HAL Id: jpa-00247507
https://hal.archives-ouvertes.fr/jpa-00247507
Submitted on 1 Jan 1991
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.
Linear response of self assembling systems: mean field solution
F. Lequeux
To cite this version:
F. Lequeux. Linear response of self assembling systems: mean field solution. Journal de Physique II,
EDP Sciences, 1991, 1 (2), pp.195-207. �10.1051/jp2:1991155�. �jpa-00247507�
J
Phys.
II1(1991)
195-207 ', F#VRIER 1991, PAGE 195Classification
Physics
Abstracts05 20 36.20 61 25
Linear response of self hssembling systems
: meanfield solution
F
Lequeux
Laboratoire de
Spectiomktrie
etd'Iii1agerie
Ultrasonores(*),
Universitk LouisPasteur,'Institut
Le Bel, 4 rue B1alse Pascal, 67070
Strasbourg
Cedex,Franci
(Received
3 August 1990,accepted
mfinal form
29Octobei
1990) ''~ ~.
Rksumk. Nous ktudions
lq
rkponse hndaire desystdmes
vivants, c'est-I-dire s~associant et se coupant Itkqulhbre thermodynamlque
Une approximation de champ moyen donne une solutionanalytique
pour l'kvolutiontemporelle
de la reponse. Nous en ddduisons une expressionanalytique
de la viscositk despolymdres
vivants Nous trouvons de nouvelles lois d~kchelles pourcette vlscositk dans la hmlte des temps de vie courts devant les temps de relaxations propres Ces
lois
dkpendent
defagon
non kvidente de~l'exposant
du temps de relaxation par rapport I lalongueur
de la chaine Nous vknfions par un calcul exact l'aspect critique de,larkponse
dans cette hmlte.,,
Abstract. We
study
the linear response ofliving
systems, which break and recombinerandomly
The mean field approximation gives ananalytical
solution for the time evolution of the response As an example, the viscosity of living polymers is computed Newscaling
laws for viscosity are found in the case of very short lifetimes The laws do notdepend
in an obvious wayon the exponent of the chain relaxation time with respect to its
length
The critical behaviour is confirmedby
an exact calculation in the limit of very short lifetimes,
1. Inboduction.
New «
living
» systems, such asliving polymers (sulfur [1],
selenium[2],
grant micelles[3]j
orphysical gels (See [4],
forinstance)
are now available forphysicists.
All theseSystems
containelementary
bodies(e
gmonomers)
These bodies self-assemble in variousgeometnes (we
say in various
states)
The word«living»
means that a givenbody,
atthermodynamical
equilibrium, changes randomly
anddiscontinuously
of stateduring
time For instance, amonomer in a
living polymer system belongs
to a chain whoselength
varies in timeby
breaking
andrecombining
with an other chain.Hence,
atthermodynamical equilibnum,
every state has a finite
lifetime,
and we can define a jump rate from one state of another Weare interested in the linear response of iuch
systems
to any extensiveperturbation,
such as strain stressrelaxation,
electricbirefnngence,
time correlation functionsThere is a fundamental difference between the effects of finite lifetime and of the
polydispersity
Apolydisperse
system response is theaveraged
response over all the states.(*) CNRS UA 851
196 JOURNAL DE
PHYSIQUE
II lV° 2Conversely,
for aliving system,
the response is acomplex
combination ofjump
rates and of thekesponse
of each state : abody
in a state A may relax in its own state, but also in any otherstates after some
jumps
We assume the linear
approximation
Hence we consider that both thedensity
of states andthe transition
probabilities
arealways
theequilibrium
ones. It is then crucial to obtain a conservation law for the responseduring
a recombination. This gtves theequations
for the response relaxation.In order to solve this set of equations, we make a mean field
approximation
This consists intaking
thebody
thatchanges
of state ascarrying
the mean response of thesystem.
Wegive
also the expression of the mean fieldapproximation
error. We solve the equations and get ageneral
expression of the Fourier transform of the macroscopic response. We then deduce theviscosity
and the vlscoelastic modulus ofliving systems.
We obtain an expression for the response function that is ageneralisation
of that of apolydisperse system.
We calculate theviscosity
ofliving polymers.
Ourapproach,
is limited tosingle exponential decays
for each aggregate, and gtves different results from those obtainedby
Cates[5]
in the case of the reptatlon process We discuss the limit cases, where the life time is smallcompared
to themean relaxation time. We denve the
scaling
laws of theviscosity
as a function of theexponent
of the relaxation time withrespect
to thelength.
We confirm some of these resultsby
an exactcalculation of the response in the lbnit case where the lifetime vanishes.
2. Definition of the system.
We consider a
system
constitutedby
bodies(monomers
orelses)
which self-assemble in differentpatterns
atthermodynamical equihbnum.
Let us call A the state of abody,
and S the ensemble of the states.Typically,
A is ageometncal
parameter which characterizes amorphology
of theaggregates
It can be thelength
ofpolymer,
or the size of a cluster(see
Fig, I).
For each state A, there is adensity n(A )
ofaggregates.
We normalizen(A) by
:j.h(A ) V(A )
dA = I(1)
s
where
V(A )
is the number of bodies that constitute anaggregate
of type A. Theintegral
overS is the
integral
over all the states.In the absence of
change
between states, I-e- for an idealaggregate
which would stay itsFig
I-Living
systems are systems where bodies self assemble in aggregates that break and recombine at the thermodynarnicalequihbriurn
A describes themorphology
of each type of aggregate B is the elementarybody.
N 2 LINEAR RESPONSE OF SELF ASSEMBLING SYSTEMS 197
pnmary state, the response to a
perturbation
woulddecay
in time We hmJt ourselves to asingle exponential decay
The response «~ of abody
state Adecays according
tobW~ W~
~
T~~~~~
and T~~ is the relaxation time of the state A Let us now define the transition
probabilities.
We will first consider the case where one
body changes
statealone,
withoutcomb1nlng
with others. We will show later that the case of a manybody
association amounts to that of asingle body
state.For each
pair
of states(A, A'),
there is a flux of bodies from A to A' Let~
be the transition
probability.
The
equation
of conservation for the number of bodies is thenbn(A V(A n(A V(A n(A') V(A')
dA'~~ ~(A-s) ~~~
~
s T(A-A)
where T~~ _~~ is the lifetime of state A and is gtven
by
~
A~-
s)
S'
~
~~
')
~~~
We suppose here that the measure
(in
themeaning
of theintegral)
of a state isinfinitely
small with respect to that of the whole system.Thus, integrals
over S areequivalent
tointegrals
over S-
(A).
We have to assume a law for the response conservation
dunng
the state transition. This is acrucial
hypothesis
as it is necessary to set asimple
rule in order to continue thestudy.
We limit ourselves to the case where this rules can beexpressed
as :trA<(t)
=
~r~(t) (5)
where « is the response
by body,
A and A' are the initial and final state of ajump taking place
at time t. This amounts to assume for instance that transitions take
place
very fastLet us consider now the case where two or more
aggregates
can combine. The transitionsoccur either
by
combination of twoaggregates,
orby splitting
of anaggregate.
First,
twoobjects
with states Ai and A
~ can stick and jump to the state A~
(see Fig. 2).
The combinedprobability
of this event is :n(Ai) n(A2)
~(AI,A~-A~)
For the inverse transition, we have
s1mllarly
:~(~3)
~(A3~AI,~2)
One could also take into account other non linear terms without
large changes
in what198 JOURNAL DE
PHYSIQUE
II N 2if' ~~ ~'
~(i,,iz~~~)
~~3
_
<
_
"~
fi lJ3-1~,ij
~
,
Fig. 2 Two aggregates
(at
state Aj and A~) combine into an aggregate at state A
(and reciprocally,
the aggregate at state A~ separates m two aggregates of type Al and A~) ~ ~
follows. We can then define the transition rate of a
body
from one state to anotherby
:i
n(A~) dA~
,
=
(6)
~(Ai-A~)
s ~(Ai>A2-A~) In
addition,
the conservation rule for the response is :~r~~(t).
VjA~)
= ~r~~ V(A i)
+~r~~.
V(A~)
Hence the rates of the response
traitsition during
the interaction can beexpressed
for each statesimply
using(6)
:n(A1) «~~.
V(A i) n(A~) «~~.
V(A~)
for state I and for, state 2
(7)
~(AI~~3) ~(A2-A3)
For many
body interactions,
and in the case of an extensive response, wecan'get
asingle
state like expression forthiiesponse
fluxesby
averaging, for eachaggregate,
over all the others aggregates which are able to associate with it We can do thisonly
because we are atequilibrium,
and all the quantities involved(state
densities and transitionrate)
are constant We nowcompute
the evolution of the macroscopic response.3.
Equations
for the time evolution of the system.An intuitive way to
compute,
the evolution of suchliving
systems is to consider the timeresponse of one
body
thatchanges
states, and toaveraged
over all thepossible
time evolutions This leads of course to thefollowing
functionalintegral
for ~the macroscopic responsei~r(t)>
=
PA
(i) ~r(t
=
0)
arIA It))
D(A (t)) (8)
where
P~
is the response propagator for abody belongtng
to a state Avarying
~ in time, ar thedinsity
oftime-dependent
state(t), D(A (t))
is theintegrand
over all thepossible
time evolutions. We canexpand
theintegral
and express it as a converging series of classicalintegral.
We will use this formalism in asimple limiting
case in section 8Another way for
computing
the response evolution is tobaljnce
the response fluxes for alllV° 2 LINEAR RESPONSE OF SELF ASSEMBLING SYSTEMS 199
states. Bodies
leaving
the'state at~time f carry the response «~(t).
We wnte the balance ofthe trans1tlon rates and of the response relaxations
Using (2), (3)
and(7)
we obtain :b~r> n(A V(A
~r>n(A V(A )
~r>n(A V(A
~r> n(A') V(A')
dA'bt ~~ T«> T(A-s)
~
s>
T(A-A)
and
becadse n(A
andV(A )
are constant, we have :3«A
«~ «~, n(A ') V(A ')
dA '~~ T
~
s' ~(A' ~
(~ )
" ~ ~~~with the rule :
= +
(10)
~A ~ "A ~(A s)
We have obtained the
complete
set oflinear'equations
for the evolution of therdsponse.
A direct solution would require an inversion of a nonsymmetncal operator,
and would be verydifficult except for some
peculiar
casesWhen writing equation
(9),
we recall that we consider eachbody
at a state as carrying themean response of bodies at state A. We
neglect
both thedisinbuilon
of responses ata
given
state, and[he spatial
correlation intothi
aggregates. The secondassumptiin'is justified by ihe
fact that internal decorrelitions aregenerally
faster than responserilaxations.
For instance in the Rouse model[6,
7~aid
even for reptation ofliving polymers
[7~ ends effects aregenerilly
small
compared
to the whole chainrelaxatiin
Wepostulate
thatin
general
bothapproii-
mations are not constraining. In
fact,
inthe,
case ofrejtation,
end effects have a mareimportant relaxation,
asexp1alned
in 74. Mean field
approximation.
Equation (9)
describes how bodies at state Arelax,
leave the state, and how some others reach A carrying their own response. The mean fieldapproximation
consists in assuming that thebody
coming m statej
carries the average response of thesystem.
Thisapproximation
doesnot take into account correlations between several states It can
onlj describe
a system
fir
which transitions between states arediscontinuous,
likeliving polymers
withbriakmg,
orliving Cayley
-trees, but notliving polymers
which growlonger
and shorterincrementally'or
'living'networks.
We will discuss thevalidity
of the mein field approximation later.The mean field approximation consists into
substituting
in(9)
~«A,
n(A') v(A')
dA~ i
~,~ ~~,~
~~~,~
~~, ~~~~s< TjA<-A)
~
T(s-A)
s,"
~which is
equal
to(«)
/T~~~ where
(«),
the mean response isaveraged
over all the bodies :j«>
=
«(A n(A ) v(A
dA(12)
In order to compute the value of T~~_ ~j, we
uie
the samemein
field approximation in the balance for bodies at a gtven state Thlscorresponds
tosubstituting
in theequation (3)
i~~~'~ ~'~~'~~~~ by ln(A') V(A')dA',
s' T(A>-A) T(s-A)
s<
200 JOURNAL DE
PHYSIQUE
II lV° 2Since-the state
density
is normalized(I),
relation(3)
atequihbnum gives,
in the mean fieldapproximation
:,
~
nlA VIA
~
l
~i~~
~(A -s)
T(s-A)
Thls gtves a definition of T~~_~~ which is coherent with
body
number balanceand
response balance. We deduce from(13), (11)
and(9)
the mean fieldequation
b£rA £rA
(£r)
£rA £rA
j ~)
= + = +
(14)
bt T~A T(A _~) f T(A ~)
We see
immediately
thespecific
effect of theliving systems
if «~ relaxes faster(without
statetransitions)
than(
«)
in theliving system,
then(
« ~ «~ and the relaxation at state A will be slowed down
by
state jumps. A contrario, if the isolated state A tends to relax slower than all the system, relaxation will be acceleratedby
state transitions.5. Discussion of the mean field
approximation.
The mean field
approximation
consists inreality,
inpostulating
thatT~~ ~,~ does not
depend
on A
',
butonly
on A It is clear that wenegleut
correlationduring
the transitions. For instance, for aliving polymer
which growslonger
and shorterby
increments, there are transitionsjust
between the
polymer
ofneighbounng lengths.
In such a case, the mean fieldapproximation
iscompletely
irrelevant.Conversely,
the approximation isquite gold
forliving polymers
which break and recombine at randomIt is easy to compute from
[8] (see
also AnnexI)
the transition rates from a chain oflength ii
to a chain oflength i~
2 ci if
ii
~
i~
T~i~_i~~ c~n(f~ ii)
iff~
z~ii
where ci and c~ are
respectively
the break and the association rates Thus the mean fieldapproximation
isgood
for smallf~,
but not for recombination(see Fig. 3).
We nowgive
ananalytical
expression for the error on the response due to mean fieldapproximation.
Mean field self consistency error
We can find an
analytical
expression from(9)
for theexalt-
mean response.
Integrating (9)
on A gtves
~(~
"
~~
~~~ ~'~~
dA +
~~'~~~'( ~'~~'~
~~'~~(15)
S A s s' (A'-A) f
Integratlng
now the mean fieldequation (14)
on A, we haveaj«)
=I«A n(A V(A )
«~n(A V(A )
dAdA +
(16)
bt
s
fA
s
~(A s)
So the
systematic
error E due to mean fieldapproximation
on the time denvative of the macroscopic response is denved from(13)
and(14)
:E=
~~~~~~'~~~~~~~~~~'~~~~(«Aj- «A~)dAidA~dA~
sj s~ s~ ~(Ai ~)
lV° 2 LINEAR RESPONSE OF SELF ASSEMBLING SYSTEMS 201
ii I
Fig
3. The mean field approximation consists insubstitutlnj
a constant to I/T~t~ t~j The continuous line represents the actual value of the transition
probability
for livingpolymers
and the broken line, thevalue taken in the mean field approximation
which vanishes if
T~~, ~ does not
depends
on A'. We will show later how tocompute
« in themean field
approximation. Equation (7)
allow tocompute
the error once the mean field response is calculated. The formalism descnbed here can then bedeveloped
in order toperform
aperturbation
calculation about the mean field6. Resolution of the mean field
equation.
It is now
simple
to compute selfconsistently («)
from(14),
because the integration of(14)
over S gtves the time denvative of
(« )
in function of(«)
But we firstperform
the Fourier transform of(14) multiplied by
the Heaviside functionY(t)
:tW~A(W )
"A(t
"
°)
"
~
~~~
+
~~(~) ~~
)
~~~~A
where WA
(w )
is the Founer transform ofY(t). «A(t).
«~
(t
=
0)
is the1nltial value of the response Ingeneral,
one takes «~(0)
constant, but forsome
special systems,
it can be necessary to use other initial conditions. Hence the Fourier transform of the response at a state A is :WA
(w )
=
~~~~~
+
~~~ ~°'~
(18)
lW +
~/~A
~(A-s)(lW
+I/TA) "(~ )
and integrating over all the A,
weigthlng by n(A) V(A),
we obtain the self consistentexpreision
for(&)
:n(A V(A ) trA(o)
~~
,~ n(A ) V(A )
dA~"~ ~°'~
~~~~~ iw +
I/Y~
~ ~ ~°s T(A
-s)(iw
+I/fA)
This gtves the modulus G* in the case where « is the stress response, assuming
«~
(0) independent
ofIWT~A T(A
-s)n(A ) V(A )dA G*(iw)
=
Go
~~'~"~ ~~~ ~~~ ~ ~"~ ~ ~~~ ~~~(20)
(iwT~~
+ I T~~ _~~n(A ) V(A )
dAlW~aA ~(A -s) + ~aA + ~(A-s)
202 JOURNAL DE PHYSIQUE II N 2
where
Go
is a constant(Go
is theplateau modulus),
and the zero shear rate viscosity has thesimple
expressionn
(A V(A ) f~
dAn =
Go (21)
~
~)jl~ n(A ) V(A )
dA(See fop
instance in[9],
theK~amers Kjoenig
relations in linearvlscoelasticity, for
theprecise rilations
between response and_modulus)
,
Limiting
cases.Long I)etime.
In the caseqf a,system
with very slow trans1tlons, in theljmit
T~~ ~~ ~ co,
equ(tion (19) becoiles
,
n(A ) V(A )
trA(0)
~ ~
(22)
l~)~~°'~"
iw+I/T~A
This is of course the response of a
polydisperse system,
and the Fourier inverse transformgtves the classical result
,.
" l J
~ " ~
(~ )
"(~ )
"A(°)
eXp(- t/T~A )
dA.~ ,
Cqnstant
relaxation time. In the case where the relaxation time does notdepend
on the state A, the modulus must notdepend
6n the lifetime Indeed in(20)
T~~ can be factorized andwe obtain
, iwT~~
Go
~i G*
(iw )
= .,-.
,
iwT~~ + I
which is of course the modulus of a
single
time relaxationsyiteni
,(nfimte
time relaxation In the case of infinite timerelajations
of all the states, we have(&)
~p = 0 infinite and in fact the Founer transform isirrelevant'because integrals diverge
This is evident because the
system
does not relax.~ <,
7.
Viscosity
ofliving polymers.
,~lfe
use the model of~randombreaking
and ends recombination~ as in[8]
The lifetime of a chain oflength f
is calculated in annex I The rule of stress co@ervation isconsistint
with(7), V(f) being simply f.
We suppose that the relaxation time of the chain scales withlength
:~'
(fjf)d
'~T~ " T
f ~
where the
exponent
«equals
3 for areptation
model[10,
7] in the case of melt or semi-diluteliving polymers,
2 for the Rousemodel,
and3/2
or 3v/2
for the Zimm model[6, 11,
7~ in thecase of dilute solution
Imphcitely,
whiletaking
thesemodels,
we assume that end contribution isnegligible
We will discuss m aforthcoming publication
the consequences of thespecific
enddynamics
lV° 2 LINEAR RESPONSE OF SELF ASSEMBLING SYSTEMS 203
Thus the
viscosity
is deduced from(21),
and we get, with x =iii, (
=
T~/T~
and for the lifetimeT =
T~/(x
+2) (see
in AnnexI,.Eq.'(33))
:lm
~«+l~-x~~
q
~GOT~
°~~~~'~~~~~ (23)
j*
xe~~dxo~t+x~. (x+2)
in the limit
t
~ co, q reduces tolw x~+~e~~dx
llm q
=
Go
T~ °~
=
Go
T~r(«
+2) (24)
'~"
o
xe~~dxThus
I
scales likeT~ iq the case T~ « T~ We are in the limit of
a,polydispersj system
On the other hand in the limit
(
~
0,
one obtains m~~-x~~~'
~
x+2
GOT«C
q((~0)"G0~« (25)
m
~j-x~~
T ~I '~:
~
(+x~(2+x)
where C is the
integral
atthe'numerator
and I that of the denominator Theintejral
Idiverges
for(~0 when,«
m2. We have then threescaling
laws for the viscosity in~the limit T~ « T~
a)
For« <
2,
q((
~
0)
~ T~ and has then a finite value for
(
=
0.
In the case «,= 0 we found
(as previously explained)
q=
Go
T~ whatever T~,b)
For .«=
2,
using achange
ofvariable,
we have, , ~
j*
y e~ ~~dy j"
e ~~dY
ln(a )
o +
(2
+aY)
Y~i Y
with a
=
(~'~.
Hence the
scaling
law for K= 2 is
~J q
(<
- 0) T~/In (T~/T~ (26)
c)
ForK ~
2,
wehive.
, j
1=j(2-«)/«, j"'
Ye~~dY
.. o +
(2
+ay) y~
with a
=
( ~'~.'Thi
second factoris an
integral
which has a finite limit for a= 0. We have
then the
scaling
law for K & 2'l
(~
- 0
)
~ T
i~ T)~
~~~~(27)
Figure
4 illustrates these different cases.204 JOURNAL DE PHYSIQUE II lV° 2
Fig.
4 The different behaviours, in alog log
representation, of the viscosity q as a function of ( theratio of the mean lifetime and the mean relaxation time, for different values of the exponent «, the mean
relaxation time remainJng constant For w
=
0, q is constant For w = 2 there is a
loganthmlc
divergences,
for w~ 2 the slope of log q as a function'of
log
( is (w2)/w
Remark the
scaling
law for viscosity, in the case Km 2 is obtainedby estimating
the denominatordivergence
for(
= 0 In tits
limit,
the effect of very small chains is veryimportant.
If thescaling
law for the relaxation time of the very small chains is different from that for thelonger chains,
the abovepredictions
are nolonger
validAlso,
if the mlnlmumchain
length
is non zero, there is a cut off near x= 0 in the
integral,
which does notdiverge
for
(
=
0,
and thescaling
law is invalid This effect can be important in semidilutesystems,
because for chainsshorter
than the meshsize
f,
the relaxation will becompletely,
different from that for chainslonger
thanf.
Theapproach
above allows us to compute viscosity in suchcases
Our results differ from those of Cates
[5],
who findphysically
in the case ofreptation («
= 3
),
q~ T
('~
Tj'~, whereas we find q T)'~ T('~.
This is due to the fact that we do not take into account the wholereptatlon
process. In the tube model[10],
the ends and the middlepart
of the chain relax with different rates.Moreover,
the relaxation function is not asingle exponential,
but a morecomplicated
function. This is consideredby
Cates[5],
but not in ourpaper.
However,
in the case ofsingle exponential decay
process, our model should wellapplied,
as for instance in the case of rod-likeliving polymers [12].
8. Exact results for
(
= 0.
We
present
now another way to derive the critical behaviour of the response near(
= 0 and« =
2. We use the direct functional
integral
calculation(8) presented
in 3. In the limit(
=
0, during
any finite space oftime,
abody
will runssuccessively
over every state of the system, and can therefore be considered as a «mean »object
: we can substitute theintegration paths (t) by
a meanpath
running over the entiresystem,
andoccupying
each stateduring
at timeproportional
to thebody density. Actually
we use theErgodic theorem,
the whole system isfully
describedby
eachbody, dunng
a time infinitesimalcompared
to thatof the response relaxation. We also make use of the
property
of commutation of thepropagators P~~i~.
Thepropagator
is aproduct
ofpropagators
of(2)
and isl'A(t)
" eXp(- j~
~~l. (28)
0 ~ WA(t)
N 2 LINEAR RESPONSE OF SELF ASSEMBLING SYSTEMS 205
Hence,
in the limit(
=
0, equation (8)
becomes :j~r(t)j
= ~r
(0),
exp(- @
dA(29)
s
~«A
where
IT (A )
dA= t and T
(A )
isproportional
to thebody de~lsity n(A ) V(A )
We use thes
normalization convention of
equation (I)
and obtain :j~r(t)>
=
~r(0)
exp(-
t~(~ ) "(~ )
dA
(30)
s ~«A
The response for
(
=
0
decays exponentially
if theintegral
converges In the case ofliving polymers,
theintegral diverges
for « m 2(see
Sect.7)
We find also a cntical behavlour near« = 2. This shows
clearly
that the results of the mean fieldtheory
are consistent with the exact solution in the limit(
= 0.
Moreover,
in thislimit,
theviscosity
can be deduced from(30)
and is :~
~~
~~~
~~°
n(A V(A
~~
~~~~
s
A
For
comparison,
theviscosity
in the limit(
= co derived from equation
(22)
isq
(<
= co
)
=
Go n(A ) V(A
T~~ dAThis leads to the
following physical
remark for(
= co the response is a
superposition
of all individual responses mparallel
On the otherhand,
for( =0,
the response is asuperposition in series of the individual responses.
Let us remark that the exact
viscosity
in the limit(
= 0
(31)
isexactly
the same than theexpression denved in the mean field
approximation
fromequation (21)
in the limitT~A -s) " 0 This confirms that these mean field results for
hung polymers
q((=0)=0
for «m2and
q
((
= 0
)
# 0 and finite forK < 2
are not artefacts.
The cntical behaviour of the response near
(
= 0 and K
=
2 is confirmed
by
this exact calculation. We have also shown that whatever thecomplexity
of the system, the responsedecays exponentially
in the limit(
=
0. This result is in
agreement
withexpenmental
data gtven for instance in[13] (for
someconcentrations)
; for moredetails,
see[14].
Let us remark that we have used the
properties
of the Ppropagators
which requires thatone
aggregate
has asingle exponentiel decay
Generahsation to morecomplex
systems must becarefully developped.
9. Conclusion.
In this paper, we have studied the linear response of
living systems,
such asliving polymers.
A206 JOURNAL DE
PHYSIQUE
II lV° 2mean field
approximation
can beapplied only
if the aggregates break and recombinerandomly,
and if the response is extensiveduring
transitions.According
to the mean fieldtheory,
we calculate ananalytical
expression for thecomplex
modulus. We alsosuggest
aperturbative
calculation near the mean fieldsolution,
in order to take into account transition correlations. Weapply
our model toliving polymers
and observe that the viscosity, in the casewhere the mean
hfetimq
T~ is smallcompared
to the mean stress relaxation timeT~, scales as
T)~
~)'~ T)'~,
where K is thescaling
lawexponent
for T~ with respect to the chainlength,
for «~ 2
Conversely,
for K <2,
theviscosity
has a finite hmite for T~.- 0. These results are confirmed in the limit T~= 0
by
an exact calculations. The response has anexponential decay,
and the mean field viscosity in this limit is the exact one. All these results evidence the existence ofcomplex
interferences inliving
processes, between size,lifetime,
and relaxation time.In this
work,
we have limJted the model to asingle exponintial dicay
for isolatedajgregatei. ~our approich
will be extendedin a
forthcoming pubhiation,
inorder'to
includeend
effects'in living jolymers, by taking
into, acc6untthi poiitions
of monomers within the chainsAcknowledgments.
I am
grateful
to R.Hocquart
and YThinet,
for many discussions and a carefulreading
of themanuscript, to S J. Candau for
stimulating encouragements
We benefit ofhelpful
discussions from M. E.
pates
within the framework of, E-E C Grant number SO*0288-C(E.D B.)
Annex I
Recombinafion
dynandcs
ofliving po1jnlers.
' '~~~
We take the model of Cates and Turner end associations This leads to the
following
equation(see [5]) (where
ci is the breakprobability
and c~ the combinationprobability)
:,
<
~~
~~~ ~= cl
I
p
(I
2 c~p
(I ) j~
p(x)
dx + 2cl j~
p(I
+x)
dx +at
~ ~
f
+ C2
o
P(X)
p(~ X)
dX(32)
The first term is due to destruction
by
scission, the second to destructionby combination,
the third to creationby
scission and the fourth to creationby
combination.At, equlhbnum,
we have thedensity
:P
(I)
= P exp
(- iii)li~
',
with
I
A~~)~
where is thedensity
inpolymer
ci
~~~
~ -,:
The lifetime of a chain of
length I
is calculatedby
suppressing the two lasttennis
of equation(32).
The term in cj is thedispantion
rate due to scission, the term in c~ to recombination :im
~(~
+ 2I)
jj
~ ~~jp
~~~-
2 c~ P(I)
~
P ~~~ ~~
".~
N 2 LINEAR RESPONSE OF SELF ASSEMBLING SYSTEMS 207
So the lifetime of a chain is T~~ _~~ =
l/(ci(I
+ 2I)).
It can also be calculated from(4)
and(6).
We take T~ =I/ci I
and have then~~~ ~~ =
~~/(x
+2)
with x=
iii (33)
References
[I]
PETSCHEK R G, PFEUTY P, WHEE~ER J. C,Phys
Rev A 34.3(1986)
2391[2] FAIVRE G, GARDISSAT J L., Macromolecules 19
(1986)
1989[3] JINDA~ V, KA~US J., PI~SI H, HOFFMANN H, LINDNER P, J
Phys
Chem 94(1990)
3129 andreferences therein
,
PORTE G, APPE~~ J,
Europhys
Lett 12(1990)
190[4] PEzRON E, LEIB~ER L., PINCUS P A,
Polymer
29.6(1988)
l105[5] CATES M E., Macromol 20 (1987) 2289.
[6] ROUSE P. E, J Chem
Phys
21(1953)
1272[7] DOI M., EDWARDS F, The Theory of polymer dynamics (Oxford University Press) 1986
[8] TURNER M. S, CATES M E, J
Phys
51(1990)
307[9] FERRY J D, Viscoelastic properties of polymers Qviley, New York) 1980
[10] DE GENNES P G., J Chem Phys 55
(I97I)
572[1Ii
ZIMM B H, J Chem Phys 24(1956)
269 [12] CATES M E, to bepublished.
[13] SHIKATA T, HIRATA H, KOTAKA T, Langmuir 3
(1987)
I08I, 4(1988)
354, 5(1989)
398.[14] CANDAU S. J, HIRSH E, ZANA R, DELSANTI M, Mol Langmuir 5
(1989)
1225,KERN F, ZANA R, CANDAU S J, to be