You are on page 1of 29

Advances in Colbid and Interface Science, 34 (1991) l-29

Elsevier Science Publishers B.V., Amsterdam

PARTICLE FLOCCULATION BY ADSORBING POLYMERS

ERIC DICKINSONI and LEIF ERIKSSON2

IProcter Department of Food Science, University of Leeds, Leeds LS2 9JT


(United Kingdom)

2Institute for Surface Chemistry, P.O. Box 5607, S-114 86 Stockholm (Sweden)

CONTENTS
Abstract .............................................................. 1
1. Introduction .......................................................... 1
2. Kinetics of bridging flocculation ..................................... 4
3. Interactions between adsorbed polymer layers: experiment.............1 0
4. Interactions between adsorbed polymer layers: theory.................1 3
5. Non-bridging mechanisms of flocculation..............................1 5
6. Factors affecting floe structure.....................................1 8
7. Neutron scattering studies of floe structure.........................2 1
8. Computer simulation of floe structure .............................. ..2 3
References ......................................................... ..26

ABSTRACT
The flocculation of colloidal particles by adsorbing polymers is a
phenomenon of considerable scientific and industrial importance. This
article describes recent developments in the field with particular reference
to flocculation of negatively charged particles by high-molecular-weight cat-
ionic polyelectrolytes by polymer bridging or charge neutralization. Experi-
mental and theoretical studies of interactions between adsorbed polymer layers
give insight into the favourable conditions for bridging flocculation. Neutron
scattering and computer simulation give insight into the relationship between
floe structure and the mechanism of flocculation.

1. INTRODUCTION
Many polymers have a strong tendency to adsorb at the surface of solid
dispersed particles. Polymer adsorption may lead to colloid stability or to
particle flocculation. Which occurs in any particular situation depends on
factors such as the nature of the particle surface, the concentration and
chemical structure of the polymer, the solvent quality of the dispersion medium,

and so on. The flocculation of colloidal particles may also be induced by non-
adsorbing polymers. This article discusses some recent advances in understand-
ing the mechanism of particle flocculation by polymers with emphasis on the

factors affecting floe structure.

ooo1&86/91/$10.15 0 1991- Elsevier Science Publishers B.V.


The physico-chemical principles underlying the adsorption of homopolymers
are becoming reasonably well understood (ref. l-7). A key feature of macro-
molecular behaviour is that, in molecules with a large number of segments, a
relatively small segmental interaction energy can produce a large effect on
the molecule scale. So, even though individual segments may be only weakly
attracted to an interface with a binding energy Es << kT (where kT is the
thermal energy), whole polymer molecules may have a strong binding energy
nbEs >> kT (where nb is the number of bound segments). This leads to adsorption
that is effectively irreversible since the Boltzmann factor for desorption,
exp(-nbEe/kT), is negligibly small. The total adsorbed amount is in practice
constrained to be constant, essentially invariant to changes in polymer concen-
tration in the bulk phase over the experimental timescale. For flexible poly-
mers in the so-called "weak adsorption" limit (ref. 8), the average configura-
tions of adsorbed molecules are not very different from the random chain con-
figurations adopted in bulk solution. This is achieved by having only a frac-
tion of the segments directly bound to the surface ("train" segments) with the
rest protruding into the bulk solution as "loops" and "tails". Strong segmental
binding (Es >> kT), as may occur with highly charged polyelectrolytes adsorbing
on surfaces of opposite charge, leads to a flat configuration with a major part
of the segments in trains.
An important commercial application of polymeric flocculation of colloids
is in industrial and municipal waste water treatment processes (ref. 9). Typi-
cally, the polymers used are high-molecular-weight cationic polyelectrolytes
and the particles flocculated may range from clay particles to microorganisms.
Recent advances in understanding the mechanism of particle flocculation has come
from experimental studies of model colloidal systems flocculated by cationic
water-soluble polymers (ref. 10-23). Many other applications of polymeric
flocculation can be found in the technological processes of mineral processing
(ref. 24-27), separation of bacterial suspensions (ref. 28-33), sewage treat-
ment (ref. 34-38), paper making (ref. 39-47),brewing (ref. 48,49), ceramics
fabrication (ref. 50,51), pharmaceutical preparation (ref. 52,53), agrochemical
formulation (ref. 54) and food colloid stabilization (ref. 55,56).
Generally the flocculation processes in the technical systems are complicated
because the suspended particle populations normally have heterogenous properties.
Also, in many cases,the systems- especially those of biological origin-contain
biopolymers of nonionic, anionic or amphoteric nature adsorbed to the surfaces
and/or dissolved in the dispersion medium. The polymers and particles will inter-
act with added cationic polymer flocculants in either a more or less complex
way (ref. 29,30,33,35,36,41-44).
Mixing of oppositely charged polyelectrolytes may in the case of weak poly-
electrolytes result in complex coacervation which means separation of the system
3

into a polymer-rich phase and a polymer deficient phase (ref. 57,58). The
ratios between the different polyions are largely the same in both phases.
In the case of stronger polyelectrolytes, formation of polyelectrolyte com-
plexes will occur (ref. 59). The latter may, depending on component concentra-
tions and ionic strength, be in the form of dissolved complexes, colloidal dis-
persions or macroscopic precipitates (ref. 43,44).
It is thus obvious that such types of polymer-polymer interactions occur-
ring in the particulate systems may, depending on the conditions, give rise to
a considerable number of different polymer-particle interactions and hence
aggregation mechanisms. This may give disadvantages such as higher demands of
flocculant (ref. 29,30,36,41,42) and restrictions in flocculant efficiency and
floe properties (ref. 30). However, under suitable conditions it can be used to
an advantage asin the case of dual polymer retention systems in papermaking
(ref. 60). In this technique one combines the strong adsorption and anchoring
effect of highly cationic polyelectrolytes of low molecular weight with the
bridging effect of high-molecular-weight anionic polyelectrolytes. Since this
complex field is far from well understood, it will, in spite of its technologi-
cal importance, not be considered further in this paper. In order to give a
reasonably well established view, we will instead be restricted to systems with
a single type of polymer.
A simple overview of how polymers can affect the aggregation and stability

behaviour of an idealized colloidal suspension is illustrated in Fig. 1, taken


from Napper's monograph on polymeric stabilization (ref. 61). When adsorbing
polymer is added to a stable dispersion of (bare) particles at low polymer con-
centration (i.e., below that giving saturation coverage), the predominant effect
is bridging flocculation. At polymer concentrations for which particles are ful-
ly coated with polymer adsorbed from a good solvent, the polymer layers on dif-
ferent particles repel one another; this corresponds to the condition of steric
stabilization. When the sterically stabilizing layer is thin, the effect of the
short-range van der Waals interparticle attraction may lead to weak coagulation
for the same reason that (strong) coagulation occurs on addition of excess
electrolyte to non-polymer-coated charge-stabilized dispersions. Stronger dis-
placement coagulation may occur if the sterically stabilizing layer is comple-
tely displaced from the particle surface, e.g., by low-molecular-weight surfac-
tant. When the particles are large, or the solvent quality is reduced to poorer
than theta conditions, sterically stabilized particles may exhibit secondary-
minimum-type flocculation. Under certain circumstances, however, it is possible
to prepare a sterically stabilized dispersion in worse than theta solvent con-
ditions; this is referred to by Napper (ref. 61) as enhanced stabilization.
Addition of non-adsorbing polymer to bare particles or sterically stabilized
particles may lead to depletion flocculation or phase separation. A large con-
WEAK
FLOCCULATION COAGULATION

ION

ENHANCED
STA8ILIZATION

DEPLETION
OEPLE TION >
FLOCCULATION
STA~IL~ZAT~~N

Fig. 1. Schematic representation of the ways by which polymers can affect


colloid stability. [Reproduced with permission from ref. 611.

centration of non-adsorbing polymer may lead putatively to depletion (re)stabi-


lization. Addition of a second adsorbing polymer at low concentrations to a
sterically stabilized dispersion leads to bridging heteroflocculation.
This article is mainly concerned with the flocculation of colloidal particles
by adsorbing polymers. Typically, the mechanism involved is bridging floccula-
tion, or, with a polyelectrolyte of opposite net charge to that of the particle,
flocculation by charge neutralization. The article draws mainly on material

published during the past few years , although mention is necessarily made to
important earlier work in order to put recent research into its proper context.
Fuller descriptions of this earlier work can be found in previous reviews
(ref. 3,61-63) or the original papers.

2. KINETICS OF BRIDGING FLOCCULATION


In systems where there is no substantial polymer-particle charge neutraliza-
tion, the flocculation of colloidal particles by adsorbing polymer is generally
considered to occur via a bridging mechanism. That is, the floes are held to-
gether by bridges of individual polymer molecules spanning between the surfaces
of two or more particles. Bridging flocculation is favoured when the adsorbed
5

polymer molecules have long chains ("tails") dangling into the dispersed phase,
and when the amount adsorbed is below saturation coverage. Under these condi-
tions, when two particles collide, the free polymer tail on one becomes attached
to the free surface on the other.
Several different processes occur when adsorbing polymer molecules are added
to a stable dispersion in sufficient quantity to destabilize the colloid (ref.
64): mixing of polymer molecules amongst the particles, polymer adsorption at
the particle surface, rearrangement of initially adsorbed polymer towards the
"equilibrium" adsorbed state, collision of particles to form floes, and break-
up of floes in shear flow. To these processes can also be added reflocculation
after floe break-up and rearrangement within the floe. The dynamics of the
flocculation process is described by three characteristic time-scales: t, for
polymer adsorption, t, for adsorbed polymer rearrangement, and t, for particle
collision. While the macromolecular reconformation time t, is sensitive to the
detailed chemistry of polymer, particle surface and continuous phase, the ad-
sorption and collision times, ta and tc , may be estimated to a reasonable ap-
proximation using the classical equations of second-order reaction kinetics
(ref. 64-66). That is, we have

ta = -[ln (1 - f) 1 / kl,$O , (1)


t, = l/kllNo, (2)
where NO is the (initial) particle concentration, and f is the fraction of
polymer required to form stable floes. If the polymer-particle and particle-
particle encounters are diffusion-controlled, the rate constants, k12 and kll,
are given by the Smoluchowski expression for perikinetic coagulation,

kij = (2kT/3n) (ai + aj)2/aiaj, (3)

where k is Bolzmann's constant, T is the temperature, n is the viscosity of


the medium, and ai and aj are the radii of the colliding entities. In applying
eqn.(3) to polymer adsorption, it is assumed that each polymer molecule may
be represented as an effective hard sphere. For flocculation in shear, eqn.(3)
may be replaced by the Smoluchowski expression for orthokinetic coagulation,

kij = (46/3) (ai + aj)3, (4)

where G is the shear-rate.


The relative values of t, and t, have been estimated by Gregory (ref. 64,65)
for typical polymeric flocculation conditions. For particles of radius 1 urn
in water at ambient temperature, the diffusion-limited adsorption rate from
eqn.(3) exceeds the orthokinetic rate from eqn.(4) with G = 50 s-l only when
the effective hardsphere polymer radius is small ($ 50 nm). However, although
the absolute adsorption rate for high-molecular-weight polymers (radius > 100
6

nm) is faster in the orthokinetic case, it is very much slower than the floccu-
lation rate as long as the polymer radius is smaller than that of the particles.

However, if the polymer has larger dimensions than the particle, which is pos-
sible with the very high molecular weight products that are nowadays available,
the adsorption rate constant will be higher than that for the particle colli-
sions. In the perikinetic situation the flocculation is always slower than the
adsorption except when the particles and the polymers have the same effective
radius in which case the rates should be the same. In a stirred suspension with
particles of larger size than the polymers,a particle may undergo several colli-
sions with other particles before it acquires sufficient polymer for strong
enough attractive forces (e.g. by bridging) to occur. This is the explanation
for the observed time-lag between addition of polymer and floe formation in some
studies of polymeric flocculation (ref. 64,65). In this discussion it is very
important also to consider t,. Unfortunately, little information is available on
reconformation rates (ref. 64-66) and only a few studies to measure them have
been made (ref. 45,67). However, with sufficiently high particle-number densi-
ties, the rate of particle collision is faster than the rate of reconformation.
Thus, with the polymer far from its equilibrium conformation a sufficient num-
ber of bridges may be formed to give large floes with a polymer amount consider-
ably lower than that which could be adsorbed by the available particle surface.
It has been shown experimentally that the adsorbed amount decreases with in-
creasing particle concentration because adsorption sites are trapped within the
floes (ref. 68). This could explain the findings of Gill and Herrington (ref.
14) that,in a system with relatively high concentration of kaolin particles,the
adsorbed amount of an oppositely charged polyelectrolyte decreased with in-
creasing molecular weight which is in contrast to what is normally found in more
diluted systems (ref. 16). High-molecular-weight polyelectrolytes form floes by
bridging at lower concentrations than smaller ones (ref. 16), and with t,<<t,
more adsorption sites would be trapped within the floes. Larger molecules would
also be more restricted from entering into the trapped adsorption sites. This
matter will be discussed further later on in this paper.
It is also important to remember that most practical flocculation processes
are carried out in turbulent conditions. The particles and polymers are trans-
ported by turbulent eddies at much higher rates than in the laminar flow case
modelled by the Smoluchowski therory. The rates are critically dependent on the
particle and polymer sizes in relation to the turbulence microscale which is
determined by the energy dissipation rate. This area is far from well investi-
gated and understood (ref. 65) but a number of studies have shown its importance
(ref. 9,45,47,69-72).
In reality, not all particle-particle collisions lead to floe formation.
This was allowed for in the early work of La Mer and coworkers (ref. 73-77) by
assuming that the flocculation rate is given by

(5)
Jf = k11EN2,
where kll is the (Smoluchowski) collision frequency, N is the particle number
density, and E is the collision efficiency factor defined by

E = e(l - e). (6)

In eqn. (6), e is the fractional coverage of particle surface by adsorbed


polymer, and E is simply proportional to the product of the equivalent concen-
trations of bare and polymer-coated particles in the dispersion. The maximum
flocculation rate is predicted to occur at half coverage (e = 0.5).
Various authors (ref. 26,78-80) have recently proposed extensions to the
simple La Mer theory whilst maintaining the same semi-phenomenological approach.
A model in which the colliding particles and adsorbed polymers have freedom to
reorientate themselves into configurations favourable to bridging was intro-
duced by Hogg (ref. 78). The collision efficiency is given by

E = 1 _ e2n- (1 - 9)2n, (7)

where n is the number of adsorption sites per particle. For particles with a
large number of active sites, eqn. (7) gives Ez 1 for e # 0 or e # 1. For
n = 1, eqn. (7) reduces to

E = 2 e (1 - e), (8)

which is the correct formulation of the La Met- model if the statistics of


collisions between bare and coated surface patches are properly taken into
account. (That is, at e = 0.5, only 50 % of collisions are successful, as com-
pared with 100 % for normal Smoluchowski kinetics). Moudgil and coworkers (ref.
26,79) have taken account of particle surface heterogeneity by assuming that
there is a mixture of active and non-active sites, and that only active sites
can interact with polymer. In the Moudgil model, the fraction 6 of active
sites is equated with the fraction of surface covered at saturation. On colli-
sion, a bare patch does not necessarily bridge with a polymer-coated patch of
another particle: only if the bare patch is active does bridging occur. The
resulting collision efficiency factor is of the form

E = 2 e 62 (1 - e), (9)

which reduces to eqn. (8) when all surface sites are active (d = 1). In yet
another approach, Molski (ref. 80) has considered a kinetic model that includes
aggregation in the absence of polymer (coagulation), aggregation of sterically
stabilized particles (weak flocculation), as well as (strong) bridging floccu-
lation. For these three processes, the probabilities of particle-particle
sticking at collision are taken as Q, e and unity,respectively. This leads to
a collision efficiency factor of the form
8

E=l-(l- CL) (1 - e)2 - (1 - !3)e2. (IO)

With no coagulation (CX = 0) and no weak flocculation (6 = 0), eqn. (10) reduces
back to eqn. (8).
The La Mer model and its various extensions are deficient in a number of
ways. They do not allow for any enhancement of flocculation rate which might be
expected for capture of particles by polymer chains dangling from the surface
and thereby increasing the effective collision radius (ref. 81). Nor do they
take account of the fact that charged colloidal particles cannot approach arbi-
tralily closely on collision due to interparticle electrostatic repulsion. This
means that, even if a bare patch of active sites does exist, the polymer tail
may not physically be able to bridge between the two particles. This problem has
been addressed by Kashiki and Suzuki (ref. 82) using a model in which the floccu-
lating polymer is represented as a sphere of diameter 6, and the colloidal par-
ticles are represented as spheres with closest approach distance L calculated
from DLVO theory. The necessary condition for bridging flocculation to occur
effectively is 6 > L. This model is able to interpret the dependence of the
flocculation rate on the polymer molecular weight (which affects 6) and the
ionic strength (which affects L). In the case of polyelectrolytes, the ionic
strength also affects the dimensions of the polymer coil and hence the adsorp-
tion rate according to the discussion by Gregory referred to above (ref. 64,65).
Also absent from all these theories (ref. 73-82) is any allowance for polymer
rearrangement at the particle surface after initial adsorption, or indeed
hydrodynamic effects on adsorption and collision.
For bridging flocculation in charged systems, Fleer and coworkers (ref. 66)
have suggested that there are two limiting situations corresponding to so-called
"equilibrium" flocculation and "non-equilibrium" flocculation. In the former
case, particle collision occurs after the adsorbed polymer layer has relaxed
(tc >> t,), whereas in the latter case, nonrelaxed polymer tails become
bridged with other particles before the adsorbed polymer configuration has had
time to flatten (tc << tr). The behaviour is illustrated schematically in
Fig. 2. As with the treatment of Kashiki and Suzuki (ref. 82), the condition
for significant flocculation to occur is that the adsorbed polymer layer
thickness should exceed the closest approach distance L =: 2 k-l, where IC-~ is
the thickness of the electric double-layer. With low-molecular-weight polymers,
the layer thickness is too small for bridging irrespective of the relative
values of t, and t,, but addition of electrolyte after relaxation may lead to
equilibrium flocculation if 6 > 2 K-I. According to Pelssers et al. (ref. 66),
bridging may not occur even with high-molecular-weight polymers if ta >> tr,
since the number of polymer tails extending beyond 2 t~-l may be less than the
high
collision
rate

-U-

/\
fiequilibrium flocculation
low collision rate -
t

[O/
low attachment rate

no flocculation
A

sol\ additionA
\

no flocculalion equilibrium flocculation

Fig. 2. Illustration of the equilibrium and non-equilibrium flocculation of


charged colloidal particles by nonionic polymer. [Reproduced with permission
from ref. 661.

apparent required threshold. With ta < tr particles with active nonrelaxed


polymer tails are generated in appreciable concentration for the onset of non-
equilibrium flocculation. Whereas equilibrium flocculation is a slow second-
order process, non-equilibrium flocculation is not second-order, but is strong-
ly biased towards the formation of larger aggregates. This is because, once
most of the polymer has become adsorbed, there is a gradual reduction in the
number of active particles and hence a reduction in the flocculation rate.
For high-molecular-weight polymers, it is estimated (ref. 66) that t, is of
the order of several seconds.
In orthokinetic flocculation, floes of a limiting size are produced due to
the break-up of large aggregates by the shear field (ref. 76). Generally
speaking, aggregates formed by bridging flocculation are stronger than those
formed by salt-induced coagulation, a lthough once broken bridging floes may not
reform due to polymer relaxation. Glasgow and Ping Hsu (ref. 70) found that
kaolin floes produced with a high-molecular-weight polyacrylamide were more
than twice as strong as those formed with FeC13. Van de Ven (ref. 83) has esti-
mated that a shear-rate of G = lo3 s-1 is required to sever a single -C-C-
linkage on a polymer chain bridging between two 10 pm radius particles.
10

3. INTERACTIONS BETWEEN ADSORBED POLYMER LAYERS: EXPERIMENT


There has been considerable progress during the past few years in the
direct measurement of forces between polymer layers adsorbed on cleaved mica
surfaces. Using apparatus introduced by Israelachvili and Tabor (ref. 84) and
carefully developed and refined by Israelachvili (ref. 85,86),forces can be
measured to a sensitivity of 10 nN using a spring deflection technique, and
surface separations can be measured to better th-n 0.1 nm using optical inter-
ferometry. Experimental results are presented as plots of F(D)/R versus D,
where F is the force, D is the distance of closest approach, and R is the radius
of curvature of the cylindrical mica contours brought together in an orthogonal
cross-cylinder configuration. The absence of contamination from dust, polymers
or surfactants is indicated with hydrocarbon solvents by monotonic attraction
into adhesive contact between the mica surfaces. Highly symmetrical non-polar
solvents can give short-ranged oscillatory structural forces. Electrolyte solu-
tions give long-ranged repulsive interactions.
A collection of force versus distance data for polystyrene (and poly-or-
methylstyrene) adsorbed on mica from cyclohexane is shown in Fig. 3. The data
are taken from the recent review of Pate1 and Tirrell (ref. 87). Experimental
temperatures for the results in Fig. 3 range from 21C to 26C; this corre-
sponds to poor solvent conditions. It has been established (ref. 4) that, under
these conditions, the measured adsorbed layer thickness scales as the radius of
gyration, rg. In presenting data for polymers of different molecular weights in
a single plot, it is therefore appropriate to measure distance D in units of
rg. The repulsive force at close separations [D< (2/3)rg] is due to the com-
pression of chain configurations in the narrow gap which is entropically un-
favourable. The repulsion at close separations is a reflection of the condition
of constrained equilibrium. True equilibrium would involve the complete expul-
sion of polymer molecules from the gap, but this does not occur over the normal
experimental time-scale. At large separations (D & 3rg) the measured force is
negligible compared with the experimental scatter, but for D < 2rg attraction
sets in as the two layers interpenetrate. The maximum attraction occurs at
D zz rg. Under these poor solvent conditions, there are two contributions to
the measured attraction : (i) an enthalpic osmotic contribution due to enhanced
polymer-polymer attraction as compared with polymer-solvent attraction, and
(ii) a bridging interaction due to polymer molecules becoming adsorbed simul-
taneously to both surfaces. With polystyrene adsorbed from cyclohexane at
37C (theta solvent conditions), an attraction is still observed at D" rg,
but it is weaker than that shown in Fig. 3. This is interpreted as unambiguous
evidence for polymer bridging (ref. 88), since the osmotic attraction should be
absent under conditions of thermodynamic ideality. The strength of the bridging
interaction is enhanced by deliberately reducing the surface coverage below its
11

01
V 2 PS-CH
*3
n 4
4 5 P aMS-CH

F/R
(tWm> O

-500 I-

-1000 /-

-150( IL-
-0 1 2 3 4

Dhg

Fig. 3. Measurements of interactions between layers of polystyrene (PS) or


poly-a-methylstyrene (PctMS) adsorbed from cyclohexane (CH) below the theta
temperature (35C). The force F divided by the surface curvature R is plotted
against the separation D divided by the radius of gyration rg: (1) PS-CH,
6 x 105 daltons, 24C; 2) PS-CH, 6 x 105 daltons, 23C; (3 PS-CH, lo5 daltons,
21C; (4) PS-CH, 9 x 10 5 daltons, 26C; (5) PclMS-CH, 9 x 10 b daltons, 25C.
[Reproduced with permission from ref. 871.

normal level by incubating in polymer solution with the two surfaces close
together (D < 100 Ftmf (ref. 89,901.
Polymer bridging was studied directly by Granick et al. (ref. 91) in experi-
ments comparing the force between two saturated layers of poly-a-methylstyrene
with that between one saturated layer and one bare mica surface. From the
results shown in Fig. 4, we see that there is a 25-fold increase in the
maximum attractive force when one surface is left uncovered. The results are
consistent with an individual segment binding energy of Es =: (1/3)kT. In the
12

-3x10
10 20 30
D (nm)
Fig. 4. Force F/R versus distance D for mica surfaces immersed in cyclohexane
at 25C. (a) Each mica sheet carries 3 mg m-* of poly-a-methylstyrene. (b) One
mica sheet carries 3 mg m-* of poly -a-methylstyrene; the other is bare. [Repro-
duced with permission from ref. 911.

same study, it was observed (ref. 91) that the attractive minimum continued to
deepen during several minutes in contact, which indicates that long-time confi-
gurational changes are involved in the development of bridging interactions.
Evidence for polymer bridging in aqueous electrolyte is provided by the data
of Klein and Luckman (ref. 92) for polyoxyethylene adsorbed on mica. The F(D)
profiles have a similar form to those plotted in Fig. 3, but the attractive
13

minimum is about an order of magnitude smaller. Short-range attraction has


also been observed (ref. 93) with adsorbed layers of polypeptides and proteins,
although here the interaction is not thought to be due to bridging.

4. INTERACTIONS BETWEEN ADSORBED POLYMER LAYERS: THEORY


Two classes of theoretical model have been developed in recent years to
describe polymer adsorption and the interaction between adsorbed layers. These
are commonly referred to as the "mean field" approach and the "scaling" approach.
The central assumption of the mean field theory of polymer solutions is that
there is essentially a homogeneous distribution of segments so that a segment
placed at a certain location is acted upon by an average potential consistent
with the bulk polymer concentration (ref. 94). Correlations and fluctuations
in local segment densities are not allowed for in the mean field theory, and
so the theory is not applicable to highly swollen polymer coils in good solvents.
The central assumption of the scaling theory is the treatment of the polymer
solution as a collection of self-avoiding sub-chains called "blobs" whose size
in a semi-dilute solution decreases with increasing polymer concentration (ref.
95). Scaling theory is useful for weakly overlapping long chains in good sol-
vents, where the mean field approach gives incorrect results, but the segment
densities in adsorbed polymer layers are typically an order of magnitude too
high for scaling laws to apply (ref. 96).
A statistical mechanical theory which takes explicit account of the con-
figurations of adsorbed polymer chains is the Scheutjens-Fleer mean-field
lattice model (ref. 97,98). A polymer chain in the vicinity of a plane surface
is described by a random walk on a lattice with each step weighted according
to the local entropy of mixing and an energy factor describing nearest neigh-
bour interactions. There is a concentration gradient of segments in the direc-
tion z perpendicular to the surface, and this gives rise to a mean field which
depends on z. The space ajoining a single plane surface is divided into paral-

lel lattice layers 1, 2, .. . . i, .. . . M where i = 1 is the surface layer and


i = M is the bulk solution. In any layer i, di is the polymer segment volume
fraction and dp = 1 - di is the solvent volume fraction. The segmental weight-
ing factor is given by

p = 6: ev CX(<@l>- <&>)I exp (Xs) , (i = 1)


(11)
i
{ 4; exp [X(<Bi> - <El:>)], (i > 1)

where the parameter X is the Flory-Huggins polymer-solvent interaction energy


(ref. 94) and the parameter Xs is related to the net adsorption energy per
segment (ref. 99). The angular brackets in eqn.(ll) refer to weighted averages
over the layers i - 1, i and i + 1, i.e.,
14

<Bi> = Xl$i-1 + XOBIj + Al&i+1 3 (12)


where ~0 is the fraction of possible steps within the same layer, and XI is
the fraction into an adjacent layer. The segmental weighting factor Pi, to-
gether with the bond weighting factors h0 and XI, determine the probability
of any conformation occurring in the concentration gradient. The numerical
iteration procedure for determining the concentration profile {di) is described
in detail in the original papers (ref. 97,100).

Polymer bridging by adsorbing polymer in the gap between two plane surfaces
is predicted by the Scheutjens-Fleer theory (ref. 96,98,100). Fig. 5 shows
a plot of the interaction free energy as a function of the surface-to-surface
distance for lOOO-segment chains in a theta solvent ( xs = 1.0, X = 0.5) at
bulk polymer concentrations of PIP = 10-12, tip = 10-6 and dp = 10-Z. In very
dilute solution, there is an attractive energy of a few per sent of kT per
lattice site. Multiplied by a factor of lo3 lattice sites for a pair of colloi-
dal particles, this corresponds to an attraction consistent with what we expect
for bridging flocculation. In concentrated solution, the minimum disappears
and the two surfaces are sterically stabilized. The solid curves in Fig. 5
refer to the situation of constrained equilibrium in which polymer is trapped
in the gap, whereas the dashed curve refers to true equilibrium. In the latter
case, attraction is found at all concentrations, which means that, if complete
equilibrium were to occur, it would be impossible to stabilize colloidal disper-
sions with homopolymers. In poor solvents, the theory predicts an extra attrac-
tive contribution due to the enthalpic osmotic interaction, in agreement with
the experimental direct force results in Fig. 3. In theta solvents, the
attraction goes away at surface coverages above 3 monolayers; in athermal
solvents (X= 0), it disappears at coverages above I.5 monolayers (ref. 96).
In order to include the effects of segment density fluctuations in a statis-
tical description of adsorption, the mean-field approach must be replaced by
computer simulation (ref. 2). A general Monte Carlo procedure for generating
the configurations of adsorbed chains on a tetrahedral lattice was developed
by La1 and coworkers (ref. 101,102). Using this simulation technique, Clark and
La1 (ref. 103) studied the bridging of a IOl-segment chain between a pair of
plane walls. The data show that the probability of bridging increases with the
segment-surface interaction energy E, for weak adsorption, but decreases with
Es for strong adsorption. The optimum binding energy for bridging is Es Z 0.5
kT. This is large enough to facilitate adsorption, but not so large as to
restrict the extension of loops and tails from either surface. The simulations
were only carried out under athermal solvent conditions, although in principle
it would be straightforward to include segment-segment interactions of any
arbitrary degree of complexity.
15

Af
-ix
0.0

-0.c

-0.0

-0.0

Fig. 5. Bridging interaction between adsorbed polymer Tayers (1000 segments,


X= 0.5, xs = 1.0) according to Scheutjens-Fleer mean-field theory. The free
energy per lattice site nf (in units of kT) is plotted against the plate separa-
tion M (in lattice units). The solid curves refer to constrained equilibrium
at three polymer volume fractions: (a) BP = lo-l2 (3 monolayers); (b) BP = 10-6
(3.5 monolayers); (c) d, = 10-B (5 monolayers). The dashed curve refers to full
thermodynamic equilibrium. [Reproduced with permission from ref. 961.

5. NON-BRIDGING MECHANISMS OF FLOCCULATION


It is observed experi~ntally and predicted theoretically that sterically
stabilized suspensions will exhibit secondary-minimum-type flocculation if the
solvent quality is reduced to poorer than theta conditions (ref. 104). This
type of incipient flocculation can be induced, for instance, by increasing the
temperature of a dispersion stabilized by polyoxyethylene chains, or by adding
16

ethanol to a dispersion stabilized by polyhydroxystearic acid (ref. 105).


Flocculation induced by changing the temperature is readily reversible, and the
process can be treated as a phase transition using simple thermodynamic argu-
ments (ref. 106). The equilibrium between a dense flocculated phase and a
dilute dispersed phase is the colloidal analogue of the liquid-vapour transi-
tion (ref. 107), and sterically stabilized dispersions exhibiting coexisting
gas- and liquid-like colloidal phases have been observed experimentally (ref.
108,109). Statistical mechanical treatments of phase equilibria in polymer-
stabilized dispersions have recently been described by Gast et al. (ref. 110)
and Canessa et al. (ref. 111).
Bridging flocculation is not the only mechanism whereby a colloidal disper-
sion may be destabilized by addition of adsorbing polymer. In many instances,
the flocculation of charged colloidal particles by polyelectrolytes of opposite
net charge is more adequately described by a charge neutralization or charge
reversal mechanism (ref. 10). This is particularly the case when highly charged
water-soluble cationic polymers are mixed with suspensions of negatively
charged particles. At low or moderate ionic strength, theory predicts (ref.
112,113) that these polymers adopt a flat configuration at the interface, with
no significant loops and tails , and therefore no capacity for interparticle
bridging by polymer chains. Charge neutralization is associated with a large
segment-surface binding energy, but elementary geometric considerations show
that the particle surface charge density is not neutralized uniformly by the
adsorbed polymer. Rather, the resulting particle surface has a mosaic pattern
of positively and negatively charged patches as depicted in Fig. 6.

Fig. 6. Sketch of adsorbed cationic polymers on a colloidal particle of low


negative surface charge density. [Reproduced with permission from ref. 101.
17

This "electrostatic patch" model (ref. 114) was invoked by Gregory (ref. 10)
to account for an enhanced rate of flocculation observed with cationic poly-
mers of moderately high molecular weight. If positively charged patches are
of sufficient size, it is envisaged that encounters with negatively charged
patches on other particles will be sufficiently favourable to cause floccula-
tion, even though the net charge on the particles, were it to be uniformly
distributed, would be large enough to give electrostatic stabilization. Elabo-
rating on the electrostatic patch model, Mabire et al. (ref. 12) were able to
explain in a semiquantitative way the effects of molecular weight and polymer
charge density on settling rate and clarification of silica suspensions. At
high ionic strength, the mechanism is less likely to operate because of the
short range of the attraction between patches and the more extended adsorbed
polymer configurations which would tend to favour conventional bridging. This
is also the case when the charge density of the polymer is low.
Jonsson and coworkers (ref. 115,116) did Monte Carlo and mean-field studies
of two charged surfaces with short polyelectrolyte chains as counterions. The
chains were grafted at one end to the surfaces. They found attraction over a
wide range of parameter values due to an entropically driven bridging across
the gap between the surfaces. The term "bridging" as used by the authors does,
however, not mean that the polyelectrolyte chains have to be attached to both
of the surfaces as is understood in the traditional meaning of the term. It is
enough apparently if the chains are stretched over the midplane between the
surfaces.
The flocculation observed when non-adsorbing polymer is added to a colloidal
dispersion is now generally referred to as depletion flocculation. The physical
origin of the phenomenon was first recognized by Asakura and Oosawa (ref. 117)
for the case of two parallel plates immersed in a solution of rigid molecules.
When the plate separation is less than the solute molecular diameter, the
region between the plates must necessarily be composed of pure solvent. The
resulting osmotic pressure difference between the solvent in the gap and the
solution outside leads to an attractive force between the plates (corresponding
to an attractive energy of the order of kT for two colloidal particles). With
flexible macromolecules replacing rigid solute molecules, there is an additional
contribution to the interparticle force arising from the loss of configurational
entropy of the chains in the neighbourhood of the interface. Nevertheless, the
overall view of the phenomenon as a solute molecule excluded-volume effect re-
mains valid. In recent years there has been some success in describing the
general experimental trends of depletion flocculation in terms of statistical
mechanisms (ref. 61,110,111,118). The effect is more readily induced by higher
molecular weight polymers and larger particles. Recent theories (ref. 110,118)
are able to predict the complete phase diagram, including solid-liquid as well
18

as liquid-liquid transitions, although it should be noted that the dense solid


phase may take a long time to reach the equilibrium crystalline state, e.g.,
8 months for a dextran + polystyrene latex system (ref. 119). The factors
favouring depletion flocculation are similar to those favouring thermodynamic
phase separation into distinct polymer-rich and particle-rich-phases-both
phenomena are reversible, and both have their origin in the fact that polymer
molecules are forced to compete with colloidal particles for the available
space. Nevertheless, Napper argues emphatically (ref. 61) that depletion floccu-
lation and phase separation are radically different phenomena, in that the
colloidal particles remain "stable" if they undergo phase separation, yet appear
to be aggregated after undergojng depletion flocculation. In practice, however,
it may be quite difficult to distinguish the two if there is any tendency to-
wards repulsion in the polymer-particle pair interaction.

6. FACTORS AFFECTING FLOC STRUCTURE


Floe structure depends on the strength of poly~r-particle interactions, on
the nature of the flocculation mechanism, and on the ratio between particle
diameter and particle-particle separation in the floes. In the separation of
solid particles from liquid media by flocculation and sedimentation, three
desirable characteristics are identified (ref. 9). Firstly, the floes should be
large enough to sediment properly. Secondly, they should be strong enough to
withstand the potentially disruptive hydrodynamic forces that accompany sedimen-
tation and shear flow. Thirdly, the floe structure and permeability should be
such that dewatering can minimize the volume of separated solid material (e.g.,
sludge from waste-water treatment). In practice , it is not easy to achieve all
the desirable characteristics because of conflicting requirements in different
stages of the separation process. For instance, mixing in some form is required
to promote orthokinetic flocculation, but the shear-rates applied may result in
local hydrodynamic forces that exceed the floe yield stress, producing small
fragments which do not readily reflocculate. Another problem is that large floes
tend to be highly permeable, whereas floes of low permeability are the most
desirable for solid-liquid separation.
It is generally found that floe density (ref. 9,15,120,121) and strength (ref.
122) decrease with increasing floe size. On the basis of experimental results in
a system of quartz particles (2-10 urn) and high-molecular-weight nonionic poly-
acrylamide. Klimpel and Hogg (ref. 15) postulated three growth regions within
the density/size relationship. The first contains small, low-porosity microflocs
with nearly closepacked primary particles. In the second region, floes are
formed from the microflocs as basic units. The density/size follows a simple
power-law relationship of the type
19

(13)

where E is the porosity of an agglomerate of size X, Xc is a characteristic


floe size and ~1 is a constant typically around unity. The third region consists
of large floes of high porosity whose properties are sensitively dependent on
the prior history of growth. The density/size relationship depends on a number
of factors such as flocculant dosage (ref. 120,121), nature of the flocculant
(ref. 12l),primary particle size (ref. 15),and amount of mixing (ref. 15).
A useful parameter in characterizing the structure of large-scale floes and
sediments is the so-called fractal dimensionality introduced originally by
Mandelbrot (ref. 123). Clusters of particles produced by Brownian aggregation
processes are fractal in that their characteristic fractal dimensionality df

is less than the Euclidean dimensionality of 3. That is, the average radius R
is given by

R- nl/df , (14)

where n is the number of particles in the floe. The parameter df is useful for
characterizing the structure of colloidal sediments produced experimentally
(ref. 15,124) or by computer simulation (ref. 125). In a low-density sediment,
three spatial scales of structure may be identified: (a) short-range order from
packing and excluded volume effects, (b) medium-range disorder associated
with the fractal-type character of Brownian aggregation processes, and (c)
long-range uniformity for a material that is homogeneous on a macroscopic
scale. This description is formalized by defining a normalized particle-par-
ticle distribution function (ref. 126,127)

g(r) , (2a d r < Y)

G(r) = (r/E)dfc3, (V < r $ 5) (15)


r 1. (r > 5)

In eqn. (15), g(r) is the short-range radial distribution which can be deter-
mined in a scattering experiment (see below), a is the particle radius, and
the parameters y and 5 denote the lower and upper ranges of the fractal-
scaling regime. Computer simulation has shown (ref. 125-128) that, whereas the
short-range structure as characterized by g(r) is sensitive to the inter-
particle forces, the longer-range structure as characterized by df is relative-
ly unaffected by changes in the colloidal interactions. Renormalization group
theory shows (ref. 129) that the cross-over r from dense structure to fractal
structure is proportional to P- 0.85 where P is the sticking probability in
diffusion-limited aggregation (P = 1 for fast Smoluchowski coagulation, and
P + 0 for reaction-limited coagulation).
With polymeric flocculation, the chemical nature of the polymer is an
important factor affecting the structure and strength of aggregates. For the
20

technologically significant case of negatively charged particles flocculated


by cationic polymers, electrostatic interactions vary in strength depending on
the ionic strength of the dispersion medium and the charge density of the poly-
mer. It is valuable to distinguish between three situations (ref. 19): (i)
high-charge polymers, (ii) low-charge polymers, and (iii) intermediate-charge
polymers.
In suspensions containing high-charge polymers, the distance between charged
groups on the polymer molecule is considerably less than that between charged
groups on the particle surface. This leads to a mosaic of positive patches on
the surface (ref. 10,12,114). At low ionic strength, strong attraction between
oppositely charged patches produces extensive irreversible aggregation via
strong and inflexible particle-particle connections. The resulting floes have a
rigid, open structure, as obse-rved experimentally by Stewart and Sutton (ref.
13). This is the type of structure found with fractal aggregates formed by salt-
induced coagulation of an electrostatically stabilized dispersion in the ab-
sence of polymer (1.75 \< df < 2.05). Floes of this sort are appropriate for
filter-pressing where a rigid porous structure and the absence of free fine
particles are essential. With high-charge polymers at high ionic strength, the
electrostatic interaction between oppositely charged patches is reduced, and
this leads to smaller and weaker floes.
In suspensions containing low-charge polymers, the structure of the adsorbed
layer and the flocculation mechanism are similar to what are found with non-
ionic polymers. Instead of the flat configuration with a large part of the seg-
ments in trains, as found with high-charge polymers, there is considerable for-
mation of loops and tails due to the reduced segment-segment charge repulsion
in bulk solution. The presence of dangling tails gives bridging flocculation at
not too high a polymer concentration. It appears (ref. 16) that the bridging
mechanism operates at cationic charge densities below about 15%. Particles
aggregated by polymer bridges have larger freedom of local movement than patch-
aggregated particles. This looser interparticle connection enables rearrange-
ment into more compact floe structures of lower potential energy. Another differ-
ence is that mechanical energy may be dissipated in bridging floes; this in-
creases their shear resistance as compared with the rigid floes produced by
high-charge polymers.
For polymers of charge density between the high- and low-charge polymers,
it is possible for the average spacing of charged groups on the cationic poly-
mer molecule to match closely that on the particle surface. In such a situation,
opposite charges are effectively neutralized by adsorption of the highly extend-
ed polymer molecules in a flat (all-train) configuration. This means that
the contribution to flocculation from electrostatic patch attraction is much
less than with high-charge polymers, and that the contribution from bridging
21

is much less than with low-charge polymers. So, when the ionic strength is low,
the intermediate-charge polymers give smaller and weaker floes than either the
high-charge or low-charge polymers. At high ionic strength, however, the inter-
mediate-charge polymer will tend to adopt a more coiled structure in bulk solu-
tion, which will make adsorption in a flat configuration less likely. This leads
to a more pronounced heterogeneous patch structure and greater extension of seg-
ments from the surface, with the result that the average floe size and strength
are increased to about the same levels as for high- and low-charge polymers
under similar conditions. These arguments are based on studies in dilute sys-
tems where t, < t,. In more concentrated systems where t, << t, the above men-
tioned effect of polymer charge density may be less clearly distinguished.
Two other points about charge effects are worthy of brief mention at this
juncture. Firstly, the polymer charge density can have a significant effect on
flocculation efficiency even if the driving force for polymer adsorption is not
primarily electrostatic. This is because, as the charge density increases, the
polymer adopts a more extended conformationin solution, which makes polymer
bridging more effective (ref. 25). Secondly, even if the interaction between
the adsorbing polyelectrolyte and the surface is initially predominantly
electrostatic, the substantially nonpolar character of the polymer may mean
that the particle-particle interaction responsible for flocculation involves
a mainly hydrophobic interaction between polymer molecules adsorbed on differ-
ent particle surfaces. Parazak et al. (ref. 18) have suggested a "hydrophobic
patch" mechanism which they think may be more important than either electro-
static patch flocculation or bridging flocculation under certain circumstances.

7. NEUTRON SCATTERING STUDIES OF FLOC STRUCTURE


The application of small-angle neutron scattering to investigate the struc-
ture of polymer-flocculated monodispersed silica particles has been reported
by Cabane and coworkers (ref. 130-132). The technique is based on the fact that
the interference pattern of the scattered radiation is determined by the distri-
bution of separations between particle centres in the floes. Scattering from
the polymer makes a negligible contribution.
Information on floe structure is contained in the structure factor S(Q).
The scattering vector Q depends on the wavelength of the neutrons and the scat-
tering angle. The Fourier transform of S(Q) is the radial distribution function
g(r), where g(r) dr is the probability of finding another particle at a dis-
tance between r and r + dr from a given particle (ref. 133). For a liquid-like
colloidal dispersion, g(r) is zero for r less than the particle diameter, it
has a major peak at r = d0 corresponding to the average nearest-neighbour pair
separation, and it decays away in an oscillatory fashion with successively
smaller ones atr" 2d0 ("second shell") and r Z 3d0 ("third shell") reaching
22

g(r) = 1 at large separations. Such a system shows a peak in S(Q) near Q =


2rr/dO. For an isotropic system of n spherical particles, the scattered intensity
I(Q) is proportional to the square of the single-particle amplitude AI(Q):

I(Q) = n[Al (Q) I2 S(Q). (16)


The single-particle scattering amplitude depends on the difference in scatter-
ing densities for the dispersed phase and the dispersion medium, the volume
of the particle, and the particle form factor (ref. 134)

P(Q) = rt3/(QR131 [sin(QR) - (QR) cos (QR)112 I (17)

where R is the particle radius. In scattering from a colloid with short-range

liquid-like order, there is a peak in I(Q) corresponding to the peak in S(Q) at


Q" 2a/dO. In scattering from a fractal-type structure, however, I(Q) follows
a simple power-law decay according to the distribution of distances around a
given particle (ref. 135).
Plots of log [I(Q)] against log Q have been examined (ref. 130-132,135) for
dispersions flocculated by polymers of variable charge density. Two types of
aggregates could be distinguished. Type I aggregates are formed without full

neutralization or screening of the charge borne by the particles. The scatter-


ing curve I(Q) has a short-range order peak at Q" 2n/dO and a strong depres-
sion in I(Q) for Q < 2n/dO. In these floes, produced by uncharged or low-charge
polymers of high molecular weight, each particle has a full co-ordination shell
of nearest neighbours at r = d0 Z 3R. At very small Q (<< 2n/dO), however,

I(Q) follows a power-law decay, corresponding to fractal-type structure at


large distances.
Type II aggregates are formed through neutralization or screening of the
negative charges on the silica particle surface. This corresponds to patch
flocculation by intermediate or high-charge polymers, and also to flocculation
at low pH, near the isoelectric point of silica, or at high ionic strength.
The scattering curve shows high intensity at low Q and a slow power-law decay
which is characteristic of tenuous objects full of voids at all length scales.
At Q " 2~/2R, there is a weak shoulder corresponding to a first co-ordination
shell made mostly of voids with a few neighbours stuck directly in contact.
These type II aggregates are fractal-like, with no short-range liquid-like
order. Their structure is very similar to that for aggregates formed from
charge-stabilized suspensions by addition of excess electrolyte (df 1.75).
The neutron scattering experiments indicate that floes formed by bridging
(type I aggregates) have a different short-range structure from those formed
by electrostatic patch neutralization (type II aggregates). The close contact
of surfaces, which occurs with charge neutralization, prevents any relative
motion of particles after collision. This results in a rigid, branched, tenuous
23

structure at all length scales, including the shell of first neighbours, since
holes are not filled up, but are permanently screened by the rest of the aggre-
gate. In contrast, bridging by long macromolecules, with retention of short-
range repulsion, allows loosely connected spheres to reorganize themselves into
lower potential energy states, thereby making space for additional spheres to
become bound. This results in floes with short-range liquid-like order in
which each particle is surrounded by a large number of neighbours. The co-
ordination number increases with the molecular weight of the bridging low-

charge polymer, but the interparticle spacing d0 is controlled by the residual


electrostatic repulsion and is independent of the polymer molecular weight
(ref. 132). This contrasts with the type II aggregates whose structure is
wholly independent of the polymer molecular weight.

8. COMPUTER SIMULATION OF FLOC STRUCTURE


While there has been a lot of activity in recent years in the computer
simulation of particle aggregation into large-scale fractal-type structures
(ref. l36,137),there has been relatively little attention directed so far to-
wards the structure of polymer-flocculated aggregates. Nevertheless, some of
the results obtained with existing models can be of some value in interpreting
experimental behaviour in polymer-containing systems. For instance, the differ-
ence between the type I and type II aggregates noted by Cabane and coworkers
(ref. 130-132) from neutron scattering experiments shows a resemblance to the
difference between computer simulated aggregates of Ansell and Dickinson
(ref. 127,128) for the irreversible Brownian coagulation of colloidal particles
with and without a potential energy barrier (a small "primary maximum" in DLVO
theory). That is, aggregates formed from particles interacting with a small
primary maximum and a secondary minimum show short-range liquid-like structure
in g(r) similar to that found with the bridging floes examined by neutron
scattering. On the other hand, aggregates formed from particles with just
short-range attraction in the interparticle potential have a fractal-type
structure over all length scales similar to the experimental floes produced
by electrostatic patch neutralization. Both computer simulation and scattering
experiments indicate that the condition for short-range liquid-like ordering
is the ability of the floe to rearrange after the particles have come close
together but are not in direct contact.
At a very simple level, a system of polymer molecules + dispersed particles
can be modelled as a binary mixture of small and large spheres interacting with
a square-well potential of the form

m, (r < d12)
ulz(r1 = E , !d12 < r 6 D12) (18)
i 0, (r > D12)
24

where dl and d2 are the small and large sphere diameters, d12 = (dl + d2)/2
is the unlike hard-core diameter, and e is an attractive energy of fixed range
D12. When the polymer-particle interaction energy E is of the order of just a
few kT, equilibrium properties can be determined by Monte Carlo simulation
(ref. 138) in a periodic cell with configurations generated according to stan-
dard procedures. The use of the square-well potential Ceqn. (ET)] removes any
ambiguity in clearly distinguishing between adsorbed and unadsorbed polymers,
and between flocculated and unflocculated particles. That is, if two colloidal
particles (diameter d2) both have centres lying within a distance D12 of the
same polymer molecule (diameter dl), then they are considered to be bridged
together in the same floe. When the well width tends to zero (012 + d12), the
model reduces to a binary mixture of hard spheres of different sizes inter-
acting with an unlike sticky hard-sphere potential (ref. 139). The dependence
of the average cluster size on system composition can be calculated from
simple algebraic expressions for this type of sticky sphere bridging floccula-
tion. Recent calculations show (ref. 140) that relatively weak attractive
interactions (1 or 2 kT) may cause extensive weak reversible bridging floccuTa-
tion (and gelation) in concentrated dispersions.
The square-well potential model, as given by eqn.(l8), exhibits bridging
flocculation and depletion when studied by Monte Carlo simulation
flocculation
in either two or three dimensions (ref. 141). Fig. 7 shows configurations of
a concentrated polymer + particle system in two dimensions with dl = 1, d2 = 7,
and D12 = 4.5. Picture (a) is a "snapshot" of an assembly of 500 polymer mole-
cules (volume fraction 81 = 0.070) + 50 colloidal particles ($62 = 0.342) with
a polymer-particle square-well attraction of E = -3.0 kT. Bridging of the big
spheres by the small ones is clearly evident, and the size of the largest floe
in the system is of the order of the size of the basic simulation cell. Con-
figuration (b) refers to an assembly of 950 polymer molecules (61 = 0.133) +
50 colloidal particles (82 = 0.342) with a polymer square-well repulsion of
E = + 2.0 kT. Here, depletion flocculation of the large spheres is evident,
with regions rich in polymer separated from particle-rich regions depleted of
polymer. Both types of simulated floes, whether characterized by bridging
(Fig. 7 a) or depletion (Fig. 7 b), have a rather disordered structure. A
similar type of structure has also been found (ref. 142) in Monte Carlo simu-
lation of the reversible flocculation of a single-component square-well system
(i.e., with no added polymer) for small values of the energy parameter E . Com-
pact floes are formed, however, when the system exhibits phase separation
(liquid-gas) at larger values of E .
A distinct weakness of the models described above is the treatment of the
polymer as a small sphere with no chain-like character. A preliminary two-
25

Eig. 7. Monte Carlo configurations of a two-dimensional system of polymers +


particles with square-well polymer-particle potential [eqn. (18)]. (a) Bridging
flocculation with E = -3 kT. (b) Depletion flocculation with e = + 2 kT.
[Reproduced with permission from ref. 1411.

dimensional simulation of irreversible particle aggregation by polymer chains


has been studied by Jullien et al. (ref. 143). Based on the analysis of a single
190-particle floe, the computed fractal dimension is significantly larger than
that for reaction-limited cluster-cluster aggregation. It appears, therefore,
that the effect of the polymer is to increase the aggregate packing density com-
pared with that for irreversible aggregation in the absence of polymer.Further
work on this sort of simulation model is required, preferably in three dimen-
sions, in order to study the effect on floe structure of rearrangements after ad-
26

sorption, and to examine more complex types of particle-particle, particle-


polymer and polymer-polymer interactions.

REFERENCES

1 Yu.S. Lipatov and L.M. Sergeeva, "Adsorption of Polymers", Wiley, New York,
1974.
2 E. Dickinson and M. Lal, Adv. Molec. Rel. Int. Proc., 1980, 17, 1.
3 B. Vincent and S.G. Whittington, in "Surface and Colloid Science", ed. E.
Matijevie, Plenum, New York, 1982, vol. 12, p. 1.
4 A. Takahashi and M. Kawaguchi, Adv. Polym. Sci.,1982, 46, 1.
5 M.A. Cohen Stuart, T. Cosgrove and B. Vincent, Adv. Colloid Interface Sci.,
1986, 24, 143.
6 P.-G. de Gennes, Adv. Colloid Interface Sci., 1987, II_, 189.
G.J. Fleer, Surfactant Sci. Ser., 1988, 27, 105.
: P.-G. de Gennes, Macromolecules, 1981, 14, 1637.
9 L.A. Glasgow, Chem. Eng. Prog., 1989, 8x8), 51.
10 J. Gregory, J. Colloid Interface Sci.,T973, 9, 448; 1976, 55, 35.
11 M.D. Sikora and R.A. Stratton, Tappi J., 1981, 64, 97.
12 F. Mabire, R. Audebert and C. Quivoron, J. Colloid Interface Sci., 1984,
97, 120.
13 R.F. Stewart and D. Sutton, in "Morphology and Properties of Particle/
Polymer Suspensions", eds. L.L. Hench and D.R. Ulrich, Wiley, New York,
1986, p. 455.
14 R.I.S. Gill and T.M. Herrington, Colloids Surf., 1986, 2, 51; 1987, 25,
297; 1987, 8, 41; 1988, 2, 331.
15 R.C. Klimpel and R. Hogg, J. Colloid Interface Sci., 1986, 113, 121.
16 G. Durand-Piana, F. Lafuma and R. Audebert, J. Colloid Interface Sci.,
1987, 119, 474; Prog. Colloid Polym. Sci., 1988, 76, 278.
17 T.K. Wang and R. Audebert, J. Colloid Interface Sci., 1987, 119, 459;
1988, 121, 32.
18 D. Parazak, C.W. Burkhardt, K.J. McCarthy and M. Stehlin, J. Colloid
Interface Sci., 1988, 123, 59.
19 L. Eriksson, B. Alm and L. Alden, in "Flocculation and Dewatering", eds.
B.M. Moudgil and B.J. Scheiner, Engineering Foundation, New York, 1989,
p. 179.
H. Kage, Y. Matsuno and K. Higashitani, Can. J. Chem. Eng., 1988, 66, 728.
N.J.D. Graham, Water Res., 1988, 22, 1229.
R. Pelton and D. Lawrence, ColloidPolym. Sci., 1989, 267, 907.
Y. Otsubo and K. Watanabe, J. Colloid Interface Sci., 1989, 127, 214.
M.J. Littlefair and N.R. Lowe, Int. J. Miner. Proc., 1986, 17, 187.
R. Jin, W. Hu and X. Hou, Colloids Surf., 1987, a, 317.
B.M. Moudgil, B.D. Shah and H.S. Soto, Miner. Metall. Proc., 1987, 27;
J. Colloid Interface Sci., 1987, 119, 466.
H. Bustamante and P.R. Rutter, Chem. Eng. Sci., 1987, 42, 809.
22: G.P. Treweek and J.J. Morgan, J. Colloid Interface Sci, 1977, 60, 258.
29 W.C. McGregor and R.K. Finn, Biotech. Bioeng., 1979, 2, 873.
30 L. Eriksson and A.-M. Hlrdin, in "Flocculation in Biotechnology and
Separation Systems" , ed. Y.A. Attia, Elsevier, Amsterdam, 1987, p. 441.
31 G.V. Tarasova, A.Ya. Teslenko, E.N. Lazarenko and V.V. Alekseeva, Colloid
J. USSR, 1986, 47. 629.
32 A.A. Baran, ColEids Surf., 1988, 31, 259.
33 I. Agerkvist, L. Eriksson and S.-O. Enfors, Enzyme Microbial Techn., In
press.
34 A.E. Steiner, D.A. McLaren and C.F. Forster, Water Res., 1976, lo, 25.
35 J.T. Novak and B.-E. Haugan, J. Wat. Pollut. Control Fed., 1980, 52, 2571.
36 L. Eriksson, Water Sci. Tech., 1987,~2, 859.
37 L. Eriksson and A.-M. Hardin, Water Sci. Tech., 1984, 16, 55.
27

38 L. Eriksson and B. Aim, in "Proceedings of 5th Internatjonal Conference on


Advanced Wastewater Treatment and Reclamations, Kracow, Poland, 25-27 Sept.
1989.
39 T. Lindstrom, C. S&remark and L. Eklund, Pulp Paper Mag. Transactions,
1977, 3, 114.
40 T. Lindstrlim and L. Wagberg, Tappi J., 1983, 66, 83,
41 6. Striim, P. Barla and P. Stenius, Sven. Papperstidn., 1979, 82. 408.
42 G. Strom, P. Barla and P. Stenius, Sven. Papperstidn., 1982, g, RlOO.
43 G. Striim and P. Stenius, Colfoids Surf., 1981, 2, 357.
44 6. Strom, P, Barla and P. Stenius, Colloids Surf., 1985, l3_, 193.
45 L. Wagberg, Ph.D. thesis, Royal Institute of Technology, Stockholm, Sweden,
1987.
46 L. Wagberg and T. LindstrQm, Colloids Surf., 1987, 27, 29.
47 M. Falk, L. Odberg, L. Wlgberg and G. Risinger, Col%ids Surf., 1989, 40,
115.
48 H. Nishihara and T. Toraya, Agric. Biol. Chem., 1987, 51, 2721.
49 J.C. Kihn, C.L. Masy and M.M. Mestdagh, Can. J. MicrobFl., 1988, 34, 773.
50 K. Kendall, N.McN. Alford, W.G. Clegg and J.D. Birchall, Nature (London),
1989, 339, 130.
51 D.C. Agrawal, R. Raj and C. Cohen, J. Amer. Ceram. Sot., 1989, 11, 2148.
52 S. Shimabayashi, M. Okuda.and M. Nakagaki, Chem. Pharm. Bull., 1988, 36,
1257.
53 S. Shimabayashi, M. Okuda, M. Yagiu and M. Nakagaki, Colloid Polym. Sci.,
1987, 265, 1099.
54 D.J. Wedlock, I.J. Fabris and J. Grimsey, Colloids Surf., 1990, 9, 67.
55 E. Dickinson, in Gums and Stabilisers for the Food Industry", eds. G.O.
Phillips, 0.3. Wedlock and P.A. Williams, IRL Press, Oxford, 1988, vol. 4,
p. 249.
56 I. Shomer, J. Sci. Food Agric., 1988, 42, 55.
57 H.G. Bungenberg de Jong, in "Colloid Science II, ed. W.R. Kruyt, Elsevier,
Amsterdam, 1949, p. 335.
58 A. Veis, in ~Biological Polye~ectrolytes~, ed. A. Veis, Marcel Dekker,
New York, 1970, p. 211.
59 E. Tsuchida and K. Abe, Adv. Polym. Sci., 1982, 45, 1.
60 T. Petljl, Kemia-Kemi 1980, 7, 261.
61 D.H. Napper, "Polymeric Stabylization of Colloidal Dispersions", Academic
Press, London, 1983.
62 J. lykfema, Adv. Colloid Interface Sci., 1968, 2, 65.
63 B. Vincent, Adv. Colloid Interface Sci., 1974, 2, 193.
64 J. Gregory, in "The Effect of Polymers on Dispersion Properties", ed. Th.F.
Tadros, Academic Press, London, 1982, p. 301.
65. J. Gregory, Colloids Surf., 1988, 2, 231.
66 E.G.M. Pelssers, M.A. Cohen Stuart and G.J. Fleer, Colloids Surf., 1989,
38, 15.
67 M.A. Cohen Stuart and H. Tamai, Colloids Surf., 1988, 31, 265.
68 H. Tanaka, L. Ddberg, L. Wagberg and T. Lindstriim, J. Colloid Interface
Sci., 1990, f34, 219.
69 Y. Argaman and W.J. Kaufman, J. Sanitary Engineering Div. Proc.American
Chem. Sot. Civ. Eng., 1970, x, 223.
70 L.A. Glasgow and J.Y.H. Ping HSU, AIChE Journal, 1982, 28_, 779.
71 G.B.J. De Boer, G.F.M. Hoedemakers and 0. Thoenes, Chem. Eng. Res. Des.,
1989, 67, 301.
72 G.B.J. De Boer, C. De Weerd and D. Thoenes, Chem. Eng. Res. Des., 1989,
67, 308.
73 V.K. La Mer and R.H. Smellie, Jr., J. Colloid Sci.. 1956, 11, 704.
74 R.H. Smellie, Jr., and V.K. La Mer, J, Colloid Sci., 1958, 23, 589.
75 T.W. Healy and V.K. La Mer, J. Phys. Chem., 1962, 66, 1835.
76 W.K. La Mer and T.W. Healy, Rev. Pure Appl. Chem., 1963, l.& 112.
77 V.K. La Mer, Disc. Faraday Sot., 1966, 42_, 248.
78 R. Hogg, J. Colloid Interface Sci., 1984, 102, 232.
28

B-19. Moudgil and T.V. Yasudevan, J. Colloid Interface Sci., 1989, 127, 239.
A. Molski, Colloid Polym. Sci., 1989, 267, 371.
W.E. Walles, J. Colloid Interface Sci., 1968, 2f, 797.
I. Kashiki and A. Suzuki, Ind. Eng. Chem. Fundam., 1986, 2& 444.
T.G.M. van de Ven, J. Colloid Interface Sci., 1981, &, 290.
J.N. Israelachvili and D. Tabor, Proc. Roy. Sot. (London), 1972, m,
19.
J-N. Israelachvili, Proc. Natl. Acad. Sci. USA, 1987, 84, 4722.
J.N. Israelachvili, Act. Chem. Res., 1987, 0, 415.
S.S. Pate1 and M. Tirrell, Annu. Rev. Phys. Chem., 1989, 4& 597.
J. Klein and P. Pincus, Macromolecules, 1982, 3, 1129.
J.N. Israelachvili, M. Tirrell, J. Klein and Y. Almog, Macromolecules,
1984, 11, 204.
90 Y. Almog and J. Klein, J. Colloid Interface Sci., 1985, m, 33.
S. Granick, S. Pate1 and M. Tirrell, J. Chem. Phys., 1986, 85, 5370.
;: J. Klein and P.F. Luckham, Nature (London), 1984, 308, 836.
93 T. Afshar-Rad, A.I. Bailey, P.F. Luckham, W. Ma~naughtan and D. Chapman,
Colloids Surf., 1988, 3l_, 125.
94 P.J. Flory, "Principles of Polymer Chemistry", Cornell Univ. Press, Ithaca,
1953.
95 P.-G. de Gennes "Scaling Concepts in Polymer Physics", Cornell Univ. Press,
Ithaca, 1979.
96 G.J. Fleer, J.M.H.M. Scheutjens and M.A. Cohen Stuart, Colloids Surf.,
1988, 3, 1.
97 J.M.H.M. Scheutjens and G.J. Fleer, J. Phys. Chem., 1979, 83, 1619;
1980, 3, 178.
98 G.J. Fleer and J.M.H.M. Scheutjens, J. Colloid Interface Sci., 1986, 111,
504.
99 A. Silberberg, J. Chem. Phys., 1968, 48, 2835.
100 J.M.H.M. Scheutjens and G.J. Fleer, Macromolecules, 1985, I8_, 1882.
101 M. Lal, K.A. Richardson, D. Spencer and M.A. Turpin, ACS Symp. Ser., 1975,
8, 16.
I02 A.T. Clark and M. Lal, J. Chem. Sot., Faraday Trans. 2, 1978, 2, 1857.
103 A.T. Clark and M. Lal, in "The Effect of Polymers on Dispersion Properties,
ed. Th.F. Tadros, Academic Press, London, 1982, p. 169.
104 Th.F. Tadros, in "The Effect of Polymers on Dispersion Properties", ed.
Th.F. Tadros, Academic Press, London, 1982, p. 1.
105 D.H. Napper, Trans. Faraday Sot., 1968, 64, 1701.
106 D.H. Napper, J. Colloid Interface Sci., 1977, 3, 390.
107 E. Dickinson, Annu. Rep. Prog. Chem., Part C, Royal Society of Chemistry,
London, 1983, p. 3.
108 J. Long, D.J.W. Osmond and B. Vincent, J. Colloid Interface Sci., 1973,
42_, 545.
109 P.W. Rouw, A, Vrij and C.G. tie Kruif, Colloids Surf., 1988, 2, 299.
110 A.P. Gast, C.K. Hall and W.B. Russel, Faraday Discuss. Chem. Sot., 1983,
f6, 189; J. Colfoid Interface Sci., 1983, 96_, 251; 1986, 109, 161.
111 E. Canessa, M.J. Grimson and-M. Silbert, Phys. Chem. Liq., 1988, 18,
287; Molec. Phys., 1989. 67, 1153.
112 J. Lyklema and G.J. Fleer, Colloids Surf., 1987, 25, 357.
113 J. Papenhuijzen, H.A. van der Schee and G.J. Fleer, J. Colloid Interface
Sci., 1987, 104, 540.
114 D.R. Kasper,?i-Theoretical and experimental investigations of the floccula-
tion of charged particles in aqueous solutions by polyelectrolytes of
opposite charge", Ph.D. thesis, California Institute of Technology, 1971.
115 T. Akesson, C. Woodward and B. Jiinsson, J. Chem. Phys., 1989, 3, 2461.
116 S. Miklavic, C.E. Woodward, B. Jbnsson and T. Akesson, Macromolecules,
Accepted for publication.
117 S. Asakura and F. Oosawa, J. Chem. Phys., 1954, 22, 1255.
118 B. Vincent, J. Edwards, S. Emmett and R. Croot, Colloids Surf., 1988, 31,
267
119 P.D. Pate1 and W.B. Russel, 3. Colloid Interface Sci., 1989,131, 192.
29

120 N. Tambo and Y. Watanabe, Water Res., 1979, 13, 409.


121 B.M. Moudgil and T.V. Vasudevan, Mineral Metall. Process., 1989, 5, 142.
122 D.H. Bathe and S.H. Al-Ami, Water Sci. Tech., 1989, 21, 529.
123 B.B. Mandelbrot, "Fractals: Form, Chance and Dimensi,n", Freeman, San
Francisco, 1977.
124 M. Zrinyi, M. Kabai-Faix and F. Horkay, Prog. Colloid Polym. Sci., 1988,
z, 165.
125 E. Dickinson, Colloids Surf., 1989, 39, 143.
126 E. Dickinson, J. Colloid Interface Sci., 1987, 118, 286.
127 G.C. Ansell and E. Dickinson, Faraday Discuss. Chem. Sot., 1987, @, 167.
128 G.C. Ansell and E. Dickinson, J. Chem. Phys., 1986, 85, 4079; Phys. Rev.,
1987, s, 2349.
129 T. Nagatani, Phys. Rev., 1989, 40A, 7286.
130 K. Wong, B. Cabane and R. Duplezx, J. Colloid Interface Sci., 1988, 123,
466.
131 K. Wong, B. Cabane and P. Somasundaran, Colloids Surf., 1988, 30, 355.
132 B. Cabane, K. Wong, T.K. Wang, F. Lafuma and R. Duplessix, ColEid Polym.
Sci., 1988, 266, 101.
133 E. Dickinsonzn "Colloid Science", ed. D.H. Everettt, Specialist Periodic
Reports, Royal Society of Chemistry, London, 1983, vol. 4, p. 150.
134 B. Jacrot, Rep. Prog. Phys., 1976, 2, 911.
135 B. Cabane, K. Wong and R. Duplessix, ACS Symp. Ser., 1987, 384, 312.
136 R. Jullien and R. Botet, "Aggregation and Fractal Aggregate= World
Scientific, Singapore, 1987.
137 P. Meakin, in "Phase Transitions and Critical Phenomena", eds. C. Domb and
J.L. Lebowitz, Academic Press, London, 1988, vol. 12, chap. 3.
138 M.P. Allen and D.J. Tildesley, "Computer Simulation of Liquids", Clarendon,
Oxford, 1987.
139 J.W. Perram and E.R. Smith, Proc. Roy. Sot. (London), 1977, @, 193.
140 E. Dickinson, J. Chem. Sot., Faraday Trans., 1990, 3, 439.
141 E. Dickinson, J. Colloid Interface Sci., 1989, 132, 274.
142 E. Dickinson, C. Elvinqson and S.R. Euston, J. Chem. Sot., Faraday Trans.
2, 1989, 8& 891. -
143 R. Jullien, R. Botet and P.M. Mors, Faraday Discuss. Chem. Sot., 1987, 83,
125.

You might also like