1: \documentclass[prl,showpacs,twocolumn,amsmath,amssymb]{revtex4}
2: %\documentclass[aps,prl,preprint,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
3: \usepackage{graphicx}% Include figure files
4: \usepackage{bm}% bold math
5:
6: \def\br{\mathbf{r}}
7: \def\bk{\mathbf{k}}
8: \def\bq{\mathbf{q}}
9: \def\bQ{\mathbf{Q}}
10: \def\bR{\mathbf{R}}
11: \def\im{\mathrm{Im}\,}
12: \def\re{\mathrm{Re}\,}
13:
14: \begin{document}
15:
16: \title{Nonuniform mixed-parity superfluid state in Fermi gases}
17:
18: \author{K. V. Samokhin$^{1}$ and M. S. Mar'enko$^{2}$ }
19:
20: \affiliation{$^{1}$ Department of Physics, Brock University,
21: St.Catharines, Ontario, Canada L2S 3A1\\
22: $^{2}$Department of Physics and Astronomy, Hofstra University,
23: Hempstead, New York 11549, USA}
24: \date{\today}
25:
26:
27: \begin{abstract}
28: We study the effects of dipole interaction on the superfluidity in
29: a homogeneous Fermi gas with population imbalance. We show that
30: the Larkin-Ovchinnikov-Fulde-Ferrell phase is replaced by another
31: nonuniform superfluid phase, in which the order parameter has a
32: nonzero triplet component induced by the dipole interaction.
33: \end{abstract}
34:
35: \pacs{74.20.Rp, 74.20.-z, 03.75.Ss}
36:
37: \maketitle
38:
39: Bardeen-Cooper-Schrieffer (BCS) theory is the standard model of
40: superconductivity and superfluidity in fermionic systems. The
41: central concept of the BCS model is that of a Cooper pair composed
42: of fermions with opposite momenta and spins. If there is a
43: mismatch between the Fermi surfaces of spin-up and spin-down
44: fermions, e.g. because of the Zeeman splitting in magnetic field,
45: then the formation of the Cooper pairs costs energy, which results
46: in the suppression of the critical temperature. It was shown by
47: Larkin and Ovchinnikov \cite{LO64} and Fulde and Ferrell
48: \cite{FF64} (LOFF) that the pair-breaking effect of the Fermi
49: surface splitting can be reduced if the Cooper pairs have a
50: nonzero center-of-mass momentum. The resulting nonuniform
51: superconducting state turns out to be more favorable at low
52: temperatures than the uniform BCS state.
53:
54: Due to the sensitivity of the nonuniform state to disorder and the
55: orbital effects, the attempts to find it in superconductors have
56: been unsuccessful (at least until the recent observation of the
57: features consistent with the LOFF state in the phase diagram of
58: the heavy-fermion compound CeCoIn$_5$ \cite{CeCoIn5}). The recent
59: surge of interest to the nonuniform superconducting and superfluid
60: states, see e.g. Refs. \cite{Comb01,MMI05,SR06,SMPM05,SM06,KJT06},
61: has been stimulated by the experimental progress in ultracold
62: atomic Fermi gases, such as ${}^6$Li and ${}^{40}$K. The
63: Fermi-surface splitting in these systems is due to the populations
64: of atoms in different hyperfine states being unequal
65: \cite{Zwier06,Part06}. When a pairing interaction between the
66: fermions is turned on, the system becomes formally equivalent to a
67: neutral and perfectly clean superconductor in a Zeeman field.
68:
69: In this Letter we consider the effects of the long-range and
70: anisotropic interaction between atomic dipole moments in a Fermi
71: gas with population imbalance. The magnetic dipole interaction is
72: quite small for alkali atoms but can be considerably enhanced to
73: become experimentally observable for atoms with large dipole
74: moments \cite{Stuh05}. Another possibility is to control the
75: magnitude of electric dipole moments by applying external electric
76: field, as suggested in Ref. \cite{MY98}. In either case, we assume
77: that the dipole interaction represents a correction to the
78: dominant isotropic $s$-wave pairing. We will show that increasing
79: the population imbalance produces a novel nonuniform mixed-parity
80: (NMP) superfluid state, which has higher critical temperature than
81: the LOFF state.
82:
83: We consider a homogeneous Fermi gas consisting of two different
84: species of atoms of equal mass $m$. For instance, in ${}^6$Li one
85: can have a mixture of the hyperfine states
86: $|+\rangle\equiv|m_s=1/2,m_i=1\rangle$ and
87: $|-\rangle\equiv|m_s=1/2,m_i=0\rangle$ \cite{SHSH96}, where $m_s$
88: and $m_i$ are the electron and nuclear spin projections
89: respectively. The free-particle Hamiltonian has the form
90: $H_0=\sum_{\bk}(\xi_{\bk}\delta_{\alpha\beta}-h\sigma_{3,\alpha\beta})
91: c^\dagger_{\bk\alpha}c_{\bk\beta}$, where
92: $\xi_{\bk}=(\bk^2-k_F^2)/2m$, $k_F$ is the Fermi momentum, $h$ is
93: the Fermi surface splitting due to unequal particle concentrations
94: in the states labelled by $\alpha,\beta=\pm$, and $\sigma_3$ is
95: the Pauli matrix (we set $\hbar=k_B=1$). We assume that the atomic
96: dipoles are fully polarized, e.g. by an external magnetic field
97: along the $z$-axis: $\bm{\mu}=\mu_d\hat z$. The interaction
98: between two atoms at distance $\bR$ is
99: \begin{equation}
100: \label{V-int}
101: U(\bR)=-\lambda_0\delta(\bR)+\mu_d^2\frac{1-3\hat R_z^2}{R^3}.
102: \end{equation}
103: where the first term describes a short-range attraction with the
104: coupling constant $\lambda_0=4\pi|a|/m>0$, and the second term is
105: the dipole interaction. The $s$-wave scattering length $a<0$
106: depends on the field, diverging in the strong-coupling regime near
107: the Feshbach resonance. The effects of the dipole interaction
108: (\ref{V-int}) have been extensively studied for Bose gases, see
109: e.g. Ref. \cite{YY04} and the references therein.
110:
111: In the BCS regime the pairing Hamiltonian, which takes into
112: account the possibility of the Cooper pairs having a nonzero
113: momentum $\bq$, can be written in the following form:
114: \begin{eqnarray}
115: \label{H-int}
116: H_{int}&=&\frac{1}{2}\sum_{\bk,\bk',\bq}V(\bk,\bk')
117: \Gamma_{\alpha\beta\gamma\delta}\nonumber\\
118: &&\times c^\dagger_{\bk+\frac{\bq}{2},\alpha}c^\dagger_{-\bk+\frac{\bq}{2},\beta}
119: c_{-\bk'+\frac{\bq}{2},\gamma}c_{\bk'+\frac{\bq}{2},\delta},
120: \end{eqnarray}
121: where $V(\bk,\bk')=-\lambda_0+4\pi
122: \mu_d^2(k_z-k'_z)^2/|\bk-\bk'|^2$ is the Fourier transform of Eq.
123: (\ref{V-int}) \cite{contact}. The matrix $\Gamma$ has the
124: following nonzero elements:
125: $\Gamma_{++++}=\Gamma_{+--+}=\Gamma_{-++-}= \Gamma_{----}=1$,
126: which is due to the fact that the pairing interaction depends only
127: on the total particle densities $n(\br)=n_+(\br)+n_-(\br)$.
128:
129: Treating the Hamiltonian (\ref{H-int}) in the mean-field
130: approximation, one arrives at the self-consistency equation which
131: contains a divergent momentum integral due to the absence of
132: ultraviolet cutoff in $V(\bk,\bk')$. This divergence is removed by
133: the renormalization of the interaction potential \cite{GMB61}, in
134: which the bare interaction is replaced by the two-particle
135: scattering amplitude, and an effective cutoff $\omega_c$, of the
136: order of the Fermi energy $\epsilon_F$, is introduced. We use a
137: BCS-like sharp cutoff and represent the pairing interaction in a
138: factorized form:
139: $V(\bk,\bk')=-\theta(|\xi_{\bk}|-\omega_c)\theta(|\xi_{\bk'}|-\omega_c)
140: \sum_i\lambda_i\phi_i(\hat\bk)\phi^*_i(\hat\bk')$. The symmetry
141: factors $\phi_i(\hat\bk)$ are the eigenfunctions and the coupling
142: constants $-\lambda_i$ the eigenvalues of the operator $\hat V$
143: defined by the kernel $V(k_F\hat\bk,k_F\hat\bk')$. The symmetry
144: factors are even (odd) for the singlet (triplet) pairing and
145: satisfy the orthonormality condition:
146: $\langle\phi^*_i(\hat\bk)\phi_j(\hat\bk)\rangle=\delta_{ij}$,
147: where the angular brackets denote the average over the spherical
148: Fermi surface.
149:
150: The effect of the dipole interaction on superfluidity is two-fold:
151: in addition to breaking the rotational symmetry of the system and
152: introducing gap anisotropy, it also leads to the possibility of
153: triplet pairing. We include one singlet and one triplet pairing
154: channel with the highest critical temperatures. One can show that
155: the lowest eigenvalues of $\hat V$ in both cases are achieved for
156: the eigenfunctions that do not depend on the azimuthal angle:
157: $\phi(\hat\bk)=f(\cos\theta)$, where the function $f(s)$ satisfies
158: the integral equation $\int_{-1}^1ds'K(s,s')f(s')=-\lambda f(s)$,
159: with the kernel $K(s,s')=-\lambda_0/2+\pi \mu_d^2|s-s'|$. In the
160: singlet case the solution is $f(s)\propto\cos(\kappa s)$, while in
161: the triplet case $f(s)\propto\sin(\kappa s)$, where
162: $\kappa=\sqrt{2\pi\mu_d^2/\lambda}$. For the singlet pairing, the
163: normalized symmetry factor has the form
164: \begin{equation}
165: \label{phi-even}
166: \phi_s(\hat\bk)=\sqrt{\frac{2}{1+\frac{\sin 2\kappa_s}{2\kappa_s}}}
167: \cos(\kappa_s\cos\theta),
168: \end{equation}
169: where $\kappa_s$ satisfies the equation
170: \begin{equation}
171: \label{kappa-eq}
172: \kappa_s\tan\kappa_s=\frac{2\pi\mu_d^2}{\lambda_0-2\pi\mu_d^2},
173: \end{equation}
174: from which one obtains $\lambda_s=2\pi\mu_d^2/\kappa_s^2$. In the
175: limit of weak dipole interaction, $\mu_d^2\ll\lambda_0$, we find
176: $\phi_s(\hat\bk)\simeq 1+\kappa_s^2(1-3s^2)/6$, and
177: $\lambda_s\simeq\lambda_0-2\pi\mu_d^2$, i.e. the dipole
178: interaction leads to a downward renormalization of the coupling
179: constant in the singlet channel. For the triplet pairing, one can
180: show that the lowest eigenvalue corresponds to $\kappa=\pi/2$, so
181: that
182: \begin{equation}
183: \label{phi-odd}
184: \phi_t(\hat\bk)=\sqrt{2}\sin\left(\frac{\pi}{2}\cos\theta\right),
185: \end{equation}
186: and $\lambda_t=(8/\pi)\mu_d^2$. Note that the coupling constant
187: and the symmetry factor in the triplet channel are not sensitive
188: to the value of $\lambda_0$ and therefore are the same as in the
189: pure dipolar case \cite{BMRS02}.
190:
191: The singlet and triplet contributions to the pairing Hamiltonian
192: (\ref{H-int}) can now be explicitly separated:
193: \begin{equation}
194: \label{H-st}
195: H_{int}=-\lambda_s\sum_{\bq}B_s^\dagger(\bq)B_s(\bq)
196: -\lambda_t\sum_{\bq}\mathbf{B}_t^\dagger(\bq)\mathbf{B}_t(\bq),
197: \end{equation}
198: where
199: $B^\dagger_s(\bq)=(1/2)\sum_{\bk}\phi_s(\hat\bk)(i\sigma_2)_{\alpha\beta}
200: c^\dagger_{\bk+\bq/2,\alpha}c^\dagger_{-\bk+\bq/2,\beta}$ and
201: $\mathbf{B}^\dagger_t(\bq)=(1/2)\sum_{\bk}\phi_t(\hat\bk)(i\bm{\sigma}\sigma_2)_{\alpha\beta}
202: c^\dagger_{\bk+\bq/2,\alpha}c^\dagger_{-\bk+\bq/2,\beta}$ are the
203: pair creation operators. The representation (\ref{H-st})
204: emphasizes the full invariance of the pairing interaction with
205: respect to rotations in the ``pseudospin'' space spanned by the
206: states $|+\rangle$ and $|-\rangle$. Decoupling $H_{int}$ yields
207: the gap function
208: \begin{equation}
209: \label{Delta}
210: \Delta_{\alpha\beta}(\hat\bk,\br)=(i\sigma_2)_{\alpha\beta}
211: \psi(\br)\phi_s(\hat\bk)+(i\bm{\sigma}\sigma_2)_{\alpha\beta}
212: \mathbf{d}(\br)\phi_t(\hat\bk),
213: \end{equation}
214: where the complex scalar $\psi$ and the complex vector
215: $\mathbf{d}$ are the order parameters of the singlet and the
216: triplet pairing respectively.
217:
218: In the absence of population imbalance $h=0$, and the critical
219: temperatures for the singlet and triplet pairing are given by the
220: standard BCS expressions: $T_a\equiv
221: T^{(a)}_{c0}=(2e^\mathbb{C}/\pi)\omega_ce^{-1/N_F\lambda_a}$,
222: where $a=s,t$ and $N_F$ is the density of states at the Fermi
223: surface per one hyperfine state (all three components of
224: $\mathbf{d}$ have the same critical temperature). Of the most
225: interest to us is the limit of weak dipole interaction, in which
226: $T_s\gg T_t$.
227:
228: The phase diagram at $h\neq 0$ can be obtained from the free
229: energy, which is represented as an expansion in powers of the
230: order parameter components: ${\cal F}={\cal F}_0+{\cal
231: F}_2[\psi,\mathbf{d}]+...$ (${\cal F}_0$ is the free energy in the
232: normal state). To find the critical temperature $T_c(h)$ of the
233: second-order phase transition into the superfluid state, or
234: inversely the critical Fermi-surface splitting $h_c(T)$, it is
235: sufficient to keep only the quadratic terms in ${\cal F}$. Using
236: Eq. (\ref{H-st}) we find that the contributions to ${\cal F}_2$
237: from $\psi$ and $d_z$, which describe the pairing between fermions
238: of different species, are decoupled from $d_\pm=(d_x\pm
239: id_y)/\sqrt{2}$, which correspond to the intra-species pairing:
240: \begin{eqnarray}
241: \label{F2}
242: {\cal F}_2=\sum_\bq
243: \biggl[\left(\begin{array}{cc}
244: \psi^*_\bq & d^*_{z,\bq} \\
245: \end{array}\right)
246: \left(\begin{array}{cc}
247: A_{ss} & A_{st} \\
248: A_{ts} & A_{tt} \\
249: \end{array}\right)
250: \left(\begin{array}{c}
251: \psi_\bq \\
252: d_{z,\bq} \\
253: \end{array}\right)\nonumber\\
254: +A_+|d_{+,\bq}|^2+A_-|d_{-,\bq}|^2\biggr].
255: \end{eqnarray}
256: The coefficients in this expression have the following form:
257: $A_{ab}=A_{ba}=N_F[\delta_{ab}\ln(T/T_a)+I_{ab}]$,
258: $A_\pm=N_F[\ln(T/T_t)+I_{tt}(h=0)\pm\delta I]$,
259: \begin{eqnarray}
260: \label{Iab}
261: I_{ab}=
262: \biggl\langle\phi_a(\hat\bk)\phi_b(\hat\bk)\re\Psi\biggl(\frac{1}{2}+
263: \frac{iW}{4\pi T}\biggr)\biggr\rangle
264: -\delta_{ab}\Psi\left(\frac{1}{2}\right),
265: \end{eqnarray}
266: where $\phi_{s,t}(\bk)$ are defined by Eqs. (\ref{phi-even}) and
267: (\ref{phi-odd}), $\Psi(x)$ is the digamma function,
268: $W=v_F\hat\bk\bq-2h$, $v_F$ is the Fermi velocity, and $\delta
269: I\propto N^\prime_Fh$ is proportional to the difference between
270: the densities of states at the Fermi levels for the two fermionic
271: species. Since at $h\neq 0$ the symmetry in the ``pseudospin''
272: space is reduced to rotations about the $z$-axis, different
273: components of $\mathbf{d}$ appear at different temperatures.
274:
275: The critical temperatures for $d_+$ and $d_-$ are found from the
276: equations $A_+=0$ and $A_-=0$ respectively. If one neglects the
277: band asymmetry $N_F'$ then the phase transition is not affected by
278: the population imbalance, otherwise $T_{c,\pm}(h)=T_t\mp O(\delta
279: I)$, so that a nonunitary triplet state with $d_+=0,d_-\neq 0$ is
280: realized.
281:
282: The effect of the Fermi-surface splitting on the other two
283: components of the order parameter, $\psi$ and $d_z$, is more
284: interesting. Because of the presence of the off-diagonal matrix
285: elements in $\hat A$, see Eq. (\ref{Iab}), the singlet and triplet
286: pairing channels can be mixed at $\bq\neq 0$ and $h\neq 0$,
287: producing the NMP state, in which both $\psi$ and $d_z$ are
288: nonzero. The critical temperature is obtained from the equation
289: $\det\hat A(\bq)=0$, after maximization with respect to $\bq$. The
290: calculation is facilitated by the observation that the functions
291: (\ref{Iab}) can be expressed in terms of two dimensionless
292: variables $\bQ=v_F\bq/2h$ and $z=h/2\pi T$. At any given $0\leq
293: z\leq\infty$, we find
294: \begin{equation}
295: T_c(h)=T_s\max\limits_\bQ e^{{\cal I}(\bQ,z)},
296: \end{equation}
297: where
298: \begin{eqnarray*}
299: {\cal I}=-\frac{1}{2}(I_{ss}+I_{tt}+r)+\frac{1}{2}\sqrt{(I_{ss}-I_{tt}-r)^2+4I_{st}^2}.
300: \end{eqnarray*}
301: The critical band splitting is given by $h_c(T)=2\pi zT_c$. If the
302: maximum of ${\cal I}$ is achieved at $\bQ=\bQ_c$, then the wave
303: vector of the superfluid instability is $\bq_c=2h_c\bQ_c/v_F$. The
304: parameter $r=\ln(T_s/T_t)>0$ characterizes the relative strength
305: of pairing in the singlet and triplet channels.
306:
307: In the absence of dipole interaction, $r=\infty$, the pairing is
308: isotropic, and ${\cal
309: I}(\bQ,z)=\Psi(1/2)-\langle\re\Psi(1/2-iz+iz\hat\bk\bQ)\rangle$.
310: At $z\leq z_{\mathrm{LOFF}}\simeq 0.30$, this function has a
311: maximum at $\bQ=0$, therefore the phase transition occurs into the
312: uniform superfluid state. At $z>z_{\mathrm{LOFF}}$, i.e. at
313: $T<T_{\mathrm{LOFF}}\simeq 0.56T_s$, the maximum of ${\cal I}$ is
314: achieved at $|\bQ|\neq 0$, which corresponds to the LOFF state.
315:
316: Since the dipole interaction lifts the spherical degeneracy of the
317: critical temperature, the cases $\bq\perp\hat z$ and
318: $\bq\parallel\hat z$ should be considered separately. In our
319: numerical analysis we use the following values of the parameters,
320: appropriate for the weak dipole interaction in the BCS limit:
321: $(N_F\lambda_0)^{-1}=1.0$ and $\mu_d^2/\lambda_0=0.1$, which gives
322: $T_t\simeq 0.02T_s$ and $r\simeq 3.93$.
323:
324: For $\bq=q\hat z$, we find that at $z\leq z_{\mathrm{NMP}}\simeq
325: 0.23$ the maximum of ${\cal I}$ is at $\bQ=0$, therefore
326: $I_{st}=0$ and the phase transition occurs into the uniform
327: singlet state. In contrast, at $z>z_{\mathrm{NMP}}$, i.e. at
328: $T<T_{\mathrm{NMP}}\simeq 0.68T_s$, the critical temperature has
329: two degenerate maxima at $\bQ=\pm Q_c\hat z\neq 0$, which
330: corresponds to the phase transition into the state
331: \begin{equation}
332: \label{NMP}
333: \left(\begin{array}{c}
334: \psi \\
335: d_z \\
336: \end{array}\right)=
337: \eta_1
338: \left(\begin{array}{c}
339: 1 \\
340: \rho \\
341: \end{array}\right)e^{iq_cz}+
342: \eta_2
343: \left(\begin{array}{c}
344: 1 \\
345: -\rho \\
346: \end{array}\right)e^{-iq_cz}.
347: \end{equation}
348: Here $\rho=-I_{st}/[\ln(T_c/T_t)+I_{tt}]$, and the weights
349: $\eta_{1,2}$ of the two plane-wave components are determined by
350: minimizing the full nonlinear free energy ${\cal F}$, see below.
351: For $\bq\perp\hat z$, there is no singlet-triplet mixing terms,
352: and at $T<T_{\mathrm{LOFF}}$ one obtains the LOFF state with
353: $\psi$ modulated in the equatorial plane and $d_z=0$. The
354: transition line for this state is determined by the maximum of
355: ${\cal I}(\bQ,z)=-I_{ss}$, which yields the critical temperature
356: slightly higher than in the isotropic LOFF case but still lower
357: than in the NMP case.
358:
359:
360: \begin{figure}
361: \includegraphics[width=7.2cm]{PhaseDiagram}
362: \caption{Critical temperature vs Fermi-surface splitting
363: in a homogeneous Fermi gas with dipole interaction.
364: The solid line corresponds to the transition into
365: the NMP state with $\bq\parallel\hat z$, the dashed line corresponds to the
366: LOFF state with $\bq\perp\hat z$, and the short-dashed line is the instability line
367: of the uniform BCS singlet state. At the lowest temperatures, $T<T_t$, the nonunitary
368: triplet state (TSC) is realized.}
369: \label{Phase Diagram}
370: \end{figure}
371:
372:
373: Thus we come to the conclusion that it is the NMP state
374: (\ref{NMP}) that has the highest critical temperature below
375: $T_{\mathrm{NMP}}$, see Fig. \ref{Phase Diagram}. One can show
376: that as $r$ varies from $+\infty$ (no dipole interaction, $T_t=0$)
377: to $0$ (strong dipole interaction, $T_t=T_s$), $T_{\mathrm{NMP}}$
378: moves from $T_{\mathrm{LOFF}}$ towards $T_s$. The parameter $\rho$
379: is the measure of the triplet component admixture, which is zero
380: at $T=T_{\mathrm{NMP}}$ and increases as temperature decreases
381: [for the parameter values used above, $\rho(T=0)\simeq 0.23$].
382:
383: In order to determine the spatial structure of the NMP phase just
384: below the critical temperature, one needs to evaluate the
385: fourth-order terms in the free energy density:
386: \begin{equation}
387: \label{F4}
388: F_4=\beta_1(|\eta_1|^4+|\eta_2|^4)+\beta_2|\eta_1|^2|\eta_2|^2.
389: \end{equation}
390: The coefficients $\beta_{1,2}$ are functions of temperature. The
391: above expression is positive definite if $\beta_1>0$,
392: $\beta_2>-2\beta_1$. At $\beta_2>0$ the minimum is achieved in the
393: state $(\eta_1,\eta_2)\sim(1,0)$ or $(0,1)$, while at $\beta_2<0$
394: one has $(\eta_1,\eta_2)\sim(1,e^{i\varphi})$, where $\varphi$ is
395: an arbitrary phase. Near $T_{\mathrm{NMP}}$ one can set
396: $\rho=q_c=0$ in the expressions for $\beta_{1,2}$, which gives
397: $\beta_1=\beta_2/4=-(N_F\langle\phi_s^4\rangle/32\pi^2T_{\mathrm{NMP}}^2)\re
398: \Psi^{\prime\prime}(1/2-iz_{\mathrm{NMP}})>0$. Therefore, the NMP
399: phase transition in the vicinity of $T_{\mathrm{NMP}}$ is of the
400: second order, with $(\psi,d_z)\sim(1,\rho)e^{iq_cz}$. The
401: determination of the full phase diagram, including the spatial
402: structure of the order parameter at all $T<T_{\mathrm{NMP}}$, is
403: much more difficult, because of the competition between the NMP,
404: LOFF, and uniform BCS phases. We leave this problem for future
405: studies.
406:
407: The origin of the NMP instability can be elucidated using the
408: Ginzburg-Landau expansion of the free energy (\ref{F2}) at small
409: $h$ and $\bq=q\hat z$ in the vicinity of $T_s$. In addition to the
410: usual uniform and gradient terms, the energy density contains
411: mixed-parity terms, which are linear in gradients \cite{helical}:
412: \begin{equation}
413: \label{GL energy}
414: F_2=\psi^*\hat A_s\psi+d_z^*\hat A_td_z+i\tilde
415: Kh(\psi^*\nabla_zd_z+d_z^*\nabla_z\psi),
416: \end{equation}
417: where $\hat A_a=N_F(T-T_a)/T_a-K_a\nabla^2_z$,
418: $K_a=7\zeta(3)N_F\langle\phi_a^2 v_z^2\rangle/16\pi^2T_a^2$, and
419: $\tilde K=7\zeta(3)N_F\langle\phi_s\phi_t v_z\rangle/4\pi^2T_s^2$.
420: Because of the last term, which favors a spatial modulation of the
421: order parameter, at
422: $h>h_{\mathrm{NMP}}=\sqrt{N_FK_s(T_s-T_t)/T_t\tilde K^2}$ the
423: superfluid phase transition is of the NMP type. In contrast, the
424: onset of the singlet LOFF instability is marked by the coefficient
425: $K_s$ changing sign at $h_{\mathrm{LOFF}}>h_{\mathrm{NMP}}$.
426:
427: To summarize, we found that in a homogeneous Fermi gas with
428: $s$-wave attraction, the atomic dipole interaction produces
429: triplet pairs that can mix with the singlet order parameter in a
430: spatially modulated superfluid state. If the densities of atoms in
431: different hyperfine states are unequal then the nonuniform
432: mixed-parity state has higher critical temperature than the LOFF
433: state, even at weak dipole interaction. To relate these findings
434: with the cold atom experiments the effects of the trapping
435: potential should be included. Another interesting open question
436: concerns the fate of the NMP state in the strong-coupling regime
437: near the BCS-BEC crossover.
438:
439: Finally, we would like to remark that the possibility of a
440: nonuniform singlet-triplet mixing is not restricted to polarized
441: Fermi gases. For example, in a fully isotropic system the pairing
442: interaction $V(\bk,\bk')$ in Eq. (\ref{H-int}) depends only on the
443: angle between $\bk$ and $\bk'$ and can therefore be factorized in
444: terms of the spherical harmonics $Y_{lm}(\hat\bk)$. In the singlet
445: $s$-wave channel the order parameter is a scalar $\psi$, with
446: $\phi_s(\hat\bk)=Y_{00}(\hat\bk)=1$, while in the triplet $p$-wave
447: channel the order parameter is now represented by three vectors
448: $\mathbf{d}_m$, with $m=0,\pm 1$, corresponding to
449: $\phi_{t,m}(\hat\bk)=Y_{1m}(\hat\bk)$. When expressed in terms of
450: the Cartesian components of $\hat\bk$, the triplet order parameter
451: becomes a 3$\times$3 complex matrix $A_{i\alpha}$ \cite{VW90}. In
452: the presence of magnetic field $\mathbf{h}$, there is a
453: contribution to the free energy density of the form
454: $ih_i(\psi^*\nabla_\alpha A_{i\alpha}-\mathrm{c.c.})$, which is
455: invariant under all required symmetry operations: orbital and spin
456: rotations, time reversal, and inversion. Similarly to the last
457: term in Eq. (\ref{GL energy}) this can lead to the NMP instability
458: at sufficiently high $h$.
459:
460: This work was supported by the Natural Sciences and Engineering
461: Research Council of Canada.
462:
463:
464:
465: \begin{thebibliography}{99}
466:
467: \bibitem{LO64}
468: A. I. Larkin and Yu. N. Ovchinnikov, Zh. Eksp. Teor. Fiz.
469: \textbf{47}, 1136 (1964) [Sov. Phys. -- JETP \textbf{20}, 762
470: (1965)].
471:
472: \bibitem{FF64}
473: P. Fulde and R. A. Ferrell, Phys. Rev. \textbf{135}, A550 (1964).
474:
475: \bibitem{CeCoIn5}
476: H. A. Radovan, N. A. Fortune, T. P. Murphy, S. T. Hannahs, E. C.
477: Palm, S. W. Tozer, D. Hall, Nature \textbf{425}, 51 (2003).
478:
479: \bibitem{Comb01}
480: R. Combescot, Europhys. Lett. \textbf{55}, 150 (2001).
481:
482: \bibitem{MMI05}
483: T. Mizushima, K. Machida, and M. Ichioka, Phys. Rev. Lett.
484: \textbf{94}, 060404 (2005).
485:
486: \bibitem{SMPM05}
487: A. Sedrakian, J. Mur-Petit, A. Polls, and H. M\"uther, Phys. Rev.
488: A \textbf{72}, 013613 (2005).
489:
490: \bibitem{SR06}
491: D. E. Sheehy and L. Radzihovsky, Phys. Rev. Lett. \textbf{96},
492: 060401 (2006).
493:
494: \bibitem{KJT06}
495: J. Kinnunen, L. M. Jensen, and P. T\"orm\"a, Phys. Rev. Lett.
496: \textbf{96}, 110403 (2006).
497:
498: \bibitem{SM06}
499: K. V. Samokhin and M. S. Mar'enko, Phys. Rev. B. \textbf{73},
500: 144502 (2006).
501:
502: \bibitem{Zwier06}
503: M. W. Zwierlein, A. Schirotzek, C. H. Schunck, and W. Ketterle,
504: Science \textbf{311}, 492 (2006).
505:
506: \bibitem{Part06}
507: G. B. Partridge, W. Li, R. I. Kamar, Y. Liao, and R. G. Hulet,
508: Science \textbf{311}, 503 (2006).
509:
510: \bibitem{Stuh05}
511: J. Stuhler, A. Griesmaier, T. Koch, M. Fattori, T. Pfau, S.
512: Giovanazzi, P. Pedri, and L. Santos, Phys. Rev. Lett. \textbf{95},
513: 150406 (2005).
514:
515: \bibitem{MY98}
516: M. Marinescu and L.You, Phys. Rev. Lett. \textbf{81}, 4596 (1998).
517:
518: \bibitem{SHSH96}
519: H. T. C. Stoof, M. Houbiers, C. A. Sackett, and R. G. Hulet, Phys.
520: Rev. Lett. \textbf{76}, 10 (1996).
521:
522: \bibitem{YY04}
523: S. Yi and L. You, Phys. Rev. Lett. \textbf{92}, 193201 (2004).
524:
525: \bibitem{contact}
526: The Fourier transform of the dipole interaction also contains a
527: constant term, which has been included in the renormalization of
528: $\lambda_0$.
529:
530: \bibitem{GMB61}
531: L. P. Gor'kov and T. K. Melik-Barkhudarov, Zh. Eksp. Teor. Fiz.
532: \textbf{40}, 1452 (1961) [Sov. Phys. -- JETP \textbf{13}, 1018
533: (1961)]; V. N. Popov, \emph{Functional Integrals and Collective
534: Excitations} (Cambridge University Press, 1987).
535:
536: \bibitem{BMRS02}
537: M. A. Baranov, M. S. Mar'enko, V. S. Rychkov, and G. V.
538: Shlyapnikov, Phys. Rev. A \textbf{66}, 013606 (2002).
539:
540: \bibitem{helical}
541: Similar mixed-parity terms were discussed phenomenologically in V.
542: P. Mineev and K. V. Samokhin, Zh. Eksp. Teor. Fiz. \textbf{105},
543: 747 (1994) [JETP \textbf{78}, 401 (1994)].
544:
545: \bibitem{VW90}
546: D. Vollhardt and P. W\"olfle, \emph{The Superfluid Phases of
547: Helium 3} (Taylor and Francis, New York, 1990).
548:
549: \end{thebibliography}
550:
551: \end{document}
552: