13.

Coagulation

13.1. The nature of coagulation

Coagulation relies on aggregation of fine particles suspended in water into a larger
unit called coagulum which, due to gravity force, settles down. Figure 13.1. presents a
simplified picture of coagulation.

coagulation

time
Fig. 13.1. Coagulation

Coagulation of particles is applied in mineral processing, clarification of water,

desliming aqueous suspensions for reuse in flotation and upgrading of ores containing
very fine particles. When coagulation involves particles of different materials, the
process is called heterocoagulation. When coagulation takes place as a result of adhe-
sion to air bubbles and solid particles, it is called flotation, while adhesion of fine to
large particles is called slime coating. Coagulation can be carried out selectively and
non-selectively. During selective coagulation one kind of particles present in suspen-
sion undergoes coagulation, while others do not. Removal of coagulated material,
forming a sediment, from uncoagulated particles is accomplished by decantation,
drainage, syphonation, etc.
Although coagulation process seems to be simple, its description is complex since
coagulation depends on many parameters (Fig. 13.2) which can be divided into those
effecting probability of particle collision PZ, governing particle adhesion Pa, and pa-
rameters effecting stability of newly formed coagulates Pstab Also time significantly
 13. Coagulation 399

influences coagulation process. The main parameters governing coagulation, groupped

into Pz, Pa, and Pstab, are related and form a pyramid-like structure.

Pa

W
Vmax

VA,VR,VS
A, ψ , κ , x, θ , H
ni,ε , ζ ,ν ν ,cs,pH,γ
xc / xj V*max
λ .....
hd H, cs
Pz
Pstab
Pk = PzPaPstab
Fig. 13.2. Static delineation of coagulation. Meaning of symbols: Pz – probability of collision of particles,
Pa – probability of adhesion of particles, Pstab – probability of stability of coagula, W – stability ratio, Vmax
– energy barrier of coagulation, H – distance between particles, VA – energy of dispersion interactions, VR
– energy of electrostatic interactions, Vs – energy of structural interactions, A – Hamaker constant, ψ –
edl potential, 1/κ – edl thickness, xc – collision radius, x – particle size, θ – hydrophobicity (contact an-
gle), n – refraction index, ε – dielectric constant, νν – characteristic frequency, cs – electrolyte concentra-
tion, pH – acidity of the solution, γ – surface energy, T – temperature, g – gravity, n1 – concentration of
particles, V*max – energy barrier of peptization, ρ – particle density, λ – wavelength of electromagnetic
beam, hd – system hydrodynamics

Figure 13.2 shows that many different parameters contribute to the probability of
coagulation Pk, and they can be expressed as a product of probabilities of sub-
processes:
Pk = Pz Pa Pstab. (13.1a)
Relationships between particular probabilities and physical parameters of coagula-
tion system are not fully known. Generally, Pz can be expressed as a function of size
of coagulating particles (xi and xj), as well as of their collision radius xz and hydrody-
namics hd, Pa as a function of the so-called stability ratio W while Pstab as a function
energy barrier of decoagulation Vmax.
Pz = f (xc, xi, xj, hd), (13.1b)

Pa = f (W) = 1/W, (13.1c)

Pstab = f (V*max). (13.1d)

 400 Part III. Separation processes

It means that the main parameter of coagulation process is stability ratio W. As it

was shown in Fig. 13.2 the main parameter is dependent on other parameters. The net
of related parameters of coagulation may contain as many as 100 parameters
(Fig.13.2).
One of the goals of mineral processing is separation of minerals. Particles to be se-
lected have to vary in the value of the main parameter. In the case of coagulation it is
stability factor W. Stability ratio W is a complex parameter, which depends on electro-
static, molecular and structural interaction between particles (see next chapters).
Coagulation is a dynamic process, while Pz, Pa and Pstab are static parameters as
they are not a function of time. Time, as parameter which determines the results of
separation, can be incorporated by appropriate definition of total probability of coagu-
lation Pk, which can be defined as:
dn
−
Pk = dt , (13.2)
z z ni n j

where:
(dn/dt) – coagulation rate,
ni i nj – number of particles i and j in a unit volume,
zz – number of collisions of particles i and j in a unit time.
Introducing Eq. (13.2) into Eq. (13.1a) gives
dn
− = Rij = z z ni n j Pz Pa Pstab , (13.3)
dt
that is a general equation of coagulation (Kruyt, 1952; Chander i Hogg, 1987).

13.2. Adhesion of particles

The mechanism of adhesion of particles during coagulation has been explained by
the DLVO theory. Acronym DLVO was derived from the names of Russian (Derjagin,
and Landau, 1941), and Dutch (Verwey and Overbeek, 1948) researchers who investi-
gated that phenomenon. According to the DLVO theory, coagulation depends on dif-
ferent forces interacting during coagulation. Verwey and Overbeck considered those
interactions in terms of energy, while Derjagin and Landau used forces instead. Both
approaches are equally useful for the description of coagulation. The energy approach
is usually more practical, therefore it will be applied in this work. The use of forces is
more suitable for interpretation of coagulation when the measurement and calculations
of forces are performed. Special terms disjoining force and disjoining pressure were
introduced in 1936 by Derjagin (1989) for forces operating during coagulation. This
emphasizes the importance of force approach to coagulation.
 13. Coagulation 401

The DLVO theory considers aggregation of particles which initially are considera-
bly far away from each other. Coagulation can be schematically expressed by the
equation:

+ = 13.4

Free enthalpy for two particles which are considerably distant from each other
(H=∞), can be expressed as G∞. After coagulation, when particles stay close to one
another at a minimum distance equal the size of atoms in their crystalline lattice, their
free enthalpy is Gh. Thus, free enthalpy of coagulation (ΔGk) equals the difference of
free enthalpy after forming coagulum and the energy before coagulation.
ΔGk = Gh – G∞ . (13.5)
Since the interaction of approaching particles is influenced by various forces, in-
cluding molecular (m), electrostatic (el), structural (s) and other (Derjagin 1989), free
enthalpy of coagulation is given by:
ΔGk = ΔGk d + ΔGk el + ΔGk s + ΔGk other. (13.6)
According to Eq. (13.6), many factors influence coagulation. As it will be shown in
subsequent sections, the most important are Hamaker constant as a measure of mo-
lecular forces, potential zeta as a measure of electrostatic interactions, and contact an-
gle, i.e. hydrophobicity of interacting particles as a measure of structural forces.

13.2.1. Molecular interactions

Molecular interactions, also called dispersion interactions, are always present be-
tween material objects. Energy of these interactions can be quite considerable and suf-
ficient for maintaining sheets of certain material, for instance graphite, in the form of a
compact crystal. Simple calculations of Isrealachvili (1985) indicate that two plain-
parallel plates having the Hamaker constant equal to 1.10-19 J (as for instance talc and
mica) adhere to one another in the vacuum with a great force leading to pressure of
7.108 Pa. Dispersion forces are responsible for interaction of atoms and particles in the
gas phase. Due to the existence of dispersion forces gases can be liquefied. Detailed
description of gaseous state requires application of real gas equations, which constitute
an extension of ideal gas equation, with corrections for dispersion forces and the vol-
ume of molecules. The term “dispersion interactions” originates from the observation
that these forces are responsible for dispersion, i.e. for dependency of refraction index
on the wavelength and dielectric constant on frequency of electric field (Israelachvili
1985).
According to Derjagin (1989) the origin of dispersion force can be described as fol-
lows: “Everywhere and always, including in vacuum and at absolute zero there are
 402 Part III. Separation processes

spontaneous quantum fluctuations of the electromagnetic filed, even in the absence of

external sources. After local fluctuations have arising, they serve as a source of virtual
(non-energy-bearing) electromagnetic waves. These waves, when they reach an inter-
face and are partially refracted, form traveling waves in the neighboring phase and in
part form standing wave that decay exponentially with increasing distance from the
interface. As a result, the wave field, characterized by the tensor of the mean square
components of the electromagnetic field, is uniform and isotropic in the bulk phases at
sufficient distances from the interface. Close to the interface, however, this tensor is
nonuniform and anisotropic. If the interlayer between the two phases is made suffi-
ciently thin, the transition zones with nonuniform distribution of electromagnetic fluc-
tuations begin to overlap. The wave field is changed; in particular, the component of
the tensor Enn2 normal to the interface, is changed. This upsets the equilibrium and, in
the general case, gives rise to a disjoining pressure (or molecular attraction), tending to
change the thickness of the interlayer (or gap) between the outer phases; the disjoing
pressure may be balanced by external forces”
The molecular interaction of a spherical particle (phase) 1 with a particle (phase) 2,
through another phase (phase 3), when particles radius r is considerably larger than the
distance between particles H, is expressed by:
A132 R
ΔGk , d = ΔG132
d
= VA = − , (13.7)
12 H
where:
H – distance between particles
R – particle radius
A132 – the Hamaker constant (in joules).
To calculate the conditions of coagulation, as well as to describe other processes
like flotation or foam properties, it is necessary to know the Hamaker constant. Posi-
tive values of the Hamaker constant indicate repulsion while negative ones attraction.
The Hamaker constant determines the energy of dispersion interaction between parti-
cles and other objects like drops, bubbles, thin films, foams, or adsorptive layers. The
Hamaker constant is defined as:
A = π2 C ρ*1 ρ*2, (13.8)
where:
ρ*1 and ρ*2 – number of atoms per volume of objects 1 and 2, respectively
C – proportionality constant between the potential of atom–atom pair and the
distance between atoms, also called the London constant.
Since C is of order 10-77 J/m6 while ρ* is of order 3.1028 m-3, the typical value of
the Hamaker constant for condensed (solid and liquid) phases interacting through the
vacuum (A131= A101=A11) is around 10-19 J (Israelachivili, 1985). Vacuum, as a medium
for interaction of particles 1 and 2, is usually denoted as 3, or 0, or no symbol is used).
 13. Coagulation 403

For instance, the Hamaker constant for different substances interacting in vacuum,
ranges from 3.8.10-29 J for Teflon to about 5.10-20 J for heavy metals (Drzymala, 1994).
The Hamaker constant values for different materials are presented in Table 13.1.

Table 13.1. Hamaker constant A11 for selected materials

collected by Drzymala (1994) and other authors
A11 A11 A11
Material Material
(×1020 J) (×1020 J) (×1020 J)
n-pentane (C5H12) 3.8b mica 10.0b MoS2 (molibdenite) 13.3e, 9.1c
b c
Teflon ([C2F4]n) 3.8 MgO (periclase) 10.5 S (sulfur) 23c
b d
Acetone (CH3COCH3) 4.1 CaCO3 (calcite) 10.1 Fe2O3 (hematite) 23.2a
b c
Ethanol (C2H5OH) 4.2 AsS (realgar) 12.0 C (graphite) 23.8a
a c
Water (H2O) 4.38 FeS2 (pyrite) 12.0 SnO2 (cassiterite) 25.6a
b c
n-octane (C8H18) 4.5 CaO (lime) 12.5 Si (silicon) 25.6a
b c
n-dodecane C12H26 5.0 FeCr2O4 (chromite) 14.0 FeAsS (arsenopyrite) 27c
n-tetradecane (C14H30) 5.0b ZnS (sphalerite) 14.0c As2S3 (auripigment) 28.4a 15c
b f
Benzene (C6H6) 5.0 CdS (greenockite) 15.3 C (diamond) 28.4a
b a
n-heksadecane (C16H34) 5.1 Al2O3 (corundum) 15.5 Cu (copper) 28.4a
b a
Cyklohexane (C6H12) 5.2 AgI (iodirite) 15.8 Ge (germanium) 30.0a
a c
KCl silvine 6.2 Sb2S3 (metastibnite) 16.0 TiO2 (rutyl) 31.0a
a a
CnH2n +2 (paraffin) 6.3–7.3 SiO2 (quatz) 16.4 PbS (galena) 33c
b a
Polystyrene 6.5 BaSO4 (barite) 16.4 Ag (silver) 40.0a
CaF2 (fluorite) 7.2 TiO2 (anatase) 19.7a Hg (mercury) 43.4a
Bornite (Cu5FeS4) 7.4c Cu2S (chalcocite) 21.0c Au (gold) 45.5–50a
b a
Poli(vinyl chloride) 7.5 Fe (iron) 21.2 CuS (covelline) 2.8c (?)
c a
Pirrothite (FeS) 8.4 Pb (lead) 21.4 [Fe, Ni]9S8) pentlandite 3.3c (?)
c a
Talc (Mg3[(OH)2Si4O10]) 9.1 Sn (tin) 21.8 CuFeS2 (chalkopyrite) 3.3c (?)
a) Visser (1972), b) Israelachvili (1985), c) Lins i współ. (1995), d) Hunter (1987), e) Ebaadi (1981),
f) Krupp et al., (1972). Symbol ? denotes uncertain data

Determination of the Hamaker constant is not a simple task. There are two ap-
proaches to calculation of the Hamaker constant known as microscopic or macro-
scopic. Gregory (1969), for example, used the microscopic approach. The Hamaker
constant for two identical particles (designated as 1) interacting in a condensed me-
dium 3, i.e. A131, is given by the equation (Sonntag 1982).
2
27 hνν ⎛ ε1* − 1 ε 3* − 1 ⎞
A131 = ⎜ * ⎟
64 ε 3* ⎜ ε + 2 − ε* + 2 ⎟ , (13.9)
⎝ 1 3 ⎠
while for two substances interacting in vacuum (0 or no symbol) (Gregory, 1969) it is
described by the relation:

27 hν 1νν 2ν ⎛ ε1* − 1 ⎞ ⎛ ε 2* − 1 ⎞
A12 = A102 = ⎜ ⎟⎜ ⎟. (13.10)
64 (ν 1ν + ν 2ν ) ⎜⎝ ε1* + 2 ⎟⎠ ⎜⎝ ε 2* + 2 ⎟⎠
 404 Part III. Separation processes

For two identical substances interacting in vacuum according to (Gregory, 1969)

the equation assumes the form:
2
27 ⎛ ε* −1 ⎞
A101 = A11 = hνν ⎜⎜ 1* ⎟
⎟ (13.11)
64 ε
⎝ 1 + 2 ⎠
where:
ε1* – terminal (not static as is claimed by Sonntag, 1982) dielectric permeability for
phase 1 particle,
ε2* – terminal dielectric permeability for substance 2 particle,
ε3* – terminal dielectric permeability for condensed medium, that is phase 3,
νν – characteristic frequency.
The terminal values of dielectric constant ε1*, ε2*, or generally εn*, for different
substances can be calculated from the index of refraction n0 within the range of visible
light, applying the Maxwell equation εn=n02. It should be stressed that εn is not static
dielectric constant ε0 referring to zero frequency of electric field. For instance, static
dielectric constant for water is 81, while terminal dielectric constant amounts 1.77,
since n = 1.323 (εn*=n02=177). The n0 value can be obtained by interpolation of n
within a visible range to zero frequency of the electric field or similar dependences
(Fig. 13.3.).
Characteristic frequency νv of a substance can be also determined from the rela-
tionship between refraction index n and frequency within visible light range, accord-
ing to the equation:
n2 + 2 3π me M νν2 3π me M ν 2
= B ν 2
ν − B ν 2
= − 2 , (13.12)
n2 −1 e2 Na ρ s e Na ρ s
where:
me – electron mass
e – electron charge
Na – Avogadro number
s – effective number determining dispersion of electronic oscillators
M – molar mass (g /mol)
ρ – density, g /cm3.
Plotting the relation:
n2 + 2
y=
n2 − 1
as a function of ν2, within the range of visible light, results in a straight line of inclina-
tion B, which enables determination of s. Knowing parameter s and the value of inter-
section of straight line with axis y, vv can be determined. The same value of intersec-
 13. Coagulation 405

tion enables determination of n0, and then, from the Maxwell equation ε0*. Figure
13.3a shows determination of the values of s, vv, n0 and ε0* for polystyrene.

s = 13.7
3.00 ν ν = 2.62×10 15s-1
n0 = 1.563
(n2+2)/(n2-1)

2.95

2.90 polistyrene

2.85
10 20 30 40
ν 2 ×10-28
a

ν ν = 1.7×10 15s-1
n589nm = 2.0
(n2-1)

cassiterite

2
0 20 40 60 80 100
(n2-1)ω 2(1030)
b
Fig. 13.3. Determination of data for calculation of the Hamaker constant basing on refractive index, a) the
Gregory (1969) method of calculating s and εo, as well as no and νν, needed for calculation of the
Hamaker constant, b) the Hough-White (1980) method of νν determination

The macroscopic approach was proposed by Lifszyc (1954, 1955). For phase 1 in-
teracting with phase 2 through phase 3 as layer of thickness H, the Hamaker constant
is included in the following equation (Dzialoszynski et al., 1959)
 406 Part III. Separation processes

h
Πm ( H ) = − ×
2π 2 c*3

⎧⎡ ⎛ ⎞ ⎤
−1
⎫
⎪⎢ ( s1 + p )(s2 + p) exp⎜ 2 pξ H ε 3 ⎟ − 1⎥ + ⎪ (13.13)
⎪⎢⎣ ( s1 − p )(s2 − p ) ⎜ c ⎟ ⎪
∞∞
⎪ ⎝ * ⎠ ⎥
⎦ ⎪
× ∫∫ p 2ξ 3ε 33 / 2 ⎨
⎪⎡⎛ −1
⎬ dpdξ ,
⎞⎛ ⎞ ⎛ ⎞ ⎤ ⎪
⎪⎢⎜ s1 + p ε1 ⎟⎜ s2 + p ε 2 ⎟ exp⎜ 2 pξ H ε 3 ⎟ − 1⎥ ⎪
0 1

⎪⎩⎣⎢⎜⎝ ε 2 ⎟⎠⎜⎝ ε 3 ⎟⎠ ⎜⎝ c* ⎟
⎠ ⎦⎥ ⎪⎭
in which:

ε1 ε
s1 = − 1 + p 2 s2 = 2 − 1 + p 2 , (13.14)
ε3 ε3
Πm(H ) – force of interaction caused by dispersion forces related to unit area (minus
sign denotes attraction of plates), also called molecular component of dis-
joining pressure
h –the Planck constant
ε1, ε2, ε3 – functions dependent on imaginary frequency ω´´ = iξ (i is imaginary num-
ber)
ξ – frequency of electromagnetic field
c* – light speed in vacuum.
This equation is quite complicated, therefore many simplified versions were de-
rived. For example, for plane-parallel particles 1 and 2 in any arbitrary dispersion me-
dium 3, for short distances between particle surfaces, the Hamaker constant is ex-
pressed as:
∞
3h ⎡ ε1 − ε 3 ⎤ ⎡ ε 2 − ε 3 ⎤
A102 = ∫⎢ ⎥⎢ ⎥ dξ ,
16π 2 0 ⎣ ε1 + ε 3 ⎦ ⎣ ε 2 + ε 3 ⎦
(13.15)

where: ε1, ε2, ε3 area real functions of complex frequency ω = ω´ + iξ, while ω denotes
angular frequency (ω = 2πν), tj. ε (ω) = ε´ω) + iε´´(ω).
This equation still remains complicated, therefore many simplified equations for
the Hamaker constants were introduced, which in 1980 were collected by Sonntag
(1982). The most useful for mineral processing applications can be the equation intro-
duced by Israelachvili (1985) which is based on the Lifszyc equations and presented in
the analytical form. For non-metallic substances, when a particle interacts with a parti-
cle of a different material 2 through phase 3, the expression for the Hamaker constant
assumes the following form:
A132 = A132 ,ν = 0 + A132 ,ν >0 , (13.16)
 13. Coagulation 407

3 ⎛ ε1 − ε 3 ⎞ ⎛ ε 2 − ε 3 ⎞
A132 ,ν = 0 = kT ⎜⎜ ⎟⎟ ⎜⎜ ⎟⎟ , (13.17)
4 ⎝ ε1 + ε 3 ⎠ ⎝ ε 2 + ε 3 ⎠

A132 ,ν > 0 =
3h ν ν (n 2
1 )(
− n32 n22 − n32 ) . (13.18)
8 2 (
n12 + ) (
n32 1 / 2 n22 + ) {(n
n32 1 / 2 2
1 + )
n32 1 / 2 + (
n22 + ) }
n32 1 / 2

For two identical objects interacting through medium 3 the Israelachvili equation is
simplified to:

A131 = A131,ν = 0 + A131,ν > 0

3 ⎛ε −ε ⎞
= kT ⎜⎜ 1 3 ⎟⎟ +
3h ν ν
2
n12 − n32 2
,
( ) (13.19
4 ⎝ ε 1 + ε 3 ⎠ 16 2 n12 + n32 3 / 2 ( )
while for two identical object interacting in vacuum, the Hamaker constant takes the
form:
2
3 ⎛ ε −1 ⎞ 3hνν (n 2 − 1) 2
A = A11 = A11,ν = 0 + A11,ν > 0 = kT ⎜ ⎟ + , (13.20)
4 ⎝ ε + 1 ⎠ 16 2 (n 2 + 1)3 / 2

where ε1 = 1 and n3 = 1.
In the equations from (13.16) to (13.20):
ε – static relative dielectric constant (for water 81),
n – refractive index for light in visible range,
νν – characteristic frequency equal to the main absorption frequency in the ultraviolet
region. Aν =0 is not higher than 3/4 kT, that is 3·10–21 J at 300 K.
For metals the expression for the Hamaker constant is slightly different because:
ε (ν ) = 1 − ν 2p /ν 2 , (13.21)

ε (iν ) = 1 + ν 2p /ν 2 , (13.22)

thus
A = (3 / 16 2 ) hν p , (13.23)

where νp is the plasma frequency of free electron gas.

In order to calculate the Hamaker constants using the Israelachvili equation, it is
necessary to know the value of characteristic frequency, i.e. frequency of radiation at
which the absorption within UV range takes place, as well as index of refraction
within the visible range. According to Lins et al. (1995), vv can be calculated on the
basis of the relation between index of refraction and frequency and the Hough-White
equation (1980):
 408 Part III. Separation processes

1
(n 2 − 1) = (n 2 − 1) ω 2 + CUV , (13.24)
ωUV
2

where:
n – refractive index of light at appropriate frequency of visible rays
ω – angular frequency of radiation in the visible region for appropriate wave-
length λ, at which n was measured (rad s–1)
ω UV – angular frequency of absorption in the ultraviolet region.
ω UV is connected with characteristic frequencyνν by the relationship:
νν = ω UV/2π. (13.25)
To determine vv, the dependence of (n -1) and (n -1)ω should be plotted and ap-
2 2 2

proximated with a straight line. The slope of the line, as it results from equation
(13.24), has the value l/ωUV, which is recalculated to the value of characteristic fre-
quency vv using Eq. (13.25). Determination of vv for cassiterite is presented in Fig.
13.3b. Knowing vv and the refraction index, the Hamaker constants can be calculated
by means of the Israelachvili equation (Eqs (13.19) and (13.20)), substituting n value
determined at 589 nm. Parameter Av=0 is calculated on the basis of static relative di-
electric constants.
Lins et al. (1985) showed that Eq. (13.24) provides estimation of the Hamaker con-
stant for opaque and light absorbing minerals, for which a direct measurement of in-
dex of refraction is not possible. Then, the reflection data for a mineral in two different
media should be taken into account e.g. in air and oil. Reflection coefficient Rn is re-
lated to refraction index n in the following way:
(n − N ) 2 + K a2
Rn = , (13.26)
(n + N ) 2 + K a2
where:
Ka – absorption coefficient
N – reflection coefficient (for air N = 1).
Combination of two equations (13.26) for a mineral in two media, e.g. in air and
oil, results in:
( N 2 − 1)
2
n=
N (1 + Rn , oil ) (1 + Rn ,air )
. (13.27)
−
1 − Rn ,oil (1 − Rn ,air )

Calculated value n is real only when the value of Ka is not an imaginary number.
To check this, Ka should be calculated from the relation:
 13. Coagulation 409

Rn ( n + 1) 2 − ( n − 1) 2
K a2 = . (13.28)
(1 − Rn )

More details regarding determination of refraction index of the Hamaker constant

for opaque substances can be found in the work by Lins et. al. (1995). The Hamaker
constants can be also calculated basing on dispersion component of surface energy
using equation (12.7).
To calculate and describe coagulation and heterocoagulation Hamaker constants
A131 or A132, are used. They reflect interactions between three different phases. These
constants can be determined with Eqs (13.16) and (13.23). Their approximated values
can be determined using equations (Israelachvili, 1985):
A12 ≈ A11 A22 , (13.29)

A131 = A313 ≈ A11 + A33 − 2 A13 = ( A11 − A33 ) 2

(13.30)

A132 ≈ ( A11 − A33 )( A22 − A33 , ) (13.31)

A132 ≈ ± A131 A132 . (13.32)

Table 13.2. Hamaker constants A131 for two objects of the same material (1) interacting through another
material (3)
Interacting media Hamaker constant A131
phase 1 phase 3 phase 1 (×1020 J)
Air water air 3.70a
Octane water octane 0.41a
Water hydrocarbon water 0.34–0.54a
Polistyrene water polistyrene 0.95a
Quartz (fussed) water quartz (fussed) 0.83a
Teflon water Teflon 0.33a
Mica water mica 2.0a
Ag, Au, Cu water Ag, Au, Cu 30–40a
Si water Si 36b
Al2O3 water Al2O3 4.12b
MgO water MgO 1.6b
SiO2 (cryst.) water SiO2 (cryst.) 1.7b
CaCO3 water CaCO3 2.23b
CaF2 water CaF2 1.04b
Oxides water oxides 1.76–4.17b
TiO2 water TiO2 1.1c
a) Hough and White (1980), b) Ross and Morrison (1988), c)Yotsumoto and Yoon (1993a).

According to Israelachvili (1985), these formulas are not accurate for the systems
of high dielectric constants and featuring high values of Av=0, like water. In such a case
 410 Part III. Separation processes

it is better to use more accurate equations proposed by Gregory (1969), Pashley

(1977), Hough and White (1980).
Some Hamaker constants for two objects made of the same material and interacting
through a different medium (3) are presented in Table 13.2.

Table 13.3. Hamaker constants A132, for two objects of different materials (1 and 2) interacting through
phase 3
Interacting media Hamaker constant A132
phase 1 phase 3 phase 2 (×1020 J)
Water octane air 0.51a
Octane water air –0.24a
Quartz (fussed) water air –0.87a
Quartz (fussed) octane air –0.70a
CaF2 liquid He He vapor –0.59a
Polistyrene water gold 2.98b
Selenium water MgO 2.83b
a) Israelachvili (1985), b) Ross and Morrison (1988).

It can be seen in the table that the Hamaker constants can be either positive or
negative. Negative value indicates that dispersion forces repel particles, while positive
particle attraction. It means that two identical particles always attract each other, while
two different objects can either repel or attract one another.
Molecular interactions in coagulation depend not only on the Hamaker constant,
but also on the distance between objects and their shapes. Table 13.4 shows different
formulas used for calculation of free enthalpy of objects interaction (ΔGd). These for-
mulas are general and refer to interaction both in vacuum and between objects 1 and 2
through another substance 3, therefore, symbol x as subscript at ΔGd and the Hamaker
constant A was introduced, which can stands for phases 131 or 132. If phase 3 is vac-
uum, x can indicate interactions between identical objects (11) or two different objects
in vacuum (12).
The relations presented in the table sometimes require, especially in the aqueous
medium, some corrections, especially at greater separation. The corrections take into
account the effect of retardation of molecular interactions. Retardation of interaction
becomes considerable when time needed for fluctuation of electric field during propa-
gation from one particle to another is comparable with the fluctuation period. For co-
agulation process, according to Shenkel and Kitchener (1960), dispersion interaction
considering retardation will assume the form:
A131R
ΔG131
d
= VA = − fo , (13.33)
12 H
where:
 13. Coagulation 411

1
fo =
1 + 1.77 P , (13.34)

while P = 2πH/λo. λo is the length of electron wave of oscillation which is usually as-
sumed to be 100 nm. The equation is valid for P ≤ 0.5. When P > 0.5, the following
equation should be used:
2.45 2.17 0.59
fo = − + . (13.35)
5P 15P 2 35P 3
Table 13.4. Equations for free enthalpy of dispersion interactions of objects of different shape (Israe-
lachvili, 1985). x denotes interacting phases, for instance phase 1 with phase 2 through phase 3 (132).
When interactions occur through vacuum the symbol is 11 (identical objects) or 12 (different objects)

Interacting objects Formula Units

C
Two atoms ΔG d = − 6 (C is a constant) J
H
Ax ( R1 R2 )
Two spheres ΔG xd = − J
6 H ( R1 + R2 )
Ax
Two flat parallel slabs ΔG xd = − J/m2
12πH 2
AR
Sphere and slab ΔGxd = − x J
6H
Ax R1R2
Two perpendicular cylinders ΔGxd = − J
6H

There are many formulas for the coefficient of retardation f0 (Czarnecki, 1986).

13.2.2. Electrostatic interactions

Interactions between two electrically charged ions can be described using the Cou-
lomb law (Wróblewski and Zakrzewski, 1984) according to which the interaction en-
ergy is inversely proportional to the distance between ions. This dependence is not
fulfilled in the case of particles interacting in aqueous suspensions because of diffused
electric double layer (edl). When the charged particles approach each other in water,
there is at first penetration and then redistribution of charge and potential in the edl.
Since the edl penetration is complicated, either the penetration with a constant charge
or constant potential for simplicity is usually considered. It is believed that superposi-
tion of both electrical double layers most often occurs at constant surface potential and
is accompanied by a decrease in the surface charge. When ions passing from the sur-
face into the solution is inhibited, because the rate of approaching particles is higher
than the speed of establishing ionic equilibrium, the edl penetration can take place at
 412 Part III. Separation processes

constant charge to maintain constant potential, while the surface potential becomes
altered. According to Sonntag (1982) both ways lead to practically identical formulas
for the energy of particles interactions.
General formula for the electrostatic interaction energy ΔGel of spherical particles
of radiuses R1 and R2, having different values and edl potential signs Ψ1 and Ψ2, when
particle radius is of considerable value comparing to the thickness of diffused edl part
1/κ, was derived by Hogg et al. (1966). Its form in the SI system is as follows (Hunter,
1989).

πεε 0 R1R2 (ψ 12 + ψ 22 ) ⎧ 2ψ 1ψ 2 ⎡1 + exp( −κH ) ⎤

ΔGel = VR = ⎨ 2 ln ⎢ ⎥
( R1 + R2 ) ⎩ (ψ 1 + ψ 2 ) ⎣1 − exp( −κH ) ⎦
2
(13.36)
+ ln [1 − exp(−2κH )] },
where:
ΔGel – change of free enthalpy for the system caused by electrostatic interactions
(frequently symbol VR is used), J
ψ1 – electrostatic potential of particle, V
ψ2 – electrostatic potential of another particle, V. When particles are identical
ψ1 = ψ2 = ψs
R1 – radius of spherical particle, m
R2 – radius of the other spherical particle, m. When particles are identical,
R1 = R2 = R
ε – dielectric constant of medium (usually water), also called relative dielectric
permeability (dimensionless quantity, for water ε = 81, at 293,2 K)
ε0 – dielectric permeability of vacuum, 8.854187817·10–12 C2 N–1 m–2 (CRC, 1998)
1/κ – Debye radius also called the thickness of edl, m
H – distance between objects, m. For spherical particles it is the distance between
most separated spots of particles.
The expression for the thickness of the edl (Debye radius) 1/κ (m) has the form
(Wiese et al., 1976):
1/ 2
⎛ 2 z 2e 2 ne ⎞
κ = ⎜⎜ ⎟
⎟ = 3.29 ⋅ 10 9 z c s , m −1 , at 298 K,
⎝ εε 0 kT ⎠ (13.37)

where:
cs – concentration of electrolyte in the bulk solution, mol/dm3
z – integer number 1, 2, 3 denoting the valence of the counter ion (without + or –
sign), which has major contribution to compensating edl
ne – electrolyte concentration in the bulk solution as number of pairs of ions per unit
volume, m–3
 13. Coagulation 413

e – elemental electric charge, coulomb (C)

k – Boltzmann constant (1,38·10–23 joule/kelvin, J/K)
T – absolute temperature, kelvin (K).
Positive interaction energy ΔGel indicates repelling, while negative one - attraction.
Particle repelling in water, due to electrostatic edl forces, takes place when interacting
particles are of identical sings of surface potential. Since ions present in the edl are not
mathematical points, in the equations for electrostatic interaction of particles surface
potential is replaced by potential in the Stern layer of the edl (ΨS). The Stern potential,
in turn, is closed to zeta potential. Therefore, zeta potential of particle is used for cal-
culation and description of electrostatic interactions between objects in mineral proc-
essing and the ones which take place in nature, including coagulation. It is a correct
approximation for low values of surface potentials. It should be stressed, however, that
in literature there are many incorrect modifications of the formula for calculating VR.
They result from wrong adaptations of the original equations derived in the electrical
units system cgs for the requirements of the SI system. In the SI units a unit of poten-
tial is volt (V), of charge is coulomb (C), and the dielectric constant is dimensionless
since it is expressed as a relative value in relation to the dielectric permeability in vac-
uum and is equal to 8.85.10-12 C2N-1m or AsV-1m-1. Since 1 joule (J)=1 volt x coulomb
(C) and As (Ampere-second) = J/V and F=C/N, dielectric permeability in vacuum can
be also expressed as 8.85·10-12 AsV-1 or Fm-1. A certain indication as to the correctness
of the equation including dielectric permeability is the presence of vacuum permeabil-
ity symbol (most often εo) beside the symbol of the dielectric constant for a particle or
medium (most often ε).
Changing equations from the cgs electrical system to SI is achieved by replacing ε
with 4 πεε0. This problem is discussed in detail in the chapter on the SI units.
Equation 13.36 describes electrostatic interactions for heterocoagulation process,
yet it can be applied for the description of coagulation, since during coagulation inter-
action between identical particles takes place and then R1=R2=R and Ψ1=Ψ2=Ψ.
For identical spherical particles of low potential and high κR values the following
approximating equation can be used (Hutner, 1987):
ΔGel = VR = 2πRεε 0ψ S2 ln{1 + exp (−κH )} , (13.38)
where Ψ2 is the Stern potential, usually replaced with the zeta potential (ζ).
For low values of κR<5 the edl is very extended. For this case Verwey and Over-
beek (1948) introduced a special equation which is highly complicated. Therefore, it is
easier to use graphs which can be found in the work of Verwey and Overbeek (1948),
as well as in the monograph by Hunter (1987). A simplified expression (with up to
40% error) for κR<6 takes the form of:
ΔGel = VR = 2πRεε 0ψ S2 exp (−κH ) . (13.39)
 414 Part III. Separation processes

Coagulation processes sometimes take place between spherical and flat particles or
between a spherical particle and a flat surface. Then, the expression for electrostatic
interactions, according to Sjollem and Busscher (1990) takes the form of:
⎧ 2ψ ψ ⎡1 + exp(−κH ) ⎤
ΔGel = VR = πεε 0 R (ψ 12 + ψ 22 ) ⎨ 2 1 2 2 ln ⎢ ⎥
⎩ (ψ 1 + ψ 2 ) ⎣1 − exp(−κH ) ⎦
(13.40)
⎫
+ ln [1 − exp (−2κH )] ⎬,
⎭
which means that interactions, expressed in joules, are twofold stronger than in the
case of two spherical particles. When a flat and parallel surface interact with each
other, but featuring different surface potentials, the expression for the energy of elec-
trostatic interactions at constant potential VΨ can be obtained using numerical proce-
dures, while that for low potentials according to the expression (Hutner 1987):

ΔGel = VRψ =
εε 0κ
2
[(ψ 2
1 ]
+ ψ 22 )(1 − ctgh κH ) + 2ψ 1ψ 2 cosec hκH , (13.41)

where ΔGel for plates is in J/m2.

For identical surface potentials of both plates (Ψ1=Ψ2=Ψ) Eq. 13.41 is reduced to:
VRψ = 2εε 0κψ 2 (−κH ) J/m2. (13.42)

This equation is also true for interactions at constant charge when potentials are
low and the two edl slightly superimpose (the plates are far from each other) (exp(-κ
H)<<1) and then interactions can be described by an approximate equation (Sonntag,
1982):
2 −2
64ne kT ⎛ zeψ ⎞ ⎛ zeψ ⎞
ΔGel = VR = exp (−κH ) ⎜ exp − 1⎟ ⎜ exp + 1⎟ , (13.43)
κ ⎝ 2kT ⎠ ⎝ 2kT ⎠
where nc is the number of pairs of ions per 1m2 and thermodynamic potential of elec-
trostatic interactions ∆Gel is expressed in J/m2.
Table 13.5 presents approximated formulas for energy of electrostatic interactions
VR between the objects of different geometry, placed in medium of dielectric constant
constants ε (after Russel et al., 1989).
The equations for electrostatic interactions contain, as a rule, surface potential (Ψ,
Ψo). Since ions in the edl have certain volume, surface potential is approximated with
the Stern potential. This potential, in turn, is approximated with zeta potential, which
is easily measurable. The zeta potential can be measured as a function of pH and ionic
strength of the solution in which coagulation takes place. The surface, Stern and zeta
potentials are mutually related and their values can be calculated. The zeta potential
depends on ionic strength of the solution as well as on its pH. At constant ionic
 13. Coagulation 415

strength zeta potential changes with the pH until it reaches a constant value, which can
be constant or slightly dropping towards the pH axis. A change of the zeta potential
with pH for many systems can be empirically described using the equation:
ζ max
ζ =2 − ζ max . (13.44)
1 + e f *( pH − iep )
In Eq. (13.44) coefficient ƒ* regulates ζ variations from iep to ζmax. A diagram
showing typical shape of the zeta potential is presented in Fig. 13.4.

Table 13.5. Approximate formulas for energy of electrostatic interactions ΔGel = VR

between objects having different geometry in medium of a given dielectric constant ε
(Russel i et al., 1989)
Geometry limitation interaction energy VR
Two parallel slabs overlap 64kTneκ −1 tanh 2 (0.25ψ ) exp( −κH )
2
⎛ kT ⎞ 2
Two spheres constant potential 2πεε 0 ⎜ ⎟ Rψ ln (1 + exp[−κH ])
⎝ ze ⎠
2
⎛ kT ⎞ 2
Two spheres constant charge − 2πεε 0 ⎜ ⎟ Rσ 0 ln (1 − exp[ −κH ])
⎝ ze ⎠
2
⎛ kT ⎞ R2
Two spheres linear overlap 4πεε 0 ⎜ ⎟ ψ 2 exp (−κH )
⎝ ze ⎠ H + 2 R
2
⎛ kT ⎞
Two spheres overlap 32πεε 0 ⎜ ⎟ R tanh (0.25ψ ) exp ( −κH )
2

⎝ ze ⎠

ζ , mV
ζ max (+)

+
iep

0
f*

–
ζ max (-)

pH
Fig. 13.4. Typical relations between zeta potential and pH
with characteristic terms such as ζmax, iep and curvature coefficient f *
 416 Part III. Separation processes

Table 13.6. Data for delineation of change of zeta potential as a function of pH in water for various ma-
terials
Mineral pHiep ζmax f* Ionic strength Source
γ–Al2O3 8.9 39 1.16 0.01 M KNO3 Wiese and Healy, 1975
γ–Al2O3 8.9 76 1.27 0.0001 M KNO3 Wiese and Healy, 1975
TiO2 5.8 –42 1.46 0.01 M KNO3 Wiese and Healy, 1975
TiO2 5.8 –79 1.29 0.0001 M KNO3 Wiese and Healy, 1975
Fe2O3 6.0 55 1.85 * Pugh, 1974
Fe3O4 6.7 43 1.10 * Gray et al., 1994
SiO2 (Fisher) 3.7 –14 0.82 0.1 M NaCl Cerda and Non-Chhom, 1989
SiO2 (Fisher) 3.7 –89 0.86 0.001 M NaCl Cerda and Non-Chhom, 1989
Anthracite 8.0 –44 0.88 * Ney, 1973
Sulfur 2.2 –32 2.10 * Ney, 1973
Talc 2.0 –49 0.76 * Ney, 1973
Bacteria** 3.3 –43 1.00 0.001 M NaCl Sadowski, 1998
Ice (D2O) ~3.0 –77 1.45 0.0001 M NaCl Drzymala et al., 1999
Ice (D2O) ~3.3 –31 1.00 0.001 M NaCl Drzymala et al., 1999
Hexadecane 3.1 –70 1.00 0.001 M NaCl Stachurski and Michalek, 1996
Diamond ~3.0 –77 1.45 0.001 M NaCl Shergold and Hartley, 1982
Air 3.3 –43 1.00 0.001M NaCl Li and Somasundaran, 1992
* In solutions without salt, it can be assumed that the ionic strength is 10–4 kmol/m3, except when the
concentration of the reagent which regulates pH is greater. In such a case the ionic strength is determined
by the pH value. ** Nokardia sp.

Table 13.6 provides the iep, ζmax and ƒ* constants for different materials in water.
A relation between the zeta potential and pH, as well as between the surface potential
and pH can be derived on the basis of the equation describing electrical double layer
(edl). The dependence between the zeta potential as well as surface potential and pH
for the Grahame model of edl is, after Smith (1976), diagrammatically shown in Fig.
13.5.
The slope of the surface potential curve versus pH for non-metals is expressed by:

⎡ dψ 0 ⎤ 2.303kT kT 1 dσ 0
⎢ d ( pH ) ⎥ =− −
⎣ ⎦ σ 0 →0 e 2 N s e 2 Θc d ( pH ) . (13.45)

Numerical value of the first right hand term of Eq. (13.45) at 250C (298 K) is 59
mV per pH unit and is equal to the Galvani potential Φ drop between the bulk of the
solid and the solution and is also called the Nernst slope. Since surface potential Ψ0 is
lower than the Galvani potential by the so-called potential chi (χ), Ψ0= Φ-χ , and it
changes its value less than 59 mV per pH unit, while this difference is expressed by
the second segment of equation (13.45).
 13. Coagulation 417

solid particle rigid part diffusive part bulk solution

electrical charge σ o σβ σd 0
capacity K1 K2 Cd
potential ψo ψβ ψd
ζ 0

plan name S ISP OSP

(S - surface. Planes of rigid part of edl : ISP - inner, OSP - outer)

Fig. 13.5. The Grahame model of electrical double layer (edl)

The dependence of the change of zeta potential and pH is very complicated for the
Grahame model shown in Fig. 13.5. For a constant ionic strength it is expressed by the
relation:
⎡ dς ⎤ ⎡ dψ ⎤ ⎡ dψ 0 ⎤
⎢ d (pH) ⎥ = exp(−κΔ) ⎢ d ⎥ ⎢ ⎥ , (13.46)
⎣ ⎦ς →0 ⎣ dψ 0 ⎦ ς → 0 ⎣ d (pH) ⎦ ς → 0
where:
−1
⎡ dψ d ⎤ ⎡ ⎛ 1 ⎞⎤
⎢ ⎥ = ⎢1 + Cd0 ⎜⎜ ⎟⎟⎥ (13.47)
⎣ dψ 0 ⎦ ς → 0 ⎢⎣ ⎝ K1 + K 2 ⎠⎥⎦
⎡ dψ 0 ⎤
and relation ⎢ ⎥ is described by Eq. (13.45).
⎣ d ( pH ) ⎦ σ 0 → 0
Equation (13.46) is simplified. A precise procedure of its derivation and a list of
assumptions can be found in the paper by Smith (1976). In Eqs (13.45) and (13.46) the
symbols have the following meaning:
ψ 0 – surface potential
ψ d – potential in the plane of beginning of the diffusive part of edl
σ0 – surface charge
K1 – integral capacity of internal part of rigid edl
K2 – integral capacity of external part of rigid edl
Cd – differential capacity of diffusive part of edl close to iep
k – Boltzman constant
T – absolute temperature
e – elemental charge
Ns – total number of surface groups available for formation of electrical charge
 418 Part III. Separation processes

Θc – fraction of surface sites at pzc occupied by either positive or negative charge

(Θc = Θ+ = Θ–)
Δ – distance between the slipping plane where the zeta potential is measured and
the plane of ψd potential (ζ = ψ d exp (–κ Δ)
κ – Debye radius.
The above equations are useful for calculation of various potentials of the edl as
well as calculations connected with coagulation.

13.2.3. Structural interaction

Structural interactions are especially important in aqueous systems. They result
from the location of water molecules at the water-solid interface and depend mainly on
hydrophilicity and hydrophobicity of the surface. Investigations show that hydropho-
bic structural interactions are considerable at the contact angle, measured through the
aqueous phase, greater than 640. Hydrophilic structural forces exist for hydrophilic or
slightly hydrophobic materials when the angle is less than 15o. For particles of me-
dium hydrophobicity, i.e. for 150<θ0<640 structural interactions can be neglected.
Forces and energies of structural interactions decrease exponentially with the de-
creasing distance. Several ways of calculating energy of interactions caused by the
structural forces are available in literature. According to Derjagin and Churaev (1989)
the free enthalpy ΔGs of structural interactions for flat particles approaching one an-
other parallel can be expressed by the relation:
⎛ H⎞ ⎛ H⎞
ΔGS = Kl exp⎜ − ⎟ = ES0 exp⎜ − ⎟ , (13.48a)
⎝ l ⎠ ⎝ l ⎠
and for spherical particles (Lu et al., 1991):
⎛ H⎞
ΔGS = K *lR exp⎜ − ⎟ . (13.48b)
⎝ l ⎠
Table 13.7. Parameters of interactions for identical plane-parallel particles in water
(after Skvarla and Kmet, 1991)
0
ES l Hydrophobicity θ
Substance
(mJ/m ) 2 (nm) (degree)
Quartz 1.2 0.85 0
Mica 39.8 0.17 0
Montmorillonite 144 2.2 0
CTAB –22.0 1.0 65
DDOA –58.0 1.2 94
DMDCHS –0.4 13.5 100
CTAB –cetylotrimethylammonium ion, DDOA – dioctadecyldimethylamonium ion, DMDCHS – di-
methylodichlorosilane
 13. Coagulation 419

Typical values Es0 and l after Skvarla (1991) are presented in Table 13.7. In equa-
tions (13.48) and (13.48b) K and K* are interaction constants, while l is a correlating
parameter as well as the parameter determining the layer thickness of oriented water
molecules on the surface. Interaction constant K is negative for hydrophobic structural
forces, which means that hydrophobic interactions cause particles attraction, while for
hydrophilic interactions K is positive, which indicates particles repulsion.
Another approach for calculation of structural interactions between particles and
other objects was proposed by van Oss et al. (1990)
⎛H −H ⎞
ΔGHAB = πRlΔGHAB0 exp ⎜ 0 ⎟, (13.49)
⎝ l ⎠
where:
ΔGHAB – energy of structural interaction as a function of the distance between two
spherical objects
ΔGHAB0 – energy of structural interaction for two plane-parallel surfaces when they
attain minimum separation H0. ΔGHAB0 is determined by means of contact an-
gle
l – distance of existence of structured water adjacent to surface
R – particle radius.
Van Oss et. al. (1990) assumed that for water l = 1nm. Theoretically, for water
molecules without hydrogen bonds l amounts about 0.2 nm, but since 10% of water
molecules are bound through hydrogen bonds, it can be assumed that l equals 1 nm.
The interaction energy, for the particles in contact can be calculated on the basis of the
van Oss-Good-Chaudhury (1988) theory according to which:
ΔGHAB0 = −2γ 12AB . (13.50)

Next, γ 12AB can be calculated from the relation:

( )
γ 12AB = 2 γ 1+γ 1− + γ 2+γ 2− − γ 1+γ 2− − γ 2+γ 1− , (13.51)

in which γ 1− , γ 1+ , γ 2− and γ 2+ , or generally γ − and γ + , are non-additive parameters

of participation of non-dispersion interactions of the total interaction energy

γ = γ LW + γ AB = γ LW + 2 γ +γ − . (13.52)

The values of these parameters can be determined from contact angles measured in
standard liquids. The values of γ- and γ+ for standard liquids and some minerals are
shown in Table 13.8.
 420 Part III. Separation processes

− +
Table 13.8. Values of γ and γ for probing liquids and selected minerals. The surface energy of the
material is equal to the sum of dispersion (LW, sometimes denote as d) and nonpolar donor-acceptor (AB)
components, that is γ = γ + γ AB (mJ/m)
LW

γ LW AB + _
Substance γ γ γ γ Source
Water 72.8 21.8 51 25.5 25.5 van Oss et al., 1990
Decane 23.9 23.9 0 0 0 van Oss et al., 1990
Diiodomethan 50.8 50.8 0 0 0 van Oss et al., 1990
Glycerol 64.0 34.0 30 3.92 57.4 van Oss et al., 1990
Formamide 58.0 39.0 19 2.28 39.6 van Oss et al., 1990
Hectorite ~40 39.9 ~0 ~0 23.7 van Oss et al., 1990
Sphalerite 60.4 52 8.4 0.2 88.3 Duran et al., 1995
Silica gel 53.3 41.2 12.1 0.7 51.3 Hołysz, 1998
Barite ~55 48.1 ~6.7 02±0.1 56.1 Hołysz and Chibowski, 1997

Another approach to the description of hydrophobic structural interactions between

spherical particles was proposed by Mao and Yoon (1997). According to them, the
energy of hydrophobic interactions can be described by a function similar to the rela-
tions for the dispersion interactions, where hydrophobic interaction constant is intro-
duced instead of the Hamaker constant:
R K131
ΔGh= − , (13.53)
12 H
where K123 is hydrophobic interactions constant.
On the basis of the data by Yoon and Lutterell (1998), the dependence between hy-
drophobic interaction constant and contact angle can be expressed using the following
empirical relation:
–log K131 = 3.2 cos θa + 18.2, (13.54)
where θa is the advancing contact angle for water measured through the aqueous
phase.
In the case of interactions between different spherical objects such as air bubble
and particles, the equation for hydrophobic interactions is of the form:
R1 R2 K132
ΔGS , h = − , (13.55)
6( R1 + R2 ) H 0

where K132 is a constant and K132 = K131K132 .

Other equations for calculating structural interactions can be found in literature.
Yotsumoto and Yoon (1993a, 1993b) applied double exponential dependence for the
expression of the energy of hydrophobic structural interactions for the spherical parti-
cles:
 13. Coagulation 421

R
ΔGS ,w = VS = {Ca Da exp(− H/Da ) + Cb Db exp(− H/Db )} , (13.56)
2
where:
Ca and Cb – constants at the exponential term
Da and Db – distance of disappearance of hydrophilic structural interactions.
The equations presented above indicate that the delineation of structural interac-
tions has been neither theoretically nor experimentally established. Most of the obser-
vation and equation for the structural forces should be reexamine because they were
established before a significant role of microbubbles, commonly present at interfaces,
was realized.

13.2.4. Other interactions

Other interactions between objects in a medium can be caused by impurities, espe-
cially by the presence of air. Presently there has been a broad discussion in the scien-
tific literature on effects caused by structural forces and air present in the form of mi-
crobubbles.

13.2.5. Stability factor W

Molecular VA, electrostatic VR , structural VS (hydrophobic Vs,h and hydrophilic Vs,w)
and other (due to air) interactions are dependent on the distance between approaching
particles and are usually plotted as shown in Fig. 13.6. The lines representing energies
can be of different shape because they depend on many parameters. Each kind of in-
teraction can be of either positive or negative sign.

+ DLVO

VR
Interaction energy,V

VS, h Vt = VR + VA + VS

distance, H

VS, w VA

Fig. 13.6. Illustration of the DLVO theory and a set of possible shapes of lines representing energies of
interactions between particles during coagulation
 422 Part III. Separation processes

When particles interact, their interactions superimpose and since they can be ex-
pressed as energies, their values sum up:
ΔGk = ΔGd + ΔGel + ΔGs + ΔGother, (13.57)
which is often written as:
Vt = VA + VR + VS + Vother. (13.58)
Fig. 13.7 shows typical interaction curve Vt illustrating the results of calculations
based on the DLVO theory.

+
Vt
ineraction energy, Vt

Vmax
distance, H
VII

Vmax – energy barier

– VI V I – primary minimum
V II– secondary minimum

Ho H∞

Fig. 13.7. Possible shape of the curve representing total interaction energy of particles Vt during coagula-
tion resulting from the DLVO theory

The DLVO coagulation curves have different shapes. There are some characteristic
points for interactions. One is the zero energy of interaction, which occurs when parti-
cles are at long distances from one another H∞. Another is the so-called first minimum
which is located at the beginning of the diagram at the distance Ho which equals the
distance between atoms in a solid. Any further approach of particles or atoms is not
possible as they are exposed to immense repelling forces, equal to those which cause
incompressibility of solids.
Equation 13.57 includes the interactions resulting from compressibility through the
ΔGothers term, which is usually not taken into consideration in the DLVO theory.
Therefore, the first minimum in the diagrams of DLVO curves in not present. The
course of the DLVO curves between the mentioned terminal points may assume dif-
ferent shapes. When particles have their surfaces electrically charged with opposite
signs, they are hydrophobic, and VA is negative, then all components of the interaction
 13. Coagulation 423

energy are negative and a barrierless coagulation takes place. In such a case the shape
of total interaction curve Vt between points H∞,and H0 is simple. It does not contain
any maximum or minimum since negative interaction energy indicates attraction of
particles. In other cases the curves possess extremes, such as the first and the second
minimum. Between the first and second minimum there exist energy barrier, Vmax. It is
a factor which causes that coagulation becomes slow or does not take place at all, as
the number of particles possessing sufficient energy to overcome that barrier is the
lower the higher barrier is. The value of barrier Vmax can be expressed in joules. Cus-
tomarily, the energy barrier is expressed in kT units which represent ratio of the bar-
rier energy in joules to kT unit at the same temperature. At 300 K 1 kT =4.14.10-21 J.
The area of energy barrier determines the ability of coagulation, which is character-
ized by the stability factor W:
∞
1 Vt
W = 2R ∫r
2R
2
exp
kT
dr , (13.59)

where:
R – particle radius
r – distance between centers of particles determining the radius of the region in
which cross section a particle can collide with another particle (r ≅ 2R).
Parameter W can assume the values from 1 for rapid coagulation, when each colli-
sion leads to adhesion, to very high numbers, e.g. 4.105, which corresponds to energy
barrier Vmax equal to 15kT. In some cases, for small values of the stability factor W, its
value can be slightly lower than 1 and then it should be assumed as 1.
There exist very useful approximated analytical dependence between stability fac-
tor W and energy barrier of coagulation (Sonntag, 1982).
1 ⎛V ⎞
W≈ exp ⎜ max ⎟ , (13.60)
2 Rκ ⎝ kT ⎠
in which κ means the Debye parameter.
This equation is not fulfilled at Vmax = 0 but basing on definition the value of W is,
at this point, equal to 1.
Factor W is the measure of ability of coagulation and it is defined in the following
way:
number of collisons between particles
W= (13.61)
number of collisions leading to coagulation .
Factor W determines the probability of particle adhesion as a result of collisions
since Pa = 1/W. This is a static definition of stability factor W. Its kinetic and kinetic-
space definitions based on the collisions ration is:
 424 Part III. Separation processes

vf jf
W= = , (13.62)
vs js
where:
v – rate of coagulation
j – flux (for instance number of collisions per unit time and surface area)
s and f – symbols representing fast and slow coagulation.
It can be concluded from the definition of the stability factor W that coagulation
can be viewed and described on different levels: statically, kinetically and as time–
space relations.
The investigation by Prieve and Ruckenstein (1980) proved that W can be also ex-
pressed in an empirical form:
W = exp [0.92{(Vmax/kT) – 1}] (for (Vmax/kT) ≥ 3). (13.63)

Table 13.9. Half-life t1/2 for a hypothetical emulsion containing 1 µm droplets having energy barrier Vmax
located at one radius from the surface of drop (after Friberg, 1991)
Energy barrier Stability ratio
Vmax /kT t1/2 W = exp[0.92{(Vmax/kT) – 1}]
(in kT units) (for (Vmax/kT) ≥ 3)*
0 0.8 s ~2
10.0 2.0 h 3.94·103
15.0 1.3 d 3.92·105
17.5 154 d 3.91·106
20.0 5.1 y 3.90·107
50.0 5.5·1013 y 3.78·1019
* Empirical equation of Prieve and Ruckenstein (1980). s - second, h - hour, d - day, y - year

Table 13.9 provides typical W and Vmax values for a hypothetic colloid system con-
sisting of particles 1µ in size. In this table the half-life period of a suspension, calcu-
lated from Eq. (13.63) is shown. Detailed calculations of a relation between Vmax and
W can be found in the work of Verwey and Overbeek (1948, p. 169).

13.3. Stability of coagulum

The issue of stability of coagulum is a part of coagulation delineation because it de-
termines the probability of stable coagulum formation (Pstab). This issue is considera-
bly less known than the adhesion of particles, described in detail in the previous chap-
ter. Generally, coagulation can be either reversible or irreversible. Irreversible coagu-
lation takes place when the aggregates cannot be destroyed, that is peptized. Irreversi-
ble coagulation usually takes places when particles coagulate due to salt addition, and
after diluting the suspension, the coagulum does not disintegrate. The degree of coagu-
lation reversibility depends on many factors, mainly at which minimum, the first or
 13. Coagulation 425

the second one, coagulation has taken place, as well as what forces effect coagulation.
According to Derjagin (1989) coagulation is reversible when it takes place at the sec-
ond minimum. This minimum is usually shallow and the energy needed to release par-
ticles from energy well is not high. However, when coagulation takes place at the first
minimum, the energy barrier of peptization V*max is considerable, because the sum of
coagulation energy barrier Vmax and the depth of energy well of the first minimum is VI
is high (Fig. 13.8). According to Derjaguin (1989) the proof of coagulation irreversi-
bility at the first minimum is the fact that initial coagulation phases can be reversed by
dilution, while aged coagulation system cannot be peptized, since a considerable part
of particles has already been in the state of the first minimum coagulation. If coagula-
tion is irreversible and coagulating particles are small, it can be assumed that Pstab
equals 1 or approximately 1. If the particles are of a large size and coagulation takes
place at the second minimum, this probability can be much lower. The issues con-
nected with Vmax are discussed in the work by Derjagin (1989).

+
V*max – energy barrier of peptization
interaction energy, Vt

Vmax
distance, H
VII

VI Vmax – energy barrier

– VI – primary minimum
VII – secondary minimum

Fig. 13.8. The DLVO curve. The energy barrier of peptization is shown as V*max

13.4. The probability of particle collision in coagulation

process
The probability of particle collision during coagulation depends on particle size
and hydrodynamics in the suspension. The probability of coagulation of a large parti-
cle x1 with a smaller one x2 in a liquid medium is determined by:
Pz = π xc2 /π x12 = (xc/x1)2. (13.64)
The equation results from the assumption that not all particles x2 which meet parti-
cle x1 on their way collide, because such a collision is guaranteed only for the particles
 426 Part III. Separation processes

being within a limited collision radius xc. The right-hand term of expression (13.64) is
to the second power, because the probability of collision depends on the ratio between
surface column of liquid limiting collision and the particle surface x1.
The probability of collision, as any other probabilities, can assume the values from
zero to one. In the case of rapid coagulation Pc equals 1 because each collision leads to
coagulation meaning that other probabilities, including adhesion and stability, also
equal 1. In other cases the probability of collision is lower than 1 and depends on the
mechanism of collisions. There are only few works thoroughly describing coagulation
including probability of collision. Such an equation should have a general form de-
rived from the probability of coagulation (Eq. 13.1a):
dn
− = z z ni n j Pz Pa Pstab , (13.65)
dt
where:
dn/dt – coagulation rate
Pz, Pa i Pstab – probability of collision, adhesion, and stability, respectively
zz – number of collisions of particle i and particle j, which concentrations
are ni and nj, per time unit.
Chander and Hogg (1987) showed that the formula for coagulation rate is:

dn ⎡ ⎛ x j ⎞⎤
Rij = − = k m n i n j x im ⎢ Pm ⎜ ⎟ ⎥ Pa Pstab
dt ⎢⎣ ⎜⎝ x i ⎟⎠ ⎥⎦ . (13.66)

Therefore, the product of probability of collision and number of collisions is:

⎛ xj ⎞
z z Pz = k m xim Pm ⎜ ⎟ , (13.67)
⎝ xi ⎠
where xi and xj denote the size of primary particle and coagulum, respectively.
The values of particular parameters for collisions caused by Brownian movements
and by stirring, are presented, after Chander and Hoggs (1987) in Table 13.10.

Table 13.10. Parameters useful for calculation of number of collisions during coagulation governed by
formula (13.66) (after Chander and Hogg, 1987)
Collision mechanism m km Pm(xj/xi)
2kT
Brown motion 0 (2 + xi/xj + xj/xi)
3η
_
Mixing 3 G s /6 (1 + xj/xi)3
_
k – Boltzmann constant, T – absolute temperature, η – liquid viscosity, G s – average shear rate.
 13. Coagulation 427

13.5. Kinetics and hydrodynamics of coagulation

In mineral processing coagulation is performed in aqueous media containing fine
particles forming suspension. The behavior of the suspension depends on the proper-
ties of particles and aqueous solution. The particles forming the suspension are sub-
jected to gravity forces and they settle according to the equations presented in the
chapter dealing with classification of particles in water. If particles are very small,
they are only weekly effected by the gravity force an undergo chaotic brownian mo-
tion. Such suspensions are known as colloidal suspension, colloids or sols. The
brownian motion result from the diffusion forces. They keep similar concentration of
particles in the whole suspension. The size of colloidal particles is usually about one
micrometer or less. When particles are of tens of micrometers, they form suspension in
which both the gravity and diffusion forces operate. Diffusion of particles resulting
from the brownian motion is determined by the Fick law (Laskowski, 1969):
dc kT dc
jd = − D =− , (13.68)
dh 6πηR dh
where:
D – diffusion coefficient
k – Boltzmann constant
T – absolute temperature
dc/dh – change of concentration with the position (height) of particle
η – medium viscosity
c – concentration of dispersed particles in mole/m3 or kg /m3,
R – particle radius
π = 3.14.
The sedimentation stream jg resulting from gravity forces (Laskowski, 1969) is
given by:
mgc
j g = uc = , (13.69)
6πηR
where:
u – settling velocity of particles in a liquid medium which can be calculated using,
for instance, the Stokes equation
g – gravity constant
m – real mass of particle in a medium equal to 4/3πR3 (ρp – ρw)
ρp – particle density
ρw – medium density.
Remaining symbols are the same as in Eq. 13.68).
When the particles are subjected to the forces of similar magnitude, then jd = jg,
and it leads to:
 428 Part III. Separation processes

dc mg
=− dh . (13.70)
c kT
After integrating Eq. (13.70) the Perrin equation is obtained
⎛ mgh ⎞
c (h) = c exp ⎜ − ⎟. (13.71)
⎝ kT ⎠
It results from the Perrin equation, that the particle concentration c(h) at different
heights in non-agitated suspensions depends on particle mass. It can also be concluded
from this equation that particle concentration at different heights will be similar if the
particle mass is not large.
Forming fine colloidal particles consumes considerable energy needed for creation
of new surface. Such systems feature high total energy and the systems with excessive
amount of energy are not stable and undergo various alterations, aiming at energy re-
duction. The main process taking place in suspensions to lower total energy of the sys-
tem involves particle binding, i.e. coagulation. Another process is recrystallization
leading to increasing size of thick particles and dissolving fine particles.
The fine particles are of higher total energy than the large ones. It is described by
the Thomson (Lord Kelvin) equation (Adamson, 1963):
cR 2γ Vz
RT ln = , (13.72)
co R
where:
cR – solubility of fine particles having radius R
co – solubility of coarse particles
γ – surface energy of material
Vz – molar volume of particle
k – Boltzmann constant
T – temperature, K.
This equation is used for determination of surface energy of the materials. During
recrystallization the system diminishes its energy, since fine particles of high surface
energy vanish forming larger particles of lower surface energy. This process is usually
very slow and, therefore it is not taken into account when discussing stability and co-
agulation of the suspensions. Another, the most important process of diminishing en-
ergy of suspensions, is coagulation. This process takes time. The description of co-
agulation kinetics is rather complicated since clusters of two particles appear first, and
then, after collision with other single particles or clusters finally form coagulum. Ac-
cording to the Smoluchowski theory of coagulation (Sontag, 1982), formula (13.73)
can only describe the loss in the number of original particles in the initial period of
coagulation.
 13. Coagulation 429

dn1
− = 8πD1rn12 = k1n12 , (13.73)
dt
where:
dn1/dt – rate of disappearance of primary particles
D1 – diffusion coefficient of primary particle
r – distance between centers of two particles being in contact (r = 2R)
k1 – rate constant
n1 – concentration of particles in suspension (number of particles in cm3).
A considerable element of the Smoluchowski theory of coagulation is the so-called
half-life period t1/2 of suspension also known as coagulation time, within which the
number of all particles in 1 cm3 decreases to its half, which can be expressed by the
relation:
t1/2 = (4π D1r n1)–1. (13.74)
Inserting appropriate quantities to Eq. (13.74), the equation for the time of half-
coagulation during a rapid coagulation is obtained:

3η 2·1011
t1 / 2 = = (seconds)
4kTn1 n1 . (13.75)

These equations have been derived following the assumption that coagulation is
caused by diffusion and each collision between particles leads to coagulation. They
describe rapid non-barrier coagulation in which any other forces between particles
which could prevent coagulation do not exist. More details regarding the kinetics of
rapid coagulation according to Smoluchowski can be found in the works and books on
colloid chemistry (for instance Sonntag, 1982).
When coagulation is retarded by energy barrier Vmax, usually caused by the pres-
ence of electrical charge on the surface, the disappearance of primary particles is de-
termined by the equation (Sonntag, 1982):
dn1 8πD1n12
− = , (13.76)
dt W
where W is stability ratio (Hunter, 1987)
∞
⎛ V ⎞ db
2
∫
W = 2 exp⎜ t ⎟ 2 ,
⎝ kT ⎠ b
(13.77a)

and b = r/R and r = 2R.

Another form of this expression is presented by Eq. (13.59). It was proved that it
can be made dependent on Vmax and assume an approximate form, which was pre-
sented in Eq. (13.60).
 430 Part III. Separation processes

1 ⎛V ⎞
W ≈ exp ⎜⎜ max ⎟⎟ , (13.77b)
2 Rκ ⎝ kT ⎠
where:
κ – Debye parameter
Vmax – energy barrier.
It should be noted that for Vmax=0, W=1, while for Vmax>0, W>1. Coagulation
strongly depends on the concentration of salts added to the suspension, for instance for
its destabilization. Reerink and Overbeek (Hunter, 1987) showed that there exist both
experimental and theoretical relation between stability factor W and salt concentration
cs
log W = − ka log cs + kb , (13.78)
where ka and kb are constant.
It results from Eq. (13.78) that stability of suspensions depends on salt concentra-
tion to a certain power ka, method of coagulation, as well salt type.
Rapid and slow coagulation, in which the driving force is diffusion, is called
perikinetic coagulation. Coagulation process can be accelerated through agitation of
the system. The process of coagulation supported by agitation is called true coagula-
tion or ortocoagulation. Agitation of the suspension causes the flow of particles and
their collisions, which can become a dominating driving force of the process. The flow
of particles resulting from agitation is characterized by mean gradient of velocity G ,
and its value depends on the intensity of agitation and the medium in which it takes
place. Expressions for mean velocity gradient for two systems, with and without agita-
tion, are presented in Table 13.11.

Table 13.11. Expressions for average velocity gradient for different reaction vessels (after Weber, 1972)
Vessel type Formula
1/ 2
⎛ P ⎞
reacting G = ⎜⎜ ⎟⎟
⎝ Vη ⎠
1/ 2 1/ 2
⎛ Qρ1 gh f ⎞ ⎛ gh f ⎞
Flowing type with baffles G = ⎜⎜ ⎟
⎟ = ⎜⎜ ⎟
⎟
⎝ Vη ⎠ ⎝ vt ⎠
1/ 2
⎛ C D Aµ ρ1 vr3 ⎞
Stationary with paddle stirrer G =⎜ ⎟
⎜ 2Vη ⎟
⎝ ⎠
G – mean velocity gradient, s–1, g – gravity, P – power used for stirring the suspension, hf – head loss
in the tank, t – mean residence time of liquid in the tank, ν – kinematic viscosity of suspension,
V – volume of liquid in the reactor,η – viscosity of suspension, Q – flow intensity, ρ 1 – liquid density,
CD – resistance coefficient dependent on impeller shape and flow mode, Ał – cross section of the impeller
blade in the plane perpendicular to the direction of movement, vr – relative rate of blade in relation to
liquid (from 0.5 to 0.75 of blade speed).
 13. Coagulation 431

According to Chander and Hogg (1987), half-life of ortokinetic coagulation can be

expressed as:
t1/2 = ~1,5/Gs ϕ, (13.79)
where:
Gs – mean shear rate, 1/s
ϕ – volume share of solid particles in the suspension.
Kinetics of coagulation can be also derived from the equations describing particle
flow in a unit of time. Relations used for this purpose are presented in Table 13.12.

Table 13.12. Expressions for flux (j) for different system

Flux Formula Source
Diffusion due to kT dc
jd = 1
brownian motion 6 πη R dh
Gravitational sedi- mg
jg = c 1
mentation 6 πη R
Fast coagulation jf = 8πDRn1 2
∞
8πDn1
∫
dr
Slow coagulation js = , W = 2 R exp (Vt /kT ) 2 2
W r
2R
⎧ µ* ⎫
⎧4 ⎫⎪ 3 ⎪
∫
Coagulation with stir- j m = jα = ⎨ (1 + q ) 3 n2 G s R13 ⎬ ⎨ x2* z * ( x *2 ) dx2* ⎬ 3
ring ⎩3 ⎭ ⎪ (1 + q ) 3 ⎪⎭
⎩ 0

Rotating disc
([ ] )−1 ddH ⎧⎨F1 ⎡⎢ ddHn + 12 (H + 1) 2 PeF2 n + RFkTex n ⎤⎥ ⎫⎬
Pe = H + 1 F3 n
4
⎩ ⎣ ⎦⎭
Pe = 2f0R ωd D∞ νk *)
3 3/2 –1 –1/2

*) Peclet number (Pe) is the ratio of convection flow to diffusion flow

Sorce: 1 – Laskowski (1969), 2 – Hunter (1987); 3 – van de Ven (1982); 4 – Dabros and Czarnecki
(1980), Dąbroś et al., (1977), Adamczyk and Dabros (1978).
c – concentration of dispersed particles, for instance in g/cm3, R – particle radius, m – real mass of parti-
cle in a medium, η – viscosity of particles, g – gravity, k – Boltzmann constant, T – absolute tempera-
ture, f0 – universal constant for rotation disc, ωd – angular rate of rotation of disc, νk – kinematic viscosity,
Gs – mean shear rate, π = 3.14, dc/dh – change of concentration with change of location h, D – diffusion
coefficient, D∞ – diffusion coefficient for particle infinitely distant from the disc surface, n = n x /n∞ –
dimensionless concentration of particles, n∞ – concentration of particles infinitely far from the disc sur-
face, n2 – concentration of particles (number of particles in unit volume), H – dimensionless distance
equal to H/R (H is the distance between disc surface and the particle), Fex – outer force, its component
parallel to axis H , j – number of particles of radius R1 ketch per unit time by particle with radius R1, α –
effectivness of ketch by orthokinetic coagulation, ł* = ł/R1 (ł maximum value of x2, to which the cross
section of ketch operates, z* = z/R1, where z is a function providing upper limit of cross section of the
ketch, q= ł* – 1, r – radius of sphere around the particle where the particle can collide with other particles
(r ≅ 2R), F1, F2, F3 – universal functions resulting from particle-surface interaction having a large curva-
ture given in Dabros and Czarnecki (1980), Dabros et al., (1977), Adamczyk and Dabros (1978).
 432 Part III. Separation processes

Table 13.12. also shows basic equations needed for the description of heterocoagu-
lation of particles on flat rotating surfaces for the technique called the rotating disc
method. This approach is often used for the research describing ability of the system
to coagulate, heterocoagulate, as well as slime coating and carrier methods of enrich-
ment.
Equations for kinetics of coagulation considering hydrodynamics are very compli-
cated but only kinetic equations combined with hydrodynamics, allow to obtain full
description of coagulation. According to Chander and Hogg (1987) the rate of coagu-
lation at which particular objects (original particles of coagulum) of xi and xj, bind
with one another to form a large coagulum, are determined by the equation:
dn1
− = Rij = kij ni n j Eij
dt , (13.80)
where:
ni and nj – concentration numbers for particles (or coagulas) xi and xj
kij – collision frequency coefficient
Eij – product of probabilities of adhesion and stability (resistance) of coagulum
to disrupture.
The collision frequency coefficient depends on the type of the driving force of the
collision and the size of particle. It can be presented as:
⎛x ⎞
kij = km xim Pm ⎜⎜ j ⎟⎟ , (13.81)
⎝ xi ⎠
where:
km – a constant for a given mechanism
Pm(xj/xi) – a function of particle size ratio (power m is an integer number dependent
on the mechanism of collision).
Coagulation is usually carried out in two ways: with and without agitation. The lat-
ter one involves particle collision resulting from the Brownian motion, while coagula-
tion with agitation is due to shearing forces. The values of factors m and Pm, as well as
km, in Eq. (13.81) have already been presented in Table 13.10. Both models of coagu-
lation do not have simple solution (Sonntag, 1982), but it can be proved that the mean
coagulum size in relation to the initial size xo undergoes further alteration according to
the relation:
1/ 3
⎛ t ⎞
x = x0 ⎜⎜1 + ⎟
⎟
⎝ t1 / 2 ⎠ , (13.82)
when collisions result from the Brownian motion, and
 13. Coagulation 433

x = xo exp(0.23 t/t1 / 2 )
(13.83)
for systems with agitation. It results from these equations that agitation considerably
accelerates coagulation. This can be especially observed for suspensions containing
hydrophobic particles which are difficult to coagulate. This process is called the shear-
coagulation (Warren, 1992).

13.6. The factors effecting coagulation

Stability of suspensions can be determined in different ways. It can be the meas-
urement of turbidity at appropriate wavelength of light (Fig. 13.9a) (Yotsumoto and
Yoon, 1993a), relative stability as turbidity quotient at the initial time and at certain
time of coagulation (Fig. 13.9b) (Sadowski and Smith, 1985), the height of the turbid-
ity line as function of time (Fig. 13.9c) (Bodman et al., 1972), the speed of particle
falling based on straight-line range of coagulation shown in Fig. 13.9c (Fig. 13.9d)
(Bodman et al., 1972), solid concentration at certain height or part of liquid (Fig.
13.9e) (Pugh, 1974), and stability ratio W being a quotient of maximum settling rate
under certain condition and settling rate when the settling is maximum (Fig. 13.9f)
(Wiese and Healy, 1975), as well as other methods.
The stability of suspensions depends on their electrical potential in the Stern layer
or a slipping plane, as well as on dispersion or structural interactions. Stability of sus-
pensions is easy to regulate with pH regulators and with soluble salts which strongly
effect the zeta potential of particles.
It is shown in Fig. 13.9 a, b, e, and f that suspensions are not stable close to their
iep and pzc points. This results from the fact that their zeta potential is at iep equal
zero and there is no electrostatic repulsion between particles. On the other hand, the
increase in suspension stability at high values of zeta potential results from the appear-
ance of energy barrier which, after Yotsumoto and Yoon (1993a), has been shown in
Fig. 13.10.
Stability of two- and multi-mineral suspensions depends on the parameters pre-
sented in Fig. 13.9 and, first of all, on the sign of zeta potential. If the sign of zeta po-
tential of both kinds of particles is opposite, a strong attraction of particles leads to a
non-barrier heterocoagulation.
If coagulation is carried out by the change of pH or addition of potential-determining
ions (Fig. 13.9), a rapid coagulation takes place not only at iep, but also in its vicinity.
This results from the fact that electrostatic repelling of particles at low Stern’s poten-
tials or zeta potentials are compensated by dispersion forces. Generally, coagulation
takes place when the zeta potential of particles is lower than 25 mV and accurate value
of the potential depends on the Hamaker constant of particle and the magnitude of
structural forces.
 434 Part III. Separation processes

stability, m t = 5 min/m t = 0 minx100%

100
150
SiO 2
2×10-3 NaCl
130
80
turbidity, τ

2×10-2 NaCl Fe 2 O3
110

90
60
2×10-1 NaCl BaSO 4
70
TiO 2

50 40
2 4 6 8 10 12 2 4 6 8 10 12 pH
pH iep SiO2 iep Fe3O4 iep BaSO4

a b
0.12 TiO 2
80 TiO 2
height of solids level, cm

pH = 2.5 0.10
intermediate
60 0.08

0.06
40 high
0.04
low
se

0.02
20
0.00
*, o, • – solids content 0 0.01 0.02 0.03 0.04
0
0 5 10 15 solids content, ml TiO 2 / ml water
coagulation time, min

c d
3 6
50 10-4 KNO3
Fe2O3
solids content in water, %

5
20 TiO 2 Al2O3
2
stability ratio, W

4
10
1
5
3

0 2
pzc I pzc pzc II

-1 1
0 2 4 6 8 10 12 14 -experiment, - -theory
pH 0
5 6 7 8 9 10
pH
e f
Fig. 13.9. Methods of suspension stability measurements and presentation
 13. Coagulation 435

Consideration by Derjagin (1989), based on the classical theory DLVO, showed

that the conditions under which coagulation, caused by potential-generating ions, takes
place are determined by the relation:
ψ s2
= const , (13.84)
A131κ c
where:
κ – the Debye parameter for a given salt concentration
ψs –the Stern or zeta potential, also dependent on salt concentration.
The relationship agrees with the empirical formula of Eilers–Korff
ζ c2
= const , (13.85)
κc

in which ζ c is the zeta potential for a given salt concentration.

a b
vt τ
x x

x x
x x

x x

H pH
Fig. 13.10. Comparison of the magnitude of energy barrier of interaction of particles Vt (a) with suspen-
sion stability (b); τ stands for turbidity of titanium oxide (rutile) suspension, determined experimentally
(after Yotsumoto and Yoon, 1993a)

If coagulation is caused by electrolyte (salt) addition, then the ksk (critical concen-
tration of coagulation), which indicates the salt concentration necessary for coagula-
tion, depends on electrical charge of the salt ions. According to Derjagin (1989) and
DLVO theory for water at 250C at high surface potentials of particles, the ksk amounts
(Hunter, 1989):
87·10 −40
ksk (mol/dm3 ) = , (13.86)
z 6 A131
2

where z is a positive number 1, 2, 3… and means counter ion valance which mainly
contributes to the compensation of the edl charge. For positively charged surfaces it is
an anion, while a cation is for a negatively charged surface. In Eq. (13.86) the
 436 Part III. Separation processes

Hamaker constant A131 is expressed in joules. This dependence is also in agreement

with the empirical equation described by Hardy and Schultz, which says that the ksk is
proportional to minus sixth power of counter ion valance and it does not depend on
surface potential, i.e.:
6 6
⎛ 1⎞ ⎛1⎞
(ksk )
z =1 : (ksk )z =2 : (ksk )z =3 = 1 : ⎜ ⎟ : ⎜ ⎟ = 1 : 0.016 : 0.0014
6
(13.87)
⎝ 2 ⎠ ⎝ 3⎠
This rule is fulfilled for instance for As2S3 and FeOOH colloids (Sonntag, 1982),
but it does not function for any other systems. Therefore, a new formula for an accu-
rate description of concentration and valence of electrolyte ions on suspension ksk has
still to be worked out. Recently, the Hardy-Schultz relation has been derived following
theoretic considerations and it also takes into account the ions of different valence
(Hus and Kuo, 1995). According to Glazman and Barboj (Laskowski, 1969) coagula-
tion takes place when the following condition is fulfilled:
ψ s z = const , (13.88)
while according to Hunter (1987) who used the classical DLVO theory, it is:
⎛ψ 4 ⎞
ksk = f ⎜⎜ 20 ⎟⎟ , (13.89)
⎝z ⎠
where:
z – valence of counter ion
ψ0 – surface potential.
Taking the above discussion into account it can be stated that the issue of accurate
determination of the conditions of coagulation is complicated and to obtain a complete
story it is necessary to incorporate structural forces, as well as to consider the fact that
the potential zeta is used for calculation instead of the surface or Stern potentials. In-
disputable is the fact that a rapid coagulation always takes place at the iep point and at
its vicinity.
Even more difficult is the delineation of heterocoagulation conditions which in-
volves different particles, since both dispersion forces and electrical ones can be posi-
tive and negative. When the Hamaker constant for the interaction of different particles
in water is positive and the signs of the surface potential are opposite, then a non-
barrier heterocoagulation occurs. However, when the Hamaker constant is positive
while the potentials are of the same sign, heterocoagulation, under the influence of pH
or other potential determining ions, will be governed by the equation (Derjagin, 1989):
ψ 1ψ 2
= const . (13.90)
A132 κ c
 13. Coagulation 437

When dispersion forces are positive (repelling) and surface potentials are of the
same sign, the system is usually stable. There exist, however, regions of electrolyte
concentrations where the above rule does not apply. If the potentials of the same sign
are considerably different as to their absolute values and when the ionic force is of a
low value, the energy barrier can appear at a long distance between particles, thus the
dispersion forces will not effect the interactions and the repelling forces become pro-
portional to the lower potential. When the distances are short, an energy minimum can
appear, which indicates attraction (Derjagin, 1984) and the force acting on a unit sur-
face area is determined by the equation:
Πmin ∝ (ψ1 – ψ2 )6 /(A131)2. (13.91)

13.7. The effect of other substances on the stability of sus-

pensions
The stability of suspensions is effected not only by potential-determining ions and
electrolytes in the form of soluble salts, but also other factors, like hydrolyzing
cations, solids precipitating from the solution (precipitates), surfactants and polymers
(Fig. 13.11).
100
a
Me +3 Me +2 Me +
ions
0
suspension turbidity

100
b
precipitates

0
100
c
surfactants
0
100
d

polymers
0
10 -8 10 -6 10 -4 10 -2 1
coagulant concentration, kmol/m 3
Fig. 13.11. Influence of concentration of various species on coagulation
(after Weber, 1972)
 438 Part III. Separation processes

Part a of the figure shows the already discussed rule stating that coagulation of a
negatively charged surface due to multivalent cations takes place at considerably
lower concentrations than that for monovalent cations. There are two mechanisms of
their action. The first one involves lowering the zeta potential as a result of diminish-
ing edl thickness, while another is caused by a specific sorption of ions and their hy-
drolyzed form.
Many multivalent cations of metals in aqueous solutions undergo reaction with wa-
ter, i.e. hydrolysis, forming hydrocomplexes. Hydrolysis of multivalent cation be-
comes considerable when the pH of the solution is regulated by adding, for instance,
trivalent iron ions (Fe(III)) forming Fe3+ ions and hydrolyzed ions, i.e. hydrocom-
plexes such as Fe(OH)2+, Fe(OH)2+ and Fe(OH) 30.
The concentration of particular hydrocomplexes depends on the quantity of added
salt of multivalent cation and pH of the solution. The Ca(II) ions behave similarly and
their distribution is presented in Fig. 13.12b. The formed hydrocomplexes undergo
strong specific adsorption on mineral surfaces. According to Fuerstenau and Han
(1988) especially strong sorptive properties are offered by monohydroxycomplexes
Me(OH)n+, (e.g. Ca(OH)+, Pb(OH)+, Fe(OH)2+ for Fe(III), Fe(OH)+ for Fe(II)). A
maximum sorption of a monohydroxycomplex on solids takes place at pH, when there
is maximum concentration of these ions in the solution (Fig. 13.12a).
It is believed that the sorption of monohydroxycomplexes of hydrolyzing metal
ions on mineral surfaces follows the reaction of exchange:

M-O-H + HOM+ → M - O M+ + H2O (13.92)

or through hydrogen bounding

M-O-H + HOM + → M - O H .... O - M +

(13.93)
H

The specific sorption of the hydroxycomplexes of metals causes that the surface
potential of negatively charged particles and accompanying potential zeta, become
altered towards positive values. The higher concentration of metal monohydroxycom-
plex, the more considerable alteration of zeta potential, which, as it is shown in Fig.
13.13, is connected with the pH of the solution. At a high sorption, the zeta potential
of particles becomes identical with zeta potential of hydrolyzing metal ions and the
maximum value of the electrokinetic potential is recorded for pH at which monohy-
droxycomplex concentration is the highest. At slightly lower pH there exists an addi-
tional isoelectric point, called the iep of reversion.
The suspension can be destabilized by adding other substances or precipitation of
particles from the solution. Then, heterocoagulation takes place (Fig. 13.11b). Also
 13. Coagulation 439

surface active substances can be used for coagulation of suspensions (Fig. 13.11c).
Their applicability depends on the kind of surfactant. The best are those undergoing
significant adsorption on particle surface resulting in the reduction of the zeta potential
of particle. Some surfactants additionally cause strong hydrophobization of particles,
and the enhancement of coagulation through hydrophobic interactions. However, a too
high concentration of a surfactant leads to coagulum disintegration. The phenomenon
takes place due to adsorption of a second layer of the surfactant in such a way that its
polar part extends into the solution, which provides particle surfaces with high electri-
cal charge, and thus, also with stability.
40
adsorption, mol/cm 2×10 10

SiO2
30

100 ppm Ca 2+
20

10

0
4 6 8 10 12 14
pH
a
-03
10

Ca 2+
concentration, kmol/m 3

-04
10
CaOH +

-05
10

Ca(OH) 2(s)
-06
10
8 9 10 11 12 13 14
pH
b
Fig. 13.12. Adsorption of calcium ions on quartz as a function of pH of solution
at a total concentration of calcium ions equal to 100 ppm (a), distribution diagram dependent on solution
pH (b). Total concentration of Ca2+ = 10–3 kmol/m3 (after Fuerstenau and Hana, 1988)
 440 Part III. Separation processes

The stability of the suspension can be described and stability factor W and can be
calculated in the presence of surfactants using classical coagulation theories. In such a
case it is necessary to modify the expressions for the energy of particular interactions,
taking into account that the surfactants have been adsorbed on particles. According to
Schick (1987) for nonionic surfactants the following relations are valid:
πN a c 2f Z 2 kT
VS = (0.5 − χ F ) [3 ( R + δ ) − 0.5Z ] , (13.94)
3V ρ a2

2
⎛ kT ⎞
VR = 32πεε 0 ⎜ ⎟ ( R + δ )ψ exp− κH ,
2
(13.95)
⎝ ze ⎠

VA = −1/12( A ...), (13.96)

where:
VA – a complex function dependent on combination of the Hamaker constant for par-
ticle, particles with adsorbed layer and distance between particles
V – partial molar volume of water
Na – the Avogadro number
ρa – density of adsorbed layer
δ – thickness of adsorbed layer
cf –surfactant concentration in adsorbed layer
Z – distance of penetration of adsorbed on particles surfactants
χF – the Flory–Huggins parameter of interactions of polymer–solvent
κ – the Debye radius
ε – dielectric constant
ε0 – dielectric permeability of vacuum
ψ –electrical potential
H – distance
R – particle radius.
Suspensions can be also destabilized using polymers (Fig. 13.11d) of long enough
chains, i.e. high molecular mass of one million or more. Such polymers cause coagula-
tion which is called flocculation. It should be noticed, however, that too high concen-
tration of the polymer can lead to restabilization of the suspension as a result of very
high electrical charge on particle surface, originating from ionic functional groups of
the polymer.
Coagulation and heterocoagulation processes are very important from a practical
point of view. But they can be both useful and harmful. Slime coating relying on ad-
hesion of fine gangue particles to useful ore particles due to the forces described by
DLVO theory is an example of harmful coagulation. Slime coating causes a decrease
 13. Coagulation 441

in the selectivity of separation. For instance, hydrophobic surfaces can be rendered

hydrophilic as a result of a contact with clay minerals. Such a phenomenon can be also
observed in the flotation of coal or metal sulfides. In some cases slime coating is use-
ful and it is used in the so-called carrier methods of enrichment. They depend on con-
trolled adhesion of useful particles to larger particles of other mineral. Sometimes
coarse particles are added into the ore, then coarse particles carrying fine particles on
their surface are separated.

+
M(OH) 2
zeta potential, ζ , mV

iepSiO 2 iepM(OH) 2
0

SiO2

– increase of CM2+

pH
Fig. 13.13. Zeta potential of quartz as a function of pH and concentration of bivalent hydrolyzing cation
(James and Healy, 1972)
Coagulation and peptization are also accompanied by other processes or effects
known as syneresis, coacervation, ticsothropy, creaming and gelling.
Coacervation is a phenomenon analogical to coagulation but taking place in poly-
meric colloidal suspensions treated with salts. Such a behaviour can be seen in protein
solutions. Syneresis is a phenomenon of increasing compactness of structures of co-
agulating material. Coagulum is formed due to coagulation and with time coagula be-
come more compact as a result of ordering and increasing number of contact points,
which finally leads to higher strength of aggregates. Leaving them alone for a long
time can create an unified particle. Transformation of colloid silica into quartz
(Sonntag, 1982) following the pattern: silica → silica gel → opal → chalcedone →
quartz is an example of syneresis. The term sol has been less and less frequently used
to refer to colloid suspension of solids in water. The term gel denotes a colloid system
which underwent coagulation forming a strong net–like structure which leads to its
characteristic jelly form. The most frequent subject to gelling are organic colloids
(starch), less often inorganic ones (silica). Gel–like form of a colloid results from ani-
sotropic properties of particles, which causes formation of three–dimensional struc-
tures. With time, the gel undergo consolidation, i.e. synersis. Micelle - the term used
previously to describe a particle with its electrical double layer structure around the
 442 Part III. Separation processes

colloid particle while presently it is used for description of association of surfactant

ions and particles. Coalescence – is a phenomenon of aggregation taking place be-
tween small objects in colloidal systems. Is has a broader notion than coagulation,
since it involves not only solids but also liquids. Coalescence of oil emulsion leads to
separation of phases into oil and, separately, water phases.

13.8. Selective coagulation

When zeta potential of two minerals are of the same sign, the suspension can be ei-
ther stable or coagulation of one of the components can take place. Such a behavior of
particles is a base for selective coagulation. Selective coagulation in the quartz–
hematite system after Pugh (1974, 1992) is shown in Fig. (13.14).
3
solids concentration, %

0
iep SiO2 iep Fe2O3

-1
0 2 4 6 8 10 12 Fig. 13.14. Stability of aqueous
pH suspensions of quartz and sepa-
rately hematite as a function of
a pH of solution (a) and the range of
3 selective coagulation of hematite
stability (b). After Pugh 1974 and 1992
selective coagulation
2 of hematite
solids density, %

heterocoagulation
0

Fe2O3 + SiO2
-1
0 2 4 6 8 10 12
pH
b
 13. Coagulation 443

Within pH range of 1-7 the suspension of quartz and hematite is unstable and hetero-
coagulates due to negatively charged quartz particles and positively charged hematite
particles. In a weak alkaline solution, at pH about 7.5, the hematite particles are slight
negatively charged and they are stabile. This situation provides the conditions for se-
lective coagulation of hematite. Above pH 8.5 both minerals are highly negatively
charged and stable.
Selective coagulation can be also carried out in the quartz–rutile (Pugh 1974,
1992), and coal–ash forming minerals systems (Yoon et al., 1991). In the case of coal,
coagulation takes place within a narrow pH range about 10 (Fig. 13.15).
coal coagulation

heterocoagulation stability

100 15

80 12
coal recovery, %

60 9 ash content, %

40 6

ash
20 3
recovery

0 0
2 4 6 8 10 12
pH
Fig. 13.15. Range of pH stability, coal coagulation, and its heterocoagulation
(after Yoon et al., 1991)

13.9. The structure of coagula

Coagula are formed during coagulation. Characteristic structures of coagula ap-
pear when the surface of particle is made of different crystallographic planes of differ-
ent surface properties. For example they can form the so called “house of cards” struc-
ture (Fig. 13.16). Among possible structures, the “face–to–face”, “face –to-edge” and
 444 Part III. Separation processes

“edge-to-edge” formations can be distinguished. They represent the first order struc-
tures which can, like bricks, be arranged as complicated form of a second order. Van
Olphen (1963) lists the following structures: edge-to-plane coagulated and aggregated,
plane-to-plane face coagulated but not aggregated.
The structure of coagula is complicated and there is a lack of accurate definitions
for their description. A list of different formulas used for the description of particle
shape can be found, for instance in the work by Sztaba (1964).

a b
Fig. 13.16. Structure of coagula. Random structure of spherical particles (a)
and house-of-cards structure for flat particles (b) (Hunter, 1993

Czarnecki (1986) applied some of them to determine the particle shape factor ζk
3 3
1
ξk =
aπ 2
∑ (a − a )
i
2
sign ∑ (a − a )
i
3

i =1 i =1 , (13.97)

where:
a1, a2 and a3 – inertia moment of particle in relation to its main axis of rotation
a = (a1 + a2 + a3)/3.
The first square element of the equation represents deviation from spherical shape,
while the second one, which is to the third power, enables distinguishing between ob-
long and oblate forms. Factor ζk defined in this way bears the following properties: for
spherical particles ζk=0, for rods ζk>0, while for disc coagules ζk<0. The term sign in
Eq. 13.97 determines the sign of the function, since sign (ym)=ym /│ym│(=+1).
The shape of coagula can be also described with the use of fractals. Fractal geome-
try, introduced by Mandelbrot (1977, 1982), describes objects which are of the same
appearance at different magnifications. Details regarding fractals can be found in
many works, including the article by Nowak (1992). A fractal character, repeated at
different levels of structure organization, can be assigned to particle aggregates.
 13. Coagulation 445

The parameter to describe coagula is fractal dimension d which is the measure of

coagulum density or the degree of particle binding in coagulum. This fractal dimen-
sion of a coagulum is expressed by the relation (Hutner, 1993)
d = log x1/log x2, (13.98)

where:
x1 – quotient of coagulum mass increase
x2 – quotient of coagulum size (length) increase.
Formula (13.98) can be also written in another form (Feder, 1988; 1988; Koylu et
al. 1995, Adachi et al., 1990), e.g.
nk = kk (Rk/R)d, (13.99)

where:
nk – number or primary particles in coagulum
Rk – gyration radius of coagulum
R – radius of individual particle in coagulum
kk – dimensionless coefficient.
According to Hunter (1993), fractal dimension of a coagulum d can be estimated
with the use of the measurement of intensity of light diffused by coagulum, since:
ln I = bλQr–d = b (4πnp sin (θ /2)/λ)–d, (13.100)
where:
bλ – constant
Q – vector of light scattering
θλ – angle of scattering
λ – wavelength
np – number of particles.
Equation (13.100) is valid for Qr values within the range R<<Qr-1<<Rk, where R is
a particle radius, while Rk is an aggregate radius. Therefore, this equation becomes
useful as soon as noticeable coagulation takes place and a plot of ln I as a function of
Qr becomes linear with the slope equal to d.
Regular flat (two-dimensional) particle packing in a coagulum would provide frac-
tal dimension 2, while regular three-dimensional fractal dimension 3. Actual value of
fractal dimension of spherical coagulum amounts from 1.7 (Koylu, 1995) to 2.2 (Ada-
chi, 1990) i.e. about 2, which indicates their relatively loose packing.
 446 Part III. Separation processes

Literature
Adachi Y., Ooi S., 1990. Geometrical structure of a floc, J. Colloid Interface Sci., 135(2), 374–384.
Adamczyk Z., Dabros T., 1978. On the convective diffusion of particles under electrical double – layer
forces, Colloid Interface Sci., 64 (3), 580–583.
Adamson A., 1963. Chemia fizyczna powierzchni, Warszawa, PWN.
Bodman S., W., Shah Y.T., Skriba M., 1972. Settling of flocculated suspension of titanium dioxide and
alum mud in water, Ind. Eng. Chem. Process Des. Develop., 11 (1).
Cerda C., Non-Chhom K., 1989. The use of sinusoidal streaming flow measurements to determine the
electrokinetic properties of porous media, Colloid and Surfaces, 35, 7–15.
Chander S., Hogg R., 1987. Sedimentation and consolidation in destabilized suspensions, in: Flocculation
in Biotechnology and Separation Systems, Y.A. Attia (ed.), Elsevier, Amsterdam, 279–296.
CRC, 1986/87. Handbook of chemistry and physics, CRC Press, Boca Raton.
CRC, 1997/98. Handbook of chemistry and physics, CRC Press, Boca Raton.
Czarnecki J., 1986. The effect of surface inhomogeneities on the interactions in colloid systems and col-
loid stability, Advances in Colloid and Interface Science, 24, 283–319.
Dabros T., Adamczyk Z., Czarnecki J., 1977. Transport of particles to a rotating disk surfcae under an
external force field, Colloid Interface Sci., 62(3), 529–541.
Dabros T., Czarnecki J., 1980. Wpływ oddziaływań elektrycznych na kinetykę heterokoagulacji, Fizyko-
chemiczne Problemy Mineralurgii, 12, 47–55.
Derjagin B.V., 1989. Theory of stability of colloids and thin films, Consultants Bureau, New York.
Derjagin B.V., Churaev N.V., 1989. The current state of the theory of long-range surface forces, Colloids
and Surfaces, 41, 223–237.
Derjagin B.V., Landau L., 1941. Theory of the stability of strongly charged lyophobic sols and of the ad-
hesion of strongly charged particles in solutions of electrolytes, Acta Physicochim., USSR, 14,
633–662.
Drzymala J., 1994. Hydrophobicity and collectorless flotation of inorganic materials, Advances in Col-
loid and Interface Sci., 50, 143–185.
Drzymala J., 1999. Characterization of materials by Hallimond tube flotation, Part 3. Maximum size of
floating and interacting particles, Int. J. Mineral Processing 55, 203–218.
Drzymala J., Sadowski J., Hołysz L., Chibowski E., 1999. Ice/water interface: zeta potential point of zero
charge, and hydrophobicity, J. Colloid Interface Sci., 220, 229–234.
Drzymala J., Laskowski J., 1980. Pomiary elektrokinetyczne w mineralurgii, Fizykochemiczne Problemy
Mineralurgii, 12 (35–45).
Duran J.D.G., Guindo M.C., Delgado A.V., Gonzalez-Caballero F., 1995. Stability of monodisperse zinc
sulfide colloidal dispersions, Langmuir, 11, 3648–3655.
Działoszyński I.E., Lifszyc E.M., Pitajewski L.P., 1959. Teoria sił van der Waalsa, Žurnal Eksp. Teor.
Fiz., 37, No. 1, 229–241.
Ebaadi S.H., 1981. Van der Waals inetractions beween surfactant-coated and bare colloidal particles,
Colloids Surfaces, 2, 155–168.
Feder J., 1988. Fractals, Chap. 2, Plenum Press, New York 1988.
Friberg S.E., , 1991. Emulsion Stability, in: Emulsions – a fundamental and practical approach, J. Sjob-
lom (ed.), NATO Series (C), Vol. 363, Kluwer Academic Pub., Dordrecht, 1–24.
Fuerstenau M.C., Han K.N., 1988. Activators, in: Reagents in Mineral Technology, Surfactant Science
Series, Vol. 27, P. Somasundaran i B.M. Moudgil (eds.), Dekker, 411–428.
Gray S.R., Langberg D.E., Gray N.B., 1994. Fine mineral recovery with hydrophobic magnetite, Int.
Journal of Mineral Processing, 41, 183–200.
Gregory J., 1969. The calculation of Hamaker constants, Advances in Colloid and Interface Sci., 2, 396–
417.