1: \documentclass[letterpaper,twocolumn,showpacs,
2: prb,floatfix
3: %superscriptaddress%
4: ]{revtex4}
5:
6: \usepackage{graphicx}
7: \usepackage{amssymb}
8: \usepackage{amsmath,amsfonts,latexsym}
9: \usepackage{array,tabularx,color}
10: \usepackage[normalem]{ulem}
11: %\usepackage{array,tabularx}
12: %\usepackage{dcolumn} % Align table columns on decimal
13: %\usepackage{groupedaddress} % point
14:
15: %\usepackage{hyperref} %don't get compounds citations e.g. [1-5]
16: %\usepackage{comment}
17:
18:
19:
20: \DeclareMathOperator{\tr}{tr}
21: \newcommand{\vect}[1]{\mathbf{#1}}
22: \newcommand*{\cL}{{\cal L}}
23: \newcommand*{\br}{\mathbf{r}}
24: \newcommand*{\bp}{\mathbf{p}}
25: \newcommand*{\bq}{\mathbf{q}}
26: %\newcommand{\fmm}[1]{{\color{red}#1}}
27:
28: \begin{document}
29:
30: \title{Spinodal decomposition in polarised Fermi superfluids}
31:
32: \author{A. Lamacraft}
33: \email{austen@virginia.edu}
34: \affiliation{Rudolf Peierls Centre for Theoretical Physics, 1 Keble
35: Road, Oxford OX1 3NP, UK}
36:
37: \author{F. M. Marchetti}
38: \email{fmm25@cam.ac.uk}
39: \affiliation{Rudolf Peierls Centre for Theoretical Physics, 1 Keble
40: Road, Oxford OX1 3NP, UK}
41:
42: %\date{\today}
43: \date{January 28, 2007} % to fix in the last version!
44:
45:
46: \begin{abstract}
47: We discuss the dynamics of phase separation through the process of
48: spinodal decomposition in a Fermi superfluid with population
49: imbalance. We discuss this instability first in terms of a
50: phenomenological Landau theory. Working within the mean-field
51: description at zero temperature, we then find the spinodal region in
52: the phase diagram of polarisation versus interaction strength, and
53: the spectrum of unstable modes in this region. After a quench, the
54: spinodal decomposition starts from the Sarma state, which is a
55: minimum of the free energy with respect to the order parameter
56: \emph{at fixed density and polarisation} and a maximum at fixed
57: chemical potentials. The possibility of observing non-trivial domain
58: structures in current experiments with trapped atomic gases is
59: discussed.
60: \end{abstract}
61:
62: \pacs{03.75.Kk, 03.75.Ss, 64.75.+g}
63:
64: \maketitle
65:
66: The ordering of matter into different phases is a central
67: preoccupation of many areas of physics, from condensed matter to
68: cosmology. Hand in hand with the \emph{existence} of different phases
69: goes the question of the dynamical processes responsible for their
70: formation, which may be equally important in determining what is
71: observed in a given situation. Recent experimental advances in the
72: creation of degenerate atomic gases have begun to realize the
73: prospect of a rich variety of new phases in atomic matter, involving
74: the hyperfine degrees of freedom, mixtures of different species, or
75: spatial order on optical lattices. With each new phase comes the
76: dynamical issue of how that phase will appear under laboratory
77: conditions. One advantage offered by atomic systems is that the
78: characteristic timescale $\hbar/k_BT$ at nanokelvin temperatures is in
79: the convenient millisecond range.
80:
81: The possibility of tuning interparticle interactions in a controlled
82: manner has proven to be of particular significance
83: lately. Magnetically tuned Feshbach resonances have permitted the
84: experimental investigation of the crossover from a Bose-Einstein
85: condensate (BEC) of diatomic molecules to the
86: Bardeen-Cooper-Schrieffer (BCS) limit of weakly-bound Cooper pairs of
87: fermionic
88: atoms~\cite{regal2004,zwierlein2004,chin2004,bourdel2004,kinast2004,zwierlein2005}.
89:
90: It appears that when the numbers of atoms of the two species
91: undergoing pairing are equal, the system forms a condensate with
92: smoothly varying properties at low temperatures. With unequal numbers
93: (we will call such a system `polarised') there is the possibility of
94: phase separation into a superfluid of low polarisation (favoured by
95: pairing) and a normal fluid of higher
96: polarisation~\cite{zwierlein2006,partridge2006,zwierlein2006_2,shin2006,partridge2006_2,bedaque2003,sheehy2006}.
97:
98: With the occurrence of phase separation in these systems now
99: established, it is crucial to examine the dynamics of this process,
100: and that is the purpose of this Letter. Working within a mean-field
101: approximation, we find that the dome of phase separation at
102: temperatures below the tricritical point~\cite{parish2006,gubbels2006}
103: contains a \emph{spinodal} region where phase separation proceeds via
104: a linear instability. In many recent papers this region is incorrectly
105: identified with the phase coexistence
106: region~\cite{chien2006,iskin2006,pao2006,pao2006b} (see
107: Fig.~\ref{fig:spinodals}).
108: %
109: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
110: \begin{figure}
111: \centering \includegraphics[width=0.46\textwidth]{f-pfinT-mphase_diagram.eps}
112: \caption{(Colour online) Zero temperature mean-field phase diagram for
113: magnetisation $m/n$ versus interaction strength $1/k_Fa$. The
114: phase separated region (PS) can be decomposed into two
115: regions --- the spinodal or unstable region and the
116: metastable region (darker shade) --- divided by the spinodal
117: normal (sp-N) line (dashed) and the spinodal superfluid
118: (sp-SF) line (dot-dashed). In addition, the PS spinodal phase
119: is divided in half by the line where the superfluid density
120: $Q$ is zero (thin solid). The tricritical point (orange
121: circle) is at $m/n=1$ and $1/k_Fa\simeq 2.37$. Inset:
122: schematic finite temperature phase diagram of
123: $T_c/\varepsilon_F$ versus $m/n$ for a fixed interaction
124: strength $1/k_Fa$. Arrows indicate possible quenches into the
125: spinodal region.}
126: \label{fig:spinodals}
127: \end{figure}
128: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
129: %
130:
131: We discuss the dynamics initiated by a quench into the spinodal region
132: by finding the unstable density modes associated with the endpoint of
133: the quench. The modes with the fastest growth rate give the
134: characteristic size of the resulting domains of superfluid and normal
135: fluid. Though our description of these processes will not be
136: quantitatively correct in the crossover region where the system is
137: strongly interacting, we emphasise that this behaviour is a generic
138: feature of systems possessing this type of phase diagram. In
139: particular, our system shares many similarities with the problem of
140: $^{3}$He-$^{4}$He mixtures (see e.g. \cite{gunton,hoffer1986}). We
141: will confine ourselves to the early stages of spinodal decomposition
142: characterized by the exponential growth of unstable modes. The
143: emergence of a coarsening regime at later times, where domains scale
144: with the time since the quench~\cite{bray1994}, is a fascinating
145: possibility that we will leave for future work.
146:
147:
148: The rest of this paper is organized as follows. In the next section we introduce a simple phenomenological model for the polarized system. We use this model to show how, with a simplifying assumption, mean-field calculations within the grand canonical ensemble may be applied to the early stages of spinodal decomposition. By adding dynamical assumptions, the equation satisfied by the sound velocity can be inferred. In section~\ref{sec:micro} we provide a microscopic mean-field calculation of the spectra of unstable modes, which are discussed in some detail in section~\ref{sec:discuss} before we conclude.
149:
150: \section{Phenomenological model}
151:
152: We begin by discussing the phase
153: diagram in phenomenological terms starting from a model free energy
154: depending on the superfluid order parameter $\Delta$ and the density
155: difference $m=n_{\uparrow}-n_{\downarrow}$, following a similar approach to the $^{3}$He-$^{4}$He system~\cite{hohenberg1979}
156: %
157: \begin{multline}
158: f_m (\Delta,m) \equiv \frac{r}{2} |\Delta|^2 + u |\Delta|^4 +
159: v|\Delta|^6 \\ + \frac{1}{2}\chi_n^{-1} m^2+\gamma m|\Delta|^2 \; .
160: \label{landau_model}
161: \end{multline}
162: %
163: The potential for $\Delta$ is the simplest one that can describe a
164: first order transition. We are interested in $r<0$ so that for $m=0$
165: there is always superfluid order $\Delta\neq 0$. Further, the coupling
166: to $m$ has the obvious physical meaning that larger polarisations
167: discourage pairing. We ignore for the moment additional couplings to
168: the total density $n=n_{\uparrow}+n_{\downarrow}$~\footnote{One can in
169: general show that, at the mean-field level, the free
170: energy~\eqref{landau_model} is a function of $m/n$ only, and
171: conversely the grand canonical potential~\eqref{landau_delta} is
172: function of $h/|\mu|$ only.}.
173:
174: Minimising~\eqref{landau_model} on the order parameter, gives the
175: potential
176: %
177: \begin{equation}
178: F(m)\equiv \min_\Delta f_m(\Delta,m) \; .
179: \label{eq:mpot}
180: \end{equation}
181: %
182: Where the transition is first order, the phase coexistence region
183: $(m_{\text{cr-SF}},m_{\text{cr-N}})$ is obtained by the usual tangent
184: construction~\cite{callen}. Inside the coexistence region one
185: identifies a spinodal region $(m_{\text{sp-N}},m_{\text{sp-SF}})$,
186: where the susceptibility $\partial_m^2 F(m)<0$, and phase separation
187: proceeds via the growth of unstable modes. In general this represents
188: an extremely difficult dynamical problem. We will make the simplifying
189: assumption that following a quench into the spinodal region, the order
190: parameter relaxes rapidly do its minimum at fixed $m$, while
191: $m(\vect{r})$, being a conserved quantity, begins to develop
192: inhomogeneities on a much longer timescale. For the case of trapped
193: gases, we will review the validity of this approach \emph{a posteriori}.
194:
195: Often it is convenient, particularly in many body calculations, to
196: work instead with the grand canonical potential $f_h(\Delta,h)$,
197: obtained from $f_m$ in the usual way
198: %
199: \begin{multline}
200: f_h(\Delta,h) \equiv \min_m \left[f_m(\Delta,m) - hm\right] \\
201: =\frac{1}{2}\tilde{r} |\Delta|^2+ \tilde{u}|\Delta|^4 +
202: v|\Delta|^6-\frac{1}{2}\chi_nh^2 \; ,
203: \label{landau_delta}
204: \end{multline}
205: %
206: where $\tilde{r}= r+2\gamma h\chi_n$ and $\tilde{u} =
207: u-\frac{1}{2}\chi_n \gamma^2<0$. If we assume $u>0$, then for small
208: $\gamma$ a second order transition occurs when $\tilde r$ changes
209: sign. Increasing $\gamma$ causes the transition to become first order
210: as $\tilde u$ becomes negative. Minimising $f_h(\Delta,h)$ on $\Delta$
211: gives the thermodynamic free energy $\Omega(h) \equiv \min_\Delta
212: f_h(\Delta,h)$. At some critical $h=h_{\text{cr}}$ there is a
213: discontinuity in the derivative (see Fig.~\ref{fig:spinodal}). In a
214: uniform phase $m=- \partial_h \Omega$ so the boundaries of the phase
215: coexistence region are $m_{\text{cr-N,cr-SF}}=-\partial_h
216: \Omega|_{h_\text{cr}^{\pm}}$. The two phases correspond to the two
217: minima of $f_h(\Delta,h)$: one at $\Delta=0$, the normal phase, and
218: the superfluid phase at finite $\Delta$. One can continue past
219: $h_{\text{cr}}$ on the metastable minimum, rather than the true
220: minimum. Such states are linearly stable, with a positive
221: susceptibility $- \partial_h^2\Omega$ ($= [\partial_m^2 F(m)]^{-1}$),
222: since one may easily see that
223: %
224: \begin{equation}
225: \frac{\partial^2\Omega}{\partial h^2} =
226: \frac{\partial^2f_h}{\partial h^2} -
227: \left(\frac{\partial^2f_h}{\partial \Delta^2}\right)^{-1}
228: \left(\frac{\partial m}{\partial \Delta}\right)^2\; ,
229: \label{sus_rel}
230: \end{equation}
231: %
232: and the first term on the right hand side is $-\chi_n<0$ --- the
233: normal state susceptibility is assumed positive. Since the curvature
234: of $\Omega(h)$ remains negative, $m$ takes values inside the
235: coexistence region but the system remains uniform. This is of course
236: not the equilibrium state of the system, but the other phase must
237: nucleate in order for phase separation to occur.
238:
239: %
240: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
241: \begin{figure}
242: \centering
243: \includegraphics[width=0.46\textwidth]{Sarma-free-en.eps}
244: \caption{(Colour online) Mean-field free energy $\Omega (h/\mu)$ for
245: the Hamiltonian~\eqref{eq:ham} versus $h/\mu$ at unitarity
246: $1/k_Fa=0$ ($\mu>0$) obtained evaluating the free energy
247: density $f_h (\Delta/\mu,h/\mu)$ respectively at the local
248: minimum for $\Delta/\mu\simeq 1.15$ (green), at $\Delta=0$
249: (blue) and at the local maxima (red), i.e. the Sarma
250: state. The value of $h/\mu$ corresponding to the
251: spinodal-normal point (sp-N), the spinodal-superfluid point
252: (sp-SF) and the critical value for phase separation (cr) are
253: indicated. At unitarity, the corresponding values of the
254: magnetisation are respectively $m_{\text{cr-SF}}/n =
255: m_{\text{sp-SF}}/n = 0$, $m_{\text{sp-N}}/n \simeq 0.73$ and
256: $m_{\text{cr-N}}/n \simeq 0.93$ (see
257: Fig.~\ref{fig:spinodals}). Inset: Plots of the free energy
258: density $f_h (\Delta/\mu,h/\mu)$ for the values of $h$
259: corresponding to $h_{\text{sp-N}}$, $h_{\text{sp-SF}}$ and
260: $h_{\text{cr}}$. Note that the thermodynamic free energy
261: corresponding to the Sarma state has a positive curvature
262: indicating instability. The cusp structure shrinks to a point
263: at the tricritical point.}
264: \label{fig:spinodal}
265: \end{figure}
266: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
267: %
268:
269: This situation changes when the metastable minima merge with the
270: maximum, for $h$ corresponding respectively to $m_{\text{sp-N}}$ and
271: $m_{\text{sp-SF}}$ for the metastable superfluid and normal phases. As
272: already discussed, inside the spinodal region we seek the
273: \emph{constrained minimum} of $f_m(\Delta,m)$ with respect to $\Delta$
274: at fixed $m$, and the spectrum of unstable modes about this point. How
275: should this programme be implemented for a many-body calculation that
276: provides instead the grand canonical potential $f_h(\Delta,h)$?
277: Fortunately, one may easily show that the stationary points of $f_m
278: (\Delta,m)$ with respect to $\Delta$ coincide with those of
279: $f_h(\Delta,h)$ \emph{at corresponding values of $h$ and $m$}. This is
280: because $f_m(\Delta,m)$ may be written as
281: %
282: \begin{equation*}
283: f_m(\Delta,m)=f_h(\Delta,h^*(\Delta,m))+h^*(\Delta,m)m \; ,
284: \label{Fform}
285: \end{equation*}
286: %
287: where $h^*(\Delta,m)$ is the value of $h$ that maximises
288: $f_h(\Delta,h)+hm$. Thus $m=-\partial_h\Omega|_{h=h^*}$. We can then
289: easily see that the conditions, $\partial_\Delta f_m(\Delta,m) = 0$
290: and $\partial_\Delta f_h(\Delta,h)=0$ are equivalent when
291: $h=h^*(\Delta,m)$.
292:
293: In the spinodal region the unstable constrained minimum corresponds to
294: a \emph{maximum} of $f_h$, as may be seen from Eq.~\eqref{sus_rel}:
295: $\partial^2\Omega/\partial h^2>0$ requires $\partial^2f_h/\partial
296: \Delta^2<0$. In the context of the mean-field theory for the paired
297: fermion system, this corresponds to the solution of the
298: self-consistent equation in a magnetic field discovered by
299: Sarma~\cite{sarma1963}.
300:
301: These results are readily adapted to the presence of a trap potential
302: $V(\vect{r})$ using the following simple approximation: At each point
303: in space we find the \emph{constrained minimum} of the order parameter
304: for fixed local magnetisation $m(\vect{r})/n(\vect{r})$ corresponding
305: to the density profiles before the quench. Spinodal decomposition
306: therefore occurs where the local magnetisation lies in the spinodal
307: region of the homogeneous phase diagram. Moving out from the centre of
308: the trap corresponds to a vertical trajectory in
309: Fig.~\ref{fig:spinodals} which will be displaced horizontally by a
310: quench in $1/k_Fa$.
311:
312: To make contact with the microscopic analysis of the next section, let us add appropriate dynamics to the above model. Expanding $\Delta_0+\delta\Delta$ around some value we write a phenomenological quadratic Lagrangian for the longitudinal (amplitude) and transverse (phase) modes
313: %
314: \begin{eqnarray}\label{model_L}
315: \cL=B\delta\Delta_L\dot{\delta\Delta_T}+\frac{R}{2}(\dot{\delta\Delta_T})^2-\frac{Q}{2}\left(\nabla\delta\Delta_T\right)^2\nonumber\\-\frac{A}{2}(\delta\Delta_L)^2-2\gamma m\Delta_0\delta\Delta_L.
316: \end{eqnarray}
317: %
318: Eq.~(\ref{model_L}) includes all terms up to second order in the derivatives except for a $(\dot{\delta\Delta_T})^2$ term that will not change the sound velocity. $\delta\Delta_L$ is coupled to the density difference $m$ as specified in the model Eq.~(\ref{landau_model}). $m$ is written in terms of the distribution function of the majority quasiparticle distribution function (at zero temperature there are no minority quasiparticles)
319: %
320: \[m(\br)=\sum_\bp n_\uparrow(\bp,\br),\]
321: %
322: which obeys the Boltzmann equation
323: %
324: \[
325: \left[\partial_t +\vect{v}_F\cdot\nabla_{\br}-2\gamma \Delta_0\nabla_\br\delta\Delta_L\cdot \nabla_\bp\right]n_{\uparrow}(\bp,\br,t)=0.
326: \]
327: %
328: where $\vect{v}_F=v_F\hat\bp$ is the Fermi velocity. The linearized solution is expressed as
329: %
330: \begin{eqnarray*}
331: n_{\uparrow}(\bp,\bq,\omega)=\frac{\bq\cdot\vect{v}_F}{w-\vect{v}_F\cdot\bq}2\gamma \Delta_0\delta\Delta_L(\bq,\omega)\delta(\epsilon_\bp-\mu)\nonumber\\
332: m(\bq,\omega)=2\gamma \Delta_0\nu(\mu)L(\omega/|\bq|)\delta\Delta_L(\bq,\omega)
333: \end{eqnarray*}
334: %
335: where $L(x)=\frac{x}{2}\log\frac{x+1}{x-1}-1$ is the Lindhard function, and $\nu(\mu)$ the Fermi surface density of states. The dispersion relation of the linearized modes is then given by a solution of
336: %
337: \begin{eqnarray*}
338: \left|\begin{array}{ccc}A & iB\omega & 2\gamma\Delta_0 \\-iB\omega & Qq^2-R\omega^2 & 0 \\2\gamma\Delta_0\nu(\mu)L(\omega/|\bq|) & 0 & -1\end{array}\right|=0,
339: \end{eqnarray*}
340: %
341: or
342: \begin{equation}\label{model_dispersion}
343: \left(A+4\gamma^2\Delta_0^2\nu L(c_s)\right)\left(Q-Rc_s^2\right)-B^2c_s^2=0.
344: \end{equation}
345: %
346: We will see that an equation of the same form emerges from the microscopic analysis of the next section, where the solutions will be further analyzed.
347:
348: \section{Microscopic calculation}\label{sec:micro}
349:
350: We turn now to the analysis of the microscopic problem described by
351: the Hamiltonian
352: %
353: \begin{multline}
354: \hat{H} - \sum_{\sigma = \uparrow,\downarrow} \mu_\sigma
355: \hat{n}_{\sigma} = \sum_{\vect{k}\sigma}
356: \left(\epsilon_{\vect{k}} - \mu_{\sigma}\right)
357: c_{\vect{k} \sigma}^\dag c_{\vect{k} \sigma}^{\vphantom{\dag}} \\
358: + \frac{g}{V} \sum_{\vect{k},\vect{k}',\vect{q}}
359: c_{\vect{k}+\vect{q}/2 \uparrow}^\dag c_{-\vect{k}+\vect{q}/2
360: \downarrow}^\dag c_{-\vect{k}'+\vect{q}/2 \downarrow}^{\vphantom{\dag}}
361: c_{\vect{k}'+\vect{q}/2 \uparrow}^{\vphantom{\dag}}\; ,
362: \label{eq:ham}
363: \end{multline}
364: %
365: where $\epsilon_{\vect{k}}=k^2/2m$ (we set $\hbar=1$) and where the
366: scattering length is introduced in the usual way:
367: %
368: \begin{equation}
369: \frac{1}{g} = \frac{m}{4\pi a} - \frac{1}{V} \sum_{\vect{k}}
370: \frac{1}{2\epsilon_{\vect{k}}} \; .
371: \end{equation}
372: %
373: The condition $\partial^2f_h/\partial \Delta^2=0$, corresponding to a
374: divergent susceptibility, was used to obtain the spinodal lines in
375: Fig.~\ref{fig:spinodals}. As explained before, the unstable modes are
376: to be found from the matrix response function (dynamical
377: susceptibility):
378: %
379: \begin{multline}
380: \hat\chi(\vect{r}-\vect{r}',t-t') \\ = -i\begin{pmatrix}
381: \langle\left[n(\vect{r},t),n(\vect{r}',t')\right]\rangle &
382: \langle\left[n(\vect{r},t),m(\vect{r}',t')\right]
383: \rangle\\\langle\left[m(\vect{r},t),n(\vect{r}',t')\right]\rangle &
384: \langle\left[n(\vect{r},t),m(\vect{r}',t')\right]\rangle\end{pmatrix}\; .
385: \label{eq:response}
386: \end{multline}
387: %
388: Finding the spectrum of collective modes requires us to solve the
389: equation
390: %
391: \begin{equation}
392: \det \hat{\chi}^{-1}(\vect{q},\varepsilon (\vect{q})) = 0,
393: \label{det}
394: \end{equation}
395: %
396: which defines the dispersion relation $\varepsilon(\vect{q})$. The
397: unstable modes correspond to $\Im \varepsilon(\vect{q}) > 0$. In
398: practice, the response matrix~\eqref{eq:response} can be found making
399: use of a path integral formulation, by expanding the action
400: %
401: \begin{gather}
402: S[\Delta,\mu,h] = -\frac{1}{g} \int_0^{\beta} d\tau \int d\vect{r}
403: |\Delta|^2 - \tr \ln \hat{\mathcal{G}}^{-1}\\\nonumber
404: \hat{\mathcal{G}}^{-1} =
405: \begin{pmatrix} \partial_{\tau} -\frac{\nabla^2}{2m} -\mu - h & -\Delta
406: \\ -\Delta^* & \partial_{\tau} +\frac{\nabla^2}{2m} +\mu -
407: h\end{pmatrix} \; ,
408: \end{gather}
409: %
410: up to second order in fluctuations of $\mu (\vect{r},\tau) = \mu +
411: \delta \mu (\vect{r},\tau)$, $h (\vect{r},\tau) = h + \delta h
412: (\vect{r},\tau)$ and $\Delta(\vect{r},\tau) = \Delta + \delta
413: \Delta^{L} (\vect{r},\tau) + i \delta \Delta^{T} (\vect{r},\tau)$
414: around their mean-field values. Here, we have introduced $\mu =
415: (\mu_{\uparrow} + \mu_{\downarrow})/2$ and $h = \mu_{\uparrow} -
416: \mu_{\downarrow}$.
417:
418: By completing the squares in
419: $\delta\Delta^{L}$ and $\delta \Delta^{T}$, one can easily obtain:
420: %
421: \begin{multline}
422: \hat{\chi} (\vect{q},i\omega_h) =\begin{pmatrix} \Pi_{\mu\mu} &
423: \Pi_{h \mu} \\\Pi_{\mu h} & \Pi_{h h}\end{pmatrix}\\
424: -\begin{pmatrix}\Pi_{\mu\Delta^L} & \Pi_{\mu \Delta^T} \\\Pi_{h
425: \Delta^L} & \Pi_{h \Delta^T}\end{pmatrix} \hat{\mathcal{D}}
426: \begin{pmatrix}\Pi_{\Delta^L\mu} & \Pi_{\Delta^L h}
427: \\\Pi_{\Delta^T \mu} & \Pi_{\Delta^T h}\end{pmatrix}\; ,
428: \label{mf_resp}
429: \end{multline}
430: %
431: where we have introduced the polarization operators
432: %
433: \begin{equation*}
434: \Pi_{ab} (\vect{q},i\omega_h) = \frac{1}{V\beta}
435: \sum_{\vect{k},i\epsilon_n} \tr \,\tau_a
436: \hat{G}_{\vect{q}+\vect{k},i\epsilon_n+i\omega_h}^{(0)}\tau_b
437: \hat{G}_{\vect{q},i\epsilon_n}^{(0)} \; ,
438: \end{equation*}
439: %
440: with $\tau_\mu=\sigma_3$, $\tau_h=\openone$,
441: $\tau_{\Delta^L}=\sigma_1$, and $\tau_{\Delta^T}=\sigma_2$, and where
442: the order parameter propagator is
443: %
444: \begin{equation}
445: \hat{\mathcal{D}}^{-1} = -\frac{2}{g} \openone + \begin{pmatrix}
446: \Pi_{\Delta^L\Delta^L} & \Pi_{\Delta^L \Delta^T} \\\Pi_{\Delta^T
447: \Delta^L} & \Pi_{\Delta^T \Delta^T}\end{pmatrix} \; .
448: \end{equation}
449: %
450: We omit the explicit expressions for these quantities, as they are
451: straightforward generalisations of the expressions found e.g. in
452: Ref.~\onlinecite{engelbrecht1997} to the case $h\neq 0$. It is clear that
453: Eq.~\eqref{mf_resp} is the generalisation of Eq.~\eqref{sus_rel} to
454: the full matrix response and to nonzero $\vect{q}$ and $\varepsilon$.
455:
456:
457: %
458: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
459: \begin{figure}
460: \centering
461: \includegraphics[width=0.46\textwidth]{comp_unst.eps}
462: \caption{(Colour online) Upper panels: Unstable modes frequency
463: $\omega/\varepsilon_F$ ($\omega(\vect{q}) \equiv
464: -i\varepsilon(\vect{q})$) versus momentum $q/q_F$ for
465: different values of the interaction strength $1/k_Fa$ across
466: the spinodal region and for fixed polarisation
467: $m/n=0.5$. Note that gaps correspond to complex
468: frequencies. Lower panel: Plot of the most unstable mode
469: frequency $\omega/\varepsilon_F$ (plus green), momentum
470: $q/q_F$ (times blue) and of the sound velocity $2c_s/v_F$
471: (solid black) across the spinodal region for fixed
472: polarisation $m/n=0.5$.}
473: \label{fig:unstable}
474: \end{figure}
475: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%
476: %
477:
478: The mean-field approximation to the mode spectrum is the solution of
479: Eq.~\eqref{det}, with the response matrix given by the
480: expression~\eqref{mf_resp}. Here, $\mu$ and $h$ are chosen so that $n$
481: and $m$ take the desired values, and inside the spinodal region,
482: $\Delta$ is taken at the Sarma value corresponding to the maximum of
483: $f_{h} (\Delta,h)$ --- as discussed, the stationary points of $f_{h}
484: (\Delta,h)$ coincide with those of $f_m (\Delta, m)$. Evidently, a
485: sufficient condition for a solution of Eq.~\eqref{det} is the
486: occurrence of a pole at $\varepsilon(\vect{q})$ in the order parameter
487: propagator, so we must solve $\det
488: \hat{\mathcal{D}}^{-1}(\vect{q},\varepsilon(\vect{q}))=0$. In general,
489: a numerical solution is called for. At the spinodal lines, however, a
490: diverging susceptibility implies a vanishing sound velocity, which we
491: can find by expanding in $\varepsilon$ and $\vect{q}$:
492: %
493: \begin{equation}
494: \hat{\mathcal{D}}^{-1} \simeq \begin{pmatrix} A+P(\varepsilon/q) &
495: iB\varepsilon \\-iB\varepsilon & Qq^2-R\varepsilon^2 \end{pmatrix}
496: \; ,
497: \label{eq:expan}
498: \end{equation}
499: %
500: where $A= \frac{1}{V} \sum_{\vect{k}}
501: \Theta(E_\vect{k}-h)\frac{\Delta^2}{2E_\vect{k}^3}$, $ B =
502: \frac{1}{V} \sum_{\vect{k}}
503: \Theta(E_\vect{k}-h)\frac{\epsilon_{\vect{k}} -\mu}{4E_\vect{k}^3}$,
504: $R=\frac{1}{V} \sum_{\vect{k}}
505: \frac{\Theta(E_\vect{k}-h)}{8E_\vect{k}^3}$ and
506: %
507: \begin{gather*}
508: Q = \frac{n}{8m\Delta^2}\left[1-\frac{h}{2
509: \sqrt{h^2-\Delta^2}}\frac{\epsilon_+^{3/2} +\epsilon_-^{3/2}}{\epsilon_F^{3/2}}\right]\\
510: %
511: P(\varepsilon/q) = \frac{\Delta^2}{2h^2} \left[\nu_{+}
512: L(\varepsilon/v_{+} q) + \nu_{-} L(\varepsilon/v_{-} q)\right] \; ,
513: \end{gather*}
514: %
515: with $E_{\vect{k}} = \sqrt{(\epsilon_{\vect{k}} -\mu)^2+\Delta^2}$,
516: and where $\nu_{\pm}=\frac{m \epsilon_{\pm}}{\pi^2}v_{\pm}^{-1}$ and
517: $v_{\pm}=dE_k/dk|_{\epsilon_{\pm}}$ are the density of states and
518: velocity at the two solutions $\epsilon_{\pm}$ of $E_{\vect{k}}=h$
519: (take $\epsilon_{\pm}=0=\nu_{\pm}$ if there is no solution). Note that the phase
520: stiffness $4 m \Delta^2 Q$ is the superfluid density~\cite{pao2006},
521: which changes sign inside the spinodal region (see
522: Fig.~\ref{fig:spinodals}).
523:
524: The higher order terms in the $(1,1)$ entry of
525: Eq.~\eqref{eq:expan} do not affect the sound velocity $c_s$, which is
526: the solution of
527: %
528: \begin{equation}\label{true_sound}
529: \left[A + P(c_s)\right] \left(Q - Rc_s^2\right) - B^2c_s^2 =0 \; .
530: \end{equation}
531: %
532: For obvious reasons, this analysis closely parallels that of
533: Ref.~\onlinecite{santamore2004} for Bose-Fermi mixtures, and Eq.~\ref{true_sound} reproduces the form of Eq.~\ref{model_dispersion} found earlier.
534:
535: \section{Discussion}\label{sec:discuss}
536:
537:
538: At the spinodal lines $A+P(0)=0$. This is the same condition that was
539: used in Refs.~\onlinecite{chien2006,chen2006} to identify the unstable
540: region, although the possibility of metastability was
541: ignored. $A+P(0)<0$ inside the spinodal region, and one can
542: distinguish two cases: When $Q>0$ the sound velocity $c_s$ is pure
543: imaginary, while for $Q<0$ it is in general complex. As a consequence,
544: the unstable mode spectrum changes in character as one moves across
545: $Q=0$ (see Fig.~\ref{fig:unstable}). Approaching the region of negative superfluid density $Q<0$
546: from the region $Q>0$, a second imaginary modes appears (see panel (c)
547: in Fig.~\ref{fig:unstable}) and the two merge (panel (b)): The
548: resulting `gap' region corresponds to complex frequencies, implying a
549: `flickering' component to the instability. The most
550: unstable mode always corresponds to a pure imaginary
551: frequency, however.
552: %
553: While the most unstable mode frequency and wavevector go to zero on
554: the superfluid side of the spinodal region, the instability towards a
555: Fulde-Ferrell-Larkin-Ovchinnikov (FFLO) phase means that the characteristic
556: wavevector does not go to zero on the normal
557: side~\cite{sheehy2006}. Indeed, one may view the early stages of
558: spinodal decomposition as a transient FFLO state.
559:
560:
561: The characteristic length and time scales at which inhomogeneities
562: appear as the precursor of phase separation are determined by the most
563: unstable modes. At unitarity $1/k_Fa=0$ and for $T_F=1\mu$K, this time
564: scale is of order of $400\mu$s, which is on the same scale as the
565: condensate formation time, as measured in
566: Ref.~\onlinecite{zwierlein-form}. The length scale is roughly $1/k_F \simeq
567: 0.1\mu$m. Both scales become larger as one approaches the spinodal
568: lines, which will always occur somewhere in the trap. The unitarity
569: region may be a suitable place to observe the late stages of spinodal
570: decomposition and the possible existence of a coarsening regime.
571: %beyond the validity of the linear approximation used here.
572:
573: In conclusion, we have studied the early stage dynamics of phase
574: separation in polarised Fermi superfluids. We expect that the
575: investigation of these instabilities is within reach of current
576: experiments.
577:
578:
579:
580: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
581: %\begin{acknowledgments}
582: We are grateful to P. Eastham for help with the numerics and to TCM
583: group in Cambridge for the use of computer resources. FMM would like
584: to acknowledge the financial support of EPSRC.
585: %\end{acknowledgments}
586: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
587:
588: %\bibliography{Letter}
589:
590: \begin{thebibliography}{28}
591: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
592: \expandafter\ifx\csname bibnamefont\endcsname\relax
593: \def\bibnamefont#1{#1}\fi
594: \expandafter\ifx\csname bibfnamefont\endcsname\relax
595: \def\bibfnamefont#1{#1}\fi
596: \expandafter\ifx\csname citenamefont\endcsname\relax
597: \def\citenamefont#1{#1}\fi
598: \expandafter\ifx\csname url\endcsname\relax
599: \def\url#1{\texttt{#1}}\fi
600: \expandafter\ifx\csname urlprefix\endcsname\relax\def\urlprefix{URL }\fi
601: \providecommand{\bibinfo}[2]{#2}
602: \providecommand{\eprint}[2][]{\url{#2}}
603:
604: \bibitem[{\citenamefont{Regal et~al.}(2004)\citenamefont{Regal, Greiner, and
605: Jin}}]{regal2004}
606: \bibinfo{author}{\bibfnamefont{C.~A.} \bibnamefont{Regal}},
607: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Greiner}}, \bibnamefont{and}
608: \bibinfo{author}{\bibfnamefont{D.~S.} \bibnamefont{Jin}},
609: \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{92}},
610: \bibinfo{pages}{040403} (\bibinfo{year}{2004}).
611:
612: \bibitem[{\citenamefont{Zwierlein et~al.}(2004)\citenamefont{Zwierlein, Stan,
613: Schunck, Raupach, Kerman, and Ketterle}}]{zwierlein2004}
614: \bibinfo{author}{\bibfnamefont{M.~W.} \bibnamefont{Zwierlein}},
615: \bibinfo{author}{\bibfnamefont{C.~A.} \bibnamefont{Stan}},
616: \bibinfo{author}{\bibfnamefont{C.~H.} \bibnamefont{Schunck}},
617: \bibinfo{author}{\bibfnamefont{S.~M.~F.} \bibnamefont{Raupach}},
618: \bibinfo{author}{\bibfnamefont{A.~J.} \bibnamefont{Kerman}},
619: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{W.}~\bibnamefont{Ketterle}},
620: \bibinfo{journal}{Phys.\ Rev.\ Lett.} \textbf{\bibinfo{volume}{92}},
621: \bibinfo{pages}{120403} (\bibinfo{year}{2004}).
622:
623: \bibitem[{\citenamefont{Chin et~al.}(2004)\citenamefont{Chin, Bartenstein,
624: Altmeyer, Riedl, Jochim, Denschlag, and Grimm}}]{chin2004}
625: \bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Chin}},
626: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Bartenstein}},
627: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Altmeyer}},
628: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Riedl}},
629: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Jochim}},
630: \bibinfo{author}{\bibfnamefont{J.~H.} \bibnamefont{Denschlag}},
631: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{Grimm}},
632: \bibinfo{journal}{Science} \textbf{\bibinfo{volume}{305}},
633: \bibinfo{pages}{1128} (\bibinfo{year}{2004}).
634:
635: \bibitem[{\citenamefont{Bourdel et~al.}(2004)\citenamefont{Bourdel, Khaykovich,
636: Cubizolles, Zhang, Chevy, Teichmann, Tarruell, Kokkelmans, and
637: Salomon}}]{bourdel2004}
638: \bibinfo{author}{\bibfnamefont{T.}~\bibnamefont{Bourdel}},
639: \bibinfo{author}{\bibfnamefont{L.}~\bibnamefont{Khaykovich}},
640: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Cubizolles}},
641: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Zhang}},
642: \bibinfo{author}{\bibfnamefont{F.}~\bibnamefont{Chevy}},
643: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Teichmann}},
644: \bibinfo{author}{\bibfnamefont{L.}~\bibnamefont{Tarruell}},
645: \bibinfo{author}{\bibfnamefont{S.~J.~J.~M.~F.}~\bibnamefont{Kokkelmans}},
646: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Salomon}},
647: \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{93}},
648: \bibinfo{pages}{050401} (\bibinfo{year}{2004}).
649:
650: \bibitem[{\citenamefont{Kinast et~al.}(2004)\citenamefont{Kinast, Hemmer, Gehm,
651: Turlapov, and Thomas}}]{kinast2004}
652: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Kinast}},
653: \bibinfo{author}{\bibfnamefont{S.~L.} \bibnamefont{Hemmer}},
654: \bibinfo{author}{\bibfnamefont{M.~E.} \bibnamefont{Gehm}},
655: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Turlapov}}, \bibnamefont{and}
656: \bibinfo{author}{\bibfnamefont{J.~E.} \bibnamefont{Thomas}},
657: \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{92}},
658: \bibinfo{pages}{150402} (\bibinfo{year}{2004}).
659:
660: \bibitem[{\citenamefont{Zwierlein
661: et~al.}(2005{\natexlab{a}})\citenamefont{Zwierlein, Schunck, Stan, Raupach,
662: and Ketterle}}]{zwierlein2005}
663: \bibinfo{author}{\bibfnamefont{M.~W.} \bibnamefont{Zwierlein}},
664: \bibinfo{author}{\bibfnamefont{C.~H.} \bibnamefont{Schunck}},
665: \bibinfo{author}{\bibfnamefont{C.~A.} \bibnamefont{Stan}},
666: \bibinfo{author}{\bibfnamefont{S.~M.~F.} \bibnamefont{Raupach}},
667: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{W.}~\bibnamefont{Ketterle}},
668: \bibinfo{journal}{\prl} \textbf{\bibinfo{volume}{94}}, \bibinfo{eid}{180401}
669: (\bibinfo{year}{2005}{\natexlab{a}}).
670:
671: \bibitem[{\citenamefont{Zwierlein
672: et~al.}(2006{\natexlab{a}})\citenamefont{Zwierlein, Schirotzek, Schunck, and
673: Ketterle}}]{zwierlein2006}
674: \bibinfo{author}{\bibfnamefont{M.~W.} \bibnamefont{Zwierlein}},
675: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Schirotzek}},
676: \bibinfo{author}{\bibfnamefont{C.~H.} \bibnamefont{Schunck}},
677: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{W.}~\bibnamefont{Ketterle}},
678: \bibinfo{journal}{Science} \textbf{\bibinfo{volume}{311}},
679: \bibinfo{pages}{492} (\bibinfo{year}{2006}{\natexlab{a}}).
680:
681: \bibitem[{\citenamefont{Partridge
682: et~al.}(2006{\natexlab{a}})\citenamefont{Partridge, Li, Kamar, Liao, and
683: Hulet}}]{partridge2006}
684: \bibinfo{author}{\bibfnamefont{G.~B.} \bibnamefont{Partridge}},
685: \bibinfo{author}{\bibfnamefont{W.}~\bibnamefont{Li}},
686: \bibinfo{author}{\bibfnamefont{R.~I.} \bibnamefont{Kamar}},
687: \bibinfo{author}{\bibfnamefont{Y.}~\bibnamefont{Liao}}, \bibnamefont{and}
688: \bibinfo{author}{\bibfnamefont{R.~G.} \bibnamefont{Hulet}},
689: \bibinfo{journal}{Science} \textbf{\bibinfo{volume}{311}},
690: \bibinfo{pages}{503} (\bibinfo{year}{2006}{\natexlab{a}}).
691:
692: \bibitem[{\citenamefont{Zwierlein
693: et~al.}(2006{\natexlab{b}})\citenamefont{Zwierlein, Schunck, Schirotzek, and
694: Ketterle}}]{zwierlein2006_2}
695: \bibinfo{author}{\bibfnamefont{M.~W.} \bibnamefont{Zwierlein}},
696: \bibinfo{author}{\bibfnamefont{C.~H.} \bibnamefont{Schunck}},
697: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Schirotzek}},
698: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{W.}~\bibnamefont{Ketterle}},
699: \bibinfo{journal}{Nature} \textbf{\bibinfo{volume}{442}}, \bibinfo{pages}{54}
700: (\bibinfo{year}{2006}{\natexlab{b}}), \eprint{cond-mat/0605258}.
701:
702: \bibitem[{\citenamefont{Shin et~al.}(2006)\citenamefont{Shin, Zwierlein,
703: Schunck, Schirotzek, and Ketterle}}]{shin2006}
704: \bibinfo{author}{\bibfnamefont{Y.}~\bibnamefont{Shin}},
705: \bibinfo{author}{\bibfnamefont{M.~W.} \bibnamefont{Zwierlein}},
706: \bibinfo{author}{\bibfnamefont{C.~H.} \bibnamefont{Schunck}},
707: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Schirotzek}},
708: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{W.}~\bibnamefont{Ketterle}},
709: \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{97}},
710: \bibinfo{pages}{030401} (\bibinfo{year}{2006}).
711:
712:
713: \bibitem[{\citenamefont{Partridge
714: et~al.}(2006{\natexlab{b}})\citenamefont{Partridge, Li, Liao, Hulet, Haque,
715: and Stoof}}]{partridge2006_2}
716: \bibinfo{author}{\bibfnamefont{G.~B.} \bibnamefont{Partridge}},
717: \bibinfo{author}{\bibfnamefont{W.}~\bibnamefont{Li}},
718: \bibinfo{author}{\bibfnamefont{Y.~A.} \bibnamefont{Liao}},
719: \bibinfo{author}{\bibfnamefont{R.~G.} \bibnamefont{Hulet}},
720: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Haque}}, \bibnamefont{and}
721: \bibinfo{author}{\bibfnamefont{H.~T.~C.} \bibnamefont{Stoof}},
722: \bibinfo{journal}{\prl} \textbf{\bibinfo{volume}{97}}, \bibinfo{eid}{190407}
723: (\bibinfo{year}{2006}{\natexlab{b}}).
724:
725: \bibitem[{\citenamefont{Bedaque et~al.}(2003)\citenamefont{Bedaque, Caldas, and
726: Rupak}}]{bedaque2003}
727: \bibinfo{author}{\bibfnamefont{P.~F.} \bibnamefont{Bedaque}},
728: \bibinfo{author}{\bibfnamefont{H.}~\bibnamefont{Caldas}}, \bibnamefont{and}
729: \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Rupak}},
730: \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{91}},
731: \bibinfo{pages}{247002} (\bibinfo{year}{2003}).
732:
733: \bibitem[{\citenamefont{Sheehy and Radzihovsky}(2006)}]{sheehy2006}
734: \bibinfo{author}{\bibfnamefont{D.~E.} \bibnamefont{Sheehy}} \bibnamefont{and}
735: \bibinfo{author}{\bibfnamefont{L.}~\bibnamefont{Radzihovsky}},
736: \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{96}},
737: \bibinfo{pages}{060401} (\bibinfo{year}{2006}).
738:
739:
740:
741: \bibitem[{\citenamefont{Parish et~al.}()\citenamefont{Parish, Marchetti,
742: Lamacraft, and Simons}}]{parish2006}
743: \bibinfo{author}{\bibfnamefont{M.~M.} \bibnamefont{Parish}},
744: \bibinfo{author}{\bibfnamefont{F.~M.} \bibnamefont{Marchetti}},
745: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Lamacraft}},
746: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{B.~D.}
747: \bibnamefont{Simons}}, \eprint{cond-mat/0605744 (Nat. Phys. to appear)}.
748:
749: \bibitem[{\citenamefont{Gubbels et~al.}(2006)\citenamefont{Gubbels, Romans, and
750: Stoof}}]{gubbels2006}
751: \bibinfo{author}{\bibfnamefont{K.~B.} \bibnamefont{Gubbels}},
752: \bibinfo{author}{\bibfnamefont{M.~W.~J.} \bibnamefont{Romans}},
753: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{H.~T.~C.}
754: \bibnamefont{Stoof}}, \bibinfo{journal}{Phys. Rev. Lett.}
755: \textbf{\bibinfo{volume}{97}}, \bibinfo{pages}{210402}
756: (\bibinfo{year}{2006}).
757:
758: \bibitem[{\citenamefont{Chien et~al.}(2006)\citenamefont{Chien, Chen, He, and
759: Levin}}]{chien2006}
760: \bibinfo{author}{\bibfnamefont{C.-C.} \bibnamefont{Chien}},
761: \bibinfo{author}{\bibfnamefont{Q.}~\bibnamefont{Chen}},
762: \bibinfo{author}{\bibfnamefont{Y.}~\bibnamefont{He}}, \bibnamefont{and}
763: \bibinfo{author}{\bibfnamefont{K.}~\bibnamefont{Levin}},
764: \bibinfo{journal}{\prl} \textbf{\bibinfo{volume}{97}},
765: \bibinfo{pages}{090402} (\bibinfo{year}{2006}).
766:
767: \bibitem[{\citenamefont{Iskin and S\'a~de Melo}(2006)}]{iskin2006}
768: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Iskin}} \bibnamefont{and}
769: \bibinfo{author}{\bibfnamefont{C.~A.~R.} \bibnamefont{S\'a~de Melo}},
770: \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{97}},
771: \bibinfo{pages}{100404} (\bibinfo{year}{2006}).
772:
773: \bibitem[{\citenamefont{Pao et~al.}(2006{\natexlab{a}})\citenamefont{Pao, Wu,
774: and Yip}}]{pao2006}
775: \bibinfo{author}{\bibfnamefont{C.-H.} \bibnamefont{Pao}},
776: \bibinfo{author}{\bibfnamefont{S.-T.} \bibnamefont{Wu}}, \bibnamefont{and}
777: \bibinfo{author}{\bibfnamefont{S.-K.} \bibnamefont{Yip}},
778: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{73}},
779: \bibinfo{pages}{132506} (\bibinfo{year}{2006}{\natexlab{a}}).
780:
781: \bibitem[{\citenamefont{Pao et~al.}(2006{\natexlab{b}})\citenamefont{Pao, Wu,
782: and Yip}}]{pao2006b}
783: \bibinfo{author}{\bibfnamefont{C.-H.} \bibnamefont{Pao}},
784: \bibinfo{author}{\bibfnamefont{S.-T.} \bibnamefont{Wu}}, \bibnamefont{and}
785: \bibinfo{author}{\bibfnamefont{S.-K.} \bibnamefont{Yip}},
786: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{74}},
787: \bibinfo{pages}{189901(E)} (\bibinfo{year}{2006}{\natexlab{b}}).
788:
789: \bibitem[{\citenamefont{Gunton et~al.}(1983)\citenamefont{Gunton, Miguel, and
790: Sahni}}]{gunton}
791: \bibinfo{author}{\bibfnamefont{J.~D.} \bibnamefont{Gunton}},
792: \bibinfo{author}{\bibfnamefont{M.~S.} \bibnamefont{Miguel}},
793: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{P.~S.} \bibnamefont{Sahni}},
794: \emph{\bibinfo{title}{Phase Transitions and Critical Phenomena}},
795: vol.~\bibinfo{volume}{8} (\bibinfo{publisher}{Academic Press, London},
796: \bibinfo{year}{1983}).
797:
798: \bibitem[{\citenamefont{Hoffer and Sinha}(1986)}]{hoffer1986}
799: \bibinfo{author}{\bibfnamefont{J.~K.} \bibnamefont{Hoffer}} \bibnamefont{and}
800: \bibinfo{author}{\bibfnamefont{D.~N.} \bibnamefont{Sinha}},
801: \bibinfo{journal}{Phys. Rev. A} \textbf{\bibinfo{volume}{33}},
802: \bibinfo{pages}{1918} (\bibinfo{year}{1986}).
803:
804: \bibitem[{\citenamefont{Bray}(1994)}]{bray1994}
805: \bibinfo{author}{\bibfnamefont{A.~J.} \bibnamefont{Bray}},
806: \bibinfo{journal}{Adv. Phys.} \textbf{\bibinfo{volume}{51}},
807: \bibinfo{pages}{481} (\bibinfo{year}{1994}).
808:
809:
810: \bibitem[{\citenamefont{Hohenberg and Nelson}(1979)}]{hohenberg1979}
811: \bibinfo{author}{\bibfnamefont{P.~C.} \bibnamefont{Hohenberg}}
812: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{D.~R.}
813: \bibnamefont{Nelson}}, \bibinfo{journal}{\prb} \textbf{\bibinfo{volume}{20}},
814: \bibinfo{pages}{2665} (\bibinfo{year}{1979}).
815:
816: \bibitem[{\citenamefont{Callen}(1985)}]{callen}
817: \bibinfo{author}{\bibfnamefont{H.~B.} \bibnamefont{Callen}},
818: \emph{\bibinfo{title}{Thermodynamics and an Introduction to
819: Thermostatistics}} (\bibinfo{publisher}{John Wiley \& Sons, New York},
820: \bibinfo{year}{1985}).
821:
822: \bibitem[{\citenamefont{Sarma}(1963)}]{sarma1963}
823: \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Sarma}}, \bibinfo{journal}{J.
824: Phys. Chem. Solids} \textbf{\bibinfo{volume}{24}}, \bibinfo{pages}{1029}
825: (\bibinfo{year}{1963}).
826:
827: \bibitem[{\citenamefont{Engelbrecht et~al.}(1997)\citenamefont{Engelbrecht,
828: Randeria, and {S\'a de Melo}}}]{engelbrecht1997}
829: \bibinfo{author}{\bibfnamefont{J.~R.} \bibnamefont{Engelbrecht}},
830: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Randeria}}, \bibnamefont{and}
831: \bibinfo{author}{\bibfnamefont{C.~A.~R.} \bibnamefont{{S\'a de Melo}}},
832: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{55}},
833: \bibinfo{pages}{15153} (\bibinfo{year}{1997}).
834:
835: \bibitem[{\citenamefont{Santamore et~al.}(2004)\citenamefont{Santamore, Gaudio,
836: and Timmermans}}]{santamore2004}
837: \bibinfo{author}{\bibfnamefont{D.~H.} \bibnamefont{Santamore}},
838: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Gaudio}}, \bibnamefont{and}
839: \bibinfo{author}{\bibfnamefont{E.}~\bibnamefont{Timmermans}},
840: \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{93}},
841: \bibinfo{pages}{250402} (\bibinfo{year}{2004}).
842:
843: \bibitem[{\citenamefont{Chen et~al.}(2006)\citenamefont{Chen, He, Chien, and
844: Levin}}]{chen2006}
845: \bibinfo{author}{\bibfnamefont{Q.}~\bibnamefont{Chen}},
846: \bibinfo{author}{\bibfnamefont{Y.}~\bibnamefont{He}},
847: \bibinfo{author}{\bibfnamefont{C.-C.} \bibnamefont{Chien}}, \bibnamefont{and}
848: \bibinfo{author}{\bibfnamefont{K.}~\bibnamefont{Levin}},
849: \bibinfo{journal}{Phys. Rev. A} \textbf{\bibinfo{volume}{74}},
850: \bibinfo{pages}{063603} (\bibinfo{year}{2006}).
851:
852: \bibitem[{\citenamefont{Zwierlein
853: et~al.}(2005{\natexlab{b}})\citenamefont{Zwierlein, Schunck, Stan, Raupach,
854: and Ketterle}}]{zwierlein-form}
855: \bibinfo{author}{\bibfnamefont{M.~W.} \bibnamefont{Zwierlein}},
856: \bibinfo{author}{\bibfnamefont{C.~H.} \bibnamefont{Schunck}},
857: \bibinfo{author}{\bibfnamefont{C.~A.} \bibnamefont{Stan}},
858: \bibinfo{author}{\bibfnamefont{S.~M.~F.} \bibnamefont{Raupach}},
859: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{W.}~\bibnamefont{Ketterle}},
860: \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{94}},
861: \bibinfo{pages}{180401} (\bibinfo{year}{2005}{\natexlab{b}}).
862:
863:
864:
865: \end{thebibliography}
866:
867: \end{document}