Hydrophobically Modified Polymers 2002tesis

Hydrophobically Modified Polymers
Rheology and Molecular Associations
Leif Karlson
Avhandling fr Filosofie Doktorsexamen

Matematisk-Naturvetenskapliga Fakulteten
Avhandlingen kommer att frsvaras vid en offentlig disputation fredagen den 4 oktober 2002 kl. 13.15 i
hrsal C, Kemicentrum, Lund
Leif Karlson 2002

Thesis
Physical Chemistry 1
Center for Chemistry & Chemical Engineering
Lund University
P.O. Box 124
SE-221 00 Lund
Sweden
Contents
List of Papers
Chapter 1 Introduction
1.1 Hydrophobically modified polymers in paint

1.2 References Chapter 1
Chapter 2 Hydrophobically modified polymers

2.1 Structure and synthesis of hydrophobically modified polymers
5
11
13
14
2.1.1
HEUR thickeners
14
2.1.2
HM-EHEC
15
2.1.3
Comb HEUR thickeners
18
2.2 Hydrophobically modified polymers in aqueous solution
19
2.2.1
HM-PEG in aqueous solution
23
2.2.2
HM-EHEC in aqueous solution
26
2.2.3
Interaction between HM-Polymers and surfactants
28
2.2.4
Clouding
30
2.2.4.1
Cloud point of HM-PEG
32
2.2.4.2
Cloud point of HM-EHEC
33
Chapter 3 Inhibition of hydrophobic associations as a tool to study

cross-linking mechanisms
34
37
3.1 Inhibition of hydrophobic interactions by changing solvent quality
38
3.2 Inhibition of hydrophobic interactions by addition of surfactant
40
3.3 Inhibition of hydrophobic interactions by addition of cyclodextrin
41
3.3.1
Structure and properties of cyclodextrin
42
3.3.2
Formation of inclusion complex between lipophilic guest

molecules and cyclodextrin
43
Cyclodextrin and HM-Polymers
45
3.3.3.1
Cyclodextrin and HM-EHEC
46
3.3.3.2
Cyclodextrin and HM-PEG
49
3.3.3
55
Main conclusions
57
Popular summary (in Swedish)
58
Acknowledgements
61
List of commercially available hydrophobically modified polymers
62
List of Papers
I
Rheology of an aqueous solution of an end-capped poly(ethylene glycol) polymer

at high concentration
Karlson, L.; Nilsson, S.; Thuresson, K. Colloid Polym. Sci. 1999, 277, 798-804.
II
Clouding of a cationic hydrophobically associating comb polymer

Thuresson, K.; Karlson, L.; Lindman, B. Colloid and Surfaces A: Physiochem. Engin.
Aspects 2001, 201, 9-15
III
Phase behavior and rheology in water and in model paint formulations thickened
with HM-EHEC: influence of the chemical structure and the distribution of
hydrophobic tails
Karlson, L.; Joabsson, F.; Thuresson, K. Carbohydrate Polymers 2000, 41, 25-35.
IV
A rheological investigation of the complex formation between hydrophobically

modified ethyl (hydroxy ethyl) cellulose and cyclodextrin
Karlson, L.; Thuresson, K.; Lindman, B. Carbohydrate Polymers 2002, 50, 219-226.
Cyclodextrins in HM-PEG Solutions. Inhibition of Rheologically Active PolymerPolymer Associations

Karlson, L.; Thuresson, K.; Lindman, B. Submitted
VI
Complex formed in the system hydrophobically modified polyethylene glycol /

methylated -cyclodextrin / water. An NMR diffusometry study
Karlson, L; Malmborg, C.; Thuresson, K.; Sderman O. Submitted
Chapter 1
Introduction
Aqueous solutions thickened with polymers are common in our daily

life. Shampoo, for instance, is a water-based solution of surfactants
that should have high viscosity, since a low viscosity would mean that
it would flow between the fingers when you poured it out of the bottle.
In cooking there are many examples of how water-soluble polymers
are used for thickening. Starch from potatoes or corn can be used for
thickening of a sauce and gelatin gives the jelly consistency to many
desserts. Polymers are also used as thickener in many low fat
products.
Some
pharmaceutical
formulations
are
water-based
systems that gain their flowing properties from polymers.

Water-borne paint is another example of an aqueous system that has
to be thickened to behave in the way we want. In fact the use of
Hydrophobically Modified Polymers (HM-P) in paint is the basis for
this thesis and has therefore got a separate section (section 1.1
below).
The aim of this thesis is to provide useful knowledge for the
development of new hydrophobically modified polymers with
improved properties primarily for the paint application. In order to fulfill
this goal the first part of the work is dealing with how hydrophobic
modification influences the properties of the polymers in solution
(Paper I, II, and III). In the second part of the thesis the thickening
mechanisms of HM-polymers in aqueous systems have been
investigated (Paper IV, V and VI).
The discussion in this thesis is based upon two types of HMpolymers, Hydrophobically Modified Ethyl Hydroxyethyl Cellulose
(HM-EHEC) and Hydrophobically Modified Ethoxylated Urethane
(HEUR). HM-EHEC is an example of a HM-polymer with a watersoluble backbone, and hydrophobic groups attached along the
backbone (Figure 1.1.a). HM-EHEC has a relatively high molecular
weight (mw) and the thickening mechanism of HM-EHEC may include

contributions both from chain entanglement and associations
between different hydrophobic parts of the molecule.
Figure 1.1. Schematic

illustration of the structure
of a HM-EHEC and b
HEUR. White necklace
represents hydrophilic
monomers and the bold lines
represent hydrophobic
groups.
Hydrophobically modified Ethoxylated Urethane (HEUR) polymers

have a water-soluble backbone with relatively low mw and
hydrophobic groups attached at both ends of the backbone (Figure
1.1.b). In a solution of a HEUR polymer the thickening effect relies
mainly on hydrophobic associations and entanglements are expected
to be of very small importance.
One way to obtain information about the thickening mechanisms of
HM-polymers in aqueous systems is to synthesize both the HMpolymer as well as the unmodified version of the same polymer and
study the difference in solution behavior. This has been the subject of
numerous studies.1-7
Another way to study the thickening mechanism is by addition of a
third component capable of selectively inhibiting one or more of the
mechanisms that contribute to the thickening effect. For instance it is
well known that, depending on the concentration, addition of
surfactant can either increase or decrease viscosity of a solution of a
HM-polymer.1,8-23 At high surfactant concentrations associations
between hydrophobic parts of the polymer chains are disrupted.
However, this method is unselective and is expected to inhibit all
types of hydrophobic interactions (including both interactions from
polymer hydrophobic tails as well as from hydrophobic patches of the
main chain). A much more selective method to disrupt only some
types of hydrophobic interactions is offered by addition of
cyclodextrins, a group of cyclic substances with a hydrophobic cavity

in
an
otherwise
hydrophilic
molecule.24,25
In
an
aqueous
environment the hydrophobic cavity of the cyclodextrin can host a

hydrophobic molecule or a hydrophobic part of a molecule provided
that it fits into the geometry of the cavity. A hydrophobic group of a
HM-polymer that has formed a complex with a cyclodextrin molecule
does not take part in the thickening mechanism.26-28 In this way it is
possible to distinguish between the contributions to the hydrophobic
associations by different parts of the HM-polymer.
1.1 Hydrophobically modified polymers in paint

This section will summarize some properties that are important for the
paint industry and that can be controlled by the choice of thickener. A
water borne paint consists of several ingredients and an example of a
simple recipe for a water borne paint can be found in Table 1.1. Even
though the thickener constitutes less than 1 % of the paint it is a very
important ingredient since it influences many of the paint properties.
High mw EHEC
Viscosity
HM-EHEC

viscosity profile for three
model paints formulated with
0.45 %w/w high mw EHEC,
0.9 %w/w low mw EHEC, or
0.45 %w/w HM-EHEC
respectively
Low mw EHEC
-3
10
-2
10
-1
10
10
10
10
10
10
-1
Shear rate (s )
Table 1.1. Example of a simple

recipe for a water borne paint
Normally water borne paint is formulated aiming at a certain Stormer

viscosity. The Stormer viscosity corresponds to the viscosity at a
shear rate (10 100 s-1) similar to the shear rate when stirring the
paint in the can, or when pouring the paint. A correct Stormer
Ingredient
Water
(wt)
viscosity is also important when loading the brush since a too low
242
viscosity means that the paint will drip off the brush. The Stormer
viscosity is adjusted by the amount of the polymer. For this reason
Thickener
Defoamer
thickening efficiency of the polymer. Conventional thickeners, with a
Dispersing Agent
high molecular weight, (mw) normally have a high thickening
Preservative
the polymer concentration may vary widely and depends on the
efficiency and give the required Stormer viscosity with a small

addition of the thickener. However at the same time they give a
Filler
110
strongly shear thinning behavior (Figure 1.2). This means that the low
Pigment
180
shear viscosity (<2 s-1) is high whereas the high shear viscosity (>104
Binder (Latex)
455
s-1) is low. Many important paint properties are influenced by the

shear profile. The low shear viscosity (<2 s-1) is important since it
influences the sedimentation of particles in the can. It also influences
the flow properties in the paint film after application of the paint. The
leveling is improved by a decreased low shear viscosity (Figure 1.3)
but on the other hand the newly applied paint film will start to sag on a
Figure 1.3. Example of a

panel from a leveling test. In
this test the surface should
be as smooth as possible.
vertical surface if the low shear viscosity is too low (Figure 1.4). The
high shear viscosity influences the thickness of the paint film during
roller application, since the shear rate in the thin layer between the
surface and the roller is high (>104 s-1). Increased high shear viscosity
means that the applied paint film is thicker resulting in better hiding
properties (Figure 1.5) and thereby reducing the number of coats
required. The main advantage of a conventional high molecular
weight thickener is the low concentration that is needed and thereby
they become cost effective. However, the strong shear thinning
behavior that results in bad leveling and bad hiding power is a
Figure 1.4. Example of a

panel from a sagging test. In
this test the thickness of the
paint film gradually
increases from the top line to
the bottom line. The sagging
is measured as the film
thickness where the paint
starts to sag.
problem.
A less pronounced shear thinning viscosity profile can be obtained by
using a thickener with a lower mw and compared to the high mw
thickeners the leveling and the hiding power are improved. A
disadvantage with this approach is that in order to achieve a required
Stormer viscosity a much higher polymer concentration is required,

which generates a higher cost.
In
general
hydrophobically
modified
polymers
combine
high
thickening efficiency with a less marked shear thinning viscosity

profile. By varying the length of the hydrophobic groups and
molecular weight of the polymer the viscosity / shear profile can be
controlled. The associative thickeners have a strong thickening effect
and give the required Stormer viscosity already at low addition levels.
Actually in most cases their thickening efficiency is comparable to
what is achieved with non-associative thickeners with a high mw. Both
high shear and low shear viscosities are influenced. Compared to the
type of conventional thickeners with high mw the HM-P:s have a much
Figure 1.5. Result from a

hiding power test
less shear thinning profile (Figure 1.2).

In the paint industry, HM-P:s are often referred to as associative
thickeners. Here hydrophobically modified cellulose derivatives (HMHEC and HM-EHEC), HEURs and HM-acrylates are the most
commonly used associative thickeners. There are also important
differences within the group of associative thickeners (Figure 1.6).The
HEUR thickeners together with low mw HM-acrylates give the lowest
tendency to shear thinning. They have the lowest low shear viscosity
and they retain a virtually constant viscosity up to high shear rates
where the viscosity suddenly drops off. HM-HEC, HM-EHEC and high
mw HM-acrylates show rheology profiles that are in-between the
HEUR thickeners and the non-associative thickeners. The less shearthinning behavior of the associative thickeners results in improved
hiding power and leveling properties.
3%w/w HM-EHEC
, (mPa s)
10
10
2%w/w HM-acrylate
3.5%w/w HM-PEG
Figure 1.6. Viscosity, ,

(filled symbols) and complex
viscosity, *, (open symbols)
as a function of shear rate
for three different HMpolymers
10
-1
10
-3
10
10
-2
-1
10
. 10
-1
10
10
10
(s )
One important advantage of HM-polymers is that the spatter from the

roller when the paint is rolled on a wall or a ceiling is drastically
reduced when the paint is thickened with a HM-polymer compared to
when a conventional thickener is used (Figure 1.7). Improved gloss is
another important parameter that is influenced by the use of an
associative thickener compared to when non-associative ones are
used. In light of this the associative thickeners seem to be a good
choice.
It has, however, to be recognized that with the associative thickeners
the properties of the paint may change quite dramatically. The major
problem for associative thickeners is their sensitivity to variations in
coating composition. Changes in type of latex, surfactant or cosolvent concentration, or addition of colorants, can have a
pronounced effect on paint viscosity. This is due to the thickening
mechanism of the associative thickeners that to a large extent is
dependent on associations between the hydrophobic groups on the
Figure 1.7. Results of
spatter tests with two paints
thickened with HM-polymer
(upper) and conventional
thickener (lower).
thickener, since these also associate with other ingredients in the

paint. The associations are very sensitive to variations in paint
composition. For example the monomer compositions of latex
particles, type of surfactant, and the surfactant concentration all have
a large impact on the paint viscosity. The non-associative thickeners
rely mainly on chain entanglements which are much less influenced
by changes in paint composition.
As will be discussed in section 2.2.3 the addition of surfactants can

either increase or decrease the viscosity of the associative thickener
solution depending on the surfactant concentration in the solution and
what type of surfactant is used. One problem is that the surfactant
content and the type of surfactants included in the paint are often
unknown, even to the paint producer. A large fraction of the surfactant
details behind the commercial production of latex are well-hidden
content in the paint originates from the synthesis of the latex, and
secrets. During the production of paint more surfactant is often added
as a wetting agent for the pigment or to improve the stability of the
paint. Normally the surfactant concentration in the paint is on a level
above where the viscosity maximum occurs, as exemplified in Figure
1.8. Additional surfactant therefore causes a reduction of the
viscosity. Paints formulated with HEUR thickeners are in general the
most sensitive to addition of surfactant since associations of
csurf

illustration of the viscosity ,
, of a HM-polymer solution
as a function of surfactant
concentration, csurf,.
hydrophobic groups are the only effective thickening mechanism for

the HEUR thickeners in the concentration range used in paint
formulations. Hydrophobically modified acrylates and cellulose
derivatives are less sensitive since they obtain a considerable part of
their thickening power from chain entanglements.
Colorants used for tinting the paint contain high amounts of
surfactant. The additions of colorants can have a strong impact on
viscosity. In the worst case a paint can lose as much Stormer
viscosity as 30 to 40 KU (30 to 40%) when tinted to a deep-tone
color.
Color acceptance is another parameter of great importance to the
paint industry. A tinted paint can show variations in shade depending
on the magnitude of the shear during the application of the paint. Bad
color acceptance appears as brush-marks, which make the surface
look striped, when the paint is applied with varying shear force from
the paintbrush. In the paint industry the color acceptance is evaluated
in a 'rub-out test' in which one part of the surface of the painted chart
is rubbed while another part is untouched (Figure 1.9). The color
acceptance is judged by means of differences in shade between the
two parts (Figure 1.10). The color acceptance problem becomes
more pronounced when hydrophobic pigments are used. The color

acceptance has been attributed to phase separation caused by the
polymer but the problem is not fully understood.
When formulated in paints the associative thickeners are often used
in combinations, both with other associative thickeners and / or nonassociative thickeners. One example is when a HEUR thickener is
added to a paint thickened with a high mw non-associative thickener
to increase the high-shear viscosity.29 But formulating a paint with
several different thickeners can be full of uncertainties since mixtures
Figure 1.9. The rub-outtest for color acceptance
of polymers often phase separate. The phenomenon with phase

separation is even more pronounced if one of the polymers is
hydrophobically modified and the other is not.30 This is probably the
cause of some of the flocculation problems that occur when
Figure 1.10. Results of color

acceptance test. For a good
result the paint should be as
little affected as possible by
the rub-out test. The far
right panel shows a good
result whereas the far left
panel shows a relatively
poor result.
10
associative thickeners are tested in paint formulations that contain

more than one thickener.

(1)
Williams, P. A.; Meadows, J.; Phillips, G. O.; Senan, C.

Cellulose: Sources and Exploration 1990, 37, 295-302.
(2)
Thuresson, K.; Lindman, B. J. Phys. Chem. 1997, 101, 64606468
(3)
Wang, K. T.; Iliopoulos, I.; Audebert, R. Polymer Bulletin 1988,

20, 577-582.
(4)
Valint, J., P.L.; Bock, J. Macromolecules 1988, 21, 175-179.
(5)
Bock, J.; Siano, D. B.; Valint Jr., P. L.; Pace, S. J. In Polymers

in aqueous media; Glass, J. E., Ed.; American Chemical
Society: Washington DC, 1989; Vol. 223, p 411-424.
(6)
Glass, E. J. Coatings Technology 2001, 73, 79-98.
(7)
Winnik, M. A.; Yekta, A. Current Opinion in Colloid & Interface

Science 1997, 2, 424-436.
(8)
Gelman, R. A. In 1987 International dissolving

Conference; TAPPI, Ed. Geneva, 1987, p 159-165.
(9)
Magny, B.; Iliopoulos, I.; Audebert, R.; Piculell, L.; Lindman, B.

Progr. Colloid Polym. Sci. 1992, 89, 118-121.
Pulps
(10) Iliopoulos, I.; Wang, T. K.; Audebert, R. Langmuir 1991, 7, 617619.

(11) Annable, T.; Buscall, R.; Ettelaie, R.; Shepherd,
Whittlestone, D. Langmuir 1994, 10, 1060-1070.
P.;
(12) Loyen, K.; Iliopoulos, I.; Olsson, U.; Audebert, R. Progr. Colloid
Polym. Sci. 1995, 98, 42-46.
(13) Piculell, L.; Thuresson, K.; Ericsson, O. Faraday Discuss. 1995,
101, 307-318.
(14) Chen, M.; Glass, J. E. Polym. Mater. Sci. Engin. 1995, 73, 449450.
(15) Aubry, T.; Moan, M. J. Rheol 1996, 40, 441-448.
(16) Piculell, L.; Guillemet, F.; Thuresson, K.; Shubin, V.; Ericsson,
O. Adv. Colloid Interface Sci. 1996, 63, 1-21.
(17) Persson, K.; Wang, G.; Olofsson, G. J. Chem. Soc. Faraday
Trans. 1997, 90, 3555-3562.
(18) Panmai, S.; Prud'homme, R., K.; Peiffer, D., G.; Jockusch, S.;
Turro, N., J. Polym. Mater. Sci. Engin. 1998, 79, 419-420.
(19) Nilsson, S.; Thuresson, K.; Hansson, P.; Lindman, B. J. Phys.
Chem. 1998, 102, 7099-7105.
(20) Olesen, K. R.; Bassett, D. R.; Wilkerson, C. L. Progress Organic
Coatings 1998, 35, 161-170.
(21) Jimnez-Rigaldo, E.; Selb, J.; Candau, F. Langmuir 2000, 16,
8611-8621.
(22) Chronakis, I. S.; Alexandridis, P. Marcomolecules 2001, 34,
5005-5018.
11
(23) Steffenhagen, M. J.; Xing, L.-L.; Elliott, P. T.; Wetzel, W. H.;

Glass, J. E. Polym. Mater. Sci. Engin. 2001, 85, 217-218.
(24) Immel, S.; Lichtenthaler, F. W. Starch/Strke 1996, 48, 225232.
(25) Connors, K. A. Chem. Rev. 1997, 97, 1325-1357.
(26) Akiyoshi, K.; Sasaki, Y.; Kuroda, K.; Sunamoto, J. Chemistry
Letters 1998, 1998, 93-94.
(27) Zhang, H.; Hogen-Esch, T. E.; Boschet, F.; Margaillan, A.
Langmuir 1998, 14, 4972-4977.
(28) Gupta, R. K.; Tam, K. C.; Ong, S. H.; Jenkins, R. D. In XIIIth
International Congress on Rheology Cambrige, UK, 2000, p
335-337.
(29) Howard, P.; Leasure, E.; Rosier, S.; Schaller, E. J. Coating
Technology 1992, 64, 87-94.
(30) Tsianou, M.; Thuresson, K.; Piculell, L. Colloid Polym. Sci.
2001, 279, 340-347.
12
Chapter 2
Hydrophobically modified polymers
Hydrophobically
modified
water-soluble
polymers
(HM-P)
are
polymers with hydrophobic groups chemically attached to a

hydrophilic polymer backbone. They are often also referred to as
associative polymers or associative thickeners. The first studies on
HM-P were made by Strauss and coworkers more than 50 years ago.
They are described in a review article.1 The work was done with
hydrophobically modified polyelectrolytes. The idea behind the
studies was that since soap molecules associate to form micelles in
aqueous solution also surfactants chemically grafted to a water
soluble polymer would form micelles. That indeed was what they
found. In addition they found that the polysoaps gave unique
solubilizing effects and a surprisingly large increase of the viscosity to
an aqueous solution. These two effects of HM-P are widely utilized.
The largest application for HM-P is as rheology modifier in water
borne paint. Landoll and his coworkers described the first associative
thickeners for water borne paint in the eighties.2-4 They worked with
hydrophobically modified (hydroxyethyl) cellulose (HM-HEC) which is
a
nonionic
cellulose
ether.
hydroxyethyl
cellulose
(HM-EHEC),
hydrophobically
modified
(HEUR)
hydrophobically
modified
ethoxylated
urethanes
Hydrophobically
and
modified
ethyl
polyacrylates (HM-PA) are other examples of associative thickeners

that
have
been
developed
for
the
paint
application.
The
hydrophobically modified cellulose derivatives are, still after 20 years,

the largest class of associative thickeners for water borne paint.
13
2.1 Structure and synthesis of hydrophobically

modified polymers
Depending on how the hydrophobic groups are situated in the
molecule HM-polymers can be divided into two categories. The first
has the hydrophobic groups attached at the ends of the polymer
backbone and they are referred to as hydrophobically end-capped
polymers (Figure 1.1.b). The second category has the hydrophobic
groups grafted along the polymer backbone. These are called comb
like HM-polymers (Figure 1.1.a).
2.1.1 HEUR thickeners

Hydrophobically
modified
ethoxylated
urethanes
(HEURs)
are
examples of end-capped water-soluble polymers. They consist of a
hydrophilic polyethylene glycol (PEG) segment in the middle with

hydrophobic groups attached at both ends. Compared to other
polymers used as thickeners the molecular weight (mw) of a HEUR
thickener is normally relatively low, 15,000 to 50,000.5 Often the
molecular weight distribution of a commercially available HEUR is
broad due to the synthesis procedure used for the manufacture of the
polymer. Polyethylene glycol of low molecular weight, e.g. 6000, is
reacted with a slight excess of diisocyanate. The resulting polymer
b
picture of the synthesis of
HEUR-polymers. The
necklaces represent
polyethylene glycol chains.
Bold lines represent
hydrophobic end groups and
filled balls represent
diisocyanate groups or
diurethane linkages.
chains with isocyanate groups at both ends are then reacted to a long
chain alcohol (Figure 2.1.a).5 A way to synthesize a HEUR with a
more narrow distribution is offered by the reaction of an alcohol
ethoxylate to diisocyanate (Figure 2.1.b). The HEUR-polymer from
this process has a polydispersity index (weight average molecular
weight (Mw) / number average molecular weight (Mn)) of about 1.1.
This type of HEUR has been used in the present studies and is
referred to as Triblock or HM-PEG. It should be mentioned that even
though the present polymers have a low polydispersity index, model
HEUR thickeners with even lower polydispersity index (Mw/Mn =1.01)
have been synthesized.6 Here the starting material was PEG, with
narrow molecular weight distribution, which was reacted to alkyl
p-toluene sulphonate at both ends.
14
2.1.2 Hydrophobically modified EHEC

The
base for
ethyl
hydroxyethyl
cellulose
(EHEC)
and
for
hydrophobically modified EHEC (HM-EHEC) is cellulose, one of the

most common natural polymers. Cellulose is a polysaccharide built up
from 1,4-anhydroglucose units (AHG). The cellulose molecules in
native cellulose form large crystalline regions, and therefore cellulose
is insoluble in water. To make cellulose soluble it has to be modified
to split up the crystalline packing. The process for making cellulose
derivatives starts with an alkalization step. The alkalization has two
purposes. Firstly by introducing charges into the molecules, the
cellulose swells. This makes individual cellulose chains available for
the chemical reaction. Secondly it also acts as catalyzation for the
modification reactions. During the synthesis of EHEC the alkalized
cellulose is modified by a reaction with ethylene oxide and then with
HO
ethyl chloride. Both reaction steps are performed at elevated
OH
temperature. Since both ethylene oxide and ethyl chloride are volatile
O
O
compounds a pressurized reaction vessel is required.
Na
Each AHG has three hydroxyl groups available for reaction. The
O
+
reaction of one ethylene oxide molecule to one of the hydroxyl groups

on an AHG results in a new hydroxyl group that is also reactive
HO
(Figure 2.2). The newly formed hydroxyl group has a reactivity

comparable to that of the hydroxyl groups on the AHG which means
that besides the reaction of the hydroxyl groups on the AHG there is
also a chain growth reaction going on. The outcome is that short oligo
(ethylene oxide) chains are formed.7 The molar substitution of
ethylene oxide (MSEO) is the average total number of ethylene oxide
OH
O
O
O
O
Na
Figure 2.2. The reaction of

ethylene oxide to alkalized
cellulose.
groups per AHG (Figure 2.3). For practical reasons the upper limit for
MSEO is about 2.5 to 3 since the efficiency of the reaction decreases
dramatically above that level due to side reactions. Up to this point
about 70 % of the ethylene oxide reacts with cellulose to form ether
groups. The remainder forms glycols by reaction with water, or ethers
of glycols by reaction with ethyl chloride.7
15
HO
Figure 2.3. Possible

structure element of an
OH
O
O
O
EHEC molecule.
represents a hydroxyethyl
group. Ethyl groups are
represented by bold lines.
In this example
MSEO= (4+3+0+2+1)/5 =2
DSethyl=(2+2+0+1+1)/5=0.8
OH
O
HO
O
HO
OH
OH
HO
O
O
HO
OH
O
O
O
OH
In contrast to the reaction with ethylene oxide where new hydroxyl

groups form, the ethyl chloride reaction consumes sodium hydroxide
+
and the hydroxyl group that has reacted to an ethyl chloride is
Na
O
terminated for further reaction (Figure 2.4). The number of hydroxyl
OH
O
groups per AHG that has reacted is expressed as degree of
substitution (DS) and the figure ranges from 0 to 3. Practically the
OH
upper limit for DSethyl is about 1 since the water solubility of the final
EHEC polymer decreases dramatically with increasing DSethyl.8 Of
Cl
course the reaction does not give a perfectly homogeneous

+
Na Cl
process for EHEC gives an uneven distribution of the hydroxyethyl
O
OH
O
O
O
substituent-distribution over all AHGs. It is likely that the synthesis

and ethyl substituents. Therefore the numbers of DSethyl and MSEO are
average values. Segments of anhydroglucose units that have a high

OH
degree of ethyl substituents are slightly hydrophobic. In water solution

the ethyl groups can give rise to hydrophobic interactions provided
Figure 2.4. The reaction of

ethyl chloride to alkalized
hydroxyethyl cellulose.
that they are situated in long sequences. This is an origin of the

backbone associations and the reason why the unmodified EHEC is
surface active and shows an associative behavior.9,10 The situation
is similar for other short hydrophobic groups (C6 or shorter) where an
anhydroglucose unit bearing hydrophobic groups can be seen as a
hydrophobic monomer unit of a copolymer. The cellulose backbone is
relatively stiff and the associations from the short hydrophobic groups
are too weak to force the polymer backbone to bend into a loop
where the hydrophobic groups could intra-aggregate. Instead the
result is inter-associations between hydrophobic segments on
16
different polymer chains, which can be detected as increased solution

viscosity.3,11,12 If the polymer concentration or the flexibility of the
polymer backbone changes the situation may be different.
By reacting aliphatic groups to the EHEC polymer a hydrophobically
modified EHEC is obtained (Figure 2.5). The HM-EHEC obtained in
this way is an example of a comb like HM-P. It has hydrophobic
groups grafted along the water-soluble EHEC backbone. Only a small
amount of hydrophobic groups are required to totally change the
properties of the polymer.3,11 In our study less than 1% of the
glucose units of the EHEC backbone have hydrophobic groups
attached and this was enough to substantially change the solution
properties as compared to those of the corresponding unmodified
EHEC.
HO
O
O
HO
O
O
O
O
O
HO
O
HO
HO
O
HO
O
O
OH
O
O
HO
OH
O O
R=
O
O
(C16)
(C14)
O
O
HO
(C12)
HO
(NP)
OH
O
x
O
O
O
HO
n = 1 or 2
(C1618)
Figure 2.5. Possible structure segment of the HM-EHEC:s studied in paper III.
R=(NP) for HM-(NP)-EHEC, R=(C12) for HM-(C12)-EHEC, R=(C14) for HM-(C14)-EHEC,
R=(C16) for HM-(C16)-EHEC, and R= a blend of (C16) and (C18) for HM-(C1618)-EHEC
17
In paper III we have investigated the effect of various chain lengths of

the hydrophobic groups. Alkyl groups varying from C12 to C16 or a
blend of C16 and C18 or nonylphenol have been used. The HM-EHEC
polymers that were obtained with these hydrophobic groups are
referred to as HM-(C12)-EHEC, HM-(C14)-EHEC, HM-(C16)-EHEC,
HM-(C16-18)-EHEC and HM-(NP)-EHEC, respectively.
The values of MSEO, DSethyl, and MShydrophobe for the HM-EHEC:s
included in this study are presented in table 2.1.2
Table 2.1.2. The substitution degrees of ethylene oxide (MSEO), ethyl (DSethyl), and hydrophobic tails
(MShydrophobe) of each of the polymer samples given as average numbers of substituents per repeating glucose
unit. Independent repeated determinations render an uncertainty in the numerical values of about 5%. The
abbreviations given in the 'Hydrophobic group' column refers to the unmodified parent EHEC (0), HMEHEC modified with, nonylphenol groups (NP), C12 groups (C12), C14 groups (C14), C16 groups (C16), and
with C16 C18 groups (C1618). The values for concentration of hydrophobic groups in the solution, chydrophobe,
are calculated for 1% w/w solutions.
Hydrophobic
group
mw/AHG
(g/mol)
chydrophobe
(mmolal)
277.0
0.8
0.008
279.7
0.28
2.1
0.8
0.0086
279.9
C14
2.1
0.8
0.0082
280.0
C16
2.1
0.8
0.0081
280.1
C1618
2.1
0.8
0.009
280.9
MSEO
DSethyl
2.1
0.8
NP
2.1
C12
MShydrophobe
0.29
2.1.3 Comb HEUR

The comb-like HEUR polymers have some interesting properties but
they have not yet received much attention. This may be because they
are complicated to synthesize in a well-characterized way.5 One
possible route to synthesize them is offered by reacting ethoxylated
monoalkylamines (EMAA) to a diisocyanate (Figure 2.6). It is a step
growth reaction and by changing the reaction conditions the
molecular weight of the polymer is varied. The molecule consists of a
number of EMMA-units, each bearing one hydrophobic group and
one amine function. This means that the polymer at low pH has a
18
positive net charge, located close to each hydrophobic group. Two

comb HEUR:s with this structure were studied in Paper II.
+
+
+
+

illustration of the synthesis
process for comb HEUR
polymers. White necklace
represents a sequence of
hydrophilic monomers and
the bold lines represent the
hydrophobic groups. Filled
balls represent diisocyanate
monomers and balls with a
plus sign represent
protonated amino groups.
Their alkyl group is in both cases a C12 chain, while the length of the
polyethylene oxide spacer between the alkyl groups has been varied.
The polyethylene oxide chains on the alkylamine contain on average
51 or 74 units, respectively. The Mw was estimated at about 25 000
for both of the polymers, indicating that they on average consist of
roughly four units. The way they have been produced suggests that
they should have a wide distribution in molecular weight and it was
found that Mw/Mn was about 2.2 for both these polymers.
2.2 Hydrophobically modified polymers in

aqueous solution
The behavior of the polymer molecules in solution depends to a large
extent on the polymer concentration, c. To describe how the behavior
of a HM-P varies with the polymer concentration it is easier to start
the discussion on the behavior of the unmodified parent polymer. The
polymer concentration interval can be divided into three different
regimes, the dilute, the semidilute and the concentrated regime
(Figure 2.7).13 In the dilute regime c is low and the mean centrum to
centrum distance between the polymer coils is larger than the mean
radius of a single polymer coil denoted as the radius of gyration, Rg.
19
The individual polymer chains are expected to move independently of

each other in the solution.
Figure 2.7. Polymer

concentration intervals
dilute solution (c<c*),
semidilute solution (c>c*)
and concentrated solution
(c>>c*)
c<c*
c>c*
c>>c*
In the semidilute regime Rg is larger than the mean distance between

the coils. Since the total volume of all polymer coils exceeds the
volume of the solution the polymer coils are forced to overlap and the
concentration where this occurs is often referred to as the overlap
concentration and is denoted c*. The chain of one polymer molecule
will entangle with other polymer molecules (Figure 2.8). The result is
entanglements of the polymer chains and the formation of a transient
polymer network which can be detected as a dramatic increase in the
Figure 2.8. Entanglements
of polymer molecules.
viscosity of the polymer solution. The overlap concentration can

roughly be estimated as the reciprocal of the intrinsic viscosity,
c*1/[], and is for most polymers in the region 0.1 to 10 %w/w. The
importance of the entanglements to the dynamics increases with
increasing polymer concentration. The chemical structure of the
polymer is very important for the coil size and thereby for the behavior
of the polymer in solution. An increased mw results in larger coils and
more chain entanglements, which can be seen as increased
viscosity.13 The coil size is also influenced by the chemical
composition of the backbone. A polyethylene glycol based polymer is
more flexible than a polymer with a cellulose origin and has therefore
a smaller coil size.14 The repulsion between the ionic groups makes
the polymer backbone of a polyelectrolyte stiff. The electrostatic
repulsion is strongly influenced by the ionic strength in the solution.
The fact that the viscosity of a polyacrylate solution decreases when
20
salt is added can be explained by reduced coil sizes due to increased

flexibility of the polymer chains.11,15
In the concentrated region the system consists of highly entangled
polymer chains. The behavior of the polymer molecules is more
similar to that in a polymer melt than to the behavior in the polymer
network in the semidilute solution.
Describing the behavior of hydrophobically modified polymers it is
important to notice that according to the properties of the unmodified
analogue the HM-P molecule also has the possibility to associate with
other HM-P molecules. The association of the hydrophobic groups is
very similar to self-association of surfactants. To minimize the contact
between water and hydrophobic groups the hydrophobic groups
associate to each other and form a water-poor domain, which is the
Figure 2.9. Associations of

hydrophobic groups of HMpolymer molecules.
interior of a micelle. The surface of the micelle is covered by the

hydrophilic polymer backbone. In aqueous solution the hydrophobic
groups of a hydrophobically modified polymer associate with each
other resulting in physical bonds holding different parts of the polymer
chains together (Figure 2.9). In a snapshot picture it can be described
as a cross-linked gel but in contrast to covalent bonds the physical
bonds are reversible. They break and reform continuously. A
hydrophobic group on one polymer molecule can either take part of
an
intra-molecular
association,
i.e.
it
interacts
with
a)
another
hydrophobic group on the same polymer chain, or interacts with a

hydrophobic group on another polymer molecule (inter-molecular
association) (Figure 2.10). At low concentrations the probability for
interaction between different HM-polymer molecules is small. Intraaggregation results in a reduced coil size.1,16-20 The intrinsic
viscosity for a HM-P is therefore often lower than for the unmodified
analogue
of
the
same
polymer.
Upon
increasing
polymer
concentration inter-molecular associations become more important

and the three-dimensional network is formed. This gives rise to a
dramatic increase of the solution viscosity. The onset concentration of
inter-molecular
association
is
often
well
below
the
b)
Figure 2.10. Illustrations of

inter-molecular association
(a) and intra-molecular
associations (b).
overlap
concentration, c*, of the corresponding unmodified polymer with the

same molecular weight.21,22
21
The strength of the hydrophobic interactions between polymer chains

is influenced by:
the length of the hydrophobic groups
the molar substitution of hydrophobic groups (MShydrophobe)
the distribution of the hydrophobic groups along the polymer

backbone.
Longer hydrophobic groups give an increased residence time of a

hydrophobic group within the micelle and also increased lifetime of
the aggregates of hydrophobic groups. This was illustrated by Sau et
al, who found that if two identical polymers are substituted with
different hydrophobic groups the polymer with the longer hydrophobic
groups gives the highest viscosity to a water solution.4 The results in
( cP)
section 2.2.2 also illustrate this.

500
The influence of MShydrophobe on the solution viscosity can be divided
400
into three different regions: At low MShydrophobe there is a positive

correlation between MShydrophobe and viscosity. This can be explained
300
2
200
by an increased number of inter-connection points holding the

polymer network together. Depending on the structure of the polymer
100
backbone and the length of the hydrophobic groups there is a
0
0
% w/w C 12 hydrophobe
Figure 2.11. Brookfield

viscosity of 2% w/w solution
of HM-HEC substituted with
1,2-epoxydodecane as a
function of degree of
hydrophobic modification.
Reproduced from 3
viscosity maximum somewhere typically in the range of 1 to 5

hydrophobic groups per 100 repeating units of the polymer backbone
if a comb like polymer is investigated (Figure 2.11).3 The reason for
the decrease is a conversion of intermolecular associations to intramolecular association and a gradual degradation of the polymer
network.20,23 At even higher MShydrophobe the HM-P becomes insoluble
in water.
The synthesis of HM-P is often performed in a two-phase system
where one phase is an aqueous solution of the polymer backbone
and the other phase consists of the hydrophobic reagent. This gives
rise to a HM-P with a more or less blocky distribution of the
hydrophobic groups along the polymer backbone. Depending on the
type of hydrophobic groups and the length of the hydrophobic
segments the more blocky structure can favor the formation either of
22
intra-associations or inter-associations (Figure 2.10). Provided that

the hydrophobic associations are strong they can force the polymer
backbone to adopt conformations that give rise to intra-molecular
associations. Selb et al have shown that for HM-P with C16-alkyl
groups the viscosity of the polymer with the blocky structure can be
4
10
random distribution of the hydrophobic groups (compare to the right

part of the diagram in Figure 2.11).23,24 This is in contrast to what is
(cP)
several times less than that of the corresponding polymer with a more
10
10
described in section 2.1.2 for short hydrophobic groups.
Hydrophobically modified ionic polymers like HM-PA are strongly

influenced by the salt content in the solution. As mentioned above
increasing salt concentration reduces the repulsion between ionic
groups on the polymer backbone. At the same time the addition of
salt makes the solvent more polar which promotes the hydrophobic
associations (Figure 2.12). At low salt concentrations the increased
interchain cross-linking predominates leading to a viscosity increase.
1
2
csalt (%w/w)
Figure 2.12. Viscosity as a

function of NaCl
concentration for 2% w/w
solution of HM-polyacrylate
substituted with C18
hydrophobic groups (3% of
the repeating units covered).
Reproduced from11
At higher ionic strength the electrostatic effects prevail and a

reduction in the viscosity occurs (Figure 2.12).15,25
2.2.1 HM-PEG in aqueous solution

The commonly accepted mechanism for the association of the HEUR
thickeners is somewhat different from the one for the comb like
polymers described in the previous section (Figure 2.13).6,26-28 At
very dilute conditions the HM-polymer molecules exist as free
molecules
(unimers)
or
as
oligomers
with
low
aggregation
numbers.27 With increasing polymer concentration the polymer

molecules start to form small micelle-like structures with the
hydrophobic parts of the thickener looping back into the micelle,
forming flower-like structures. The onset of micelle formation
generally occurs already at polymer concentrations far below the
overlap concentration of the unmodified analogue of the polymer (c*).
The formation of micelles becomes more cooperative with increasing
length of the hydrophobic groups.28 The unfavorable entropy caused
23
by bending the hydrophilic backbone into a loop conformation

opposes the micelle formation. Consequently the formation of flower
micelles is favored by longer hydrophobic groups and by increased
length of the PEO-spacer as can be seen as a decrease of the
concentration where aggregation starts to occur. Fluorecence
quenching techniques have been used to determine the average
number of hydrophobic groups per micelle (NR) on a variety of HEUR
polymers.28,29,30,31 It was found that flower micelles are very
uniform in size and contain in the range of 20 to 30 hydrophobic
groups per micelle. Over a wide concentration range NR is
independent of the polymer concentration. NR for HEUR polymers is
considerably lower compared to the aggregation number for related
surfactants forming spherical micelles, which is typically 60 to 80.13
This can probably be explained by the fact that the polymer
backbones of HM-PEG are large head groups which limits the
number of hydrophobic groups that can participate in the same
micelle.
Unimers
Flower
Micelles
Clusters
Network
increased cHM-PEG
Figure 2.13. Schematic representation of the self-aggregation of HM-PEG as function of increasing

cHM-PEG
24
With increasing polymer concentration the average distance between

the flower micelles becomes smaller and larger aggregates are
formed. The flower micelles can be seen as building blocks for the
formation of larger aggregates. Transient bridges consisting of
HM-polymer molecules with one hydrophobic group in one micelle
and the other end in a neighboring micelle are formed resulting in
clusters of micelles. The driving force for this cross-linking is the
lowering in free energy achieved by allowing some of the thickener
molecules to attain more flexible conformations of the hydrophilic
backbones with no strict need for looping back. In the case of long
hydrophobic groups the aggregation into clusters starts at a
concentration far below close packing of micelles. For the situation
where the attraction forces are weaker (shorter hydrophobic groups)
the aggregation starts at higher concentrations, but still below the
concentration for close packing of micelles. Semenov et al predicted
that these systems at concentrations below close-packed micelles
would phase separate into one phase containing closely packed
micelles and one phase impoverished in polymer.22 However, in our
studies on aqueous solutions of HM-PEG with C16-18 hydrophobic
groups (the structure is described in paper I) no macroscopic phase
separation occurred at room temperature. Instead a microscopic
phase
separation
has
been
suggested
with
polymer
rich
microdomains (clusters) in a diluted bulk phase.6,27,28 Contrary to

the micelles, which have rather well defined aggregation numbers, it
is reasonable that the clusters appear in a wide range of sizes and
that the average cluster-size increases with increasing HM-PEG
concentration.6,27,32 The polymer concentration inside the clusters
differs from the average concentration in the solution. One indication
for the solution being inhomogeneous is given by the phase behavior
of triblock solutions which is further described in section 2.2.4.1.
Upon increasing polymer concentrations the distances between
different clusters become smaller which gives the possibility for the
polymer chains to more frequently connect micelles located in
different clusters and a three-dimensional network that extends over
macroscopic distances is formed. This can be detected as a dramatic
increase of the solution viscosity. It occurs at a polymer concentration
25
where the solution is still likely to be very inhomogeneous with large

concentration fluctuations. Due to the large concentration fluctuations
the polymers that connect micelles located in different clusters and
have to span polymer depleted regions are likely to be rare. In
contrast the inter-micellar links within the clusters are much more
numerous. All the physical bonds are temporary and the clusters
continuously break and reform. Therefore the HM-P:s that are
involved in forming bridges between the clusters at one moment can
change to be an intra-micellar link at the next moment .
2.2.2 HM-EHEC in aqueous solution

In paper III we found that the hydrophobic group chain length had a
dramatic effect on the low shear viscosity of aqueous solutions of
4
HM-EHEC. The C12-group only has a minor effect on the viscosity,
(mPa s)
10
and experiments with shorter hydrophobic groups (not presented)
10
have shown that the hydrophobic groups should have at least 12

carbon atoms to have any noticeable effect on the viscosity. By
10
(0) (NP) (C12) (C14) (C16)
Figure 2.14. Viscosity of 1%

w/w solutions of HM-EHEC
with varying length of the
hydrophobic groups. (0)
represents unmodified EHEC
increasing the length of the hydrophobic chains from C12 to C16 the
viscosity increased two orders of magnitude (Figure 2.14). This is in
good agreement with results from earlier studies.4,25,26 This effect is
ascribed to the residence time of the hydrophobic chains in the
polymer micelles, which increases for longer hydrophobic groups
and results in slower motions of the polymer molecules and thereby a
higher viscosity.26,33
For grafted HM-P with low MShydrophobe, like HM-HEC and HM-EHEC,
the average number of hydrophobic groups per micelle (NR) is low.
The low aggregation number is likely to result from the polymer chain
being a very large head group. The relatively stiff backbone from
cellulose ether prevents formation of loops and the consequence is
that only a small number of hydrophobic groups can take part in the
formation of each micelle. NR for HM(NP)-EHEC and HM-HEC
micelles have been determined to be about five to ten9,33 compared
to 60 to 80 for surfactants forming spherical micelles13 and 20 to 30
for the more flexible HEUR thickeners. The consequence is that
26
rather poor micellar structures are formed with a high degree of

contact between water and hydrophobic groups.
From what has been discussed above, it follows that there are at
least three types of interpolymer crosslinks, that contribute to the
formation of the three dimensional network of a HM-EHEC solution
(Figure 2.15). Apart from chain entanglements and associations
(a)
between hydrophobic side groups also associations of hydrophobic

segments of the polymer backbone play an important role. The
hydrophobic segments on the EHEC and HM-EHEC backbones have
been ascribed to patches with high substitution density of ethyl
groups, described in section 2.1.2.34 Earlier when different
HM-EHEC batches have been compared it has been assumed that
(b)
since all studied HM-EHEC:s were synthesized according to the

same process the substitution pattern should be similar and that the
interactions of hydrophobic backbone segments contribute almost
equally for all HM-EHEC:s. To give a clearer picture of the
contribution from the different types of crosslinks it would be helpful to
(c)
have methods to study the contributions separated from each other.

This will be discussed in chapter 3.
Figure 2.15. Interpolymeric

cross-links in HM-EHEC
solutions. (a) chain
entanglements, (b)
associations between
hydrophobic side-chains and
(c) associations between
hydrophobic segments of the
polymer backbone.
27
2.2.3 Interaction between hydrophobically modified polymers
and surfactants

illustration of the influence
of surfactant concentration
on the viscosity of solutions
of HM-polymers.
log csurf
Hydrophobically modified polymers in aqueous solution interact

strongly with surfactants leading to the formation of mixed micelles. At
concentrations of HM-P corresponding to the semidilute regime of the
unmodified parent polymer it is found that the viscosity passes via a
pronounced maximum when the surfactant concentration is gradually
increased (Figure 2.16).17,18,35-40 The degree of interaction is
determined both by the structure of the surfactant and the nature of
the polymer. As described in section 2.2.1 and 2.2.2 the micellar
structures of HM-P normally have low aggregation numbers
compared to surfactant micelles and the consequence is a quite large
degree of contact between water and the hydrophobic groups. At low
surfactant concentrations, already far below the cmc of the surfactant,
the surfactant molecules are incorporated in the existing micelles
from the HM-P. Incorporation of surfactant molecules into the micelles
reduces the water hydrocarbon contact. This increases the activation
energy for detachment of a hydrophobic group from the micelle
thereby increasing the residence time of the hydrophobic groups
within the micelles thus leading to stronger associations.9,40 The
viscosity in an aqueous solution of a HM-P depends on the number of
interconnecting links in the network and on the relaxation time. A
28
changed viscosity can be the result of a variation of either of these

parameters, or both. For the end-modified polymers an increased
number
of
active
links
has
been
observed29,41
while
for
hydrophobically modified cellulose ethers the effect of increased

viscosity upon addition of surfactant is suggested to be caused mainly
by increased relaxation times.9,33,42 Besides the increased viscosity
the stronger association can also be detected as a dramatic shift to
lower TCp (compare section 2.2.4.2).
At surfactant concentrations above the viscosity maximum the
number of micelles in the solution increases. This results in an
increased ratio between micelles and hydrophobic groups of the
polymer. In this process the decreased viscosity is a consequence of
the physical network losing some of its connectivity. At high surfactant
concentrations where the number of micelles exceeds the number of
polymer hydrophobic groups in the system there is only one polymer
hydrophobic group in each micelle. At this stage the viscosity is
independent of the surfactant concentration and has a value that is
even lower than for the HM-P solution before addition of surfactant.
How strong the effect is depends on the structure of the surfactant.
Normally nonionic polymers interact more strongly with anionic
surfactants than with nonionic or cationic surfactants. In line with this
it has been found that anionic surfactants give the most pronounced
viscosity increase and also the largest reduction of the viscosity at
excess surfactant.34
Hydrophobically modified polyelectrolytes, for instance HM-PA,
interact strongly with oppositely charged surfactants. The interaction
is caused by a combination of electrostatic attraction and hydrophobic
forces. The strength of the hydrophobic associations increases with
increasing length of the hydrophobic groups on the polyelectrolyte.
With long hydrophobic groups the hydrophobic interactions can be
strong enough to overcome the electrostatic repulsion between the
polymer backbone and surfactants of the same charge resulting in a
net attraction.36,43
29
2.2.4 Clouding
For most substances the solubility increases with increasing
temperature. This is not the case for EHEC and HEUR thickeners.
They both belong to a family of polyethylene oxide containing
substances that have a reversed relationship between solubility and
temperature.44 The solubility of these substances decreases with
increasing temperature. At temperatures above a critical value a
water solution containing any of these polymers phase separates into
one polymer rich phase and one phase depleted in polymer. The
phase separation can be detected by the scattering of light resulting
in a cloudy appearance of the solution. The temperature where the
solution first becomes hazy is referred to as the cloud point
C O
H
O C
C C
HH
H
O
phenomenon have been done. One reasonable explanation builds on
H
C
temperature below TCp results in a one-phase situation and a

transparent solution. Many attempts to explain the reversed solubility
temperature, TCp. The process is reversible and decreasing the
conformational changes of the polymer molecules with changing

C
Figure 2.17. Different

conformations of an ethylene
oxide group. Conformation
A has low energy and is
more polar compared to
conformation B.
temperature. The polyethylene oxide chain has a large number of

possible conformations. The conformation with the lowest free energy
in a polar environment (conformation A in Figure 2.17) has a low
statistical weight. At low temperature the low energy conformation will
dominate. Conformation A has a large dipole moment. With
increasing temperature other conformations with higher energy but
also with higher statistical weight will be more and more important.
The higher energy conformations have a lower dipole moment and
conformation B in Figure 2.17, for instance, has virtually no dipole
moment. The consequence is that the polyethylene oxide chain
becomes less and less polar with increasing temperature. This gives
an increasing tendency to phase separation since water-polymer
interactions become less favorable with increasing temperature.44,45
Phase behavior studies give the possibility to study the influence of
other substances on the interaction between the polymer chains.
Addition of a third water soluble component can have a large impact
on the TCp.45 For instance most salts decrease the TCp (salting out)
but some salts with large anions, e.g. I- and SCN-, have the opposite
30
effect (Figure 2.18). The addition of a salt that does not interact with
the polymer molecules results in a more polar environment and
thereby stronger hydrophobic interactions and increased tendency for
phase separation. On the contrary the large polarizeable anions I- and
SNC- interact with the unpolar parts of the polymer molecules
resulting in an increased entropic penalty of phase separation.
90
80
NaSCN
(TCp ) (C)
TCp (C)
70
60
50
NaCl
40
30
80
70
60
10
csurf (mmolal)
cSalt (M)
Figure 2.18. TCp as a function of salt
Figure 2.19. TCp as a function of sodium
concentration for 0.9% w/w solution of
dodecyl sulphate concentration for 0.9%
EHEC. Reproduced from45
w/w solution of EHEC. Reproduced

from45
Surfactants are another type of substance that strongly influences the

phase
separation
temperature.
Depending
on
the
surfactant
concentration, csurf, and type of surfactant, addition of surfactants can

either increase or decrease the TCp. Upon progressively increasing
50
(SDS), TCp is found to decrease initially (Figure 2.19). At slightly
40
higher csurf the TCp passes through a minimum and at even higher csurf
TCp increases. The trend is similar for addition of other micelle forming
surfactants provided that the surfactant molecules associate with the
polymer. If there is no association between polymer and surfactant
the result can be a segregative phase separation with the polymer
enriched in one phase and the surfactant in the other phase.
Since TCp is strongly influenced by added surfactants it is also likely
that other surface-active ingredients have a large impact on the
phase separation temperature. It is therefore not sufficient to measure
(Pa s)
the surfactant concentration, csurf, of the ionic surfactant C12SO4Na
30
20
10
20
30
40
50
T (C)
Figure 2.20. Complex

viscosity as a function of
temperature for a model
paint thickened with EHEC.
The phase separation can be
detected as a step decrease
in viscosity.
the TCp in water to predict the phase separation temperature for a
31
60
paint. Since the paint is a dispersion of particles rather than a clear

solution it is not possible to use the normal cloud point measurements
to detect the phase separation. As illustrated in Figure 2.20 the phase
separation can instead be determined as a dramatic viscosity
decrease when the temperature is increased.
2.2.4.1 Cloud point of HM-PEG
120
100
80
TCp (C)
Figure 2.21. TCp as a

function of polymer
concentration for C1618(EO)140 polymer (open
circles) and for C1618(EO)140-IPDU-(EO)140-C1618
polymer (filled circles).
60
40
20
0
1
0
cpolymer (% w/w)
HM-PEG with hydrophobic groups at both end show a dramatic drop

in the TCp compared to unmodified PEG or PEG that is only modified
at one end (Figure 2.21). 6,30,46 TCp also strongly depends on the
polymer concentration and the cloud point curve as a function of
polymer concentration passes via a minimum. The effect of TCp
depression by introducing hydrophobic groups to the polymer
structure is very strong and cannot be explained only by the small
shift in hydrophobic/hydrophilic balance between the polymers. It is
more likely that it depends on the strength of the hydrophobic
associations holding the polymer network together and restricting the
swelling of the polymer matrix. The formation of one concentrated
phase in equilibrium with one phase depleted in polymer requires the
hydrophobic associations to be strong enough to compensate for the
entropic loss following the formation of the concentrated phase.
32
2.2.4.2 Cloud point of EHEC and HM-EHEC

For EHEC TCp is correlated to MSEO and DSethyl, and TCp increases with
60
point is dramatically influenced by the introduction of hydrophobic

groups on the EHEC polymer. As an example TCp decreased by 15C,
from 65 to 50C, when on average about one out of 120 glucose units
of the unmodified EHEC (0) was grafted with nonyphenol groups
(NP). On a typical HM-EHEC molecule this corresponds to five to ten
hydrophobic groups. As can be seen in Figure 2.22 the shift in TCp is
even stronger when the EHEC is modified with alkyl groups. The
longer the alkyl chain, the more pronounced is the shift in TCp. The
TCp (C)
increasing MSEO and decreases with increasing DSethyl.8 The cloud
40
20
0
(0) (NP) (C12) (C14) (C16)
Figure 2.22. TCp of 1% w/w

solutions of HM-EHEC with
varying length of the
hydrophobic groups. (0)
represents unmodified EHEC
large difference in TCp between the polymers indicates that the

strength of the hydrophobic association is much larger for longer
hydrophobic groups. Also the effect on the solution viscosity of the
polymers reveals large variations in the strength of the associations
(compare section 2.2.2).47
33

(1)
Strauss, U. P. In Polymers in aqueous media; Glass, J. E., Ed.;

American Chemical Society: Washington DC, 1989; Vol. 223, p
317-324.
(2)
Landoll, L. M. In U.S. Patent 4228277; Hercules Inc.: United

States, 1979.
(3)
Landoll, L. M. J. Pol. Sci. 1982, 20, 443-455.
(4)
Sau, A. C.; Landoll, L. M. In Polymers in aqueous media; Glass,

J. E., Ed.; American Chemical Society: Washington DC, 1989;
Vol. 223, p 343-364.
(5)
Glass, E. J. Coatings Technology 2001, 73, 79-98.
(6)
Alami, E.; Rawiso, M.; Isel, F.; Beinert, G.; Binana-Limbele, W.;
Francois, J. In Hydrophilic polymers. Performance with
environmental acceptance; Glass, J. E., Ed.; American
Chemical Society: Washington, DC, 1993; Vol. 248, p 343-362.
(7)
Brandt, L. In Ullmann's Encyclopedia of Industrial Chemistry;

Wiley-VCH Verlag GmbH: Weinheim, Germany, 2002.
(8)
Thuresson, K.; Karlstrm, G.; Lindman, B. J. Phys. Chem.

1995, 99, 3823-3831.
(9)
Thuresson, K.; Soderman, O.; Hansson, P.; Wang, G. J. Phys.

Chem. 1996, 100, 4909-4918.
(10) Cabane, B.; Lindell, K.; Engstrm,

Macromolecules 1996, 29, 3188-3197.
S.;
Lindman,
B.
(11) Wang, K. T.; Iliopoulos, I.; Audebert, R. Polymer Bulletin 1988,

20, 577-582.
(12) Hill, A.; Candau, F.; Selb, J. Progr. Colloid Polym. Sci. 1991,
84, 61-65.
(13) Evans, D. F.; Wennerstrm, H. The colloidal domain. Where
physics, chemistry, biology, and technology meet. Second
Edition; Wiley-VCH:, 1999.
(14) Polymer Handbook; 2 ed.; John Wiley & Sons, Inc: New York,
1975.
(15) Williams, P. A.; Meadows, J.; Phillips, G. O.; Senan, C.
(16) Bock, J.; Siano, D. B.; Valint Jr., P. L.; Pace, S. J. In Polymers
(17) Tanaka, R.; Meadows, J.; Phillips, G. O.; Williams, P. A.
Carbohydrate Polymers 1990, 12, 443-459.
(18) Gelman, R. A. In 1987 International dissolving
Pulps
(19) Aubry, T.; Moan, M. J Rheol 1994, 38, 1681.

(20) Tam, K. C.; Farmer, M. L.; Jenkins, R. D.; Bassett, D. R. J.
Polym. Sci. B: Polym. Phys. 1998, 36, 2275-2290.
34
(21) Schaller, E. Surface Coatings Australia 1985, 6-13.

(22) Semenov, A. N.; Joanny, J.-F.; Khokhlov, A. R. Macromolecules
1995, 28, 1066-1075.
(23) Volpert, E.; Selb, J.; Candeau, F. Macromolecules 1996, 29,
1452-1463.
(24) Selb, J.; Candau, F. In Associative polymers in aqueous media;
Glass, J. E., Ed.; American Chemical Society: Washington DC,
2000; Vol. 765, p 95-108.
(25) Wang, T. K.; Iliopoulos, I.; Audebert, R. In Water-soluble
polymers: synthesis, solution properties and applications;
Shalaby, Ed.; American Chemical Society: Washington D.C.,
1991, p 218-231.
(26) Annable, T.; Buscall, R.; Ettelaie, R.; Whittlestone, D. J. Rheol.
1993, 37, 695-726.
(27) Alami, E.; Almgren, M.; W., B. Macromolecules 1996, 29, 22292243.
(28) Winnik, M. A.; Yekta, A. Current Opinion in Colloid & Interface
Science 1997, 2, 424-436.
Trans. 1997, 90, 3555-3562.
(30) Alami, E.; Almgren, M.; Brown,
Macromolecules 1996, 29, 5026-5035.
W.;
Francois,
J.
(31) Vorobyova, O.; Winnik, M. A. In Associative polymers in

aqueous solution; Glass, J. E., Ed.; American Chemical
(32) Xu, B.; Li, L.; Yekta, A.; Masoumi, Z.; Kanagalingam, S.;
Winnik, M., A.; Zhang, K.; Macdonald, P., M.; Menchen, S.
Langmuir 1997, 13, 2447-2456.
(33) Piculell, L.; Nilsson, S.; Sjstrm, J.; Thuresson, K. In
Assosciatve polymers in aqueous media; Glass, J. E., Ed.;
American Chemical Society: Washington DC, 2000; Vol. 765, p
317-335.
(34) Thuresson, K.; Lindman, B. J. Phys.Chem. 1997, 101, 64606468.
(35) Shaw, K. G.; Leipold, D. P. J. Coatings Technology 1985, 57,
63-72.
(37) Dualeh, A. J.; Steiner, S. A. Macromolecules 1990, 23, 251255.
(38) Magny, B.; Iliopoulos, I.; Audebert, R.; Piculell, L.; Lindman, B.
Progr. Colloid Polym. Sci 1992, 89, 118-121.
(39) Nystrom, B.; Thuresson, K.; Lindman, B. Langmuir 1995, 11,
1994-2002.
Chem. 1998, 102, 7099-7105.
35

P.;

101, 307-318.
(43) Iliopoulos, I. Curr. Opin. Colloid Interface Sci. 1998, 3, 493-498.
(44) Karlstrm, G. J. Phys. Chem. 1985, 89, 4962-4964.
(45) Karlstrm, G.; Carlsson, A.; Lindman, B. J. Phys. Chem. 1990,
94, 5005-5015.
(46) Thuresson, K.; Nilsson, S.; Kjoniksen, A.-L.; Walderhaug, H.;
Lindman, B.; Nystrom, B. J. Phys. Chem. 1999, 103, 14251436.
(47) Thuresson, K.; Joabsson, F. Colloids and Surfaces A:
Physicochem. Eng. Aspects 1999, 151, 513-523.
36
Chapter 3
Inhibition of hydrophobic associations as a
tool to study cross-linking mechanisms
Many studies have tried to investigate how the hydrophobic

modification influences the solution properties of a polymer. Of course
it is possible to get an idea of the strength of the associations of
hydrophobic groups by synthesizing both the hydrophobically
modified polymer and its unmodified analogue. This was the
approach of many of the early studies.1-5 However, if both polymers
are synthesized in separate reactions it is possible that their
structures differ by more than just the hydrophobic modification.
Sometimes this problem can be circumvented by using the
unmodified polymer as starting material in the synthesis. It is likely
that with this approach the HM-P and the parent polymer differ in
molecular weight since an additional reaction step often leads to a
degradation of the polymer backbone. By decoupling the polymer
network it is possible to gradually move in the direction of the
unmodified system. The decoupling can be achieved by changing the
solvent quality or by the addition of a third component (co-solute).
Surfactants and cyclodextrins are examples of co-solutes that
dramatically change the strength and number of the associations.
Different information can be achieved by using the different methods.
By changing the solvent quality or by the addition of an excess of
surfactant all types of hydrophobic associations can be decoupled. It
would therefore be desirable to have a method to specifically
disconnect associations caused by hydrophobic side chains. The
addition of cyclodextrin, on the other hand, offers the possibility to
specifically decouple the associations originating from hydrophobic
side chains. This is for example particularly useful for the evaluation
of HM-EHEC since the associative interactions originate both from
associations of hydrophobic segments of the polymer backbone and
from associations of hydrophobic side groups. The different methods
are discussed in more detail in the following sections.
37
3.1 Inhibition of hydrophobic interactions by

changing solvent quality
Hydrophobically modified polymers have a much stronger tendency to
associate in water than in other (less) polar solvents, e.g. alcohols
and glycols.4,6 This is to be expected since the driving force for
association is to minimize contact between the hydrophobic moieties
of the HM-P and the solvent molecules and this becomes less
important when the polarity of the solvent is reduced. It is also in
agreement with what is found for self-assembly of surfactants.7 Upon
gradual addition of a less polar solvent to an aqueous polymer
solution the intermolecular hydrophobic associations are broken since
it becomes less important to avoid the contact between the
hydrophobic tails and the solvent. In Figure 3.1 the viscosity of 1 %
w/w HM-(C14)-EHEC and 1 % w/w HM-(NP)-EHEC solutions are
given as a function of the concentration of diethylene glycol
monobutylether (BDG), cBDG, in the solvent. Since the viscosity of the
solvent changes with changing ratio between BDG and water the
viscosity is presented as the relative viscosity, rel = / solvent, where
solvent in each point is the viscosity of the solvent at that specific

BDG/water ratio.
10
rel
HM-(C14)-EHEC
10
Figure 3.1. The influence of

cBDG on the viscosity of 1%w/w
solutions of HM-EHEC.
HM-(NP)-EHEC
1
10
10
15
20
25
30
cBDG (%w/w)
A saturation level where rel is independent of cBDG is reached at

about 15 % w/w BDG. Above that BDG concentration the relative
38
viscosity of the polymer solution is constant. In paper III a BDG/water

ratio of 20 / 80 of was used, and in the following text viscosity
measured in such a solution is referred to as BDG. The fact that both
HM-(C14)-EHEC and HM-(NP)-EHEC have a BDG that is almost the
same as BDG for the corresponding unmodified EHEC indicates that
the hydrophobic interactions from the hydrophobic modification are
totally decoupled. A way to represent the influence by the
hydrophobic interaction is QBDG which is the ratio between the value of
the (Newtonian) viscosity in water to that observed in water/BDG,
BDG (equation (3.1)).
QBDG =
BDG
(3.1)
QBDG can be regarded as a phenomenological measurement of the

influence of hydrophobic associations on the viscosity in the aqueous
solution. In this way different polymer samples (regarding chemical
structure of the hydrophobic tails, modification degree, modification
pattern etc.) can be ranked. With this method it is evident that the
unmodified EHEC also has a contribution to the viscosity originating
from hydrophobic interactions. This was observed as a small but
significant QBDG of about 1.2 (Figure 3.2). Since this polymer has no
hydrophobic grafts the origin of the interactions has to be sought
elsewhere. As described in section 2.2.2 the uneven distribution of
ethyl substituents results in hydrophobic segments along the
backbone and it is likely that the blocky structure causes hydrophobic
associations. The low QBDG of 1.2 indicates that the strength of these
interactions is much weaker than those given by the hydrophobic
grafts, provided that the length of the hydrophobic groups is C12 or
longer. Since all hydrophobic associations are disconnected and
chain entanglements are the only remaining interpolymer cross-links
in the solution, BDG can be used as a measure of the chain
entanglement contribution to the viscosity.
39
QBDG
Figure 3.2 shows the influence of the length of the hydrophobic tails
16
14
12
10
8
6
4
2
0
on the solution viscosity and QBDG of some HM-EHEC:s. The strength

of the association of the hydrophobic grafts is strongly dependent of
the length of the hydrophobic groups as can be seen from the QBDG.
The values of QBDG for HM-EHEC grafted with short hydrophobic
(0)
(NP)
(C12)
(C14)
Figure 3.2. QBDG for

HM-EHEC with different
hydrophobic groups. From left
to right; unmodified EHEC (0),
HM-EHEC modified with
nonylphenol groups (NP), with
C12 and C14 groups.
groups (C12 or NP) are 2.5 and 3.4 respectively whereas it increases
dramatically for the longer hydrophobic groups (C14).
With this method it is not possible to separate contributions to the
viscosity from associative interactions of different origin, since it was
found that the contribution from grafted hydrophobic groups as well
as the contribution from a hydrophobic polymer backbone was
affected by the addition of BDG.

addition of surfactant
In section 2.2.3 is described the influence of surfactant on the
associative behavior of HM-polymers. At high concentration of
surfactant the number of micelles exceeds the number of hydrophobic
groups of the polymer which means that on average each micelle
contains only one hydrophobic group from a HM-polymer, as
illustrated in Figure 3.3. The result is that the hydrophobic
associations between HM-polymers are decoupled and the polymer
network is disconnected. This can be detected as a decreased
solution viscosity and increased self-diffusion of the polymer
molecules.4,8-23 The viscosity and self-diffusion in a solution of a
HM-polymer at excess surfactant are expected to attain the same
values as for a solution of the corresponding unmodified polymer
(provided that the molecular weight is the same). This has for
Figure 3.3. Schematic picture
of HM-P at high csurf where the
associations between
hydrophobic side chains are
decoupled by surfactant.
example been illustrated for HM-HEC and unmodified HEC.19

As described in section 2.2.2 the hydrophobic associations of
HM-EHEC consist of associations between hydrophobic side chains
as well as interaction of hydrophobic segments of the polymer
backbone. Thuresson et al have shown that not only HM-EHEC but
40
also unmodified EHEC is affected by the addition of surfactant.5 The

suggested explanation is that the surfactant associates both to the
hydrophobic segments of the main chain and to hydrophobic side
chains. At excess surfactant the viscosity of solutions of the
hydrophobically modified polymer and the unmodified parent polymer
attains the same value. In analogy with the effect of addition of BDG,
discussed above, the observation that the viscosity for a solution of
the parent polymer is lower compared to when no surfactant is added
indicates that the associations from hydrophobic segments of the
backbone are decoupled by addition of surfactants. Surfactants
cannot be used to selectively decouple any of the types of
hydrophobic associations. Similarly to the solvent approach (section
3.1) it is therefore not possible to distinguish between the contribution
from hydrophobic associations of the polymer backbone and the
contribution from associations of hydrophobic side chains by this
method.

cyclodextrin
Cyclodextrin (CD) is a cyclic molecule with a hydrophobic cavity
OH
(Figure 3.4 and 3.5). CD binds selectively to hydrophobic molecules
OH
decouple the association caused by hydrophobic groups grafted to

the polymer backbone. The deactivation of hydrophobic associations
by CD gives unique information about the association mechanisms of
HM-polymers that cannot be achieved by deactivation at excess
OH HO
OH
O
HO
HM-polymers the CD molecules bind primarily to hydrophobic side- or

Therefore addition of CD provides a unique possibility to specifically
or parts of molecules that fit into the cavity. In aqueous solutions of

end-groups and not to hydrophobic segments of the backbone.
OH
OH
O
HO
OH
O
HO
HO
OH
O
HO
OH
HO
O
OH
O
HO
Figure 3.4. Chemical structure

of -cyclodextrin.
surfactant or by changed solvent quality.
41
3.3.1 Structure and properties of cyclodextrin

Cyclodextrins (CD:s) are cyclic oligomers of -D-glucose. Three
different CD:s, denoted -, -, or - cyclodextrin, are naturally
occuring and they consist of 6, 7 or 8 glucose units respecively.24,25
They are synthesized by enzymatic degradation of starch. Their
chemical structure is very rigid and the three-dimensional shape can
be described as a shallow truncated cone with a cavity in the center
Hydrophobic cavity
H
O
H
C
C O
H
O
O
H C
C
H
C
H C H
O
H

representation of the geometry
of a cyclodextrin molecule.
extending from one end to the other (Figure 3.5).24,25 The exterior of
the cone is hydrophilic since all the hydroxyl groups of the AHGs are
located there while the cavity has non-polar properties. The size of
the cavity varies depending on whether it is -, -, or - cyclodextrin.
Some useful physical properties of the different cyclodextrins are
listed in table 3.1.
By substitution, the physical properties of the cyclodextrins can be
changed. Substitution with methyl- (M-) or hydroxypropoxyl (HP-)
groups has been used to increase the solubility of CD in organic
solvents. As a natural consequence of the location of hydroxyl groups
the substituents will be located on the rims of the molecule, resulting
in an increase in the height of the torus. More surprising is that the
diameter of the cavity is reduced by the derivatization.26 The overall
result of the derivatization is that the cavity volume increases. Acetyl
(Ac-) is used to increase the solubility of -cyclodextrin in water,
which in its natural form has a quite poor aqueous solubility.
Somewhat unexpected it is found that also slighly hydrophobic
substituents like methyl and hydroxypropyl increase the water
solubility (Table 3.2.). A reduced possibility to form crystalline
structures is the most probable reason (compare with native cellulose
that is insoluble while methyl cellulose and hydroxypropyl cellulose
are soluble).27
42
Table 3.1. Properties of -, -, and - Cyclodextrin.

Cyclodextrin
Per-O-methyl-
Per-O-methyl-
Number of
glucose units
6
7
8
6
7
Molecular
weight
972
1135
1297
1224
1429
Cavity
diameter ()
4.7 5.3
6.0 6.5
7.5 8.3
4.2
5.8
Torus hight
()
8
8
11
11
Data obtained from26 and24

Table 3.2. Solubility in water at 25C for -, -, and - cyclodextrin substituted with methyl (M-),
hydroxypropyl (HP-), and acetyl- (Ac-) groups
Cyclodextrin
M-M-HP-Ac-M-HP--
a)
b)
Degree of
substitution
1.8
1.8
0.75
1
1.8
0.6
Solubility in
water at 25C
(g/100ml)
14a)
2a)
23a)
388b)
300b)
200b)
220b)
330b)
180b)
Data obtained from24

Data supplied by Dr Stephan Neuman, Wacker-Chemie GmbH, Germany
3.3.2 Formation of inclusion complex between lipophilic guest

molecules and cyclodextrin
In an aqueous solution a less polar guest molecule readily substitutes
the polar water molecules inside the cavity provided that the unpolar
molecule has the correct dimensions to fit within the cavity (Figure
3.5). This hydrophobic attraction drives the formation of an inclusion
complex. The complex formation has frequently been studied and
surfactant/cyclodextrin systems especially have received a lot of
attention. Various methods, e.g. calorimetry or surfactant selective
electrodes, have been used to determine complex constants.28-34
The complex constants for the formation of an inclusion complex
43
between -CD or -CD and some commonly used surfactants are

listed in Table 3.3.

representation of the inclusion
of a lipophilic group into the
cavity of a cyclodextrin
molecule. The filled balls
represent water molecules.
The changed shape of the cavity, as a result of the derivatization of

the CD, influences the ability for the modified CD:s to form a complex
with another substance. Therefore the complex constants for the
modified CD:s differ from the constants from the corresponding
unmodified CD:s.26 An increased length of the cavity often results in
a stronger tendency for complex formation. On the other hand a
reduced cavity diameter from the derivatization results in a reduced
ability to form complex with bulky hydrophobes e.g. aromatic
groups.26
The fact that lipophilic molecules can hide inside the cavity of an
otherwise
hydrophilic
and
water-soluble
molecule
has
given
cyclodextrins many technical applications. One obvious application is

to enhance water solubility of poorly soluble substances but it has
also been used to mask unpleasant odors and tastes and to reduce
the vapour pressure of volatile organic compounds dissolved in
water.24
44
Table 3.3. Complex formation constant, K1 (mM-1), for -CD and -CD in combination with sodium
dodecyl sulphate (SDS), sodium tetradecyl sulphate (STS), sodium hexadecyl sulphate (SHS), dodecyl
trimethyl ammonium bromide (DTAB), tetradecyl trimethyl ammonium bromide (TTAB) and cetyl
trimethyl ammonium bromide (CTAB).
Surfactant
Reference
K1
-CD
-CD
DTAB
23.7
30
TTAB
61.0
39.8
30
CTAB
99.2
67.7
30
SDS
25.6
33
STS
48.2
33
SHS
53.3
33
3.3.3 Cyclodextrin and HM-P

The hydrophobic tails of an HM-polymer in an aqueous solution can
form inclusion complexes with added cyclodextrin molecules. This
leads to a disruption of the physical bonds holding the three
dimensional polymer network together (Figure 3.6). This can be
detected as a reduction in the viscosity of the polymer solution. This
is similar to the effect of the addition of excess surfactant or by
changes of the solvent quality to a less polar system as discussed in
section 3.1 and 3.2. Eisenhart and Lau and their coworkers were the
first to report the viscosity reducing effect by the addition of
cyclodextrin in two patents.35,36 They used the inhibition of
hydrophobic
interactions
to
reduce
the
viscosity
in
highly
concentrated solutions of associative thickeners. The reduced

viscosity is desired during production and handling (pumping etc.) of
the thickener or at other occasions when the polymer is present at
high concentration and therefore gives very high viscosity. The
complexation is reversible and by addition of e.g. a surfactant with
higher affinity to the cyclodextrin the thickening effect can be
regained.
In some later papers it has been reported that the degree of
association in solutions of hydrophobically modified polymer can be
Figure 3.6. A schematic

representation of the
disruption of the polymer
network following the complex
formation between
cyclodextrin and polymer
hydrophobic tails.
45
controlled by addition of cyclodextrin.37-39 The viscosity is reduced

with increasing CD concentration (cCD) and levels off at a CD/HM-P
ratio where all hydrophobic interactions are inhibited.37,38 At excess
CD the HM-P molecules are expected to be unable to associate to
each other. This can be used if the molecular weight of the
HM-polymers should be determined by techniques such as light
scattering methods or by gel permeation chromatography (GPC).
Islam and coworkers have demonstrated how the use of cyclodextrins
simplifies
the
hydrophobically
determination
modified
of
the
polyacrylate
molecular
by
weight
preventing
of
self-
association.40
3.3.3.1 Cyclodextrin and HM-EHEC

In paper IV we have studied the formation of an inclusion complex in
aqueous solution between cyclodextrin and the hydrophobic groups
grafted on EHEC. We found, in agreement with earlier studies37,38
that in the region where cCD is low compared to the total concentration
of polymer hydrophobic groups in the solution (chydrophobe), the viscosity
decreases with increasing CD concentration in the solution (Figure
3.7). At cCD > chydrophobe the viscosity levels off and attains a constant
value. Three different cyclodextrins, methyl--CD, -CD and methyl-CD, were used in combination with two HM-EHEC samples with
either nonyl phenyl (HM-(NP)-EHEC) or tetradecyl (HM-(C14)-EHEC)
hydrophobic groups. By representing the complex formation within a
Langmuir adsorption model and assuming that 1:1 nut and bolt
complexes are formed the concentration of adsorption sites, B, and
the complex constant, K, can be determined. A detailed description of
how equation (3.2) is derived is found in the appendix of paper IV.
B + cCD + 1 / K
( B + cCD + 1 / K ) 2
BcCD

2
4
= 1
0
B
46
(3.2)
The viscosity without CD is represented by 0 and the viscosity at

excess CD by . From fitting equation (3.2) to our experimental data
points ( ( ) ( 0 ) vs. cCD) with K and B as fitting parameters, B
and K could be determined.
1.0
( )/(0 )
0.8
8
0.6
0.4
0.2
0.0
0
Figure 3.7. Relative viscosity

as a function of the
concentration of methylated-cyclodextrin, cCD, of 1% w/w
solutions of HM-EHEC. Open
symbols represent HM-(NP)EHEC and filled symbols
represent HM-(C14)-EHEC.
The full lines represent a fit of
Equation (3.2) to the data.
cCD (mmolal)
In table 3.4 it can be seen that the complex constant, K, is very much
influenced both by the shape of the polymer hydrophobic group and
the structure of the CD. For the HM-EHEC with C14-hydrophobic
groups the highest values of K are found for the methylated
cyclodextrins. As described in chapter 3.3.1 methylation of a CD
makes the cavity deeper and narrower. This indicates that the long
and relatively thin C14 hydrophobe fits better into the longer and more
narrow cavity of a methylated cyclodextrin. The values of K are
slightly lower but in the same range as those found for the complex
formation between CD and surfactants containing C14 alkyl groups
(see table 3.3.). This is reasonable since the backbone of HM-EHEC
is an extremely large and bulky head-group that is likely to oppose
the complex formation.
Compared to the HM-(C14)-EHEC the values of K are in general lower
for the HM-(NP)-EHEC. On the other hand it seems that the more
bulky nonyl phenol group fits best into the wider cavity of the -CD as
indicated by the highest K for -CD in combination with HM-(NP)EHEC.
47
Table 3.4. Complex formation constant, K, and concentration of adsorption sites, B, obtained
by fitting Equation (3.2) to the experimental data. is the viscosity at excess CD and 0 is the
viscosity when no CD is added.
HM-EHEC
NP
C14
CD
(mmolal-1)
(mmolal)
(mPa s)
(mPa s)
M-
2.7
0.27
440
115
22.6
0.25
440
105
M-
17.0
0.26
440
105
M-
44.0
0.32
1439
50
11.2
0.31
1439
80
M-
66.0
0.30
1439
80
M-
45
45
For HM-(C14)-EHEC the concentration of binding sites, B, obtained

from the model (table 3.4) almost perfectly matches the concentration
of hydrophobic tails (chydrophobe = 0.29 mmolal) obtained from chemical
analysis. This gives an indication that all hydrophobic groups are
potentially important for the formation of the polymer network and that
all hydrophobic tails can form a complex with CD. With NP the
situation is different, and it can be concluded that the values of B are
lower than the total concentration of hydrophobic groups for HM-(NP)EHEC. Judging from the values of B and (chydrophobe = 0.28) 5 to 10% of
the hydrophobic groups are not available for complex formation with
CD. The nonylphenol used for the synthesis of HM-(NP)-EHEC is of
technical quality that contains both mono- and di-nonyl phenol. In
mono-nonyl phenol the nonyl group can be situated either in ortho
position or para position on the phenol ring. It is possible that the 5 to
10% that is not available for complex formation consists of di-nonyl
phenol and ortho-nonyl phenol. They have the most bulky structure
and are therefore more difficult to fit into the cavity of CD.
Considering the size of the hydrophobic segments of the backbone of
HM-EHEC it is reasonable to assume that CD is not capable of
decoupling the associations of such segments. The fact that the
viscosity of a solution of the unmodified EHEC was not affected at all
by addition of CD suggests this.
48
The viscosity at excess CD, , also tells something about the ability
of the CD to decouple the polymeric network. For M--CD the value
of is almost equal to the viscosity for a solution with the same
polymer concentration of the unmodified EHEC (HM-(0)-EHEC) with
the same molecular weight. This is an indication that all associations
that stem from the grafted hydrophobic groups are disconnected. We
note that is higher than BDG (Figure 3.1) where also associations
between hydrophobic segments are disconnected. For the other
combinations is somewhat higher than the viscosity of the
unmodified polymer and especially for the HM-(NP)-EHEC this is
obvious. The reason is that all polymer hydrophobic tails are not
available for complex formation with CD in these cases. From the
values of and by using equation (3.2) the fraction of hydrophobes
in the solution of HM-(NP)-EHEC that is not available for complex
formation can be estimated to be about 16%, which is quite close to
what was found above when B was compared to chydrophobe.
CD offers a selective way of decoupling the associations of
hydrophobic side chains, provided that the hydrophobic side chains
have a structure that fits into the cavity, but leaving the associations
from hydrophobic patches of the backbone intact. The quotient, QCD,
between and 0 (equation 3.3) can be used as a phenomenological measurement of the contribution to the viscosity caused by
association between hydrophobic side chains.
QCD =
(3.3)
3.3.3.2 Cyclodextrin and HM-PEG

In analogy with the results for HM-EHEC the addition of cyclodextrin
to an aqueous solution of HM-PEG results in a degradation of the
polymer network as indicated by a reduction of the solution viscosity
and an increased mean self-diffusion coefficient of HM-PEG (DHM-PEG)
(Figure 3.8). In papers V and VI we have studied the degradation of
the polymer network in the HM-PEG system in the polymer
49
concentration range 3 to 10% w/w. We adopted the same model as

we used for the HM-EHEC CD system (equation 3.2).
Figure 3.8. Relative viscosity

(filled symbols) and mean selfdiffusion coefficient (open
symbols) of a 3% w/w solution
of HM-PEG as a function of
the concentration of
methylated--cyclodextrin, cCD.
/0
0.6
-11
10
-12
10
-13
0.8
10
0.4
0.2
D (m /s)
1.0
0.0
0
cCD (mmolal)
Low concentration of CD
Figure 3.9 shows that the viscosity decreases dramatically with the
addition of methylated -cyclodextrin (M--CD) to the HM-PEG
solution. The change is most pronounced at small additions of CD,
below 1 mmolal. In an attempt to determine the number of binding
sites, B, in the same way as described for the CD /HM-EHEC system
in section 3.3.4 equation (3.2) was fitted to the viscosity data points
( / 0 vs. cCD). The best representation of the experimental results
was obtained for B = 0.4 mmolal which constitutes only 10% of the
total number of hydrophobic groups. The results show that
deactivation of the first few hydrophobic associations has a much
stronger influence on the viscosity than would be expected if all
associations were equally important for the viscosity. This is
supported by the measurements of DHM-PEG also included in Figure
3.9. The increase in DHM-PEG is steep at cCD < 0.5 mmolal and at higher
concentrations it levels off. This shows that it is enough to terminate
about 10% of the hydrophobic tails to change the viscosity and DHMPEG
almost to the levels achieved at excess CD where the network is
totally decoupled.
50
1.0
10
-11
10
-12
10
-13
/0
0.6
0.4
0.2
D (m /s)
0.8
0.0
0
B1
Figure 3.9. Relative viscosity,

/0, (filled symbols) and mean
self-diffusion coefficient,
DHM-PEG, (open symbols) as a
function of the concentration
of methylated--cyclodextrin,
cCD, for a 3% w/w solution of
HM-PEG. The full line
represents a fit of Equation
(3.2) to the relative viscosity
data. B1 was obtained by
extrapolation to /0=0 from
the behavior at low cCD
cCD (mmolal)
To explain this we must go back to the model of the network

formation of HEUR thickeners in aqueous solution (section 2.2.1). 3%
w/w HM-PEG is in the region where the HM-PEG is expected to be
present in a percolated network built of clusters of flower micelles
(Figure 2.13). At this HM-PEG concentration the solution is expected
to be inhomogeneous with rather large concentration fluctuations
where inter-micellar links inside the clusters are numerous while the
polymers that connect micelles located in different clusters are rare. It
is likely that the polymers that connect different clusters give a
relatively more important contribution to connectivity of the network
and therefore are more important to the viscosity and DHM-PEG than the
polymers involved in associations inside the clusters. The dramatic
change in / 0
and DHM-PEG can be understood if primarily
hydrophobic associations responsible for connecting different clusters

are deactivated at low cCD.
Viscosity measurements show that B is virtually independent of the
polymer concentration in the concentration range between 3 and 10
%w/w (B = 0.45 mmolal at 3% w/w, B = 0.42 mmolal at 5% w/w and B
= 0.53 mmolal at 10% w/w). This indicates that the number of
linkages between the clusters stays almost unchanged with
increasing polymer concentration whereas the clusters grow in size.
This has been suggested before by Alami et al.41
51
Intermediate concentrations of CD
At intermediate concentrations, where B < cCD < chydrophobe, a new region
appears, as can be seen in Figure 3.10. The changes in / 0 and in
DHM-PEG are much less dramatic in this region. The break-point
between the two regions in the viscosity curve almost coincides with
what is found from the self-diffusion measurements. It is reasonable
that it is the size of the decoupled clusters that influences the
viscosity of the solution and DHM-PEG in this region indicating that the
size of the clusters decreases with increasing concentration of CD.
This indicates that it is the inter-micellar linkages inside the clusters
that are disconnected leading to a degradation of the clusters into
separate micelles and further into separate polymer molecules
bearing a CD molecule at each end. The distribution in self-diffusion
coefficients, , reflects the size distribution of the polymer
aggregates.42,43 As can be seen in Figure 3.12, decreases with
increasing cCD. This also indicates that the clusters are degraded
since the clusters are expected to have a broad distribution in sizes
while the size of the micelles is rather uniform.
0.6
-11
10
-12
10
-13
0.4
0.2
0.0
0
52
10
0.8
/0
Figure 3.10. Relative viscosity,

/0, (filled symbols) and mean
self-diffusion coefficient,
DHM-PEG, (open symbols) as a
function of the concentration
of methylated--cyclodextrin,
cCD, for a 3%w/w solution of
HM-PEG in the intermediate
region of cCD.
1
B(3%)
cCD (mmolal)
D (m /s)
1.0
High concentration of CD
In the region cCD > chydrophobe both the viscosity and DHM-PEG are
expected to be independent of cCD and to be on same level as that
found in a solution containing the corresponding unmodified PEG with
similar
molecular
10
weight.
In
fact
this
is
what
-9
we
found.
10
-10
10
-11
10
-12
10
-13
DPEG for unmodified PEG
DHM-PEG, DCD (m /s)
DCD for free CD
0
0
10
15
20
25
Figure 3.11. Mean selfdiffusion coefficients for

HM-PEG, DHM-PEG; (open
circles) and for M--CD, DCD,
(triangles) and the distribution
in DHM-PEG, , (filled circles) as
a function of CD concentration
in 3%w/w solution of HM-PEG.
The lower dashed line
represents the mean selfdiffusion coefficient for
unmodified PEG (DPEG) (mw=
20000g/mol) in 1% w/w
solution of PEG. The upper
dashed line represents DCD
when no HM-PEG or PEG is
present.
30
cCD (mmolal)
Since the self-diffusion at high cCD does not attain a plateau value
until cCD is above twice B this indicates that more than one
cyclodextrin molecule can bind to each hydrophobic group. Another
explanation could be that the complex formation is not quantitative
and free cyclodextrin is present in the solution in this region.
The self diffusion of M--CD (DCD) has also been measured (Figure
3.11). DCD changed moderately with cCD and reached a plateau at
about 10 to 15 mmolal. The value of DCD at the plateau was low
compared to DCD in a solution of CD where no HM-PEG was present.
Experiments where HM-PEG was substituted by unmodified PEG
showed that interactions between the PEG chain and CD are of minor
importance. This indicates that the reduction of DCD at high cCD is
mainly caused by obstruction effects.
The fraction of CD that is bound to HM-PEG (Pb) can be determined
by the use of equation 3.4 where DCD,obs is the observed self-diffusion
53
of CD at the actual cCD and DCD,free is the self-diffusion of CD at excess

CD.
DCD ,obs = Pb DHM PEG + (1 Pb ) DCD , free
(3.4)
The fraction of bound CD decreases with increasing cCD. A calculation

at cCD = 10.7 mmolal gave the result that the average number of
bound CD per hydrophobic group (CD/hydrophobe) was 1.4 which
supports that more than one CD molecule can bind to each
hydrophobic group. In line with this Olson et al have shown by NMRmeasurements that two or even more -CD molecules can bind to a
C12-hydrophobic group attached to a PEG chain.44
A model for the degradation of the HM-PEG network
From the results presented above a model for the degradation of the
polymer network in the HM-PEG solution is suggested (Figure 3.12).
In the region cCD < B the CD primarily breaks the linkages between
different clusters. In the region B < cCD < chydrophobe where the change in
viscosity is less pronounced the viscosity is mainly influenced by the
size of the clusters. At high CD concentration, cCD > chydrophobe, the
HM-PEG appears mainly as small aggregates or as individual
molecules with the hydrophobic groups hidden inside the interior of

representation of the suggested
model for the degradation of
HM-PEG network with
cyclodextrin
/0
the CD molecules.
cCD
54

(1)
Wang, K. T.; Iliopoulos, I.; Audebert, R. Polymer Bulletin 1988,

20, 577-582.
(2)
Valint, J., P.L.; Bock, J. Macromolecules 1988, 21, 175-179.
(3)
Bock, J.; Siano, D. B.; Valint Jr., P. L.; Pace, S. J. In Polymers

(4)
Williams, P. A.; Meadows, J.; Phillips, G. O.; Senan, C.

(5)
Thuresson, K.; Lindman, B. J. Phys. Chem. 1997, 101, 64606468
(6)
Gelman, R. A.; Barth, H. G. Adv. Chem. Ser. 1986, 213, 101110.
(7)
Jnsson, B.; Lindman, B.; Holmberg, K.; Kronberg, B.

Surfactants and polymers in aqueous solution; John Wiley &
Sons Ltd: Chichester, England, 1998.
(8)
Gelman, R. A. In 1987 International dissolving

(9)
Magny, B.; Iliopoulos, I.; Audebert, R.; Piculell, L.; Lindman, B.

Progr. Colloid Polym. Sci 1992, 89, 118-121.
Pulps

P.;
(12) Loyen, K.; Iliopoulos, I.; Olsson, U.; Audebert, R. Progr. Colloid
Polym. Sci. 1995, 98, 42-46.
101, 307-318.
(14) Aubry, T.; Moan, M. J. Rheol. 1996, 40, 441-448.
(15) Piculell, L.; Guillemet, F.; Thuresson, K.; Shubin, V.; Ericsson,
O. Adv. Colloid Interface Sci. 1996, 63, 1-21.
Trans. 1997, 90, 3555-3562.
(17) Macdonald, P., M. In Polymeric materials: Sci. Eng. Spring
meeting1997; ACS, Ed. San Francisco, 1997; Vol. 76, p 27-28.
(18) Panmai, S.; Prud'homme, R., K.; Peiffer, D., G.; Jockusch, S.;
Turro, N., J. Polym. Mater. Sci. Engin. 1998, 79, 419-420.
Chem. 1998, 102, 7099-7105.
(20) Olesen, K. R.; Bassett, D. R.; Wilkerson, C. L. Progress Organic
Coatings 1998, 35, 161-170.
(21) Jimnez-Rigaldo, E.; Selb, J.; Candau, F. Langmuir 2000, 16,
8611-8621.
55
(22) Chronakis, I. S.; Alexandridis, P. Marcomolecules 2001, 34,

5005-5018.
(23) Steffenhagen, M. J.; Xing, L.-L.; Elliott, P. T.; Wetzel, W. H.;
Glass, J. E. Polym. Mater. Sci. Engin. 2001, 85, 217-218.
(24) Loftsson, T.; Brewster, M. E. J. Pharm. Sci. 1996, 85, 10171025.
(25) Connors, K. A. Chem. Rev. 1997, 97, 1325-1357.
(26) Immel, S.; Lichtenthaler, F. W. Starch/Strke 1996, 48, 225232.
(27) Wentz, G. Angew. Chem. Int. Ed. Engl. 1994, 33, 803-822.
(28) Amiel, C.; Sebille, B. J. Inclusion Phenomena Molecular
Recognition in Chem. 1996, 25, 61-67.
(29) Mwakibete, H.; Bloor, D. M.; Wyn-Jones, E. Langmuir 1994, 10,
3328-3331.
(30) Mwakibete, H.; Bloor, D. M.; Wyn-Jones, E.; Holzwarth, J. F.
Langmuir 1995, 11, 57-60.
(31) Junquera, E.; Tardajos, G.; Aicart, E. Langmuir 1993, 9, 12131219.
(32) Funasaki, N.; Yodo, H.; Hada, S.; Neya, S. Bull. Chem. Soc.
Jpn. 1992, 65, 1323-1330.
(33) Park, J. W.; Song, H. J. J. Phys. Chem. 1989, 93, 6454-6458.
(34) Sasaki, K. J.; Christian, S. D.; Tucker, E. E. Fluid Phase
Equilibria 1989, 49, 281-289.
(35) Eisenhart, E. K.; Johnson, E. A. In U.S. Patent 5137571; Rohm
and Haas: United States, 1992.
(36) Lau, W.; Shah, V. M. In U.S. Patent 5376709; Rohm and Haas:
United States, 1994.
(37) Akiyoshi, K.; Sasaki, Y.; Kuroda, K.; Sunamoto, J. Chemistry
Letters 1998, 93-94.
(38) Zhang, H.; Hogen-Esch, T. E.; Boschet, F.; Margaillan, A.
Langmuir 1998, 14, 4972-4977.
(39) Gupta, R. K.; Tam, K. C.; Ong, S. H.; Jenkins, R. D. In XIIIth
International Congress on Rheology Cambrige, UK, 2000, p
335-337.
(40) Islam, M. F.; Jenkins, R. D.; Bassett, D. L.; Lau, W.; Ou-Yang,
H. D. Macromolecules 2000, 33, 2480-2485.
(41) Alami, E.; Almgren, M.; W., B. Macromolecules 1996, 29, 22292243.
(42) Nydn, M.; Sderman, O. Macromolecules 1998, 31 (15), 49905002.
(43) Nydn, M.; Sderman, O.; Karlstrm, G. Macromolecules 1999,
32, 127-135.
(44) Olson, K.; Chen, Y.; Baker, G. L. J. Polym. Sci. Part A: Polym.
Chem. 2001, 39, 2731-2739.
56
Main conclusions
One intention of this thesis has been to support the development of improved associative
thickeners for water borne paint and it is my opinion that novel information has been obtained.
It has been shown that the viscosity of HM-PEG solutions as a function of polymer
concentration passes via a maximum. At concentrations above 50% w/w the viscosity decreases
considerably. This was referred to a gradual transition from a state containing micelle-like
structures to a more meltlike state (Paper I). This is important when the goal is to have high
concentration of polymer while keeping the viscosity moderate, and may be utilized to minimize
handling and transportation costs of the product.
The dynamics, and the strength, of hydrophobic associations of hydrophobically modified
polymers in aqueous solution are very much influenced by the length of the hydrophobic
groups. Longer hydrophobic groups give, due to slower dynamics and increased relaxation
times, an increased viscosity. When formulated in a paint a HM-polymer with long hydrophobic
groups gives a more elastic consistency compared to when a HM-polymer with shorter
hydrophobic groups is used. (Paper III)
In an aqueous solution a cyclodextrin (CD) molecule can form an inclusion complex with a
hydrophobic group on a HM-polymer. This prevents the hydrophobic group from associating
with other hydrophobic groups, and it leads to a degradation of the physically cross-linked
polymer network. This can be detected as a reduction of the viscosity. At excess CD the
viscosity attains the same value as for a solution of the unmodified polymer with the same
molecular weight. This can be used to deduce the part of the total thickening effect that has its
origin in associations of hydrophobic side chains (Paper IV). This observation has already been
implemented in analysis methods for quality control in the production of HM-EHEC.
In a HM-PEG solution it is enough to terminate only a small fraction of the total amount of
associative linkages to reduce the viscosity almost to the same level as that for a solution of an
unmodified PEG. The results were confirmed by self-diffusion measurements. The changes in
viscosity and self diffusion are for instance much more dramatic compared to what can be
observed when surfactant is added. The suggested interpretation is that it is primarily
hydrophobic associations involved in connecting different clusters of micelles that are
disconnected (Paper V and VI). These results have supplied new information that can be useful
for the understanding of the thickening mechanisms of HM-PEG, both in water solution and in
more complicated systems like a paint.
57
Populrvetenskaplig sammanfattning
Tre begrepp som r viktiga fr denna avhandling r viskositet,

polymer och hydrofob grupp. Ett materials viskositet r ett mtt p hur
trgt eller hur ltt materialet flyter. Lg viskositet betyder att materialet
flyter ltt medan hg viskositet betyder att det flyter trgt. En polymer
r en stor molekyl, som bildas genom kemisk reaktion dr sm
molekyler, monomerer, kopplas samman till en mycket strre
kedjemolekyl. Vissa polymerer r lsliga i vatten och kan anvndas
som frtjockare fr vattenbaserade system, d.v.s. att de hjer
viskositeten hos vattenlsningen. Begreppet hydrofob grupp antyder
att den inte tycker om vatten. (hydro- r ett frled som anger att ngot
innehller eller har samband med vatten och -fob kommer av phobos
som p grekiska betyder 'fruktan', 'skrck'.) I sjlva verket r det s
att det r vattenmolekylerna som hellre omger sig med andra vattenmolekyler n att komma i kontakt med den hydrofoba gruppen. Fr att
minimera kontakten med vatten sker sig den hydrofoba gruppen till
andra hydrofoba grupper i lsningen. Man sger att de hydrofoba
grupperna associerar till varandra.
Figur 1. Schematisk bild av
polymermolekyler som
trasslar in sig i varandra
En bra bild fr att frst hur frtjockningen med polymerer gr till r

en tallrik spagetti. Trdarna av spagetti trasslar in sig i varandra och
det r svrt att rra runt med gaffeln. Polymermolekylerna i en lsning
upptrder p samma stt. De r lnga trdar som trasslar in sig i
varandra och hindrar varandra frn att rra sig vilket resulterar i en
frhjd viskositet (Figur 1). Om man skr spagettin i mindre bitar gr
det lttare att rra omkring med gaffeln. P samma stt r det med
polymerlsningar. Korta polymermolekyler (lg molekylvikt) frtjockar
mindre n lnga polymermolekyler.
I en hydrofobmodifierad polymer (HM-polymer) har en liten mngd
hydrofoba grupper reagerats fast lngs polymerkedjan. De hydrofoba

polymermolekyler med
hydrofoba grupper som
associerar till varandra.
58
grupperna associerar till varandra och ger tvrbindningar mellan

polymerkedjorna (Figur 2). Det betyder att alla polymerkedjorna
hnger ihop i ett enda stort ntverk. Resultatet blir en avsevrd
frhjning av viskositeten. Stora hydrofoba grupper ger starkare

tvrbindningar n sm grupper och drfr hgre viskositet. I liknelsen
med spagetti kan man sga att de hydrofoba grupperna r som riven
ost som klistrar ihop spagettin och gr det nnu svrare att rra runt.
Vattenlsningar frtjockade med polymerer r vanliga i vrt dagliga
liv. Ett exempel r schampo, som r en vattenlsning som br ha hg
viskositet. Om den inte hade det skulle den rinna ut mellan fingrarna
nr den hlldes ur flaskan och ner i handen. Andra exempel kan man
hmta frn matlagningen. Strkelse frn potatis eller majs, anvnds
fr att reda (frtjocka) sser och gelatin anvnds i mnga efterrtter
fr att ge dem dess konsistens.
Vattenbaserad mlarfrg r ytterligare ett exempel p en vattenlsning som mste frtjockas fr att den skall uppfra sig som vi vill. I
Figur 3. Frgen rollas p en

svartvitrutig panel nr
tckfrmgan skall bedmas.
en frg med fr lg viskositet sjunker alla partiklar snabbt till botten p

burken och nr man mlar kan man bara ta lite frg i penseln om inte
frgen skall droppa. Fr att frgen skall f rtt viskositet tillstts
vattenlsliga polymerer.
Polymerer med hg molekylvikt r effektiva frtjockare vilket betyder
att bara lite polymer behver tillsttas fr att ge den nskade
viskositeten. Nackdelen r att frg frtjockad med polymer med hg
molekylvikt har dlig tckfrmga vilket betyder att man mste gra
flera strykningar fr att f bra tckning (Figur 3). Andra nackdelar r
Figur 4. Panel frn

utflytningsfrsk. Bra
utflytning ger en jmn yta
medan dlig utflytning ger en
yta med tydliga linjer efter
penseldrag.
att frgen har dlig utflytning d.v.s. att den mlade ytan fr mrken av
penseldrag (Figur 4) och att den skvtter mycket nr man rollar den
p vggen eller i taket (Figur 5). Polymerer med lgre molekylvikt ger
bttre frgegenskaper men i gengld mste mycket mer polymer
tillsttas fr att man skall f nskad viskositet.
Figur5. Nr man skall avgra hur
mycket en frg skvtter rollas
frgen p vggen. P ett svart
papper som har placerats
horisontellt en bit nedanfr kan
man avgra hur mycket frgen har
skvtt. Panelen till vnster r ett
exempel p nr en frg skvtter
lite medan frgen som anvnts till
panelen till hger skvtter mycket.
59
Nr HM-polymerer anvnds som frtjockare i frg ger de en

Hydrofobt
hlrum
kombination av de goda egenskaperna frn polymerer med hg och

lg molekylvikt. Samtidigt som de ger bra frgegenskaper ssom bra
tckfrmga, bra utflytning och lite skvtt ger de hg frtjockningseffekt d.v.s. lite polymer behver tillsttas.
Hydrofobmodifierade polymerer frtjockar bde genom intrassling av
polymerkedjorna
Hydrofil
utsida
(spagetti)
och
genom
associationer
mellan
hydrofoba grupper (smlt ost). Arbetet i denna avhandling har gtt ut

p att frska frklara hur frtjockningen gr till och hur polymerens

en cyklodextrinmolekyl
struktur pverkar dess egenskaper. Ett stt att studera detta som jag
har anvnt i det hr arbetet r att tillstta cyklodextrin till
vattenlsningar av polymerer och se hur det pverkar lsningarnas
viskositet. Cyklodextrinmolekylen liknar en mutter i formen (Figur 6).
P utsidan r den hydrofil (tycker om vatten) medan hlet i mitten r
hydrofobt (tycker inte om vatten). En hydrofob grupp p polymeren
kan gmma sig inuti hlrummet p en cyklodextrinmolekyl frutsatt att
den inte r fr stor fr att f plats i hlet. Det finns olika cyklodextriner
med olika storlek p hlrummet. Med rtt cyklodextrin fr det bara
plats en hydrofob grupp i varje cyklodextrinmolekyl och bara en
cyklodextrin fr plats p varje hydrofob grupp. En hydrofob grupp som
har gmt sig inuti hlrummet i en cyclodextrinmolekyl kan inte lngre
delta i att bilda tvrbindningar. Resultatet blir att polymerntverket
faller snder och viskositeten sjunker. Eftersom varje cyclodextrinmolekyl tar hand om en hydrofob grupp kan man bryta tvrbindningarna i polymerntverket p ett mycket kontrollerat stt och
drmed f en detaljerad bild av hur frtjockningen gr till.

hur polymerntverket bryts
ner av cyklodextrin.
60
Acknowledgements
I wish to thank:
-
Sture and Magnus, my bosses during these years, for giving me the opportunity to use
a lot of working hours for this work and for the great freedom in choosing the subjects of
my study, almost total freedom as long as it concerned associative thickeners. I share
your conviction that this work someday will pay back in the form of new and improved
products.
my supervisors, Krister and Bjrn for excellent guidance, encouragement and patience.
Even though my first attempt to do a PhD came to nothing you encouraged me to try a
second time. Special thanks for the hospitality Krister, Maria and Thea have shown
during my many visits to Lund.*
all the people at the paint lab and the analysis lab at Akzo Nobel Surface Chemistry. I
could not have done this without the help from you. Especially, Barbro and Myran,
your results are all over this thesis.
all the people at Physical Chemistry 1 for the stimulating and friendly atmosphere at the
department. The co-authors Fredrik, Susanne, Carin and Olle for measurement
results, discussions and advice. Majlis, Monica, Martin and Maria for all small things
you have helped me with.
the people who contributed in putting the thesis together. Peter interpreted the paint
results and helped with the confusing terminology of the paint industry. David, our man
in England, helped with an extra spell and grammatical check of this thesis. I love the
comments, like water poor domain doesnt sound like English but I dont have an
alternative!
my family for all the love, support and patience.
*Krister and I, in Kristers

and Marias living room,
making the last adjustments
before submitting one of the
papers
61
List of commercially available hydrophobically modified polymers

used as associative thickeners in the paint industry.
More than 100 associative thickener products exist of which the majority is sold in very small volumes.
This is not a complete list of all products but a selection of some of the most commercially important. The
column Producers comment contains information the producers use to characterize their products. All
data are obtained from internet.
Product
Producers comment
Solvent
Producer
Acrysol TT 615
HASE
high low shear visc/low high shear visc
water
R & H10
Acrysol TT 935
HASE1
high low shear visc
water
R & H10
Acrysol DR 1
HASE1
water
R & H10
Acrysol DR 73
HASE1
high low shear visc/high high shear visc
water
R & H10
Acrysol RM 5
HASE1
water
R & H10
Acrysol RM 55
HASE1
water
R & H10
Acrysol DR 72
HASE1
high low shear visc/ pseudoplastic
water
R & H10
Acrysol RM825
HEUR2
KU efficient
BDG/w6
R & H10
Acrysol SCT275
HEUR2
KU efficient
BDG/w6
R & H10
Acrysol RM2020
HEUR2
high low shear viscosity
water
R & H10
Acrysol RM 8 W
HEUR2
KU efficient
water
R & H10
Acrysol RM 12 W
HEUR2
high low shear visc/ pseudoplastic
water
R & H10
Aquaflow NLS-200
HM-PE4
low shear efficient
BDG/w6
Aqualon
Aquaflow NLS-210
HM-PE
low shear efficient
BDG/w
Aqualon
Aquaflow NHS-300
HM-PE4
high shear efficient
water
Aqualon
250 600 mPa s (1% solution)
none
ANSC11
Bermocoll EHM200
HM-EHEC
Bermodol HAC 2000
HASE1
high medium shear visc.
water
ANSC11
Bermodol HAC 2001
HASE1
newtonian
water
ANSC11
Bermodol PUR 2102
HEUR2
high low shear visc.
BDG/w6
ANSC11
Bermodol PUR 2110
HEUR2
newtonian..
none
ANSC11
Bermodol PUR 2130
HEUR2
newtonian.
water
ANSC11
Bermodol PUR 2150
HEUR2
high low shear visc. surfactant
w/surf.7
ANSC11
DSX 1514
HEUR2
low structural viscosity
BTG/w8
Cognis
DSX 1550
BDG/w
Cognis
Cognis
DSX 2000
HEUR
structural viscosity
4
newtonian
BDG/w
newtonian
HM-PE
DSX 3000
HM-PE
DSX 3256
HEUR
HEUR
DSX 3290
Natrosol Plus 100
Natrosol Plus 330
62
Type
pseudoplastic
w/diluent
Cognis
9
Cognis
w/diluent
Cognis
HM-HEC
5 25 cP (1% solution)
none
Aqualon
HM-HEC
none
Aqualon
Product
Type
Natrosol Plus 340
HM-HEC5
Natrosol Plus 430
Rheolat 255
Rheolat 278
Rheolat 420
HM-HEC
Comment
Solvent
Producer
none
Aqualon
5000 9000 cP (1% solution)
none
Aqualon
HEUR
HEUR
antisettling
BDG/w
antisettling
water
HASE
antisettling
BDG/w
Elementis
Elementis
Elementis
pseudoplastic
BTG/w
Mnzing12
Tafigel PUR 40
HEUR
Tafigel PUR 45
HEUR2
newtonian
BTG/w8
Mnzing12
Tafigel PUR 50
HEUR2
pseudoplastic
water
Mnzing12
Tafigel PUR 60
HEUR2
strongly psedoplastic
BTG/w8
Mnzing12
Tafigel PUR 61
HEUR2
strongly psedoplastic
water
Mnzing12
Ucar Polyphobe 202
HASE1
highly associative
water
Dow
Ucar Polyphobe 203
HASE
low associative nature
water
Dow
HASE
low high shear viscosity
water
Dow
HASE
water
Dow
Ucar Polyphobe 205

Ucar Polyphobe 206
1
Hydrophobically modified polyacrylate (Hydrophobically modified Alkali Swellable Emulsion)

Hydrophobically modified urethanes
3
Hydrophobically modified ethyl hydroxyethyl cellulose
4
Hydrophobically modified polyether
5
Hydrophobically modified hydroxyethyl cellulose
6
mixtures of diethyleneglycol monobutylether and water
7
water with surfactant
8
mixture of triethyleneglycol monobutylether and water
9
water with viscosity reducing agent
10
Rohm & Haas
11
Akzo Nobel Surface Chemistry
2
12
Mnzing Chemie
63
Colloid Polym Sci 277:798804 (1999)

Springer-Verlag 1999
L. Karlson
S. Nilsson
K. Thuresson
Received: 23 October 1998

Accepted in revised form: 10 March 1999
L. Karlson
Akzo Nobel Surface Chemistry AB,
S-44485 Stenungsund, Sweden
S. Nilsson K. Thuresson (&)
Physical Chemistry 1,
Center for Chemistry and
Chemical Engineering,
Lund University, P.O. Box 124,
S-22100 Lund, Sweden
e-mail: krister.thuresson@fkem1.lu.se
Fax: +46-46-2224413
SHORT COMMUNICATION
Rheology of an aqueous solution

of an end-capped poly(ethylene glycol)
polymer at high concentration
Abstract Generally it is observed

that the viscosity of an aqueous
solution of a hydrophobically modied polymer increases with concentration; however, here it is
shown that the viscosity prole of
an end-capped poly(ethylene glycol)
polymer passes through a maximum. Thus, a substantial decrease
in viscosity is observed at high
concentrations (50 wt%). The observation is suggested to be due to a
gradual change, on the molecular
level, from a structure containing

micellelike structures that are interconnected via polymer bridges to a
more meltlike state, where micro
segregation in hydrophilic and
hydrophobic regions is less
pronounced.
Key words End-capped poly(ethylene glycol) Viscosity
Rheology Phase behavior
Microstructure
Introduction
Experimental
End-capped poly(ethylene glycol) polymers (HM-PEG)

belong to a family of associative thickeners that nd
frequent use in paint formulations [1]. Due to shipping
costs the most ecient way is to transport the product
as a dry powder from manufacturer to user. However,
due to slow dissolution kinetics, which is a complication
for the user, a trade-o has to be made and the
thickener is generally shipped in an aqueous solution. It
is generally found that such a solution has a high
viscosity [24], and to have a solution that is easy to
handle the high viscosity has to be reduced. A reduction
can be obtained in various ways, where addition of a
less polar solvent or addition of a surfactant can be
noticed [5, 6]. As a complication, the added cosolute
may disturb the nal paint formulation, and for this
reason it is desirable to reduce its concentration. In the
present investigation it is shown that the viscosity as a
function of HM-PEG concentration passes through a
maximum and decreases strongly at concentrations
above about 50 wt%. This suggests that a method to
reduce the viscosity compatible with the demands is
simply to increase the dry content.
Materials and sample preparation

The preparation of the HM-PEG, which has a triblock structure
(TB), was started by ethoxylation of a mixture of unsaturated
alcohol chains (C16-C18). The reaction continued until each of them
contained, on average, 140 repeating ethylene oxide (EO) units.
This compound (C18EO140), which has a diblock structure (DB),
may be viewed as a surfactant molecule with a polymeric-sized
headgroup (Fig. 1). In a reaction between isophoronediisocyanate
(IPDI) and C18EO140, two entities of C18EO140 were connected
head-to-head to obtain one HM-PEG polymer (C18EO140-IPDUEO140C18). During the reaction IPDI was present in small excess
and dibutyltindilaurate was used as a catalyst. Unreacted IPDI was
eliminated by termination with small amounts of ethanol.
In the following, HM-PEG will be referred to as the TB polymer
or TB, and DB will be the term used for the C18EO140 compound.
By using size-exclusion chromatography with well-characterized
PEG fractions as calibration standards the weight-average molecular weight (Mw = 11 000) and the polydispersity index (Mw/
Mn = 1.1) of the TB polymer could be determined. The TB was
used without any further purication.
Samples for rheology and phase studies, each containing a
volume of about 10 ml, were prepared by weighing the components
directly into glass tubes. These were sealed with Teon tightened
screw caps. Because the dry TB material has a uy structure
and the water in which this powder should be dissolved had a
much smaller volume, the dissolution process was sometimes
799
TB polymer a crystalline phase appears at low temperatures. This

phase border was determined by decreasing the temperature of
each sample in steps of 2 C. The temperature of the samples was
controlled to within 0.1 C by immersion in a jacketed glass
vessel connected to a temperature-controlled water bath. At each
temperature the samples were allowed to equilibrate for at least
12 h before visual inspection. Phase separation which led to the
formation of a crystalline phase in equilibrium with the micellar
phase was observed as small white ``regions'' due to scattered light
from the crystallites which were formed within the sample.
Rheological experiments
Fig. 1 The chemical structure of the end-capped poly(ethylene glycol)

(TB) investigated together with the structure of the building blocks
(DB and IPDI) from which the TB polymer was synthesized
complicated, especially for samples with a high concentration of the
TB polymer; however, by using a centrifuge the process could be
sped up and after a few hours of centrifugation all samples
appeared homogeneous. After completed dissolution the samples
were left to stand overnight before any measurements were
performed.
Phase behavior
In the present investigation only a brief study of the phase behavior
was performed. This was done in order to ensure that the viscosity
was determined under conditions where the solution was a single
phase. The cloud-point-curve data (dening the border between the
one- and two-phase regions at low TB concentration, cf. Fig. 2) is
reproduced from an earlier publication [4]. By using X-ray
diraction it has been found that at high concentrations of the
Fig. 2 Partial phase diagram for the TB polymer. The inset shows a
blow-up of the two-phase region (2U) at low concentrations. Here the
two liquid phases in equilibrium are isotropic with one phase enriched
in polymers, while the other is depleted. In the two-phase region at
high concentrations a crystalline phase containing only small amounts
of water is balanced by an isotropic phase containing a fairly high
concentration of the TB polymer. The phase diagram is dominated by
a micellar one-phase region (1U)
Oscillatory-shear and steady-shear experiments were conducted in

a Physica UDS 200 rheometer that was equipped with an automatic
gap setting and a plateplate geometry with a diameter of 25 mm.
The temperature was controlled to within 0.1 C with a Peltier
plate and the samples were protected from water evaporation by
using a solvent trap during the measurements. Before carrying out
any oscillatory-shear measurements each sample was checked to
ensure it was within the linear viscoelastic region where the storage
(G) and the loss (G) moduli are independent of the applied stress.
In the samples investigated this regime was easily identied because
it was typically found to extend over a rather wide shear-stress
range (Fig. 3). While the data presented in Fig. 3 were obtained for
one sample at one temperature (20 wt% at 40 C), similar behavior
was observed for all samples
investigated. The complex viscosity
p
was calculated using g G02 G 002 =x, where x 2pf is the
angular frequency, and f is the frequency of the oscillation in hertz.
The viscosity, g, of the samples was also obtained with steadyshear measurements. It was found that by increasing the shear rate
the tendency increased for air to enter the measuring gap and
replace a signicant volume fraction of the sample. This eect
became more pronounced when the TB concentration was
increased. If the viscosity is presented as a function of the shear
rate this artifact is observed as a pronounced shear-thinning
behavior. An alternative method to identify unphysical data is to
plot the viscosity as a function of the shear stress, r (Fig. 4b). In
this representation a certain shear stress sometimes corresponds
to two dierent viscosity values. In Fig. 4a, for each measurement
we have chosen to omit viscosity data obtained at shear rates
exceeding the maximum shear stress (Fig. 4b). The maximum shear
rate given by this method agreed with that determined by careful
examination of when uid was exiting the measuring gap.
Therefore, for the data presented in Fig. 4a, the shear-rate range
decreases when the TB concentration increases.
Fig. 3 The storage (G0 ) and loss (G00 ) moduli, and the damping factor
tan(d) = G00 =G0 given as a function of the applied stress (r). The data
were obtained at a TB concentration of 20 wt%, the frequency was
1 Hz, and the temperature was kept constant at 40 C
800
tion of the TB polymer in the lower-concentration

phase is still relatively high (65 wt% at 15 C and
85 wt% at 40 C). The very concentrated phase contains TB polymer in the crystalline state. The crystalline
structure of this phase has been detected by X-ray
diraction measurements. By increasing the temperature the tendency of this latter phase separation can be
decreased so that the one-phase region extends to
higher concentrations (Fig. 2). A deeper examination of
the phase behavior was beyond the scope of the present
investigation.
Steady-shear measurements versus oscillatory-shear
measurements
Fig. 4 a The viscosity for ve dierent TB concentrations obtained

from oscillatory-shear experiments and from steady-shear experiments. The complex viscosity, g* , (symbols) and the steady shear
viscosity, g, (thin lines) are shown. The temperature was 15 C. b The
steady-shear viscosity for samples containing 3 wt% (circles) and
30 wt% (squares) TB polymer as a function of the applied stress, r.
Filled symbols refer to the data points used in Fig. 4a, while the open
data points have been dropped as being erroneous (see text)
Results and discussion

In this investigation rheology measurements were
conducted at two dierent temperatures (15 and
40 C); however, at low TB concentrations only the
temperature slightly below room temperature (15 C)
could be used, while at high concentrations the
measurements could only be performed at the high
temperature (40 C). The reason for this is the phase
behavior of the aqueous TB solution. Generally, PEGbased compounds in aqueous solution have a reversed
temperature dependency and a phase separation into
one phase rich and one phase depleted in PEGcontaining molecules is induced by increasing the
temperature [7]. This is also the case with the present
TB polymer, and the phase-separation temperature
passes through a minimum corresponding to about
20 C at a TB concentration of about 0.6 wt% (Fig. 2)
[4]. On the other hand, at high concentrations the
solution separates into one phase very concentrated in
TB polymer (virtually 100%) which is in equilibrium
with a phase of a lower concentration. The concentra-
At suciently low frequencies or shear rates all samples

obeyed the CoxMertz rule [8] and, consequently, g*
had the same numerical value as g (Fig. 4a). At higher
frequencies or shear rates this was not the case, and the
two methods gave dierent results. As indicated in the
Experimental section one reason for such an observation
may be the penetration of air bubbles into the measuring
gap; however, great care was taken to only present data
that represent material properties of the polymer
solution. In contrast to the oscillatory-shear measurements, which all were performed at small deformations
(linear viscoelastic region), the steady-shear mode results
in a more severe perturbation of the sample [9]. At rest,
the equilibrium situation may be pictured in the
following way. Virtually all polymer chains are associated in aggregated structures of various sizes, and
provided the concentration is suciently high, a percolated dynamic network structure reaches throughout the
solution [4, 9]. The eect of the ow may be that the
percolated network is fragmented into smaller aggregated structures [9]; however, the fragments, which are
oriented in the ow prole, may still contain many
associated polymer chains. This process may be seen as a
virtual reduction of the polymer concentration (i.e. the
number of polymer chains contributing to the viscosity
via intermolecular associations within the percolated
dynamic network is decreased). Another eect of the
ow may be that polymer chains, which are due to
hydrophobic associations serving as links in the network, are stretched. The stretched-chain conformations
provide energy to overcome the potential barrier for
disengagement. In this way the exit rate from a polymer
polymer junction is increased (i.e. higher dynamics) [10].
Both these eects are expected to decrease the viscosity
and, thus, the shear-thinning region should rst be
detected in the steady-shear mode; however, at the
lowest concentration (3 wt%) the opposite was observed, and the shear-thinning region was rst detected
with the rheometer put in the oscillatory-shear mode
(Fig. 4a). A tendency for shear-thickening can also be
801
discerned from the steady-shear measurement in an

intermediate shear-rate range at this concentration.
Others have reported similar observations for solutions
of associating polymers, with a shear-thickening region
before the shear-thinning part [9, 1113]. This may be
rationalized along the same lines as Witten and Cohen
[14, 15] used to explain the shear-thickening behavior
of ionomers in organic solvents. They concluded that
the eect was due to a rearrangement from intra- to
interaggregation when individual polymer chains were
stretched in the ow prole (i.e. the number of polymer
chains contributing to the viscosity via intermolecular
associations is increased). Recently Tam et al. [9] argued
that ow-induced interaggregation also increases the
fraction of polymers that participate in the percolation
network; however, at even higher shear rates the normal
shear-thinning behavior is always observed (Fig. 4a).
We note that in the unperturbed state the fraction of
polymer chains in the loop conformation (i.e. both ends
of the polymer chain bind to the same hydrophobic
domain) is expected to increase when the concentration
decreases. Therefore, in a limited concentration interval
the shear-thickening eect is expected to increase with
decreasing concentration.
passes through a maximum, and a decrease corresponding to more than one decade can be observed at high TB
concentrations. In order to rationalize these observations it may be valuable to follow the concentration
dependency of the evolution of G0 and G00 (as a function
of frequency) (Fig. 6). An indirect way to probe polymer
dynamics is to determine the inverse angular frequency,
s* = 1/2pf *, where G0 and G00 intersect. At times
shorter than this characteristic time a solution has a
response which is mainly elastic (this may be interpreted
as the polymer chains not having the time to relax to
their new equilibrium positions), while at longer times
viscous behavior prevails. However, it is important to
realize that s* is not expected to correspond to any
characteristic time on a molecular level, but rather it
should be used to indicate dierences between samples.
It is likely that s* is inuenced by factors such as
entanglements and intermolecular associations. The
Inuence of dynamics on the viscosity

Figure 5 reports g* from the low-frequency limit
(x 0), where all solutions had Newtonian behavior
(frequency-independent viscosity). At 15 C the viscosity
seems to reach a plateau, and within the concentration
range investigated, no signicant decrease can be
observed; however, at this temperature the concentration range is limited, while at 40 C, where higher
concentrations are accessible, the Newtonian viscosity
Fig. 5 The Newtonian viscosity as a function of TB polymer

concentration. The data points were obtained from oscillatory-shear
experiments via the complex viscosity, g*, when the angular
frequency, x, approached 0 s)1. The lines are calculated as the
product of G* and s* (see text). Dots (d) and the full line refer to
15 C, while open circles (s) and the dashed line were obtained at
40 C
Fig. 6 a Storage, G0 , and loss, G00 , moduli as a function of angular

frequency, x. Measurements at eight dierent concentrations of the
TB polymer are given, and in order to have a separation between the
dierent measurements the curves were vertically shifted by the shift
factors given. For each TB concentration the storage modulus is
indicated, and the line above each G0 curve represents the corresponding loss modulus. The concentration corresponding to each pair
(of G0 and G00 ) is given in the table. From bottom to top the TB
concentration is 3, 6, 10, 20, 30, 35, 40, and 55 wt%. The temperature
was kept constant at 15 C. b The same as in a but the measurements
were performed at 40 C. Each pair of G0 and G00 curves corresponds
to the concentrations given in the table. Note the shift factors
802
coordinates (s* and G*) of the intersection points of G0

and G00 for the dierent samples at the two temperatures
are shown in Fig. 7. Generally, for polymer solutions the
expectation is that the dynamics and the viscosity of the
solution are coupled. For solutions containing nonassociating polymers the normal case is that an increased
entanglement situation follows an increasing polymer
concentration. Thus, an increase both in the Newtonian
viscosity as well as in s* is anticipated with increasing
concentration. This is indeed the case at low-to-intermediate TB concentrations, while in line with what was
observed from the viscosity (Fig. 5), Fig. 7 clearly shows
that, with the present TB polymer, s* passes through a
maximum, and a decrease is observed at high concentrations; however, the concentration corresponding to
the onset of the decrease diers somewhat for the two
parameters (with the decrease in s* located at a lower
concentration). Figure 7 also shows that G* increases in
the concentration range investigated. In a system where
the rheological response is determined by a single
relaxation process the product of the characteristic time
of that process, s, and the value of the storage modulus
in the high-frequency limit, G1 , is proportional to the
Newtonian viscosity [16]: g sG1 . In Fig. 5, together
with the data of the Newtonian viscosity, we have
included the product s*G*. Despite the absolute values
diering, the evolution with concentration is similar.
Here we notice that a dierence is expected since
G* G1 and, furthermore, it is reasonable that, at least
at high concentrations, this system is associated with
more than one relaxation process. This would in turn
imply that s* s.
Inuence of structure on the viscosity
Here a more thorough discussion regarding the microstructure of the solution is valuable. Relative-viscosity
Fig. 7 s* (circles and dots) and G* (open and lled squares) given from
the ``cross-over'' points in Fig. 6a. Filled symbols refer to 15 C, while
open symbols refer to 40 C. The ``cross-over'' points in the 10 and
55 wt% solutions at 40 C were determined by extrapolating G0 and G00
measurements on the corresponding DB system (from

which the TB polymer is produced see Material and
sample preparation section and Fig. 1) and a nonlinear
least-squares t of a hard-sphere model to that data,
suggest that this solution approaches maximum packing
at a DB concentration of about 56 wt%. Of course one
has to be careful with absolute values given by the hardsphere model since the micelles denitely have a soft
potential; however, it should be without doubt that at
fairly low concentrations the solution is lled with
micelles. This conclusion is also suggested by the
intrinsic viscosity, [g] = 47 cm3/g, of the DB system,
which suggests that the micelles overlap at a concentration of 2.1 wt% (calculated from the reciprocal intrinsic
viscosity [g])1 = 0.021 g/cm3). The intrinsic viscosity
was extracted from measurements in the concentration
range 0.15 wt% of the DB polymer. Because the
critical micelle concentration, (cmc) of this compound
is expected to be very small, [g] refers to a solution
containing DB micelles. (The cmc of a similar DB
compound, C18EO84, was reported to be 10)3 wt% [17].)
We note that the dierent numerical values from the two
dierent methods (56 and 2.1 wt%, respectively) are to
be expected. The hard-sphere model results in a smaller
radius of the micelles than that given by intrinsic
viscosity data, which give the hydrodynamic radius.
Furthermore, the suggestion that the solution is lled
with micelles at rather low concentrations is in line with
a recent investigation regarding structural and dynamic
properties of aqueous solutions of the related compound
C17EO84 [18]. It was found that the micellar phase at low
concentration was followed by a close-packed micellar
phase at higher concentration (transition region 10
20 wt%). From an earlier publication it is known that
the present TB compound also assembles in selfaggregated structures [4]; these are expected to have a
strong resemblance to the DB micelles. At very low TB
concentrations the average distance between ``polymeric
micelles'' is long, and there is a high probability that a
TB polymer has both ends in the same micelle, with the
PEG middle block being in a loop conformation;
however, this is not the case in the present investigation
because higher concentrations are employed. With
increasing concentration intermicellar bridges become
increasingly important, and this is the main reason for
an increased viscosity with increasing TB concentration.
At even higher concentrations the bridges are expected
to reach not only nearest micelles but also micelles
further away [19]; thus, at intermediate concentrations
the solution may be viewed as consisting of polymeric
micelles connected via these bridges to ``clusters'', and at
higher concentrations the solution becomes more entangled with ``interpenetrating clusters''. We conclude that
the observation that s* and the viscosity increase with
increasing concentration at low-to-intermediate TB
concentrations is reasonable.
803
If the increase in s* and the viscosity is straightforward to explain, it is more dicult to rationalize the
decrease in these parameters at high concentrations. The
rst alternative which comes to mind is a transition from
a close-packed micellar phase containing spherical
aggregates in cubic symmetry, to a hexagonal phase,
and later a lamellar phase. This is the normal behavior
when increasing the concentration in systems containing
amphiphilic compounds [20]. The structural change is
expected to reduce the viscosity because the symmetry in
the solution changes. Rods in hexagonal symmetry and
lamellas can slide relative to each other. One example
of where the viscosity has been reported to decrease
by passing the phase border from cubic micellar to
hexagonal is in aqueous solutions of Pluronic polymers
[21, 22]. [A Pluronic polymer has the structure
(EO)n(propylene oxide)m(EO)n.] In these investigations
increasing the temperature while keeping the concentration xed changed the microstructure in the Pluronic
solution; however, such a phase transition should also
result in birefringence of the samples when observing
them between crossed-polarized glass plates. No birefringence could be detected for any of the TB samples.
Perhaps the lack of less-curved aggregates (hexagonal
and lamellar structures) could be anticipated taking into
account the large middle block of the TB molecules. The
associated big headgroup area at the polar/nonpolar
interface should tend to give a high spontaneous
curvature towards oil favoring spherical aggregates.
This is a general trend with increasing headgroups that
has been observed both experimentally and in mean-eld
calculations in systems containing the Pluronic type of
polymer [23, 24]. At the present stage we do not have an
unambiguous explanation for the decreased s* and
viscosity at high TB concentrations, but we believe that
the observation can be ascribed to the fact that the
solution approaches a ``meltlike'' state. Thus, the border
between polar and nonpolar regions becomes blurred,
and the solution becomes more homogeneous on a
molecular level. In this view, and referring to Figs. 5 and
7, the viscosity decreases as a result of a gradual change,
starting with a decreased lifetime of the hydrophobic
moieties in the polymeric micelles and ending with
vanishing interpenetrating clusters, clusters, and micelles.
Conclusions
In this paper we have shown that the viscosity changes in
a nonmonotonous way with increasing concentration of
an HM-PEG polymer (TB) in aqueous solution, and at
high concentrations the viscosity decreases substantially.
At low and intermediate TB concentrations the increase
in viscosity was interpreted in terms of the formation of
polymeric micelles, which were connected to clusters and
later, at higher concentrations, interpenetrating clusters.
The decrease in viscosity at high concentrations was
suggested to be due to a gradual transformation towards
a meltlike state, and at this stage the organization of the
solution into polar and nonpolar regions should be less
pronounced.
The observation should be important, for instance,
for manufacturers of polymers of the TB type. Such
polymers are generally shipped to consumers in aqueous
solution, and a low viscosity solution with a high dry
content is desired. This has, for instance, been obtained
by adding dierent cosolutes (as exemplied by nonpolar
solvents and surfactants) to solutions containing a
moderate TB concentration. The observation in this
investigation suggests that a high dry content and a
relatively low viscosity may be obtained by simply
increasing the TB concentration; however, with the
present TB polymer the highest concentrations (and thus
the lowest viscosity) are not easily accessible due to the
formation of a crystalline phase at low-to-intermediate
temperatures. Here we note that the latter property may
also be of interest. Under conditions where the concentrated sample phase-separates, it follows that the viscosity of the micellar phase (which is the main part of the
sample) increases substantially due to a decrease in
the HM-PEG concentration in that phase. By lowering
the temperature of a solution containing at least 80 wt%
HM-PEG to below 10 C an increase in the viscosity
corresponding to about one decade can be expected (as
suggested by Figs. 2, 5 in combination). This is a
property that may be useful in applications where a
``gelation'' by lowering the temperature is sought.
Acknowledgements S.N. and K.T. thank the Center for Amphiphilic Polymers for nancial support. The rheometer was funded
by a grant from Nils and Dorthi Troedsson's Research Foundation.
References
1. Glass JE (ed) (1989) Polymers in aqueous media. American Chemical Society, Washington, D.C.
2. Jenkins RD (1990) Ph.D. Thesis. Lehigh University, Bethlehem, USA
3. Annable T, Buscall R, Ettelaie R,

Whittlestone D (1993) J Rheol
37:695726
4. Thuresson K, Nilsson S, Kjniksen
A-L, Walderhaug H, Lindman B,
Nystrom B (1999) J Phys Chem B

103:14251436
5. Annable T, Buscall R, Ettelaie R,
Shepherd P, Whittlestone D (1994)
Langmuir 10:10601070
804
6. Karlson L, Joabsson F, Thuresson K

(1998) Carbohydr Polym (submitted)
7. Karlstrom G (1985) J Phys Chem
89:49624964
8. Cox WP, Merz EH (1958) J Polym Sci
28:619622
9. Tam KC, Jenkins RD, Winnik MA,
Basset DR (1998) Macromolecules
31:41494159
10. Tanaka F, Edwards SF (1992) J NonNewtonian Fluid Mech 43:247271
11. Bock J, Siano DB, Valint PL Jr, Pace
SJ (1989) In: Glass JE (ed) Polymers in
aqueous media. American Chemical
Society, Washington, D.C., pp 411424
12. Walderhaug H, Hansen FK, Abrahmsen S, Persson K, Stilbs P (1993) J Phys
Chem 97:83368342
13. Jenkins RD, Silebi CA, El-Aasser MS

(1991) In: Schulz DN, Glass JE (eds)
Polymers as rheology modiers. American Chemical Society, Washington,
D.C., pp 222233
14. Witten TA, Cohen MH (1985) Macromolecules 18:19151918
15. Witten TA (1988) J Phys (Paris)
49:10551063
16. Doi M, Edwards SF (1986) The theory
of polymer dynamics, 1st edn. Oxford
University Press, New York
17. Yaminsky VV, Thuresson K, Ninham
BW (1998) Langmuir (accepted)
18. Hakansson B, Hansson P, Regev O,
Soderman O (1998) Langmuir 14:5730
5739
19. Semenov AN, Joanny J-F, Khokhlov

AR (1995) Macromolecules 28:1066
1075
20. Evans DF, Wennerstrom H (1994) The
colloidal domain: where physics, chemistry, biology and technology meet, 1st
edn. VCH, New York
21. Schillen K, Glatter O, Brown W (1993)
Prog Colloid Polym Sci 93:6671
22. Hvidt S, Jrgensen B, Brown W,
Schillen K (1994) J Phys Chem
98:1232012328
23. Wanka G, Homann H, Ulbricht W
(1994) Macromolecules 27:41454159
24. Linse P (1993) J Phys Chem 97:13896
13902
Colloids and Surfaces

A: Physicochemical and Engineering Aspects 201 (2002) 9 15
www.elsevier.nl/locate/colsurfa
Clouding of a cationic hydrophobically associating comb

polymer
K. Thuresson a,*, L. Karlson b, B. Lindman a
a
Physical Chemistry 1, Center for Chemistry and Chemical Engineering, Lund Uni6ersity, P.O. Box 124,
SE-221 00 Lund, Sweden
b
Akzo Nobel Surface Chemistry AB, SE-444 85 Stenungsund, Sweden
Received 16 June 1999; accepted 29 September 2000
Abstract
A novel cationic hydrophobically associating comb polymer, containing poly(oxyethylene) chains in the back-bone,
is described and investigated with respect to the aqueous solubility by cloud point measurements. Cloud points were
investigated as a function of polymer charge, poly(oxyethylene) chain length and concentrations of added NaCl and
could be interpreted in terms of the aqueous behavior of poly(oxyethylene) chains, general electrostatic effects,
salting-out effects for nonionic polymer chains and polymer association. 2002 Elsevier Science B.V. All rights
reserved.
Keywords: Associating polymer; Hydrophobically modified polymer; Comb polymer; Phase behavior; Cloud point
1. Introduction
Hydrophobically associating polymers have in
recent years become a very active field of research.
Both block and graft copolymers as well as more
complex architectures have been investigated,
with both fundamental and applied aspects as a
starting point [1 4]. With respect to applications,
rheology control and adsorption on solid surfaces
are in the foreground, but control of phase behavior is always of key importance.
* Corresponding author. Tel.: + 46-46-2220112; fax: +4646-2224413.
E-mail address: krister.thuresson@fkem1.lu.se (K. Thuresson).
There is an enormous variation in polymers

investigated with respect to architecture, backbone, substituents and charge but due to limitations in methods of synthesis and modification as
well as in characterization it is difficult to systematically vary the different parameters of polymer
chemical structure, which are relevant. Instead
progress is dependent on the synthesis and
physico-chemical studies of polymers with features which are novel with respect to previous
work.
The present study introduces a novel polymer,
characterized by alkyl chains grafted onto a backbone composed of poly(oxyethylene) (POE)
chains and cationic ammonium groups; the backbone is obtained by coupling fatty amines with
0927-7757/02/$ - see front matter 2002 Elsevier Science B.V. All rights reserved.
PII: S 0 9 2 7 - 7 7 5 7 ( 0 0 ) 0 0 7 7 5 - 5
10
K. Thuresson et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 201 (2002) 915
POE chains by diisocyanate. Furthermore, an initial physico-chemical characterization with respect

to phase behavior is presented. The polymer displays a clouding behavior in water at elevated
temperature and the conditions of clouding are
investigated and give an insight into interpolymer
association.
2. Experimental
2.1. Materials and sample preparation

The amphiphilic polymers used in this investigation were synthesized at Akzo Nobel Surface
Chemistry AB, Stenungsund, Sweden. The preparation was started by ethoxylating fatty amines.
The amines were produced from coconut oil that
mainly contains saturated hydrocarbons (C12).
Two different polymer fractions were made, and
the ethoxylation reaction was continued until the
fatty amine contained, on average, 51 or 74 repeating ethylene oxide units, respectively. This
compound, which has a diblock structure (DB),
may be viewed as a surfactant molecule with a
polymeric-sized head group (Fig. 1). This surfactant is positively charged at low pH, while it is
uncharged at high pH, due to the titrating aminegroup located between the C12 hydrocarbon tail
and the poly(oxyethylene), POE, headgroup. In a
condensation reaction between the surfactant and
isophoronediisocyanate, IPDI, a hydrophobically
modified poly(oxyethylene) polymer with a comb
structure was produced (Fig. 1). During the reaction IPDI was present in small excess and
dibutyltindilaurat was used as a catalyst. Unreacted IPDI was eliminated by termination with
small amounts of ethanol. A rough estimate of the
weight-average molecular weight (Mw) was determined by size-exclusion chromatography (SEC)
on a packing material made from styrene-divinylbenzene. Three columns were coupled in series
(Phenogel 10 000 A, (600 7.8 mm), Phenogel
1000 A, (600 7.8 mm) and Phenogel 500 A,
(600 7.8 mm) (Phenomenex)) and the mobile
phase was tetrahydrofuran. The flow-rate was 1.0
ml min 1, the injection volume 100 ml and the
temperature 25C. For detection a refractive index
detector was used. A narrow standard calibration

was made with PEG/PEO with Mp = 4250,
12 600, 27 250 and 50 400. The Mw was about the
same for both polymer fractions and was estimated to be roughly 25 000. Thus, each polymer
chain is composed of, on average, four DB
molecules. Apart from a distribution in the length
of the POE chains of the DB compound, the
number of DB molecules in a polymer chain may
vary significantly. It is not surprising that the
polydispersity index of Mw/Mn : 2.2 suggests a
wide distribution in molecular weight. Below, the
two different polymer fractions will be referred to
as -(OE)51- and -(OE)74-, respectively.
Before use the polymers were purified in several
steps. In order to remove hydrophobic compounds, such as unreacted fatty amines or alcohol-terminated
isophoronediisocyanate,
the
polymer sample was allowed to swell or partly
dissolve in acetone for a few hours under vigorous
Fig. 1. The chemical structure of the investigated comb copolymer together with the structure of the ethoxylated amine
(DB) and the coupling agent, isophoronediisocyanate (IPDI),
from which the final polymer was synthesized. The polymer
was made with two different lengths of the POE block (x =51
or 74 OE-units, respectively). Calculated from the molecular
weight, Mw =25 000, size-exclusion chromatography (SEC)
suggests n = 4.
stirring. Later the polymer was precipitated by an

addition of hexane. After repeating this procedure
(three to five times) the precipitate was collected
and carefully dried, and later dissolved in water to
a concentration of a few percent. This solution
was dialyzed against deionized water, which had
been passed through a Millipore water purification unit. This treatment reduces the amount of
low molecular weight impurities (such as salt and
oligo(oxyethylene)). Finally the polymers were recovered by freeze-drying. The samples for the
phase studies, each containing a volume of about
3 ml, were prepared by weight from a polymer
stock solution (with a concentration of ca. 10
wt.%), Millipore water, and NaCl. Glass tubes
sealed with Teflon tightened screw caps were used.
Each sample was equilibrated at least overnight
before measurement. The pH was adjusted by
adding small volumes of either HCl or NaOH
(with a concentration of approximately 1 M) with
a microsyringe.
2.2. Phase beha6ior

The temperature of the samples was controlled
to within 9 0.1C by immersion in a jacketed
glass vessel connected to a temperature-controlled
water-bath. At temperatures significantly differing
from room temperature there may be a large
difference between the temperature in the water
bath and the actual temperature in the sample
cell. Therefore, a thermocouple was used for a
direct measurement in the sample. The cloud
point temperature, Tcp, was determined by visual
inspection of the sample. The temperature was
varied in steps of 1C, and Tcp was taken as the
mean value from determinations where the temperature was increased or decreased, respectively.
This procedure rendered an estimated uncertainty
of 91C. During temperature equilibration and
measurement, the sample was stirred with a magnetic bar. For samples where temperatures below
0C had to be used a mixture of water and
ethanol was used as the cooling medium. In order
to facilitate the determination of Tcp at low temperatures, the sample cell was sprayed with ethanol on the outside to prevent condensed water
from freezing.
11
3. Results and discussion

Many nonionic polymers, as well as nonionic
surfactants of the oligo(oxyethylene) type, show a
reverse temperature dependence in their solubility
in aqueous systems, i.e. solubility decreases with
increasing temperature [5,6]. In the region of the
onset of phase separation, the scattering of light
from the solutions increases dramatically; the solutions get cloudy, hence the notion of a cloud
point. The cloud point gives useful information
on the molecular interactions and, since it can be
accurately and easily measured, it offers a very
useful characterization of these systems.
Probably the simplest example of clouding of
an aqueous polymer system is offered by poly(oxyethylene); the phase behavior of POE water
has been investigated in detail over the entire
mixing range for different molecular weights [7,8].
The phase diagram displays the expected behavior
of a closed-loop two-phase region (thus with mutual miscibility between polymer and water both
at low and at high temperatures, but not at
intermediate), which increases in magnitude, both
along the temperature axis and along the composition axis, as polymer molecular weight increases.
Nonionic surfactant phase behavior has similarities with that of POE, but the behavior is more
complex due to the self-assembly which leads to
several additional phases, than the water-enriched
and the polymer-enriched solution phases [9].
Clouding is also displayed quite generally by
nonionic graft and block copolymers containing
oxyethylene groups. The phase behavior is intermediate between what is displayed by POE and
nonionic surfactants, the complexity being determined mainly by the self-assembly of the polymer.
In general, the more strongly associating the polymer the lower is the cloud point.
The solubility and the clouding of nonionic
polymers and surfactants is strongly dependent on
the addition of cosolutes [1014]. Depending on
whether there is an enrichment or depletion of a
cosolute at the polymer molecules, either an increased, i.e. increased cloud point, or decreased
solubility can result [15]. Ionic surfactants are the
most efficient in raising the cloud point and associate to the nonionic micelles or the nonionic
12
Fig. 2. The cloud point temperature, Tcp, as a function of the

polymer concentration for the two different comb copolymers.
The NaCl concentration was kept constant at 100 mM. The
insert shows the same data in a semi-logarithmic representation.
polymer and effectively bring about a charging;

for a charged system the important contribution
from the mixing entropy of the counterions
strongly promotes solubility [16]. However, this
effect is strongly reduced by added electrolyte.
Solubility can also be strongly enhanced by
charging up the polymer itself by the same mechanism. Therefore, polyelectrolytes are generally
highly soluble and do not show clouding at elevated temperature. However, for many polyelectrolytes, the solubility is strongly reduced in the
presence of electrolyte.
In the present investigation, we have considered
in some detail the clouding of a novel polymer,
which contains long POE chains, at the same time
as it is charged; the charge of the polymer can be
varied by varying the pH of the system. For two
polymers with different lengths of POE chains, we
have investigated the clouding as a function of
polymer concentration, pH and concentration of
added salt.
Generally it is thus found that poly(oxyethylene)-containing compounds have a reversed temperature dependency with a decreased
solubility at elevated temperatures. This is also
the case with the present amphiphilic polymers
(Fig. 2). At low concentrations there is a pronounced decrease in Tcp with increasing polymer
concentration. At approximately 1 wt.% polymer
a minimum is reached, and at higher concentra-
tions Tcp increases. The polymer containing the

longest poly(oxyethylene) blocks, -(OE)74, has the
highest Tcp of the two investigated polymer fractions. It is obvious that the solubility decreases as
a result of a decreased length in the hydrophilic
OE block and the concomitant increased influence
of the hydrophobic IPDI units and fatty chains.
The data in Fig. 2 were obtained at a rather
high concentration of screening electrolyte in order to reduce the expected strong influence of
charged amine groups. At polymer concentrations
slightly above 1 wt.%, Tcp is rather invariant with
concentration. Therefore, a polymer concentration of Cp = 2 wt.% was chosen for further investigations of the influence of a screening electrolyte
and a variation in pH.
As can be expected, a strong decrease in Tcp
was observed at NaCl concentrations up to about
CNaCl = 30 mM, while at higher concentrations
the decrease in Tcp is less pronounced. The first
steep part can be ascribed as a screening of the
electrostatic forces, which in turn is expected to
decrease the solubility of the polymer chains; as
salt is added the entropic penalty of phase separation arising from the counterions is reduced. The
rather slow decrease in Tcp at NaCl concentrations above CNaCl = 50 mM is not caused by a
simple electrostatic effect. This weak decrease in
solubility is ascribed to a salting-out effect. Such a
behavior is indeed observed when NaCl is added
to aqueous solutions of a range of non-charged
compounds, like POE and nonionic cellulose
ethers. The solubility is reduced on addition of
cosolutes depleted in the vicinity of the polymer.
As indicated in Section 2, the charge density of
the present polymer compounds can be varied by
a change in the pH. Because the charge is associated with a titrating amine-group, the polymers
can be expected to have either a zero-charge or to
be polyelectrolytes with a positive charge. At low
pH the polymer is expected to be highly charged,
while an increasing pH is expected to reduce the
charge density. In line with the expectation of a
titrating amine-group the pH-range where the
charge changes strongly seems to be located at
easily accessible pH-values, around pH 7.58.5
(Figs. 4 and 5). The effect of screening electrolyte
is pronounced at high charge densities (low pH),
while at high pH when the polymer chains approach zero-charge the effect of salt is small and
Tcp is virtually the same independent of the NaCl
concentration. (A closer examination reveals
small variations. This will be discussed below.)
For electrostatic reasons, the high charge end of
the titration curve (low pH) extends to lower
pH-values when the concentration of screening
electrolyte is decreased (Fig. 4). This effect can be
traced to the fact that titrating groups on each
polymer chain are close enough not to be independent. It is interesting to see that also at very
high concentrations of screening electrolyte, at
which electrostatics should have a vanishing contribution to the solubility, an increasing pH (and
decreasing charge density) decreases the solubility
of the comb polymers. We will return to this issue
below.
In Figs. 6 and 7, we have assembled data for
the polymer concentration dependency of Tcp for
two different fixed pH values (pH 6 and 10). At
these two pHs, the polymer chains are expected to
have a high and a low charge density, respectively
(cf. Figs. 4 and 5). However, at the employed salt
concentration, CNaCl =100 mM, long range electrostatic repulsions are screened and the polymers
should have a similar behavior as related nonionic compounds. In these figures we have also
included the Tcp for an uncharged end-capped
poly(ethylene glycol). From earlier investigations,
it is known that this polymer forms micellar-like
structures. Thus, the hydrophobic part of the
molecules is protected from water exposure by the
hydrophilic poly(ethylene glycol) head group,
which forms the corona of the micelle. In a first
approximation such micelles are expected to have
a Tcp similar to that of an unmodified
poly(ethylene glycol) of a corresponding molecular weight. Following this the cloud point temperature of the micelles is expected to be
Tcp ] 100C. The pronounced difference to the
measured data points was ascribed to the bifunctionality of the end-capped polymer [17]. Due to
connectivity this renders an attraction between
micelles that promotes a formation of a concentrated phase. A reduced Tcp is expected. When the
polymer concentration increases this mechanism
becomes less important (because the average dis-
13
tance between micelles decreases), and Tcp again

increases (strongly).
A similar behavior is anticipated for the present
comb polymers. However, this is a more complicated system, with a pH-sensitive charge density
which influences the solubility strongly when
CNaCl B 50 mM (cf. Fig. 3). To rationalize the
data in Figs. 6 and 7, we may speculate that
despite the fact that long-range electrostatic repulsions are screened at CNaCl = 100 mM, there is a
non-negligible contribution from the electrostatic
repulsions in the headgroup region of the micelles.
This opposes aggregation, and should, therefore,
favor solubility at concentrations where the restricted swelling mechanism otherwise operates.
This explains the difference in Tcp between pH 6
and 10 (Figs. 6 and 7), and, therefore, also the
variation with pH at high salt concentrations
which was pointed out in connection with Figs. 4
and 5. In this context, a close examination of
Figs. 4 and 5 also reveals that changing the salt
concentration from 100 to 300 mM has a somewhat more pronounced influence at low pH (high
charge density of the polymer) than at high. This
observation can possibly be related to an increased aggregation situation and accompanying
connectivity. However, Tcp is also influenced at
high pH (where the polymer is expected to be
uncharged). We relate the latter observation to a
salting-out effect.
Fig. 3. The cloud point temperature, Tcp, as a function of

added NaCl for the two different comb copolymers at a
concentration of 2 wt.%. The insert shows the same data in a
semi-logarithmic representation.
14
Fig. 4. The cloud point temperature, Tcp, for the comb copolymer with the short PEO-block (containing 51 U) as a function
of pH at three different NaCl concentrations (0, 100, and 300
mM). The arrows indicate samples that had Tcp \ 100C. The
polymer concentration was kept constant at 2 wt.%.
Figs. 6 and 7 also comprise data from Fig. 2. It

follows that when the pH was not adjusted the
-(OE)74- polymer produces solutions with pH
slightly above pH 6 (cf. Fig. 7). A comparison
with data presented in Fig. 5 suggests a pH 6.57
in a 2 wt.% solution (data point marked in Fig.
7). The corresponding data for the -(OE)51- polymer suggests that this polymer produces solutions
with a higher pH. By comparison to Fig. 4 the
marked data point in Fig. 6 (2 wt.% polymer)
corresponds to pH 8.1.
If the qualitative behavior of the present comb
polymers is similar to that of the end-capped PEG
polymer, with a decrease in Tcp at low concentrations and thereafter an increase at higher concen-
Fig. 5. As in Fig. 4 but for the comb copolymer with the long
PEO-block (containing 74 U). With this polymer, Tcp was not
determined at CNaCl = 0 mM.
Fig. 6. The cloud point temperature, Tcp, for the comb copolymer with the short PEO-block (containing 51 U) as a function
of polymer concentration for two fixed pH values (pH 6 or 10,
respectively). Also included is the data from Fig. 2, where the
pH was not controlled. The NaCl concentration was kept
constant at CNaCl =100 mM. Included in the figure is Tcp for
an ordinary end-capped PEG polymer (from Ref. [17]).
trations, there are also differences. The

end-capped polymer has a much more pronounced minimum in Tcp, and we assign this to
the strong attraction between micelles that is expected from the C18 moieties. This association is
also expected to produce rather well-defined micelles. With the present comb polymers, the hydrophobic attractions are weaker (C12 moieties).
We may also speculate that due to geometrical
constraints and the expected chemical differences
between different polymer chains, the micellar
structures that form are less well defined. One
obvious complication to the aggregation process
Fig. 7. As Fig. 6 but for the comb copolymer with the long
PEO-block (containing 74 U).
is that two POE chains in the corona accompany

each hydrophobic tail aggregated in a micelle.
15
Acknowledgements
K. Thuresson thanks the Centre for Amphiphilic Polymers (CAP) for financial support.
4. Conclusions
References
A novel hydrophobically-associating cationic
comb copolymer is described. Main features of
the copolymer are alkyl chains grafted onto a
backbone containing poly(oxyethylene) chains
and a variable (with pH) cationic charge. The
solubility in aqueous systems shows a complex
behavior with features of both a polyelectrolyte
and of a POE copolymer. The temperature-dependent solubility, as studied by cloud point measurements, has the following main features:
Solubility
decreases
with
increasing
temperature.
Solubility increases with increasing lengths of
the POE blocks.
Solubility increases with increasing charge density of the polymer.
Solubility decreases with increasing NaCl concentration both for charged and uncharged
polymer and both at low and high salt
concentration.
The observations can be understood in terms of
a combination of the effects of interpolymer association, of a decreased polarity of the polymer at
higher temperatures and two types of electrostatic
effects. Salt addition has dual effects: For a
charged polymer there is a strongly decreased
solubility due to small additions of salt, which can
be understood as a general electrostatic effect
related to the entropy of the counterion distribution. In addition, there is a general salting-out
effect, displayed at high salt concentration and
also for the nonionic polymer, attributed to the
depletion of ions in the vicinity of the polymer
chains. The latter effect is well-established for a
wide range of nonionic polymers, including POE
and cellulose ethers.
[1] J.E. Glass, Polymers in aqueous media, in: Advances in

Chemistry Series, vol. 223, American Chemical Society,
Washington, DC, 1989.
[2] D.N. Schulz, J.E. Glass, Polymers as rheology modifiers,
in: ACS Symposium Series 462, American Chemical Society, Washington, DC, 1991.
[3] M.A. Winnik, A. Yekta, Curr. Opin. Colloid Interf. Sci. 2
(1997) 424 436.
[4] I. Iliopoulos, Curr. Opin. Colloid Interf. Sci. 3 (1998)
493 498.
[5] F.E.J. Bailey, J.V. Koleske, Poly(ethylene oxide), Academic Press, New York, 1976.
[6] G. Karlstro m, B. Lindman, Phase behavior of nonionic
polymers and surfactants of the oxyethylene type in water
and in other polar solvents, in: S.E. Friberg, B. Lindman
(Eds.), Organized Solutions, Marcel Dekker, New York,
1992.
[7] S. Saeki, N. Kuwahara, M. Nakata, M. Kaneko, Polymer
17 (1976) 685.
[8] G. Karlstro m, J. Phys. Chem. 89 (1985) 4962 4964.
[9] D.J. Mitchell, G.J.T. Tiddy, L. Waring, T. Bostock, M.P.
MacDonald, J. Chem. Soc. Faraday Trans. 1 (79) (1983)
975.
[10] G. Karlstro m, A. Carlsson, B. Lindman, J. Phys. Chem.
94 (1990) 5005 5015.
[11] B. Lindman, A. Carlsson, G. Karlstro m, M. Malmsten,
Adv. Colloids Interf. Sci. 32 (1990) 183 203.
[12] P. Bahadur, K. Pandya, M. Almgren, P. Li, P. Stilbs,
Colloid Polym. Sci. 271 (1993) 657 667.
[13] K. Pandya, K. Lad, P. Bahadur, J.M.S. Pure Appl.
Chem. A30 (1993) 1 18.
[14] K. Thuresson, S. Nilsson, B. Lindman, Langmuir 12
(1996) 2412 2417.
[15] L. Piculell, S. Nilsson, Prog. Colloid Polym. Sci. 82 (1990)
198 210.
[16] L. Piculell, B. Lindman, G. Karlstro m, Phase behavior of
polymer surfactant systems, in: J.C.T. Kwak (Ed.), Polymer Surfactant Systems, Marcel Dekker, New York,
1998, pp. 65 141.
[17] K. Thuresson, S. Nilsson, A.-L. Kjniksen, H. Walderhaug, B. Lindman, B. Nystro m, J. Phys. Chem. B 103
(1999) 1425 1436.
CARP 1351
Carbohydrate Polymers 41 (2000) 2535

www.elsevier.com/locate/carbpol
Phase behavior and rheology in water and in model paint formulations

thickened with HM-EHEC: influence of the chemical structure and the
distribution of hydrophobic tails
L. Karlson a, F. Joabsson b, K. Thuresson b,*
a

Physical Chemistry 1, Center for Chemistry and Chemical Engineering, Lund University, PO Box 124, SE-221 00 Lund, Sweden.
Received 13 November 1998; received in revised form 24 February 1999; accepted 28 April 1999
Abstract
The phase behavior and rheology of aqueous solutions of hydrophobically modified ethyl(hydroxyethyl) cellulose have been investigated.
Effects of variations in the chemical structure of the hydrophobic tails grafted to the polymer backbone were followed. When the length of the
polymer hydrophobic tails was increased the effects caused by association between different polymer chains became more pronounced. This
was manifested by an increased tendency of the solution to phase separate, a higher viscosity, and a more elastic rheological response. The
higher elasticity and viscosity was ascribed to slower polymer dynamics following from stronger hydrophobic associations. A separation of
chemically different polymer chains into two coexisting phases was strongly promoted by modification with long hydrophobic tails. It was
found that one of the coexisting phases contained highly substituted polymer chains, while in the other phase, less substituted polymer chains
were found. It is proposed that this type of phase separation occurs because the highly substituted polymer chains have a pronounced
tendency to form a network.
Model paint formulations prepared with the different polymers showed that an increasing length of the polymer hydrophobic tails slowed
down the dynamics of the formulation. This was manifested as a higher thickening efficiency (a smaller amount of polymer material was
needed to obtain the desired viscosity), and a more pronounced shear-thinning behavior of formulations comprising polymers with long
hydrophobic tails. Compared to the simpler systems, which only contained polymer and water, the model paint formulations were less prone
to phase separation. It is suggested that, in the paint formulation, surfactants, latex particles, pigment, and fillers increase the number of
possible association sites for the polymer hydrophobic tails. q 2000 Elsevier Science Ltd. All rights reserved.
Keywords: Hydrophobically modified polymer; Associative thickener; Ethyl(hydroxyethyl) cellulose; EHEC; Phase behavior; Viscosity; Rheology; Paint
formulation
1. Introduction
Associative polymers are a class of polymers that have
found a widespread use in technical formulations. Their
main function is as thickening compounds, and thus, they
are added in order to obtain a desired consistency. Polymers
belonging to this group are amphiphilic, containing lyophilic as well as lyophobic segments (Glass, 1989; Landoll,
1982). In an aqueous solution hydrophilic segments are
responsible for the hydration and swelling, while the hydrophobic parts of the polymer chain minimize the contact with
water by assembling in aggregates (cf. the micellization
process of surfactants) (Cabane, Lindell, Engstrom &
* Corresponding author. Tel.: 1 46-46-222-0112; fax: 1 46-46-2224413.
E-mail address: krister.thuresson@fkeml.lu.se (K. Thuresson)
Lindman, 1996). As a result, hydrophobic modification of

the polymer chains may decrease the viscosity of its
aqueous solution. This is commonly seen at low concentrations where different polymer chains have a low probability
to interact. This observation has been attributed to intraaggregation of individual polymer chains, and the accompanying reduced size of the polymer coils (Bock, Siano,
Valint & Pace, 1989; Gelman, 1987; Strauss, 1989; Tanaka,
Meadows, Philips & Williams, 1990). However, at higher
concentrations inter-aggregation becomes increasingly
more important, and already at a concentration in the vicinity of the overlap concentration, c p, of the unmodified
parent polymer the viscosity is increased substantially by
hydrophobic modification (Schaller, 1985; Semenov,
Joanny & Khokhlov, 1995; Shaw & Leipold, 1985).
The polymer used in the present investigation,
hydrophobically modified ethyl(hydroxyethyl) cellulose
0144-8617/99/$ - see front matter q 2000 Elsevier Science Ltd. All rights reserved.
PII: S0144-861 7(99)00067-3
26
L. Karlson et al. / Carbohydrate Polymers 41 (2000) 2535
(HM-EHEC), is a technically important associative thickener that can be obtained in different grades. A common
feature of different HM-EHEC fractions is that they all
are cellulose ethers to which a low amount of hydrophobic side chains have been attached by a chemical
reaction (Bostrom, Invarsson & Sundberg 1992). A typical grafting density of hydrophobic moieties is less than
5 per 100 repeating glucose units of the polymer chain.
A growing commercial interest, but also more basic
scientific interests regarding associative polymers, has
recently prompted investigations regarding HM-EHEC.
These studies dealt with the rheology and phase behavior of HM-EHEC in an aqueous solution. Focus was
on the influence by a third additive (such as salt, alcohol (Thuresson, Nilsson & Lindman, 1996), or surfactant (Joabsson, Rosen, Thuresson, Piculell & Lindman,
1998; Thuresson & Lindman, 1997; Thuresson, Lindman & Nystrom, 1997). Comparisons were always
done with corresponding solutions based on the unmodified analogue in order to trace the effects of the modification step. Less work has been devoted to effects of a
variation in the hydrophobic modification (density or
type). Therefore, we have in the present investigation
varied the length of the hydrophobic moiety in a
controlled way, and followed changes in phase behavior
and rheology of the quasi-binary solution (polymer 1
water). From a technical point of view it is important to
know how an associative thickener influences more
complex solutions. For this reason, some simple experiments (stability and rheology) in model paint formulations were also performed. Here interactions of the
polymer chains with surfactants, latex particles,
pigments, fillers, etc. can be anticipated.
2. Experimental
2.1. Materials
The cellulose used for the synthesis of the (HM-)EHEC
polymers was supplied by Borregaard (Borregaard SVS
19T). Chloroethane was supplied by Hydro Polymers AB
(Stenungsund, Sweden). Epichlorohydrin of Puriss grade
was obtained from Fluka. Tintetrachloride (SnCl4) of Puriss
grade was purchased from Merck. Epoxyethane (EO) and
nonylphenol (NP) were supplied by Akzo Nobel Surface
Chemistry AB (Stenungsund, Sweden). Dodecanol (C12),
tetradecanol (C14), hexadecanol (C16), and a mixture of
hexadecanol and octadecanol (C1618) were all obtained
from Condea. Their trade names are Nacol 1296, Nacol
1495, Nacol 1695 and Nafol 1618, respectively. All have
a purity which is better than 95% (w/w), except Nafol 1618
which is sold with the specification that beside 63 ^ 4% (w/w)
hexadecanol and 33 ^ 4% (w/w) octadecanol, the mixture
may also contain , 0.2% (w/w) C12, , 2% (w/w) C14, ,
3% (w/w) C20, and , 0.2% (w/w) C22. Diethyleneglycol
monobutyl ether (BDG) was of 97%-purity and purchased

from BDH. Acetone of pro analysis grade was obtained
from Merck.
The following components were used to formulate the
model paints; a defoamer from BYK Chemie (trade name
Byk 022), a dispersing agent from Rohm & Haas (trade
name Tamol 731), a preservative from Angus Chemie
GmbH (trade name Canguard), a filler from Omya (trade
name Hydrocarb), a pigment from Kronos AS (trade name
Kronos 2190), and a binder from Vinamul (trade name
Vinamul 3650).
2.2. Synthesis
The hydrophobically modified EHECs were synthesized
according to a standard procedure (Bostrom et al., 1992).
2.3. Purification
The purification of the synthesized polymers from byproducts originating from the reaction process was made
in several steps. First the polymer products were washed
with hot water containing 10% (w/w) Na2SO4. The temperature (958C) of the water was far above the cloud point of the
polymer. At these conditions only a small fraction of the
polymer material is soluble in water. In a recent investigation it was found that at equilibrium (at such high temperatures) the polymer-rich phase, which has a concentration of
2030% (w/w), is in equilibrium with a phase of a much
lower polymer concentration containing only about 20% dry
weight of the total polymer material (Joabsson et al., 1998).
Salt, and in particular salt with highly charged ions, like
SO422, is expected to reduce the solubility of EHEC significantly. By taking into account that the equilibrium situation
is far from reached we expect to lose insignificant quantities
of the polymer via the rinsing water, while reducing the
content of NaCl and low molecular weight poly(ethyleneglycol) (PEG). After washing with hot water each polymer
sample was dried and ground into granules with a size of ca.
0.3 mm. To get rid of remaining PEG and hydrophobic
material which not has been successfully attached to the
polymer chains, the dry granules were washed with acetone
under vigorous stirring for 30 min. After washing, the
product was collected by vacuum filtration with a Buchner
funnel. This procedure was repeated two or three times.
After the final acetone aliquot had been removed by filtration the polymer powder was dried, and later dissolved in
water to a concentration of ca. 1% (w/w). This solution was
centrifuged during 1 h at ca. 10 000g to eliminate water
insoluble impurities, such as non-modified cellulose. In
the final step of the purification procedure, the supernatant
was dialyzed against an excess of Millipore water for
several days with a repeated exchange of the dialysis
water, to get rid of small water soluble impurities (remaining salt, traces of PEG etc.). The dialysis was performed by
using a Spectra/Porw membrane tubing with a molecular

Table 1
The substitution degrees of ethyleneoxide (MSEO), ethyl (DSethyl), and
hydrophobic tails (MShydrophobe) of each of the polymer samples given as
average numbers of substituents per repeating glucose unit. Independent
repeated determinations render an uncertainty in the numerical values of
about 5%. The abbreviations given in the Hydrophobe column refers to
the unmodified parent EHEC (0), HM-EHEC modified with, nonylphenol
groups (NP), C12 groups (C12), C14 groups (C14), C16 groups (C16), and with
C16 C18 groups (C1618). Values within brackets are determined for polymer
samples after the dialysis, while the other numbers refer to the polymer
samples before the dialysis step
Hydrophobe
MSEO
DSethyl
MShydrophobe
0
NP
C12
C14
C16
C1618
2.1
2.1
2.1
2.1
2.1
2.1
0.8
0.8
0.8
0.8
0.8
0.8
0
(0.0079)
0.0086 (0.0090)
0.0082 (0.0074)
0.0081 (0.0068)
0.0091
weight cut off of 68000. After the dialysis the polymer

material was recovered by freeze-drying.
2.4. Characterization
The degrees of substitution of hydroxyethyl and ethyl
groups, MSEO 2.1 and DSethyl 0.8, were determined
by gas chromatography following degradation of the
products with HBr in glacial acetic acid (Stead & Hindley,
1969. The given values (MSEO and DSethyl) correspond to the
average numbers of EO and ethyl groups per sugar unit,
respectively. The substitution degrees of hydrophobic tails
(C12, C14, C16, C16 18), MShydrophobe, were determined by a
procedure given by Landoll (1982). In one HM-EHEC
sample (hydrophobically modified with NP), MShydrophobe
was determined by measuring the absorption of light at
275 nm with a UV-vis spectrophotometer using phenol in
aqueous solution as a reference. All numbers (MSEO, DSethyl,
MShydrophobe) for the six different polymers are summarized
in Table 1. Below the different polymers will be referred to
as EHEC, HM(NP)-EHEC, HM(C12)-EHEC, HM(C14)EHEC, HM(C16)-EHEC, and HM(C1618)-EHEC, respectively.
27
2.5. Methods
All samples (except the model paint formulations) were
prepared by weighing the different components into test
tubes which were sealed with Teflon tightened screw caps.
If possible, the samples were carefully mixed for an
extended time (several days) at a temperature where they
did not phase separate. After complete mixing each sample
was equilibrated at the temperature of interest.
The model paint formulations were prepared by slowly
adding the thickener (the (HM-)EHEC polymer) to the
water under vigorous stirring with a magnetic bar. After
about 3 h the thickener was completely dissolved. A
pigment paste was prepared by adding part of the defoamer, the dispersing agent, the preservative, the filler, and the
pigment to the thickener solution. The mixture was
dispersed with a mixer from Diat. The rotating part of the
mixer had a diameter of 4 cm and the mixing was performed
at 3000 rpm for 20 min. The binder and the remaining part
of the defoamer were then added to the pigment paste under
vigorous stirring. Finally, the paint was stirred for another
20 min (see also Table 3).
(HM-)EHEC polymers in aqueous solution have a
reversed temperature dependency and a lower critical
consolute temperature. The behavior is unexpected from
simple thermodynamic considerations, and in passing we
note that different theoretical models have dealt with this
subject (Goldstein, 1984; Karlstrom, 1985; Karlstrom,
Carlsson & Lindman, 1990; Kjellander & Florin, 1981).
Because of the concomitant scattering of light when the
solution phase separates this temperature is often referred
to as the cloud point, Tcp. In the present investigation the Tcp
was taken as the temperature where the solution first became
hazy as determined by visual inspection. The determinations
were conducted on 1% (w/w) solutions in spectrophotometer cuvettes immersed in a temperature controlled
water bath. The Tcp was approached by increasing the
temperature in steps of 28C with a waiting time of 20 min
after each temperature step to allow for thermal equilibrium
before observation of the samples.
In the phase studies the samples were carefully equilibrated and macroscopically separated at the temperature
Table 2
The relative weight of the two coexisting phases (including water) and the concentration of HM-EHEC, cp, in the top and the bottom phases in samples that
phase separated macroscopically at 308C. The values of the average hydrophobic modification degree, MShydrophobe, of the HM-EHEC chains in each phase is
also given. TP refers to top phase, BP to bottom phase and mean refers to the average value of both phases. Two numbers given in the same box
corresponds to independent phase separations and composition analyses. Upper and lower halves is for HM-EHEC substituted with C14 and C16 hydrophobic
moieties, respectively
Weight fraction, BP
C14
42% (w/w)
30% (w/w)
C16
47% (w/w)
52% (w/w)
cp, TP
cp, BP
cp, mean
MShyd, TP
MShyd, BP
MShyd, mean
0.83% (w/w)
0.83% (w/w)
1.37% (w/w)
1.33% (w/w)
1.05% (w/w)
0.99% (w/w)
0.006
0.010
0.010
0.013
0.009
0.011
0.62% (w/w)
0.59% (w/w)
1.48% (w/w)
1.35% (w/w)
1.03% (w/w)
0.99% (w/w)
0.004
0.004
0.011
0.010
0.009
0.009
28
Fig. 1. (a) The viscoelastic response (G 0 and G 00 vs. angular frequency) of a 2% (w/w) EHEC solution (0) and a 2% (w/w) HM-EHEC solution modified with
C14 chains (C14). The cross over angular frequencies used to calculate the characteristic times t p are indicated. For the EHEC solution the cross over was
estimated by a linear extrapolation of G 0 and G 00 . Note that G 0 and G 00 for HM(C14)-EHEC (upper set of curves) have been shifted upwards. The insert shows the
suggested evolution of G 0 and G 00 by the Maxwell model (see text). G 1 Pa and t p 1 s were used in the calculation. The measurements were performed at
38C. (b). The complex viscosity, h p, from oscillatory shear experiments vs. angular frequency for 2% (w/w) aqueous solutions of the different (HM-)EHECs.
The angular frequency, 2p f #, used to calculate the characteristic time t # is indicated for the solution prepared with HM(C14)-EHEC. The full lines represents,
from bottom to top, unmodified EHEC (0), HM-EHEC modified with nonylphenol groups (NP), with C12, C14 and C16 groups. The measurements were
performed at 38C.
of interest (308C). If needed, the separation process was

speeded up with a centrifuge. During centrifugation the
temperature was kept at 30 ^ 0.58C. The polymer concentration in each phase was determined by freeze-drying and
weighing. The average hydrophobic modification degree of

the polymer material in each phase was determined by either
of the two procedures described above.
The rheological measurements were performed with a
StressTech rheometer from Rheologica equipped with a

4 cm, 18 cone and plate system. The temperature of the
sample during a rheological measurement was controlled
to within ^ 0.18C by an external water bath connected to
the measuring geometry. Measurements were performed at
(HM-)EHEC concentrations of 1% (w/w) or 2% (w/w). For
samples with a polymer concentration of 1% (w/w), and for
measurements on the solvent mixtures containing BDG and
water at different mixing ratios, the rheometer was put in the
constant shear mode. For these solutions the Newtonian
plateau, characterized by a shear rate independent viscosity,
was always reached.
In order to slow down polymer dynamics and to facilitate
the determination of the cross-over times, t p and t #,
(defined below) a (HM-)EHEC concentration of 2% (w/w)
was employed. The rheometer was put in the oscillatory
shear mode, and all measurements were performed in the
linear regime, where the response was independent of the
applied stress. An oscillatory shear measurement reports the
storage, G 0 , and loss, G 00 , moduli as a function of the
frequency, f, of the oscillation.
The complex viscosity, h p,
p
p
was calculated as h G2 1 G 002 =2pf . The low frequency
limit characterized by a Newtonian behavior (viscosity
independent of f) was reached for all samples except for
the aqueous solution containing 2% (w/w) HM(C16)EHEC (see Fig. 1(b)).
In this investigation two methods, which both use a crossover from a viscous to an elastic behavior of the solution,
have been used to indirectly probe polymer dynamics. t p is
the characteristic time corresponding to the inverse angular
frequency, 1/2pf p, where G 0 G 00 . At frequencies higher
than f p the response is mainly elastic. An elastic response
indicates that on the experimental time scale, the polymer
chains do not have the time to relax to the new equilibrium
position due to the applied deformation. In the low-viscous
samples (i.e. the samples prepared with EHEC, HM(NP)EHEC, and HM(C12)-EHEC) this frequency is not directly
accessible with the present rheometer. Here, the presented t p
values have been estimated by linearly extrapolating the
evolution of G 0 and G 00 vs. frequency in a loglog representation, Fig. 1(a). An alternative way to probe the crossover from
a viscous to an elastic behavior of the solution is to determine
the angular frequency at which the viscosity profile (h p vs.
2pf) changes from being Newtonian to shear thinning, Fig.
1(b). The characteristic time, t# 1=2pf # , corresponding to
the inverse of this angular frequency, is for the present samples
longer than t p, and the two methods become complementary. Indeed t p and t # follow the same trend (cf. Fig. 6).
If the rheological data from a polymer system are determined by only one relaxation time, the simplest model of a
viscoelastic fluid, namely the Maxwell model, can be used to
describe the frequency dependencies of the dynamic moduli.
Then the evolutions of G 0 and G 00 with f are given by:
G 0 G
2pf 2
1 1 2pf 2
and
G 00 G
2pf
:
1 1 2pf 2
29
Here G is the plateau value of G 0 at high frequencies. In this

model t p equals t # In particular it follows from the Maxwell
model that log(G 0 ) has the slope 1 2 as a function of log(f) and
log(G 00 ) has the slope 1 1 in the terminal zone (low frequencies), while at frequencies above t p, log(G 00 ) has the slope 2 1
while G 0 approaches G . However, this model cannot satisfactorily describe our data as is evident from the different
shapes of the curves (compare data in Fig. 1(a) with insert).
Rather the frequency dependencies of the dynamic moduli
indicate that a range of relaxation times has to be used to
give a proper representation. In light of the expected polydispersity regarding the molecular weight as well as the chemical
structure of the (HM-)EHEC polymers this is not surprising.
Thus, the reason for the differences in t p and t # can probably
be traced to the fact that the relaxation in solutions of
(HM-)EHEC is due to a range of different processes with
different characteristic times.
In the industry one method to characterize paints is to
measure the viscosity at one low and one high shear rate.
The two values are referred to as the Stormer and ICI
viscosities, respectively. Here the Stormer viscosity was
measured with a Stormer viscometer at a shear rate in the
range 10100 s 21. The values are given in Krebs units
(KU). The conversion of the numerical values from KU to
the SI-quantity Pa s can to our knowledge only be made
with an approximate formula (Reisser, M. & Oleinitec,
A.B., private communication). 1 These measurements were
made in accordance with the ASTM D 562-81 standard. The
ICI viscosity was obtained at a shear rate corresponding to
approximately 12 000 s 21 and is usually given in Poise.
However, here we have chosen to present these data in
Pa s to facilitate comparison with other measurements
(10 P 1 Pa s). The measurements were made in accordance with the ASTM D 4287-94 standard.
The model paints were characterized according to color
acceptance (ASTM D 5326), leveling (ASTM D 4062-81),
spatter resistance, gloss (ASTM D 523-89), and storage
stability. For the color acceptance test a universal colorant
was added to the paint formulation. After shaking the paint
for 5 min it was applied to a white chart. A rub-out test in
which one part of the surface of the painted chart was
rubbed while another part was untouched was performed.
The paint was judged by means of color differences between
the two different parts, and by a comparison with standard
charts the paint formulation was given a rating of 110; 1
corresponds to a poor resistance, while 10 to an excellent
resistance to mechanical rubbing.
Leveling of the model paint formulations was measured
by the Lenata draw-down method. The results were
1
Over the viscosity range of 0.22.1 Pa s the following equation may be
used to convert Stormer viscosity values from Krebs units to Pa s: ln(KU)
1.1187 1 0.8542 ln(193.8 Pa s 1 36) 2 0.0443[ln(193.8 Pa s 1 36)] 2.
Over the viscosity range of 2.15.0 Pa s it is more appropriate to use:
ln(KU) 1.8118 1 0.596 ln(193.8 Pa s 1 36) 2 0.0206[ln(193.8 Pa s 1
36)] 2. Here KU corresponds to the viscosity at 258C in Krebs units, while
Pa s is the viscosity at 258C in Pa s.
30
Fig. 2. The cloud point of a collection of different EHEC samples (1% (w/w)) plotted against the fraction of hydroxyethyl groups, MSEO/(MSEO 1 DSethyl). The
data is reproduced from Thuresson et al. (1995). An open circle and a filled circle represent the unmodified EHEC and the HM(NP)-EHEC corresponding to the
samples in the present investigation, respectively. The dashed line is only drawn as a guide to the eye.
compared with standards and given ratings from 1 to 10

where 1 is a low degree of leveling and 10 is an excellent
degree of leveling.
The spatter resistance was measured by applying the paint
formulation with a roller onto a wall with an area of
0.25 m 2 at a frequency of 60 strokes/min. The spatter was
collected on a test chart located below the rolled surface,
and the result was compared with a standard. Again the
paint formulation was given a rating from 1 to 10 where 1
is a poor performance and 10 corresponds to a low spatter
and excellent spatter resistance.
To evaluate the gloss properties of the paint formulation it
was applied to a glass panel. The reflectance was measured
after 10 days at an angle of 608 and is reported in percentage
of the reflectance of a standard surface of polished glass.
Fig. 3. The cloud point, Tcp, for the different (HM-)EHEC samples in 1%
(w/w) aqueous solutions. From left to right; unmodified EHEC (0), HMEHEC modified with nonylphenol groups (NP), with C12, C14 and C16
groups, and with C16 C18 groups(C1618). The one phase region of the last
sample, HM(C1618)-EHEC, is not experimentally accessible (Tcp , 08C).
The storage stability of the paint was estimated by a

repeated measurement of the Stormer viscosity after 28
days of storage at 508C.
3. Results and discussion
This section is divided into three parts, each of which
contains a discussion of the effect of hydrophobic modification of the polymer chains.
3.1. Phase behavior
It has been found that for EHEC (without hydrophobic
modification) the Tcp can be correlated to MSEO and DSethyl
in a way such that Tcp increases with increasing MSEO, while
it decreases with increasing DSethyl, Fig. 2 (Thuresson, Karlstrom & Lindman, 1995). It was also found that the conversion of EHEC to HM(NP)-EHEC had a drastic influence on
Tcp. This is also found in the present investigation, Fig. 3.
The longer the aliphatic chain, the more pronounced is the
shift in Tcp (relative to that of the unmodified EHEC). It is
important to note that the overall hydrophilic/hydrophobic
balance is only slightly changed by a variation in hydrophobic tail length, and consequently this cannot be the explanation to the pronounced shifts in Tcp. It is therefore reasonable
to invoke a mechanism in which the phase separation is
influenced by the association of polymer hydrophobic tails
into micellar aggregates (Thuresson & Joabsson, 1999).
These connect different polymer chains into a network.
The swelling of the polymer matrix is then restricted by
finite extension of the part of the cellulose chains connecting
different micellar aggregates. This favors the formation of
one phase that is concentrated in polymer. This phase is in
equilibrium with a phase depleted in polymer (cf. a bridging
flocculation in a polymer/particle system or the restricted
31
Table 3
Recipe for the preparation of the model paints formulation. x is the weight
per mille (w/w) of thickener in the model formulation. The different
components are added to the mixture in the same order as listed in the
table (top to bottom). Note that the defoamer is added in two portions.
For details, see Section 2.5
Water ( w/w)
Thickener, HM-EHEC ( w/w)
Defoamer, Byk 022 ( w/w)
Dispersing agent, Tamol 731 ( w/w)
Preservative, Canguard ( w/w)
Filler, Hydrocarb ( w/w)
Pigment, Kronos 2190 ( w/w)
Binder, Vinamul 3650 ( w/w)
Defoamer, Byk 022 ( w/w)
P
( w/w)
Fig. 4. (a) The bar to the left in each collection (gray) reports the viscosity
of the polymer in water, and the bar to the right (striped) reports the
viscosity in a 80:20 mixture of water and BDG (see text). To compensate
for the higher viscosity of the solvent (water/BDG compared to water), the
numerical values for the viscosity in water/BDG mixtures were divided
with 2.7 before presentation (cf. insert Fig. 5). The line reports on the
Q-value, which is the ratio between the viscosity of the aqueous polymer
solution and the viscosity of the polymer dissolved in the water/BDG
mixture. The different groups of bars correspond from left to right to,
unmodified EHEC (0); EHEC modified with nonylphenol groups (NP),
with C12, C14, and C16 groups. All solutions contain 1% (w/w) polymer
and the measurements were performed at 38C. The reported values correspond to the viscosity at the Newtonian plateau and they were obtained with
the rheometer put in the constant shear mode. (b) Same as Fig. 4(a), but at
208C. At this temperature it was impossible to measure the viscosity of the
aqueous solution of HM(C16)-EHEC due to a phase separation. Note that
the solution of this polymer in the water/BDG mixture has a one phase
behavior.
swelling of a covalently crosslinked gel). The outlined

scenario requires the hydrophobic associations to be strong
enough to overcome the decrease in entropy that accompanies the formation of the phase concentrated in polymer.
However, the unfavorable reduction of the entropy is significantly reduced if both phases contain approximately the
same polymer concentration. This has recently been
discussed by Annable and Ettelaie (1994) in an experimental and theoretical study, based on a simple FloryHuggins
approach. They discovered that a mixture of a hydrophobically modified polymer with its unmodified analogue might
phase separate. After the phase separation the two different
243.2 2 x
x
2
6.5
1
110
180
454.3
3
1000
polymers are enriched in separate phases. Below we will

refer to this behavior as a segregative phase separation, or
alternatively that the polymers phase separate segregatively
(Piculell & Lindman, 1992).
Based on this knowledge, mixtures prepared with the
HM(C16)-EHEC and the HM(C14)-EHEC polymers, respectively, were macroscopically phase separated at 308C into
two liquid phases in equilibrium. In each sample, the two
phases had similar volumes (Table 2). Each phase was
analyzed with respect to polymer concentration and the
average hydrophobic modification degree. Despite the
uncertainty in the data, the trend is clear and it was found
that the samples separated segregatively with one phase
containing HM-EHEC chains with a higher substitution
degree than the chains residing in the other phase. Thus,
the HM-EHEC sample contains a range of polymer molecules differing in molecular weight, a normal situation for
all polymer samples, and also in chemical composition, and
the binary aqueous solution has a behavior reminiscent of
a multi-component system. This conclusion was also drawn
in a previous publication in which the phase behavior of
EHEC and HM(NP)-EHEC at high temperatures was investigated (Joabsson et al., 1998). In connection with this
discussion it is valuable to comment on the fact that
although Tcp for a 1% (w/w) HM(C14)-EHEC solution is
reported to be ca. 328C (Fig. 3), a macroscopic segregative
phase separation is observed already at 308C (Table 2). This
reflects the difficulties in observing this type of phase
separation by means of the clouding phenomenon. We
think the disagreement is due to the fact that in the vicinity
of the phase separation temperature the two phases have
similar compositions (segregative phase separation with
similar polymer concentrations), differing mainly in a
small variation of hydrophobic modification as shown
above. Thus, at such conditions the two phases can be
expected to have a similar refractive index and light is not
scattered (no clouding is observed). The conjecture is that
there is a risk of overestimating the true phase separation
temperature by visual detection of Tcp. In the present HMEHEC systems we estimate the uncertainty in the phase
32
Fig. 5. The influence on viscosity of the composition of the solvent in solutions of associating HM-EHECs as exemplified with 1% (w/w) solutions of HMEHEC modified with nonylphenol groups (NP) or with C14 groups (C14). It can be seen that the viscosity levels off at an approximate weight mixing ratio of
BDG:water corresponding to 15:85. The data are presented as a relative viscosity (the viscosity of the polymer solution divided with the viscosity of the solvent
mixture). The insert reports the viscosity of the BDG:water mixture at different compositions. All measurements were performed at 208C with the rheometer
put in the constant shear mode.
separation temperature, by using the clouding phenomena,

to ^ 58C.
3.2. Rheology in quasi-binary solutions
Usually the expectation is that the viscosity of an aqueous
solution of a hydrophobically modified polymer increases
with the length of the hydrophobic tails. The higher viscosity has been ascribed to slower motions of individual polymer chains as the residence time of the hydrophobic tails in
polymer micelles increases (Annable, Buscall, Ettelaie &
Whittlestone, 1993; Leibler, Rubinstein & Colby, 1991).
The present results follow that trend. Measurements
performed on aqueous solutions at two different temperatures (3 and 208C) are reported in Fig. 4. The viscosity for
the HM(C16)-EHEC solution at 208C could not be measured
as a natural consequence of the phase separation which
occurred already at a much lower temperature (cf. Fig. 3).
Fig. 4 also gives the viscosity of the different polymers in a
solvent mixture composed of 80% (w/w) water and 20% (w/
w) BDG. The knowledge that a saturation level of BDG
usually is observed was decisive when the ratio between
water and BDG was chosen. This is exemplified in Fig. 5,
where data for 1% (w/w) HM(C14)-EHEC and 1% (w/w)
HM(NP)-EHEC solutions are presented. Initially the viscosity decreases strongly, but at about 15% w/w BDG the
decrease levels off and above that BDG concentration the
relative viscosity of both the investigated polymer solutions
is constant within the experimental uncertainty. The effect
of BDG can be rationalized by comparison with the wellknown influence of surfactants on aqueous solutions of
hydrophobically modified polymers. These are regarded
to break intermolecular hydrophobic associations by
individually dissolving hydrophobic patches (or hydrophobic tails) within micellar-like aggregates (Piculell, Thuresson & Ericsson, 1995). Indeed, BDG has an amphiphilic
structure and is expected to adsorb at polar/non-polar interfaces. As a consequence of decreased hydrophobic associations, addition of BDG can also be anticipated to decrease
the tendency to segregative phase separation. In particular
the phase separation of an aqueous solution of HM(C16)EHEC was inhibited in the water/BDG mixture and the
viscosity of that solution could be determined (Fig. 4(b)).
It is also interesting that all (HM-)EHEC polymers, independent of the nature of the polymer hydrophobic tail, give
virtually the same viscosity when they are dissolved in the
water/BDG mixture. This indicates that the different
(HM-)EHEC polymers have similar molecular weights,
and that the differences observed between the aqueous solutions can be referred to variations in the aggregation process
of the polymer hydrophobic tails. Thus, the Q-value, which
is the ratio between the value of the (Newtonian) viscosity
in water to that observed in water/BDG, can be regarded as a
phenomenological measurement of the influence of hydrophobic associations on the viscosity of the aqueous solution.
In this way different polymer samples (variation in chemical
structure of the hydrophobic tails, variation in modification
degree, variation in modification pattern etc.) can be ranked.
In line with this, we note that the unmodified EHEC has a Qvalue close to 1. However, comparisons of Q-values should
be done with care because the evolution of Q with polymer
molecular weight has not been investigated.
Generally a shift towards longer relaxation times is
expected when the polymer chains in a solution become
more entangled (increased polymer concentration or higher
polymer molecular weight) or when the life time of
33
Table 4
Summary of the observations in the model paint formulations. The different columns refer to unmodified EHEC (0), HM-EHEC modified with nonylphenol
groups (NP), with C12, C14, and C16 groups, and with C16 C18 groups (C1618). The numbers given for color acceptance, spatter resistance, and leveling are
ratings obtained by visual comparisons to standards. 1 is poor and 10 is excellent. x is the (w/w) of the thickener that is needed to obtain the desired Stormer
viscosity
x ( w/w)
Stormer viscosity (KU)
ICI viscosity (Pa s)
Colour acceptance
Spatter resistance
Levelling
Gloss (%)
Storage stability (KU)
Consistency
NP
C12
C14
C16
C1618
7.5
110
0.18
10
4
1
24
123
flowing
4.5
113
0.15
8
7
1
27
120
flowing
5.75
112
0.16
6
7
1
26
119
flowing
4.5
112
0.11
6
7
1
24
120
flowing
4
113
0.10
6
7
1
26
121
jelly
3.9
109
0.10
6
6
1
24
119
jelly
intermolecular hydrophobic associations becomes longer

(in case of hydrophobically modified polymers). Extracted
from an oscillatory shear experiment, which reports indirectly on polymer dynamics, the change in the characteristic
times, t p and t #, may be taken to reflect such variations. At
experimental times exceeding t p and t # the viscous behavior of the solution dominates, while at shorter times the
elastic behavior prevails. Indeed, t p and t # increase with an
increasing length of the hydrophobic tail (Fig. 6). We stress
that t p and t # not are expected to correspond to any characteristic time of any process on a molecular level, but we
do believe that they capture trends correctly and therefore
can be used to compare different HM-EHEC polymers. We
note that a shift towards longer relaxation times, following
an increased hydrophobicity of the hydrophobic tails, will
on a fixed experimental time scale (as in applications) show
up as a more elastic behavior of the solution.
influence the aggregation process. The phase separation

may be inhibited by an increased number of sites where
the polymer hydrophobic tails can adsorb. However, at
this stage it is too early to draw any further conclusions.
Another interesting result, which follows from Table 4, is
that HM-EHEC with long hydrophobic tails give paints with
a more elastic consistency as compared with paints prepared
with thickeners with shorter hydrophobic tails. This correlates to the slower relaxation process with increasing
strength of the inter-molecular associations which was
observed as a higher elasticity by oscillatory shear measurements already in the binary solutions. The slower relaxation process of solutions prepared with HM-EHEC modified
with long aliphatic chains is also reflected in the viscosity
values reported in Table 4. The more pronounced shearthinning behavior of solutions of HM-EHEC comprising
long hydrophobic tails follows from lower ICI viscosity
values.
3.3. Model paint formulations

A recipe of the formulations is given in Table 3. The
concentration of (HM-)EHEC in the final mixtures varies
because a paint is formulated aiming at a certain (Stormer)
viscosity, rather than a certain polymer concentration (Table
4). The amount of polymer that is needed decreases when
the polymer hydrophobic tails becomes longer. In other
words; HM-EHECs with strongly associating hydrophobic
tails have a high thickening efficiency. It is interesting to
note that despite the rather extreme conditions which the
model paint formulations were exposed to (508C for 28
days was used in the storage stability test) no phase separation was observed. Rather the model paint formulations had
a good stability and only minor changes in the viscosity with
time could be detected (compare the storage stability
column with the Stormer viscosity column in Table 4).
This is in direct contrast to the results obtained in the
simpler binary solutions. We recall that certain HMEHEC solutions phase separated already at room temperature (see Fig. 3). This can be taken as an indication of the
fact that surfactants, pigments, latex particles, fillers etc.
Fig. 6. The variation of t p (to the left in each collection) and t # (to the right
in each collection) with the length of the hydrophobic tail. The cross over
times were determined as illustrated in Fig. 1. From left to right; unmodified
EHEC (0), HM-EHEC modified with nonylphenol groups (NP), with C12,
C14, and C16 groups. The line over each bar represents an estimated error.
The polymer concentration was kept constant at 2% (w/w) and the temperature was 38C.
34
4. Conclusions
One of the reasons why hydrophobically modified polymers have appeared on the market, and sold as thickeners, is
that they often have been found to provide a better thickening efficiency in aqueous solutions than their unmodified
analogues. In this paper we have in a systematic way varied
the chemical structure of the hydrophobic moieties that
were grafted to the polymer backbone when ethyl(hydroxyethyl) cellulose, EHEC, was converted to obtain the hydrophobically modified HM-EHEC. By changing the length of
hydrophobic tails the strength of the associations between
different polymer molecules was modulated. At the conditions in the present investigationHM-EHEC with a
substitution degree of MShydrophobe 0.008 dissolved in an
aqueous solution to a concentration of 1% (w/w)we
observed that the hydrophobic tails should contain more
than 12 carbon atoms to give a thickening efficiency significantly better than that of the unmodified analogue (see Fig.
4(a)). It was also found that the strong hydrophobic
associations that accompanied the modification with long
hydrophobic tails resulted in solutions that were more
elastic on a given time scale. This may be a problem in
applications (such as water borne paints). Thus, in the
design of an effective associative thickener, other aspects
apart from high viscosity following strong hydrophobic
associations have to be taken into account, and it is not
sufficient to rank different HM-polymers only by thickening
efficiency.
The literature reports that, provided inter molecular associations are strong enough; increasing water content of an
aqueous solution of HM-EHEC induces a phase separation.
Under such circumstances one of the phases contains
virtually pure water, while most of the polymer material
can be found in the other phase. The observation was rationalized in terms of a behavior reminiscent of restricted swelling of a covalently bonded gel (Thuresson & Joabsson,
1998). In the present investigation we found that also a
modification of the polymer chains with tails of a quite
modest hydrophobicity can induce a phase separation. The
reason is an inhomogeneous distribution of hydrophobic
tails among the polymer chains. At such conditions, the
solution has a tendency to phase separate segregatively
into one phase containing polymer chains with a high substitution degree, and the other phase containing polymer
chains with a lower substitution degree. The mechanism is
related to that of the restricted swelling, but in addition the
tendency towards phase separation is increased because
both phases have similar polymer concentration.
Finally, it is interesting to note that when the HM-EHEC
polymers were in model paint formulations the tendency to
segregative phase separation was decreased. One explanation to this observation can be that surfactants, latex particles, pigment, and fillers supply association sites for the
polymer hydrophobic tails. This is expected to promote a
single-phase behavior if the reason for a phase separation
can be traced to a restricted swelling of the HM-EHEC

matrix.
Acknowledgements
F.J. and K.T. thank the Centre for Amphiphilic Polymers
(CAP) for financial support. Dr. Lennart Piculell is gratefully acknowledged for comments on this manuscript.
References
Annable, T., Buscall, R., Ettelaie, R., & Whittlestone, D. (1993). The
rheology of solutions of associating polymers: comparison of experimental behavior with transient network theory. Journal of Rheology, 37
(4), 695726.
Annable, T., & Ettelaie, R. (1994). Thermodynamics of phase separation in
mixtures of associating polymers and homopolymers in solution.
Macromolecules, 27, 56165622.
Bock, J., Siano, D. B., Valint Jr, P. L., & Pace, S. J. (1989). Structure and
properties of hydrophobically associating polymers. In J. E. Glass (Ed.),
Polymers in aqueous media, Advances in Chemistry (pp. 317411).
Washington, DC: American Chemical Society.
Bostrom, P., Ingvarsson, I., & Sundberg, K. (1992). Water soluble nonionic
cellulose ethers and their use in paints. US Patent 5140099, Berol Nobel
AB, Stenungsund, Sweden.
Cabane, B., Lindell, K., Engstrom, S., & Lindman, B. (1996). Microphase
Separation in polymer 1 surfactant systems. Macromolecules, 29,
31883197.
Gelman, R.A. (1987). Hydrophobically modified hydroxyethylcellulose.
TAPPI International Dissolving Pulps Conference, Geneva, Switzerland, pp. 159165.
Glass, J. E. (1989). Polymers in aqueous media, Advances in Chemistry.
Goldstein, R. E. (1984). On the theory of lower critical solution points in
hydrogen-bonded mixtures. Journal of Physical Chemistry, 80, 5340
5341.
Joabsson, F., Rosen, O., Thuresson, K., Piculell, L., & Lindman, B. (1998).
Phase behavior of a clouding nonionic polymer in water. Effects of
hydrophobic modification and added surfactant on phase compositions.
Journal of Physical Chemistry B, 102 (16), 29542959.
Karlstrom, G. (1985). A new model for upper and lower critical solution
temperatures in poly(ethylene oxide) solutions. Journal of Physical
Chemistry, 89 (23), 49624964.
Karlstrom, G., Carlsson, A., & Lindman, B. (1990). Phase diagrams of
nonionic polymerwater systems. Experimental and theoretical studies
of the effects of surfactants and others cosolutes. Journal of Physical
Chemistry, 94 (12), 50055015.
Kjellander, R., & Florin, E. (1981). Water structure and changes in thermal
stability of the system poly(ethylene oxide)water. Journal of Chemical Society, Faraday Transactions 1, 77, 20532077.
Landoll, L. M. (1982). Nonionic Polymer Surfactants. Journal of Polymer
Science, Part A: Polymer Chemistry, 20, 443455.
Leibler, L., Rubinstein, M., & Colby, R. H. (1991). Dynamics of reversible
networks. Macromolecules, 24 (16), 47014707.
Piculell, L., & Lindman, B. (1992). Association and segregation in aqueous
polymer/polymer, polymer/surfactant and surfactant/surfactant
mixtures: similarities and differences. Advances in Colloid and Interface Science, 41, 149178.
Piculell, L., Thuresson, K., & Ericsson, O. (1995). Surfactant binding and
micellation in polymer solutions and gels: binding isotherms and their
consequences. Faraday Discussions, 101, 307318.
Schaller, E. J. (1985). Rheology modifiers for water-borne paints. Surface
Coatings Australia, 22, 613.

Semenov, A. N., Joanny, J. F., & Khokhlov, A. R. (1995). Associating
polymers: equilibrium and linear viscoelasticity. Macromolecules, 28,
10661075.
Shaw, K. G., & Leipold, D. P. (1985). New cellulosic polymers for rheology control of latex paints. Journal of Coating Technology, 57 (727),
6372.
Stead, J. B., & Hindley, H. (1969). A modified method for the analysis of
oxyethylene/oxypropylene copolymers by chemical fission and gas
chromatography. Journal of Chromatography, 42, 470475.
Strauss, U. P. (1989). Hydrophobic polyelectrolytes. In J. E. Glass (Ed.),
Polymers in aqueous media, Advances in Chemistry (pp. 317324).
Tanaka, R., Meadows, J., Phillips, G. O., & Williams, P. A. (1990). Viscometric and spectroscopic studies on the solution behaviour of hydrophobically modified cellulosic polymers. Carbohydrate Polymers, 12,
443459.
Thuresson, K., & Joabsson, F. (1999). Phase separation induced by dilution
35
in a system of a hydrophobically modified polymer. Colloids and

Surfaces, A: Physicochem. Engng Aspects, 151, 513523.
Thuresson, K., Karlstrom, G., & Lindman, B. (1995). Phase diagrams of
mixtures of a nonionic polymer, hexanol and water. An experimental
and theoretical study of the effect of hydrophobic modification. Journal
of Physical Chemistry, 99, 38233831.
Thuresson, K., & Lindman, B. (1997). Effect of hydrophobic modification
of a nonionic cellulose derivative on the interaction with surfactants.
Phase behaviour and association. Journal of Physical Chemistry, 101,
64606468.
Thuresson, K., Lindman, B., & Nystrom, B. (1997). Effect of hydrophobic
modification of a nonionic cellulose derivative on the interaction with
surfactants. Rheology. Journal of Physical Chemistry, 101, 64506459.
Thuresson, K., Nilsson, S., & Lindman, B. (1996). Influence of cosolutes on
phase behaviour and viscosity of a nonionic cellulose ether. The effect
of hydrophobic modification. Langmuir, 12 (10), 24122417.
Cyclodextrins in HM-PEG Solutions.

Inhibition of Polymer-Polymer Associations
L. Karlson,*,1 K. Thuresson,2 and B. Lindman.2
1
Physical Chemistry 1, Center for Chemistry and Chemical Engineering, Lund University,
P.O. Box 124, SE-221 00 Lund, Sweden.

* To whom correspondence should be addressed.
fax: +46 303 839 21
e-mail: leif.karlson@akzonobel.com
Abstract
In an aqueous solution of a hydrophobically end-modified PEG-polymer, HM-PEG, the
thickening effect is dependent on intermolecular hydrophobic associations and the formation
of a network structure. In the present investigation cyclodextrin, CD, has been added to an
aqueous HM-PEG solution and a decrease in Newtonian viscosity has been followed. The
decreased viscosity is referred to polymer-polymer associations becoming less numerous
when complexes between polymer hydrophobic tails and CD become more frequent;
CD-decorated polymer hydrophobic tails have no possibility to contribute to the network. It
was found that deactivation of the first few hydrophobic tails has very large consequences for
the viscosity. A termination of a fraction as small as 10% (or below) of the total amount of
polymer hydrophobic tails may reduce viscosity to a level almost corresponding to that of the
unmodified parent polymer. This can be understood by taking into account that a solution of
a HM-PEG polymer is expected to be inhomogeneous with large concentration fluctuations,
and that the viscosity is likely to be strongly decreased by reducing the probability for
hydrophobic associations responsible for connecting different clusters. The effect of CD on
rheology is very different for different architectures of hydrophobically modified polymer, in
particular between graft copolymers and end-capped ones. This can be understood from the
differences in net-work structure.
-1-
Introduction
Hydrophobically modified polymers (HM-P:s) or water-soluble associative polymers
(WSAP:s) are used in a variety of technical formulations that we meet in our daily life.
Examples are water-borne paints and shampoos.1,2 A common reason for adding a HM-P to
a formulation is that it gives a different rheological behavior as compared to a normal
hydrophilic thickener. Another reason to choose a HM-P is that the amphiphilic properties
can help to increase stability of dispersions.
To be able to predict performance of a hydrophobically modified polymer in an application it
is important to have knowledge about the effect of the hydrophobic modification on
macroscopic properties. Such knowledge is also interesting for more fundamental reasons
and is often correlated with molecular interactions when HM-P:s are subject to more basic
research. One way to obtain information about the effect of a hydrophobic modification is to
synthesize both the unmodified and the hydrophobically modified version of the thickener.
This is a route that has been employed in several investigations.3-9 Besides that this
requires extra synthesis work it is difficult to control the reaction so that the only difference
between the two polymers is the hydrophobic modification. Therefore it would be desirable if
the effect caused by a hydrophobic modification instead could be studied by inhibition and
decoupling of the hydrophobic associations.
In a recent paper, we reported on the addition of diethyleneglycol monobutylether (BDG) to
aqueous polymer solutions to obtain such decoupling.9 In other investigations, surfactants
have been added at high concentration, which decouples hydrophobic polymer-polymer
associations by encapsulating each polymer hydrophobic tail in a micelle.10,11 Another
efficient way, which will be employed here, to decouple the hydrophobic associations is
offered by addition of cyclodextrin (CD). CD is a cyclic oligomer of glucose with the shape of
a truncated cone that has a hydrophilic exterior and a hydrophobic cavity in the center,
Figure 1. Three different sizes are available; -, -, and -cyclodextrin consist of six, seven,
or eight glucose units, respectively. In aqueous solutions CD molecules form nut and bolt
(or inclusion) complexes with substances containing lipophilic groups, e.g. surfactants or
HM-P:s, provided that the hydrophobic group has a shape that fits in the cavity. Principles of
such complex formation have been studied thoroughly by several groups, and in particular
has complexation between CD and surfactants been in focus.12-18
-2-
O
O
O
OH HO
O
HO
O
O
OH
O
O
Hydrophobic cavity
HO
O
O
OH
HO
O
O
C
C O
H
O
O
O
H C
H
H C H
O
H
Figure 1. To the left is shown the chemical structure and to the right a schematic
representation of the geometry of an -cyclodextrin molecule.
CD has also been used together with HM-P with the purpose to reduce the viscosity of an
aqueous solution. This was described for the first time in the beginning of the nineties.19,20
Here the purpose was to reduce viscosity of a highly concentrated aqueous solution of a
HM-P thickener to facilitate incorporation of the thickener into technical formulations, e.g.
paint. More recently the concept of controlling the self-association of HM-P by addition of CD
has been more thoroughly studied in two papers by Zhang et al and Akiyoshi et al. .21,22 A
conclusion from these investigations was that CD molecules form complexes with the
hydrophobic end groups of the HM-P polymers, which means that polymer hydrophobic tails
are hidden within the CD-cavities, and only the hydrophilic outer shell of the CD molecules is
exposed to the aqueous environment. In HM-P solutions the viscosity then decreases since
the three-dimensional polymer network is disrupted when the possibility to form interpolymeric hydrophobic associations is reduced. The concept to encapsulate the hydrophobic
tails of HM-P with CD has also been used to avoid disturbing hydrophobic interactions when
properties of individual polymer molecules were in focus and investigated with techniques
such as gel permeation chromatography and static light scattering.23
In a recent investigation we focused on how the viscosity in an aqueous solution of a graft
copolymer, hydrophobically modified ethyl (hydroxy ethyl) cellulose (HM-EHEC), decreased
in the transition region where the concentration of CD (cCD) was lower than the total
concentration of hydrophobic groups (chydrophobe), and the effect of various cyclodextrins was
-3-
investigated.24 The complex formation and the concomitant disruption of the polymer
network was followed by measuring the viscosity as a function of the CD concentration. The
data were rationalized in a simple association model from which the effective binding
constant could be extracted together with the concentration of binding sites. We found a
reasonable agreement between the model and the data, and in particular the concentration
of binding sites equaled the concentration of hydrophobic tails. The interpretation of this was
that all hydrophobic tails were important for the network formation in the HM-EHEC system.
In the present study we have used methylated -cyclodextrin (M--CD) to decouple
associations in aqueous solutions of an associating polymer with a different architecture, a
water-soluble
polymer
hydrophobically
end-capped,
the
choice
of
polymer
being
hydrophobically modified poly (ethylene glycole), HM-PEG. The viscosity has been
rationalized within the same model as we used for the HM-EHEC/CD system, and it was
found that initially the decrease in viscosity is much stronger than what is expected from
results in the previous investigation. The observations can be understood by taking into
account that the HM-PEG solution is likely to be inhomogeneous with large concentration
fluctuations.
Experimental
Materials
Hydrophobically end-modified polyethylene glycol (HM-PEG) with the structure C1618-EO140IPDU-EO140-C1618 was used in this study. C1618-EO140 denotes an ethoxylate of a mixture of
unsaturated alcohols (C16 to C18), and IPDU represents an isophorone diurethane group
connecting two ethoxylated alcohol molecules. The synthesis and characterization methods
are described elsewhere.25 The weight average molecular weight (Mw = 13.500) and the
polydispersity index (Mw / Mn = 1.1) were determined by size exclusion chromatography
(SEC). The HM-PEG was purified from low molecular weight impurities (salt originating from
the catalyst, low molecular weight PEG etc.) by dialysis; a 3% w/w solution of HM-PEG was
dialyzed against an excess of Millipore water for several days with repeated exchange of the
dialysis water. The dialysis was performed by using Spectra/Por molecularporous
membrane tubing with a molecular weight cut off of 6-8 000. After the dialysis the polymer
material was recovered by freeze-drying.
Methylated -cyclodextrin (M--CD) was supplied by Wacker-Chemie (under the trade name
Cyclodextrin Alpha W6 M1.8). The degree of methylation per glucose unit was 1.6 1.9, as
-4-
given by the supplier. The M--CD was of pharmaceutical quality and was used without
further purification. To all samples water of Millipore quality was used.
Methods.
Aqueous HM-PEG solutions of three concentrations, 3, 5 or 10%w/w, were prepared, without
and with 0.5%w/w M--CD. These six stock solutions were prepared by weighing the
components in test tubes that were sealed with Teflon tightened caps. Before proceeding,
the stock solutions were left to equilibrate for at least 24 hours. Furthermore, in order to
facilitate preparation of test solutions with a CD concentration below 0.1%w/w, a dilution with
respect to CD of the more concentrated stock solutions was made. From these, in all nine
stock solutions, samples with desired compositions were prepared by weight. The M--CD
concentration was varied in the range 0.002 to 0.5%w/w (corresponding to 0.018 to 4.5
mmolal). Before any rheological measurements were started the final samples were left to
equilibrate at room temperature for at least 12h.
The rheological measurements were performed with a StressTech rheometer from
Rheologica, Sweden. A 4 cm, 1 cone and plate geometry was used, and the temperature of
the sample was controlled to within 0.1C by an external water bath. Measurements were
performed at 20C, both as continuous, and oscillatory, shear measurements. The viscosity,
or *, was determined as a function of the shear rate, or 2 f, where f denotes the

frequency. Within a range of shear rates the viscosity data from the two methods coincide,
and are independent of or 2 f, Figure 2. Values from this Newtonian plateau are reported
in the following figures.
-5-
10
, (Pa s)
10
10
-3
10
10
-2
10
-1
10
.
10
10
10
-1
, 2f (s )
Figure 2. Complex viscosity, * (filled diamonds), from oscillatory shear measurements and
viscosity, (open circles), obtained from continuous shear measurements for a solution
containing 10% HM-PEG (14.8 mmolal hydrophobes) and 0.005 % w/w CD (0.047 mmolal).
Model Considerations
Our results were interpreted in a simple model were CD-molecules are regarded to bind to
the hydrophobic tails of the polymer chains with a complex formation constant K. It is
assumed that 1:1 nut and bolt complexes are formed, and we represent this complex
formation within a Langmuir adsorption model. The concentration of adsorption sites, B, in
the model is restricted by the concentration of polymer hydrophobic tails, and cannot exceed
this value. To obtain the K and B values from our rheological data we have assumed that
G nk b T .26,27 kb is the Boltzmann constant, and T is the absolute temperature.

Furthermore, we have assumed that the characteristic time, , of the relaxation process, that
is important for the viscosity at the Newtonian plateau, is independent of the CD
concentration, cCD. By this latter assumption the main contribution to the viscosity in the
HM-PEG solutions is regarded to stem from associations of the polymer hydrophobic tails,
and the effect from entanglements of the PEG backbone is regarded insignificant. I.e. a
change in viscosity at the Newtonian plateau is only dependent on the concentration of
rheologically active chains, n ( n ). In fact, an earlier study of HM-PEG has shown that
disruption of hydrophobic associations is likely to be the main contributor to the relaxation
time,27 and since the coil size of an unperturbed PEG chain with similar molecular weight
suggests an overlap concentration far above the concentrations used in this study the effect
from chain entanglements is likely to be negligible.28
-6-
From this follows that:24

= 1 = 1
0
B + cCD + 1 K
(B + cCD + 1 K ) Bc
CD
2
(1)
Here 0 and are the viscosities that are obtained without CD and at excess CD,
respectively, and is the fraction of occupied binding sites in the Langmuir model.
In a previous paper we have studied the complex formation between a graft hydrophobically
modified polymer, HM-EHEC, and CD.24 In this work we fitted equation (1) to our
experimental data points ( ( ) ( 0 ) vs. cCD) with K and B as fitting parameters. In that
way we could determine K for several combinations of different hydrophobic groups and
CDs. We also found a very good correlation between the number of adsorption sites, B, and
the total amount of hydrophobic groups in the solution, chydrophobe. This observation was not
too surprising and this was taken as an indication of all hydrophobic groups being equally
important and contributing in a similar way to the network formation and to the viscosity.
Figure 3 is obtained by using equation (1) and illustrates how the viscosity () as a function
of cCD is influenced by a variation in B and K . From this figure it appears that the initial
behavior at low cCD is largely determined by B, and provided K has a sufficiently high value
(>10 mmolal-1) B strongly influences the curve in a large viscosity range. In such a situation
the initial behavior (at low cCD) is well represented by a straight line with the slope of -1/B
(see Equation 2). The viscosity at excess CD () is expected to be virtually independent of
cCD and to be the same as that found in a solution containing an unmodified polymer with a
corresponding molecular weight. K influences the curve in the intermediate region where a
transformation from the slope 1/B to the plateau value at high cCD (slope 0) appears. High
values of K cause a very abrupt transition, while a low value leads to a more extended
transition. In our previous work we determined K for the complexation between M--CD and
C14-alkyl hydrophobes of HM-C14-EHEC to K=44 mmolal-1,24 and since similar hydrophobes
(C16-18) are used also in the present polymer we have no reasons to believe that K is lower
here. The value of B can therefore be obtained from the initial behavior via a simplified
equation:
c

1 CD
0 0
B
(2)
-7-
1
0.9
K=44 mmolal-1
B=14.8 mmolal
/ 0
0.8
0.7
0.6
0.5
0.4
K=3 mmolal-1
B=4.4 mmolal
0.3
0.2
0.1
0
0
K=44 mmolal-1
B=4.4 mmolal
10
15
20
cCD (mmolal)
Figure 3. Viscosity as a function of the CD concentration as calculated from Equation 1.

Complex constants, K, 3 or 44 mmolal-1 and concentrations of adsorption sites, B, 4.4 or 14.8
mmolal, respectively, have been used. These B-values equal the concentration of HM-PEG
hydrophobic tails at HM-PEG concentrations of 3 or 10 %w/w, respectively. The dashed line
is a linear extrapolation of the trend in the region 0 cCD 0.5 B for K= 44 mmolal-1 and B =
4.4 mmolal.
Results and Discussion

The full line in Figure 4 is obtained from Equation 1 with values of B=4.4 mmolal (the
concentration of polymer hydrophobic tails in a 3%w/w HM-PEG solution), and K=44
mmolal-1 (the binding constant that was obtained for a C14 aliphatic chain in combination with
M--CD in our previous investigation).24 However, the experimental data are very different
and it is obvious that initially CD influences the viscosity much stronger than expected from
these values of B and K. At low cCD only a small fraction of the hydrophobic tails can be
deactivated (from stoichiometrical considerations), but the effect on the viscosity is dramatic.
Obviously deactivation of the first few hydrophobic associations has a much stronger
influence on the viscosity than what could at first be expected. Actually, by changing the
value of B from 4.4 mmolal to 0.4 mmolal, a value that is suggested by the initial slope, and
by keeping K=44 mmolal-1 constant a much better representation of the experimental data is
obtained (dotted line). This is a quite surprising result since this means that it is enough to
eliminate only about 10% of the hydrophobic tails to reduce the viscosity to a level virtually
-8-
corresponding to that at excess CD. As was explained in the Experimental section this is
based on the assumption that 1:1 complexes are formed. Since the hydrophobic tails are
relatively long we note that there is a possibility that higher complexes may form. Olson et al
have shown by NMR-measurements that two or even more -CD molecules can bind to a
C12-hydrophobic group attached to a PEG chain.29 This tendency is likely to be more
important at high cCD. This can however not explain the behavior that was observed since
formation of higher complexes would actually mean that the 10% is an overestimation.
1.0
K= 44 mmolal-1
B =4.4 mmolal
/ 0
0.8
0.6
0.4
K= 44 mmolal-1
B =0.4 mmolal
0.2
0.0
0
cCD (mmolal)
Figure 4. Relative viscosity, /0, as a function of the CD concentration in a solution with

3%w/w of HM-PEG. Filled circles represent the experimental data. The full line represents
the theoretical viscosity calculated from Equation 1 with K= 44 mmolal-1 and B = 4.4 mmolal,
or with B = 0.4 mmolal-1 (dashed line).
To be able to rationalize this observation a general discussion about structure in a HM-PEG

solution, and its concentration dependency, is needed.30 Micelle-like structures may appear
already at concentrations of about 10-3 % w/w.31,30 Since the triblock structure of the
HM-PEG chains results in an attraction between micelles,32 clusters that contain many
micelles are likely to form. While micelles probably have fairly well defined aggregation
numbers the clusters may appear in a wide range of sizes, and it seems reasonable that the
average cluster-size increases with concentration.33-35 Below the concentration where
clusters start to interact and connect to each other the viscosity of the solution is likely to be
low, while above this concentration the viscosity increases rapidly. At this stage the solution
is often referred to as containing a three-dimensional network that extends over macroscopic
distances. The structure as a function of the concentration that follows from the above
discussion is schematically illustrated in Figure 5. Regions with higher concentration of
-9-
polymer correspond to clusters of micelles, and within these clusters inter-micellar links are
likely to be numerous, while polymers that connect micelles located in different clusters have
to span polymer depleted regions and are more rare. One indication of that the solution is
inhomogeneous is given by the phase behavior, Figure 6. A solution of the corresponding
diblock polymer, which has a chemical structure corresponding to half the triblock polymer,
has a phase behavior very much resembling that of an unmodified PEG polymer with a
phase separation at high temperatures. This is expected since the hydrophobic tails are
hidden in the core of micelles and it is only the PEG part that is exposed towards the
aqueous solution. Despite that similar micellar structures are expected to form with the
triblock polymer a solution based on HM-PEG has a much more pronounced tendency to
phase separate, and a phase concentrated in polymer is obtained in equilibrium with a phase
depleted in polymer. This situation appears already at slightly elevated temperatures, and
only a small change of the PEG/solvent interaction is needed to induce phase separation.
The very pronounced difference in phase behavior between the di- and tri-block polymers
can be traced to the attraction between micelles.36 The molecular picture that emerges is
that the solution is likely to be inhomogeneous with large concentration fluctuations, and at
intermediate concentrations a percolated network forms via relatively few bridges between
different clusters. This picture is valid in the concentration range that has been investigated.
Unimers
Flower
Micelles
Clusters
Network
increased cHM-Peg
Figure 5. Schematic representation of the self-aggregation of HM-PEG as function of

increasing cHM-PEG.
-10-
120
100
Tcp (C)
80
60
40
20
0
0
10
c (% w/w)
Figure 6. Partial phase diagram for HM-PEG. The filled circles represent the triblock and the
open circles represent the diblock. Reproduced from 36
With this in mind we are now in a position to explain the fact that the viscosity was affected
much stronger than what was first expected from Equation 1. In order to reduce viscosity
strongly it is only needed to decouple associations between different clusters, and since
these are only expected to involve a small fraction of the total number of HM-PEG chains a
strong initial decrease in viscosity on addition of CD can be understood. This could be the
explanation to that a CD concentration corresponding to only about 10% of the hydrophobic
groups has to be added to the 3 wt% HM-PEG solution to give a viscosity that is virtually the
same as in a solution containing the unmodified PEG.
One reason for this behavior could be that bridges between clusters correspond to HM-PEG
chains that are more stretched than links between micelles within the clusters. This means
that the former break and reform more frequently and are more likely to be presented to CD
molecules. Furthermore, the fact that the solution is depleted in polymer between clusters
may result in a relatively high CD concentration here, and termination of HM-PEG chains
located in this region increases also for this reason. A cartoon-picture of a HM-PEG solution
containing a low CD concentration could then, as illustrated in Figure 7, be viewed as
discrete clusters covered on the surface by CD. It should however be noted that we do not
believe that this is a static situation but rather Figure 7 should be seen as a snapshot. For the
sake of completeness we have to mention the possibility that the phenomenon (with a rapidly
decreasing viscosity) possibly could be shear-induced, and the network is being broken
-11-
down into fragments that orient in the flow. However, we find this unlikely since identical
results have been obtained with many different measuring geometries of the rheometer
(cone and plate of different diameters, plate-plate, and double gap), and special care was
paid to perform control measurements also at very low shear rates. The results were also not
affected by whether the samples had been subject to pre-shear or not, and also with the
/0
rheometer in the oscillatory shear mode identical results were obtained.
cCD
Figure 7.The figure shows a schematic representation of the binding of CD to HM-PEG

hydrophobic tails. Already at rather low concentrations of CD the percolated network
structure is eliminated because hydrophobic associations between clusters are inhibited. At
higher CD-concentrations also clusters and individual micelles are expected to be
disengaged by CD.
Measurements have also been performed at other polymer concentrations (5 and 10% w/w
polymer). As a matter of fact a similar result but even more pronounced deviation from the
expected value of B was found at the two other investigated concentrations, Table 1 and
Figure 8. The value of B obtained from the experimental data is virtually unaffected by an
increased HM-PEG concentration (in the investigated range). This means that the fraction
B/chydrophobe decreases with increasing HM-PEG concentration, and at the highest HM-PEG
concentration this ratio is as low as 4%. This may mean that the number of polymer chains
that participate within one cluster grows with increasing polymer concentration (increasing
cluster size), which also has been suggested before.34
An increasing size of the decoupled clusters would be likely to give a contribution to the
viscosity of the solution. Indeed, in a closer look it can be seen that the experimental curves
can be divided into three different regions, instead of two, Figure 8. This behavior becomes
-12-
more pronounced with increasing HM-PEG concentration, and in the second region, located
at intermediate CD concentrations, the change in viscosity is less dramatic than the initial
steep decrease. We refer the decrease in this intermediate region to disengagement of
individual clusters and micelles by CD. While B was obtained by extrapolation to / 0 = 0 at
low CD concentration, a similar extrapolation can be made in this intermediate region to
obtain B2. B2 would then be connected to the actual HM-PEG concentration in the solution.
Indeed, B2 is in contrast to B dependent of the HM-PEG concentration, Table 1.
Table 1. Data from three different HM-PEG concentrations. B and B2 were obtained by
extrapolation to /0=0 in the low cCD and intermediate cCD regions respectively.
Concentration HM-
Concentration
PEG (%w/w)
Hydrophobic tails,
B (mmolal)
B/ chydrophobe
B2 (mmolal)
chydrophobe (mmolal)
3
4.4
0.45
10%
2.42
7.4
0.42
6%
3.30
10
14.8
0.53
4%
4.39
-13-
1.0
3%
/ 0
0.8
0.6
0.4
0.2
0.0
0
B2
cCD (mmolal)
1.0
5%
/ 0
0.8
0.6
0.4
0.2
0.0
0
B2
cCD (mmolal)
1.0
10%
/ 0
0.8
0.6
0.4
0.2
0.0
0
cCD (mmolal)
B2
Figure 8. Relative viscosity, /0, as a function of CD concentration in solutions with 3% w/w,

5% w/w or 10% w/w HM-PEG, respectively. B-values were obtained by extrapolation to / 0
= 0 from the behavior at low CD concentration (data represented as filled circles has been
used in the extrapolation). The B2 values were obtained by extrapolation to / 0 = 0 from the
behavior at intermediate CD concentrations (open circles). Data represented by open
squares have not been used in the extrapolations.
-14-
In the structural picture that evolved above, clusters are connected with a rather small
number of bridges, and by adding CD these bridges are deactivated. This is related to the
situation for which the percolation theory has been developed.37 This was quite recently
used to rationalize rheological data in a related system, where microemulsion droplets were
connected into a network with HM-PEG.38 At a certain concentration of bridges (here
between clusters) a percolated network is anticipated, and at this point the viscosity is
expected to increase rapidly (diverge) since an infinite network that extends over
macroscopic distances forms. Below the percolation threshold the viscosity can be seen as a
summation of contributions from all different cluster sizes and is given by:37
0 ((c HM PEG cCD )c (c HM PEG cCD ))0.7
(3)
In Equation 3 we have used the concentration variable (c HM PEG cCD ) since a HM-PEG
molecule that is terminated with CD at one end is expected to be incapable of forming a
bridge between clusters. At all three investigated HM-PEG concentrations (3, 5, and 10 %)
we expect a percolated network at cCD = 0. Following the discussion above, in the initial
stages addition of CD terminates hydrophobic tails of HM-PEG polymers that are important
for the percolated network (between clusters), leaving the clusters almost unaffected. Thus,
in the present view the percolation threshold, (c HM PEG cCD )c , becomes dependent on the
HM-PEG concentration, and the specificity of CD to preferentially deactivate bridges between
clusters implies that a pure HM-PEG solution should always have a higher viscosity than a
CD containing sample with the same effective concentration (c HM PEG cCD ) , Figure 9. In the
figure it can be seen that the percolation theory provides a good representation of CDcontaining solutions. It is interesting to see that the viscosity as a function of HM-PEG
concentration (without CD) has a rather different behavior, with a different functional form.
Obviously Equation 3 can not be used to represent the decrease in viscosity upon dilution of
a pure HM-PEG solution (in the investigated concentration range). This may be taken as an
evidence of that the structure and its dependency on the concentration variable in solutions
of HM-PEG is different if the solution also contains CD.
-15-
100
(Pa s)
10
0.1
0.01
0
(cHM-PEG- cCD) (mmolal)
Figure 9. The figure shows the viscosity as a function of the effective concentration variable
(cHM-PEG cCD), see text. Three sets of data with varying cHM-PEG (3%w/w (), 5%w/w () and
10%w/w ()) are shown. The full lines are best fits to Equation 3 for each HM-PEG
concentration, and dotted vertical lines represent the corresponding percolation thresholds
(cHM-PEG - cCD)c. The viscosity as function of cHM-PEG without CD is represented by ().
Conclusions
The most surprising observation in the present investigation is that it is enough to terminate a
rather small fraction (about 10% in a 3% HM-PEG solution) of the HM-PEG hydrophobic tails
with CD molecules to reduce the viscosity to a level virtually corresponding to that at excess
CD. To rationalize this observation, advantage was taken of that concentration fluctuations
are likely to be substantial in a HM-PEG solution of this concentration and that viscosity, in
this view, becomes strongly dependent on HM-PEG chains that connect different clusters of
micelles. By deactivating these latter associations, the viscosity decreases rapidly.
Another interesting observation is that the fraction of HM-PEG chains that are active in
interconnecting different clusters seems to decrease with an increasing polymer
concentration (in the investigated range). This may be taken as an indication that sizes of the
clusters increase with an increasing HM-PEG concentration.
-16-
Acknowledgement
This investigation was sponsored by the Center for Amphiphilic Polymers (CAP).
References
(1)
Glass, J. E. Polymers in Aqueous Media; American Chemical Society: Washington,

DC, 1989; Vol. 223.
(2)
Glass, J. E. Hydrophilic Polymers; Performance with Environmental Acceptability;

American Chemical Society: Washington, DC, 1996; Vol. 248.
(3)
Strauss, U. P. Hydrohpobic polyelectrolytes. Polymers in aqueous media; American

Chemical Society: Washington DC, 1989; Vol. 223; pp 317-324.
(4)
Williams, P. A.; Meadows, J.; Phillips, G. O.; Senan, C. Cellulose: Sources and
Exploration 1990, 37, 295-302.
(5)
Landoll, L. M. J. Polym. Sci. 1982, 20, 443-455.
(6)
Tanaka, R.; Meadows, J.; Phillips, G. O.; Williams, P. A. Carbohydrate Polymers

1990, 12, 443-459.
(7)
Picton, L.; Muller, G. Macromol. Symp. 1997, 114, 133-138.
(8)
Joabsson, F.; Rosen, O.; Thuresson, K.; Piculell, L.; Lindman, B. J. Phys. Chem.
1998, 102, 2954-2959.
(9)
Karlson, L.; Joabsson, F.; Thuresson, K. Carbohydrate Polymers 2000, 41, 25-35.
(10)
Nilsson, S.; Thuresson, K.; Hansson, P.; Lindman, B. J. Phys. Chem. 1998, 102,
7099-7105.
(11)
Piculell, L.; Nilsson, S.; Sjstrm, J.; Thuresson, K. Assosciatve polymers in aqueous
media; American Chemical Society: Washington DC, 2000; Vol. 765; pp 317-335.
(12)
Funasaki, N.; Yodo, H.; Hada, S.; Neya, S. Bull. Chem. Soc. Jpn. 1992, 65, 13231330.
(13)
Junquera, E.; Tardajos, G.; Aicart, E. Langmuir 1993, 9, 1213-1219.
(14)
Ma, Z.; Glass, J. E. Polym. Matrl. Sci. Engin. 1993, 69, 494-495.
(15)
Mwakibete, H.; Bloor, D. M.; Wyn-Jones, E. Langmuir 1994, 10, 3328-3331.
(16)
Mwakibete, H.; Bloor, D. M.; Wyn-Jones, E.; Holzwarth, J. F. Langmuir 1995, 11, 5760.
(17)
Park, J. W.; Song, H. J. J. Phys. Chem. 1989, 93, 6454-6458.
(18)
Wan Yunus, W. M. Z.; Taylor, J.; Bloor, D. M.; Wyn-Jones, E. J. Phys. Chem. 1992,
96, 8979-8982.
(19)
Eisenhart, E. K.; Johnson, E. A. Method for improving thickeners for aqueous

systems. In U.S. Patent; Rohm and Haas Company: United States, 1992.
(20)
Lau, W.; Shah, V. M. Method for improving thickeners for aqueous systems. In U.S.
Patent; Rohm and Haas Company: United States, 1994.
(21)
Zhang, H.; Hogen-Esch, T. E.; Boschet, F.; Margaillan, A. Langmuir 1998, 14, 49724977.
-17-
(22)
Akiyoshi, K.; Sasaki, Y.; Kuroda, K.; Sunamoto, J. Chemistry Letters 1998, 93-94.
(23)
Islam, M. F.; Jenkins, R. D.; Bassett, D. L.; Lau, W.; Ou-Yang, H. D. Macromolecules
2000, 33, 2480-2485.
(24)
Karlson, L.; Thuresson, K.; Lindman, B. Carbohydrate polymers 2002, 50, 219-226.
(25)
Karlson, L.; Nilsson, S.; Thuresson, K. Colloid Polym. Sci. 1999, 798-804.
(26)
Green, M. S.; Tobolsky, A. V. J. Chem. Phys. 1946, 14, 80-89.
(27)
Annable, T.; Buscall, R.; Ettelaie, R.; Whittlestone, D. J. Rheol. 1993, 37, 695-726.
(28)
Gregory, P.; Huglin, M. B. Makromol. Chem. 1986, 187, 1745-1755.
(29)
Olson, K.; Chen, Y.; Baker, G. L. J. Polym. Sci. Part A: Polym. Chem. 2001, 39, 27312739.
(30)
Winnik, M. A.; Yekta, A. Current Opinion in Colloid & Interface Science 1997, 2, 424436.
(31)
Yaminsky, V. V.; Thuresson, K.; Ninham, B. W. Langmuir 1999, 15, 3683-3688.
(32)
Semenov, A. N.; Joanny, J.-F.; Khokhlov, A. R. Macromolecules 1995, 28, 10661075.
(33)
Alami, E.; Rawiso, M.; Isel, F.; Beinert, G.; Binana-Limbele, W.; Francois, J.
Hydrophilic polymers. Performance with environmental acceptance; American
Chemical Society: Washington, DC, 1993; Vol. 248; pp 343-362.
(34)
Alami, E.; Almgren, M.; W., B. Macromolecules 1996, 29, 2229-2243.
(35)
Xu, B.; Li, L.; Yekta, A.; Masoumi, Z.; Kanagalingam, S.; Winnik, M., A.; Zhang, K.;
Macdonald, P., M.; Menchen, S. Langmuir 1997, 13, 2447-2456.
(36)
Thuresson, K.; Nilsson, S.; Kjoniksen, A.-L.; Walderhaug, H.; Lindman, B.; Nystrm,
B. J. Phys. Chem. 1999, 103, 1425-1436.
(37)
Stauffer, D.; Coniglio, A.; Adam, M. Adv. Polym. Sci. 1982, 44, 103 -158.
(38)
Bagger-Jrgensen, H.; Coppola, L.; Thuresson, K.; Olsson, U.; Mortensen, K.

Langmuir 1997, 13, 4204-4218.
-18-
Complex formed in the system hydrophobically modified polyethylene

glycol / methylated -cyclodextrin / water. An NMR diffusometry study.
L. Karlson,*,1, C. Malmborg,2 K. Thuresson2 and O. Sderman2
1
Physical Chemistry 1, Center for Chemistry and Chemical Engineering, Lund University, P.O.
Box 124, SE-221 00 Lund, Sweden.

*
To whom correspondence should be addressed.
fax: +46 303 839 21

e-mail: leif.karlson@akzonobel.com
Abstract
In aqueous solutions hydrophobically modified polyethylene glycol (HM-PEG) forms a
transient polymer network held together by intermolecular hydrophobic associations. In the
present investigation we have used NMR-diffusometry to study how the addition of
methylated -cyclodextrin (M--CD) influences the polymer network. The addition of M--CD
resulted in an increased mean self-diffusion of HM-PEG, DHM-PEG, which is referred to a
degradation of the polymer network when hydrophobic associations are disrupted due to
complex formation between the hydrophobic groups of HM-PEG and M--CD. Addition of
small amounts of M--CD results in a dramatic increase in DHM-PEG. Upon further addition of
M--CD the increase in DHM-PEG is less dramatic and at excess M--CD, DHM-PEG levels off and
equals the mean self diffusion coefficient for unmodified PEG with the same molecular
weight. The suggested interpretation is that the addition of the first molecules of M--CD
mainly reduces the probability for hydrophobic associations inter-connecting different clusters
of polymer micelles whereas at higher M--CD concentrations a disengagement of the
individual clusters into separate HM-PEG molecules becomes important.
-1-
Introduction
Associative polymers (APs) are commonly used as rheology modifiers in water based paint
formulations. Compared to conventional thickeners APs offer a flow behavior closely
resembling Newtonian flow over a wide range of shear rates. When applying the paint this
leads to improved leveling and a better hiding power of the paint. Another important
advantage offered by the use of APs compared to conventional thickeners is a dramatic
reduction of the roller spatter.
One class of APs frequently used is the hydrophobically modified urethanes (HEURs). The
HEUR thickeners used in technical applications are normally composed of diurethane linked
polyethylene glycol blocks with hydrophobic alkyl end-groups.1 The viscosity-enhancing
effect of HEUR originates from the formation of a transient polymer network held together by
physical linkages of assembled hydrophobic groups. A large number of investigations have
been carried out in order to obtain information about the association mechanisms of HEUR
polymers. The work is summarized in two recent review articles.1,2 The synthesis procedure
of technical HEUR thickeners results in a broad distribution in molecular weights.1 This
synthetic route and the obtained molecular weight distribution is attractive from a practical
point of view, while interpretation of results from experiments aimed at physical
characterization becomes more complicated. To simplify interpretations from fundamental
physico-chemical investigations of HEUR polymers, polymers with narrow molecular weight
distributions have been synthesized and used for this purpose.3,4 HEUR polymers with
narrow molecular weight distributions are often referred to as hydrophobically modified
polyethelen glycol (HM-PEG) or Triblock polymers. It is now commonly accepted that
HM-PEG form flower like micelles at low concentration.1,2
Upon increasing polymer concentration aggregated structures are formed. First micelles
form, which become connected into clusters, and at higher polymer concentrations a
percolated network is formed, which spans the entire sample. The three dimensional network
is held together by polymer molecules with one hydrophobic end in a micelle in one cluster
and the other end in a micelle in a neighboring cluster.5,3,4,2 (Figure 1)
One way to study the properties of the network is offered by NMR self-diffusion
measurements.6,7 With this technique, the molecular displacements of the species in the
samples may conveniently be followed. Such data convey information about polymer size
polydispersity, including possible effects of association.8,9 For instance, for an extensively
associated polymer solution, one expects slow diffusion and also a broad distribution in the
values of the diffusion coefficients, provided that the life-time of the polymers in the
-2-
associated structures is long compared to the time-scale of the measurements.8 Any effects
that would decrease the degree of association then leads to a increase in the self-diffusion
coefficients as well as a to a narrower distribution of diffusion coefficients.
Unimers
Flower
Micelles
Clusters
Network
increased cHM-PEG
Figure 1. Schematic representation of the self-aggregation of HM-PEG as function of

increasing cHM-PEG.
A fruitful way to study association mechanisms of a polymer network in aqueous solution is

by gradually decoupling the network by the addition of a third component. An excess of
surfactant has been used for this purpose in many studies. An alternative route of decoupling
the polymer network is offered by cyclodextrin (CD). CD is a cyclic oligomer of glucose. The
CD molecules either consist of 6, 7 or 8 glucose units and they are denoted -CD, -CD or
-CD, respectively. As shown in Figure 2, the CD molecule has the shape of a truncated
cone. The exterior of the molecule is hydrophilic while the interior forms a lipophilic cavity. In
aqueous solutions CD molecules form inclusion complexes with lipophilic substances
provided that the lipophilic molecules have a shape that fits in the cavity. Examples of
complex formation constants found in the literature for cyclodextrin in combination with some
different surfactants are presented in table 1.
-3-
Table 1. Complex formation constant, K1 (mM-1), for -CD and -CD in combination with
sodium dodecyl sulphate (SDS), tetradecyl sulphate (TDS), hexadecyl sulphate (HDS),
dodecyl trimethyl ammonium bromide (DTAB), tetradecyl trimethyl ammonium bromide
(TTAB) and cetyl trimethyl ammonium bromide (CTAB).
Surfactant
Reference
K1
-CD
-CD
DTAB
23.7
10
TTAB
61.0
39.8
10
CTAB
99.2
67.7
10
SDS
25.6
11
TDS
48.2
11
HDS
53.3
11
O
O
O
OH HO
O
HO
O
O
OH
O
O
HO
O
O
OH
HO
O
O
O
O
Hydrophobic cavity
Figure 2. To the left is shown one example of the chemical structure of a methylated
-cyclodextrin molecule and to the right a schematic representation of the geometry of the
same molecule.
The effect of adding CD to an AP-solution is a reduction of the viscosity, and the use of CD
to reduce the viscosity of AP-solutions was first described in two patents by the Rohm and
Haas Company.12,13 Later, more fundamental studies to explain the effect of CD on
HM-polymer solutions have been performed.14-16 The CD molecule is believed to form a
nut and bolt complex with a lipophilic group of the HM-polymer. This means that when a
polymer hydrophobic tail is hidden within the CD-cavity, this polymer is deactivated and
does not contribute to the connectivity. This process will eventually result in a partial or
complete disruption of the three-dimensional polymer network.
-4-
In an earlier paper we have reported on the effect of methylated -cyclodextrin (M--CD) on

aqueous solutions of HM-PEG in the polymer concentration range from 3 to 10% w/w. The
degradation of the polymer network was followed by viscosity measurements, and by
rationalizing the data in a simple model the concentration of binding sites could be extracted.
It was found that the number of polymer hydrophobic tails that needed to be blocked to
substantially reduce the viscosity only constituted a small fraction of the total number of
hydrophobic groups in the system. This result was in sharp contrast to what was earlier
found for the system containing hydrophobically modified ethyl (hydroxyethyl) cellulose
(HM-EHEC), where a good agreement between the calculated concentration of binding sites
and the total number of hydrophobic groups in the system was found.17 The findings were
interpreted in the following way. In a HM-PEG solution, the dynamics and the viscosity are
strongly dependent on the polymer linkages connecting different clusters, and by adding CD
the probability of forming these cluster-spanning linkages reduces dramatically.
In the present work, self-diffusion of the different components in the system has been studied
in order to confirm the suggested model. Both self-diffusion of HM-PEG and of M--CD as a
function of the CD-concentration have been followed for a constant amount of polymer. In
line with previous investigations the data have been used to obtain information about
polymer transport dynamics and association mechanism.18-23
Experimental
Materials
Hydrophobically end-modified polyethylene glycol (HM-PEG) with the structure C1618-EO140IPDU-EO140-C1618 was used in this study. C1618-EO140 denotes an ethoxylate of a mixture of
unsaturated alcohols (C16 to C18), and IPDU represents an isophorone diurethane group
connecting two ethoxylated alcohol molecules. HM-PEG was synthesized and characterized
according to procedures described elsewhere.24 The weight average molecular weight
(Mw =13500) and polydispersity index (Mw/Mn = 1.1) were determined by size exclusion
chromatography (SEC). Before use, the HM-PEG was purified from low molecular weight
impurities, such as salt and low molecular weight PEG, by dialysis. A 3% solution of the
polymer was dialyzed against water of Millipore quality for a week with repeated exchange of
the water. For the dialysis Spectra/Por molecularporous membrane tubing with a molecular
weight cut off of 6-8000 was used. After the dialysis the polymer was recovered by freeze
drying.
-5-
The cyclodextrin used for this study was methylated alpha-cyclodextrin (M--CD) supplied by
Wacker-Chemie under the trade name Cyclodextrin Alpha W6 M1.8. The degree of
substitution of methyl groups was 1.6 1.9 according to the producer. The material was used
as supplied, without any further purification.
Polyethylene glycol with molecular weight of 20,000 (PEG 20,000) was obtained from Merck
and was used without further purification.
Samples
The NMR samples were prepared by using D2O (99.8%) supplied by Dr. Glaser AG Basel,
Switzerland, as the solvent. Three stock solutions were prepared with HM-PEG
concentration 3% w/w and M--CD concentrations of 0, 0.2 (1.78 mmole/kg) and 3.35 % w/w
(29.87 mmole/kg), respectively. The stock solutions were prepared in test tubes, sealed with
Teflon tightened caps, and then equilibrated for 24 h at room temperature. From the stock
solutions, samples with desired compositions were prepared by weight, directly in NMR
tubes that were flame sealed. The samples that were not measured immediately after the
preparation were stored at -18C in order to minimize degradation. Before the measurements
the samples were left to equilibrate in room temperature for at least 24 h. Visual inspection of
the samples before the measurements showed that they were clear and homogenous.
Two stock solutions containing 1 % w/w PEG 20,000 were prepared, of which one also
contained 1 % w/w M--CD. From these two stock solutions the final samples containing
1 % w/w PEG 20,000 and M--CD to a concentration of 0, 0.25, 0.5, 0.75 and 1 % w/w,
respectively, were prepared directly in the NMR tubes, which were flame sealed.
Methods
Pulsed Field Gradient (PFG) NMR 1H experiments were performed on a 200 MHz Bruker
DMX spectrometer equipped with a Bruker DIFF-25 gradient probe driven by a Bruker
BAFPA-40 unit.25 The temperature was 25C. In this study, we have performed Hahn spinechoes for measurements on PEG and used the stimulated-echo technique for
measurements on CD. Typical values used for have been 0.5 - 3 ms and for 20 - 100 ms
(for the meaning of and see next section).
-6-
Data evaluation
For a system of monodisperse molecules, the normalized PFG NMR echo signal is given by:
I (k ) = exp(kD)
(1)
where k is given by k = 2 g 2 2 ( / 3) . is the magnetogyric ratio of the proton, is the

duration of the field gradient pulses, and , the diffusion time, is the distance between the
leading edges of the gradient pulses. D, finally, is the self-diffusion coefficient of the studied
species. Clearly, for a monodisperse system, a semilogarithmic plot of I vs. k should yield a
straight line.
For a polydisperse system, Equation 1 above is not longer valid.26 Instead, the signal
becomes a sum of signal decays from species of different molecular weights:
I (k ) = pi exp( kDi )
(2)
where pi is the number fraction of molecules with diffusion coefficient Di. This situation
manifests itself in curved lines, when echo intensities are plotted vs. k in a semilogarithmic
plot.
A common way to treat polydispersity effects in PFG NMR studies is to assume a continuous
distribution of diffusion coefficients, P(D) and to write:
I (k ) = P( D) exp( kD)dD
(3)
A convenient form for P(D) is constituted by a log-normal distribution:
(ln( D) ln( D0 )) 2
exp
P( D) =
2
D 2
(4)
where D0 is the median value for the diffusion coefficient and is the width of the distribution.
The mean value of the diffusion coefficient is obtained from D = D0 exp 2 2 .26
We have used Equation 1 above in evaluating data obtained from M--CD, as the echo
intensities from this species gave rise to linear semi-logarithmic plots and Equation 2 for all
polymer diffusion experiments. In both cases, we have used nonlinear least-squares fitting
procedures (using in-house developed soft-ware, based on the Matlab-package) to extract
the relevant parameters.
-7-
Results and discussion

Figure 3 displays a semilogarithmic plot of the mean self-diffusion coefficient, DHM-PEG, for a
3% w/w solution of HM-PEG as a function of the concentration of methylated -cyclodextrin
(cCD). The polymer concentration of 3% w/w is far above the onset of polymer aggregation
and the solution (without cyclodextrin) is expected to contain a percolated network of flower
micelles (compare Figure 1). Also given in Figure 3 are the widths of the obtained
distributions of polymer diffusion coefficients. In the pure HM-PEG system (without
cyclodextrin), DHM-PEG has a value of 2.5 x 10-13 m2/s while the value of implies a broad
distribution of diffusion coefficients. This implies that the life-time of the associated structures
is long compared to the time-scale of the experiments, which is of the order of 100 ms.
Moreover, the size of the associated structures varies considerably, from essentially infinite
for those who take part in the percolated network, to smaller numbers for those who are not
part of the sample-spanning network.
-9
10
DCD for free CD

-10
DHM-PEG , DCD (m /s)
10
-11
10
DPEG for unmodified PEG
-12
10
-13
10
0
0
10
15
20
25
30
cCD (mmolal)
Figure 3. Mean self-diffusion coefficients for HM-PEG, DHM-PEG; (open circles) and for
M--CD, DCD, (triangles) and the distribution in DHM-PEG, , (filled circles) as a function of CD
concentration in 3%w/w solution of HM-PEG. The lower dashed line represents the mean
self-diffusion coefficient for unmodified PEG (MW= 20000 g/mol) in 1% w/w solution of PEG.
The upper dashed line represents DCD when no HM-PEG or PEG is present.
An interesting observation that we will first discuss is the fact that the width of the distribution
of diffusion coefficients is considerably lower for the sample without any added CD. Upon
addition of small amounts of CD, first increases. We interpret this observation as follows.
-8-
There are two diffusion mechanisms for HM-PEG in these solutions. In one, an individual
polymer performs a random walk between neighboring polymer micelles, in which process
one of the hydrophobic end-caps leaves a micelle and enters an adjacent micelle. The
second end-cap subsequently transfers to the new micelle. Thus the polymer moves in much
the same fashion as a geometer. The second process involves diffusion flow of individual
micelles or clusters of micelles. The two diffusion mechanisms are independent and their
contributions to the observed polymer diffusion are therefore additive. While the first process
depends on the (average) distance between micelles and the solubility of end-caps in water,
the second depends on the size of the clusters. Thus, the second process is expected to give
rise to polydispersity effects, while the first process is expected to be characterized by a
single diffusion coefficient. Without CD present, a majority, if not all, polymer molecules take
part in the percolated network, and the diffusion process is dominated by the first of the two
proposed mechanisms. As CD is added, finite clusters are formed (see further discussion
below) and as a consequence the diffusion rate increases and the width of the diffusion
coefficient distribution increases.
Figure 3 shows that DHM-PEG increases dramatically with the addition of methylated
-cyclodextrin (M--CD) to the HM-PEG solution. This increase is accompanied by a
substantial decrease in the width of the diffusion coefficient distributions. The change is most
pronounced at small additions of CD, below 1 mmole. This corresponds well with the results
of the viscosity measurements reported in a previous paper.17 The combined results of the
self-diffusion and the viscosity experiments at low concentrations of CD are presented in
Figure 4. At low cCD we expect each CD molecule to bind to one hydrophobic end-group of
the HM-PEG and that the hydrophobic group that has formed complex with a CD molecule is
no longer available for hydrophobe-hydrophobe associations. CD opposes the associative
behavior of HM-PEG, and the results suggest a disintegration of the polymer network since
the viscosity decreases and the diffusion increases. Since the total concentration of
hydrophobic groups (chydrophobe) is 4.4 mmolal in the solution, only about 20% or less of the
hydrophobic groups can be covered by a cyclodextrin molecule in this concentration range.
Supported by the viscosity measurements we interpret the initial steep increase of DHM-PEG as
a degradation of the polymer network into separate clusters with dangling PEG chains, each
decorated with a CD molecule, sticking out in the solution. This interpretation would imply,
that M--CD preferentially binds to hydrophobes that are part of HM-PEG that binds the
clusters together.
-9-
1.0
-11
10
-12
/0
10
0.4
0.6
D (m /s)
0.8
-13
10
0.2
0.0
0
cCD (mmolal)
Figure 4. Mean self-diffusion coefficient for HM-PEG, DHM-PEG; and relative viscosity, /0, as
a function of CD concentration in 3%w/w solution of HM-PEG. B was obtained by
extrapolation to /0=0 from the behavior at low CD concentration.17
At about 0.5 mmolal of added CD, DHM-PEG reaches a second region where the increase in
DHM-PEG is slower compared to the initial steep increase. (Figure 5) The CD concentration,
where the break-point between the two regions is found, corresponds well to what was found
in the viscosity measurements. Included in the figure is the concentration of binding sites,
B=0.45 mmolal, which was found to be needed to be inhibited in the rheological

measurements to disengage the network and reduce viscosity strongly.17 In the viscosity
measurements we also found a second region were the decrease in viscosity was more
moderate. This concentration range is located almost in the same interval as that of the
second region of the self-diffusion increase. We referred this more moderate viscosity
decrease to a disengagement of individual clusters into separate HM-PEG molecules, and
finally, virtually every hydrophobic group is covered by a cyclodextrin molecule. DHM-PEG
continues to increase, albeit very moderately, up to cCD = 10 mmolal which is more than twice
the total concentration of hydrophobic groups. One possible explanation is that the complex
formation is not quantitative and at this stage free cyclodextrin is present in the solution,
although we feel this is less likely, on account of the large binding constant for alkyl chains to
CD (table 1). Another, perhaps more likely reason is that more than one cyclodextrin
molecule bind to each hydrophobic group. Here it should be noted that the viscosity
measurements indicated a total disruption of the polymer network already at cCD about 2.5
- 10 -
mmolal. We note that the viscosity at cCD close to chydrophobe is almost equal to the viscosity of
the solvent and small variations in the viscosity are therefore difficult to detect.
At high cCD the polymer network is expected to be totally disconnected and it can be seen
from Figure 3 that DHM-PEG adopts a constant value of about 3 x 10-11 m2/s. In addition, the
widths of the distribution of diffusion coefficients are narrow here. This is in good agreement
with results obtained for the unmodified polyethylene glycol (PEG) with a molecular weight of
20,000 g/mole where we measured D to 3.3 x 10-11 m2/s. This also corresponds well to what
has been previously reported for the self diffusion of PEG.26,27
-11
DHM-PEG (m /s)
10
-12
10
-13
10
10
15
cCD (mmolal)
Figure 5. Mean self-diffusion coefficient for HM-PEG, DHM-PEG; as a function of CD

concentration in 3%w/w solution of HM-PEG. The dashed line represents B obtained from
Figure 4.
As noted above, the distribution in self-diffusion coefficients, , decreases with the addition of
M--CD indicating a reduction of the size distribution of the polymer aggregates. This is in
line with the discussion above. At excess M--CD the reduction levels off at a constant value
which most likely reflects the diffusion coefficient of individual HM-PEG molecules that are
decorated at both ends with M--CD. Here the value of would then reflect the molecular
weight distribution of HM-PEG.
We have also measured the self-diffusion of M--CD (DCD). (Figure 3) A single exponential
decay fits well to the measured decay of the intensity, I, as a function of k for CD also when
HM-PEG is present. This suggests that the exchange rate of a CD molecules bound to a
hydrophobic group is fast relative to the experimental time scale (around 100 ms). In the
- 11 -
investigated concentration regime, DCD varies moderately and reaches a plateau value at
around 10-15 mmolal CD. Furthermore, DCD at the plateau in the solution containing
HM-PEG is about 1 x 10-10 m2/s. This is low compared to what was found for a solution of
M--CD without HM-PEG where DCD was found to be 2.5 x 10-10 m2/s. The difference is
probably due to an obstruction to the movement of non-bound CD molecules exerted by the
HM-PEG chains despite the rather low volume fraction of the polymer. For hard spheres at
low volume fractions , the obstruction effect is given by D = D0 1
, where D and D0
2
are the diffusion coefficients in the presence and absence of obstructing spheres,
respectively. However, in the present case, we have a situation where CD molecules have to
diffuse through a medium of essentially overlapping polymer chains. In addition, the polymer
network may be considerably inhomogeneous on the relevant length-scales. This causes a
reduction in the diffusion coefficient which is considerably higher than simple obstruction
theories would imply.9,28 This hypothesis is supported by measurements of DCD in the
presence of unmodified PEG (MW= 20,000 g/mol) that shows a visible obstruction effect
already at concentration of unmodified PEG (cPEG) of 1 %w/w, see Figure 6 (please note that
the concentration of PEG is lower for the data presented in Figure 6, compared to the data
presented for HM-PEG). We also stress the fact that DCD in 1%w/w PEG is almost
independent of cCD shows that the complex formation between unmodified PEG and CD is of
minor importance and that the interaction is mainly an obstruction effect.
-10
2.4x10
-10
DHM-PEG , DCD (m /s)
2.0x10
-10
1.6x10
-10
1.2x10
-11
8.0x10
-11
4.0x10
10
15
20
25
30
35
40
cCD (mmolal)
Figure 6. Mean self-diffusion coefficient for unmodified PEG (squares) and for M--CD
(triangles) as a function of CD concentration in 1%w/w solution of unmodified PEG.
- 12 -
We have calculated the fraction of bound CD to HM-PEG (Pb) at two different cCD according
to Equation 5 where DCD,obs is the observed self-diffusion of CD at the actual cCD and DCD,free
is the self-diffusion of un-associated CD (which we take from data in Figure 3 at excess CD).
DCD,obs =Pb DHM-PEG + (1-Pb) DCD, free
(5)
Unfortunately, measurements of DCD below cCD = 2 mmolal was not possible due to low
signal to noise ratio and therefore calculations could only be done above B. As expected the
fraction of bound CD decreases with increasing cCD (table 2). The calculation of bound CD
molecules divided by the total number of hydrophobic groups (CD/hydrophobe) at cCD = 10.7
mmolal give another indication that more than one CD molecule can bind to each
hydrophobic group, although one has to be careful with such a conclusion, since the value of
DCD,free is uncertain, as it will depend on the structure and inhomogenities in the polymer
network, which properties presumably changes upon addition of CD.
Table 2 Data from two different concentrations of M--CD. The values of Pb are obtained
from calculations using Equation (1).
(m2/s)
3.0 10-11
Pb
CD/hydrophobe
(mmolal)
2.7
0.73
0.4
10.7
5.8 10-11
0.58
1.4
cCD
DCD,obs
- 13 -
Conclusions
In this investigation, methylated -cyclodextrin (M--CD) has been added to an aqueous
solution of hydrophobically modified polyethylene glycol (HM-PEG). Addition of M--CD had
a large impact on the mean self-diffusion coefficient of HM-PEG (DHM-PEG). When the
concentration of M--CD (cCD) increased, DHM-PEG first increased rapidly. At intermediate cCD
the increase in DHM-PEG was less dramatic and at excess CD DHM-PEG attained a constant
value. These results confirm the model suggested from the results of rheological
measurements in an earlier study on the same system.17 It is likely that DHM-PEG, as well as
the viscosity, is strongly dependent on HM-PEG chains that connect different clusters of
micelles and that the complex formation is primarily deactivating these associations at low
cCD. At higher cCD the complex formation results in a degradation of the clusters and micelles
into separate HM-PEG molecules where all hydrophobic end groups are hidden in
cyclodextrin molecules.
It was also possible to determine the mean self-diffusion coefficient of M--CD (DCD) in the
system. The results have been used to establish the fraction of M--CD that is bound to
HM-PEG. The results indicate that more than one M--CD molecule bind to each
hydrophobic group at high concentration of M--CD.
References
(1)
Glass, E. J. Coatings Techn. 2001, 73, 79-98.
(2)
Winnik, M. A.; Yekta, A. Current Opinion in Colloid & Interface Science 1997, 2, 424436.
(3)
Alami, E.; Rawiso, M.; Isel, F.; Beinert, G.; Binana-Limbele, W.; Francois, J. Model
hydrophobically end-capped poly(ethylene oxide) in water. In Hydrophilic polymers.
Performance with environmental acceptance; Glass, J. E., Ed.; American Chemical
Society: Washington, DC, 1993; Vol. 248; pp 343-362.
(4)
Alami, E.; Almgren, M.; W., B. Macromolecules 1996, 29, 2229-2243.
(5)
Annable, T.; Buscall, R.; Ettelaie, R.; Whittlestone, D. J. Rheol. 1993, 37, 695-726.
(6)
Callaghan, P. T. Principles of Nuclear Magnetic Resonance Microscopy; Clarendon

Press: Oxford, 1991.
(7)
Callaghan, P. T.; Lelievre, J. Biopolymers 1985, 24, 441-460.
(8)
Nydn, M.; Sderman, O. Macromolecules 1998, 31 (15), 4990-5002.
(9)
Nydn, M.; Sderman, O.; Karlstrm, G. Macromolecules 1999, 32, 127-135.
(10)
Mwakibete, H.; Bloor, D. M.; Wyn-Jones, E.; Holzwarth, J. F. Langmuir 1995, 11, 5760.
(11)
Park, J. W.; Song, H. J. J. Phys. Chem. 1989, 93, 6454-6458.
- 14 -
(12)
Eisenhart, E. K.; Johnson, E. A. Method for improving thickeners for aqueous

systems. In U.S. Patent; Rhom and Haas Company: United States, 1992.
(13)
Lau, W.; Shah, V. M. Method for improving thickeners for aqueous systems. In U.S.
Patent; Rhom and Haas Company: United States, 1994.
(14)
Akiyoshi, K.; Sasaki, Y.; Kuroda, K.; Sunamoto, J. Chemistry Letters 1998, 1998, 9394.
(15)
Zhang, H.; Hogen-Esch, T. E.; Boschet, F.; Margaillan, A. Langmuir 1998, 14, 49724977.
(16)
Gupta, R. K.; Tam, K. C.; Ong, S. H.; Jenkins, R. D. Interactions of methylated betacyclodextrin with hydrophobically modified alkali-soluble associative polymers
(HASE): effect of varying carbon chain length; XIIIth International Congress on
Rheology, 2000, Cambrige, UK.
(17)
Karlson, L.; Thuresson, K.; Lindman, B. submitted to Langmuir .
(18)
Persson, K.; Abrahmsn, S.; Stilbs, P.; Hansen, F. K.; Walderhaug, H. Colloid Polym.
Sci. 1992, 270, 465-469.
(19)
Walderhaug, H.; Hansen, F. K.; Abrahmsn, S.; Persson, K.; Stilbs, P. J. Phys.
Chem. 1993, 97, 9336-8342.
(20)
Abrahamsn-Alami, S.; Stilbs, P. J. Colloid Interface Sci. 1997, 189, 137-143.
(21)
Walderhaug, H.; Nystrm, B.; Hansen, F. K.; Lindman, B. Progr. Colloid Polym. Sci.
1995, 98, 51-56.
(22)
Macdonald, P., M. NMR diffusion studies of associating polymer interactions with

surfactant and latex; Spring meeting April 13-17, 1997, San Francisco.
(23)
Persson, K.; Wang, G.; Olofsson, G. J. Chem. Soc. Faraday Trans. 1997, 90, 35553562.
(24)
Karlson, L.; Nilsson, S.; Thuresson, K. Colloid Polym. Sci. 1999, 277, 798-804.
(25)
Stilbs, P. Progress in Nuclear Magnetic Resonance Spectroscopy 1987, 19, 1-45.
(26)
Hkansson, B.; Nydn, M.; Sderman, O. Colloid Polym. Sci. 2000, 278, 399-405.
(27)
Brown, W.; Stilbs, P.; Johnsen, R. M. J. Pol. Sci 1983, 21, 1029 - 1039.
(28)
Gregory, P.; Huglin, M. B. Makromol. Chem. 1986, 187, 1745-1755.
- 15 -

Hydrophobically Modified Polymers 2002tesis

Загружено:

Сведения о документе

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Hydrophobically Modified Polymers 2002tesis

Загружено:

Авторское право:

Доступные форматы

Hydrophobically Modified Polymers

Rheology and Molecular Associations

Avhandling fr Filosofie Doktorsexamen

Leif Karlson 2002

1.1 Hydrophobically modified polymers in paint

Chapter 2 Hydrophobically modified polymers

Comb HEUR thickeners

2.2 Hydrophobically modified polymers in aqueous solution

HM-PEG in aqueous solution

HM-EHEC in aqueous solution

Interaction between HM-Polymers and surfactants

Cloud point of HM-PEG

Cloud point of HM-EHEC

2.3 References Chapter 2

Chapter 3 Inhibition of hydrophobic associations as a tool to study

3.1 Inhibition of hydrophobic interactions by changing solvent quality

3.2 Inhibition of hydrophobic interactions by addition of surfactant

3.3 Inhibition of hydrophobic interactions by addition of cyclodextrin

Structure and properties of cyclodextrin

Formation of inclusion complex between lipophilic guest

Cyclodextrin and HM-Polymers

Cyclodextrin and HM-EHEC

Cyclodextrin and HM-PEG

3.4 References Chapter 3

Popular summary (in Swedish)

List of commercially available hydrophobically modified polymers

Rheology of an aqueous solution of an end-capped poly(ethylene glycol) polymer

Clouding of a cationic hydrophobically associating comb polymer

A rheological investigation of the complex formation between hydrophobically

Cyclodextrins in HM-PEG Solutions. Inhibition of Rheologically Active PolymerPolymer Associations

Complex formed in the system hydrophobically modified polyethylene glycol /

Aqueous solutions thickened with polymers are common in our daily

systems that gain their flowing properties from polymers.

weight (mw) and the thickening mechanism of HM-EHEC may include

Figure 1.1. Schematic

Hydrophobically modified Ethoxylated Urethane (HEUR) polymers

cyclodextrins, a group of cyclic substances with a hydrophobic cavity

environment the hydrophobic cavity of the cyclodextrin can host a

1.1 Hydrophobically modified polymers in paint

Figure 1.2. Schematic

Table 1.1. Example of a simple

Normally water borne paint is formulated aiming at a certain Stormer

thickening efficiency of the polymer. Conventional thickeners, with a

high molecular weight, (mw) normally have a high thickening

the polymer concentration may vary widely and depends on the

efficiency and give the required Stormer viscosity with a small

s-1) is low. Many important paint properties are influenced by the

Figure 1.3. Example of a

Figure 1.4. Example of a

Stormer viscosity a much higher polymer concentration is required,

thickening efficiency with a less marked shear thinning viscosity

Figure 1.5. Result from a

less shear thinning profile (Figure 1.2).

Figure 1.6. Viscosity, ,

One important advantage of HM-polymers is that the spatter from the

thickener, since these also associate with other ingredients in the

As will be discussed in section 2.2.3 the addition of surfactants can

Figure 1.8. Schematic

hydrophobic groups are the only effective thickening mechanism for

more pronounced when hydrophobic pigments are used. The color

of polymers often phase separate. The phenomenon with phase

Figure 1.10. Results of color

associative thickeners are tested in paint formulations that contain