Vous êtes sur la page 1sur 6

Noname manuscript No.

(will be inserted by the editor)

Vortex line density uctuations of quantum turbulence


Shoji Fujiyama Makoto Tsubota

Received: date / Accepted: date

Abstract We investigate the vortex line density uctuations of quantum turbulence. The scenario of quantum turbulence experimentally suggested by the Lancaster group is conrmed in the numerical simulation. The spectrum of the vortex line density uctuations with respect to frequency obeyed a -5/3 power law, which is consistent with the experiment of the Lancaster group. Based on the argument of time scales experienced by vortex rings with different sizes and on the power spectrum, the connection between self-similar structure of the vortex tangle and the power spectrum is discussed. Keywords numerical simulation vortex line density uctuations power spectrum PACS 67.30.he 67.30.hb 47.32.C- 47.37.+q

1 Introduction The universality of quantum turbulence has attracted considerable interests in recent years. Superuid, in which quantum turbulence exists, has special hydrodynamic features such as lack of viscosity and quantization of vorticity. The former one provides the different mechanism for the transition from laminar to turbulent state. The latter one distinguishes quantum turbulence from classical turbulence in the way that the vorticity can be dened without ambiguity, and simplies the mathematical descriptions of turbulence properties. Beyond these hydrodynamic differences, however, quantum and classical turbulence are connected with universal laws. For example, Kolmogorov law, a statistical law found in classical turbulence, has been conrmed in quantum turbulence both experimentally and numerically [15]. Furthermore, it is conrmed that a drag force on an oscillating object that creates quantum turbulence exhibits the similarity with the classical one [6, 7]. Recently, the experiments on quantum turbulence, which suggests another possibility of the universality, were conducted. The Lancaster group [8] measured the vortex line density uctuations of quantum turbulence in superuid 3 He at ultimately low temperatures where
Shoji Fujiyama Makoto Tsubota Department of Physics, Osaka City University, Sumiyoshi-Ku, Osaka 558-8585, Japan E-mail: fujiyama@sci.osaka-cu.ac.jp

mutual friction can be neglected. They showed that the spectrum of the line density uctuations |L( f )|2 with respect to frequency obeys a power law with exponent -5/3. On the other hand, Roche et al. [9] performed the second sound measurement of the vortex line density uctuations in superuid 4 He at relatively high temperatures where the normal uid component signicantly remains. They also conrmed the spectrum of the line density uctuations obeys the same power law. At such high temperatures, the turbulences in superuid and the normal uid components are tightly locked by mutual friction in length scale larger than the mean vortex line spacing of vortex tangle. It is remarkable that the same power spectrum was obtained in different temperature regimes, and these experiments suggest that a universal feature is underlying over these turbulences. Taking exponent -5/3 as a key feature for the universality, in this paper, we report the numerical simulation supporting the experiments of the Lancaster group [8, 10, 11], and draw some conclusions from the power law. Section 2 consists of three parts: the model describing the experiments of the Lancaster group, the formulation for calculating the vortex dynamics, and the obtained results. In Sect. 3, the power law is linked with a self-similar structure of the vortex tangle, based on the vortex ring emission from the vortex tangle.

2 Numerical Simulation Model: We start from modelling the experiments of the Lancaster group [8, 10, 11]. They created quantum turbulence by transversely oscillating a grid at T0.2 Tc where a normal uid component is negligible. According to [12], remnant vortices are pinned on the grid. An oscillatory ow around the grid induces Kelvin waves on the remnant vortices when the frequency matches the resonance of the vortices, resulting in the amplication of Kelvin waves. The elongated vortices nally reconnect to themselves and produce vortex rings, leaving the original vortices on the grid. The production of vortex rings continues as long as the grid keeps oscillating. It is expected that the production rate of the vortex rings is proportional to the grid velocity. At low grid velocity, the vortex rings propagate independently and cause no turbulence. At high grid velocity, in contrast, the vortex rings are no longer independent, and collide with and reconnect to each other. The reconnection of two rings forms a larger ring with low propagation velocity. The larger vortex ring intercepts subsequent rings, and nally forms a vortex tangle. First we conrm this scenario in numerical simulation and establish the equilibrium state of the vortex tangle. The scheme of the simulation is as follows. We set a numerical cell with periodic boundaries in transverse directions and with open ends in longitudinal direction. The cross section of the cell is 200 m 200 m and the length is 600 m. To mimic the production of vortex ring, we inject vortex ring from the left side of the cell (see Fig. 1). The size of the injected rings is 20 m in diameter, which is estimated from the dispersion relation of Kelvin wave using the grid oscillation frequency 1300 Hz. They are injected at random position and at random angle within a 20 cone around the forward direction. The time interval ! of the injection is 2 ms. Formulation: Vorticity in B phase of superuid 3 He is quantized with the circulation quantum " = h/2m3 , where h is Plancks constant and m3 is mass of 3 He atom. Core size a0 of a quantized vortex is 10 nm, which is far smaller than any length scale appearing in this paper, thus we accept the vortex lament model. Since the experiments were performed at low temperatures enough to neglect normal component, we omit the contribution of mutual friction. In the absence of mutual friction, a vortex moves with superuid, so the vortex

10 vortex line length density (m-2)

108

0 1 Time from switching on grid (s)


Fig. 1 Snapshot of the simulation. t = 20 ms on the top, t = 250 ms on the middle, and t = 500 ms on the bottom Fig. 2 Vortex line length density

dynamics is equivalent to calculating the superuid velocity at every point on the vortex. In the present case, superuid velocity consists purely of the velocity produced by the vortices themselves. The velocity on a position r by the vortex laments is described in the form of the Biot-Savart integration r v # (r ) =

" 4$

where s 1 indicates a position vector on a vortex, and the integral is performed all over the vortices L . The velocity eld in (1) diverges when s1 r since we adopt the vortex lament model. To evade this difculty, we take cut-off parameter a0 . Calculating the velocity of each point on the vortices leads us to follow the trajectory of the vortices. It is known that two vortices reconnect when they become close and cross each other [13]. In the vortex lament approximation, however, the structure of the vortex core is neglected, so such a process needs to be handled manually. The reconnection process is included in our numerical code in the way that two vortices reconnects when they become close within the numerical resolution [14]. Because of the numerical resolution, Kelvin waves of wavelength smaller than the resolution is neglected, and vortex rings smaller than the resolution are removed from the dynamics. Result: Figure 1 is the time evolution of the vortex dynamics. As soon as several vortex rings are injected, they reconnect to themselves, forming larger rings (top of Fig. 1). Larger vortex rings with the low propagation velocity trap subsequent vortex rings (middle of Fig. 1), ending up with vortex tangle (bottom of Fig. 1). The vortex line density L(t) is plotted as a function of time in Fig. 2. After 0.5 s, the vortex line density reaches a statistically equilibrium state at 6 108 m2 . Looking at the line density of the experiment 1 108 m2 [11], we can conclude the vortex tangle in numerical simulation is established as dense as that of the experiments. Since we obtain an equilibrium state from 0.5 s to 1.5 s, we conduct Fourier transform of the vortex line density with respect to frequency (Fig. 3). The spectrum is dened as

s s (s 1 r ) ds 1 , 3 s |s 1 r |

(1)

10 Line Density Power Spectrum (m-4 Hz-1) 10

15 14

1013 1012 10 10
11 10

109 10
8

10 100 Frequency (Hz)


Tint 0

1000

Fig. 3 Spectrum |L( f )|2 of the vortex line density uctuations with respect to frequency. The spectrum is plotted on a log-log scale. Dashed line indicates % f 5/3

L( f ) =

ei2$ f t L(t) dt, Tint

(2)

where Tint is a time period in which the line density uctuations are measured. The power spectrum agrees well with a f 5/3 prole, consistent with the experiment of the Lancaster group.

3 Discussion and Conclusion The origin of the line length density uctuations can be clasied as follows: 1. 2. 3. 4. 5. length change by the reconnection (inducing Kelvin waves) injection of vortex rings escape of vortex rings from the computational cell elimination of vortex rings smaller than the computational resolution expansion and contraction of vortices due to the velocity eld created by vortices.

The vortex rings are injected in the time interval 2 ms which corresponds to 500 Hz in frequency. Thus the injection of vortex rings contributes only on the peak at 500 Hz, not on the entire power spectrum. In the simulation, vortex rings with various sizesfrom the computational resolution to the size of the cellescape from the cell, so the escape of vortex rings is considered to affect the change of the line length most. In this paper, we pick up the contribution of the vortex rings that escape from the computational cell from the contributions listed above. The emergence of the power spectrum implies the existence of a self-similar structure with respect to time. Focusing on the vortex rings, we draw the conclusion that the power

spectrum is linked with the self-similar structure of the vortex tangle as follows. Each amplitude of the spectrum means how many vortex rings are emitted from the cell with the corresponding frequency. Since the propagation velocity of the ring with the radius R is written as vring " /R and the turnover time [15] of the vortex ring is R/vring R2 /" , the smaller vortex ring evolves in smaller time scale. Thus large vortex rings contribute to the amplitude of low frequency in the spectrum, and vice versa. Here, let us consider the number density of vortex rings n(R) with the radius R R + dR. Because the number of the emitted vortex rings is considered to be proportional to n(R), the decrease of the vortex length density is described as |cem n(R)dR 2$ R| Tint L( f )d f , (3)

where cem is a dimensionless constant and Tint is the time period in which the line density uctuations are measured.1 Since |L( f )|2 % f 5/3 , (3) is rewritten as n(R) = d f A f 5/6 , dR cem 2$ R (4)

where A is a normalisation constant with dimension [L2 T 2/3 ]. As the frequency is given by f " /R2 [16], then Eq. (4) becomes n(R)

" 1/6 A R7/3 . 2$ cem Tint

(5)

The time period Tint should not be involved in the nature of the structure, thus A should be written as B Tint to cancel out, where B has the dimension [L2 T 1/6 ]. The number density n(R) should have something to do with the time interval of vortex injection ! , which should be involved in B. Since the nature of vortex tangle is determined by " , B can be constructed with ! and " , then one obtains B = C" 1 ! 5/6 from dimensional analysis, where C is a dimensionless constant. Finally Eq. (5) becomes n(R) = C R7/3 . 2$ cem ("! )5/6 (6)

Equation (6) represents the self-similar structure in terms of the number density with respect to vortex ring radius, having exponent -7/3. Although we only consider the contribution of the vortex ring emission and the argument is based on the time scale in which the vortex rings experience, it is worth remarking the connection between the uctuation spectrum and the self-similar structure of quantum turbulence. In conclusion, the grid experiments conducted by the Lancaster group are modelled and the numerical simulation is performed. As a result, the scenario proposed by them is conrmed, and the vortex line length density of the numerical simulation qualitatively agrees with that of the experiments. Furthermore the vortex line density uctuations in equilibrium quantum turbulence are calculated and found to obey the power spectrum |L( f )|2 % f 5/3 , consistent with the experiment. Finally, considering the contribution of the vortex ring emission to the power spectrum and assuming time scales experienced by the vortex rings, we formulate the number density n(R), which unveils the connection between the spectrum and the self-similar structure of quantum turbulence.
1

The time period Tint appears in the formulation because the denition (2) of the spectrum includes T . int

6 Acknowledgements S. F. acknowledges the support of JSPS Research Fellowships for Young Scientists (Grant No. 217762). M. T. acknowledges the supports of Grant-in-Aid for Scientic Research from JSPS (Grant No. 21340104) and Grant-in-Aid for Scientic Research on Priority Areas (Grant No. 17071008) from MEXT.

References
C. Nore, M. Abid, M. E. Brachet, Phys. Rev. Lett. 78, 3896 (1997) J. Maurer, P. Tabeling, Europhys. Lett. 43, 29 (1998) S. R. Stalp, L. Skrbek, R. J. Donnelly, Phys. Rev. Lett. 82, 4831 (1999) T. Araki, M. Tsubota, S. K. Nemirovskii, Phys. Rev. Lett. 89, 145301 (2002) M. Kobayashi, M. Tsubota, Phys. Rev. Lett. 94, 065302 (2005); J. Phys. Soc. Jpn. 74, 3248 (2005) L. Skrbek, W. F. Vinen, in Progress in Low Temperature Physics, ed. by W. P. Halperin and M. Tsubota, vol. 16 (North-Holland, Amsterdam, 2008), pp. 195-246 7. S. Fujiyama, M. Tsubota, Phys. Rev. B 79, 094513 (2009) 8. D. I. Bradley, S. N. Fisher, A. M. Gu nault, R. P. Haley, S. OSullivan, G. R. Pickett, V. Tsepelin, Phys. e Rev. Lett. 101, 065302 (2008) 9. P. E. Roche, P. Diribarne, T. Didelot, O. Francais, L. Rousseau, H. Willaime, Europhys. Lett. 77, 66002 (2007) 10. D. I. Bradley, D. O. Clubb, S. N. Fisher, A. M. Gu nault, R. P. Haley, C. J. Matthews, G. R. Pickett, V. e Tsepelin, K. Zaki, Phys. Rev. Lett. 95, 035302 (2005) 11. D. I. Bradley, D. O. Clubb, S. N. Fisher, A. M. Gu nault, R. P. Haley, C. J. Matthews, G. R. Pickett, V. e Tsepelin, K. Zaki, Phys. Rev. Lett. 96, 035301 (2006) 12. R. J. Donnelly, in Quantized Vortices in Helium II (Cambridge University Press, Cambridge, 1991) 13. J. Koplik, H. Levine, Phys. Rev. Lett. 71, 1375 (1993) 14. M. Tsubota, T. Araki, S. K. Nemirovskii, Phys. Rev. B 62, 11751 (2000) 15. U. Frisch, in TURBULENCE (Cambridge University Press, Cambridge, 1995) 16. W. F. Vinen, J. J. Niemela, J. Low Temp. Phys. 128 167 (2002) 1. 2. 3. 4. 5. 6.

Vous aimerez peut-être aussi