Polarized Fermi gases 201/ω0τP54.543.532.521.510.5T=0.03T FT=000 0.05 0.1 0.15 0.2 0.25 0.3δX/R ↑Figure 2: The quantity 1/ω (s)D τ P determining the dampingof the dipole mode as a function of the amplitude of theoscillation for T =0andT =0.03 T F↑ .approximation. We found that the superfluid core disappears for values of the totalpolarization P =(N ↑ − N ↓ )/(N ↑ − N ↓ ) larger than the critical value P c =0.77, inagreement with the experimental results of the MIT group [2, 3]. Moreover, both thein situ column and density profiles of the two spin components agree very well with theexperimental values (see Fig. 1). It is worth noticing that the proper inclusion of interactioneffects in the normal phase is crucial to correctly describe the superfluid/normalquantum phase transition. The BCS mean-field approach would predict the incorrectvalue P c ≃ 1 for the critical polarization.Collective oscillations and collisional effectsThe functional (2) for the normal phase is particularly well suited to study the highlypolarized limit where the number of particles in the spin-down component is muchsmaller than in the spin-up component (N ↓ ≪ N ↑ ). In this limit the collective oscillationsof a trapped gas can be classified into two cathegories: the in-phase oscillations,where the motion is basically dominated by the majority component, and the out-ofphase(spin) oscillations where the minority component moves in the trap, the majorityone being practically at rest. The spin dipole frequency is given by the simple formula[6]√ (ω (s)D= ω mim ∗ 1+ 3 )5 A (2)
Polarized Fermi gases 21showing that even in the extreme limit of a single impurity interactions affect the spinfrequenciy in a sizable way by increasing the value by a factor ∼ 1.23 with respect tothe ideal gas prediction. Its measurement, together with the determination of the radiiand/or the RF spectrum [13], would consequently provide unique information on thevalue of the relevant interaction parameters in the normal phase. It is worth noticingthat differently from other quantities – e.g., radii and the RF gap[7] – the spin frequency(2) is independent of the number of atoms.A crucial point to discuss is the role of collisions which tend to damp out the spinoscillation. Indeed the spin dipole mode described above is well defined only in thecollisionless limit ω (s)D τ P ≫ 1, where τ P is the collisional time, while it becomes overdampedin the hydrodynamic regime ω (s)D τ P ≪ 1 since the spin current is not conservedby collisions. The collisional time τ P is expected to be particularly small at unitaritydue to the strong interactions, unless one considers extremely cold systems where collisionsare quenched by Pauli blocking. We have investigated this problem in closecollaboration with the team of the Niels Bohr Institut [14]. The most relevant result isshown in Fig. 2 where we report the quantity 1/ω (s)D τ P as a function of the amplitudeof the oscillation in units of the majority cloud radius R ↑ and for different values of thetemperature. While, as expected, at zero temperature the “small” amplitude oscillationsare well defined, by increaing T the effects of collisions become soon importantdue to the small value of the collective frequency. Less damping is of course expectedif one considers the higher frequency modes of the gas.[1] M. W. Zwierlein, A. Schirotzek, C. H. Schunck, W. Ketterle, Science 311, 492(2006).[2] Y. Shin, M. W. Zwierlein, C. H. Schunck, A. Schirotzek, and W. Ketterle, Phys.Rev. Lett. 97 030401 (2006).[3] Y. Shin, C. H. Schunck, A. Schirotzek, and W. Ketterle, Nature 454, 689 (2008).[4] G. B. Partridge, W. Li, R. I. Kamar, Y. A. Liao, R. G. Hulet, Science 311, 503(2006).[5] G. B. Partridge, W. Li, Y. A. Liao, R. G. Hulet, M. Haque, and H. T. C. Stoof,Phys. Rev. Lett. 97, 190407 (2006).[6] C. Lobo, A. Recati, S. Giorgini,S. Stringari, Phys. Rev. Lett. 97, 200403 (2006);