes the e ectivepotential
ab
2
we study the corresp onding symmetry breaking. Initially, there are N 1
f
global avor symmetries for b oth left and right handed fermions, as well
2
as N 1 lo cal gauge symmetries. Sup er uidity reduces the am-mountof
c
symmetry. In order to nd the unbroken generators we study the second
2 2 2
variation of the order parameter =( ), where (i =1;:::;N + N
i j i
f c
2) parameterizes the avor and color transformations. Zero eigenvalues of this
matrix corresp ond to unbroken color- avor symmetries. The corresp onding
eigenvectors indicate whether the unbroken symmetry is a pure color, a pure
avor, or a coupled color- avor symmetry.
Our results are summarized in Table 1. The two avor case is sp ecial. In
this case, the dominant order parameter do es not break the color symmetry
completely, and the avor symmetry is completely unbroken. This is the sce-
nario discussed in [3, 4 ]. Sub dominantinteractions can break the remaining
color symmetry, either with or without [3] avor symmetry breaking.
The main result is that, for three avors, weverify that color- avor lo ck-
ing is indeed the preferred order parameter. We nd that all quark sp ecies
acquire a mass gap, and b oth color and avor symmetry are completely
broken. There are eightcombinations of color and avor symmetries that
generate unbroken global symmetries. These are the generators of the diag-
c+L+R
onal SU (3) . Also, the quark mass gaps fall into representations (8+1)
of the unbroken symmetry.And,asmentioned ab ove, the singlet state is
twice heavier than the o ctet.
Note that in the present analysis, which only takes into account the lead-
ing interaction, the order parameter is completely anti-symmetric in b oth
I
color and avor. We nd . If subleading interactions are
abI
ab
taken into account, the order parameter will have the more general form
= + . This order parameter leaves the same residual
1 2
ab a b b a
symmetry.
The main qualitative results we found for three avors extend to N >
f
3. Color symmetry is always completely broken, and all quarks acquire a
mass gap. The only remaining symmetries are global coupled color- avor
symmetries. For massless quarks, chiral symmetry is sp ontaneously broken.
For an o dd number of avors, there are subleading instanton op erators that,
after the dominant gap is formed, can give an exp ectation value to op erators
of the form .For even numb ers of avors, generating a non zero h i
L R L R
is more subtle. Instantons can only give an exp ectation value to op erators 10
N N N gaps (deg) =(N N ) N
c f par c f sy m
3 2 3 (4) 3 ( ) + 3 (col)
0 0
3 3 9 (8), 2 (1) 0:80 1:27 8
0 0
3 4 18 (8), 2 (4) 0:63 1:21 6
0 0
3 5 30 (5), 2 (7), 3 (3) 0:43 1:18 3
0 0
3 6 45 (16), 2 (2) 0:80 1:27 9
0 0
Table 1: Groundstate prop erties of the s-wave sup er uid state in QCD with
N = 3 colors and N avors. N = N (N 1)N (N 1)=4 is the number
c f par c c f f
of totally anti-symmetric gap parameters. The column lab eled \gaps (deg)"
gives the relative magnitude of the gaps in the fermion sp ectrum, together
with their degeneracy.Thenumerical values of the gap and the condensation
3
energy p er sp ecies are given in units of = 36 MeV and =0:73 MeV =fm ,
0 0
resp ectively. N is the number of unbroken color- avor symmetries.
sy m
2
of the form ( ) .
L R
N = N (or a multiple thereof ) is the most favorable case, in the precise
f c
sense that in this case the condensation energy p er sp ecies is maximal. If N
f
is a multiple of N , the dominant gap corresp onds to multiple emb eddings
c
of the N = N order parameter. Wehave not studied the case N 6=3
f c c
systematically, but since color and avor are interchangeable in (2), the case
N =3forvarious numb ers of avors is covered implicitly. Also, wehave
f
not studied the interesting case N = N !1.
f c
Acknowledgement: We wish to thank S. Treiman for some helpful ques-
tions. 11
References
[1] M. Alford, K. Ra jagopal, and F. Wilczek, hep-ph/9804403.
[2] D. Bailin and A. Love, Phys. Rept. 107, 325 (1984), and references
therein.
[3] M. Alford, K. Ra jagopal, and F. Wilczek, Phys. Lett. 422, 247 (1998).
[4] R. Rapp, T. Schafer, E. V. Shuryak, and M. Velkovsky,Phys. Rev. Lett.
81 (1998) 53.
[5]J.R.Schrie er, Theory of Superconductivity, (Benjamin/Cummings,
Reading, Mass. 1964; 3rd printing, 1984).
[6] T. Schafer and F. Wilczek, hep-ph/9810509 .
[7] N. Evans, S. Hsu, and M. Schwetz, hep-ph/9808444 , hep-ph/9810514.
[8] S. Elitzur, Phys. Rev. D12, 3978 (1975).
[9] R. Rapp, T. Schafer, E. V. Shuryak, and M. Velkovsky, in preparation.
[10] E. Fradkin and S. Shenker, Phys. Rev. D19, 3682 (1979).
[11] R. Ja e, Phys. Rev. Lett. 38, 195, 617(E) (1977).
[12] C. Dover, Nuovo Cim. 102A, 521 (1989).
[13] Pro ceedings of the International Conference on Hyp ernuclear and
Strange Particle Physics (HYP '97), to b e published in Nucl. Phys. A
(1998).
[14] P. Ring and P.Schuck, The Nuclear Many-Body Problem, (Springer-
Verlag, New York, 1980), chapter 6.
[15] S. Weinb erg, Nucl. Phys. B413, 567 (1994) 12