• Aucun résultat trouvé

Short-time dynamics and statics of free and confined tethered membranes

N/A
N/A
Protected

Academic year: 2021

Partager "Short-time dynamics and statics of free and confined tethered membranes"

Copied!
18
0
0

Texte intégral

(1)

HAL Id: jpa-00248074

https://hal.archives-ouvertes.fr/jpa-00248074

Submitted on 1 Jan 1994

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.

Short-time dynamics and statics of free and confined tethered membranes

J.H. van Vliet

To cite this version:

J.H. van Vliet. Short-time dynamics and statics of free and confined tethered membranes. Journal de

Physique II, EDP Sciences, 1994, 4 (10), pp.1737-1753. �10.1051/jp2:1994229�. �jpa-00248074�

(2)

Classification

Physics

Abstracts

68.10 82.70 87.20

Short-time dynamics and statics of free and confined tethered membranes

J-H- van Vliet

(*)

Institut fur

Festk6rperforschung, Forschungszentrum Jiilich,

Postfach 1913, 52425 Jiilich~

Germany

(Received

1

February1994,

revised 17 June 1994,

accepted

5

July1994)

Abstract. Monte Carlo simulation and theoretical results for the short-time

dynamics

and the statics of

hexagonal self-avoiding

tethered membranes in free space

or confined in a slit are

compared.

In particular the initial

decay

rate of the

dynamic

structure factor and the static structure factor are determined. It 15 shown that for small

scattering

vector

lengths

the tethered membrane has the

shape

and short-time diIfu5ion constants of a flat disc. Further it 15 shown that the

decay

rate for intermediate

scattering

vector

lengths

suggests that the membrane

primarily

relaxes

through

its

out-of-plane

modes,

implying

that the membrane behaves a5 a rather 5tiIf sheet. For a membrane confined to a slit both the structure factor and the

decay

rate with the

scattering

vector

parallel

to the membrane surface are in better agreement with

theory

than for a free membrane due to suppression of

edge

fluctuations. The influence of

hydrodynamic

interactions of the membrane with the slit walls on the

decay

rate15 discussed.

l~ Introduction~

Tethered membranes can be looked upon as a two-dimensional

generalization

of linear

poly-

mers~ I.e.~ network

polymers. Experimental

realizations of such network

polymers

are, e-g-,

the

spectrine

network in the cell membranes of

erythrocytes iii,

a

graphite

oxide

crystalline

membrane [2]~ and cross-linked

monolayers

[3].

The

properties

of tethered

membranes,

as for

polymers [4,

5]~ can be discussed on

essentially

three

length

scales. The small

scattering

vector q

length

scale

qL

< 2~r, the intermediate q scale

2~r/L

< q <

2~rla

and the

high

q scale qa > 2~r~ where L is the diameter of the tethered

membrane,

and a the radius of the beads connected

by

tethers in the membrane. The diameter L of a

particular

membrane conformation is defined here as the distance between the centres of

mass of the two membrane beads which are at the maximum distance for that conformation.

(*)

Present address: Cavendish

Laboratory, University

of

~cambridge, Madingley

Road,

Cambridge

CB3 OHE, United

Kingdom

(3)

The

largest

progress in

understanding

the

physics

of

(tethered)

membranes

by theoretical,

simulation and

experimental

methods 11,

2,

6]

sofar,

has been made for the statics at intermedi- ate q. Simulation studies

suggest

that

self-avoiding

tethered membranes with excluded volume

between beads and tether

length

lo <

2v5a

are flat

[6-8].

The earlier theoretical studies on the

dynamics

of tethered membranes concentrated on

crumpled

membranes

[9, 10].

More recent theoretical work concentrated on the

dynamical properties

of flat

(tethered)

membranes and the effects of

hydrodynamic

interaction

[11, 12].

To my

knowledge

sofar no simulation work has been

presented dealing explicitly

with the influence of

hydrodynamic

interactions

(HI)

on

the

dynamics

of tethered membranes.

A

question dominating

a

large part

of the papers which have been

presented

to

date,

is if self-

avoiding

tethered membranes are flat or

crumpled.

For a

crumpled

membrane the square radius of

gyration

should scale as

R(

~-

N~/~,

and for a flat

membrane, R(

~- N

[6, 9].

A related

scaling analysis

can also be

performed

for the

eigenvalues

of the inertia tensor

(see,

e.g., [13]

).

A reliable

scaling analysis

of simulation data is

particularly dependent

on

large

membrane sizes since the

edge

fluctuations of the membrane are

large

[8].

Although by

finite size

scaling

the situation

can be

improved [7],

the most

straightforward approach, I-e-, simulating

tethered membranes for a scale of

large diameters, spanning preferably

some decades in

length,

is not very realistic

considering

the

large

number of beads and

large

relaxation times

[14]

needed to

equilibrate

membrane conformations and to

get independent

membrane

conformations,

as

compared

to

scaling problems

studied for linear

polymer

chains. Another

approach

to determine if the membrane is flat or

crumpled

is the

analysis

of the static structure factor for its

scaling

with q for the intermediate q scale. However this

approach

has not

presented

very conclusive evidence to

date,

since the observed

scaling exponent

for a flat membrane is not very different from that of a

crumpled membrane,

and for a flat membrane the

apparant

value

depends crucially

on

the details of the membrane model

[15,

16] as well as on the presence of

edge

fluctuations

[6].

Although

now the

self-avoiding

membrane is

generally

believed to be flat it seems worthwhile to

provide

a criterion for flatness which is

simpler

and does not

depend

on the intermediate

length

scale

fluctuations,

as discussed later on in detail.

Frey

and Nelson [11] worked

out,

within the framework of

Langevin dynamics,

the behaviour of the

dynamic

structure factor

parallel

and

perpendicular

to the membrane surface. In contrast to the short-time behaviour dealt with here

they

discussed the

long-time

behaviour. As

argued by Frey

and Nelson and

Lipowsky

[12] the relaxation rate

Tjj~

due to fluctuations

perpendicular

to the membrane surface with

hydrodynamic

interactions scales for the intermediate q scale as

Tjj

~

~- Q~~~~

(1)

where

(

is the

roughness exponent.

For the relaxation rate due to fluctuations

parallel

to the membrane surface

Frey

and Nelson showed

T£~

~-

q~~~ (2)

where the

roughness

exponent

(

is connected to uJ

by

the

scaling

relation

[8, 17]

(

"

((2 +1°) (3)

In this paper Monte Carlo

(MC)

simulation and theoretical results are

presented

on the

dynamics

of of tethered

self-avoiding membranes,

in

particular

for the initial

decay

rate

r(q)

of the

dynamic

structure factor. For

S(q)

results are

presented

in

particular

for

qL

< 2~r, which offers an alternative way to look at the flatness of a membrane.

Further,

data are

provided

for the situation in which a membrane is confined between two walls. The first

objective

of

(4)

this confinement is to suppress the

edge

fluctuation of the finite-size membranes

simulated,

in order to make a

comparision

with the

theory, developed

for tethered membranes

neglecting edge fluctuations,

easier [8]. Confinement restricts the

out-of-plane

fluctuations

[18] whereby

the

relatively large edge

fluctuations are

expected

to be

suppressed effectively.

The second

objective

is to demonstrate the effect of confinement on the

hydrodynamics

of a tethered

membrane.

2~

Theory~

2~1 STATICS. The model of the tethered membrane used in the simulation

study presented

here is the well-known network of

triangulated

beads and free

perimeter,

first introduced

by

Kantor et al.

[14].

Here the customary

hexagonal shaped

membrane is used

[19].

The structure factor is defined as

S(q)

+

j ~ iexP(iq

rv

)) (4)

where N is the number of beads and r~ the vector

connecting

beads I and

j.

The

angular

brackets indicate an average over all conformations.

Averages

of

S(q)

can be taken with

respect

to different orientations of q: over all

orientations,

over orientations

parallel

to the membrane surface

(perpendicular

to the smallest

eigenvector

of the inertia

tensor)

or orientations per-

pendicular

to the membrane surface

(parallel

to the smallest

eigenvector

of the inertia

tensor).

These averages will be indicated

by

the connotations

"isotropical", "parallel"

and

"perpen-

dicular"

respectively.

The two latter connotations are also indicated

by

the indices "I" and

"))" respectively.

If a membrane confined in a slit is concerned

defining

averages

parallel

and

perpendicular

to the slit makes sense. This is indicated

by

indices "XY" and "Z"

respectively.

The

isotropically averaged S(q)

calculated for the simulated membrane conformations is calculated with the usual formula [5]

S(q)

=

p ~j l~ (5)

~~

~~~i~~

For the simulated membrane conformations the average

S(q)

with the

scattering

vector

parallel

to the membrane surface

S(qi)

is an average over at least 30 random orientations of qi

(20].

For small values of q with

respect

to the diameter L of the tethered membrane

qL

< 2~r the

shape

of the tethered membrane can be

judged by comparing

it to the

S(q)

obtained

by

a

model calculation. For a flat membrane with

qL

< the size of the fluctuations

perpendicular

to the membrane surface rjj < because of the

scaling

relation rjj ~-

L~ [12, 21].

Therefore if

the membrane is flat its

isotropically averaged S(q)

should be similar to that of an

infinitely

thin disc

is, 22]

~~~~ q~)2 ~

~)~~~~

~~~

where R is the radius of the disc. The

symbol J~

denotes a Bessel function of the order x.

For a flat disc I calculated the ratio of the diameter of the disc and the radius of

gyration

~2 fl

8 ~~~

This ratio can be used as an additional criterion to

judge

how flat the membrane

actually

is.

(5)

Abraham and Nelson [8] evaluated the structure factor

analytically by approximating

the membrane

by

a disc. The basic elements in this derivation are:

I)

fluctuations

parallel

and

perpendicular

to the membrane surface are assumed

Gaussian;

it)

the summation over all distances in

equation (4)

is

replaced by

the

integration

over the

prob- ability density

function

(16/~r) f(x),

for the normalized distances on a

disc, I-e-, ) £~~

is

replaced by (16/~r) f/ fix)

dx.

The function

fix)

is

fix)

=

x(arccos(x) xfi) (8)

1

(16/~r) f(x)dx

= 1

where x

=

1/(2R)

and the distance between two

points

within the disc. This distance function

f(x)

was

long

ago introduced

by Kratky

and Porod [22] to calculate the structure factor for

an

infinitely

flat disc

(see Eq. (6)).

Since a summation over discrete

points

is

replaced by

an

integration

it is a

priori

clear that this

approach

will

only

be succesful for values of q with

ql

<

1, I-e-,

for the intermediate or small q scale.

With the identification 2R = L and the use of

equation (8)

and the

equation

Fiqjj,

qi, L

=

Jo (qi

XL

expi- (qj (xL)~~ exp(- (q j (XL )~ (9)

Abraham and Nelson [8] derived from

equation (4)

S(qjj,qi, L)

"

(16/7r) / f(X)Flqjj,qi, L)dX 11°)

Actually

the version of

equation (10) presented by

Abraham and Nelson is

slightly

different in that the

argument

of

Jo

is

multiplied by

an additional fit

parameter.

As will become clear in section 4 a

satisfactory

fit with simulation data can be made without this additional

parameter.

2~2 SHORT-TIME DYNAMICS. The

dynamic

structure factor is defined here as

S(q, t)

%

j ~

(exP(iq in(t) rj(°)i) ill)

The initial

decay

rate of

S(q, t)

is defined as

riq)

+

it j

in

(Siq, t)) i12)

Akcascu and Giirol

[23]

showed that

r(q)

within the framework of the Smoluchowski

equation

can be calculated as an ensemble average in

configuration

space.

£~~ (q D~

q

exp(iq

r~

))

~(q)

"

~

~~

(13)

~ exp iq r~

where

D~

is the diffusion tensor. A more detailed account on the derivation of the

previous

equation,

the relation of

r(q)

to

scattering experiments,

and its

application

to various

polymeric

(6)

systems

can be found in the papers of Akcascu et al.

[23],

several text books and review articles

[4, 5, 24, 25].

Equation (13)

can be used to

probe

for

2~r/L

< q <

2~rla

the

decay

rate of the internal fluctuations of the

polymerized object.

In the limit

qL

< I the short-time diffusion

constant, D,

can be extracted from

equation (13)

since

[4,

5]

iirn

rjq)/q2

= D

j14)

The

resulting expression

is

~

2~/2 ~ ~~ ~"

~~ ~~~~

~J

An

assumption underlying

the use of the Smoluchowski

equation

is that the relaxation times of the fluid modes are much shorter than the internal relaxation times of the

polymerized object.

Therefore

equation (13)

for the initial

decay

rate is useful in the

description

of the

dynamics

for

scattering experiments designed

to

study only

the internal and translational motions of the

polymerized object.

For

polymerized objects

with constrainted coordinates the relaxation

times associated with those coordinates will decrease and

eventually

become

widely separated

from the relaxation times of the other slower internal modes. If the

scattering experiment

is

designed

to

probe

these slower modes

only,

then the initial

decay

rate must be calculated with

a modified diffusion operator in which

appropriate

constraints are

imposed

at the outset as

argued by Stockmayer

et al.

[26]. If, however,

one is still interested in the relaxation of the fast internal modes associated with the constrainted coordinates

equation (13)

for the initial

decay

rate must be used.

As

argued by

Abraham and Nelson [8] the excluded volume interactions between the beads of a tethered membrane induce an

entropical bending rigidity.

As

pointed

out in the

previous paragraph

this constraint

might

influence the initial

decay

rate measured in a

scattering

ex-

periment

not

probing

the faster relaxation times related to such a

constraint, necessitating

a different

expression

for the initial

decay

rate. The interest here is in the

situation,

as considered

by Frey

and Nelson

[11]

in the derivation of

equations (I)

and

(2),

were the solvent relaxation times are assumed much shorter than the internal relaxation times on the intermediate q scale

without

making

any further distinction for internal relaxation time scales, This

assumption

is

likely

to be most valid for a flat tethered membrane at the lowest

possible entropic

bend-

ing rigidity, I-e-,

for a tether

length

lo

=

2v5a just

small

enough

to

prevent

the membrane from

crumpling

[8]. In order to evaluate

equations (13)

and

(15)

an

explicit expression

for the

diffusion tensor is needed. The diffusion tensor used here is the Oseen tensor

[4,

5]

Dzj

#

~~~ 6zj1

+

~~~~ (1

6~j

1

+ ~~ ~~~

II16)

irqa irqr~ r~~

or its

improved

version the

Rotne-Prager-Yamakawa (RPY)

tensor

[25, 27]

JJ~

-

16vI

+

~iii~11 61J) III

+

~~i~"

+

ill III ~"l~" II

~~~~

where r~ > 2a.

It is also

possible

to include HI of the membrane beads with the two walls

by

modification of the

diagonal

elements of the diffusion tensor. An account of the calculation of the

diagonal

elements of the diffusion tensor in the presence of walls and the

assumptions underlying

this

(7)

calculation is

presented

in the

Appendix.

The simulation results on

r(q)

are obtained

by evaluating equation (13) using

the RPY tensor as the diffusion tensor. For the simulation

results HI with the walls is either included or not.

I

analytically

calculated the initial

decay

rate with the q

parallel

to the membrane surface from

equation (13) by approximating

the membrane

by

a disc as done

by

Abraham and Nelson in their calculation of

S(qjj,qi,L) according

to

equation (10).

The diffusion tensor used to

calculate

r(qi)

is the Oseen tensor.

~~~~~ ~~~~ ~~~~~~~

~

~~~~~~ ~~/L ~~~~~ (~0(qiXL)

+

@@)

~~~/~) II fl~)J0(~IXL)Gdz

~~~~~ ~~~

j18)

C

=

/ f(x)dx

r = ri +

rjj

ri " XL

~/L

o =

/I

+

A(xL)2(-2

G

= exp

~- ~q( xL)~) r(

=

A(xL)~~

2

The normalization constant C for the numerator is a consequence of

replacing

a summation

over indices I

# j by

an

integration. Replacing

a summation over all indices I and

j

as for the

denominator,

results in

16/~r

as a normalization constant.

In a similar way I evaluated

equation (15). Particularly simple analytic

solutions can be obtained

by

introduction of the

following

elements:

I) modelling

of a flat

discj

r

= ri "

XL;

it) large

disc

size; 2a/L

-

0j

iii)

use of the Oseen tensor

(Eq. (16)).

Further the

isotropically averaged

diffusion constant~

D,is

obtained

by averaging

the r-h-s of

equation (15)

over all orientations and the diffusion constants

parallel~ Di,

and

perpendicular~

Djj,

to the disc surface are obtained

by averaging

the r-h-s- of

equation (15)

over the orientations

parallel

and

perpendicular

to the surface

respectively.

The result is

16kBT

~

9~r2qL

2kBT (19)

~~ @

4kBT

~'~ 3~r2qL

The relation between the diffusion constants is

D

=

)Di

+

(Djj (20)

It is a

priori

clear that these

equations

will

predict

the diffusion of tethered membranes

correctly only

in the limit

qrjj

« 1~ which however is

automatically

satisfied since

qL

< 1 for

equation (15)

as noted before. The

expression

of

Frey

and Nelson

ill]

for D has a

slightly larger prefactor

D

=

2kBT/(3~rqL)~

which can be

explained by

the use of a circular

integration

(8)

domain instead of the correct

integration

domain

prescribed by equation (8)~

in

averaging

the r-h-s- of

equation (15).

Note that for

equation (18)

and

equation (19)

HI with the walls is not included. Further it should be noted that

equations analogous

to

equations

(18~

19)

can be obtained

using

the

RPY tensor~

equation (17),

in a

straightforward

manner. The

resulting equations

however are

more

complicated

and

give

rise to

only

small corrections or, as for the

isotropically averaged D,

to

exactly

the same result.

3~ Model and simulation method~

For the Monte Carlo

simulations~

the tethered membrane

considered,

is the well-known

hexagon consisting

of a

triangular

network of N

spherical

beads [14~

19].

The radius of a bead is

a = 0.5.

Neighbouring

beads in the network are linked

by

tethers of

length

lo < 1.6. Pairwise hard-core

repulsion

of all beads exists. The choice of tether

length

and the

repulsion

assures

self-avoidance of the membrane. No

explicit bending rigidity

is introduced in the simulations.

Along

the

diagonal

of the

hexagon Lo

+1 beads are

placed

connected

by Lo

tethers. The total number of beads is N

=

(3(Lo +1)~ +1)/4.

The size of the membranes simulated is in the range

Lo

= 10-50. To my

knowledge

the

largest hexagonal triangularly

tethered membrane

simulated sofar has a size of

Lo

+1 = 75 in a MD simulation

[6, 8].

Membranes are simulated both in free space and confined to

a slit bounded

by

two

repulsive

walls. The walls of the slit

are

perpendicular

to the Z-axis and restrict the

Z-component

of the centre of each bead to be in the interval

[0, 2d],

with a

corresponding

distance between walls H = 2d + 2a.

A MC move is defined

by

an

attempted randomly

selected

displacement

for each bead. This

displacement

is

produced by selecting

a

step-size

so " 0 0.2 for each of the three

orthogonal

directions in space; the average size of a MC move

resulting

is s

=

0.svsso.

A MC step is defined as a

cycle

in which all N beads

experienced

one MC move.

For each MC move

overlap

with other beads and

eventually walls,

as well as the

tethering

constraint has to be checked before

accepting

or

rejecting

the move. The beads can be divided in four

classes;

the characteristic of each class

being

that a bead is tethered

only

to a nearest

neighbour

bead of another class. In this way it is

possible

to check the interactions of a bead

with the nearest

neigbour

beads and the next nearest

neigbour

beads

(connected

to the bead

which is moved via a

neighbour bead, I-e-,

two tether

lengths away)

after all beads of a class have been

subjected

to a MC move.

All the other interactions are checked

only

if the bead

positions

before the MC

step

are such that a

danger

exists that beads

might overlap

in this MC

step.

This is

implemented by keeping

on record a list of those

beads,

other than

neighbour

and nearest

neighbour beads,

which are at a distance < 2a +

2viso.

The distances between these beads are checked in every MC step. This list is

updated

when the sum of the

magnitudes

of the two

largest displacements

since the last

update

for any two beads

Ii-e-, including

those not on the

list)

exeeds

2vsso.

A related

algorithm

is the automatic

updating

of a Verlet

neighbour

list

[28].

To

update

the

list,

distances between beads are checked

by using

a variant of the cell index

algorithm [28].

The membrane is

projected

on the

XY,

X Z or YZ

plane.

The actual

plane

used is the

plane

for which the membrane surface area

projected

on it is the

largest.

The difference

with the usual scheme

being

that one of the three coordinates

X,

Y or Z is not used to

assign

a bead to a cell. This method is efficient due to the flatness of the membrane. A related

algorithm, whereby

the cell orientation is determined

by

the orientation of the

eigenvectors

of the membrane's inertia tensor is described

by

Grest

[13].

This scheme of classified

beads,

a list of beads and cells can be

easily programmed

in vectorizable code.

The relaxation time TR is defined as the time needed for the tethered membrane to diffuse

(9)

its own radius of

gyration.

In a Monte Carlo simulation run a membrane was

equilibrated

for

a number of MC steps at least five times

larger

than the relaxation time TR. The membrane conformation was

sampled

at intervals of MC

steps

at least twice TR. For a flat tethered membrane in free space TR *

N~/(as)~ [29].

Here the value TR

"

N2/(0.svsso)~

is used.

For a confined membrane the relaxation time is

expected

to be smaller

[18].

For the

largest

membrane

(Lo

"

50)

2

independent

runs where

performed.

For the membranes of smaller size 3

independent

runs were

performed. Quantities

of interest were calculated for each conformation

sampled

in a run

(at

least 10

samples

per

run)

and

averaged

afterwards over all

samples

in that run. Final averages and statistical errors therein were obtained

by averaging

the averages

per run over all runs.

The simulations were

performed

on a CRAY

Y/MP

M94 vector computer or a Silicon

Graph-

ics

Indigo

workstation. The CPU time in seconds has been

computed

for runs on the CRAY computer and varies from 0.020 s per MC step for

Lo

" lo to 0.055 s per MC

step

for

Lo

" 50.

4~ Results and discussion~

4~l FLATNESS OF THE MEMBRANE. The idea that a tethered

self-avoiding

membrane is

flat is confirmed

by comparison

of the

isotropically averaged

structure factor

S(q)

for a rather

large

tethered membrane in free space,

Lo

"

50,

with

S(q)

for a disk with a diameter

equal

to that of the membrane for small q in

figure

I. It is clear that for the q range of

interest, qL

< 2~r,

S(q)

for the membrane is in

good agreement

with

S(q)

of a disc.

Further confirmation of the membrane's flatness can be found in the

comparison

of the ratio

R(/L~

for a disk with that of a tethered membrane as obtained from the simulations and

presented

in table I. It is clear that for a tethered membrane in free space the ratio

R(/L~

is less than the theoretical ratio of 0.125

provided by equation (7). However, by confining

the

membrane to a slit with H

= 3.5 the simulation result for

R(/L~

is within error

equal

to the

0.8

0.6 S(q) o

off

o

0.2

0

0.02 0.08 0.12 0.16 0.2

q

Fig.

I.

Dependence

of the

isotropically averaged

structure factor

S(q)

as a function of q for small values of q. Data points:

S(q)

for

a tethered membrane with Lo = 50. Solid line:

S(q)

for

a disc with

radius 2R

= 46

according

to equation

(6).

The diameter of the disc is set equal to the diameter L of the tethered membrane obtained from the simulation.

(10)

Table I. Radius of

gyration

and diameter.

environment

Lo L~ R2 j~2/~2

free space 10 79.5 9.81 0.12

free space 20 295 29.4 0.10

free space 30 740 82.2 0.11

free space 50 2125 235 0.11

slit;

H

= S-S 20 371 41.9 0.113

slit;

H

= 3.5 20 396 49.2 0.124

Table II. Short-time diffusion constants; 11

=

D/kBT.

~~~i~~~~~~~ j~

jjsimulation jjtheory jjsimulation jjtheory j~simulation jjtheory

~ l I

it

free space 10 1.71E-2 2.02E-2 1.78&2 2.27&2 1.56E-2 1.52&2

free space 20 1.00E-2 1.05E-2 1.04&2 1.18&2 9.41E-3 7.85&3

free space 30 6.34E-3 6.62&3 6.69&3 7.45&3 5.61E-3 4.97&3

free space 50 3.85E-3 3.91E-3 4.12E-3 4.40E-3 3.30E-3 2.93&3

slit;

H = S-S 20 8.99E-3 9.33E-3 9.62&3 1.05&2 7.62E-3 7.00&3

slit;

H

= 3.5 20 8.52E-3 9.05E-3 9.24E-3 1.02&2 6.90 E-3 6.79&3

theoretical value. It should be noted that the statistical error for free membranes in the data of table I. is

approximately 10i~ higher

than for the confined situation. This

suggests

that

by

confinement the

large edge

fluctuations can be

effectively suppressed.

This

suppression

of

edge

fluctuations also increases the diameter of the membrane as

compared

to a free membrane of the same

size, I-e-, equal Lo.

A

comparison

of the short-time diffusion constants for a disk and the simulation results for

a tethered membrane in table II

give

further

support

to the idea of a flat membrane. The diffusion constants for a disk are calculated

by

substitution of the L values from the simulation

(11)

data

presented

in table I in

equation jig).

There is a

good agreement

between the diffusion constants from the simulation and the results for a disk.

Noteworthy

is that the trend observed for the

disks, Di

> D >

Djj,

is

clearly

confirmed

by

the simulation data and in accordance with the

expectance

that a flat

object

will

experience

less friction

moving parallel

to its surface than

perpendicular

to it. The smaller diffusion constants for the confined membranes are due to the

larger diameter,

see table

I,

of these membranes due to

suppression

of

edge

fluctuations.

4~2 RELAXATION ON THE INTERMEDIATE q SCALE. The relaxation modes of the membrane

are

investigated

for the intermediate q scale

2~r/L

< <

2~rla.

The

edge

fluctuations

clearly

dominate the

isotropically averaged

initial

decay

rate as shown in

figure 2, resulting

in a

scaling

relation

typical

for a

crumpled object,

r

~- Q~. A similar

explanation

has been offered

by

Abraham and Nelson for the

"crumpled"

appearance of the

isotropically averaged

structure factor

[8].

The

large

influence of the

edge

fluctuations shows also up in the

large

data scatter for

r(qjj perpendicular

to the membrane

(results

not shown

here).

The initial

decay

rate

parallel

to the membrane surface is as the

corresponding

structure factor [8] less disturbed

by edge

fluctuations. Therefore

r(qi)

can be used to

get

an

insight

in the relaxation modes of the membrane. In

figure

3 a

scaling plot

for

r(Qi)/q~~~~

versus

QIL

with

(

= 0.65 for two membrane sizes shows

clearly

that

r(qi)

'~ Q~~~~ for the

larger

values of the intermediate q scale

(first peak)

in accordance with

equation Ii ).

It

suggests

that relaxation is dominated

by

the

out-of-plane

relaxation

(bending modes) [11], I.e.,

the membrane behaves lik4 a rather stiff sheet. This statement derives further

support

from the fact that

in-plane

relaxation

(phonon modes)

of

r(Qi) according

to

equation (2)

is

surely

not

recognizable.

4~3 PARALLEL STRUCTURE FACTOR AND DECAY RATE. The

suppression

of

edge

fluctua-

tions

by

confinement between two walls leads to a better

agreement

between the simulation results for the

parallel

structure factor for not too

large

q values and the theoretical result

according

to

equations (9, 10)

than found before

by

Abraham and Nelson [8] as shown in

figure

4~

+ +

~ +

0.1 +

~

+ ~

~ + +

+

~

+

~~/

+ +

i(~)

~~/

+

ooi

oooi

o-i i to

q

Fig.

2.

Log-log plot

of the initial decay rate

~(q)

=

r(q)/kBTq~

as a function of q for both the

isotropic

average of

r(q)

(lower

curve)

and

r(q i) (upper curve)

for a tethered membrane with Lo

= 50.

(12)

~

i

~

~~_

+~ o

~

~

~ +o oi >

~

x~q~> oi

ii

11~i$& ' ~

'~/ ~ ' ~~

°

~ ~ " '

o-o

i o

~ L

Fig.

3.

Log-log

plot of the initial

decay

rate

x(qi)

"

r(qi)/kBTq )~~'

as a function of qiL with ( = 0.65:

(+)

Lo

= 30,

(0)

Lo

= 50.

o.oi S(q~)

w

o.oooi

o-i i lo

q~

Fig.

4.

Log-log

plot of the structure factor

S(qi)

as a function of qi Data

points: S(q)

for a tethered membrane with Lo

" 20 in

a slit of diameter H = 3.5. Solid line:

S(q) according

to

equations (9), (10)

with L

= 19.9, B

= 0.05 and qjj = 0. Value for L obtained from the simulation.

4. The

corresponding

initial

decay

rate,

r(Qi),

shows a similar

good agreement

with the

decay

rate calculated from

equation (18) presented

in

figure

5. Another

possibility

is to compare the theoretical curves for the confined membrane with the structure factor

S(q~y)

and

decay

rate

r(q~y) parallel

to the walls of the slit. However there is no notable difference with

S(qi)

and

r(qi)

and therefore these results are not

presented

here.

(13)

lo

i

Do $

o ~

1lq4

o-i ° S

o

Slj$

h/ W ~9

o.oi

o.ooi

o-i i lo

q~

Fig.

5.

Log-log

plot of the initial

decay

rate

~(qi)

"

r(qi)/kBTq (

as a function of qi The HI interaction with the wall is not included. Data points:

~(qi)

membrane with Lo

= 20 in a slit of

diameter H

= 3.5. Solid line:

~(qi) according

to

equation (18)

with L

= 19.9, B

= o.05,and A

= 0.62.

Value for L obtained from the simulation.

4~4 HYDRODYNAMIC INTERACTION wiTH THE WALLS. Sofar for the membranes confined

to a slit the HI with the walls has not been included in the simulation results discussed. The

reason for this is that the presence of walls sofar has been intended to diminish the

edge

fluctuations and not to represent the membrane as

being physically

confined.

However,

if the walls are considered as real

physical objects

present to

study

the influence of confinement

on membrane

dynamics,

the influence of HI with the walls is relevant.

Further,

the results obtained may

depend

on the orientation of q,

I-e-, parallel

or

perpendicular

to the membrane

surface

or to the walls.

For the short-time diffusion constants there is no difference if HI with the walls is included

or not. This is due to the

tendency

of the membrane's centre of mass to avoid the

walls,

since the

neighbourhood

of a wall reduces its translational

entropy, thereby minimizing

the HI with the walls. The influence of HI with the walls should be even less for

Di

than for

Djj

since translation

parallel

to the membrane

surface,

which is on average

parallel

to the

walls,

results in smaller HI with the walls than translation

perpendicular

to the walls.

However,

the latter effect is of

secondary importance

since both

Di

and

Djj

are not influenced

by

HI with the walls.

Also for the initial

decay

rate

parallel

to the membrane

surface, r(Qi),

or the

walls, r(q~y),

introduction of HI with the walls does not make a notable difference

(results

not shown

here).

However for q

perpendicular

to the membrane as

presented

in

figure 6,

or

perpendicular

to the walls as

presented

in

figure 7,

there is a difference. It is clear that for q > 1 the presence of HI with the walls decreases the

decay

rate. The effect is

stronger

for

r(Qjj)

than for

r(qz).

Further,

for all q values

r(Qjj)

<

r(Qz).

These results can be

easily

understood in

physical

terms

realizing

that the medium between a wall and the membrane is

pushed against

the wall

by

a local move of the membrane towards the

wall, causing

backflow of medium and

thereby

a disturbance of the

velocity

of the medium at the

positions

of membrane beads. On the other hand for a move

parallel

to the wall the

displaced

medium flows almost

freely parallel

to the walls

thereby causing considerably

less disturbance

(14)

o-i

T~q,i

o.oi

o.ooi

o-i i lo

ql

Fig.

6.

Log-log plot

of the initial

decay

rate ~(qjj =

r(qjj /kBTq(

as a function of qjj for a tethered membrane with Lo

" 20 in a slit of diameter H

= 5.5:

IQ)

HI of the membrane with the walls

included;

(+)

HI of the membrane with the walls not included.

1iq4 o-i

o.oi

o-i i lo

qi

Fig.

7.

Log-log plot

of the initial decay rate

~(qz

=

r(qz /kBTq)

as a function of qz for a tethered membrane with Lo = 20 in

a slit of diameter H

= 5.5:

IQ)

HI of the membrane with the walls included;

(+)

HI of the membrane with the walls not included.

of the

velocity

at the

positions

of the membrane

beads,

as reflected

by

the

insensitivity

of

r(qi

and

r(q~y

to HI with the walls. The effect is

larger

for

r(Qjj

than for

r(Qz ),

because an

instantaneous orientation of the membrane surface will not

always

be

parallel

to the walls of the slit and

r(qz

contains a component

perpendicular

and

parallel

to the surface

and,

as discussed

before,

the latter

component

is

hardly

influenced

by

HI with the walls. The faster diffusion of

(15)

the membrane

parallel

than

perpendicular

to its

surface, Di

>

Djj,

in combination with the

previous

notion that the orientation of the membrane surface is

only

on average

parallel

to the slit

walls, explains

that for all q values

r(Qjj)

<

r(Qz).

5.

Concluding

remarks.

It is shown that the simulation data for the

isotropically averaged

structure factor for a self-

avoiding triangularly

tethered membrane are in

agreement

with the theoretical results obtained for a disc. This

strongly suggest

that a

self-avoiding

tethered membrane is flat and not crum-

pled.

The

advantages

of

using

the structure factor for small and not for intermediate Q, are

its

insensitivity

to

edge fluctuations,

which are rather

large

for the

relatively

small membrane diameters used in

computer simulations,

and the

possibility

to compare with the

simple

model

structure factor of a disc. Under the

right conditions, I.e.,

for membrane sizes

being

neither too

large

nor too small as

compared

to the available wave

length

of the radiation

used,

this

criterion

might

also be useful in

experimental

cases,

especially

since a

scaling analysis

of data to make a distinction with a

crumpled

membrane is far from trivial.

The short-time diffusion constants obtained from

simple

model calculations on a disc and simulation data show a close

agreement, giving

further support to the statement that a self-

avoiding triangularly

tethered membrane is flat.

The

comparison

between the simulation data and theoretical curves for structure factor and initial

decay

rate of a confined tethered membrane shows that the

edge

fluctuations can be

effectively suppressed by

the walls.

It is shown that the

self-avoiding triangularly

tethered membrane in the intermediate scale is

likely

to relax

primarily by out-of-plane (bending)

modes even in the direction

parallel

to its

surface, suggesting

that the tethered membrane behaves as a rather stiff sheet. It is clear however that to use this result in the

interpretation

of a

scattering experiment

is a subtle matter. It

especially requires

that the relaxation times for the fluid are

considerably

less than for the internal relaxation modes of the membrane and that all the internal relaxation modes of the membrane can be

probed

in the

experiment.

The latter condition is

probably

best

guaranteed by

the use of a membrane which has a

bending rigidity

which is not too

large.

It should be realized that the effect of

hydrodynamics

with the walls on the

decay

rate and short-time diffusion constants of a confined membrane is based on crude

approximations

as stated before in the context of

polymer hydrodynamics [30].

HI of a membrane bead

is modelled as the HI of a

single

solid

sphere

with a wall.

Further,

in

superimposing

the

single

wall contributions to model HI with two

walls,

is a rather

strong approximations

which will

definitely

fail if the walls are too close

together [31]. Furthermore,

the HI of a tethered

membrane with a wall is dealt with as the HI of a collection of N

independently moving spheres,

that is not a trivial

assumption. Therefore,

results on HI with the walls of a tethered

membrane are

likely

to be valid

only qualitatively, especially

for small slit widths.

Having

said

that,

the effects of HI with the walls on the

dynamics

of the membrane seems

interesting enough

to warrant further more

quantitative

and detailed

investigations.

Acknowledgements.

This work was

supported by

the Commission of the

European Community

within the framework of the "Human

Capital

and

Mobility"

programme. The author thanks K. Kremer for the

critical

reading

of the

manuscript. Stimulating

discussions with K.

Kremer,

R.

Lipowsky,

R-R-

Netz,

M. Kraus and U. Seifert as well as the comments of an anonymous referee are

gratefully acknowledged.

(16)

Appendix.

The effect of HI on the diffusion tensor of a solid

sphere moving parallel

to a

single

wall is known

[31, 32]

~~~

~~~ ~ 6~

~

~~

~ ~°~ ~ ~ ~~~~

D""

=

) (-(

In

l~ ~)

+

0.9588)

~

for y

i

1

7rna l/

where y

=

a/h,

with h the distance between the centre of the

sphere

and the

wall,

and a

= x, y.

For the HI of a solid

sphere moving perpendicular

to a wall the

corresponding equation

is

[31, 33]

D~~

=

) (I (y

+

jy~

+ for y <

(22)

7rna

D~~

=

~~~

~

ln

~)

+

0.971264)

~

for y

I

6~rqa

1- y 5 y

For y = 0.76980 the short and

long

distance

approximations

for the

sphere moving parallel

to a wall are

equal.

For y

= 0.4250959 the short and

long

distance

approximations

for the

sphere moving perpendicular

to a wall are

equal. Labeling

the two

parallel

walls with a and b

respectively

and

superimposing

corrections to the HI of a

sphere

due to a

single wall,

the

diagonal

elements of the diffusion tensor for a

sphere

between two walls are

expressed

for ya <

0.76980,

yb < 0.76980 as

D~~

"

$ (1 (lva

+

vb)

+

jlva

+

vb)~) 123)

for ya §

0.76980,

yb > 0.76980 as

~~~

~~~

~6~~

~

~~

~

5

~~

b~~

~ ~

~~~~ ~

for y~ §

0.4250959,

yb I 0.4250959 as

D~~

"

) (1 (lva

+

vb)

+

jlva

+

vb)~)

and for y~

§ 0.4250959,

yb > 0.4250959 as

D~~

=

) -(y~

+

jy(

+ ~~

In ~~ +

0.971264)

)

7rna yb l/b

Due to the

symmetry

in the indices a and b of the

equations,

a and b can be

interchanged

where dictated

by

the

position

of the

sphere

with respect to the walls. The

superposition

of the corrections of the two walls to

get

the overall correction to the diffusion tensor is an

approximation suggested by

Oseen. This

approximation

assumes that the

hydrodynamic

in- teraction with a wall is not influenced

by

the presence of the other wall. As stated

by Happel

and Brenner [31] this can

only

be correct if the walls are not close

together

and have a finite extension.

They

showed

explicitly

for a bead

moving perpendicular

to a wall that the presence of a second wall leads to a difference in the correction as

compared

to the correction obtained

(17)

by

the

superposition approximation.

Nevertheless the differences are

only small, although

one

might

expect that for

a

sphere

which has a short distance to both walls the

superposition approximation

is

highly

erroneous. That situation however is not encountered for the tethered

membrane studied here.

To include the HI of the membrane beads with the walls in the diffusion tensor, either Oseen

or RPY tensor, the

diagonal

elements on the r.h.s. of either

equation (16)

or

equation (17)

should be modified as

D$"

e D"* and D]~~ e D~~

References

ill

Schmidt C-F-, Svoboda K., Lei N., Petsche

I.B.,

Berman

L.E., Safinya

C.R. and Grest G.S., Science 259

(1993)

952.

[2] Wen X., Garland

C-W-,

Hwa

T.,

Kardar M., Kokufuta E.,

Yong L.,

Orkisz M, and Tanaka

T.,

Nature 355

(1992)

426.

[3]

Stupp

S.I., Son S., Lin H-C- and Li L-S-, Science 259

(1993)

59.

[4] Doi M, and Edwards

S-F-,

The

Theory

of

Polymer Dynamics (Clarendon Press, Oxford, 1986).

[5] Burchard

W.,

Adv.

Polym.

Sci. 48

(1983)

1.

[6] Grest G-S- and Murat M.,

Computer

Simulations of Tethered

Chains,

review,

preprint,

to ap-

pear in Monte Carlo and Molecular

Dynamics

Simulations in

Polymer Science,

K. Binder Ed.

(Clarendon

Pre55, Oxford,

1994).

[7] Ho J-S- and

Baumgirtner

A.,

Phys.

Rev. Lett. 63

(1989)

1324.

[8] Abraham F-F- and Nelson

D-R-,

Science 249

(1990)

393; J.

Phys.

France 51

(1990)

2653.

[9] Cates M.E.,

Phys.

Rev. Lett. 53

(1984)

926; J.

Phys.

France 46

(1985)

1059.

[10] Muthukumar

M.,

J. Ghem.

Phys.

88

(1988)

2854.

[III

Frey E. and Nelson D-R-, J.

Phys.

I France1

(1991)

1715.

[12] Lipowsky R., Adv. Solid State

Phys.

32

(1992)

19.

[13] Grest G-S-, J.

Phys.

I France1

(1991)

1695.

[14] Kantor

Y.,

Kardar M. and Nelson

D-R-, Phys.

Rev. Lett. 57

(1986)

791;

Phys.

Rev. A 35

(1987)

3056.

[15] Abraham F-F- and Goulian

M., Europhys.

Lett. 19

(1992)

293.

[16] Goulian M., Lei N., Miller J. and Sinha

S.K., Phys.

Rev. A 46

(1992)

R6170.

ii?]

Aranovitz J.A, and

Lubensky T-C-, Phys.

Rev. Lett. 60

(1988)

2634.

[18]

Gompper

G. and Kroll D.M.,

Europhys.

Lett, 15

(1991)

783; J. Phys. I France1

(1991)

1411.

[19] Plischke M, and Boa]

D., Phys.

Rev. A 38

(1988)

4943.

[20] Grest G-S- and Murat

M.,

J.

Phys.

France 51

(1990)

1415.

[21] Nelson D-R- and Peliti L., J.

Phys.

France 48

(1987)

1085.

[22]

Kratky

O. and Porod G., J. Golloid Sci. 4

(1949)

35.

[23] Akcascu A-Z- and Gfirol H., J. Polym. Sm., Polym. Phys. Ed, 14

(1976)

1;

Benmouna M, and Akcascu A-Z- Macromolecules11

(1978)

l18?;

Akcascu A-Z-, Benmouna M, and Han C-C-,

Polymer

21

(1980)

866.

[24] Dynamic

Light Scattering:

The Method and Some

Applications,

W. Brown Ed.

(Clarendon

Press,

Oxford, 1993).

[25]

Dfinweg,

B., Stevens M, and Kremer

K.,

Structure and

Dynamics

of Neutral and

Charged Polymer

Solutions: Effects of

Long-Range Interaction,

review, preprint, to appear in Monte Carlo and Molecular Dynamics Simulations in

Polymer

Science, K. Binder Ed.

(Clarendon

Press,

Oxford,

1994).

(18)

[26]

Stockmayer

W.H, and Burchard W., J. Ghem.

Phys.

70

(1979)

3138;

Schmidt M. and

Stockmayer

W-H-, Macromolecules17

(1984)

509;

Akcascu A.Z., Hammouda H.,

Stockmayer

W-H- and Tanaka

G.,

J. Ghem.

Phys.

85

(1986)

4734.

[27] Rotne J. and

Prager S.,

J. Chem.

Phys.

50

(1969)

4831;

Yamakawa H., J. Ghem.

Phys.

53

(1970)

436.

[28] Allen M-P- and

Tildesley

D.J., Computer Simulation of Liquids

(Clarendon Press,

Oxford,

1987)

pp. 147-152.

[29] Kantor Y.,

Europhys.

Lett. 20

(1992)

337.

[30] van Vliet J-H- and ten Brinke G., Macromolecules 25

(1992)

3695.

[31]

Happel

J. and Brenner H., Low

Reynolds

Number

Hydrodynamics

2nd. Ed.

(Martinus Nijholf,

Dordrecht,

1973) chapter

7.

[32] Goldman

A-J-,

Cox R-G- and Brenner H., Ghem.

Eng.

Sm. 22

(1967)

637.

[33] Cox R-G- and Brenner

H.,

Chem.

Eng.

Sci. 22

(1967)

1753.

Références

Documents relatifs

We focus on volumetric deformations and first establish the potential energy of fully confined bubbles as a function of their radius, including contributions from gas

Figure 2: Contour maps of the atomic density of oxygens of H 2 O (O w ) within the first molecular layer of water at the fully hydroxylated (001) surface of quartz, which is

The problem we tackle in this paper is directly related to such curves: we address the existence, the regularity, and the structure of minimizers of the elastic energy among W

They renormalize the elastic coefficients and lead to a large distance anomalous elasticity, as in the case of a single tethered

Molecular dynamics simulations of the structure of gelation/percolation clusters and random tethered membranes... Molecular dynamics simulations of the structure

In order to model the confined environment of proteins within the endoplasmic reticulum, the dynamic behavior of γ -gliadins inserted inside lyotropic lamellar phases

It is shown that flat, finite (or semi-infinite) fluid membranes with zero edge energy (line tension) and zero Gaussian curvature energy are unstable with respect to

Abraharn [12] showed very conclusively in his study of perforated open membranes that the order param- eter [33], defined as the ratio of the the mean square radius of gyration over