Anda di halaman 1dari 17

Journal of Low Temperature Physics, Vol. 93, Nos.

5/6, 1993

Vortex Dynamics at the Superfluid


~-Transition

Gary A. Williams

Physics Department
University of California
Los Angeles, CA 90024

A vortex-ring theory of the superfluid 4He ~-transition is extended to


include the dynamics of the transition. The response of the vortices to an
oscillating superflow is found by solving the Fokker-Planck equation. This
allows a calculation of the superfluid relaxation time, which is in
agreement with Landau-Khalatnikov theory and with dynamic scaling. At
high frequencies the transition becomes broadened, with both the
superfluid density and the dissipation remaining finite at and above T~.
Comparison is made to earlier theories that use high-temperature
expansions and renormalization-group expansions. Applications to other
subjects such as mutual friction, high-Tc superconductors, and rapidly
quenched systems are briefly discussed.
PACS numbers: 67.40 Fd, 67.40 Vs, 67.40 Kh, 64.60 Ht.

1. I N T R O D U C T I O N

One of the most fundamental aspects of liquid 4He is the decrease


of the superfluid fraction to zero at the transition temperature T x.
Surprisingly, there has been no coherent physical picture of this process
over the entire temperature range up to T x. At temperatures well below
T x the dependence of Ps is fairly well described by the Landau theory of
roton excitations. 1 However, the precise physical nature of the roton
excitations has never been very clear, and fits of the theory to experimental

1079
0022-2291/93/1200-1079507.00/0~ 1993 Plenum Publishing Corporation
1080 G. Williams

values of the superfluid density require roton energy gaps somewhat


different from those measured in neutron scattering. 2 As the temperature
is raised to the vicinity of TZ it is often said that the Landau model "breaks
down" since it cannot describe the power-law variation of the superfluid
density Ps - (T~.'T)'67 that is experimentally observed3 in the critical
regime. In the critical regime, perturbation theories of phase transitions 4
have been developed which do predict (after some very difficult
mathematics) the correct variation of Ps and other parameters. It is a
puzzle, however, that these pertubation theories make no reference at all
to the roton excitations that control the superfluid density at only slightly
lower temperatures. Instead there is only reference to an apparently new
type of thermal excitation called a "critical fluctuation". The physical
properties of these fluctuations are left unspecified, except that they
somehow involve longer and longer length scales as the critical
temperature is approached.
With the advent of vortex-ring theories of the superfluid phase
transition 5-1 it is now possible to postulate a much more consistent
physical picture underlying the decrease of the superfluid density to zero
at T~: In this picture vortex rings are the only thermal excitations that act
to reduce the superfluid density. At low temperatures only the smallest
rings of diameter = 2/~ are excited, and these are identified as the roton
excitations of the Landau theory.7"9 The energy needed to excite these
rings is the vortex core energy of about 6 K, and this corresponds well
with measurements of the roton gap near TZ.11 As the temperature is
raised it becomes possible to excite rings with larger diameter, due to the
screening effects of the smaller rings. Finally at T~. rings of macroscopic
size are generated, and these can be identified as the "fluctuations" of the
perturbation theories.
The above scenario was first anticipated by both Feynman and
Onsager in their original papers proposing the quantization of circulation.12
To quote from Onsager:
"...we can have vortex tings in the liquid, and the thermal excitation of
Helium II, apart from the phonons, is presumably due to vortex rings
of molecular size...a more specific model of Landau's "rotons" ... As a
possible interpretation of the X-point, we can understand that when the
concentration of vortices reaches the point where they form a connected
tangle throughout the liquid, then the liquid becomes normal."

*Phonon excitations are also present, but their reduction of the bare superfluid density p,O is
negligible even near T;~, and hence it is assumed in the remainder of this paper that p o / p =_ 1.0).
Vortex Dynamics at the Superfluid ~.-Transition 1081

The first analytical implementation of this concept was carded out in the
two-dimensional theory of Kosterlitz and Thouless 13, whose ideas formed
the basis for the calculation in three dimensions. 5
In the present paper the vortex-ring theory is extended 14 to allow
a study of the dynamic properties of the superfluid phase transition. This
is of interest because most experimental probes of the superfluid involve
finite-frequency measurements. The methods employed follow those of the
two-dimensional dynamics of Ambegaokar, et al. 15 In section 2 the static
properties of the vortex-ring theory are briefly reviewed, and in section 3
the dynamic theory is developed. In section 4 the superfluid relaxation
time is calculated and comparison is made with experimental results.
Section 5 discusses possible applications of these ideas to other problems
of interest.

2. STATIC T H E O R Y

The renormalization equations for the superfluid density are most


conveniently written by defining coupling constants K~ = (h2/mZ)(psa o
/kaT ) and K = K~(a/ao), where m is the helium mass, a is the average
diameter of a vortex ring, and a o is the vortex core diameter. Using a
dimensionless length Q = ln(a/ao), the recursion relations determining the
renormalized value of K~ are found from real-space renormalization
techniques to be 5q

1 (1)
.-~ .-~ = - - ~ +AoY

3Y = [ 6 - r c Z K ( l n ( a / a ) + 1)]y (2)
O~

K r --- lim Ke -j (3)


|--~
1082 G. Williams

where y is the vortex fugacity, a c the scale-dependent effective core size,


and A o a constant. These equations are iterated to large ~ starting from
bare values K o and Yo = exp(-n2Ko C) at Q = 0, where the constant C
characterizes the vortex core energy, and K o uses the bare ps. The Flory
scaling assumption of Shenoy 6 is employed for the effective core size, a J a
= K , where 0 = 3/(d+2) = 0.6 is the Flory exponent of the self-avoiding
walk in 3D. There is now considerable justification of this value of the
exponent for closed vortex loops, as discussed later in this section. 1A6'17
It is known that the core energy constant C = 4/3 in the 3D XY model,
and the requirement that the critical coupling constant be the same as
deduced from Monte Carlo simulations, 18'2 Ko = 0.238, determines s the
value Ao = 24.4. This is reasonably close to the predicted value 1 of
4x3/3, which would require an upward shift of only about 10% in Koc;
such a change would be within the present uncertainty in the bare spin-
wave helicity modulus at To.21
The values of the non-universal parameters a o and C for the case
of liquid helium are determined from experimental inputs. C is adjusted
until the superfluid amplitude matches the experimental value 3, and a o is
then found from the known T x = 2.172 K. The results are 8 C = 1.03 and
Ko~ = 0.277, and a o = 2.3,~. The energy of the smallest ring of diameter
a o is the energy gap at T x, A/kB = n 2 Koe C T~. = 6.1K. In general, both
a o and C could be slowly varying functions of temperature and pressure,
and the effect of this would be to limit the critical regime where power-
law behavior is observable, to smaller values of Tx-T. These effects could
be incorporated in the theory by fitting to experimental results at finite
Tx-T, but this has not yet been implemented in the model.
The above recursion relations have a very simple interpretation,
similar to that of Kosterlitz and Thouless 13 in 2D. Integrating Eqs. 1 and
2 with respect to ~, and substituting Eq. 3 gives an equation for the
renormalized superfluid density,

I m2kBT
= I + Ao ~ J a 4 e -v(a> a 2 d a . (4)
P, p,~ ~ 2 a 8o ,*.

The rings are dipoles with a polarizability proportional to a 4, and the


integral in Eq. 4 is the susceptibility, a sum of the polarizability times the
number density of the thermally excited rings 5. The superfluid density is
reduced as the dipoles are oriented by an applied flow field, such that their
net backflow cancels part of the field. The energy appearing in the
Vortex Dynamics at the Superfluid ~,-Transition 1083

Boltzmann factor of Eq. 4 is the renormalized ring energy (divided by


kBT),

U(a)=rd~fKr(ln(~l+a)da+rc2KoC (5)

The energy to create a large ring is screened by the orienting effect of its
flow field on the smaller rings around it. Fig. 1 shows U as a function of
a for several different temperatures. Near T~, U is strongly reduced from
its bare form Uo=rCZKo(a/ao)ln(a/ac), becoming more nearly logarithmic in
a. It is this screening effect which eliminates the "confinement" of the
rings that would occur if their energy remained linear. The integral of
Eq.4 can diverge at large a and drive the superfluid density to zero only
if U(a) increases at a rate that is less than linear.
70 I I I

60-
t= 10 .3
50'
t = l 02
40
E3 / ~ - - - " t = l 05 -
30

10
i
0 I I I
0 500 1000 1500 2000

a/a o
Fig.1 The screened ring energy U(a) for several different values
of the reduced temperature.

By iterating Eqs. 1-3 (or equivalently Eqs. 4 and 5) to values of


a greater than the coherence length ao/Kr, it is found near T~ that Ps varies
as tv, where t is the reduced temperature t = (T~. - T)/T~. = 1 - KoJKo, and
fits for 10-4 < t < 10-6 give an exponent v = 0.67168 5: 0.00003. This
agrees very well with Shenoy's prediction 6 v = 0.6717. The iterations are
carded out using a fifth-order Runge-Kutta technique using a step size AQ
= l x l 0 -4. To get the value of the exponent to the above accuracy it was
necessary to first determine Ko to an accuracy of 10-12.
1084 G. Williams

The calculation of v above depends weakly on the value of the self-


avoiding walk exponent 0. The Flory expression 0 = 3/(d+2) is known to
be exact for d=2 and d--4, and is at the least a very good approximation
in d=3. e-expansions 22 and computer simulations z3 of open-ended walks
find values closer to 0.59 for d=3 instead of the exact 0.60. Less is known
about the case of closed walks applicable to the present calculation, but
there is at least some evidence that the Flory value might be exact for this
problem. A recent calculation ~7 valid for closed loops of oriented
polymers found the Flory expression to be the exact result of an
expansion, valid for all d < 4. There is also a recent study of closed loops
by Chattopadhyay et al., 1 who used Flory-type arguments to show that the
total perimeter length of a closed loop is related to the best-fit planar
circular loop of diameter a by

P __ (6)
a"~

Their simulations, using Chorin's algorithm 16 for generating the closed


loops, verified this form, and they found that the exponent was rather
precisely given by the Flory value for 0, to within an uncertainty of
+ 0.003.
It should be pointed out that the vortex-ring theory is not the fh'st
to demonstrate a fundamental connection between the properties of the
self-avoiding walk and the critical exponents of the n-vector models. This
was well demonstrated by Domb z4 and others 15 in high-temperature
expansions. It is interesting to note that Domb's derivation of finite-size
scaling, in which his random-walking topological objects were limited to
a maximum size equal to the lattice size, is the same physical picture that
was used for the vortex-ring model in a finite box. 21

3. D Y N A M ~ T H E O R Y

The dynamics of the superfluid transition at finite frequencies can


be investigated by finding the response of the fluid to an oscillating
velocity field. The fundamental equations of motion which must be solved
in order to study i.e. first and second sound propagation are the two-fluid
Landau equations, suitably generalized to include thermal fluctuations. 26
These are a complete set of eight equations in eight variables, which can
Vortex Dynamics at the Superfluid X-Transition 1085

be chosen to be (for example) ~ , e, Ps, and the specific heat at constant


pressure cp. In general ps and Cp are functions of temperature, pressure,
and frequency, and their functional form is an input into the Landau
equations. Since in the present theory these quantities are determined
close to T~. by the renormalization of the vortex rings, the critical dynamics
will depend on the dynamics of the vortices. This scheme for the
superfluid dynamics is exactly the same as that employed in the two-
dimensional calculation of Ambegaokar et al.,15 except that in the present
case it is extended into three dimensions.
It should be noted that an alternate method of calculating the
dynamics of the transition has been developed from the field-theoretical
r e n o r m a l i z a t i o n group. 27'28 Because this theory is based on the dynamics
of the order parameter (which is not a physical observable) instead of the
dynamics of the superfluid density, the Landau equations of motion cannot
easily be employed. Instead a series of model Hamiltonians is invoked
which use an increasing number of "coupling" terms to approximate the
equations of motion: model E and model F, z7 and "beyond model F". 2s
The dynamics of model F are known to be incomplete, as it does not have
a first sound mode. This has been remedied in calculations which go
"beyond model F", zs but it has still not been verified that these equations
are equivalent to the Landau equations. The problem in making the
comparison arises because the superfluid density is quite difficult to
calculate from the order parameter, requiring a calculation of two
derivatives of the free energy.
To calculate the frequency-dependent superfluid density it is
necessary to find the response of a vortex ring of average diameter a to a
low-amplitude oscillating flow field 7 = ~7o exp(-io)t). This is determined
by the Fokker-Planck equation for the vortex distribution function, which
is given by 29

kJ +r (7)

where p = rc2('~/m)pa2/2 is the momentum of the ring in a direction


perpendicular to the plane of the ring, and F is the energy (divided by
kBT) of the ring in the flow ~s, F = U(a) -ff'~s/k~T. With no flow the
equilibrium value of F is F o = (ao) -6 exp(-U(a)), the usual thermal
distribution function. A similar Fokker-Planck equation for the distribution
1086 G. Williams

function is present in the renormalization-group dynamics (see Eq. A1 of


Ref. 30), but the formulation in terms of the order parameter precludes a
direct comparison, as discussed above. The quantity A in Eq. 7 is the total
drag force on the ring, proportional to its perimeter P. With P given by
Eq. 6, the drag force is then A = ~t p, where T is the mutual friction
coefficient characterizing vortex-ring energy loss defined by Barenghi,
Vinen, and Donnelly. 31
For small v s there is little deviation from equilibrium,
F --F +8Fe-i,o,, and Eq. 7 can be linearized. Defining a response
function by

g(o),a)= 23 FoPVo
k~T fSFcos0
o
sin0d0 (8)

where 0 is the angle between/Y and ~7, after much algebra Eq. 7 becomes

2P2~g ap~2
(9)

g [.~ icoa2orcl..~o~r-~
1 (5) + =o.

The quantity

= rt2K __a (In (K -) + 1) (10)


r ao

is related to the ring energy U, and near T~. it becomes a constant for tings
smaller than the coherence length,_/~ ~- 5. ,The diffusion coefficient D has
been defined by D = "ykBT/IrCao(Ph/m)
2I" and as shown by Donnelly 32
this is the diffusion coefficient of the smallest ring of diameter ao. In
solving Eq. 9 a reasonable assumption is to neglect the spatial derivatives
of g. The equation is quite similar to that found in the two-dimensional
case 15, where it was shown that neglecting the derivatives only leads to a
redefinition of the diffusion coefficient by a constant factor of about 7.
Making this same approximation with Eq. 9 gives
Vortex Dynamics at the Superfluid k-Transition 1087

g(o~,a)--- (/~ - 5/2)


(11)
-i3a2D"xl'~o
~8

where D' is assumed to be shifted from D by some constant. For small


rings and low frequencies g = 1 and the the rings can follow the applied
flow in phase, but for large rings and high frequencies the diffusing rings
cannot keep up, with the real part of g falling to zero and the imaginary
part becoming large.

4. S U P E R F L U I D R E L A X A T I O N

The superfluid density at finite frequency is found from the


dynamic permeability

~(co) -- o -- 1 + ~ a~O) g(o~,a)aa (12)


p, (o)) -' oa
a

where p(0) is the static value from Eq. 4. Inserting the response function
of Eq. 11 into this expression, the superfluid density in the limit of low
frequencies (ox << 1) takes the form p s ( o ) / p = (p,(o)/p) (1 - io,'x). This
is the same form as predicted by the phenomenological Landau-
Khalatnikov theory 33, but in this case there is an explicit equation for the
relaxation time x given by
0
n a2 p , ( o ) " f Op(o) (a/ao)"r'* (13)
-- 8 D' p
J. da

A numerical evaluation TM of this expression using the static recursion


relations of Eqs. 1-3 finds that the relaxation time diverges near T~. as a
power law of the reduced temperature, x = xo t-x, and a fit of the calculated
values to this form for 10-4 < t < 10-6 yields x = 1.0075 + 0.0001. This
is in agreement with the Landau-Khalatnikov theory 33 and with dynamic
scaling 27, which predicts x = zv = 1.00752, where z = 3/2 is the dynamic
1088 G. Williams

exponent for the helium universality class in three dimensions and v =


0.67168 is the superfluid exponent found above. A value of D" --- 1.2 x
10-5 cm2/s is found from the experimental result 34 'co = 1.8 x 10~2 s. This
value for the diffusion coefficient is roughly of the order of the quantum
of circulation, as has been predicted on dimensional grounds. 15
In more recent work it has been found that the integral of Eq. 13
can be evaluated by expanding about the static fixed point, similar to the
methods employed by Shenoy.6 The integral is a moment of the integral
that generates (ps)-1, and hence will diverge with an exponent proportional
to v. The analytic evaluation indeed yields x=zv as found above, with
z=0/(1 - 0). For the Flory value 0=3/5 this gives z=3/2, as originally
inferred from the numerical results. 14 This calculation of the relaxation
time is valid only if the diffusion coefficient in Eq. 13 is not also
renormalized to zero at T x, which would imply a breakdown of dynamic
scaling. Any temperature dependence of D would considerably narrow the
critical temperature regime where the power-law behavior would be
observable, and if 1/D is weakly divergent at T x the value of z from Eq.
13 would be slightly increased. The possibility of a weak breakdown of
dynamic scaling has been noted in the renormalization group 35, although
the question of the stability of the dynamic scaling fixed point has not
been conclusively resolved. 36 If it is unstable, these theories would predict
an increase in z of less than 0.05. Experimentally, there is some evidence
that 1/D increases as Tz is approached, 11 but no divergence has been
observed. It should be mentioned that similar questions about the
renormalization of the diffusion coefficient have arisen in the two-
dimensional vortex theory37, and an extension of those calculations to the
3D case would be very interesting.
At high frequencies where co'c becomes appreciable the superfluid
density calculated from Eqs. 11 and 12 deviates from the Landau-
Khalatnikov form. Figure 2 shows the real part of ps(tO)/p for several
different frequencies. When 0yc > 1 the superfluid fraction increases above
the static value, and remains finite at and above T x. This is a finite-size
broadening of the transition, since the largest rings are overdamped and
cannot respond to the applied flow. The integral of Eq. 12 is cut off at
finite ring size, giving a broadening in complete correspondence with the
effects seen for finite box size21 and finite wavenumber. 8
The broadening of the transition is also accompanied by dissipation,
due to the rings that are maximally out of phase with the driving flow
field. Figure 3 shows the magnitude of the imaginary part of the
superfiuid fraction, which is proportional to the dissipation that would be
Vortex Dynamics at the Superfluid ~.-Transition 1089

observed in a torsion oscillator measurement15. There is a peak in the


dissipation at 0~ = 1 and then a decrease to a finite value at T~. This
behavior is very similar to that observed34 in first sound attenuation,
although the comparison is not direct because that attenuation is related to
the imaginary part of the frequency-dependent specific heat. 3s It would be
quite an interesting test of this vortex theory if the predictions of Figs. 2
and 3 for the frequency dependence of ps could also be calculated from the
dynamic renormalization group theory, which has apparently never been
attempted.

10 1

-
.
O
w

10.2

3.17 M H z ~ ~
'1:
10-3 1.0 M H z - - ~

0.1 M H z ~
Static
10 -4 I I I
10 .2 1 0 .3 1 0 -4 1 0s 1 06

Reduced temperature t

Fig.2 Real part of the superfluid density for several different


frequencies (o)/2~).

A detailed calculation of the propagation characteristrics of first and


second sound will require an evaluation of the frequency-dependent
specific heat, which plays a significant role in both modes. This will
involve calculating the free energy using the distribution function of Eq.
11 and then taking temperature derivatives. For second sound, which is
a low-frequency mode, a crud~ approximation can be made to neglect
entirely the frequency dependence of Cp. In this case inserting the low-
frequency form of ps(o~) following Eq. 12 into the Landau expression for
c 2 yields a damping coefficient determined by the imaginary part of Ps, D2
= (Czo)2X, where c20 involves only the real part of Ps. If dynamic scaling
1090 G. Williams

3x10 3 I I -

2
z
t-)

i:5 1

1 0 -2 1 0a 1 0 .4 1 0s 1 0 .8

Reduced temperature t

Fig.3 Magnitude of the imaginary part of the superfluid


fraction for several different frequencies (0~/2~).

holds then the asymptotic variation of D 2 is t -v/2, a result already well


known from the Landau-Khalatnikov theory33 and dynamic scaling. 27
The attenuation of fourth sound is controlled by the imaginary part
of Ps, since the sound speed is c 4 = (pjp)lt2 Cr This leads to an
attenuation coefficient given by tx4~ = re Re0a)/Im(la), where p is given by
Eq. 12. This attenuation is shown in Figure 4, and it becomes very large
at high frequencies. This high attenuation, combined with the high
attenuation from normal fluid unlocking,39 makes fourth sound unusable
as a probe of the effects predicted in Fig. 2. A better technique may be
to measure the impedance of a shear crystal oscillator loaded with a fairly
thick helium film. 4

5. APPLICATIONS

The dynamic theory discussed above has applications to a number


of other topics, including mutual friction near the X-point, high-Tc
superconductors, and systems that are rapidly quenched through their phase
transition.
Vortex Dynamics at the Superfluid ~.-Transition 1091

, ! , /
8000

,-
i
6000
E -- 1.0 M H z - ~ , , , J /
..) /--...,j
4000 - 0.6M H z ' . ~ ~ ~
2000

0
1 02 1 0a 1 0 .4 10 s 10 6

Reduced temperature t

Fig.4 Attenuation of fourth sound for several different


frequencies (oY2r0.

5.1 Mutual Friction

Mutual friction is the drag force exerted on a vortex line that is


moving with respect to the normal fluid. Much like the superfluid density,
the temperature dependence of the mutual friction has been a puzzle over
the entire temperature range up T~, being variously ascribed to roton
scattering at low temperature41 and to "critical fluctuations" that give rise
to the divergence observed4z near T~. The vortex model now gives a more
consistent picture of this process, since in all regimes the mutual friction
results from an exchange of energy between the vortex line and the
thermally excited vortex loops which constitute the normal fluid. As the
line moves it perturbs the distribution of loops, and due to their finite
relaxation time the interaction is dissipative. At low temperature only the
smallest loops of diameter a s are excited (the rotons) whose relaxation
time is very short, and the mutual friction is relatively small. A s the
temperature is raised larger loops are excited which cannot respond as
rapidly, leading to the divergence of the mutual friction coefficients at the
k-point.
1092 G. Williams

5.2 High-T c Superconductors

There is now experimental evidence that near T c the phase


transition of high-Tc superconductors is in the same universality class as
the helium ~-transition. 43 This was expected from the Ginzburg criterion44
in conjunction with the very short zero-temperature coherence lengths
found in these materials. Once again, this behavior is described as arising
from "thermal fluctuations". The vortex theory now provides a more
coherent picture45. In fact, the Ginzburg criterion is readily illustrated by
Eq. 4, from which the superfluid density is calculated. In the
superconducting case the quantity ps is the superfluid density from
depairing processes, while the integral of Eq. 4 is the contribution from
vortex rings. The integral will always diverge exactly at To, but the
question is over what range of temperature interval Tc-T will it dominate
the depairing contribution. For standard superconductors with large a o the
coefficient of the integral is very small, and the BCS depairing dominates
to within a few microKelvins of T e. For the high-T materials the much
smaller a o and higher T allow the vortices to dominate for several degrees
near T~. The anisotropy of the materials complicates matters, with rings
parallel to the CuO planes costing less energy to excite than rings
perpendicular to the layers.46 Shenoy has incorporated this into the vortex
theory using the anisotropic XY model47, and finds at lower temperatures
that the behavior is 2D, but that at Tc the anisotropy becomes irrelevant
and large circular loops are excited in a fully 3D transition.
The dynamics of order-parameter relaxation are just beginning to
be probed experimentally using pulsed lasers. Earlier measurements on
standard superconductors48 found a relaxation time that diverged near T~
as the inverse of the BCS energy gap, varying as (T~-T)lrz. Although the
initial measurements on the high-T materials are problematic49, there is
some indication that the divergence is faster than the BCS variation. The
above theory of vortex dynamics may be applicable to this system, and
would predict a variation of (T-T) "1. This ignores effects such as pinning
and anisotropy energies that could also be important in this case. Fisher,
et al. 44 also have pointed out that the dynamic universality class could be
different from the static class because of the plasmon excitations of the
charges.

5.3 Quenched Systems

There have been several studies of the dynamics of systems which


Vortex Dynamics at the Superfluid ~.-Transition 1093

are quenched rapidly through their critical point. Monte Carlo simulations
of the XY model have been carried out where the temperature is
instantaneously reduced to zero from a value near Tc, and the time decay
of the size and density of vortex loops is monitored.5 Zurek51 has
proposed that a cosmic-string phase transition, occurring in the rapid
cooling of the universe following the big bang, will be analogous to a fast
passage of liquid helium through TT. There have now been initial
experiments performed with helium in which a sudden pressure change is
used to rapidly quench through the 2~-point.52 The attenuation of second
sound is used to monitor the vortex density, and a relatively slow
relaxation to the new equilibrium is observed.
It should be possible to describe the behavior of these systems
using vortex methods similar to those outlined in section 3 above.
However, the main difference is that for the quenched systems the Fokker-
Planck equation probably cannot be linearized, as the variations in the
vortex distribution function are both large and rapid. For a complete
description it will be necessary to solve Eq. 6 without neglecting the
spatial derivatures of the distribution function.

ACKNOWLEDGEMENTS

Useful discussions with S. Putterman, J. Rudnick, V. Dohm, and S.


Shenoy are acknowledged. This work was supported in part by the US
National Science Foundation, grant DMR 91-20170.

REFERENCES

. J. Wilks, The Properties of Liquid and Solid Helium, (Clarendon


Press,Oxford,1967) Chap.5.
2. J. S.'Brooks and R. J. Donnelly, J. Phys. Chem. Ref. Data 6, 51
(1977).
3. A. Singsass and G. Ahlers, Phys. Rev. B30, 5103 (1984).
4. K. Wilson, Rev. Mod Phys. 49, 773 (1975); M. E. Fisher, ibid, 46,
97 (1974).
5. G. A Williams, Phys. Rev. Lett. 59, 1926 (1987); 61, 1142(E);
6. S. R. Shenoy, Phys. Rev. B40, 5056 (1989); ibid B42, 8595 (1990).
7. G. A. Williams, in Excitations in Two-Dimensional and Three-
Dimensional Quantum Fluids, eds. A.F.G. Wyatt and H. Lauter,
(Plenum Press, New York, 1991), p. 311.
1094 G. Williams

8, G.A. Williams, Phys. Rev. Lett. 68, 2054 (1992); 68, 2976 (E)
(1992).
9. G.A. Williams, J. Low Temp Phys. 89, 91 (1992).
10. B. Chattopadhyay, M.C. Mahato, and S.R. Shenoy, Phys. Rev.
B47, 5159 (1993).
11. O. Deitrich, E. Graf, C. Huang, and L. Passell, Phys. Rev. AS,
1377 (1972); K.Ohbayashi, M.Udagawa, H. Yamoshita, M.
Watabe, and N. Ogita, Physica B 165&166, 485 (1990).
12. L. Onsager, Nuovo Cimento Suppl. 6, 249 (1949); R.P.Feynman,
in Progress in Low Temperature Physics, ed. C.J. Gorter,
(North-Holland, Amsterdam, 1955), Vol. 1, p.52.
13. J.M. Kosterlitz and D. J. Thouless, J. Phys. C6, 1181 (1973).
14. G.A. Williams, Phys. Rev. Lett. 71, 392 (1993).
15. V. Ambegaokar, B.I. Halperin, D.R. Nelson, and E.D. Siggia, Phys.
Rev. B21, 1806 (1980); S. Teitel and V. Ambegaokar, Phys. Rev.
B19, 1667 (1979).
16. A. Chorin, J. Stat. Phys. 69, 67 (1992).
17. R. Kamien, J. Phys. I (Paris) 3, 1663 (1993).
18. G. Kohring, R. Shrock, and P. Wills, Phys. Rev. Lett. 57, 1358
(1986).
19. J. Epiney, Diploma thesis, ETH Zurich, 1990 (unpublished).
20. Y. Li and S. Teitel, Phys. Rev. B40, 9122 (1989); W. Janke,
Phys. Lett. A 148, 306 (1990).
21. G.A. Williams, Physica B 165&166, 769 (1990).
22. J.C. Le Guillou and J. Zinn-Justin, Phys. Rev. Lett. 39, 95
(1977).
23. N. Madras and A.D. Sokal, J. Stat. Phys. 50, 109 (1988).
24. C. Domb, J. Phys. C 3, 256 (1970); 5, 1399 (1972); 5, 1417
(t972).
25. M. Aizenman, Comm. Math. Phys. 86, 1 (1982); V. Dotsenko et
al., Phys. Rev. Lett. 71, 811 (1993).
26. S.J. Putterman, Superfluid Hydrodynamics, (North-Holland,
Amsterdam, 1974), Chap. IV.
27. P.C. Hohenberg and B. I. Halperin, Rev. Mod. Phys. 49, 435
(1977).
28. V. Dohm, J. Low Temp. Phys. 69, 51 (1987); J. Pankert and V.
Dohm, Phys. Rev. B 40, 10842 (1989).
29. R.J. DonneUy and P. H. Roberts, Phil. Trans. Roy. Soc. 271, 41
(1971)
30. B.I. Halperin, P.C. Hohenberg, and E.D. Siggia, Phys. Rev. B 13,
Vortex Dynamics at the Superfluid X-Transition 1095

1299 (1976).
31. C. F. Barenghi, R. J. Donnelly, and W. F. Vinen, J. Low Temp.
Phys. 52, 189 (1983).
32. R. J. Donnelly, Quantized Vortices in Helium H, (Cambridge
University Press, Cambridge, 1991) p. 262.
33. V. L. Pokrovskii and I. M. Khalatnikov, JETP Lett. 9, 149
(1969); I. M. Khalatnikov, Sov. Phys. JETP 30, 268 (1970).
34. R. Carey, C. Buchal, and F. PobeU, Phys. Rev. B 16, 3133
(1977); R. Williams and I. Rudnick, Phys. Rev. Lett. 25, 276
(1970).
35. C. De Dominicis and L. Peliti, Phys. Rev. B 18, 353 (1978).
36. V. Dohm, Phys. Rev. B 44, 2697 (1991).
37. R. Petschek and A. Zippelius, Phys. Rev. B 23, 3483 (1981).
38. R. A. Ferrell and J. K. Battacharjee, Phys. Rev. B 20, 3690 (1979).
39. W. Y. Tam and G. Ahlers, J. Low Temp Phys. 58, 497 (1985).
40. M. J. Lea, D. S. Spencer, and P. Fozooni, Jap. J. Appl. Phys. 26,
supp. 26-3, 53 (1987).
41. W. F. Vinen, in Progress in Low Temperature Physics, ed. C.J.
Gorter, (North-Holland, Amsterdam, 1955), Vol.3, p.1.
42. L. P. Pitaevskii, JETP Lett. 25, 154 (1977); P. Mathieu, A. Serra,
and Y. Simon, Phys. Rev. B 14, 3753 (1976).
43. M. B. Salamon, J. Shi, N. Overend, and M. A. Howson, Phys. Rev.
B 47, 5520 (1993).
44. D. S. Fisher, M. P. A. Fisher, and D. A. Huse, Phys. Rev. B
43, 130 (1991).
45. G. A. Williams, Physica B, Proceedings of LT20, to be published.
46. J. Friedel, J. Phys. (Paris) 49, 1561 (1988).
47. S. R. Shenoy, Helv. Acta Phys. 65, 467 (1992); B.
Chattopadhyay and S. R. Shenoy, to be published.
48. I. Schuller and K. E. Gray, Phys. Rev. Lett. 36, 429 (1976).
49. S. G." Han, Z. V. Vardeny, O. G. Symko, and G. Koren, Phys.
Rev. Lett. 67, 1053 (1991).
50. H. Toyoki, J. Phys. Soc. Japan 60, 1433 (1991); M. Mondello
and N. Goldenfeld, Phys. Rev. B 45, 657 (1992).
51. W. H. Zurek, Nature 317, 505 (1985).
52. P. C. Hendry, N. S. Lawson, R. A. M. Lee, P. V. E.
McClintock, and C. D. H. Williams, Physica B, Proceedings of
LT20, to be published.

Anda mungkin juga menyukai