Vous êtes sur la page 1sur 8

Physica D 237 (2008) 21952202 www.elsevier.

com/locate/physd

Is the Reynolds number innite in superuid turbulence?


Carlo F. Barenghi
School of Mathematics and Statistics, Newcastle University, Newcastle upon Tyne NE1 7RU, UK Available online 16 January 2008

Abstract Superuidity, the hallmark property of quantum uids (e.g. liquid helium, atomic BoseEinstein condensates, neutron stars), is characterised by the absence of viscosity. At temperatures which are low enough that thermal excitations can be neglected, liquid helium can be considered a perfect superuid, and one would expect that superuid turbulence were dissipationless because the Reynolds number is innite. On the contrary, experiments show that helium turbulence decays, even at these low temperatures. The solution of this apparent puzzle lies in subtle but crucial differences between a superuid and a classical Euler uid. c 2008 Elsevier B.V. All rights reserved.
PACS: 67.40.Vs; 47.27.-i; 03.75.Lm; 67.57.Fg Keywords: Superuid helium; Euler; Vortices; Turbulence

1. Introduction The motivation behind this article is the relation between the concept of inviscid, incompressible Euler uid (as in traditional textbooks of uid mechanics) and superuid helium. This relation is particularly intriguing in view of recent experiments [1,2] concerning the nature of turbulent dissipation near absolute zero. A second motivation is that research in superuidity, quantised vorticity and turbulence [3] has gone beyond the traditional context of liquid helium (the two isotopes 4 He and 3 He) and now includes ultra-cold atomic gases [4] and neutron stars [5]. Clearly the three physics communities which are involved (condensed matter physics, atomic physics and astrophysics) should benet from more contact with classical uid mechanics. The third motivation is the recognition of the great potential of cryogenics helium to produce turbulence at very large Reynolds numbers [6] and Rayleigh numbers [7], which is making classical uid mechanicists interested in issues of turbulence at very low temperatures. An example of the successful interaction between classical uid mechanicists and low temperature physicists is the recent application of the classical Particle Image Velocimetry method in liquid helium [8,9].
Tel.: +44 191 222 7307; fax: +44 191 222 8020.

E-mail address: C.F.Barenghi@ncl.ac.uk. 0167-2789/$ - see front matter c 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.physd.2008.01.010

Since the best known superuid is still the common isotope most of the following considerations will refer directly to it, unless stated otherwise. Helium is a gas at room temperature. To turn it into a liquid it is necessary to cool it down to approximately 4 K degrees above absolute zero. Upon further cooling a phase transition occurs at the critical temperature Tc = 2.1768 K (at saturated vapour pressure), marking the onset of BoseEinstein condensation [10]. Below Tc liquid helium is a quantum uid called helium II. Its strange properties are well described by the two-uid model of Landau and Tisza [11]. According to this model, helium II is the intimate mixture of two-uid components, the normal uid and the superuid. The normal uid consists of thermal excitations (similar to phonons in a solid) which carry the entire entropy and viscosity of the liquid. The superuid is related to the BoseEinstein condensate [10] and has zero entropy and viscosity. Hereafter, for the sake of simplicity, I shall ignore the difference between the superuid fraction and the condensate. The feature which is crucial in our problem is that the normal-uid fraction decreases very rapidly with temperature, and below 1 K, at temperatures which can be easily reached experimentally, helium II can be considered a pure superuid. The normal uid can be neglected in the case of 3 He too; although the critical temperature Tc for the onset of superuidity is much lower (few mK) than in 4 He,
4 He,

2196

C.F. Barenghi / Physica D 237 (2008) 21952202

turbulence experiments can be performed at very small values of T /Tc [12]. In classical uid dynamics the ratio of the magnitudes of inertial and viscous forces acting on a uid is the Reynolds number Re = UD/, where D and U are the characteristic length scale and velocity scale of the ow and is the kinematic viscosity of the uid. If Re 1 inertial forces dominate and the ow is turbulent. Thus Re measures the intensity of the turbulence. The kinematic viscosity of liquid helium is almost two orders of magnitude less than waters, so it is relatively easy to make liquid helium turbulent. The issue which we address is what happens at temperatures so small that helium II is a pure superuid. Since a superuid has zero viscosity, the Reynolds number of helium II ow is innite and one would expect a form of dissipationless turbulence. On the contrary, experiments [1,2] show that turbulence decays, even at temperatures as low as few mK, which is puzzling. The natural question is: what is the ultimate energy sink near absolute zero? To answer the question and to solve the puzzle we must identify subtle but crucial differences between a superuid and a classical Euler uid. 2. The NLSE Quantum mechanics constrains the rotational motion of the superuid to vortex lines [13]. To understand the superuid vortex structure it is convenient to consider the nonlinear Schroedinger equation (NLSE) for a weakly-interacting BoseEinstein condensate [14]: ih = t 2m h2
2

Fig. 1. Straight vortex line (a) and Kelvin wave (b).

3. Fluid dynamics interpretation of the NLSE The uid dynamics interpretation of the NLSE is based on the Madelung transformation = ReiS where R and S are the amplitude and the phase of . Substitution into Eq. (1) yields the following classical continuity equation for the superuid density s = m R 2 : s + t (s vs ) = 0, (4)

+ V0 ||2 E 0 ,

(1)

and the following (quasi) Euler equation for the superuid velocity vs = (h /m) S: s vs j vs j + vsk t xk = jk p + , x j xk (5)

where = (x, t) is the macroscopic single-particle wavefunction, V0 the (delta-function repulsive) potential of interaction between the bosons, m the mass of one boson, E 0 the energy increase upon addition of a boson, h Plancks constant and h = h/(2 ) The total energy of the condensate, dened as E tot = 2m h2 | |2 d3 x + V0 2 ||4 d3 x, (2)

where vs j is the jth component of vs . The pressure, p, and the quantum stress, jk , are dened as p= V0 2 , 2m 2 s jk = h 2m
2

2 ln s . x j xk

(6)

is a constant of motion. In the case of a trapped atomic condensate, a term of the form Vext must be added to the right-hand side of Eq. (1), where Vext is a conning harmonic potential [10]. In using the NLSE model one should keep in mind that the NLSE has quantitative predicting power in the case of atomic BoseEinstein condensates, but is only a qualitative model of helium II. The reason is that helium II is a liquid, not a weaklyinteracting gas; as far as our discussion is concerned, however, the difference should not be essential. Our aim is to compare vortex lines solutions of the NLSE model with vortex lines solutions of the classical incompressible Euler equations vs 1 + (vs )vs = t s p, vs = 0. (3)

Note that without the quantum stresses term, Eq. (5) would be the classical Euler equation (3). 4. Vortex lines and vortex tangles The solution of the NLSE which corresponds to a vortex aligned along the z-axis, as in Fig. 1(a), is obtained by letting S = in cylindrical coordinates (r, , z). The resulting velocity is vs = h , mr (7)

where is the unit vector along the direction. This is the textbook velocity eld of a vortex line in a classical Euler uid, shown in Fig. 2. As we shall see in Fig. 4, the corresponding density is zero on the axis of the vortex.

C.F. Barenghi / Physica D 237 (2008) 21952202

2197

Fig. 2. Superuid velocity eld around a vortex line.

Let C be any path of integration around the vortex line. Then the circulation around C is =
C

vs d = , 103 cm2 /s

(8)
Fig. 3. Tangle of vortex laments in a periodic box computed using the approach of Schwarz.

where the constant = h/m is called the quantum of circulation. Since S = 0, a superuid vortex line has zero vorticity but nite circulation: the core is a microscopic hole, surrounded by a macroscopic region of potential ow. A sinusoidal, or, more in general, a helical displacement of the vortex line away from the straight position is called Kelvin wave, as shown in Fig. 1(b). The wave rotates with angular frequency k 2 , where k is the wavenumber and = 2/k the wavelength. 5. Turbulence Vortex systems can be laminar or turbulent. If the vessel which contains helium II rotates at constant angular velocity (larger than some small critical value), the superuid breaks into vortex lines which are aligned along the axis of rotation [13] forming a lattice of areal density n = 2 /. Essentially the superuid mimics the classical vorticity 2 of a rotating Euler uid by making n vortices (per unit area) carrying one quantum of circulation each. Disordered systems of vortex lines are easily created by making the helium turbulent, This can be done, for example, by imposing a heat ow [15,16] or by agitating the liquid helium with towed grids [17], rotating propellers [18,19], vibrating grids [1,2] or wires [20]. Numerical simulations indicate that superuid turbulence manifests itself as a disordered tangle of vortex laments, as shown in Fig. 3. The intensity of the turbulence is usually characterised by the vortex line density L (vortex length per unit volume). The quantity = L 1/2 is the average distance between the vortex lines in the tangle. The nature of superuid turbulence and its similarities with classical turbulence is a problem which is attracting attention, as reviewed by Vinen and Niemela [21]. The particular aspect which is relevant to our problem is the experimental observation that turbulence generated at very low temperatures decays [1,2] or diffuses away [20], which is at rst surprising, given that

Fig. 4. Superuid density near a vortex line.

the superuid is usually interpreted as an inviscid Euler uid. To understand these experiments, it is necessary to understand the difference between a superuid and a classical Euler uid, expressed by the last term of Eq. (5). 6. Euler uid vs superuid In the classical Euler case we are free to assume that the uids density is constant. In the NLSE model the density near the vortex must be determined self-consistently by solving the NLSE for the amplitude of . The result is shown in Fig. 4. The density drops from its bulk value (away from the vortex) to zero (on the axis of the vortex) over a characteristic distance a 108 cm called the vortex core radius; this quantity is of the order of the superuid healing length = h / m E 0 . Fig. 4 shows that the superuid vortex core is hollow, and the presence of a vortex makes a region of liquid helium to become multiplyconnected. This means that the diverging velocity eld, vs for r 0 predicted by Eq. (7) is not a problem, because, in the same limit, s 0, whereas the classical Euler model of a vortex line breaks down on the axis.

2198

C.F. Barenghi / Physica D 237 (2008) 21952202

Fig. 5. Schematic vortex reconnection.

In general, the total circulation around an arbitrary region is preserved by the evolution until a vortex line crosses C, which causes a change of by one quantum . Note that the classical Kelvins theorem that is constant does not apply because when the vortex meets C the density becomes zero. Thus, although both the classical Euler uid and the superuid conserve the total energy, there is an important difference: in the classical Euler case the circulation cannot change and the topology of the ow is determined by the initial condition. In the NLSE case, on the contrary, vortex reconnections are possible (see Fig. 5), as rst demonstrated by Koplik and Levine [22], and the topology can change while conserving the total energy. The existence of vortex reconnections in a classical viscous uid is well known [23,24]. However, whereas in the NavierStokes equation reconnections are controlled by the viscosity and dissipate energy, in the NLSE reconnections are controlled by the quantum stress and maintain the total energy constant. A striking consequence of the ability of superuid vortices to reconnect is the following. Consider a vortex tangle initially conned within a region of radius R in innite space. The tangle consists of a number of vortex loops of any size and orientation. If the loops obeyed ordinary Euler dynamics, the initial linkage between the loops could not change, because helicity, hence the linking number, must be conserved. On the contrary, superuid vortex loops can reconnect and undergo the following unusual process of diffusion in space [25]. Suppose that a vortex reconnection creates a vortex loop which is smaller than the average distance between loops, is located near the boundary of the tangle and has circulation such that its translational velocity points out of the tangle. The small loop can escape to innity with very little probability of being re-absorbed into the tangle by another vortex reconnection, because its speed is inversely proportional to its size and most other loops are larger and slower. Once the small loop has escaped, the total vortex length of the tangle has been reduced, hence the typical spacing of loops has increased, which favours the escape of a second loop, and so on. In this way the tangle evaporates into loops, thus spreading in space. This scenario [25] is consistent with what is seen in the experiments [20]. Vortex nucleation is another phenomenon which is possible in a superuid but not in a classical Euler uid. Typically nucleation occurs near a boundary where 0, for example in the equatorial region of a rapidly moving ion [26,27], or at the edge of a trapped condensate [28] or at an intense rarefaction sound pulse [29] or at a dark soliton [30].

Vortex reconnections are special events which arise from the presence of the quantum stress Eq. (6), the term which makes Eq. (5) to differ from Eq. (3). Let D be the typical length scale of a ow. The ratio of the pressure term and the quantum stress term scales as h 2 /(m E 0 D 2 ) and becomes unity only if D . Thus the quantum stress term is important only at scales smaller than the vortex core radius, Away from vortices s is constant and Eq. (5) reduces to the classical Euler equation. The smallest macroscopic ow scale D in a superuid vortex tangle is of the order of the average distance L 1/2 between vortex lines; typical experimental values are 103 104 cm, orders of magnitude bigger than a 108 cm. We conclude that, apart from vortex reconnecting events, in most of the uid and at most times the quantum stress term in Eq. (5) is negligible, and the superuid obeys classical Euler dynamics. This consideration justies the vortex lament approach to superuid turbulence of Schwarz [31]. He modelled superuid vortices as space curves of innitesimal thickness which move under the velocity eld which each line induces upon each other. Let x be a point along a lament and x the derivative with respect to arclength. In the absence of the normal uid, the velocity of the vortex at x is given by the BiotSavart integral dx = dt 4 (x z) dz, |x z|3 (9)

which expresses Eulers dynamics in integral form. A convenient approximation to Eq. (9) which is often used is the Local Induction Approximation (LIA), dx/dt x x where = /(4 ) ln(1/(|x |a)). To implement Schwarzs approach, the computer code must introduce vortex reconnections when two vortex lines come sufciently close to each other. 7. Dissipation of kinetic energy at absolute zero Using the NLSE model, Nore et al. [32], made a discovery which shed light onto the nature of dissipation at absolute zero. They computed the temporal evolution of an arbitrary vortex conguration (a TaylorGreen ow) which evolved into a vortex tangle, and noticed that, while the total energy remained constant, incompressible kinetic energy was transformed into compressible sound energy. They also found that the energy spectrum is consistent with the k 5/3 classical Kolmogorov spectrum observed experimentally [18] above and below Tc . Further work using both the vortex lament model [33] and the NLSE [34] gave spectra consistent with the classical Kolmogorov law. 7.1. Sound radiation by vortex motion Generation of sound by vortex motion is a known classical effect. In the case of quantised vorticity, the effect can be accurately investigated in the controlled experimental conditions of trapped BoseEinstein condensates. In particular, Parker et al. [35] suggested to add a Gaussian dimple at the bottom of the harmonic trapping potential. By tuning the depth of the dimple, the sound which is radiated by the vortices can escape the dimple, whereas the vortices remain trapped in it;

C.F. Barenghi / Physica D 237 (2008) 21952202

2199

Fig. 6. Density waves created by an a single vortex which orbits a dimple trap (x and y coordinates are in units of ).

Fig. 8. Left: interaction of vortexantivortex pair (coming from the top of the gure) with a third vortex. Right: corresponding density wave; the trajectory of the vortex pair is superimposed.

7.2. The Kelvin wave cascade A mechanism to transfer energy towards small scales is the reconnection-driven cascade of smaller and smaller loops, eventually to thermal excitations [38], proposed by Feynman [39], later demonstrated numerically [40] and also modelled by a master-equation approach [41]. Here I shall describe other recently discovered mechanisms which can create very short length scales. The rst is the Kelvin wave cascade [42]. Fig. 9 shows that when vortex lines collide and reconnect, the resulting reconnection cusps generate large amplitude Kelvin waves (compared to the wavelength) which interact with each other and create further Kelvin waves of shorter and shorter wavelength. At temperatures above 1 K the friction with thermal excitations would damp out the Kelvin waves, feeding energy into the normal uid, but in the low temperature regime this energy sink does not exist. Kinetic energy is thus shifted to higher and higher wavenumbers k as in the energy spectrum shown in Fig. 10, until k is large enough that sound can be efciently radiated away. The Kelvin wave cascade is thus similar to the Richardson cascade of classical Kolmogorov turbulence. The important difference is that the energy sink is acoustic for the Kelvin cascade and viscous for the Richardson cascade. Studying the Kelvin wave cascade numerically, it is important to realize that the LIA is not suitable because it conserves length, not energy (like the exact BiotSavart law). There are still many open questions on the Kelvin wave cascade; current work attempts to determine the precise power law of the Kelvin cascade energy spectrum [4345] and what happens in the transition regime [46,47] between the Richardson cascade (k 1/) and the Kelvin wave cascade (k 1/). 7.3. Reconnection pulses Another more direct mechanism to turn kinetic energy into sound energy was found by Leadbeater et al. [48] by studying the collision of vortex rings in the NLSE model, see Fig. 11. They found that at the point of reconnection a rarefaction pulse is created, expanding in size and becoming more shallow as it moves away. Fig. 12 shows density proles of the same pulse at

Fig. 7. Density waves created by a vortexvortex pair.

this allowed Parker et al. to relate the vortex acceleration to the sound energy. Fig. 6 shows the dipolar radiation pattern generated by an off-centre vortex which moves along an orbit in the dimple; the quadrupolar pattern emitted by a corotating vortexvortex pair is shown in Fig. 7. Fig. 8 shows a two-dimensional vortexantivortex pair which travels towards an isolated vortex [36]. The vortex pair is deected, and the sound which is generated by the interaction is visible as a density ripple. After the interaction, the separation between the two vortices of the pair is reduced because some kinetic energy was turned into sound energy. In three dimensions, sound radiation can be emitted by rotating Kelvin waves [37]. The power which is radiate per unit length by a co-rotating vortexvortex pair separated by the distance is proportional to 6 . Taking (where is the average intervortex spacing deduced from the observed vortex line density L) we nd that sound radiation by moving vortices cannot account for the observed decay of superuid turbulence [21]: a much shorter length scale is necessary to radiate enough sound and explain the measurements.

2200

C.F. Barenghi / Physica D 237 (2008) 21952202

Fig. 11. Collision of vortex rings in the NLSE model. The time sequence shows the interaction of two rings, which are initially slightly offset with respect to each other.

Fig. 9. Collision of four vortex rings. t = 0: initially the rings are set to travel against each other; t = 0.59: cusps created by the vortex reconnections; t = 0.069: the cusps relax and launch large amplitude Kelvin waves along the vortices. t = 0.129: generation of Kelvin waves of larger and larger wavenumber.

Fig. 10. Energy spectrum before the collision of the four vortex rings of the previous gure, and after the collision. Note the saturated energy spectrum.

Fig. 12. Density along the z-axis for a collision of two vortex rings initially offset with respect to each other. The eleven curves correspond to different times. Just before the reconnection (bottom curve) the density is uniform except for a slight increase near the origin indicating the approaching rings. At the reconnection a rarefaction pulse is created in which the depth drops to zero. As the pulse moves away, the depth decreases.

different times following the vortex reconnection. Immediately after the reconnection the pulse is short (few times the healing length ) and intense (the density drops to zero). Later the pulse spreads out and becomes more shallow as it moves away. The kinetic energy which is transformed into sound energy of the pulse depends on the initial impact parameters, and it is maximum if the two rings collide head-on and destroy each other. Calculations involving a small number of interacting vortex rings [49] show that in general the Kelvin wave cascade and the rarefaction pulses are present at the same time. Fig. 13 shows that the total vortex length, which can be taken as a proxy for total kinetic energy, decreases with time. The sudden drops in length are caused by the creation of rarefaction pulses, and the oscillations are due to Kelvin waves. The relative importance of the Kelvin wave cascade and the rarefaction pulses depends on

the vortex line density L and how the vortex reconnection rate scales with L [50]. 8. Conclusions The vortex laments shown in Fig. 3 are similar to coherent vortex structures computed in classical turbulence [51,52]. The similarity of superuid turbulence to classical turbulence seems to go beyond pictures: the same laws of vortex dynamics are present, and the same k 5/3 Kolmogorov spectrum is observed. In many respects superuid turbulence is simpler than classical turbulence: superuid vortices are discrete laments with the same circulation and the same microscopic core structure, whereas in classical turbulence vorticity is continuous and eddies can be of any size and strength; moreover, the ow around each superuid vortex line is potential, whereas in

C.F. Barenghi / Physica D 237 (2008) 21952202

2201

vortex reconnections) against results obtained using the NLSE, in order to create a more realistic model of superuid turbulence. Acknowledgments I am grateful to W.F. Vinen for stimulating discussions. This work is supported by EPSRC grants GR/T08876/01 and EP/D040892/1. References
[1] S.I. Davis, P.C. Hendry, P.V.E. McClintock, Decay of quantised vorticity in superuid He-4 at mK temperatures, Physica B 280 (2000) 4344. [2] D.I. Bradley, D.O. Clubb, S.N. Fisher, A.M. Guenault, R.P. Haley, C.J. Matthews, G.R. Pickett, V. Tsepelin, K. Zaki, Decay of pure quantum turbulence in superuid He-3-B, Phys. Rev. Lett. 96 (2006) 035301. [3] C.F. Barenghi, R.J. Donnelly, W.F. Vinen (Eds.), Quantized Vortex Dynamics And Superuid Turbulence, Springer Verlag, 2001. [4] N.G. Parker, B. Jackson, A.M. Martin, C.S. Adams, Vortices in BoseEinstein condensates. arXiv:0704.0146, 2007. [5] N. Andersson, Modelling the dynamics of superuid neutron stars, Astrophys. Space Sci. 308 (2007) 395402. [6] R.J. Donnelly, Ultra-High Reynolds Number Flows Using Cryogenic Helium: An Overview, in: R.J. Donnelly, K.R. Sreenivasan (Eds.), UltraHigh Reynolds And Rayleigh Number Flows, Springer Verlag, 1998, pp. 128. [7] J.J. Niemela, L. Skrbek, K.R. Sreenivasan, R.J Donnelly, Turbulent convection at very high Rayleigh numbers, Nature 406 (2000) 439439. [8] T. Zhang, S.W. Van Sciver, Nature Phys. 1 (2005) 3638. [9] G.P. Bewley, D.P. Lathrop, K.R. Sreenivasan, Superuid helium: visualization of quantized vortices, Nature 441 (2006) 588588. [10] C.J. Pethick, H. Smith, Bose-Einstein Condensation in Dilute Gases, Cambridge University Press, Cambridge, 2001. [11] L.D. Landau, E.M. Lifschitz, Fluid Mechanics, Pergamon Press, 1987. [12] A.P. Finne, T. Araki, R. Blaauwgeers, V.B. Eltsov, N.B. Kopnin, M. Krusius, L. Skrbek, M. Tsubota, G.E. Volovik, An intrinsic velocityindependent criterion for superuid turbulence, Nature 424 (2003) 10221025. [13] R.J. Donnelly, Quantized Vortices in Helium II, Cambridge University Press, Cambridge, 1991. [14] P.H. Roberts, N.G. Berloff, The nonlinear Schroedinger equation as a model of superuidity, in: C.F. Barenghi, R.J. Donnelly, W.F. Vinen (Eds.), Quantized Vortex Dynamics and Superuid Turbulence, Springer Verlag, 2001, pp. 235257. [15] W.F. Vinen, Mutual friction in a heat current in liquid helium II. Experiments in steady heat currents, Proc. Roy. Soc. A 240 (1957) 114127. [16] C.F. Barenghi, A.V. Gordeev, L. Skrbek, Depolarization of decaying counterow turbulence in He II, Phys. Rev. E 74 (2006) 026309. [17] M.R. Smith, R.J. Donnelly, N. Goldenfeld, W.F. Vinen, Decay of vorticity in homogeneous turbulence, Phys. Rev. Lett. 71 (1993) 25832586. [18] J. Maurer, P. Tabeling, Local investigation of superuid turbulence, Europhys. Lett. 43 (1998) 2934. [19] P.E. Roche, P. Diribarne, T. Didelot, O. Francais, L. Rousseau, H. Willaime, Vortex density spectrum of quantum turbulence, Europhys. Lett. 77 (2007). Art. No. 66002 2007. [20] S.N. Fisher, A.J. Hale, A.M. Guenault, G.R. Pickett, Generation and detection of quantum turbulence in Superuid He-3B, Phys. Rev. Lett. 86 (2001) 244247. [21] W.F. Vinen, J.J. Niemela, Quantum turbulence, J. Low Temp. Phys. 128 (2002) 167231; J. Low. Temp. Phys. 129 (2002) 213 (erratum). [22] J. Koplik, H. Levine, Vortex reconnection in superuid helium, Phys. Rev. Lett. 71 (1993) 13751378. [23] A.K.M. Hussain, Coherent structures in turbulence, J. Fluid Mech. 173 (1986) 303356.

Fig. 13. Total vortex length (which can be interpreted as measure of kinetic energy) vs time for the interaction of a small number of vortex rings.

classical turbulence eddies have arbitrary rotation curves; nally, the superuid is inviscid, whereas a classical uid is viscous. The classical theory of homogeneous, isotropic turbulence deals with incompressible uids. In superuid turbulence density changes are relevant only when one considers effects at length scales of the order of the vortex core radius (e.g., vortex reconnections) or at length scales bigger than the vortex core but still much smaller than the average intervortex distance (eg sound emission by Kelvin waves). At length scales larger than , superuid turbulence can be considered incompressible. Indeed, for k 1/, superuid turbulence and classical turbulence seem to obey the same Kolmogorov spectrum (although numerical calculations with better resolution are needed). It is the nature of the energy sink at high wavenumbers which is very different: viscous for the classical uid and acoustic for the quantum uid. In conclusion, the NLSE can be interpreted as a way to regularise the Euler equation, removing the singularities on the axes of vortices and allowing the vortices to reconnect. Superuid turbulence at very low temperatures can be interpreted as the turbulence of an incompressible, reconnecting Euler uid, in which the energy sink at very small scales is acoustic. The solution of our puzzle is that, although the superuid has zero viscosity, the Reynolds number is innite only nominally. Dissipation exists even in the limit of absolute zero: organised kinetic energy can be turned into disorganised sound energy. Vortex reconnections are the key to understand the puzzle, because they trigger both the Kelvin wave cascade and rarefaction pulses. The approach of Schwarz is convenient because it allows the calculation of more intense vortex tangles than the NLSE, but it is incompressible, so it does not include the acoustic loss of kinetic energy which is crucial at low T . However, the nite discretisation along the laments introduces a small, unavoidable energy sink. Future work should calibrate this numerical dissipation (particularly during

2202

C.F. Barenghi / Physica D 237 (2008) 21952202 [39] R.P. Feynman, Application of quantum mechanics to liquid helium, in: C.J. Gorter (Ed.), Progress in Low Temperature Physics, vol. I, NorthHolland, Amsterdam, 1955, pp. 1753. [40] M. Tsubota, T. Araki, S.K. Nemirovskii, Dynamics of vortex tangle without mutual friction in superuid 4 He, Phys. Rev. B 62 (2000) 1175111762. [41] S.K. Nemirovskii, Evolution of a network of vortex loops in He II: Exact solution of the rate equation, Phys. Rev. Lett. 96 (2006) 015301015304. [42] D. Kivotides, J.C. Vassilicos, D.C. Samuels, C.F. Barenghi, Kelvin wave cascade in superuid turbulence, Phys. Rev. Lett. 86 (2001) 30803083. [43] W.F. Vinen, M. Tsubota, A. Mitani, Kelvin-wave cascade on a vortex in superuid He-4 at a very low temperature, Phys. Rev. Lett. 91 (2003) 135301135304. [44] E. Kozik, B. Svistunov, Kelvin-wave cascade and decay of superuid turbulence, Phys. Rev. Lett. 92 (2004) 035301035304. [45] S. Nazarenko, Kelvin wave turbulence generated by vortex reconnections, JETP Lett. 84 (2007) 585587. [46] E. Kozik, B. Svistunov, Kolmogorov and Kelvin wave cascade in superuid turbulence at T = 0: What is in between? arXiv:condmat/0703047, 2007. [47] V.S. Lvov, S. Nazarenko, O. Rudenko, Bottleneck crossover between classical and quantum turbulence, Phys. Rev. B 76 (2007) 024520024529. [48] M. Leadbeater, T. Winiecki, D.C. Samuels, C.F. Barenghi, C.S. Adams, Sound emission due to superuid vortex reconnections, Phys. Rev. Lett. 86 (2001) 14101413. [49] M. Leadbeater, D.C. Samuels, C.F. Barenghi, C.S. Adams, Decay of superuid vortices due to Kelvin wave radiation, Phys. Rev. A 67 (2002) 015601. [50] C.F. Barenghi, D.C. Samuels, Scaling laws of vortex reconnections, J. Low Temp. Phys. 36 (2004) 281. [51] A. Vincent, M. Meneguzzi, The dynamics of vorticity tubes in homogeneous turbulence, J. Fluid Mech. 258 (1994) 245254. [52] M. Farge, G. Pellegrino, K. Schneider, Coherent vortex extraction in 3D turbulent ows using orthogonal wavelets, Phys. Rev. Lett. 87 (2001) 05450110545014.

[24] M.V. Melander, F. Hussain, Cross linking of 2 antiparallel vortex tubes, Phys. Fluids A 1 (1989) 633639. [25] C.F. Barenghi, D.C. Samuels, Evaporation of a packet of quantized vorticity, Phys. Rev. Lett. 89 (2002) 155302155305. [26] T. Winiecki, C.S. Adams, Motion of an object through a quantum uid, Europhys. Lett. 52 (2000) 257263. [27] T. Frisch, Y. Pomeau, S. Rica, Transition to dissipation in a model of superow, Phys. Rev. Lett. 69 (1992) 16441648. [28] M. Tsubota, K. Kasamatsu, M. Ueda, Vortex lattice formation in a rotating BoseEinstein condensate, Phys. Rev. A 65 (2002) 02360310236034. [29] N.G. Berloff, C.F. Barenghi, Vortex nucleation by collapsing bubbles in BoseEinstein condensates, Phys. Rev. Lett. 93 (2004) 090401090404. [30] B.P. Anderson, P.C. Haljan, C.A. Regal, D.L. Feder, L.A. Collins, C.W. Clark, E.A. Cornell, Watching dark solitons decay into vortex rings in a Bose-Einstein condensate, Phys. Rev. Lett. 86 (2001) 29262929. [31] K.W. Schwarz, Three-dimensional vortex dynamics in superuid 4He: Homogeneous superuid turbulence, Phys. Rev. B 38 (1988) 23982417. [32] C. Nore, M. Abid, M.E. Brachet, Kolmogorov turbulence in low temperature superows, Phys. Rev. Lett. 78 (1997) 38963899. [33] T. Araki, M. Tsubota, S.K. Nemirovskii, Energy spectrum of superuid turbulence with no normal-uid component, Phys. Rev. Lett. 89 (2002) 145301145303. [34] M. Kobayashi, M. Tsubota, Kolmogorov spectrum of superuid turbulence: Numerical analysis of the GrossPitaevksii equation with a small-scale dissipation, Phys. Rev. Lett. 94 (2005) 065302065305. [35] N.G. Parker, N.P. Proukakis, C.F. Barenghi, C.S. Adams, Controlled vortex sound interaction in Bose-Einstein condensates, Phys. Rev. Lett. 92 (2004) 160403160406. [36] C.F. Barenghi, N.G. Parker, N.P. Proukakis, C.S. Adams, Decay of quantized vorticity by sound emission, J. Low Temp. Phys. 138 (2005) 629634. [37] W.F. Vinen, Decay of superuid turbulence at a very low temperature: The radiation of sound from a Kelvin wave on a quantized vortex, Phys. Rev. B 64 (2001) 134520134523. [38] D.C. Samuels, C.F. Barenghi, Vortex heating in superuid helium at low temperatures, Phys. Rev. Lett. 81 (1998) 43814384.

Vous aimerez peut-être aussi