HAL Id: jpa-00247810
https://hal.archives-ouvertes.fr/jpa-00247810
Submitted on 1 Jan 1993
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of
sci-entific research documents, whether they are
pub-lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diffusion de documents
scientifiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
Triple line dynamics in superfluid helium
Françoise Brochard-Wyart
To cite this version:
Classification
Physics
Abstractsfi7.70-68.10
Short Communication
lkiple
line
dynamics
in
superfluid
helium
Frangoise
Brochard-ivj'art
Institut Curie, 11 rue Pierre et Alarie Curie, 75231 Paris Cedex 05, France
(Received
14SeiJtember1992,
accepted
in final form 28October1992)
Abstract.
Superfluid
~IIe
fails to ivet anysubstrate,
as has been showntheoretically
and
experimentally.
~Vestudy
someliydrodynainic
featuresexpected
inpartial
wetting
(in
conditions where
exchanges
with the vapor arenegligible):
I)
thedynamics
ofdewetting:
thinfilms
deposited
on non wettable surfaces are metastable below a critical thickness and evolveby
nucleation and
growth
ofdry
holes. The boles arepredicted
here to open at constantvelocity.
ii)
The collective triodes of a contactline,
"triplons",
which arepropagative.
1 Tut io duct ion.
It has been sho,vn very
recently
thatsuperfluid
~IIe
fails to wet all substrates.Theoretically,
Cheng
et al.[II
predicted
that,
at T =0,
~ lie(leposited
on weak
binding
substrates such as thealkali metals could
e,;hibit,
depending
on the substratepotential,
either nonwetting
orwetting:
finite contact
angles
arepredicted
for ~ liedeposit.ed
onpotassium,
rubidium andcesium,
whilecomplete
wetting
ispredicted
forlithium,
sodiuiii andhydrogen.
In a more recent paper,they
explore
thetemperature
dependence
ofthe,vetting-non
wetting
phase
diagram.
Observationsby
Naclier andDupont-Roc
[2] of heat. flow in~lle
films in aglass
tube,
with and withoutan interior cesium
ring
show that~lle
does not ,vet cesium. Howeverwetting
is observed forRb and Ii
rings. Using
vapor pressure isothermmeasurements,
&Iukherjee,
Druist and Chan[3] indicate a
complete
absence of~IIe
adsorption
on cesium coatedgraphite
for T < 2 K.Ketola,
Wang
and lIallock [4](using
thirdsound)
have observed anomalouswetting
ofpartly
Cs-coated
glass.
Increasing
the vapor pressure,they
monitora transition between
dry
cesiumand cesium covered
by
a thinfilm,
,i=hereas Naclier observed nowetting
for~He
in coexistencewith bulk
liquid.
Thisdiscrepency
has been attributed toroughness and/or
surfaceimpurities
(e.g.:
oxygenpromoting wetting),
,vhicli are different in the twoexperiments.
In a more recentwork,
Rutledge
and Taborek [5] ,vere able to prepare very smooth cesiumevaporated
films ontothe surface of a quartz niicrolJalance an(I slio,v that
~lle
filmundergoes
awetting
transition22 JOURNAL DE
PHYSIQUE
II N°1Our aim in the
present
note is tostudy
thedynamics
ofwetting by
superfluid
helium,
as-suming:
a)
a finite contactangle
SE,
(or
anegative spreading
coefficient S = 7so-(7sL+7),
where 7;j are the bare
solid,
solid/liquid
andliquid/vapor
interracialtensions,
andb)
negli-gible exchanges
between theliqui(I
and the vapor(thus
restricting
our attention to very lowtemperatures).
If a helium film is
deposited
on a non-wettablesubstrate,
the film is unstable below a criticalthickness [6]
ec =
2K~~
sin~~
(1)
2 where Jc~~ is the
capillary
length.
~-l
~ 7(~~
~
with 7= 0.35
niN.ni~~,
p = 0.14g.cni~~,
oneexpects
K~~
+~ 0.5 mm. "Thick" films are metastable:
they
dewetI)y
nucleation andgrowth
ofdry
patches
[6, 7].
Microscopic
filmsare unstable and dewet
by
a meclianisiu ofspinodal
decomposition,
wherecapillary
waves areamplified [8-10].
Thedewetting
of viscouslicjuids
has been studiedrecently
bothexperimentally
and
theoretically:
dry
holes in the film open and aliquid
rim is formed at theperiphery
of thedry
region.
Thevelocit.y
of rim Vd~ #(0E)~
is constant intime, inversely
proportional
to theviscosity
q of theliquid
andindependent
of the initial thicknesse~.
In section 2 westudy
thedewetting
velocity
and theshape
of themoving
rim for inertialliquids.
In section
3,
we consider a contact line I)etween asolid,
liquid
helium and vapor which isdeformed
sinusoidally.
Many
capillary
probleins
involve the deformations of a contact line: forinstance the
separation
of the line fi.oni ananchoring
site can beanalysed
once undulationmodes
are understood. These modes have been studied for viscous
liquids by Ondarquhu
andVeyssid
[I
ii.
lvith silicone oilsthey
foundsingle
exponential
relaxations,
suggestive
ofpurely
viscousregimes.
we have calculated these collective mode relaxations for viscousliquids
[12].
Our aim in section 3 is to discuss t-he case of inertial
liquids.
Most of our discussion in section 3 isonly
qualitative,
andignores
all nuiuericalprefactors.
2.
Dewetting:
nucleation andgrowth
ofdry
regions,
When a
liquid
fiIni isdeposited
on a non-wettablesurface,
dry regions
nucleate and grow. Wehave
pictured
infigure
i theopening
of a(lry patch
of a radiusR(t)
surroundedby
aliquid
rim,
which troves at.velocity
U =j
The native fiIni is immobile. Theliquid
of thedry
i
region
is collecte(I in the rim of voliiueflit)
increasing
with time. We cansimply
derive themotion
equation
of theinoving
tint front the transfer of surfaceenergies
into kinetics energy of the rim.)pQU~=-SA
(3)
where A
= ~
R~
is the surface of thedry
surface of radius R andQ,
(Q
= Ae),
the volume ofthe
liquid
which has been collected in the rim. Fromequation
(3),
one finds thevelocity
U :U= ~~
=/T
~~.
(4)
pe
U
my RIM NATIVE FILM
o ~ B
E
(iii)
(
eFig.
1.Opening
of adry patch
of radiusR(t)
in a helium film of thickness e. Theliquid
is collected ina rim of circular
shape.
@E is theequilibrium
contactangle.
U is thedewetting
velocity.
The
velocity
is constant with time! Thedriving
forceacting
on the rim isconstant,
but themass of the rim increases and the
resulting
motion is uniform.Equation
(3)
assumes thate < ec. In the
general
case, one must add agravity
energycontribution,
which leads to:U =
fi.
(4')
~ e ~Typically,
with S ~W 7, oneexpects
U ~W cms~~
for ~~ ~W lo. eThe
velocity
ofdewetting
is similar to thevelocity
ofopening
of a hole in a soap film[13,
14].
The case of soap films iscomplicated by
the presence of the surfactant at the freesurface,
which
gives
rise to arelatively
complex
elastic behavior. In our case, the interfacial tension 7is
constant,
alldequation
(4)
is exact.The
shape
of rim is deduced front the motionequation
of aliquid
sheet. In the lubricationapproximation,
thevelocity
V(z)
of aliquid
sheet ofheight
((~)
isindependent
of z(plug
flow)
and the Stokes
equation
for a non-viscousliquid
is:P(
($
+V~~)
=(
7("'
(5)
The derivative of the moinentuni is balanced
by
theLaplace
pressuregradient.
Theequation
has to be modified at the two contact lines A and B with the solid substrate and the native
film,
respectively.
Because the acceleration term isnegligible
for z -0,
the staticequilibrium
conditions hold at both ends of the tint
The rim moves at
velocity
Ugiven
by equation
(4),
except
in a narrowregion
where thevelocity
slows down to zero.
Equat.ion
(5)
then showsthat,
in most of therim,
("'
=0,
I-e- the pressureis constallt arid the rim
profile
is aportion
of aperfect circle,
except
near the contact line Bin a thin
region
of extensionf.
Thehealing length
f
is deduced fromequation
(5)
by
ascaling
24 JOURNAL DE
PHYSIQUE
II N°1Integrating
equation
(5)
between A and B leads toequation
(4).
We havepictured
infigure
ithe
shape
of the rim and the smallregion
where 0 falls to zero.3. Collective uiodes of a contact line.
We consider the contact line a non-wettable
solid,
superfluid
~He
and its vapor(of negligible
pressure).
The line is deforiuedsinusoidally
with awavelength
~~(Fig.
2).
We discuss theq
resulting
oscillationassuming
a small contactangle
0E
< 1. In thislimit,
thecapillary
energyof the
weakly
disturbedline,
withamplitude
uq for a wave vector q, has been calculated firstin reference
[15]
and is of the form:fei
q =)1011ql
lttql~
(8)
y
~
2~
q
6<6~
+ + + + + XFig.
2. "Triplon" modes of a contact line.Equation
(8)
holds whengravitational
energies
arenegligible
(q
>K).
In thisregime,
the lineelasticity
cannot be described in teriils of a line tension. A tension(
wouldgenerate
an energy~W
(
q~jug
~ The distortions of the
liquid-,<apor
interfacenear the contact line extend
(in
thedirection normal to the
line)
up to dist.ances of orderq~~
This is the source of the anomalousfactor (q( in
equation
(8).
In theopposit.
liiuit(q
<K),
the deformations are screened at adistance
K~~,
and the eitergy has a standard foriu:fAq
"(7
01
K (q(~("q(~
~~~The
general
expression
interpolating
I)etu~een these two limits has been derived morerecently
[16]:
feiq
")7
°I
(fi
)
lug
~(IQ)
We derive the
dynamical
features of the mode fi.oiu the balance between the elasticrestoring
force deduced froiu
equation
lo)
and the inertial force:7
0(
(fi
«)
[uq[
= Mw~uq
(ii)
3. I CAPILLARY REGIhfE
(q
>K).
In tillslimit,
theperturbation
extends over aregion
ofsize
q~~
normal to the direction of theline,
and thescaling
form of M is:hi =
p~(.
(12)
q"
From
equation
(ii),
we find thefi.equency
w(q)
of thepropagative
modeomitting
an exactnumerical coefficient:
w(q)
=q~/~
/~
~0E.
(13)
Joanny
has calculated thefi.equency
of the oscillation ofsuperfluid
droplets
around theirequi-librium
shape
[Iii
and finds w=
'
(
in
agreeiuent
withequation
(13).
~
3.2 GRAVITY REGIhIE q <
«~~
Forlong
waveiengtiis,
the deformation extends on thecapillary
length
K~~
and the mass fi,I is now:hi
+~
p0EK~~
(14)
Equation
(10)
leads to standardpropagative
modes of avibrating
rope w = cq, where thevelocity
of the wave, c, is:~
/~
~
' ~~~~ withK~~
+~ 0.5 runt,0E
~10~~,
one finds c +~ni.s~~
4Concluding
reiuarks.We have studied some
dynainical
features ofsuperfluid
~lle
deposited
on non-wettablesub-strates at very low temperatures, where the role of vapor is
negligible.
I)
Thin films are unstable and dewetby
nucleation andgrowth
ofdry
patches,
which open at _-?constant
velocity
U=
U*0E,
where U* = g~
Contrary
to the case of viscousliquids
where~
the
dewetting
velocity
i~,asindependent
of the initial film thickness e, one expects U +~e~~/~
for inertial
liquids.
By
videorecording
theopening
of micro or macrodry
patches,
one canmonitor the filiu thickness
varying
over a broad range(100 I-100
pm).
The metastable filmsobserved
recently by
liutledge
an(I Taborek [5] do not dewetspontaneously.
However,
if ahole in the film is fornied
(I)y
a laserlJeain),
the film will dewetprobably by
the mechanismsdescribed here.
ii)
The"triplons"
iuodes of the contact line arepropagative.
As forcapillary
waves ofusual
liquids
in the inertialregime,
,ve find w+~
q~/~
for
large
wave vectors q. On the other
hand,
for undulations of siuall wave vectors, the deforiuatioiis are screened and w = cq, where26 JOURNAL DE
PHYSIQUE
II N°1Acknowledgements.
We thank P-G- de Gemies and J.F.
Joanny
forstimulating
discussions on thewetting
of~He,
C. Redon ondewetting
and the Referee for his useful comments.References
Ill
CHENGE.,
COLE li.lV., SAAAt W-F- and TREINERJ.,
Phys.
Rev. Lett. 67(1991)
1007.[2] NACHER P-J- and DUPONT-ROC J.,
Pbys.
Rev. Lett. 67(1991)
2966.[3j IiUKHERJEE S-II-, DRUIST D-P- and CIiAN M-H-W-, I. Low
Temp.
Phys.
87(1992)
l13.[4j iIETOLA
II-S-,
WANG S. and iIALLOCI( R-B-,Pliys.
Rev. Lett. 68(1992)
201.[5j RUTLEDGE J-E- and TABOREK P.
Pliys.
Rev. Lett. 69(1992)
937.[6] REDON
C.,
BROCHARD-WYART F. and RONDELEzF.,
Phys.
Rev. Lett. 66(1991)
715.[7j BROCHART-'VYART F., Dl hIEGLIO J-hi- and
QUER#
D., C. R. Acad. Sci.(Paris)
II 304(1987)
553.
[8j BROCHARD-WYART F. and DAILLANT
J.,
Can. I.Phys.
68(1990)
1084.[9j BROCHARD-&VYART F., REDON C. and SYI(Es
C.,
C. R. Acad. Sci.(Paris)
314(1992)
131.[10j
REITERG., Pltys.
Rev. Lett. 68(1992)
75.[III
ONDARCUHU T. and VEYsslk hi., Nature 352(1991).
[12j
BROCHARD-WYAfIT F. and DE GENNESP-G-,
Latigmuir
7(1991)
3216.[13] CuLicK