Aggregation of Hydrophobic - CiteSeerX

0 downloads 0 Views 2MB Size Report
In fragmentation, particles of hydrophobic products may be produced by emulsification of the product in water, if it is a liquid, or grinding it down to a very small.
Phys.

J.

II

FYance

H4lAne Olivier

Aggregation

Hydrophobic

of

Chariol

RP, Physique

CEA

mixte

Laboratoire

de

Orsay, France Rhbne-Poulenc, 93308

1997,

FEBRUARY

(~), Anwar Hasmy (~), (3) and Bernard

Lannibois

Aguerre

Equipe

(~)

319-342

Limited

Surfactant in Water

(~

(1997)

7

Robert

Botet

Cabane

(~)

Rhbne-Poulenc, 93308 des Solides, Universit6

PAGE

319

Molecules

(~,*),

Aubervilliers, France Paris-Sud, Bhtiment 510,

91405

(~)

(Received

Aubervilliers,

July1996,

28

PACS.82.70.Kj

received

jn

France

final

form

1996, accepted 25

October

11

suspensions simulation modeling; random phenomena, processes,

Emulsions

1996)

October

and

Computational

PACS.82.20.Wt

PACS.05.40.+j

Fluctuation

and

Brownian

motion

hydrophobic molecules has been studied in presence of hydrophobic grow through aggregation of clusters molecules; the growth is terminated by adsorption of surfactant. accordThe particle qizes vary of hydrophobic molecules surfactant molecules. ing to the concentrations and of Two regimes have been found for the use of surfactant Inolecules: surfactant concentrations, an efficient at low adsorbed the surfaces of the growing particles; at reg1nle where all surfactant nlolecules are on high surfactant molecules left in concentrations, a wasteful reg1nle, where surfactant are excess Attempts to reduce the particle sizes by adding increasing anlounts of surfactant become water. inefficient surfactant These results point where nlost of the added renlains in water. at are sortie between aggregation s1nlulates the conlpetition explained by a kinetic aggregation model which well reThe results of exper1nlents of hydrophobic nlolecules and adsorption of surfactant. are where aggregation is allowed to proceed un1nlpeded for a time r, and produced by s1nlulations deternlined by conditions, particle sizes are adsorption of the surfactant In these then starts. the rate of aggregation and by the value of this t1nle delay. Abstract.

of

precipitation

The

Amorphous

surfactants.

added

in

of

water

particles

Introduction

1.

Many drugs currently reach

to

the

However, the is

much

too

amount

low

range from 10~8 the hydrophobic

development

under

sites,

active

these

drug

of

for

practical

g/g

to

molecules that

can

purposes:

g/g.

10~5

The

be

are

made

must

dissolved

typical classical

of

be

large hydrophobic

carried

across

way

to

achieve

this

in

the

In

order

environment.

aqueous

an

transported solubilities for large and

molecules.

medium

aqueous

hydrophobic transport

is

molecules to

dissolve

particles, I.e. supramolecular objects which may be dispersed in water. Typical carriers are polymeric particles, liposomes, micelles [I-4]. Still, carriers the of drug that be dissolved in most carriers is small, about 10$l of the amount can of where particles Therefore there is need design free delivery carrier system, to mass. a a (*)

©

Author

Les

for

iditions

molecules

correspondence

de

Phvsioue

into

carrier

(e-nlail:

1997

botettllps.u-psud.fr)

JOURNAL

320

drug could be covered with requirements are that the particle

pure

should

small

a

size

orier of100 m2 /g,

be of the

and

PHYSIQUE

DE

the

N°2

surfactant

of

amount

should

II

and

dispersed in specific

0.2 ~J, the

be

smaller

than

amount

of

surfactant

should

be

as

water.

The

surface

area

low

as

possihle.

dispersions may be obtained, either by fragmentation [5-8], or by growth processes fragmentation, particles of hydrophobic products may be produced by emulsification the product in water, if it is a liquid, or grinding it down to a very small size, if it is a solid. with growth processes, the hydrophobic product is dissolved in a solvent that is miscible

Very

fine

[9,10]. of In

In

and

water,

then

then

It is

solution

In all

is mixed

control

to

necessary

macroscopic

a

the

with

causing the

water,

precipitation

the

to

process

molecules

insoluble

precipitate.

to

particles

colloidal

obtain

instead

of

solid.

surfactants are needed, either to help the fragmentation (emulsification aggregation of the fragmented particles, or to control growth. The to prevent of large, sometimes comparable with the the surfactant amounts necessary are hydrophobic product. This reduces the advantage of such delivery systems over the specific toxicity of the formula. In this paper, we carriers, and may cause a non problem of the use of surfactants for one particular case: dispersions made through

these

processes,

grinding) or problem is that

of

of

amount

traditional address

the

precipitation

a

process.

hydrophobic product made of molecules that are nearly insoluble hydrophobic product is dissolved in a solvent that is miscible with The hydrophobic molecules water, and the solution is mixed with a large excess of water. are desolvated. instantly Because they are insoluble in water, they stick to each other whenever they meet, and dissociation do not there is no nucleation reactions In such conditions, occur. barrier, and aggregates of hydrophobic molecules grow immediately through a Diffusion Limited Cluster-cluster Aggregation (DLCA) mechanism [11-13]. If this aggregation is allowed to continue, the aggregates will become so large that they will consider

We

in

"model"

a

(10~8g/g).

water

This

reach

macroscopic

sizes

added

in

order

control

aggregates surfactant

their stick

nanometric

needed

aggregates

As will be

their

interior

adsorbed

obtained.

be

may a

po180ned aggregation final

the

control

to

surfaces.

(VH)

longer

description of

Th18

cause

no

fall

this

not

will

(precipitation). However, surfactants out of solution may be growth. Indeed, surfactant molecules adsorb on the growing may hydrophobic tails. Hydrophobic surfaces that are fully covered with saturated surfactant with to each other; hence, aggregates that are recombine with other aggregate8. In this way, a stable dispersion of

and

through may

molecules

by

to

shown

of the

sizes

below,

the

platicized by residual solvent and would require a ratio of surfactant

is

surfactant

given by:

volume

suggests

proces8

should aggregates aggregates are often

~ ~

be

that the

where

r

is

0.06.

thickness

the

desired

particle

However,

a

of the

diameters

100

are

general finding

is

that

nm,

the

which

of

surfactant covers

their

water.

Therefore

volume

(Vs)

a

to

particles, bemonolayer coverage hydrophobic volume

~~

layer

surfactant

amount

spherical, amorphous

~~~

R

the

the amount

and

of the particles. Typically, Vs/l~H should be of the order of needed to control the growth of surfactant inefficient of a monolayer coverage. This use

and

thus

amount

R

the

the

radius

ratio

vastly in excess of surfactants have possible molecules available origins: either the surfactant two not may are the particle surfaces during the but they do they available, aggregation near are process, or not protect the surfaces efficiently. formation of nanoparticles by aggregation of hydrophobic In the following, we report the surfactants; of different competition molecules in conditions have been set for the presence between aggregation and poisoning by surfactants, and the resulting particle sizes have been of

aggregates

in such

conditions

is

SURFACTANT

N°2

measured.

In

molecules

and

poisoned

order to interpret these data, we have (or poisons) are moved surfactants

aggregation

probabilities

has

reproduced

been

AGGREGATION

LIMITED

with

designed set

a

kinetic

a

random

at

of

321

model a

on

that

rules

hydrophobic

where

lattice.

The

determine

process

of

sticking

the

all species. This simulation shows of surfactants: two regimes for the use where the regime, surfactant molecules efficiently, used but the aggregation an are proceeds to rather large sizes, and a "wasteful" regime, where a vast excess of surfactant is used aggregation at an earlier stage. The comparison with experimental results shows to stop the that the preparation of nanoparticles is usually conducted in the "wasteful" regime where of the fraction surfactant in the phase. Dispersions of nanometric remains a large aqueous particles (sizes below 100 nm) cannot be obtained in the "efficient" regime, unless the surfactant adsorbs efficiently on the growing aggregates than the hydrophobic molecules themselves. more

of

"efficient"

Materials

2.

hydrophobic product used in most precipitation experiments was cholesteryl acetate. This product was chosen because it has a very low solubility in water (2 x 10~8 g/g), a 8ubstantial solubility in polar solvents that are m18cible with water (e.g. 0.034 g/g in acetone at 23 °C), becau8e it of drugs currently in development. and resembles number It is a solid imp a obtained be of lls °C) which in number crystal [14-16], amorphou8 structures can a or in an microcrystalline 8tate. or We also used a liquid hydrophobic product, hexadecane imp 18 °C), in order to compare emulsification. the sizes of particles prepared through precipitation The solubility and through The

=

=

of

hexadecane For

deuterated

mass

x

10~l~ g lg.

experiments,

polystyrene to gift from Claude

generous AI = 15000

a

is I

water

scattering

neutron

used was

in

g/mol.

make

Picot

at

to

cores

of the

the

Institut

hereafter

noted

It is

intended the

determine

location

the

of the

hydrophobic nanoparticles. Charles Sadron; it had an

surfactant, we This polymer average

molar

PSD.

ethoxylated fatty alcohol with 12 carbons in the named C12E5, and a diblock copolymer groups, made of a styrene block (molar mass 1000) and an oxyethylene block (molar mass 1000), (type VPSE 1010). Goldschmidt hereafter PS-POE. obtained from named This polymer was hydrophobic surface, they from micellar solution When these surfactants adsorbed a on are a form monolayers where the area per molecule is 0.48 nm~ for C12E5 and 1.28 nm~ surfactant for PS-POE; the thickness of these layers are 0.5 nm and 2.5 nm, respectively. The hydrophobic molecules initially dissolved in polar solvents which were acetone (for were tetrahydrofuran and C12E5) and hexadecane cholesteryl with PSPOE), ethanol (for acetate surfactants

The

alkyl

linear

(THF) (for The

chain

PSD

variables

used

and 5

this study oxyethylene

in

were

an

hereafter

PS-POE).

and

which

define

the

composition

of the

mixtures

are

cH,

initial

concentration

of

solvent. of surfactant in the hydrophobic molecules in the solvent, and cs, initial concentration mainly g/g. of is found that the the precipitation Both expressed It in outcome was are determined by the value of the ratio cs/cH. Typical values for the composition were: cH, 10~3 molecules in of hydrophobic concentration to 3 x 10~~ g/g; ratio cs/CH, 5 x 10~~ to 10; final of solvent in water, 10~~ g/g. concentration water, 10~5 to 3 x 10~~ g/g; final

3.

Methods

3. I.

PRECIPITATION.

First, the

follows. Then

the

solution

precipitation of hydrophobic product and the surfactant injected into a large volume of The

hydrophobic was

molecules were

water.

in

water

was

dissolved

in the

At

moment

the

performed as polar solvent. of

injection,

JOURNAL

322

PHYSIQUE

DE

II

N°2

hydrophobic molecules was on the order of 105. This very high aggregation of the hydrophobic molecules, as described in determined by the relative introduction. the The aggregation stopped at a stage that was of hydrophobic molecules, cH, and of surfactant, cs. The final dispersion had concentrations droplets covered with a very small hexadecane colloidal stability in all ca8es, excepted for local supersaturation caused supersaturation

of

the

of

amount

surfactant.

QuAsi ELASTIC

3.2. were

calculated

[17,18].

It

immediate

was

SCATTERING.

LIGHT

from

their

found

that

the

The

sizes

were

sizes of the

through Quasi

measured

motions

in the

particles in the final dispersion Light Scattering (QELS)

Elastic

50 to 500

range

nm,

which

this

particles

were

for

method

is

adequate. 3. 3.

ANGLE

SMALL

NEUTRON

[17,18].

SCATTERING.

The

8tructures

of the

examined

by differences in scattering length between nuclei in the particles and nuclei in the phase. Good contrast between particles and continuous phase was achieved using D2U as the continuous phase. This contrast used to continuous was determine whether the particles were Alternatively, the hydrophobic molecules porous or dense. deuterated, while the surfactant remained hydrogenated. This and contrast water was were surfactant determine and the of the within the particles. Twcused to the location structure obtained Dll at ILL. These dimensional the instrument scattering patterns patterns were on of intensity, I, vs. magnitude, subsequently radially averaged to yield scattering curves were and to the Q, of the scattering vector. Q is related to the wavelength, I, of incident neutrons scattering angle, 9, by: through

SANS

Neutrons

scattered

are

Q The

general

(a)

features

phase

continuous

Q

of

are

as

scattering

curves

I

Sin

=

for

Ill

independent

12)

particles

dispersed

in

a

homogeneous

follows [19]:

intensity, Io, is proportional to particle mass and to the concentration in density of length, Ap, between D2U and the protonated scattering square molecules in the particles. Hence, the average of the particles may be determined; content (b) at low values, the rays scattered by nuclei at opposite ends of the particle interfere destructively. In this regime, the radius of the scattering the curvature curve measures average of gyration, Rg, of the particles. This determined by fitting the measured quantity may be intensity curve to an expansion formula. For globular particles the best expansion is the Guinier at

of the

-+

0

the

difference

formula:

~2j~2

IIQ)"Ioexp

~~

j3)

(c) at higher Q values, the intensity decay reflects interferences shorter within each distances at particle, I-e- their internal For dense globular particles (no internal structure), the structure. intensity follows Porod's law, from whidh the surface area A of the particles may be determined:

1(Q) particles only according

G3

AQ~~

hollow shells, the destructive interference is not as strong, Q~~i for rod-like particles it is Q~~ For bushy or porous dimension di the decay follows the power law Q~df Thus, from the slope fractal decay in a log-log plot, the dimensionality of the particles may be determined. For flat

or

is

to

(4) and

the

decay

particles with of the intensity

SURFACTANT

N°2

AGGREGATION

LIMITED

323

Q~ q.

d '

1,

~ll

,

'~

Q

l/Lm

~~

it

$~f~

~i

""

~

Fig.

1nlage of

TEM

1.

dispersion, aggregation of

the

a

then

it

nlost

particles,

cholesteryl

dried

was

in

and

air the

dispersion

acetate on

collodion

a

in

water.

before

nlenlbrane

recrystallization

of

Uranyl acetate was observation; this

added caused

to

the

few.

a

ELECTRON MicRoscoPY. The particle shapes and sizes also observed through were Cryo-Transmission Electron Microscopy (CRYO-TEM) [20]. Thin liquid films were prepared perforated carbon films supported on 3 mm electron microscope grids. A specimen was on prepared by applying a small (about 3 ~Jl) drop onto the grid, blotting most of it to the desired thickness (under 200 nm) and plunging it into liquid ethane at its melting point. This ultrasolid-like (vitreous) fast cooling caused vitrification of the liquid phase, I.e., specimens became These without change of phase that leads to structural cryo-specimens rearrangement. were cooling-holder (Gatan 626) of the TEM stored and transferred under liquid nitrogen, to the (JEOL 2000FX), where they were equilibrated at -170 C, and examined with an acceleration voltage of 100 kV. 3.4.

°

3.5. how

ComPosiTioN surfactants

All

VARIABLES.

these

aggregation. The iv; second, as a normalized surface area, by the surface area that would be created the particle surfaces. These quantities are the

values

of the

compared 4.

with

composition the

Experimental

laws

variables

OBSERVATION

cholesteryl acetate dispersions The image (Fig. I) shows a

range,

a

broad

and

the

two

if all

or

in

the

cs/cHi

the

kinetic

as

surface

growth

area

in

order

that

an

molecules

used

were

final

dispersions,

laws

for

these

study

to

results

particle average of the dispersion

area

surfactant

the

measured

cs,

surface

of

first,

ways:

total

the

I.e.

performed

were

loss

made

divided to

cover

different

at

variations

from volume

then

are

models.

Results

4.I.

and

cH,

predicted by the

step, TEM.

measurements

aggregation process results expressed in are

the

limit

oF

THE

distribution

THROUGH

DisPERsioNs

of

were

dried

collection sizes.

It

ELECTRON

collodion

on

a

of

globular particles,

can

also

be

seen

MicRoscoPY.

membrane

that

with

aid

sizes

extensive

In

examined in

the

a

first

through 100

aggregation

nm

and

JOURNAL

324

"l'l I

PHYSIQUE

DE

N°2

II

fit,

pot J

'~G

~

[~

Fig.

10~~ g/g and cs/cH

"

fi1nl is

This

1nlage

CRYO-TEM

2.

(cH

particles 1nlportant

held in

are

a

vitrified

features

the

are

a

cholesteryl

dispersion

acetate

fi1nl of

The

vitrified

in

appears which is also

PS-POE

with

water

water

as

unifornl

a

as

surfactant

a

background. cholesteryl acetate grey

The perforation in a carbon fi1nl, seen. film, pushed by the meniscus against the edge of the fi1nl. water good dispersion of the particles in water, their globular shapes and

The the

crystallinity.

of any

absence

across

the

of

0.1).

=

0.2 /Lm

recrystallization

of All subsequent examination shows particles shown Figure 2 in image were confined in the meniscus of the vitrified the edge of one of the perforations in the water film, near and globular amorphous non-aggregated, with carbon film. The particles structure an appear of A CRYO-TEM. confirmed observation through This shapes. in measurement was every performed on an image where the particles were uniformly the particle sizes (diameters) was dispersed in the vitrified water film; the resulting size distribution is shown in Figure 3. some

dispersions

the

molecules

were

dissolved

were

100

were

particles

times

It

shows

the

values

of the

The the

was

acetone,

acetate, A smaller.

found

average initial

radii

initial

vary

that

and

acetate

and

dispersion

the

in

times.

for

in

between

The

In

surfactant

first

a

was

set of

the

precipitated

was

experiments, the hydrophobic copolymer PS-POE. Both

diblock

in

The

water.

10~~ g/g;

PS-POE, 5 x 10~~ to 10~~ g/g; colloidal obtained dispersion was in every immediately through QELS, measured were calculated

volumes cH

of

15 and 30 nm, cu

=

the

remained

sizes

concentration

concentration

drying.

of

result

a

SizEs. the

solution

the

these

particle

as

CRYO-TEM.

PARTICLE

oF

cholesteryl

cholesteryl

were:

through

STUDY

SYSTEMATIC

4.2.

occurred

have

made

same

from

cholesteryl and

10~~ g/g.

the

acetate

volumes

They

are

than

more

over

these

radii, in

at

initial

the

concentrations

final

case.

and

then

"

of the

sizes

again

at

later

Figure

month.

one

cs/cH

concentrations

The

I, for

4

different

acetone.

between

14

x

always larger

10~ and than

the

l10

x

10~ nm3,

volumes

that

SURFACTANT

N°2

AGGREGATION

LIMITED

325

o.06

~ 0.04 w

c

0.02

~~~ 0

20

lo

30 s

Fig.

3.

Distribution

10~~ g/g vitrified

of

cs/cH

and

would

The

water.

automatically

by

a

obtained

be

particle

surfaces:

surfactant efficient

would use

molecules

of this

of

"

of particle sizes (diameters) ns nleasured CRYO-TEM on a 1). The 1nlage contained 1000 particles, unifornlly distributed particles were selected individually and their diameters s were

conlputer

as

a

cs/CH much

then

=

fi1nl

nleasured

the

cs/cH

surfactant, increased.

system;

it is:

The

=

they

rise

law for

this

and

performed experiments were surfactant, both dissolved in "

the

surfactant had been used to limit aggregation by covering the I, the particle volume corresponding to an efficient use of the be 4000 nm3. At high dilution, the particle approach the limit of volumes if all

at

is

0.4), the larger than

1nlage (cH across

progranl.

away rise

from it when

measured

the

volumes

volumes

with

were

C~

high

At

larger for

an

to

the

concentration

be the

same

at

all

cH of hydrophobic compositions cs/cH

(5)

CH.

hexadecane

ethanol.

calculated

found

was

v~v Other

40

(nm)

than

as

hydrophobic

a

surfactant those

efficient

of use

molecule,

concentration

C12E5

and

(cs/cH

"

I

and

cholesteryl acetate particles, and of surfactant; again, they follow

(cs/cH 0.04), However, at low surfactant concentration 10~~ g/g) of efficient successive observed: regimes large dilutions regime two (cH < at were a of the (denoted by arrows in Fig. 4), and at higher concentrations, again the surfactant use (rising straight line in the log-log representation of regime of wasteful use of the surfactant Fig. 4). the

described

law

in

equation (5).

"

the particle volumes generally larger than those predicted for an efficient are surfactant, and they rise with the of hydrophobic molecules in the concentration initial solution, according to equation (5). We have also initial measured systematically the particle volumes obtained with the same dilution cH but different values of the ratio cs/cH (Fig. 5). The data for and C12E5 hexadecane surfactant show a fast decrease of the that the volumes at cs/cH < 0.05, indicating average additional is used efficiently, followed by a slow decrease at cs/cH > 0.05, indicating that the surfactant is mostly wasted. In

use

summary,

of

the

JOURNAL

326

PHYSIQUE

DE

N°2

II

io~

tl

~

io~

'

D

n

57

E 10

> I

o

io~

io~

cH Fig. ispersions

surfactant; overall

dilution

squares:

S-POE,

an

slope

order it is

to

these

put

useful

to

cs/cH

notions

calculate

cH,

by

1.

Filled

of

squares:

in

arrow and

log-log scales,

of

efficient

the

amount

or

been

as

wasteful

of

predicted

a

by

fit

linear

nlodel.

kinetic

has

that

volunles

the

a

are

surfactant

of the

surface

kept

cs/cH

at

of

alues

lines

filled

use

particle

C12E5

the

indicates

the

been cholesteryl

changed. Hollow

and

hexadecane

The

has

nlolecules

has

surfactants,

the

product and

hydrophobic

to

surfactant

0.04.

"

efficient use is 1 data

the

of

of

"

at

12E5

for

The

In

cs/cH

at

and

calculated

ontaining

measured

olution,

given

a

cs/cH

the

of

and exadecane

solutions

ratio

the

acetate

basis,

of

precipitation

ade

quantitative a protected from

on

been

aggregation by

calculated the adsorption of surfactant. The surface of the particles may be area efficiently, surface radii, assuming a globular shape. If the surfactant used this was surface surfactant Thus would equal the of monolayer the molecules. all containing area area a The values of quantities is an indicator of the effectiveness of the surfactant. the ratio of both presented in Figure 6. this ratio are from

their

factant

0.05, choose

droplets

hexadecane

For

molecule

followed a

by

surface

The

wasteful

a

area

by C12E5,

covered

[21,22].

calculated

regime.

of1.28

nm~

For

ratios

show

cholesteryl surfactant

per

choose

we

a

a

surface

regime of

acetate

molecule

particles [23].

of 0.47

area

efficient

The

use

covered calculated

nm~

per

sur-

cs/cH

to

up

by PS-POE, ratios

=

we

show

a

surface is kept than what is covered by the surfactant. In regime where, surprisingly, more fact, cholesteryl acetate particles can be obtained in the absence of any added surfactant. This cholesteryl acetate the particles to repel each in is because contains ionic impurities that cause medium where dissociation ionic possible. Therefore cholesteryl dispersions is the acetate any the always regime of surfactant. in are excess In

summary,

sufficient

amounts

it is of

not

possible

surfactant:

controlling the aggregation. of the wasted surfactant; this

to in

This

in

is

one

particles of arbitrarily small fraction of the cases, only a snlall result be improved unless one cannot of the goals of the next section. obtain

such

volumes surfactant

understands

by adding is

efficient

the

fate

SURFACTANT

N°2

AGGREGATION

LIMITED

327

~~9 tl

n

a "

~

io

~

©1f

n

~S ~

~ ~ "

l~~

C

~

>

o

~

io~

o

O

io~ io

~

io

io

~

C~/C~ Fig.

of surfactant concentration particle volumes iv. Each set of data the average cs on dispersions made from the precipitation of solutions containing a given hydrophobic surfactant; the overall dilution of the solution, measured by cH, has been kept a constant, but the ratio cs/cH of surfactant hydrophobic nlolecules has been changed. Hollow dianlonds: to Cholesteryl acetate and PS-POE, at cH C12E5 0.01 g/g. Filled hexadecane and at cH squares: 0.04 g/g. Hollow hexadecane and C12E5 at cH 0.1 g/g. The lines are a guide to the eye. squares: 5.

Effect

corresponds product and

to

=

=

=

STRUCTURES

4.3.

poisoned The

.

final

of the

sition

PARTICLES

OF

aggregation

can

aggregation

voids .

less

preserved

structures

uous

collapse The

amount

than

the

of

reordering

of

Quantitative contrasts

surfactants A first without is are

The

between shown

initial

Where

is

subunits, answers

observe

to

the

in

become

to

either

surface

should the

have

or

these the

the

to

number

a

Diffusion is

there

of

structures

The and

to

process

the

of

compo-

reasons.

Aggregation of Clusters [11-13]. Are such tenreordering process by which the Limited

structures

porous

or

or

SURFACTANT.

THE

a

dense? from

protected by

the

surfactant: remain

the particle radii is usually much adsorption of surfactant (see the been trapped inside the particles by reasons? phase for kinetic

calculated

that is been

wasted did it

fractal

particles,

final

structures

that

with

aggregates

hydrophobic

amount

previous Sect.). the

the

and

interesting for

particles through

colloidal

of

(DLCA) usually produces

OF

according

characterized

This is

aggregates.

LOCATION

AND

also be

in

the

has

it

aqueous

obtained through SANS, using the appropriate questions were scattering from hydrophobic molecules and surfactants, or from

alone.

particles made of cholesteryl acetate, with or phase made of D20. In this case the contrast nanoparticles. These scattering and all the molecules that form the water curves followed by a steep decay. downward in Figure 7: they show an initial curvature measured is related to the particle sizes according to equation (3) the sizes curvature

set

of

PS-POE

experiments as

a

was

surfactant,

performed in

an

with

aqueous

JOURNAL

328

PHYSIQUE

DE

N°2

II

o

~ p

g 10° fi

~w ~

Q

_i

.

~

1

O

o Z

~

tl

io°

cs/C~ Fig.

Surface

6.

nlolecules at

cH

0.01 g lg.

=

C12E5

and

yield

a

values

at

cH

indicate

that

through

g/g.

0.1

"

area

must

A

second

set

of

surfactant, hydrogenated high ratios of

in

low

at

ratios

vanished

has

a

in

curves

was

made

there

are

size

on

the

order

of I

flat

objects (Q~~ These

tation

is

are

molecules as

follows.

particles In

is wasted; higher aggregation process.

the

with

agreement

those

particles: according to Therefore, aggregation give dense

to

of PSD, the

case

of

good

made

this

PS-POE,

hexadecane squares: would the surfactant

structures.

using is

contrast

PS-POE between

as

the

and

"

"

"

other

IQ

objects

and the

observed

only. The

that

A fit of the

law)

Power

features

surfactant

l1nlit

in

surfactant

and

surfactant

sortie

the

to

shoulder

that

are

Hollow use

if all

acetate

obtained the deuterated phase. The scattering at curves aqueous hydrophobic molecules (cs /cH 0.5 or Ii show a Q~~ decay, and Q of samples made 0.3 nm~~ (Fig. 8). The scattering near curves 0.05 to 0.25) show the but the shoulder initial law decay, same power

surfactant

surfactant

shows

with

D20.

of

features transcription of these into real decay comes from flat such structures

A initial

Q

of

range related to

that

surfactant

of

this

is

performed

phase

aqueous

(cs/cH (Fig. 8).

adsorption

the

kept

be

Cholesteryl

0.04 g lg. Efficient

indicate

values

would

that

area

dianlonds:

and C12E5 at cH = guide to the eye.

Lower

besides

experiments an

characteristic

a

by

a

surface

the

Hollow

shape of the law indicates that the particles are dense. reordering process that collapses the voids

Power

followed

are

steep decay

The

be

a

lines

unity.

scattering

QELS. equation (4), the Q~~ from

events

to

forces

other

fit of the

a

obtained

The

equal

ratio

surfaces.

hexadecane

squares:

by

divided

the

on

Filled

surface

particles, particle

of the

area

adsorbed

were

In

hollow

also

shells

is

follows.

as

hollow

to

The

The

shells.

contribute

the

Q~~

shoulder

scattering;

of the are

where location

surfactant

the of the

coherent

these

of the

Q 0.3 nm~l objects have a scattering from =

of 7.6

nm.

scattering is caused by the molecules, the interprethe nanoparticles. that cover

surfactant

monolayers

law

power at

is obtained by adding the curve scattering from small spheres with a diameter

scattering

dispersions

in terms

space as

N°2

SURFACTANT

AGGREGATION

LIMITED

329

~ io~

~

*i

°

O

§p

(

O

O

~3

~

~ d~ 10~ j ~

slope=.4

~~i

io° Q

Fig. 7. nanoparticles

Angle

Snlall

D20.

in

cholesteryl io-~ g/g).

dianlonds:

lcu

=

The

change

have

an

thickness calculated

cs/cH

"

Q~~ (Fig. 7) to Q~~ (Fig. 8) particles. The fit to the scattering

to

be

8.2

thickness,

uniform

nm

These

of 24 than

for rather

at Q (Fig. 8);

show a

and

that

as

=

indicates

of at

the

a

that

this

in

standard

a

thickness

dispersion

the

results

but

nm

the

acetate

PS-POE

with

dispersions cholesteryl containing acetate particles, without surfactant. Hollow surfactant, conlposition ratio cs/cH a

of

curves

cholesteryl

pure

particles

acetate

thickness average is much larger 0.10.

diamonds:

from

of the

interior

Scattering

Neutron

Filled

(nm~)

there

range

deviation

cs/cH

layer formed of thick

"

0.25

and

surfactant

and

is

thin

thickness

PS-POE 3.I does

no

obtained

of the

monolayer of

adsorbed

is

nm

not

surfactant with

of 5

the

in

shells

that

nm.

This

molecules, which is for the dispersion at form a monolayer of

patches.

nm~~ is

observed in the dispersions that contain a large interpretation is that this scattering originates from surfactant micelles that Indeed, the average molar mass of these micelles, meain water. are sured through gel permeation chromatography, is 300000 g/mol. From this value, the aggregation number is calculated to be 150 PS-POE molecules, and the radius of the PS core is calculated This is in good agreement with the size of the small spheres used in to be 3.9 nm. (the water swollen POE shell would the fit of the scattering contribute only at smaller curve Q values). The

excess

In

small

excess

of

scattering

surfactant

sunlmary,

micelles

the

that

surfactant are

"

an

is

dispersed

0.3

obvious

located in

water.

in

irregular

monolayers covering

the

particles

and

in

JOURNAL

330

PHYSIQUE

DE

N°2

II

io~

io~

~~3 W

~ C

~ ~

10

W ~

fi d~ ~~o

I

i ~o

Q(nm

Fig. 8. surfactant,

Scattering

~)

PS-POE dispersions containing PSD nanoparticles covered with as a Taking final concentrations equal to 5 x 10~~ g/g for both PS-POE ii; THF, 0.1 g/g. The full lines correspond to the calculated and PSD (cs/cH for scattering curves (outer dispersion containing large shells diameters beyond, standard deviation thickness 24 a on nnl, thickness, 5 nm) and snlall micelles (dianleter 7.6 nm). Filled circles: Concentrations equal to 5 x 10~~ g/g and 10~~ g/g for PS-POE and PSD, respectively (cs/cH 0.05) THF, 0.1 g/g. The dashed line corresponds to the calculated diameters for a dispersion containing large shells (outer scattering curve nnl). beyond 175, thickness 24 nnl, standard deviation thickness, 5 on in

D20.

of

curves

Hollow

circles:

=

=

Numerical

5.

Simulations

of

the

Poisoning

Growth

MODEL. coalescence have been inNumerical models for droplets growing by accounted for in mechanism previously [24], however, the poisoning have not been growth poisoning these models. introduce lattice model to simulates In the following, we a a mechanism exhibed by our experimental system. as THE

5.I.

troduced

System. We have considered lattice limited in a cubic box of edge length sites considered. For the particle species H (hydrophobic molecule) and S (surfactant) are starting configuration, time t 0, a number NH of particles H and Ns of particles S, each one of a cubic site size of unit length, are randomly and uniformly in the box, avoiding distributed Concentration, or overlaps. Each particle is then a unit cell containing six faces of unit area. related to Ns and NH by: volume fraction, of species are 5.I.I.

L.

The

Two

=

cH

~) =

L

(6a)

SURFACTANT

N°2

cs

In

algorithm

our

introduce

we

(S specie) adsorption by cluster iH (formed by possible.

Algorithm.

The

5.1.2. iteration

include

to

r

one

more

ore

a time delay particles of the

in

H

which

surfactant

species)

becomes

iterative procedure, in which at a given consists in an (under the assumption that random is is picked at or a up coefficient proportional to n°, a is the kinetic is exponent equal to -) for a compact system of n particles), according to probability distributions, for H

diffusion

its

(6b)

algorithm

3-dimensional

The

iH

cluster

a

331

)~.

"

parameter

time

a

AGGREGATION

LIMITED

surfactant

species:

~j

£~~ nj(

~~"

for S species

and

~~~~

+

ns

~n ~Jn

~~~

~~~~

~ ~~

Then, a displacement of a unit step is tried by possibilities (+1, 0, 0), (0, +1, 0), (0, 0, +1), taking (PBC). If the moving object does not try to occupy a site of periodic boundary care occupied by other object, the displacement is performed. In the opposite case, no displacement surfactant is performed and: (I) if the collision is between a free and a surface site of specie H (belonging to a cluster iH), a bond is established them provided the time t is larger between than r, and the incremented by one. (ii) if the collision takes places mass n~~ of the cluster iH is between different clusters, and two (or more) sites (H or S species) belonging to two (or more) each one with at least one free surface site, we proceeds to the "coalescence" of the clusters by configuration with minimum surface. arranging the new mass (= n~>~ + n~ii~ +.. in a compact Then, we arrange on the new surface the total number of surfactants isl + isii +.. adsorbed previously by the coalescing clusters. If this number exceeds the surface, the exceeding new surfactants distributed randomly in the empty sites of the box. (iii) no change is again are where

at

adsorbed After another

between

is

in

According

to

free

a

surfactant

a

situations,

above

the

each

below 5.1.3.

t

=

are

goes

has

been

by choosing randomly

on

is stopped at a given surface by the joisoning effect, or if there is no (7) the time t is calculated adding:

described experiments regime, i.e. when cH +

have

been

calculated

Two-Dimensional

configurations

of

if its

time more

tend in than

which

all

cluster

one

~~ £~~

+

~~~ ns

iteration.

As the diluted

algorithm

the

simulation

The

~~

at

(even

surfactant

another

and

given cluster).

surfactant.

or

the six

among conditions

saturated

are

box.

the

collision

surfactants.

direction

a

by the surface of considering one of cluster

clusters

of free

number

random

if the

made

in

the

is

ns

choosing

at

randomly

coalescence,

and

we

been

the

All

addressed results

in

the

reported

=

We

our

white

or

observe

have

percolation threshold). cs < cp (cp the end of the growth process (t tend)-

with

in

calculations

numerical

our

is

Illustration.

obtained

0, H (depicted in grey, distributed

above,

the that

box. some

show in

algorithm when

a

For

0

of

three

cluster

is


c

0.5

~ ~ 0

20

60

40

V 2.0

(b) 1.5

~

I-o

c

o.5

o-O 0

50

40

30

20

lo

V

Fig.

14.

and

Cs

All

these

Size =

+

H

weight

results

curves

adsorbed H

distribution

on or

is

H

its +

attached

front

surface.

an

At

S, leading to

each

for CH

curves

4CH (hollow circles).

to

In

=

(b) Cs

average

2000

over

each

time

a

different

event,

=

0.001

4CH,

step,

and L r

=

=

50.

In

la)

(circles),

r

r =

=

0, Cs

2500

=

2CH (black circles)

(squares)

and

r

=

8000.

s1nlulations.

investigates

all

by

coalescence

or

~

~~~~

one

state

0

possible binary events: A probability adsorption.

the

namely:

Kin,jn

~

l~~

~

~ ~

(14) ~~

338

for

of two

coalescence available

is

DE

PHYSIQUE

radii R~~

and

Rj~, provided

II

N°2

least

at

adsorption

free

one

site

cluster, and:

each

on

of

clusters

JOURNAL

kj~

s

l)

+

Rm

"

'

(R~~

~~"

II

I)

+

(15)

EIH

(assumed here to be of radius I). The terms I/R~~ are due adsorption of a surfactant of the diffusion, clusters during their and the sections to R~~ + Rj~ are just the cross Brownian path. Then one event is chosen according to these probability weights, and the time increased by the is amount: for

Stokes

dt This

leads

~

equations:

kinetic

the

to

j16)

~ ~~

=

dnj,s~ it) ~~~~~

dt

~

~

I~J',J-J'"J',s~>11 j-j',s~_~i

j',s~i,s~_~i

~

~?l~~

~JS"J,sj

=

equations

These

difficult

too

seem

RESULTS.

6.2.

There

ment.

The

is

that

advantage

geometry,

no

"~'~? "~

~'~

model

there

is

of the

is

then

more

no

are

kinetic

this

validity for ordinary three-dimensional space is not the results. In Figure 15, we have plotted the on poisoning ratio cs/cH, for T 0, and NH 256. =

for

which well

are

recovered.

for the

-IA

surfactants

the

kj,slip,sj

-illp

lip

other

And

we

one.

Note

"

used

are

obtain

that

ji~~~

Ej+j ).

+ SJ

efficiently, the

An

alternative

quite similar geometry.

the

to

study

to

way

model

described

in

approach is that it is simpler to impleMoreover, since one can write down be tempted in simple Nevertheless, cases. described in the preceding section, and its ascertained. This validity has to be checked

Smoluchowski

simulations

equations, analytical investigation can approach is a mean-field theory of the model

the

l~J

~J

analytically.

handled

the

the

so

~

be

to

simulations:

preceding section, except

the

kj,sllj,sj

+

J"SJ

~'~J

Monte-Carlo

is

~

"p +

~'~~

them

Kj',jllj',s~> llj,sj

j',s~i

these

two

cluster

average

We

that

see

exponents

versw

the

regimes (small- cs/cH,

two

of surfactants) is excess cs/cH values regime, and slightly larger than those obtained

for

-2.7

(Eq. (9))

volume

the

large- cs/cH,

and

exponents

fast.

the

are

that

small-

simulations, possibly as a of the absence of geometry in the consequence Smoluchowski approach. The behaviour of all other computed data are quite similar and are not reproduced here. This shows clearly that the Smoluchowski approach can be used in this process, and that the two regimes meet at about the same point. For example, the ratio of the surface lost by poisoning surface lost by coalescence obtained from equation (13), decreases with cH (whqn to the sp /sc, aggregation becomes faster) and increases with cs /cH (when poisoning becomes efficient). more

in

the

This

last

trend

linear

law.

most

of the

of

Carlo

Monte

surface

surface

in Figure 16: the rise explained in the following surface is lost by coalescence,

illustrated

is

This

be

can

initial sites

sites

in

in

the

an

H

molecule

geometrical

(here, algorithm

we

for

of

sp/sc

with

cs/cH

At

very

low

way..

choose cluster

therefore a

=

sizes

sc

6.5 up

m

aNH

match

to to

3).

remarkably close to a concentrations,

is

surfactant where

the All

a

the

is

values surfactants

of

number

number are

used

of to

SURFACTANT

N°2

AGGREGATION

LIMITED

339

~~3

slope=-2.7

$

~>

lope=-lA

~~o

10

c~

Fig.

log-log plot

15.

Snloluchowski to

the

different

power

the

of

quantity

calculations.

nunlerical

donlains

law

iv We

that

-1

best

fit

the

versus

NH

considered

=

data.

our

poisoning 256

This

cs/cH

ratio

and

r

=

result

curve

The

0.

front

full an

obtained lines

from

the

correspond

average

over

sp/sc

versus

1000

simulations.

io'

slope=1

~o

~

~~~~Jf

~~~~

~

~

o ° o

~

,

io~ c

Plot of the Fig. 16. poisoning ratio cs/cH the slope1.

surface obtained

lost

from

by poisoning the

to

Smoluchowski

s/c

the

~

surface nunlerical

lost

by

coalescence

calculations.

The

full

line

the

indicates

JOURNAL

340

aggregation,

stop

therefore

sp

Ns.

t

~~ ~

sc

high

very

ing

more

to

while

cH,

concentration,

surfactant

than 2

In this

molecules.

that

of

forming

an

limit

H-S

II

N°2

Thus, ~~

At

PHYSIQUE

DE

~~

(~8)

~f

cH

hydrophobic molecules probability of forming is proportional to cs.

the the bond

again proportional to cs /cH. The complete that aggregation is the dominant process

aNH

in

/sc,

variation

of sp

most

practical

form

never

H-H

an

clusters

bond

Therefore

shown

in

situations

the

is

demonstrates

cs/cH

to

sp/sc

ratio

Figure 16,

(up

contain-

proportional

is

"

4).

Discussion

7.

We

this

start

compared for

the

discussion

with

the

of the

role

with

a

from

results surfactant

in

brief

of experimental summary Finally, simulations.

numerical

precipitation

results. we

try

to

These

results

put

together

are a

then model

processes.

solution The precipitation of hydrophobic insoluble yields particles that are completely amorare Stable obtained when particles phous, dense and globular. dispersions the are are aqueous function surfactants. surfactants In these kept apart by adsorbed by impurities that or as by two parameters: stable dispersions, the particle sizes are determined the poisoning ratio cs/cH, I-e- the ratio of surfactant dilution molecules ratio to hydrophobic molecules, and the hydrophobic molecules the original solution. the of solvent in ratio I.e. to cH, The of particle sizes according to these variations parameters show that there are 2 regimes, surfactants the differ in the way used. At very low of surfactants, which concentrations are indicating that surface of the particles matches the of a monolayer of the surfactant, area area surfactant used to keep particles apart and limit their aggregation. At higher all molecules are of surfactant, the surface of the particles is much less than the of a concentrations area area monolayer of the surfactant (Fig. 6). Neutron scattering experiments on dispersions made in this regime reveal that part of the surfactant is adsorbed the particles, and part of it forms micelles in water. on excess Thus, attempts to reduce the particle sizes by adding increasing of surfactant amounts become inefficient at some point where added surfactant remains in most of the water. 7.I.

SUMMARY

molecules

7.2.

in

EXPERIMENTAL

oF

conditions

COMPARISON

oF

where

RESULTS.

they

EXPERIMENTS

wiTH

NUMERICAL

SIMULATION.

The

kinetic

model

aggregation of hydrophobic molecules and The two-dimensional poisoning by surfactant. illustration presented in Figure 9 shows qualexperimental particles covered with adsorbed surfactant itative with the results: agreement with single molecules, also covered by surfactant, Quancoexist and with surfactant. excess (Figs. 10 and II) show a regime of efficient results from the 3-dimensional simulation titative of the surfactant followed by a regime where most surfactant Comparison with is wasted. use experiments (Figs. 5 and 6) shows that, in the simulation, the regime of efficient use extends (cs/cH of 0.04); as a result, the parsurfactant concentrations 4 instead to much higher ticle volumes have been reduced to extremely low values (taking I nm as the diameter of a iv t 10 instead of 107 at the crossover). This discrepancy originates from the time monomer, delay between aggregation and surfactant adsorption, as explained below. assumed that surfactant In the efficient regime of this simulation, it was adsorption starts at molecules. there is a time the In the experiments, time as the aggregation of hydrophobic same delay between the two caused by the finite rate at which water diffuses into the solvent that A similar delay was introduced surfactants. initially contains hydrophobic molecules and in demonstrates

the

effects

of the

competition

between

"

SURFACTANT

N°2

AGGREGATION

LIMITED

341

concentrations, this simulation yields the same results aggregation leaves surface enough bind all to as one area (Fig. 10). However, at high surfactant the surfactant concentration, the particle sizes remain those obtained with no time delay, because the initial above aggregation is not prevented by Comparison surfactant. the with the shows that the actual experiments time delay is excess that used in the still much larger than simulations. A simulation conditions would in these have to handle particles of extremely large sizes. Alternatively, the hydrophobic molecule used could be rescaled to be an aggregate simulation in the containing 10~ molecules. In the wasteful regime, the experiments show that the volume per particle increases average linearly with concentration (Fig. 4). The numerical simulation reproduces this trend and provides a simple explanation for it. Indeed, the particle volume rises linearly with concentration only in the limit of long delays (Fig. 12). Therefore the rise in the final volume reflects the unimpeded aggregation during the time delay before the beginning of surfactant adsorption. previous

the

The

At

simulation.

another

linear

with

rise

surfactant

low

because

the

initial

concentration

indeed

is

expected

that

aggregation [I Ii. the results of experiments In conclusion, are allowed to proceed unimpeded for a time t,

from

equation for

Smoluchowski

the

kinetic

is

At

there is, concentration, molecules; thus aggregation

surfactant

low

well-reproduced by adsorption then point,

that

at

simulation

and

surfactant

enough

than

more

aggregation

if

of the

surface

starts.

adsorb

to

until it is limited because all remaining continues poisoned. Thus the surfactant molecules used efficiently, but the final partiare cle volumes large. At high surfactant adsorption of surfactant stops the concentrations, are immediately for Consequently, the aggregation after the time delay adsorption. aggregation numbers determined by the length of delay this time and by the rate of aggregation, I.e. are by the of hydrophobic concentration molecules. Adding more surfactant does not change these particle volumes; consequently the excess surfactant wasted. is

all

surfactant

8urfaces

7.3.

are

MODEL

and

FOR

PRECIPITATION

THE

hydrophobic

tion of

molecules

in

As in any

PRocEss.

water

can

be

divided

kinetic

stages:

three

into

precipitapropagation

the

process,

initiation,

termination.

with the solution containing the hydrophobic they hydrophobic molecules because aggregate, surfactants do bind the insoluble this mixed At however, the in solvent. this not to stage, are growing aggregates, because they are not sufficiently hydrophobic. According to Propagation proceeds through random collisions of the growing aggregates. with that is Smoluchowski volume linearly time, the equation, the average at a rate grows molecules. proportional to the of hydrophobic concentration Initiation

caused

is

and

molecules

Termination

ticles. In

The

The

practical

situations,

solution

final cause

will

be

high the

pressure most

the

may

a

set

the

dilute that

homogenizer could attractive

option.

to

it the

achieves

be

that

for

this

the

the

endeavour.

particles at will just be the

necessary between

to

be

time

delay

instantaneous to

surfactants

sets

therefore

and

will

adapted

the

dispersions

obtain

cover

aggregation, that

high

termination

surfactant

excess

Reducing

device

so

limitations

needed

of

so

is

content

severe

since rate

be

water

water

the

aggregation and usually desired to

amount

aggregation.

requires

surfactant a

reduce

dispersion further

it is

above

beyond the be efficient,

not

the

delay between

described

concentration

stage will

mixing of Immediately,

the

when

occurs

time

results

by

surfactant.

achieve

that

fine

as

are

Increasing the

aggregation

par-

particles. possible. surfactant

aggregation

Diluting sizes;

the

initial

however,

it,

reconcentrate

water

as

the

end of the

wasted.

the

on

size of the

particle

mixing of this

adsorb

final

which

may

adsorption and polar solvent.

instantaneous

and

mixing,

it

the

of If

would

JOURNAL

342

DE

PHYSIQUE

N°2

II

Acknowledgments This

work

the

BIOAVENIR

us

used

(A.H.)

the

beams

neutron

financed

program like to acknowledge

would

of ILL

by

in

Grenoble,

France.

RHONE-POULENC

support

from

It

with

CONICIT

was

the

performed

MRE

(Venezuela)

and

as

MICE.

and

of

part

One of

(France).

CNRS

References

[Ii [2] [3] [4]

[5]

F., E., Fessi H., Puisieux

G.,

Barratt

Vauthier

Al-Angary A. 1 (1992) 277.

and

C.

Halbert

P., Couvreur P., Devissaguet J-P-, Dubernet C., Fattal S., Polym. Biomater. (1994) 749. and G. W., Drug Targeting Delivery (Microencapsulation of drugs)

Couarraze

Benita

Pharm. Biopharm. 39 (1993) Allemann E., Gurny R. and Doelker E., Eur. J. Gregoriadis G., Liposomes as drug carriers (Wiley, 1988). Julienne M.C., Alonso M.J., Gomez Amoza J.L. and Benoit J.P., Drug Dev. Ind. 18

(1992)

173.

Pharmacy

1063.

Koosha F. and Muller R.H., Archiv. Pharm., suppl. 321 (1988) 680. A., Cabane B., Pharm. Res. 12 (1995) 39. Kaplun Talmon Y. and Sj6str6m B., [7] Controlled Veillard M., J. G., Chabot and Spenlehauer Vert M., Benoit J.P., F. [8] [6]

(1988)

Pharmace~t.

J.

[10]

StoInik 30

[I Ii

N.,

Alkhouri-Fallouh

[9]

S.,

(1994)

Jullien

Rel.

7

217.

R.

Davies

Roblot-Treupel

(1986)

28

M.C.,

L., Fessi H., Devissaguet

J-P-

and

Puisieux

F., Int.

125.

Illum

L., Davis S.S.,

Boustta

M.

and

Vert M., J.

Controlled

Ret.

57.

and

Botet

R., Aggregation and

Fractal

Aggregates (World Scientific, Singapore,

1987). Meakin P., Phase Critical Transitions and Phenomena, vol. 12 (Academic Press, New York, 1988) p. 335. Growth Phenomena (World Scientific, Singapore, 1989). [13] Vicsek T., Fractal [14] Wendorf J.H. and Price F.P., J. Phys. Chem. 75 (1971)18. [15] Wendorf J.H. and Price F.P., Mol. Cryst. Liq. Cryst. 25 (1974) 71. [16] Kunihisa K.S. and Gotoh M., Mol. Cryst. Liq. Cryst. 42 (1977) 97. [17] Cotton J.P., Neutrons, X-rays and Ligth Scattering, P. Linder and Th. Zemb, Eds. (ElsePublishers, 1991) p. 3. vier Science C. and G., Analyses chimiques et caractAiisations, Techniques de Weill [18] Strazielle l'IngAnieur PI, pp. 1065-1-1065-22. Surfactants Sciences Surfactants methods of investigations, solutions: [19] Cabane B., new Series, R. Zana, Ed., vol. 22 (Marcel Dekker Inc., New York, 1987) p. 57. Microz. Techniq~e10 [20] Bellare 3.R., Davis H. T., Scriven L. E. and Talmon Y., J. Electron

[12]

(1988) [21]

87.

Lichterfeld

F., Schmeling T. and Strey R., J. Phys.

Chem.

90

(1986)

[22] Olsson U., Wlrz U. and Strey R., J. Phys. Chem 97 (1993) 4535. O., private Sonneville communication. [23] Meakin P., Rep. Prog. Phys. 55 (1992) 157. [24] for a review see: [25] Family F. and Meakin P., Phys. Rev. A 40 (1989) 3836. [26] van Dongen P.G.J. and Ernst M.H., J. Stat. Phys. 50 (1988) 295.

5762.