1: % SSB ve fazovem diagramu QCD
2:
3: \chapter{Quantum chromodynamics at nonzero density}
4: \label{Chap:QCD} The physics of hot and/or dense matter is described by the
5: phase diagram of QCD. While the region of low net baryon density and high
6: temperature is being explored experimentally in heavy ion collisions, the cold
7: and very dense nuclear matter seems to exist only in the neutron stars.
8:
9: It has been known for a~long time that at sufficiently high density quarks are
10: no longer confined\footnote{In fact, the term \emph{quark confinement} loses
11: its sense once the mean distance between quarks is much smaller than the
12: confinement scale. The quarks then do not feel the long-distance strong
13: attraction and provide the appropriate degrees of freedom to describe the
14: highly squeezed matter.} and may undergo the Cooper pairing very much analogous
15: to that in ordinary superconductors \cite{Bailin:1984bm}. However, only in the
16: past decade has the phenomenon of color superconductivity attracted
17: considerable attention due to the discovery that it may appear already at
18: densities attainable in the neutron stars \cite{Rapp:1998zu,Alford:1998zt}.
19:
20: Since then, the subject has been investigated to great detail and several
21: qualitatively different phases have been found. Extensive reviews are given in
22: Refs. \cite{Rajagopal:2000wf,Buballa:2003qv,Alford:2001dt,Rischke:2003mt}. An
23: introduction to the physics of cold dense quark matter may be found in the
24: lecture notes \cite{Shovkovy:2004me,Schaefer:2005ff}.
25:
26: Despite the amount of energy devoted to the study of the QCD phase diagram,
27: there is still a~controversy regarding the structure of the ground state at
28: moderate baryon density. It seems that we are only confident that at very high
29: densities the quark matter resides in the \emph{Color-Flavor-Locked} phase
30: \cite{Alford:2002rz}. This is supported by the weak-coupling calculations from
31: first principles, which are applicable due to the asymptotic freedom of QCD.
32:
33: On the other hand, the knowledge of the moderate-density region of the phase
34: diagram is rather weak. Usually, either the weak-coupling results are directly
35: extrapolated just by running the QCD coupling, or the structure of the
36: interaction is taken over from the high-density regime and used as an input to
37: the phenomenological models such as that of Nambu and Jona-Lasinio.
38:
39: This chapter consists of two main parts. In the first one, we introduce an
40: alternative mechanism for generating the effective four-quark interaction and
41: show that it leads to an unconventional pairing in the color-sextet channel.
42: This is based on our paper \pubcite1.
43:
44: The second part, based on the recent paper \pubcite4, deals with a~different
45: approach to the QCD phase diagram. Inasmuch as we cannot attack the problem of
46: the QCD phase diagram at moderate density directly and current lattice
47: techniques fail in that region as well, it is plausible to study theories
48: similar to QCD which are amenable to both analytical and lattice calculations.
49: We describe a~simple case of such a~theory -- the two-color QCD with two quark
50: flavors -- and provide a~new setting for its low-energy description in terms of
51: the chiral perturbation theory.
52:
53:
54: \section{Single-flavor color superconductor with color-sextet pairing}
55: It is most common to describe the quark matter at moderate baryon density
56: within the Nambu--Jona-Lasinio model \cite{Buballa:2003qv}. In this approach,
57: the crucial point is the choice of the model interaction. The color and flavor
58: structure of the interaction are usually taken over from the weak-coupling
59: regime -- either perturbative (the one-gluon exchange) or nonperturbative (the
60: instanton-mediated interaction). Both these interactions share the common
61: feature that they are attractive in the color-antisymmetric channel and
62: repulsive in the color-sym\-metric one. It should, however, be stressed that
63: the arguments based on the weakly coupled QCD merely provide an
64: \emph{evidence}. There is \emph{no proof} that the strongly coupled QCD at
65: moderate density inevitably leads to the same behavior. It is therefore worth
66: exploring the alternatives.
67:
68: In this section we shall investigate the behavior of dense quark matter under
69: the assumption that the quarks pair in the color-symmetric (sextet) channel. We
70: shall for simplicity consider a~homogeneous phase of a~single-flavor quark
71: matter. The physical reasoning behind this assumption is the following. The
72: color, flavor and spin structures of the Cooper pair are connected by the
73: requirement that the Pauli exclusion principle be satisfied. This means that,
74: as long as the orbital momentum is zero, the total spin of the color-sextet
75: Cooper pair of quarks of a~single flavor must be zero. On the contrary, in the
76: color-antitriplet channel the Pauli principle requires total spin one.
77:
78: The point is that the spin and orbital momentum effects dramatically reduce the
79: energy gap i.e., the binding energy of the Cooper pair. Indeed, while -- in the
80: color-antitriplet channel -- the gap of the two-flavor spin-zero superconductor
81: at moderate density is roughly estimated as tens $\mathrm{MeV}$, the gap of the
82: one-flavor spin-one superconductor is only tens or a~hundred $\mathrm{keV}$
83: \cite{Buballa:2002wy}. In the latter case, the color-sextet pairing might
84: prevail even if the pairing interaction is quite weak.
85:
86: It is well known that while at very high density the CFL phase is the stable
87: ground state of the three-flavor quark matter, at moderate density the CFL
88: pairing is disfavored by the strange quark mass and the resulting mismatch of
89: the Fermi momenta. The $2+1$ pairing scheme is more likely. The up and down
90: flavors are bound by the strong attractive interaction in the color-antitriplet
91: channel. The strange quarks then pair with themselves and we suggest here that
92: the pairing be in the \emph{color-sextet spin-zero} channel rather than the
93: color-antitriplet spin-one channel favored by the one-gluon exchange
94: interaction.
95:
96: We first \emph{assume} the particular form of the pairing and explore its
97: impact on the symmetry of the theory. It is only later that we provide
98: a~physical motivation for the attraction in the color-symmetric channel and work
99: out the description within the Nambu--Jona-Lasinio model.
100:
101:
102: \subsection{Kinematics of color-sextet condensation}
103: Suppose that the superconducting phase is described by the order parameter
104: $\Phi$ which transforms in the $\mathbf 6$ representation of the color
105: $\mathrm{SU(3)}$ group. It is best represented by a~complex symmetric $3\times
106: 3$ matrix upon which the $\mathrm{SU(3)\times U(1)}$
107: transformations\footnote{Recall from Section \ref{Sec:lsm_for_sextet} that the
108: $\mathrm{U(1)}$ here represents the baryon number.} act as $\Phi\to U\Phi\tr
109: U$. The assumption that the ordered phase be homogeneous translates to the
110: requirement that $\Phi$ be a~spacetime-independent constant. Note that in the
111: Nambu--Jona-Lasinio model $\Phi_{ij}$ will correspond to the vacuum expectation
112: value of the bilinear operator $\psi_{\alpha
113: i}(C\gamma_5)_{\alpha\beta}\psi_{\beta j}$, but for now this interpretation is
114: not needed.
115:
116: The crucial observation is that any complex symmetric matrix $\Phi$ may be
117: brought by a~suitable $\mathrm{SU(3)\times U(1)}$ transformation to a~special
118: form $\Delta$ which is diagonal, real and positive \cite{Schur:1945ab}. We
119: shall denote its diagonal entries as $\Delta_1,\Delta_2,\Delta_3$. These cannot
120: be changed by a~unitary transformation since they are the eigenvalues of the
121: positive Hermitian matrix $(\he\Phi\Phi)^{1/2}$, and thus represent \emph{three
122: independent order parameters} of the phase.
123:
124: The presence of three order parameters makes the phase structure of the
125: color-sextet superconductor quite rich. Depending on the relative values of the
126: order parameters, several symmetry-breaking patterns may be distinguished:
127: \begin{enumerate}
128: \item \emph{All $\Delta$'s are different and nonzero.} This is the most general as
129: well as intriguing possibility. The continuous $\mathrm{SU(3)\times U(1)}$
130: symmetry is completely broken, only a~discrete $(Z_2)^3$ is left.
131: \item \emph{Two $\Delta$'s are equal and nonzero.} In this case, there is
132: a~residual $\mathrm{O(2)}$ symmetry in the corresponding $2\times2$ block of
133: $\Phi$.
134: \item \emph{$\Delta_1=\Delta_2=\Delta_3\neq0$.} Quite similar to the previous
135: case, but now the enhanced symmetry of the ground state is $\mathrm{O(3)}$.
136: \item \emph{Some of the $\Delta$'s are zero.} According to the number of
137: vanishing order parameters, there is a~residual $\mathrm{U(1)}$ or
138: $\mathrm{U(2)}$ invariance, simply meaning that the corresponding colors do not
139: participate in the pairing.
140: \end{enumerate}
141:
142: It will turn out in the following that the possibility of most interest is the
143: $\mathrm{O(3)}$-symmetric phase. Since this results in the same number of
144: broken color generators as the breaking $\mathrm{SU(3)\to SU(2)}$ by the
145: standard color antitriplet, it is worthwhile to comment on the difference
146: between these two symmetry-breaking patterns.
147:
148: The structure of the spectrum is always determined by the unbroken subgroup.
149: Now the breaking $\mathrm{SU(3)\to SO(3)}$ is isotropic so that all five broken
150: generators fall into a~single ($5$-plet) representation of $\mathrm{SO(3)}$. On
151: the other hand, in the $\mathrm{SU(3)\to SU(2)}$ case four of the broken
152: generators form a~complex $\mathrm{SU(2)}$ doublet while the remaining one is
153: a~singlet.
154:
155:
156: \subsection{Ginzburg--Landau description}
157: To determine which of the possible symmetry-breaking patterns are actually
158: realized, one has to employ a~particular model to calculate the order parameter
159: $\Phi$. Ignoring for the moment the fluctuations of the order parameter(s), we
160: have to write down the most general $\mathrm{SU(3)\times U(1)}$ invariant
161: potential, whose minimum determines the ground state. Such a~potential can
162: always be written in terms of a~certain set of algebraically independent
163: invariants. In our case there are three of them, namely $\Tr\he\Phi\Phi$,
164: $\det\he\Phi\Phi$ and $\Tr(\he\Phi\Phi)^2$.
165:
166: Restricting to quartic polynomials of the Ginzburg--Landau type, the most
167: general potential reads
168: $$
169: V(\Phi)=-a\Tr\he\Phi\Phi+b\Tr(\he\Phi\Phi)^2+c(\Tr\he\Phi\Phi)^2.
170: $$
171: Such a~potential was already investigated in Section \ref{Sec:lsm_for_sextet}.
172: It was shown that the nature of the global minimum depends on the sign of the
173: parameter $b$. When $b>0$, the order parameter $\Delta$ is proportional to the
174: unit matrix so that the ground state has the $\mathrm{SO(3)}$ symmetry. When
175: $b<0$, the minimizing configuration is such that $\Delta$ has a~single nonzero
176: diagonal entry, corresponding to the symmetry breaking pattern
177: $\mathrm{SU(3)\times U(1)\to SU(2)\times U(1)}$.
178:
179: We stress the fact that the parameters $a,b,c$ are unknown at this stage so
180: that we cannot decide which of the ordered phases is actually realized. It is,
181: however, possible to derive the Ginzburg--Landau functional from the underlying
182: microscopic model, either QCD or Nambu--Jona-Lasinio \cite{Giannakis:2001wz}.
183:
184: To account for the fluctuations of the order parameter $\Phi$, the
185: Ginzburg--Landau functional has to be enriched with derivative terms. The
186: lowest-order Lagrangian reads
187: \begin{equation}
188: \LA
189: =\alpha_e\Tr\he{\de_0\Phi}\de^0\Phi+\alpha_m\Tr\he{\de_i\Phi}\de^i\Phi-V(\Phi).
190: \label{GL_effective_Lagr}
191: \end{equation}
192: The coefficients $\alpha_e$ and $\alpha_m$ are in general different since
193: Lorentz invariance is broken by medium effects. Note that the ``kinetic term''
194: of $\Phi$ is not canonically normalized -- this is because $\Phi$ represents a
195: composite object, the Cooper pair of quarks
196: \cite{Pisarski:1999gq,Rischke:2000qz}.
197:
198: Treating the dense quark matter at moderate baryon density as a BCS-type
199: superconductor, one may next switch on the colored gauge fields perturbatively.
200: Within the effective Lagrangian \eqref{GL_effective_Lagr}, this amounts to
201: replacing the ordinary derivatives with the covariant ones,
202: $$
203: \de_{\mu}\Phi\to D_{\mu}\Phi=\partial_{\mu}\Phi-\imag g A_{\mu}^{a}
204: \left(\tfrac{1}{2}\lambda_{a}\Phi+\Phi\tfrac{1}{2}\tr\lambda_{a}\right),
205: $$
206: and adding the Yang--Mills kinetic term for the gluons. As a result of the
207: usual Higgs mechanism, both electric and magnetic gluons acquire nonzero masses
208: -- the Debye and the Meissner ones, respectively. At zero temperature, the
209: coefficients are roughly $\alpha_{e,m}\sim\mu^2/\Delta^2$ so that both Debye
210: and Meissner masses are found to be of order $g\mu$ (for detailed results and
211: their discussion see Ref. \pubcite1).
212:
213: However, as pointed out by Rischke \cite{Rischke:2000qz}, the gauged
214: lowest-order Lagrangian \eqref{GL_effective_Lagr} does not reproduce correctly
215: the mass ratios of the gluons of different adjoint colors. The reason is the
216: restriction to operators of dimension four or less we employed to construct the
217: Lagrangian \eqref{GL_effective_Lagr}. For a more proper treatment, higher-order
218: operators like $\left|\Tr(\Phi^{\dagger}D_i\Phi)\right|^2$ have to be included,
219: which also contribute to the gluon masses.
220:
221:
222: \subsection{Nambu--Jona-Lasinio model}
223: We shall now develop the description using the elementary quark fields. Here we
224: come to the point of the proper choice of the four-fermion interaction. As
225: already mentioned above, we do not take up any of the interactions commonly
226: used in literature, but rather follow a~different approach. Our motivation goes
227: back two decades to the paper by Hansson et al. \cite{Hansson:1982dv}. These
228: authors investigated the possibility of the existence of the bound states of
229: two gluons and classified the strength of the QCD-induced force by the total
230: spin and color content of the gluon pair.
231:
232: They discovered that apart from the colorless glueball, the most strongly bound
233: state is that of total spin zero which transforms as a~color octet. Such
234: a~state, of course, cannot exist as an excitation of the QCD vacuum. It might,
235: however, be a~well defined degree of freedom in the dense deconfined phase. Now
236: if it really exists, it certainly interacts with the quarks and its exchange
237: leads to the effective fermionic Lagrangian (see Fig.
238: \ref{Fig:glueball_induced_interaction}),
239: \begin{equation}
240: \LA=\bar\psi(\imag\slashed\de-m+\mu\gamma_0)\psi+G(\bar\psi\vek\lambda\psi)^2,
241: \label{NJLlagrangian}
242: \end{equation}
243: with $G>0$. (The color and spin indices are suppressed.) The proposed
244: interaction is attractive in the color-sextet channel and provides the basis
245: for the following analysis.
246: \begin{figure}
247: $$
248: \parbox{40\unitlength}{%
249: \begin{fmfgraph}(40,20)
250: \fmfset{arrow_len}{3mm}
251: \fmfset{curly_len}{2mm}
252: \fmfleftn{l}{2}
253: \fmfrightn{r}{2}
254: \fmf{fermion,tension=2}{l1,vld}
255: \fmf{fermion,tension=2}{vlu,l2}
256: \fmf{fermion}{vld,vlu}
257: \fmf{fermion,tension=2}{r1,vrd}
258: \fmf{fermion,tension=2}{vru,r2}
259: \fmf{fermion}{vrd,vru}
260: \fmf{gluon}{vld,vl,vlu}
261: \fmf{gluon}{vru,vr,vrd}
262: \fmfv{d.sh=circle,d.si=0.1w,d.fi=shaded}{vl}
263: \fmfv{d.sh=circle,d.si=0.1w,d.fi=shaded}{vr}
264: \fmf{dashes}{vr,vl}
265: \end{fmfgraph}}
266: \longrightarrow
267: \parbox{30\unitlength}{%
268: \begin{fmfgraph}(30,20)
269: \fmfset{arrow_len}{3mm}
270: \fmfleftn{l}{2}
271: \fmfrightn{r}{2}
272: \fmf{fermion}{l1,vl,l2}
273: \fmf{fermion}{r1,vr,r2}
274: \fmf{dashes}{vl,vr}
275: \fmfdot{vl}
276: \fmfdot{vr}
277: \end{fmfgraph}}
278: \longrightarrow
279: \parbox{20\unitlength}{%
280: \begin{fmfgraph}(20,20)
281: \fmfset{arrow_len}{3mm}
282: \fmfleftn{l}{2}
283: \fmfrightn{r}{2}
284: \fmf{fermion}{l1,v,l2}
285: \fmf{fermion}{r1,v,r2}
286: \fmfdot{v}
287: \end{fmfgraph}}
288: $$
289: \caption{Effective four-quark interaction induced by the exchange of the scalar
290: color-octet glueball.}
291: \label{Fig:glueball_induced_interaction}
292: \end{figure}
293:
294: We use the method outlined in Section \ref{Sec:NJL_model}. Anticipating the
295: color-sextet condensate, we split the full Lagrangian \eqref{NJLlagrangian} in
296: such a~way that the free part, which determines the propagator, reads
297: $$
298: \LA_{\text{free}}=\bar\psi(\imag\slashed\partial-m+\mu\gamma_0)\psi+
299: \frac12\bar\psi\Delta(C\gamma_5)\tr{\bar\psi}-
300: \frac12\tr\psi\he\Delta(C\gamma_5)\psi.
301: $$
302: Here $\Delta$ stands for the diagonal matrix of the order parameters.
303:
304: This Lagrangian is conveniently diagonalized with the help of the Nambu--Gorkov
305: notation,
306: $$
307: \Psi=\left(
308: \begin{array}{c}
309: \psi \\ \tr{\bar\psi}
310: \end{array}\right).
311: $$
312: We find, for each color $i$, two types of fermionic quasiparticles -- a
313: quark-like and an antiquark-like -- whose dispersion relations are
314: $$
315: E^2_{i\pm}(\vek k)=\left(\sqrt{\vek k^2+m^2}\pm\mu\right)^2+\abs{\Delta_i}^2.
316: $$
317:
318: In the mean-field approximation the gaps $\Delta_i$ are determined by the
319: requirement of the cancelation of the one-loop corrections. We obtain three
320: separate but identical gap equations. Integrating over the frequency and
321: regulating the three-dimensional integral with a~cutoff $\Lambda$ they read, at
322: finite temperature $T$,
323: $$
324: 1=\frac23G\int^{\Lambda}\frac{\dthree\vek k}{(2\pi)^3}\left(\frac1{E_+(\vek k)
325: }\tanh\frac{E_+(\vek k)}{2T} +\frac1{E_-(\vek k)}\tanh\frac{E_-(\vek k)
326: }{2T}\right).
327: $$
328:
329: Several remarks to this result are in order. First, in its derivation we have
330: not been entirely self-consistent. We compared the terms of the same structure,
331: $\bar\psi\Delta(C\gamma_5)\tr{\bar\psi}$, in the free Lagrangian and the
332: one-loop correction. The presence of the chemical potential induces, however,
333: a~similar term $\bar\psi\Delta(C\gamma_5)\gamma_0\tr{\bar\psi}$ at one loop,
334: and this has been neglected. Since the full Lorentz invariance is broken by the
335: chemical potential, it is natural that such a~term appears. To be fully
336: self-consistent, we would have to include such operators into our Lagrangian
337: from the very beginning and solve a~coupled set of gap equations for their
338: coefficients. Such an analysis was done in Ref. \cite{Buballa:2001wh}.
339:
340: Second, note that we derived three identical gap equations for the order
341: parameters $\Delta_1,\Delta_2,\Delta_3$. Since the integrands in the gap
342: equation are monotonic in $\Delta$, there is obviously only one nonzero
343: solution and thus all the gaps acquire the same value. This means that the
344: four-quark interaction we chose prefers the $\mathrm{SO(3)}$ symmetric phase
345: discussed above. This might, however, be just an artifact of the mean-field
346: approximation. Indeed, the separation of the three colors occurs only at the
347: one-loop level. The physical picture is such that the quarks of any individual
348: color generate a~mean field which is in turn felt only by the quarks of the
349: same color. It is then not surprising that all the three gaps have equal size.
350: At two or more loops the colors start to mix and this might lead to lifting the
351: degeneracy and splitting of the gaps. As shown above, if this happens the color
352: $\mathrm{SU(3)}$ invariance is completely broken. A~definite answer may be
353: given only after a~more sophisticated approximation is employed.
354:
355:
356: \section{Two-color QCD: Chiral perturbation theory}
357: We have already mentioned that realistic QCD calculations from first principles
358: are not available at moderate baryon density because of the large coupling
359: constant. The trouble is that neither are the lattice simulations. The reason
360: is that the Euclidean Dirac operator, $\mathcal
361: D=\gamma_{\nu}(\de_{\nu}-A_{\nu})+m-\mu\gamma_0$, is complex at nonzero baryon
362: chemical potential $\mu$.
363:
364: This gave rise to interest in QCD-\emph{like} theories that do not have the
365: sign problem \cite{Kogut:1999iv}. There are two distinguished classes of such
366: theories -- QCD with quarks in the adjoint representation of $\mathrm{SU(3)}$
367: and two-color QCD \cite{Kogut:2000ek}. In the following, we shall consider the
368: latter case.
369:
370: It turns out that the determinant of the Euclidean Dirac operator of two-color
371: QCD, defining the path-integral measure for the gauge bosons, is in general
372: just real. In order for it to be positive, there must be an even number of
373: quarks with the same quantum numbers \cite{Splittorff:2000mm}. Therefore, the
374: case of an even number of flavors is usually studied.
375:
376:
377: \subsection{Symmetry}
378: The key feature of the two-color QCD is the pseudoreality of the gauge group
379: generators, the Pauli matrices, $T_k^*=-T_2T_kT_2$. Assuming the quarks in the
380: fundamental (doublet) representation of the gauge $\mathrm{SU(2)}$, the
381: right-handed component of the Dirac spinor, $\psi_R$ (color and flavor indices
382: are suppressed), may be traded for the left-handed spinor
383: $\tilde\psi_R=\sigma_2T_2\psi_R^*$, the Pauli matrices $\sigma_k$ acting in the
384: Dirac space. The conjugate left-handed spinor has the same transformation
385: properties as $\psi_L$ and is used to replace the conventional Dirac spinor
386: with
387: $$
388: \Psi=\left(\begin{array}{c}
389: \psi_L \\
390: \tilde\psi_R
391: \end{array}\right).
392: $$
393:
394: The Euclidean Lagrangian of massive two-color QCD at finite chemical potential
395: thus becomes
396: \begin{equation}
397: \LA=\imag\he\Psi\sigma_{\nu}(D_{\nu}-
398: \Omega_{\nu})\Psi-m\left[\tfrac12\tr\Psi\sigma_2T_2M\Psi+\text{H.c.}\right].
399: \label{micro_Lagrangian}
400: \end{equation}
401: Now $D_{\nu}$ is the gauge-covariant derivative and $\Omega_{\nu}$ is the
402: constant external field that accounts for the effects of the chemical
403: potential. Finally, $M$ is the block matrix in the $\Psi$ space,
404: $$
405: M=\left(\begin{array}{rc}
406: 0 & 1\\
407: -1 & 0
408: \end{array}\right).
409: $$
410:
411: Using the new spinor $\Psi$ it is easily seen that instead of the naively
412: expected chiral $\mathrm{SU(N_f)_L\times SU(N_f)_R}$ symmetry, the Lagrangian
413: \eqref{micro_Lagrangian} is, in the chiral limit $m=0$ and at $\Omega_{\nu}=0$,
414: invariant under an extended group $\mathrm{SU(2N_f)}$. At zero chemical
415: potential, this symmetry is broken by the standard chiral condensate down to
416: its $\mathrm{Sp(2N_f)}$ subgroup \cite{Kogut:2000ek}.
417:
418: In the $\Psi$ notation, the standard chiral transformations correspond to
419: independent unitary rotations of the upper and lower components $\psi_L$ and
420: $\tilde\psi_R$, respectively. The new transformations in the extended group
421: $\mathrm{SU(2N_f)}$ mix these and thus break the baryon number. In terms of the
422: order parameters, these transformations rotate the chiral condensate
423: $\langle\bar\psi\psi\rangle$ into the diquark condensate
424: $\langle\psi\psi\rangle$.
425:
426: It is therefore not surprising that the chemical potential term breaks the
427: $\mathrm{SU(2N_f)}$ down to the conventional chiral subgroup
428: $\mathrm{SU(N_f)_L\times SU(N_f)_R\times U(1)_B}$. The reason is that it lifts
429: the degeneracy between the particles and antiparticles, and the transformations
430: breaking the baryon number $\mathrm{U(1)_B}$ therefore no longer leave the
431: Lagrangian invariant.
432:
433: Unlike the case of the real, three-color QCD, the two-color QCD has the
434: remarkable property that two quarks may form a~color-singlet state. This is
435: again connected to the pseudoreality of the fundamental representation of the
436: gauge group. It follows that the ordered phase with quarks Cooper-paired should
437: not be called color-superconducting, but rather just superfluid.
438:
439: On the technical level, this fact has the far-reaching consequence that the
440: superfluidity of two-color QCD\footnote{One should carefully distinguish the
441: Bose--Einstein condensation of Goldstone bosons with the quantum numbers of the
442: diquark, from the Cooper pairing of quarks near the Fermi sea. Both effects
443: result in the baryon number superfluidity, but while the former occurs in the
444: confined phase, the latter arises from the pairing interaction between
445: deconfined quarks. The nice feature of two-color QCD is that the diquark
446: condensate may be used as an order parameter in both the confined and the
447: deconfined regime.} may be investigated within the framework of the chiral
448: perturbation theory. The effective Lagrangian is constructed on the coset space
449: $\mathrm{SU(2N_f)/Sp(2N_f)}$. This effective theory has been investigated to
450: great detail, including both the loop \cite{Splittorff:2001fy} and finite
451: temperature \cite{Splittorff:2002xn} effects.
452:
453: The Goldstone bosons are, as usual, generated from the ground state by
454: space\-time-dependent symmetry transformations. In this case, they are
455: parametrized by an antisymmetric unimodular unitary matrix $\Sigma$. The
456: leading-order low-energy effective Lagrangian reads
457: \begin{equation}
458: \LA_{\text{eff}}=\frac{F^2}2\Tr(\nabla_{\nu}\Sigma\nabla_{\nu}\he\Sigma)-
459: G\re\Tr(J\Sigma),
460: \label{eff_LagrangianCHPT}
461: \end{equation}
462: where the $\nabla$'s denote the covariant derivatives,
463: $$
464: \nabla_{\nu}\Sigma=\de_{\nu}\Sigma-(\Omega_{\nu}\Sigma+\Sigma\tr\Omega_{\nu}),\quad
465: \nabla_{\nu}\he\Sigma=\de_{\nu}\he\Sigma+(\he\Sigma\Omega_{\nu}+\tr\Omega_{\nu}\he\Sigma),
466: $$
467: and $J$ is a~source field for $\Sigma$. When the quark mass is included, the
468: Goldstone bosons acquire nonzero mass $m_{\pi}$ which is related to the quark
469: mass $m$ by the Gell-Mann--Oakes--Renner relation
470: $$
471: mG=F^2m_{\pi}^2.
472: $$
473:
474: In the following we shall concentrate on the simplest case $\mathrm{N_f}=2$.
475: Here one can take advantage of the Lie algebra isomorphisms
476: $\mathrm{SU(4)\simeq SO(6)}$ and $\mathrm{Sp(4)\simeq SO(5)}$. We shall argue
477: that it is more convenient to describe the low-energy effective theory on the
478: coset space $\mathrm{SO(6)/SO(5)}$.
479:
480:
481: \subsection{Coset space}
482: The coset $\mathrm{SU(4)/Sp(4)}$ is parametrized by the antisymmetric
483: unimodular unitary matrix $\Sigma$, while the coset $\mathrm{SO(6)/SO(5)}$
484: corresponds to the unit sphere $S^5$ i.e., it is described by a~unit vector
485: $\vek n$ in the six-dimensional Euclidean space. The mapping between these two
486: formalisms is provided by the relation
487: $$
488: \Sigma=n_i\Sigma_i,
489: $$
490: where $\Sigma_i$ are a~set of six conveniently chosen matrices, satisfying the
491: identity $\he\Sigma_i\Sigma_j+\he\Sigma_j\Sigma_i=2\delta_{ij}$. One particular
492: realization of the basis matrices is given by
493: \begin{equation*}
494: \begin{split}
495: \Sigma_1=\left(
496: \begin{array}{cr}
497: 0 & -1\\
498: 1 & 0
499: \end{array}\right),\quad
500: \Sigma_2=\left(
501: \begin{array}{cc}
502: \tau_2 & 0\\
503: 0 & \tau_2
504: \end{array}\right),\quad
505: \Sigma_3=\left(
506: \begin{array}{cc}
507: 0 & \imag\tau_1\\
508: -\imag\tau_1 & 0
509: \end{array}\right),\\
510: \Sigma_4=\left(
511: \begin{array}{cc}
512: \imag\tau_2 & 0\\
513: 0 & -\imag\tau_2
514: \end{array}\right),\quad
515: \Sigma_5=\left(
516: \begin{array}{cc}
517: 0 & \imag\tau_2\\
518: \imag\tau_2 & 0
519: \end{array}\right),\quad
520: \Sigma_6=\left(
521: \begin{array}{cc}
522: 0 & \imag\tau_3\\
523: -\imag\tau_3 & 0
524: \end{array}\right).
525: \end{split}
526: \end{equation*}
527:
528: This particular choice of the basis is not accidental. The first three matrices
529: have been used in literature to denote the chiral, diquark, and isospin
530: condensate, respectively \cite{Kogut:2000ek,Splittorff:2000mm}. The physical
531: nature of the individual matrices is made more transparent by assigning to them
532: quark bilinears,
533: $$
534: \Sigma\to\tfrac12\Psi^T\sigma_2T_2\Sigma\Psi+\text{H.c.},
535: $$
536: that provide the interpolating fields for the Goldstone bosons correspondingly.
537:
538: Concretely, we find that $\Sigma_2$ and $\Sigma_4$ are real and imaginary parts
539: of an isospin singlet with baryon number $+1$, the diquark. Further,
540: $\Sigma_3,\Sigma_5,\Sigma_6$ form an isospin triplet with no baryon charge --
541: the pion. Finally, $\Sigma_1$ corresponds to the isospin singlet with no baryon
542: charge i.e., the $\sigma$ field,
543: \begin{gather*}
544: \Sigma_2\to-\tfrac12\psi^TC\gamma_5T_2\tau_2\psi+\text{H.c.},\quad
545: \Sigma_4\to-\tfrac12\imag\psi^TC\gamma_5T_2\tau_2\psi+\text{H.c.},\\
546: \Sigma_3\to-\imag\bar\psi\tau_1\gamma_5\psi,\quad
547: \Sigma_5\to\imag\bar\psi\tau_2\gamma_5\psi,\quad
548: \Sigma_6\to-\imag\bar\psi\tau_3\gamma_5\psi,\\
549: \Sigma_1\to\bar\psi\psi.
550: \end{gather*}
551:
552:
553: \subsection{Effective Lagrangian}
554: We shall now rewrite the effective Lagrangian in terms of the unit vector $\vek
555: n$. The baryon number chemical potential $\mu$ is incorporated in terms of the
556: external field $\Omega_{\nu}=\delta_{\nu0}\mu B$, where the baryon number
557: generator is represented by the block matrix
558: $$
559: B=\frac12\left(
560: \begin{array}{cr}
561: 1 & 0\\
562: 0 & -1
563: \end{array}\right).
564: $$
565:
566: Adjusting the source $J$ to reproduce the quark mass effect, the leading-order
567: Lagrangian \eqref{eff_LagrangianCHPT} becomes
568: \begin{equation}
569: \LA_{\text{eff}}=2F^2(\de_{\nu}\vek n)^2+4\imag
570: F^2\mu(n_2\de_0n_4-n_4\de_0n_2)-2F^2\mu^2(n_2^2+n_4^2)-4F^2m_{\pi}^2n_1.
571: \label{bilinear_effective_Lagrangian}
572: \end{equation}
573:
574: To determine the spectrum of the theory for a~particular value of the chemical
575: potential, one has to find the ground state by minimizing the static part of
576: the Lagrangian, and then expand the Lagrangian about the minimum to second
577: order in the fields.
578:
579:
580: \subsubsection{Normal phase}
581: For $\mu<m_{\pi}$ the static Lagrangian is minimized by the conventional chiral
582: condensate i.e., $\vek n=(1,0,0,0,0,0)$. The five independent degrees of
583: freedom may be identified with $n_2,\dotsc,n_6$, and the resulting dispersion
584: relations are
585: \begin{align*}
586: E(\vek k)&=\sqrt{\vek k^2+m_{\pi}^2}&&\text{pion triplet $n_3,n_5,n_6$},\\
587: E(\vek k)&=\sqrt{\vek k^2+m_{\pi}^2}-\mu&&\text{diquark $n_2+\imag n_4$},\\
588: E(\vek k)&=\sqrt{\vek k^2+m_{\pi}^2}+\mu&&\text{antidiquark $n_2-\imag n_4$}.
589: \end{align*}
590:
591: This result is exactly what we would expect. The pion triplet carries no baryon
592: charge so its dispersion relation is not affected at all by the chemical
593: potential. The dispersions of the diquark and antidiquark are split and the gap
594: of the diquark is getting smaller until it eventually vanishes at
595: $\mu=m_{\pi}$. At this point the Bose--Einstein condensation sets, breaking the
596: baryon number spontaneously. The diquark is the corresponding Goldstone boson.
597:
598:
599: \subsubsection{Bose--Einstein condensation phase}
600: When $\mu>m_{\pi}$, the vacuum condensate is given by $\vek
601: n=(\cos\alpha,\sin\alpha,0,0,0,0)$, where $\cos\alpha=m_{\pi}^2/\mu^2$. In the
602: excitation spectrum we again find the pion triplet, but now with the dispersion
603: $E(\vek k)=\sqrt{\vek k^2+\mu^2}$. Finally, there are two excitations, the
604: mixtures of the (anti)diquark and $\sigma$, whose dispersion relations are
605: $$
606: E^2_{\pm}(\vek k)=\vek
607: k^2+\frac{\mu^2}2(1+3\cos^2\alpha)\pm\frac{\mu}2\sqrt{\mu^2(1+3\cos^2\alpha)^2+16\vek
608: k^2\cos^2\alpha}.
609: $$
610: Note that the gap of the `$-$' solution vanishes so that this is the Goldstone
611: boson of the spontaneously broken symmetry. In accordance with the general
612: discussion in Chapter \ref{Chap:GBcounting}, its dispersion relation is linear
613: at low momentum.
614:
615: Our calculations confirm the results achieved previously in literature
616: \cite{Kogut:2000ek,Splittorff:2000mm}. The notable advantage of the
617: $\mathrm{SO(6)/SO(5)}$ formalism presented here is that it allows
618: a~straightforward physical interpretation of the various modes, being the
619: linear combinations of $n_1,\dotsc,n_6$ whose quantum numbers are well known.
620:
621: In particular, it turns out that the quantum numbers of the Goldstone boson in
622: the Bose--Einstein condensed phase change as the chemical potential increases.
623: Just at the phase transition point, it is the diquark, matching continuously
624: the diquark mode in the normal phase. On the other hand, in the extreme limit
625: $\mu\gg m_{\pi}$, it is just $n_4$, the imaginary part of the diquark. It is
626: now the linear combination of the diquark and the antidiquark and thus carries
627: no definite baryon number. This is, of course, hardly surprising since the
628: baryon number is spontaneously broken and hence is not a~good quantum number
629: anymore.
630: