• Aucun résultat trouvé

Mean field theory of n-layer unbinding

N/A
N/A
Protected

Academic year: 2021

Partager "Mean field theory of n-layer unbinding"

Copied!
10
0
0

Texte intégral

(1)

HAL Id: jpa-00247840

https://hal.archives-ouvertes.fr/jpa-00247840

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.

Mean field theory of n-layer unbinding

W. Helfrich

To cite this version:

W. Helfrich. Mean field theory of n-layer unbinding. Journal de Physique II, EDP Sciences, 1993, 3

(3), pp.385-393. �10.1051/jp2:1993100�. �jpa-00247840�

(2)

Classification

Physics

Abstracts

64.60 68.10

Mean field theory of n-layer unbinding

W. Helfrich

Institut fur Theoretische

Physik,

Freie Universit&t Berlin, Amimallee14, 1000 Berlin 33,

Germany

(Received 3J August J992,

accepted

in

final form

J6November J992)

Abstract. -A two term

Ginzburg-Landau

Hamiltonian for the free energy

density

of

n-layer

systems is in~roduced. It is used to derive

a distribution function for such systems and to obtain

scaling

laws

involving

the number of components. We discuss the

validity

of this mean-field

theory

for stretched

polymers

in two-dimensional space and for fluid membranes.

1. Introduction.

There are several theoretical treatments of the

unbinding

transition of a

pair

of fluid membranes.

Fornlally,

this

problem

is

equivalent

to that of a

single

membrane

unbinding

from

a flat Wall. The earliest

theory,

a renornlalization group

calculation,

was

presented by Lipowsky

and Leibler

ill. Utilizing

the scale invariance of thernlal membrane

undulations, they

renornlalized the

potential

which a membrane feels in front of a wall

by integrating

out

step by

step the

short-wavelength

undulations.

Assuming

a short range

repulsion

and the Van der Waals attraction of thin

plates, they

calculated the critical value of the Halnaker constant, I.e. the

strength

of Van der Waals interaction, at which

unbinding

takes

place.

Later on,

Lipowsky [2] pointed

out that the renornlalization group flow

equations

for a fluid membrane in three dimensions are identical to those for a stretched

polymer

in two dimensions

exposed

to the same one-dimensional interaction

potential.

There is a difference in the

derivations,

the

scaling being anisotropic

in the case of the 2d

polymer (or

linear

interface).

The

propagation

of the stretched

polymer

in a one-dimensional

potential

is known to

obey

a

Schr6dinger

type

equation [3, 4]. Accordingly,

the method of

finding

the

ground

state wave function of the

Schrodinger equation

and

squaring

it to obtain the

polymer

distribution function seems transferable from the

polymer

to the membrane. In a third

approach Lipowsky

and Zielinska

[5]

did a Monte Carlo

study

of membrane

pair unbinding, expressing

the interaction of the flat membranes

by

a square well

potential.

All these methods

predicted

the same critical behavior for the mean membrane

spacing

s as a function of the

strength

u

(~ 0, arbitrary units)

of an attractive

potential responsible

for

(3)

386 JOURNAL DE PHYSIQUE II N° 3

adhesion. The common result is

s -v

(u~ u)~~ (l)

with #f = I, where u~ is the critical

strength.

The critical exponent #f does not

depend

on the details of the

potential. However, equation (I )

was shown

[1, 4]

to be valid

only

if the

potential

energy of interaction between the flat membranes

drops

off faster at

large spacings

than the

squared

inverse

spacing.

This condition is

generally

satisfied

by electrically

neutral fluid membranes.

Subsequently,

we

proposed

a « reflection » model as a

quasi microscopic description

of

pair unbinding [6].

Central to it is the

assumption

that the membranes fornl

sharp

bends where

they

hit each other

(or

the

single

membrane hits the

wall).

These bends are

thought

to be confined to

the narrow

region

of

strongly negative potential

energy of membrane interaction. At the critical

point,

the

bending

energy of reflection is on average

compensated by

the

potential

energy of the well. Adhesion sets in as the well

deepens further, obeying again

the

scaling

law

(I)

with

#f =

I. The reflection model suggests that the

unbinding

transition of membranes is modified

by logarithmic

corrections of the scale invariance of membrane

shape

fluctuations. In

particular,

the

unbinding

transition appears in it to be of

weakly

first order

[6].

Very recently,

Cook-Roder and

Lipowsky[7] published

a Monte Carlo

study

of the

unbinding

of three

parallel undulating

membranes.

Finding

#f = 0.8 for three

equal

membra- nes,

they

invoked the necklace model for a bunch of three linear interfaces

[8]

to

explain

#f ~ l. In this

model,

the internlediate interface acts like a

repulsive potential

between the two outer ones.

Varying

with the

squared

inverse

spacing

of the latter, the

potential

is

marginal

in the

language

of renornlalization group

theory.

The exponent #f as obtained

by

the

Schr6dinger equation

method

depends

on the

strength

of the

potential

in this

special

case.

However,

the necklace model is not

quite appropriate

of the situation considered since attraction acts in it

only

where all three and not

just

two of the interfaces come

together.

In the

following,

we propose a

Ginzburg-Landau type theory

for the

unbinding

transition of

n-layer

systems,

designed

for both two-dimensional

polymers

and fluid membranes. The

density

defined as the inverse mean membrane

(or polymer) spacing

represents the order parameter. In a

multilayer

system

composed

of a finite number n of

layers,

a

gradient

ternl

enters even

though

fluctuations of the order parameter are

disregarded.

In the

expansion

of the

free energy

density

we omit all powers of the order

parameter beyond

the

second,

thus

limiting

ourselves to situations where the mean

layer spacing

is much

larger

than the width of the attractive internlembrane

potential.

From the

Ginzburg-Landau

Halmiltonian we calculate a

layer density

distribution function. Based on

it,

we find

equation (I)

with #f = I to

apply

also to

n-layer

systems and derive additional

scaling

laws that involve n. The

validity

of the mean

field

theory

is then

briefly

discussed in tennis of the

Schr6dinger equation

method.

Finally, arguments

are

presented

Which

suggest

that the mean field

theory

cannot be

expected

to hold

exactly

even for n

= 2 in the case of membranes.

After

finishing

the present

theory

we became aware of MiIner and Roux's mean-field

theory

of membrane

unbinding

in infinite

multilayer

systems which has

just

been

published [9].

Although quite

different from ours, it also

predicts

#f

= I for the critical

exponent

of

unbinding.

We will come back to it at the

beginning

of the discussion.

2. Mean field

theory

and

density

distribution function.

In order to describe the

unbinding

of n

polymers

or membranes we introduce the

following

free energy of

multilayer

fornlation per unit area or

volume, respectively,

f=AP~+Dl Ill

l~ (2)

(4)

The

mutually impenetrable

components are taken to be

infinitely

extended and

basically

nornlal to the z axis. Their mean number

density

is

expressed by

a continuous function p = p

(z) despite

their discreteness. It is

advantageous

to use the modified

density

P

(Z)

= I Is

(z )

,

(3)

where

s(z)

is the mean

spacing

of the

components

as a function of z, so that the membrane

thickness,

an irrelevant

quantity,

does not enter the

equations.

The first ternl of

(2)

contains the control

parameter

A of the

unbinding

transition which is

proportional

to u u~

occurring

in

equation (I)

for

pair unbinding.

It is

negative

for bound states and zero at the critical

point.

Vanishing

A means that there is neither attraction nor

repulsion

of the components in a

homogeneous multilayer

system. The factor p~ comes from the linear ternl in a mean field like

expansion

in powers of p of the total interaction energy between

adjacent polymers

or

membranes per unit

length

or area,

respectively.

There is no zero order contribution to this energy if the

potential

energy of interaction goes to zero for

large spacings.

Contributions of

higher

than linear order in p are

neglected

because we are interested in low

densities,

I-e- in

mean

spacings

much

larger

than the

region

where the interaction

potential

is

significant.

The

validity

of the one-ternl Landau

expansion

will be examined below in tennis of

microscopic

models for

polymers

and membranes. We will also estimate A in the case of

polymers

for a

particular potential.

The

p~

ternl of

(2) alone,

without the intervention of a

gradient

ternl,

would result in an

abrupt unbinding transition,

I.e. a

jump

from s=0 to s= oo as

A

changes

from

negative

to

positive.

The

gradient

ternl in

(2)

is unusual in that

(dp/dz)~

is divided

by

p. This can be understood for membranes in the

following

way. The

bending

energy of a

single

membrane

and, correspondingly,

the free energy of

purely undulatory

membrane interaction are known to be scale invariant. The main effect of a

gradient dp/dz

is a distortion of the membrane

configurations

as

compared

to a uniform distribution. It

seems therefore reasonable to expect the

gradient

term to be also scale

invariant,

which

requires

the form

(I/p (dp/dz)2.

As a tentative value of the coefficient D for membranes we may take

from

a for the

energy nsity of purely undulatory

in

a

nifornl ultilayer

system,

f =

D/s~

[10].

Here k

is

Boltzmann's

onstant, T and K the

rigidity

of

the

single embrane.

scaling

arguments

can be

used

in

the case

of

polymers

to

justify

the (I/p

) (dp/d2)~

term.

owever, scaling

is

isotropic in the olymer

case,

requiring the engths parallel

to

the polymers to be multiplied

gnifying

of s.

~

with an unknown numerical

factor,

where « is the line tension of the

polymers.

A

gradient

ternl of the fornl

(I/p ) (Vp

)~ in the free energy bulk

density

of unstretched

polymers

has been introduced some time ago

by

de Gennes

[11].

The distribution function p

(z)

is obtained

by minimizing

the

integral

F

=

I[Ap~+D (~~ )~+Mpj

dz.

(6)

P dz

Without the last ternl in the

integrand,

F would be the free energy of the whole

n-layer system,

JOURNAL DE PHYSIQUE11 T 3, N'3, MARCH iW3 is

(5)

388 JOURNAL DE PHYSIQUE II N° 3

per unit

length

for

polymers

or per unit area for membranes. The last ternl with a

Lagrange multiplier

M has to be added to take account of the constraint

p

(z) dp

= n

(7)

where n is the fixed number of

polymers

or membranes

composing

the system. The Euler

equation resulting

from 8F

= 0 is

easily

found to be

k~D~~~2p ~dz ~2D~'

~~~

With the reduced

quantities

fit i12

~ 2 D ~ ~~~

and,

for A ~

0,

~ =

~~

p

(lo)

this transfornls into

~~'l=-~2+ (~'l )~+1~. (ll)

df~ 21~ df

The solution of the second-order differential

equation

is

if the two

boundary

conditions are chosen such as to obtain a

symmetric

function

~ = ~

(c )

that tends to zero for

c

- ± oo. The universal distribution function

(12)

is

plotted

in

figure

I.

We are now in a

position

to derive

scaling

laws for

weakly

bound

n-layer systems.

The

number of

components

should

obey

n~p(0)Az (13)

where Az is a measure for the width of the distribution. Because of

(9)

and

(lo)

the factors on the

fight satisfy

p(0)~ ~ (14)

and

D l12

AZ~

,

(15)

M

so that

n

(MD

)~'~

(16)

(6)

Ti

i

o 2 3 ,

Fig. I. Plot of the function ~(c) as

given by

equation (12).

The last

equation

fumishes the

dependence

of the

Lagrange multiplier

M on n.

Eliminating

M between

(15)

and

(16) yields

A2

~

~

(17)

Accordingly,

the total width of the

multilayer

system is

inversely proportional

to both the number of

components

and the control

para1neterA.

A

corollary

of

(17)

is

p(0)~

~2-D.

(18)

The

scaling

law

(17)

leads back to

(I)

with #f = I ifs is

replaced by

Az/n. This

implies

that the

critical

exponent

of

n-layer unbinding equals

the

exponent

obtained earlier

by

various other methods for the

special

case n =

2. Of course, near the critical

point

of

unbinding

at

A

= 0 the control

parameter

will vary

linearly

with any

changes

not

only

of the interaction

potential,

but also of the line tension of the

polymers

or the

bending rigidity

of the membranes, and of temperature.

3. Discussion.

Let us first compare our mean field

theory

to that of MiIner and Roux

[9].

These authors started like us from a two-ternl

expression

for the free energy of

binding

per unit

volume,

but

;hey

considered a unifornl

multilayer

system. One ternl is

positive

and represents

purely undulatory interaction,

thus

varying

as

p~.

The other stands for the

potential

energy and varies as

pi applying

to

a uniform distribution of each membrane between its nearest

neighbors

and interaction

potentials

much more narrow than the mean

spacing.

This ternl, whose

prefactor

is the control parameter of the

phase transition,

is similar to the first term in our

equation (2).

However,

in our

theory

the

p~

tern1combines

undulatory

and

potential

interaction.

(Purely

undulatory

interaction would be associated with a nonuniforn1distribution and the

potential

(7)

390 JOURNAL DE PHYSIQUE II N° 3

energy would

always

be

negative

for a square well

potential

in front of a

wall.)

As a

result, undulatory

interaction alone does not occur but

only

the

gradient

ternl related to it which varies in effect with p instead of p~. This

change

of powers, while not

changing

the critical

exponent

#f, results in the

collapse

of infinite

multilayer

systems which our model

predicts

for all bound states

regardless

of the

strength

of

binding.

Next,

we check the

validity

of

equation (2)

which underlies our mean field

theory

for the critical behavior of

n-layer

systems. The fornlula

Ap~

for the free energy

density

of a

homogeneous multilayer system

can be made

plausible,

in the case of

polymers, by considering

the

simple

case of a

single polymer

between

parallel (linear)

walls. Jn order to

keep things simple

we assume as

usually (

q~

~)

«

l,

where q~ is the

angle

a monomer makes with the

preferred

direction. This can

always

be achieved

by choosing

a sufficient line tension

«. A

single polymer

between walls does not

undergo

the

abrupt collapse

which occurs in a

homogeneous multilayer

system when the control

parameter

becomes

negative.

Its

Schrbdinger type equation

takes the form

[3, 4]

-~~~~~~~~~'+U(z)9'=EP (19)

2 "

d2~

where

U(z)

is the

potential

energy of interaction with the walls per unit

length

and

E the

corresponding ground

state

eigenenergy. (For (19)

to be

applicable

the interaction energy

per monomer has to less than

kn.

The interaction is

conveniently represented by

square well

potentials

of

depth

U~0 and width

f~

in front of the walls. There is a critical

depth

U~ for which

P(z)

is constant outside the wells. In a

vicinity

of the critical

point

one has

to

E

"

z (U Uc (20)

where

I

» 2

f~

is the

spacing

of the walls.

(The

result is

easily

obtained

by

a

perturbation

calculation or

by

a

matching procedure

for a P that is still

nearly

unifornl outside the

wells).

Since E is a free energy, we may infer from

(20)

that the free energy of interaction per unit

length

of one

polymer

in a

multilayer

system of

(homogeneous) density

p should indeed be

Ap

=

fo(u uc)

p

(21)

in accordance with

(2).

Trying

to go a step

further,

we write down a

Schr6dinger

type

equation

for the entire

n-layer

system.

Considering only

the

(n-I)-dimensional subspace

of

configurations

with

zi + + z~ =

0,

I.e.

disregarding displacements

of the center of

gravity,

the

equation

may be written as

~~~~~

A~_1

4l +

U~yi,

,

y~_1)

4l

= E4l.

(22)

Here yi,...,y~_i are some Cartesian coordinates

(of

the same scale as

zi,...,z~),

A~ i is the

Laplacian,

and

W~yi,

..., y~ i

)

is the wave

function,

all of them in the

subspace.

U is the

potential

energy of the mutual interaction of the

polymers

and the

eigenenergy

E is the

ground

state energy, both

refemrg

to a unit

length

of the

system comprising

all

n elements.

Equation (22)

describes the

propagation

of n

mutually impenetrable polymers

with E

being

the free energy of their interaction. ln the

Schr6dinger equation analogy

the

polymers

are

represented by

n

equal

but

distinguishable particles

in one-dimensional space which

obey

zi w z~ « w z~. The

potential

energy of

interaction, U,

may

again

be

expressed by

narrow

square well

potentials acting

between nearest

neighbors.

The wells are in front of the contact

(8)

(hyper-) planes

z;

= z;~i with I

=

I,..,

n I which behave as

rigid

walls. The effective width of the

potential

well is

only Io/2~~,

so that U~ is now

larger

than in the case of a

single polymer interacting

with an actual wall. For two

polymers

and the obvious choice y =

2~ ~'~(x~ xi

)

one has

W

~

e~"~~ (23)

with

~

2

E~

«

«~ =

~

(24) (kT)

and

~r2 (U U~ f

~~

16 U~ ~

~ ~~~~

These

simple relationships

lead to the well-known result Az

~

(U~

U

)~ (26)

Unfortunately,

the situation becomes very

complex

for n

~ 2. The n I

degrees

of freedom

are

necessarily coupled

since

subsequent

contact

planes

are not

orthogonal

but make

angles

of

60°. At the same

time,

a

separation

of 4l into a radial and an

angular

function is

impossible

for

two reasons.

First,

the

potential wells,

their width

being independent

of

position, overlap

when

three or more elements are near each other.

Second,

at

large spacings

the

potential

wells,

acting

like

&functions,

deternline

(a 4lla~ )/W

on the contact

planes,

where ~ is the coordinate

(same

scale as zi,

.,

z~) along

the nornlal to the

respective

contact

plane.

The constancy of

(awla~ )/W

at

large spacings

is

incompatible

with

asymptotic separability.

If

separation

were

possible,

the radial function R

(r)

or, more

precisely,

the modified radial function

fi X

(r)

=

R(ryr

~

(27)

could be treated in the same manner as the wave function of a

single polymer feeling

a

I/r~ potential. (Both

the

separation

and the substitution of the modified function can

give

rise to an effective

potential

of this

type.)

The

I/r~ potential

in combination with a square well

potential

has been

shown,

for the

single polymer

in front of a

wall,

to result in a critical

exponent

#f that varies with its

strength [4]. Although

this

theory

does not

apply,

first

numerical results for three

polymers by

Netz and

Lipowsky

indicate critical behavior with

#f = 0.9 for n

= 3

[12].

As

yet

there is no

analytical proof

that mean field

theory

is incorrect.

How similar are the

unbinding

transitions of two-dimensional

polymers

under a line tension and fluid membranes with a

bending rigidity

? The

identity

of the renornlalization group flow

equations

for the two

systems,

which holds for any number of components, presupposes that the thernlal undulations of the membranes are scale invariant. This

assumption

is not

entirely

correct since the thermal undulations

give rise,

for

instance,

to an

absorption

of membrane area and a decrease of the effective

bending rigidity,

the correction terms

depending logarithmically

on the base area of the membrane

[13].

Both effects do not exist in the

polymer analog.

However, in the case of

fairly

stiff membranes

(K

= lo

kT)

such as

lipid bilayers they

amount to no more than a few

percent

of the area and of the

bending rigidity, respectively,

in the

absence of undulations. At their critical

points

of

unbinding

both

polymers

and membranes

(9)

392 JOURNAL DE PHYSIQUE II N° 3

appear to be reflected from each other at no cost of free energy, while the mutual contacts

give

rise to a

negative

free energy in the case of adhesion. The concept of reflection is in accordance with the

Ap~

term of the

Ginzburg-Landau expansion (2).

It is at this

point

that a

possibly

crucial difference emerges between

polymers

and membranes. The mean square

angle

which a

polymer

makes with its

preferred

direction does not

depend

on its

length.

In contrast, the mean square

angle

(q~

~)

which a membrane makes with its basal

plane

increases

logarithmically

with the size of the

piece.

With

periodic boundary

conditions one calculates for a square of size L~

[13]

(

q~

~)

=

~~

ln

~

8 "K

(28)

a

where a is a molecular

length.

In the

special

case of

purely undulatory interaction,

(q~~)

can also be estimated for

multilayer

systems. In those systems L in

(28)

has to be

replaced by

an effective

length

K i12

L

~ s

(29)

which is the lateral correlation

length

of membrane

spacing.

The numerical factor in

(29)

is

near

unity

but difficult to calculate

exactly.

Substitution of

I

for L in

(30)

results in

(q~~)

= ~~ ln

~~~

(30)

8 ""

kTa~

It follows from

equation (28)

that (q~ ~) is much smaller than

unity

for

lipid

membranes unless

their size is astronomical.

Highly

flexible

amphiphilic

membranes

(K= kT)

may

crumple

when

single

but,

according

to

(30), (q~~)

ml is

easily preserved

in

multilayer

systems.

Nevertheless, (q~

~)

can increase

by

a factor of two and more as the size or the

spacing

is made

larger

and

larger.

Let us recall that near the critical

point

of

unbinding

we

expect

the membranes to be reflected or,

rather,

deflected from each other. The deflection is

thought

to

consist,

at most

places,

in

sharp

bends of the

colliding

membranes in the narrow

region

where the interaction

potential

is

negative.

A small deflection

angle

may seem more favorable as it lowers the

bending

energy

per unit

length along

the lines of deflection and at the same time increases the width of the

strip

of membrane immersed in the well. On the other

hand,

more undulations have to be

suppressed

as the deflection

angle

gets smaller and the

strip

is widened. There is

possibly

an

optimal

deflection

angle

so that the free energy of membrane deflection

depends

on how well bulk and surface

angles

can be matched. It seems obvious that a small

spread

of q~~ is a

prerequisite

for

good matching.

If these ideas are correct, the

unbinding

transition of two or more membranes need not be

govemed by (I )

with #f = I. It could even end up as a first-order transition at some finite

spacing,

thus

avoiding

the

high energies

of

matching

a wide range of bulk

angles

to the

optimal

deflection

angle.

4. Conclusion.

Starting

from a

Ginzburg-Landau Hamiltonian,

we derived a distribution function for

weakly

bound

n-layer

systems, either stretched

polymers

in two dimensions or fluid membranes with a

bending rigidity.

The distribution function is of universal

shape

and was used to obtain

scaling

laws of the

unbinding

transition. The critical

exponent

#f of

unbinding

came out to be

unity,

regardless

of n, and the width of the distribution function was found to be

inversely

proportional

to the number of components.

(10)

It seems to be an open

question

whether or not this mean field

theory

is correct for

n ~ 2 in the case of

polymers.

For membranes its

validity

may besides be

impaired

even for

n = 2

by

a

logarithmic

correction.

However,

the mean field

theory

is

hoped

to be useful as a

first

guide

in

interpreting experimental

results

conceming

the

unbinding

of systems of more

than two components. For

instance,

it

predicts

the

unbinding

transition to occur at the same critical

potential

or temperature etc., for any number of components. This appears to agree with the

only experimental

observation of membrane

unbinding

known to date

[14].

If there

was a increase of the transition temperature with the number of

components

which varied from 2 to more than

20,

it was no more than a few °C

[15].

Acknowledgement.

a1n

grateful

to M. M. Kozlov for

solving equation (I I)

and to R.

Lipowsky

and M. M. Kozlov for discussions. This work was first

presented

at the

Aspen

Center for

Physics (Workshop

on

Self-Assembling Systems

and

Membranes,

Summer

1992).

Jn this

stimulating

environment I

was introduced to reference

[9] by

S. Milner and made aware of reference

[11] by

A. Liu.

References

[II LIPOWSKY R, and LEIBLER S,,

Phys.

Rev. Lett. 56

(1986)

2541.

[2] LIPOWSKY R.,

Europhys.

Lett. 7 (1988) 255.

[3] KROLL D, M, and LIPOWSKY R.,

Phys.

Rev, B 28 (1983) 5273.

[4] LiPowsKY R, and NIEUWENHUIzEN T, M,, J.

Phys,

A : Math, Gen, 21

(1988)

L89, [5] LIPOWSKY R, and ZIELINSKA B,,

Phys.

Rev, Lett. 62

(1989)

1572.

[6] HELFRICH W,, Phase Transitions in Soft Condensed Matter, T. Riste and D,

Sherrington

Eds.

(Plenum

Publishing Corporation,

1989).

[7] CooK-RODER J, and LIPOWSKY R,, Europhys. Lett. 18 (1992) 433, [8] FISHER M. E. and GELFAND M, P,, J. Stat. Phys. 53 (1988) 175.

[9] MILNER S. T. and Roux D., J. Phys. I France 2 (1992) 1741.

[10] HELFRICU W., Z. Natu~fiorsch. 33a (1978) 305.

[I II DE GENNES P. G., J. Chem.

Phys.

72 (1980) 4756 ;

see also HAI TANG and FREED K. F., J. Chem.

Phys.

94

(1991)

1572.

[12] NETz R. R. and LIPOWSKY R,, poster

presented

at STATPHYS 18, Berlin, August 2-8, 1992, and preprint,

[13] See, e, g,, HELFRICU W.,

Liquids

at Interfaces, Les Houches XLVIII (1988) J. Charvolin et al, Eds.

(Elsevier Science Publisher, 1990).

[14] See also GROTEUANS S. and LIPOWSKY R., Phys, Rev. A 41 (1990) 4574.

[15] MuTz M. and HELFRICH W., Phys. Rev, Lett. 24

(1989)

2883.

Références

Documents relatifs

The analysis is based on the important inequality ( JaWa 2.6) with ( INEG IN 2.7) with Gronwall Lemma and some appropriate estimates. We provide simple comments for the

form a basis for the r’4 irreducible representation of the cubic group, one must look for harmonic polynominals belonging to T4 : : those of the lowest degree

Thus we investigated whatever the same variation of total protein expression, between different donors and zones of HAM was also observed in the amount of growth factors (GF) that

This equation enables us to calculate the limiting free energy of the Sherrington-Kirkpatrick spin glass model at this particular value of low temperature without making use of

In order to check this prediction, a comparison with exact solutions of the cylindrical bending of the lattice is performed in Section 4. It reveals that only the

In order to give a clear demonstration of the additional warping effects included in the Bending-Gradient theory we compare Kirchhoff-Love, Reissner-Mindlin and Bending-Gradient

We describe athermal packings and their observed geometric phase transitions using equilibrium statistical mechanics and develop a fully microscopic, mean-field theory of the

We study the flow generated by this relaxed energy for radially non-increasing functions (functions with balls as superlevel sets).. In the first part of the paper, we prove a