1: \documentclass[prl,aps,twocolumn,showpacs,amsmath,amssymb,floatfix]{revtex4}
2: %\documentclass[prl,aps,preprint,showpacs,amsmath,amssymb,floatfix]{revtex4}
3:
4: \usepackage[dvips]{graphicx}
5: \usepackage{bm}
6: \newcommand{\rot}{\bm\nabla\wedge}
7:
8: \begin{document}
9:
10: \title{Fermi Gases in Slowly Rotating Traps:
11: Superfluid vs Collisional Hydrodynamics}
12: \author{Marco Cozzini}
13: \author{Sandro Stringari}
14: \affiliation{Dipartimento di Fisica, Universit\`a di Trento and BEC-INFM, I-38050 Povo, Italy}
15:
16: \date{\today}
17:
18: \begin{abstract}
19: The dynamic behavior of a Fermi gas confined in a deformed trap rotating at
20: low angular velocity is investigated in the framework of hydrodynamic theory.
21: The differences exhibited by a normal gas in the collisional regime and a
22: superfluid are discussed. Special emphasis is given to the collective
23: oscillations excited when the deformation of the rotating trap is suddenly
24: removed or when the rotation is suddenly stopped. The presence of vorticity in
25: the normal phase is shown to give rise to precession and beating phenomena
26: which are absent in the superfluid phase.
27: \end{abstract}
28:
29: \pacs{03.75.Ss, 67.40.Db, 03.75.Kk}
30:
31: \maketitle
32:
33: Recent experiments on ultracold atomic Fermi gases close to a Feshbach
34: resonance \cite{thomas,jin,salomon} have explicitly revealed the emergence of a
35: hydrodynamic regime which shows up in the anisotropic shape of the expanding
36: gas.
37: This behavior is the consequence of the anisotropy of the pressure gradients
38: which are stronger in the tighter directions of the trap. It should be
39: contrasted with the isotropic nature of the expansion of the non interacting
40: Fermi gas, which instead reflects the initial isotropy of the momentum
41: distribution.
42: In Ref.~\cite{menotti} the anisotropy of the expansion of a Fermi gas has been
43: suggested as a possible signature of superfluidity.
44: The analysis of Ref.~\cite{menotti} concerns, however, dilute gases far from
45: Feshbach resonances where the system, in the normal phase, is collisionless at
46: low temperature due to Pauli blocking \cite{vichi1} and consequently expands
47: differently from a superfluid.
48: If one instead works close to a Feshbach resonance the huge increase of the
49: collisional rate will favor the achievement of the hydrodynamic regime in the
50: normal phase even at low temperature \cite{thomas}.
51: At the same time the resonance effect is expected to enhance significantly the
52: value of the critical temperature for superfluidity \cite{holland}, providing
53: promising perspectives for its experimental realization.
54: Since the anisotropy of the expansion is compatible with either the collisional
55: and superfluid hydrodynamic pictures, its experimental observation cannot be
56: used as a test of superfluidity without further considerations. As a
57: consequence, it is important to identify alternative effects which permit to
58: distinguish between the two regimes.
59:
60: The purpose of the present letter is to show that the study of the collective
61: oscillations of a trapped Fermi gas rotating at low angular velocity can
62: provide a useful identification of superfluidity.
63: It is in fact well known that a superfluid cannot support vorticity, unless
64: quantized vortices are created. In the following we will consider situations
65: where vortices are absent. This can be ensured by rotating the confining trap
66: at sufficiently low angular velocities. Under these conditions the dynamic
67: behavior of a superfluid is described by the equations of irrotational
68: hydrodynamics:
69: %
70: \begin{equation} \label{eq:cont}
71: \frac{\partial{n}}{\partial t}+\bm\nabla\cdot({n}\bm{v}) = 0 \ ,
72: \end{equation}
73: %
74: \begin{equation} \label{eq:euler irrHD}
75: \frac{\partial\bm{v}}{\partial t}\!=\!
76: -\!\bm\nabla\!\left(\!\frac{v^2}{2}\!+\!\frac{V_{\text{ext}}}{M}\!+
77: \!\frac{\mu}{M}\!\right)\!=\!
78: -\!\bm\nabla\!\left(\!\frac{v^2}{2}\!+\!\frac{V_{\text{ext}}}{M}\!\right)\!
79: -\!\frac{\bm\nabla P}{M{n}} ,
80: \end{equation}
81: %
82: where $V_{\text{ext}}$ is the external potential generated by the confining
83: trap, while $P$ and $\mu$ are, respectively, the pressure and the chemical
84: potential of a uniform gas evaluated at the corresponding density. In the last
85: identity we have used the $T=0$ thermodynamic relationship
86: $\bm\nabla{P}={n}\bm\nabla\mu$. The above equations apply to dynamic phenomena
87: of macroscopic type where the local density approximation to the equation of
88: state is justified. They hold for both Bose and Fermi superfluids at zero
89: temperature. At finite temperature, below $T_c$, they should be generalized to
90: the equations of two fluid hydrodynamics (see, for example, \cite{landau}).
91: Equations (\ref{eq:cont}) and (\ref{eq:euler irrHD}) have been systematically
92: used in the last years to test the effects of superfluidity on the dynamic
93: behavior of Bose-Einstein condensed gases \cite{PS}.
94:
95: Differently from a superfluid, a normal gas can support vorticity and in the
96: collisional regime the equation for the velocity field takes the classical
97: Euler form
98: %
99: \begin{equation} \label{eq:euler rotHD}
100: \frac{\partial\bm{v}}{\partial t} =
101: -\bm\nabla\left(\frac{v^2}{2}+\frac{V_{\text{ext}}}{M}\right)-
102: \frac{\bm\nabla P}{M{n}}+\bm{v}\wedge(\rot\bm{v}) \ ,
103: \end{equation}
104: %
105: where the time dependence of the pressure should be calculated taking into
106: account the conditions of adiabaticity, following from the conservation of
107: entropy.
108: Equation (\ref{eq:euler rotHD}) differs from equation (\ref{eq:euler irrHD})
109: for the superfluid velocity because of the last term, proportional to the
110: vorticity $\rot\bm{v}$.
111: The hydrodynamic description (\ref{eq:euler rotHD}) holds provided the
112: collisional relaxation time $\tau$ is much smaller than the inverse of the
113: typical frequencies $\omega$ characterizing the dynamic phenomena under
114: investigation, fixed by the oscillator trapping frequencies: $\omega\tau\ll1$.
115:
116: Let us first suppose that the initial state of the system does not contain any
117: velocity flow (gas at rest in a static trap).
118: In this case the equations for the expansion as well as for the linearized
119: collective oscillations take the same irrotational form both in the superfluid
120: and classical cases.
121: In the following we will assume a trapping potential of harmonic shape:
122: $V_{\text{ext}}=M\,(\omega_x^2\,x^2+\omega_y^2\,y^2+\omega_z^2\,z^2)/2$.
123: If the potential is axially symmetric ($\omega_x=\omega_y=\omega_\perp$), one
124: can classify the solutions for the linearized oscillations in terms of the
125: third component $\hbar m$ of angular momentum.
126: We will consider solutions where the velocity field is linear in the spatial
127: coordinates. The solutions with $m=\pm2$ and $m=\pm1$ are surface excitations
128: of the form $\bm{v}\propto\bm\nabla(x\pm iy)^2$ and
129: $\bm{v}\propto\bm\nabla[(x\pm iy)z]$ respectively, with frequencies given by
130: %
131: \begin{eqnarray}
132: \label{eq:m+-2 irr}\omega(m=\pm2) & = & \sqrt2\,\omega_\perp \ , \\
133: \label{eq:m+-1 irr}\omega(m=\pm1) & = & \sqrt{\omega_\perp^2+\omega_z^2} \ ,
134: \end{eqnarray}
135: %
136: independent of the equation of state.
137: Assuming a power law dependence for the isoentropic equation of state
138: ($P\propto{n}^{\gamma+1}$), one easily finds analytic solutions also for the
139: $m=0$ modes which are characterized by a velocity field of the form
140: $\bm{v}\propto\bm\nabla[a(x^2+y^2)+bz^2]$.
141: The corresponding collective frequencies are given by
142: %
143: \begin{equation} \label{eq:m=0 axisymm gamma}
144: \begin{array}{l}
145: \displaystyle\omega^2(m=0) \,\ = \,\ \frac{1}{2}\,
146: \bigg[2\,(\gamma+1)\,\omega_\perp^2+(\gamma+2)\,\omega_z^2 \,\ \pm \\
147: \sqrt{4(\gamma+1)^2\omega_\perp^4+(\gamma+2)^2\omega_z^4
148: +4(\gamma^2-3\gamma-2)\omega_\perp^2\omega_z^2}\,\bigg] \ .
149: \end{array}
150: \end{equation}
151: %
152: Equation (\ref{eq:m=0 axisymm gamma}) reduces to the one derived in
153: Ref.~\cite{sandro} in the interacting Bose case ($\gamma=1$), while in the case
154: of the ideal gas ($\gamma=2/3$) it reduces to the predictions of
155: Refs.~\cite{griffin,minguzzi2,kagan} and, for spherical trapping, to the ones
156: of Refs.~\cite{baranov}.
157: For elongated traps ($\omega_z\ll\omega_\perp$) one finds
158: $\omega=\sqrt{2(\gamma+1)}\,\omega_\perp$ and
159: $\omega=\sqrt{(3\,\gamma+2)/(\gamma+1)}\,\omega_z$.
160:
161: The analysis of the collective frequencies is easily generalized to tri-axial
162: anisotropy ($\omega_x\ne\omega_y\ne\omega_z$), where one finds 3 solutions of
163: the form $\bm{v}\propto\bm\nabla(xy)$, $\bm{v}\propto\bm\nabla(xz)$,
164: $\bm{v}\propto\bm\nabla(yz)$.
165: These are the so called scissors modes \cite{david scissors} relative to the
166: three pairs of axes, with frequency $\sqrt{\omega_x^2+\omega_y^2}$,
167: $\sqrt{\omega_x^2+\omega_z^2}$ and $\sqrt{\omega_y^2+\omega_z^2}$, respectively.
168: The other three solutions have the form
169: $\bm{v}\propto\bm\nabla(ax^2+by^2+cz^2)$ and their frequencies obey the
170: equation
171: %
172: \begin{equation} \label{eq:omega6 gamma}
173: \begin{array}{c}
174: \omega^6-(2+\gamma)(\omega_x^2+\omega_y^2+\omega_z^2)\,\omega^4+ \\
175: \rule[-0.3cm]{0cm}{0.8cm}+4\,(\gamma+1)
176: (\omega_x^2\,\omega_y^2+\omega_x^2\,\omega_z^2+
177: \omega_y^2\,\omega_z^2)\,\omega^2+ \\
178: -4\,(2+3\,\gamma)\,\omega_x^2\,\omega_y^2\,\omega_z^2 \,\ = \,\ 0 \ .
179: \end{array}
180: \end{equation}
181:
182: Let us stress again that the above results hold both in the superfluid and
183: normal hydrodynamic phases.
184: In particular, if we excite the scissors mode by suddenly rotating the
185: confining trap starting from the ground state configuration, like in the
186: experiment of Ref.~\cite{foot}, the gas will oscillate with the same frequency
187: in both regimes.
188: The situation would be different in a dilute Fermi gas far from Feshbach
189: resonances where the dynamic response, in the normal phase, is collisionless at
190: low temperature and the behavior of the scissors oscillation exhibits
191: different features with respect to the superfluid case
192: \cite{david scissors,minguzzi}.
193: Although in a Fermi gas close to a Feshbach resonance the measurement of the
194: collective frequencies (\ref{eq:m+-2 irr}$-$\ref{eq:omega6 gamma}) around the
195: ground state does not provide a direct indication of superfluidity, their
196: experimental determination would be nevertheless very useful, providing an
197: accurate and quantitative proof of the achievement of the hydrodynamic regime.
198: This is best tested looking at the surface excitations, whose frequencies are
199: insensitive to the equation of state.
200:
201: We are now ready to explore the rotational properties of the system. The most
202: natural procedure is to let a deformed trap rotate at angular velocity $\Omega$
203: in the $x$-$y$ plane.
204: The angular velocity should be turned on adiabatically in order to ensure the
205: conditions of stationarity.
206: Furthermore the final angular velocity $\Omega$ should be small enough to avoid
207: the formation of quantized vortices in the superfluid phase. This condition is
208: not very restrictive since the angular velocity needed to nucleate vortices by
209: adiabatic increase of the rotation rate is very high \cite{castin}, as
210: confirmed experimentally in the case of Bose-Einstein condensates
211: \cite{chevy nucleation}.
212:
213: If the system is superfluid the stationary velocity field has the irrotational
214: form $\bm{v}=\alpha\bm\nabla(xy)$. The dependence of the coefficient $\alpha$
215: on the angular velocity $\Omega$ has been discussed in Ref.~\cite{recati} and
216: for small angular velocities reduces to $\alpha=-\epsilon\Omega$, where
217: $\epsilon=(\omega_x^2-\omega_y^2)/(\omega_x^2+\omega_y^2)$ is the deformation
218: of the trap in the $x$-$y$ plane.
219: Actually, in the limit of an axisymmetric trap ($\epsilon=0$), the velocity
220: field exactly vanishes revealing that the superfluid is not capable to rotate.
221: If instead the system is normal, the stationary velocity field takes the rigid
222: form $\bm{v}=\bm\Omega\wedge\bm{r}$, corresponding to
223: $\rot\bm{v}=2\,\bm\Omega$.
224: Notice that, while the rigid rotation is correctly described by the
225: hydrodynamic equations (\ref{eq:cont}) and (\ref{eq:euler rotHD}), the
226: thermalization process which permits its achievement starting from a gas
227: initially at rest, is not accounted for by Eq.~(\ref{eq:euler rotHD}), because
228: of the absence of viscosity terms.
229: The time needed to achieve the rigid rotation of the gas (spin-up time) was
230: calculated in Ref.~\cite{david} and in the hydrodynamic regime
231: $\omega_\perp\,\tau\ll1$ is of the order of
232: $(\epsilon^2\omega_\perp^2\tau)^{-1}$, where $\tau$ is the relaxation time
233: fixed by collisions and $\omega_\perp^2=(\omega_x^2+\omega_y^2)/2$. This
234: suggests that, in order to reach steady rigid rotation in reasonable times, the
235: system should not be too deeply in the hydrodynamic regime and at the same time
236: the deformation of the rotating trap should not be too small
237: \footnote{The presence of a residual, static anisotropy $\epsilon_{\text{st}}$,
238: that in general acts to despin the cloud even if
239: $\epsilon_{\text{st}}\ll\epsilon$, is not expected to play an important role in
240: the hydrodynamic regime, where the ratio between the spin-up and spin-down
241: times is given by $(\epsilon_{\text{st}}/\epsilon)^2$ \cite{david}.}.
242:
243: The fact that the velocity field is so different in the two regimes, gives rise
244: to different predictions for the frequencies of the collective oscillations.
245: For an axisymmetric trap the differences become particularly clear. In fact,
246: while an axisymmetric superfluid cannot rotate and the frequencies of the
247: $m=\pm2$ quadrupole oscillations are given by Eq.~(\ref{eq:m+-2 irr}), in the
248: collisional hydrodynamic case the degeneracy of these modes is lifted by the
249: presence of the rigid rotation according to the law \cite{cozzini}:
250: %
251: \begin{equation} \label{eq:m+-2}
252: \omega(m=\pm2) =
253: \sqrt{2\,\omega_\perp^2-\Omega^2}\pm\Omega \ .
254: \end{equation}
255: %
256: Equation (\ref{eq:m+-2}) is consistent with the sum rule approach developed in
257: Ref.~\cite{zambelli}, yielding the result
258: %
259: \begin{equation} \label{eq:zambelli}
260: \Delta\omega = \omega(m=+2)-\omega(m=-2) = \frac{2}{M}\,
261: \frac{\langle l_z \rangle}{\langle x^2+y^2 \rangle}
262: \end{equation}
263: %
264: for the splitting of the $m=\pm2$ quadrupole frequencies, where
265: $\langle{l_z}\rangle$ is the angular momentum per particle.
266: In the case of a superfluid no angular momentum is carried by the system
267: because of the irrotationality constraint, while in the collisional
268: hydrodynamic regime the angular momentum is given by the rigid value
269: $\langle{l_z}\rangle=M\,\Omega\langle{x^2+y^2}\rangle$ and hence
270: $\Delta\omega=2\,\Omega$.
271: Measuring the splitting $\Delta\omega$ then provides an efficient way to detect
272: the vorticity of the gas.
273: The splitting can be measured by suddenly switching off the deformation of the
274: trap.
275: Immediately after, the system will feel the axisymmetric trap
276: $V_{\text{ext}}=M\,[\omega_\perp^2(x^2+y^2)+\omega_z^2\,z^2]/2$, but will no
277: longer be in equlibrium. Actually, in linear approximation the state of the
278: system can be written in the form
279: $|0\rangle+a_+|m=+2\rangle+a_-|m=-2\rangle$,
280: where $|0\rangle$ is the new equilibrium state, while $|m=\pm2\rangle$
281: are the $m=\pm2$ quadrupole states with excitation energies
282: $\hbar\omega_{\pm}\equiv\omega(m=\pm2)$.
283: The coefficients $a_{\pm}$ are fixed by the initial conditions, including the
284: quadrupole deformation and the angular velocity $\Omega$. One finds
285: $a_++a_-=\langle{x^2-y^2}\rangle/\sqrt\sigma$ and
286: $a_+\omega_+-a_-\omega_-=2\,\Omega\langle{x^2-y^2}\rangle/\sqrt\sigma$, where
287: $\sigma$ is the quadrupole strength \cite{zambelli}.
288: The time evolution of the states $|m=\pm2\rangle$ is fixed by the
289: frequencies $\omega_{\pm}$ and a simple calculation yields the result
290: %
291: \begin{equation} \label{eq:angle}
292: \tan(2\,\theta) =
293: \frac{b_+\tan(\Delta\omega\,t/2)+b_-\tan(\omega_Qt)}
294: {b_+-b_-\tan(\omega_Qt)\tan(\Delta\omega\,t/2)} \ ,
295: \end{equation}
296: %
297: where $\theta$ is the angle of the principal axis of the gas in the $x$-$y$
298: plane, $\omega_Q=(\omega_++\omega_-)/2$ and $b_\pm=a_+\pm a_-$.
299: In the superfluid case $\Delta\omega=0$ and one finds the result
300: $\tan(2\,\theta)=(2\,\Omega/\omega_Q)\tan(\omega_Qt)$. In the collisional
301: hydrodynamic case, the angle $\theta$ exhibits an additional slow precession.
302: This is best seen by taking stroboscopic images at times $t=2\pi n/\omega_Q$,
303: with $n$ integer, at which the deformation of the atomic cloud is maximum. For
304: such times $\tan(\omega_Qt)=0$ and Eq.~(\ref{eq:angle}) yields the precession
305: law $\theta=\Delta\omega\,t/4$.
306: This precession is caused by the splitting $\Delta\omega=2\,\Omega$ and is
307: absent in the superfluid case.
308: The numerical solution of the hydrodynamic equations confirms (see Fig.1) the
309: accuracy of the prediction (\ref{eq:angle}), based on linear approximation.
310: In the numerical calculation we have chosen the value $\gamma=2/3$
311: characterizing the equation of state of an ideal gas, including the most
312: relevant case of a degenerate Fermi gas. The results are not however sensitive
313: to the value of $\gamma$, consistently with Eq.~(\ref{eq:angle}), unless large
314: values of $\epsilon$ are considered.
315: The proposed experiment is similar to the one used in \cite{chevy} to measure
316: the angular momentum of quantized vortices.
317: In that experiment the deformation of the gas was produced by suddenly
318: switching on a laser field in an almost axy-symmetric Bose-Einstein
319: condensates. This corresponds to setting $a_+=a_-$ ($b_-=0$) and
320: Eq.~(\ref{eq:angle}) reduces to $\theta=\Delta\omega\,t/4$.
321:
322: \begin{figure}
323:
324: \includegraphics[width=8.5cm]{fig1.eps}
325:
326: \caption{\label{fig:angle eps=0}Time evolution of the angle $\theta$ (measured
327: in radians and plotted modulus $\pi/2$) after the sudden switch off of the trap
328: deformation.
329: (a) Superfluid hydrodynamics; (b) collisional hydrodynamics. The trap
330: parameters are $\epsilon=0.2$, $\Omega=0.2\,\omega_\perp$,
331: $\omega_z/\omega_\perp=0.1$, while $\gamma=2/3$. The dashed line in (b) is the
332: curve $\theta=\Omega\,t/2$ (see text).}
333:
334: \end{figure}
335:
336: If instead of switching off the deformation we simply stop the rotation of the
337: deformed trap, an other interesting phenomenon takes place that is worth
338: discussing.
339: In fact the gas, due to its inertia, will first continue rotating, but will
340: soon feel the restoring force produced by the deformed confining potential,
341: generating an oscillation around the new equilibrium configuration. In the
342: superfluid this procedure will excite the usual scissors mode with frequency
343: $\sqrt2\,\omega_\perp$.
344: In the case of classical hydrodynamics, the gas will instead oscillate
345: differently.
346: Actually a remarkable property exhibited by equations (\ref{eq:cont}) and
347: (\ref{eq:euler rotHD}) of classical hydrodynamics is that they admit stationary
348: solutions with non vanishing velocity flow also when the trap is at rest in the
349: laboratory frame.
350: These solutions have the form \cite{cozzini}
351: $\bm{v}=\bm\Omega\wedge\bm{r}+\alpha\bm\nabla(xy)$ and for small $\Omega$ one
352: finds $\alpha=\epsilon\Omega$.
353: Under the condition $\Omega\gg\epsilon^2\omega_\perp$, the resulting
354: oscillation can still be described as a linear combination of the two modes
355: (\ref{eq:m+-2}) and will consist of the characteristic beating
356: \footnote{In the opposite limit $\Omega\ll\epsilon^2\omega_\perp$, the beating
357: disappears and the gas oscillates with the frequency $\sqrt2\,\omega_\perp$.}
358: %
359: \begin{equation} \label{eq:theta}
360: \theta(t)=(\Omega/\omega_Q)\,\sin(\omega_Q\,t)\,\cos(\Delta\omega\,t/2) \ .
361: \end{equation}
362: %
363: Under the same conditions, also the intrinsic deformation $\delta$ of the cloud
364: exhibits an oscillatory behavior described by the law
365: $\delta(t)-\delta_0=-(2\,\Omega\,\epsilon/\omega_Q)\,
366: \sin(\omega_Q\,t)\,\sin(\Delta\omega\,t/2)$, where
367: $\delta_0=\langle{y^2-x^2}\rangle/\langle{x^2+y^2}\rangle=\epsilon$ is
368: the deformation of the stationary configuration.
369: In Fig.~\ref{fig:comparison} we compare the behavior of the angle $\theta$ in
370: the superfluid and normal cases. The numerical results have been obtained by
371: solving the hydrodynamic equations and in the collisional case exhibit the
372: typical beating predicted by Eq.~(\ref{eq:theta}).
373:
374: \begin{figure}
375:
376: \includegraphics[width=8.5cm]{fig2.eps}
377:
378: \caption{\label{fig:comparison}Time evolution of the angle $\theta$ (measured
379: in degrees) after the sudden stop of the rotation of the trap.
380: (a) Superfluid hydrodynamics; (b) collisional hydrodynamics. Parameters as in
381: Fig.~\ref{fig:angle eps=0}.}
382:
383: \end{figure}
384:
385: Let us finally recall that the differences between the superfluid and
386: collisional regimes discussed in this letter concern genuine macroscopic
387: phenomena. The theory of superfluidity predicts also the occurrence of more
388: microscopic quantum phenomena, associated with the quantization of circulation
389: and the appearance of quantized vortices.
390: The description of quantized vortices requires theoretical schemes beyond the
391: hydrodynamic picture and has been the object of recent work in Fermi
392: superfluids in various regimes, including the BCS \cite{viverit} and the
393: unitarity limit \cite{bulgac}.
394: Their observation, similarly to the case of Bose-Einstein condensates
395: \cite{chevy}, could be again revealed by the splitting (\ref{eq:zambelli}) of
396: the quadrupole frequencies. In a Fermi superfluid, one predicts
397: $\langle{l_z}\rangle=\hbar/2$ for a single vortex line aligned along the
398: symmetry axis
399: \footnote{It is useful to compare the angular momentum of a single vortex line
400: with the rigid value predicted in the normal phase. For an ideal degenerate
401: Fermi gas in a harmonic trap one finds $\langle{l_z}\rangle=
402: (\hbar/2)(6\,N\omega_z/\omega_\perp)^{1/3}(\Omega/\omega_\perp)
403: [1-(\Omega/\omega_\perp)^2]^{-2/3}$, where $N$ is the number of atoms of each
404: spin component.}.
405: The realization of quantized vortices however requires different procedures
406: with respect to the ones discussed in the present work.
407: In particular one should likely work at higher angular velocity, close to
408: $\omega_\perp/\sqrt2$ where the superfluid becomes unstable against the
409: formation of quadrupole deformations \cite{recati,dalfovo}. Furthermore, one
410: should switch on the rotation of the trap in a non adiabatic way in order to
411: favor their nucleation.
412: This procedure has already proven to be successful in the experimental
413: realization of quantized vortices in Bose-Einstein condensates \cite{ENS}.
414:
415: \begin{acknowledgments}
416: Stimulating discussions with L.~P.~Pitaevskii are acknowledged.
417: \end{acknowledgments}
418:
419: \begin{thebibliography}{99}
420:
421: \bibitem{thomas}
422: K.~M.~O'Hara \textit{et al.}, Science \textbf{298}, 2179 (2002).
423:
424: \bibitem{jin}
425: C.~A.~Regal and D.~S.~Jin, cond-mat/0302246.
426:
427: \bibitem{salomon}
428: T.~Bourdel \textit{et al.}, cond-mat/0303079.
429:
430: \bibitem{menotti}
431: C.~Menotti, P.~Pedri, and S.~Stringari, Phys. Rev. Lett. \textbf{89}, 250402
432: (2002).
433:
434: \bibitem{vichi1}
435: L.~Vichi, J. Low. Temp. Phys. \textbf{121}, 177 (2000).
436:
437: \bibitem{holland}
438: M.~Holland \textit{et al.}, Phys. Rev. Lett. \textbf{87}, 120406 (2001);
439: E.~Timmermans \textit{et al.}, Phys. Lett. A \textbf{285}, 228 (2001);
440: Y.~Ohashi and A.~Griffin, Phys. Rev. Lett. \textbf{89}, 130402 (2002).
441:
442: \bibitem{landau} L.~D.~Landau and E.~M.~Lifshitz, \textit{Fluid Mechanics} (2nd
443: edn), Pergamon Press, Oxford (1987).
444:
445: \bibitem{PS}
446: L.~P.~Pitaevskii and S.~Stringari, \textit{Bose-Einstein Condensation},
447: Oxford University Press (2003);
448: F.~Dalfovo \textit{et al.}, Rev. Mod. Phys. \textbf{71}, 463 (1999).
449:
450: \bibitem{chevy next} F.~Chevy and S.~Stringari, to be published.
451:
452: \bibitem{sandro}
453: S.~Stringari, Phys. Rev. Lett. \textbf{77}, 2360 (1996).
454:
455: \bibitem{griffin}
456: A.~Griffin, Wen-Chin~Wu, and S.~Stringari, Phys. Rev. Lett. \textbf{78}, 1838
457: (1997).
458:
459: \bibitem{minguzzi2}
460: M.~Amoruso \textit{et al.}, Eur. Phys. J. D \textbf{7}, 441 (1999).
461:
462: \bibitem{kagan}
463: Yu.~Kagan, E.~L.~Surkov, and G.~V.~Shlyapnikov, Phys. Rev. A \textbf{55}, R18
464: (1997).
465:
466: \bibitem{baranov}
467: M.~A.~Baranov and D.~S.~Petrov, Phys. Rev. A \textbf{62}, 041601 (2000);
468: G.~M.~Bruun and C.~W.~Clark, Phys. Rev. Lett. \textbf{83}, 5415 (1999).
469:
470: \bibitem{david scissors}
471: D.~Gu\'ery-Odelin and S.~Stringari, Phys. Rev. Lett. \textbf{83}, 4452 (1999).
472:
473: \bibitem{foot}
474: O.~M.~Marag\`o \textit{et al.}, Phys. Rev. Lett. \textbf{84}, 2056 (2000).
475:
476: \bibitem{minguzzi}
477: A.~Minguzzi and M.~P.~Tosi, Phys. Rev. A \textbf{63}, 023609 (2001).
478:
479: \bibitem{castin}
480: S.~Sinha and Y.~Castin, Phys. Rev. Lett. \textbf{87}, 190402 (2001).
481:
482: \bibitem{chevy nucleation}
483: K.~W.~Madison \textit{et al.}, Phys. Rev. Lett. \textbf{86}, 4443 (2001).
484: One should not confuse the nucleation frequency with the critical frequency
485: required to ensure energetic stability to the vortex. This latter frequency is
486: significantly smaller.
487:
488: \bibitem{recati}
489: A.~Recati, F.~Zambelli, and S.~Stringari, Phys. Rev. Lett. \textbf{86}, 377
490: (2001).
491:
492: \bibitem{david}
493: D.~Gu\'ery-Odelin, Phys. Rev. A \textbf{62}, 033607 (2000).
494:
495: \bibitem{cozzini}
496: M.~Cozzini and S.~Stringari, Phys. Rev. A \textbf{67}, 041602(R) (2003).
497: The hydrodynamic formalism developed in this paper refers to Bose-Einstein
498: condensates containing many vortical lines, where one can introduce the concept
499: of diffused vorticity.
500:
501: \bibitem{zambelli}
502: F.~Zambelli, S.~Stringari, Phys. Rev. Lett. \textbf{81}, 1754 (1998).
503:
504: \bibitem{chevy}
505: F.~Chevy, K.~W.~Madison, and J.~Dalibard, Phys. Rev. Lett. \textbf{85}, 2223
506: (2000).
507:
508: \bibitem{viverit}
509: G.~M.~Bruun and L.~Viverit, Phys. Rev. A \textbf{64}, 063606 (2001);
510: G.~M.~Bruun and C.~W.~Clark, J. Phys. B \textbf{33}, 3953 (2000).
511:
512: \bibitem{bulgac}
513: A.~Bulgac and Yongle Yu, cond-mat/0303235.
514:
515: \bibitem{dalfovo}
516: F.~Dalfovo and S.~Stringari, Phys. Rev. A \textbf{63}, 011601 (2001).
517:
518: \bibitem{ENS}
519: K.~W.~Madison \textit{et al.}, Phys. Rev. Lett. \textbf{84}, 806 (2000).
520:
521: \end{thebibliography}
522:
523: \end{document}
524: