Electrokinetic properties of silica, alumina, and montmorillonite

Material Information

Electrokinetic properties of silica, alumina, and montmorillonite
Horn, John Milton, 1951-
Publication Date:
Physical Description:
xiv, 153 leaves : ill. ; 28 cm.


Subjects / Keywords:
Adsorption ( jstor )
Aluminum ( jstor )
Charge density ( jstor )
Hydrogen ( jstor )
Ions ( jstor )
Montmorillonite ( jstor )
pH ( jstor )
Sodium ( jstor )
Streaming ( jstor )
Surface areas ( jstor )
Aluminum oxide ( lcsh )
Dissertations, Academic -- Materials Science and Engineering -- UF
Materials Science and Engineering thesis Ph. D
Montmorillonite ( lcsh )
Silica ( lcsh )
bibliography ( marcgt )
non-fiction ( marcgt )


Thesis--University of Florida.
Bibliography: leaves 145-152.
General Note:
General Note:
Statement of Responsibility:
by John Milton Horn, Jr.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright [name of dissertation author]. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
000206963 ( ALEPH )
04054671 ( OCLC )
AAX3757 ( NOTIS )

Full Text








To My Lovely Wife, Barbara



Deep appreciation and many thanks are extended to the

author's graduate supervisory committee which included

Dr. G. Y. Onoda, Jr., chairman; Dr. L. L. Hench;

Dr. R. W. Gould; and, Dr. D. O. Shah. Special thanks go to

his advisor, Dr. G. Y. Onoda, Jr., without whose many

helpful and lengthy discussions, this work would not have

been possible.

The author wishes to thank Mr. Fumio Ouchi for the

Auger data presented in Chapter 7. Also, thanks go to

Mr. Peter Curreri and Mr. Jim Adair for helpful discus-

sions, and to Mr. Nick Gallantino for technical

assistance in the lab.

Finally, the author wishes to acknowledge the

National Institute of General Medical Sciences grant

#GM21056-02 and the National Science Foundation

grant #AER76-24676 for partial financial support for this





ACKNOWLEDGEMENTS....................................... iii

LIST OF TABLES ........................................ vi

LIST OF FIGURES........................................ vii

ABSTRA CT .............................................. xi

1 INTRODUCTION .............. .................. 1

FLOW IN POROUS BEDS. ......................... 8
Introduction.............................. 8
Materials and Methods..................... 11
Results and Discussion.................... 16
Conclusions.............................. 24

SOLUTION .................................... 25
Introduction ............................. 25
Procedure................................ 28
Results ................................. 35
Discussion ............................... 41
Conclusions.............................. 47

PHORESIS MEASUREMENTS ........................ 49
Introduction. ........................... 49
Materials and Methods..................... 50
Results and Discussion.................... 51
Conclusions.............................. 64



ON FUSED SILICA ............................. 66
Introduction ............................. 66
Materials................................ 70
M ethods .................................. 71
Results and Discussion................... 76
Conclusions................................ 84

CHARGE OF ALUMINA........................... 85
Introduction............................. 85
Materials and Methods...................... 87
Resu lts .................................. 90
Discussion................................ 97
Conclusions.............................. 101

MONTMORILLONITE.............................. 103
Introduction ............................. 103
Materials................................ 107
Methods.................................. 109
Results and Discussion................... 111
Conclusions ................................ 136

APPENDIX A ............................................ 140

APPENDIX B ............................................ 142

BIBLIOGRAPHY........................................... 145

BIOGRAPHICAL SKETCH................................... 153


Table Page
I Standard Free Energy of Formation Values
at 2980K (Kcal/mole)............................ 29

II Reactions Depicting Formation of the
Neutral Soluble Silicate Species,
H2SiO3, and Their Equilibrium
Constants (K) ................................. 31

III Activity Coefficients for Ionic Species
at Various Solution Concentrations............ 32

IV Calculated Values for Disturbed Layer
Thickness (t) and Soluble Si02 at
Equilibrium Using Data from van Lier
et al. (ref 22) .............................. 46

V Values of Co for Various Zeta Potentials
and Weight Percent Solids ..................... 55

VI Values of C, F and log (C/F) for Various
Zeta Potential Values.......................... 59

VII Surface Areas of A-16, A-17, T-61, C-30 DB
and Gamma Alumina Powders...................... 91

VIII Compositions of Montmorillonite Clays 1608
and 1613 in Weight Percent .................... 108

IX Chemical Conditions for Zero Zeta
Potential (ZZP) for 1608 and 1613
Montmorillonites.............................. 125


Figure Page

1 Streaming potential cell; a) PMMA tube;
b) threaded PMMA plugs; c) platinum leads
to electrodes consisting of perforated
disks with attached platinum mesh;
d) platinum electrode; e) solution flow
entrance; f) solution flow exit;
g) porous bed; h) O-ring ..................... 13

2 Streaming potential solution flow system..... 15

3 R-C circuitry used in streaming potential
experiments.................................. 17

4 Streaming potential vs. pressure using
the R-C measuring circuit.................... 19

5 Streaming potential vs. pressure without
the R-C measuring circuit.................... 21

6 Flow rate vs. pressure for fused silica...... 22

7 Surface charge density error (Ao) vs. pH in
10-1, 10-2, 10-3 M/L NaCI solution for
vitreous silica using Bolt's experimental
conditions.................................... 36

8 Surface charge density error (Ao) vs.
total surface area for amorphous silica
at pH = 7, 8, 9 and 10....................... 37

9 Surface charge density error (Ao) vs. pH
in 10-1, 10-, 10-3 M/L NaC1 solutions
for hydrated silica using Tadros's and
Lyklema's experimental conditions............. 39

10 Surface charge density error (Ao) vs.
total surface area for precipitated
silica at pH = 7, 8, 9 and 10................ 40


LIST OF FIGURES continued.

Figure Page

11 Zeta potential vs. concentration of
A1C13-6H20 for 0.00125, 0.0025, 0.025,
0.2 and 1.0% weight percent montmoril-
lonite........................................ 52

12 Apparent aluminum ion concentration Co
vs. weight percent montmorillonite for
zeta potential = 0 at pH = 4.0............... 56

13 Concentration of aluminum Co minus the
equilibrium concentration (C) of aluminum
in solution vs. weight percent montmoril-
lonite for zeta potential = +15, +10, +5,
0, -5, -10 mV at pH = 4.0.................... 57

14 Zeta potential and log adsorption density
(F) vs. equilibrium concentration (C) of
aluminum in solution for montmorillonite
at pH = 4 .0 .................................. 60

15 Log of the ratio of C and F vs. zeta
potential for montmorillonite for the
experimental data............................ 63

16 Streaming potential apparatus modified
to maintain constant flow pressure;
a) secondary reservoir; b) primary
reservoir; c) pump; d) solution flow
valve; e) cell; f) electrometer;
g) recorder; h) solution head height......... 73

17 Zeta potential vs. concentration of
sodium citrate and aluminum chloride......... 74

18 Streaming potential-pressure ratio vs.
stream time for untreated, heat treated,
base treated heat treated and base
treated and non-aluminated fused silica...... 77


LIST OF FIGURES continued.

Figure Page

19 Zeta potential vs. concentration of
sodium citrate for fused silica in
supporting electrolytes of 10-3 M/L
NaCl and 10-3 M/L NaC1, 10-4 M/L
A1C13-6H20.................................... 81

20 Streaming potential-pressure ratio vs.
stream time for aluminated fused silica
for untreated, heat treated, base
treated and heat treated and base
treated surfaces............................. 83

21 Zeta potential vs. pH for y-alumina
after aging for 0, 1, 8, 16, 33 and
97 days in water ............................. 93

22 pH of the zero point of charge (pHZPC)
vs. aging time in water for T-61, A-17,
C-30 DB, y, and A-16 aluminas................. 94

23 Solution pH vs. weight of alumina added
for A-16 alumina.............................. 96

24 Zeta potential vs. pH for unwashed T-61
alumina (coarse particles) aged in water
for 1, 2 and 3 days.......................... 100

25 Zeta potential vs. pH for 1608 and 1613
montmorillonite............................... 112

26 Zeta potential vs. pH for 1613Na and
1613Ca montmorillonite....................... 113

27 Zeta potential vs. log concentration of
NaC1, CaC12-2H20 and A1C13-6H20 solutions
for 1608 montmorillonite..................... 115

28 Zeta potential vs. log concentration of
NaCI, CaC12-2H20 and AlC13-6H20 solutions
for 1613 montmorillonite..................... 116

LIST OF FIGURES continued.

Figure Page

29 Zeta potential vs. log concentration
of NaC1, CaC12-2H20, and A1C13-6H20
for 1613Na montmorillonite.................... 117

30 Zeta potential vs. log concentration
of NaC1, CaC12'2H20, and A1C13-6H20
for 1613Ca montmorillonite................... 118

31 Zeta potential vs. pH for 1608 montmoril-
lonite in 0, 10-5 and 10-4 M/L A1C13-6H20
solutions .................................... 122

32 Zeta potential vs. pH for 1613 montmoril-
lonite in 0, 10-5, 10-4 M/L A1C13-6H20
solutions .................................... 123

33 Auger peak height ratio vs. pH for
vitreous silica after one hour exposure
to 10-4 M/L A1C13-6H20 solution.............. 128

34 Zeta potential vs. pH for 1613Mat in
10-4 M/L A1C13-6H20 solution. SSH-O
means single solution method, clay
plus Al solution aging time = 0 hours.

35 Zeta potential vs. pH for 1613Mat in
10-4 M/L Al(N03)3-9H20 solution.............. 130

36 Zeta potential of 1613 montmorillonite
in 10-4 M/L A1C13-6H20 and water at pH =
6 vs. clay equilibrium time at pH = 4,
6 an d 8 ...................................... 132

37 Solution pH vs. total weight of montmoril-
lonite added to 25 ml of water............... 135

Abstract of Dissertation Presented to the
Graduate Council of the University of Florida
in Partial Fulfillment of the Requirements for the
Degree of Doctor of Philosophy



John Milton Horn, Jr.

March 1978

Chairman: George Y. Onoda, Jr.
Major Department: Materials Science and Engineering

The surface chemistry of montmorillonite, vitreous

silica and alumina is investigated by electrokinetic

methods of streaming potential and microelectrophoresis.

Also, improved analytical techniques are developed to

obtain these electrokinetic data and adsorption

information for these oxide materials.

Many investigators find that linear streaming

potential-flow pressure relationships do not pass through

the zero potential-zero flow pressure origin. One source

of this error, electrode polarization, is eliminated

by introducing an R.C. circuit in the system. Simulta-

neous study of flow pressure versus flow rate reveals a

nonlinear curve suggesting the existence of noncreeping

flow in the pressure range studied. However, streaming

potential-flow pressure relationships remain linear and

pass through the origin when electrode polarization

effects are eliminated. Therefore, zeta potential

values calculated from the Smoluchowski equation under

noncreeping flow conditions are as valid as if creeping

flow conditions existed.

Using mass balance concepts, surface charge

density error incurred by assuming that all hydrogen

or hydroxide ions which cannot be accounted for as free

ions in solution adsorb to the surface rather than form

part of complex ions in solution is calculated for five

silica phases in three electrolyte solution concentra-

tions. The error increases as thermodynamic stability

of the silica decreases, as pH increases and as the total

surface area present in the experimental system decreases.

A new method is developed for determining adsorption

isotherms directly from electrophoretic measurements.

The unique feature is gathering zeta potential data as a

function of ion concentration for various solids

concentrations. The adsorption of aluminum ions onto

montmorillonite clay is presented as an experimental

example. The values of adsorption densities and equili-

brium concentrations of aluminum ions in solution after


adsorption determined by this method fit an analytical

form of the Stern equation.

The desorption of aluminum ions from thermally

and/or chemically treated vitreous silica surfaces is

investigated by observing the changes of the streaming

potential-flow pressure ratio as a function of streaming

time. Aluminum ions desorb from surfaces whose treatment

has created adjacent silanol groups whereas they remain

on surfaces which contain isolated silanol groups.

Analogous to interpretations of adsorption-desorption

phenomena for fine particle systems, the mechanism of

adsorption onto coarse particle surfaces is hydrogen

bonding of the aluminum ion to isolated hydroxyl groups

on the silica surface.

The aging of alumina powder surfaces is studied by

observing changes in the zero point of charge (ZPC) with

aging time in water using electrophoresis measurements.

Alumina powders age due to the changing hydration state

of the surface. Grinding of T-61 powder from relatively

coarse to fine powder causes the ZPC to shift from

pH = 7.0 to pH = 9.5. Nonalpha phase aluminas age much

more slowly but to a greater extent than alpha phase


The surface chemistry properties of montmorillonite,

an aluminosilicate clay mineral, are investigated by


electrophoresis. Ion exchange of sodium, calcium and

hydrogen ions in solution for similar cations in the

clay controls the electrokinetic behavior of the

montmorillonite. Aluminum ions are found to specifically

adsorb from solution onto the surface. However, the

degree of aluminum ion adsorption as shown by electro-

phoresis measurements depends on the equilibration time

and pH of the clay in water whereas the electrokinetic

behavior of the clay in the absence of aluminum ions in

solution is independent of equilibration time and pH

in water.



The surface chemistry of montmorillonite, an

aluminosilicate clay mineral, and its two major con-

stituents, silica and alumina, is investigated by

electrokinetic methods of streaming potential and micro-

electrophoresis. Using these techniques, many authors

have studied the changing characteristics of the elec-

trical double layer surrounding the particles or colloids

as a function of adsorption of ions from solution, ion

exchange, and/or aging of hydrated surfaces. However,

proper use of these two techniques can also yield

important information about the mechanisms of these

processes. Opportunities are provided for improving

these techniques which can then be used to develop new

methods for determining adsorption properties of oxide


The electrokinetic parameter used to describe the

nature of the electrical double layer is the zeta

potential. Essentially, the zeta potential is the

electrical potential at the Stern plane in solution. The

Stern plane separates the diffuse (Gouy-Chapman) layer of

counter ion charge (the ions in solution which electrical-

ly balance the charge on the solid surface) from the

layer of adsorbed ions (Stern layer). The magnitude of

the zeta potential is determined by the concentration of

ions in the Stern layer and/or the concentration of

ions in the diffuse layer. Zeta potentials can be

determined from streaming potentials using coarse

particles or from electrophoretic mobilities of colloids.

In the past, two practical problems of noncreeping

flow and electrode polarization have limited accurate

measurements of streaming potentials. In a streaming

potential experiment, solution is forced to flow through

a porous bed of coarse particles. A linear relationship

between the streaming potential, E, and solution flow

pressure, P, is predicted by the Smoluchowski equation

which is used to calculate zeta potentials as described

in Chapter 2. In practice a linear relationship is

usually observed. However, Chapter 2 also shows that

a nonlinear relationship exists between volumetric flow

rate and flow pressure. This suggests that noncreeping

flow exists in the pressure range used in many typical

streaming potential measurements. Since the Smoluchowski

equation assumes creeping flow, zeta potentials calculated

from streaming potential values obtained under noncreeping

flow conditions may be invalid. Chapter 2 shows why

streaming potential values obtained under commonly

encountered noncreeping solution flow conditions can

still be used to calculate zeta potential values using

the Smoluchowski equation.

The second problem commonly experienced in streaming

potential measurements is the nonzero intercept of the

streaming potential-solution flow pressure relationship.

Theoretically, when the flow pressure is zero, the

streaming potential must be zero. However, a finite

rest potential is commonly observed at zero pressure.

This problem may be due to electrode polarization.

Therefore, Chapter 2 presents a method for measuring

streaming potentials without the undesirable effects of

electrode polarization.

Calculation of surface charge densities (a) of the

solid oxide surface from potentiometric titrations using

potential determining ions is used by many investigators

to determine the variations of a with pH. In a

potentiometric titration, the concentration of hydrogen

ions before and after addition of a known amount of

titrant is recorded. The quantity of hydrogen or

hydroxide ions which cannot be accounted for in solution

is assumed by most investigators to be adsorbed onto the

solid surface. However, if dissolution of the solid

occurs, some of the ions may not adsorb to the surface,

but may be part of complex ions in solution. However,

they are not free hydrogen or hydroxide ions. Their

absence cannot be detected by pH measurement. Therefore,

absolute values of surface charge densities calculated

by most investigators are too large. For silica, surface

charge densities are negative values since the surface

is negatively charged in water above pH = 3.0. Therefore,

Chapter 3 describes the method used for aqueous silica

systems to calculate the surface charge density error due

to use of mass balance expressions which are incomplete

when hydrogen or hydroxide ions which are part of complex

ions in solution due to dissolution of the solid are


Accurate adsorption density values which are used

to calculate surface charge densities are easily

determined for oxide systems since hydrogen and hydroxide

ions are potential determining ions. However, more time

consuming and sometimes less accurate methods must be

used to determine adsorption densities of other ions on

the oxide surface. Chapter 4 presents a new method for

determining this information from electrophoresis data.

The unique aspect of this method is the use of various

solids concentrations in gathering data of zeta potential

versus apparent ion concentration in solution. From

these data, adsorption densities and equilibrium

concentrations of ions in solution after adsorption can

be determined. The Stern equation is then used to

determine if these data fit the expected form for common

adsorption isotherms.

The techniques discussed in Chapters 2 and 4 are

used to study the electrokinetic properties of an

aluminosilicate system. The system chosen was montmoril-

lonite clay, which is a major component of phosphate

slimes. The slow settling of phosphate slimes due to

fine particle size clays such as montmorillonite is a

major problem in the phosphate mining industry. Dense

coagulation of clay particles would result in faster

slime settling rates and higher final sediment densities.

The degree of coagulation of the particles is partially

controlled by their electrical double layer properties.

According to DLVO theory (1), coagulation will result

when the electrical repulsive forces between particles

are small enough to allow van der Waal's forces of

attraction to cause particle coalescence. Since zeta

potentials are a measure of the degree of these electrical

forces, Chapter 7 is devoted to finding chemical

conditions of zero zeta potential (ZZP) of the particle


To fully understand the montmorillonite system, its

two major components, silica and alumina, were studied

independently. Chapter 7 shows that montmorillonite has

two important properties. First it specifically adsorbs

aluminum ions under certain conditions. Since silica also

has this property, a detailed study of the desorption of

this ion from silica surfaces treated with chemical and

thermal agents known to alter hydration (2) is

presented in Chapter 5. This is accomplished by using

a slightly modified streaming potential technique from

that described in Chapter 2. Information obtained from

these desorption studies gives insight to the adsorption

mechanism of aluminum ions onto silica surfaces.

The second major property of montmorillonite

described in Chapter 7 is that equilibration time and pH

in water affects its capacity to specifically adsorb

aluminum ions. This appears to be a phenomenon caused by

varying degrees of hydration of the clay particles.

Another "aging effect" of this type is described in

Chapter 6 for aluminum oxide surfaces. It is possible

that hydration of clay particles is controlled in some

manner by the hydration of alumina in the clay. Therefore


Chapter 6 presents evidence of alumina surface hydration

by studying the change of the zero point of charge

with aging time in water.



The measurement of streaming potential on porous beds

is an important method for determining zeta potential (3).

The method is convenient because many materials cannot

readily be shaped as capillary tubes. A common practice

for porous beds is to use the Smoluchowski equation (4)

4T XE []

to calculate the zeta potential () from the measured

streaming potential (E), the viscosity (n), the specific

conductivity (X), the dielectric constant (E), and the

driving pressure (P).

Equation 1 was originally derived for simple

capillaries assuming laminar flow conditions. Boumans (5)

has shown that the E/P ratio in simple capillaries is

smaller under turbulent flow than under laminar flow.

A change in flow behavior with increasing pressure is

also known to occur in porous beds. The relationship

between flow rate and pressure changes from linear to

nonlinear at a certain Reynolds number. Many streaming

potential measurements on porous beds, in the past, have

been made under conditions near the linear to nonlinear

transition zone. This raises questions about the

validity of using the E/P ratio to calculate the zeta

potential when the flow condition is nonlinear. The

purpose of this investigation was to determine whether

the E/P ratio changes with increasing P, as flow changes

from linear to nonlinear. If no changes occur,

Equation 1 is valid in the nonlinear region as well as

the linear region. Only a rather narrow pressure range

was investigated since this was the range of interest in

typical streaming potential measurements.

In porous beds, a Reynolds number, Re, has been

defined (6) as

D vp 1
Re = -- (-) [2]

where D is the mean particle diameter which is equal to

400 microns for the particles used, v is the superficial

velocity found by taking the ratio of volumetric flow

rate and cross sectional area of the bed, p is the

solution density, i is the solution viscosity and E is the

bed porosity.

At Re values below around 10, flow rate and pressure

are linear (7). This is the region of creeping or Darcy

flow. At high Reynolds number, a nonlinear relationship

develops (the region of noncreeping or nonDarcy flow).

This change is a result of the formation of standing

eddies behind the particles. The Ergun equation (6)

describes the flow behavior over the range of interest:

D 2
(P P) (=) ( ) = 150 D (1-) [3]
2-7 L D (vp/) [J
vp p

where AP is the pressure drop across the bed, L is the

bed length and all other variables as defined in Equa-

tion 2.

Accurate E versus P measurements cannot be obtained

without recognizing and dealing with the experimental

problems of electrode polarization. Ball and

Fuerstenau (3) cite electrode polarization as the

probable cause for E versus P curves not passing through

the origin. Somasundaran and Agar (8) proposed that the

instantaneous change in the voltage when solution flow

is initiated is the true streaming potential. Korpi and

deBruyn (9) incorporated a recorder into their system to

aid in the measurement of the instantaneous voltage

changes during flow initiation and termination.

In the present investigation, a R-C circuit is

introduced which directly nulls the background potentials

(rest potential and electrode polarization potentials).

This greatly facilitates the measurement of the true

streaming potential, especially under conditions where

the background potential is large and varies with time.

A modified streaming potential apparatus is described

that is suitable for low pressure studies and for use on

materials that are considerably reactive. A new cell

design is introduced that is easier to pack, easier to

clean, and is less fragile than previous designs.

Materials and Methods

1. Materials

The bed materials used for this study were fused

silica* and an invert silicate glass (denoted as

"bioglass") whose preparation is noted elsewhere (10) and

whose composition is as follows: 45 wt.% Si02,

24.5 wt.% CaO, 24.5 wt.% Na20, and 6.0 wt.% P205. This

glass has certain reactive properties which makes it an

interesting material for biological implant studies (10).

The fused silica and bioglass were ground and sieved, and

the -20+45 fraction (0.0833-0.0354 cm aperture) was used

in the experiments. The fused silica was acid washed by

conventional methods used by other investigators (11).

Due to the reactivity of bioglass, no acid washing was

*Vitreosil,Thermal American Fused Quartz Co., Montville,
New Jersey.

performed. Instead, bioglass was only washed with con-

ductivity water.

The bed lengths were 4.1 cm and the cross sectional
area of the beds was 2.38 cm The bed porosity was

calculated to be 36%.

Water used to rinse the bed material and to prepare

solutions was obtained from a water deionization system*

with the following specifications: a resistivity of

1.87 x 10 ohm-cm, dissolved solids at the parts per

billion level, dissolved gases removed, and organic

removed. Since dissolved gases were removed, equilibra-

tion with air was required which caused the pH to drop to

5.5-6.0 due to absorption of CO2. This is the pH range

at which the streaming potential experiments were

performed. Sodium chloride** solutions of various

concentrations were prepared with this water and used in

the experiments.

2. Apparatus

The cell for the streaming potential system is shown

schematically in Figure 1. It consists of a thick walled

*Continental Water Conditioning Co., Inc., Gainesville,
**ACS reagent grade, Scientific Products, Ocala, Florida.

Figure 1. Streaming potential cell; a) PIMA tube;
b) threaded PMMA plugs; c) platinum leads to
electrodes consisting of perforated disks with
attached platinum mesh; d) platinum electrode;
e) solution flow entrance; f) solution flow
exit; g) porous bed; h) 0-ring.

polymethylmethacrylate (PMMA) tube with ends tapped to

receive two threaded PMMA plugs. This design allows

easier cleaning between experiments and easier packing of

the porous bed material. The platinum leads to the

electrodes* protrude completely through the PMMA plugs

and are sealed by a press-fit using a Teflon spacer. The

platinum electrode consists of a perforated disk for

easy solution flow and an attached platinum mesh for

increased electrode surface area. Because the electrodes

are able to slide through the PMMA plug, adjustments can

be made to obtain a tight packing of the bed.

The solution flow system is shown schematically in

Figure 2. The solution reservoir is a 25 liter poly-

ethylene container with a mechanical on-off flow valve.

The height of the outlet tube is varied to produce dif-

ferent hydrostatic pressures across the cell. This

design allows for E versus P measurements at pressures as

low as 1.0 cm Hg. Also, flow rates can be measured at

the outlet tube exit.

The potentials from the electrodes were measured with

an electrometer.** The output from the electrometer was

*Englehard Industries, Carteret, New Jersey.
**Keithley, Model 602.



flow valve

-- adjustable


Figure 2. Streaming potential solution flow system.

passed through an R-C circuit (details of which are dis-

cussed in the Results and Discussion section) into a strip

chart recorder.* E versus P curves were generated by

plotting the peak height value on the recorder cor-

responding to the applied pressure.

Results and Discussion

1. Circuitry for Measuring Streaming Potential

A R-C circuit shown in Figure 3 was introduced into

the system for the purpose of directly measuring only the

true streaming potential on the recorder within a 99% ac-

curacy by nulling out all background potentials. Two

different circuits can be employed by changing the switch.

When the liquid is not streaming, the switch is in

position 1. The rest potential is rapidly stored in the

capacitor, nulling the signal to the recorder. The time

constant of this circuit is 2.2 seconds and so more than

99% of the nulling occurs in 10 seconds.

After charging the capacitor with the rest potential,

the switch is turned to position 2 and flow through the

cell is initiated within two seconds. The time constant

for the new circuit is 220 seconds. Therefore, the rest

*Hewlett Packard Model 680.

10 Li

670 n Recorder
with 2 x 10I


._ output

input :

Figure 3. R-C circuitry used in streaming potential

potential stored in the capacitor does not decay more than

1% in two seconds. The potential received by the electro-

meter when flow is initiated is the sum of the rest

potential and the true streaming potential. However,

since the rest potential has already been stored in the

capacitor, the rest potential contribution of the input

voltage is nulled out and only the streaming potential is

recorded. Because of the arrangement of the resistors,

the input voltage to the recorder is 1/100 of the output

from the electrometer (which is one volt full scale).

Therefore, the recorder is kept on a 10 mV full scale


Streaming can be terminated after around five seconds

of flow. Then the switch is placed in position 1. The

capacitor again charges rapidly to the rest potential and

a new measurement can be initiated.

Using the R-C circuit, streaming potential measure-

ments were carried out on fused silica using water and

NaCl solutions of 10-5, 10 and 10- mol/L as streaming

solutions and on bioglass using water as the streaming

solution. The E versus P curves obtained are shown in

Figure 4. Within experimental error, all curves are

observed to be linear and pass through the origin.

Similar findings have been observed for a variety of

-6.0- --z

> -5.- 10 M NaCI --
o 10 M NaCI .

x -4-0- 10 M NaCI 'b

*^ Bioglass
0 -3.0- H 0
2.0 3.0 4.0

E -2.0 -

-1.0 -= \0

0 1.0 2.0 3.0 4.0

Pressure, (cm.Hg)

Figure 4. Streaming potential vs. pressure using the R-C
measuring circuit.

materials and solutions without deviations in linearity

or intersection with the origin.

Without the R-C circuit, under otherwise identical

conditions, straight lines passing through the origin

are not obtained in the E versus P curves as shown in

Figure 5. This can be attributed to the fact that rest

potential contributions were not taken into account.

2. Creeping Versus Noncreeping Flow Conditions

During the E-P measurement described above, solution

flow rates were also determined. Volumetric flow rates,

Q, versus pressure are given in Figure 6. A nonlinear

relationship between Q and P was observed.

In Equation 3, v can be converted to Q, since
Q = vfr2, where r is the radius of the bed. The Ergun

function becomes

2 1.75 2
Ap k Q + Q [4]
kP k1
1 1


k3 D 2 4
k1 P


k2 150 ( )(Tr p)
2 DP
Since = 0.36, D= 0.04 cm, L = 4.1 cm, p = 1 g/cm,

Figure 5.

Pressure (cm. Hg)

Streaming potential vs. pressure without the R-C
measuring circuit.

A IU M NOCI ..I .. ,,
10 M NaCI
8.0- -3
- 0o 10 M NaCI

-* o

o 4.0

2.0- _

0 A
0 1.0 2.0 3.0 4.0

Pressure, (cm. Hg)

Figure 6. Flow rate vs. pressure for fused silica.

p = 0.009 poise, and r = 0.87 cm, the values of kI and k2

are calculated to be 0.0040 and 51, respectively.

The data of Figure 6 fit an equation of the form

given by Equation 4 if k1 and k2 are 0.00375 and 48, re-

spectively. This is in close agreement with the re-

spective calculated values. Thus, the nonlinear behavior

follows what is expected from the earlier flow studies.

From Figures 4 and 6, it can be seen that the E/P

ratio remains unchanged even though solution flow

becomes nonlinear. The streaming potential is

proportional to the streaming current. The streaming

current is given by the integral over space of the product

of the charge density and velocity (projected in the net

flow direction) at every point (12), and so this integral

must be proportional to pressure. However, the charge

density distribution does not change with pressure. This

strongly suggests that the velocity of the liquid at every

point near the surface increases in direct proportion

with the pressure. This conclusion would be in agreement

with the current views that streamline flow near the

particle surfaces remains even after standing eddies



In the porous beds that were studied, noncreeping

flow did not appreciably affect streaming potential-flow

pressure relationships at the pressures studied. This

suggests that streaming potential experiments in the

past which have been performed under similar noncreeping

flow conditions would have had the same E/P ratios as

would have been measured under creeping flow conditions.

For those cases, the calculated zeta potentials would have

been as valid under noncreeping flow as under creeping

flow conditions.

Electrode polarization effects were nulled out by

using a R-C circuit which allowed only the true streaming

potential to be recorded. This method greatly facili-

tated the measurement of streaming potentials of reactive

materials such as bioglass.

Combined results of the electrode polarization and

noncreeping flow studies showed that electrode polari-

zation and not noncreeping flow was the reason for E

versus P curves not passing through the origin as found

by previous investigators. This was demonstrated by the

fact that E versus P curves were linear and passed through

the origin under conditions of noncreeping flow as long

as electrode polarization effects were accounted for,

vis., by using the described R-C circuitry.



Calculations of surface charge densities are made

from adsorption density values of potential determining

ions on oxide surfaces (13-16). Potentiometric titration

of the aqueous-oxide system with potential determining

ions are used to determine adsorption density values as

a function of pH. For aqueous-silica systems, different

negative values of surface charge density at any given

pH are found depending on the phase of silica used in the

titration. Using precipitated silica, Tadros and

Lyklema (17) found that absolute values of surface charge

densities were an order of magnitude higher than those of

Bolt (14), who studied amorphous silica. Tadros and

Lyklema suggest that a gel structure exists on the

precipitated silica surface and that extension of

surface and counter ion charge inside the pores of this

gel structure causes higher charge densities. However,

Yates and Healy (18) suggest that precipitated silica does

not have a gel structure in the surface but that it

contains an incompletely condensed layer of polysilicic

acid. They also suggest that surface roughness con-

tributes to high inner layer capacities creating high

surface charges due to slight interpenetration of

potential determining and adsorbed counter ions.

An alternative hypothesis for anomalously high

surface charge densities can be investigated by examining

the mass balance equation which describes an aqueous-

oxide system in a titration experiment. The real

adsorption density (Freal) is defined as the excess moles

per unit of surface area of hydrogen over hydroxide ions

on the solid surface. The correct mass balance equation

for this value is

[(H-OH)initial+AH-(H-OH) solution-C]V
real A [5]

where (H-OH)initial is the total excess moles of hydrogen

over hydroxide species per unit volume of solution in the

system before titrant addition, AH is the moles of titrant

per unit volume of solution added to the initial system,

(H-OH)solution is the excess moles of free hydrogen over

hydroxide ions per unit volume in solution, V is the

solution volume, A is the total surface area of oxide

powder used in the titration and C is the excess moles

per unit volume of hydrogen over hydroxide ions which are

part of complex species in solution due to dissolution

of the solid. Essentially Equation 5 says that those

hydrogen or hydroxide ions after titrant addition which

cannot be accounted for as free ions in solution or in

solution as part of complex species must be adsorbed to

the solid surface. Only relative values of Freal can

be determined if titrations are performed in one elec-

trolyte concentration since the quantity (H-OH) initial

is unknown. However, by performing titrations at various

electrolyte concentrations, several adsorption density

versus pH curves can be obtained which intersect at one

point. In the absence of specific adsorption of counter

ions to the oxide surface, rreal is zero at this point.

Therefore, from Equation 5, (H-OH)initial can be

determined and absolute values of Freal can be calculated.

Many workers choose to neglect the C term in Equa-

tion 5 since they assume that the concentration of complex

species which form in solution is negligible compared to

the concentration of free hydrogen or hydroxide ions in

solution. However, this may not always be true especially

under conditions of a thermodynamically unstable solid

phase in solutions of high pH. If "C" is neglected, then

[(H-OH) initia+AH-(H-OH) olution]V
apparent A

where apparent is the adsorption density calculated from

potentiometric titration data by most investigators. If

Equation 6 is subtracted from Equation 5, then

Ar = r r = [7]
real apparent A [7]

C will be the amount of error incurred in the adsorption

density value when these hydrogen or hydroxide ions

assumed to be free in solution actually compose part of

complex solution species. Therefore, this chapter serves

to determine C in terms of experimental parameters so

that real apparent and, therefore, surface charge

density error real Capparent, can be calculated for the

silica water system.


1. Formation of Complex Species

It is possible to calculate the equilibrium solu-

bility of silica from thermodynamic information. All

that is required is knowledge of the free energy of

formation of the solid and soluble species at the

appropriate temperature. Table I shows the standard free

energy of formation for five solid silica phases, a sodium

silicate, a neutral aqueous silica species and two aqueous

complex ionic species. Also shown are the values for

Table I. Standard Free Energy of Formation
Values at 2980K (Kcal/mole)

N Species AG (Kcal/mole)
1 Si02(quartz) -192.4

2 Si02(cristobalite) -192.1

3 Si02(trydymite) -191.9

4 Si02(vitreous) -190.9

5 Si02(hydrated) -187.8

H2Si03(aq) -242.0

HSiO^ ) -228.36

SiO(aq) -212.0

Na2SiO3(c) -341.0

Na(aq) -62.6

H20(aq) -56.69

H 0

OHaq) -37.6

Sources: Pourbaix, ref. 14 and
Dickerson, Gray and Haight,
ref. 20.

H20 Na H and OH (19,20) Equa-
2(aq)' a(aq)' (aq)' (aq) 20) Equa-
tions are written describing the formation of the neutral

soluble silica species (H2Si03). Since five different

silica phases are considered, five equations yielding

five equilibrium constants (K[N] where N = 1-5) are

obtained. The reactions and values for their equilibrium

constants are shown in Table II.

The reactions for the formation of the various ions

from H2SiO3 species are shown in Equations 8-10. Their

equilibrium constants are calculated from data in Table I.

H2SiO3 = HSiO3 + H KA = 1.01 x 10-10 [8]

H2SiO3 = SiO + 2H KB = 9.92 x 1023 [9]

2Na+ + H SiO3 t Na2SiO3 + 2H+, KC = 2.81 x 10-27 [10]

Activity coefficients for the various ions were

incorporated into the final calculations of surface

charge density error. The values for these coefficients

were obtained from calculations by Klotz (21) using the

Debye-Hueckel theory for strong electrolytes. Table III

shows the values for ions used in the calculations at

several ionic strengths. The values for HSiO3 and Si03

were not directly available from Klotz (21), but were

chosen for ions of similar size as HSiO and SiO3.

Table II. Reactions Depicting Formation of the
Neutral Soluble Silicate Species,
H2Si03, and Their Equilibrium
Constants (K)

Reaction K[N (where N=1-5 from Table I)
Si2(quartz)+H2H2SiO3 K 1 = [H2Si03] = 6.31 x 10-6

Si02(crist.)+H20=H2Si03 K 2 = [H2Si03] = 1.05 x 10-5

Si02(trid.) +H20=H2SiO3 K 3 = [H2SiO3] = 1.47 x 10-5

Si02(vit.) +H20=H2SiO3 K 4 = [H2Si03] = 7.49 x 10-5

Si2(hyd.) +H20=H2SiO3 K 5 = [H2Si03] = 1.47 x 10-2

Table III.

Activity Coefficients for Ionic Species
at Various Solution Concentrations

Solution Concentration (moles/liter)
Species 0.1 0.01 0.001
HSiO3 0.750 0.898 0.964

SiO 0.360 0.660 0.867

H+ 0.830 0.914 0.967

Na+ 0.770 0.901 0.964

Source: Klotz, ref. 21

The excess of hydrogen of hydroxide species which

is part complex species is directly related to the

concentration of each complex species in solution. From

Equations 8-10 it can be seen that for every HSiO3 ion

formed, the excess of hydrogen over hydroxide is -1.

Accordingly, for Si3 the excess is -2 and for Na2SiO3

it is -2. Therefore,

C = -([HSiO3] + 2[Si03] + 2[Na2Si03]) [11]

2. Development of the Working Equation for Ao

By writing the equilibrium constant expressions for

Equations 8-10 and substituting K[N] from Table II for

[H2Si03], the equilibrium concentration of each complex

species can be written in terms of experimental

parameters. That is,

[HSiO3] + K[N] [12]

S KB K[N] [13]
[SiO2 [13]
2 +2

KC K[N] YNa+ [Na ]
[Na2Si3] = 2 2 [14]
YH+ [H ]

Now, C in Equation 7 can be expressed in terms of experi-

mental parameters:

V KA K[N] + 2KB K[N]

S HSiO YH+ [H+] Si0 YH+ [H]
3 3

+2 2
2KC K[N] [Na +]2 +
+ 2 2 [15]
YH+ [H ]

Adsorption densities can be converted to surface charge

densities by the relation

a = 106FT [16]

where F is Faraday's constant (9.65 x 104 coul/mole) and

106 is the number of microcoulombs per coulomb. There-


6 V
Ao = 10 F A(C) [17]

where C is the term in parentheses in Equation

Calculations of Ao were made for 10-3, 102 and 10-

M/L NaC1 solutions for the five different silica species

for a pH range of 7-10 in increments of 0.1 pH units.

Since the calculations are numerous, a computer program

was written and is appended. The program is written

specifically for the silica system. However, as the

appendix shows, the program can be generalized for any

material as long as the free energy of formation data for

the various reacting species are known.


Figure 7 shows the calculated values for Aa as a
03 -2 -1
function of pH for 10-3, 102, 10 M/L NaC1 solutions

for vitreous silica using Bolt's (14) experimental

conditions of surface area and solution volume. It can

be seen that as the ionic strength of the electrolyte

solution increases, the absolute value of AC increases

for any given pH value. This trend is consistent with

the experimental surface charge densities measured by


Comparison of the calculated values with Bolt's

absolute values in Table I of his paper (14) for amorphous

silica shows that little error is incurred in neglecting

complex species formation in solution. This is not an

unexpected result since Bolt used a high surface area

to volume ratio in his experiments. Therefore, Bolt's

data represents real as well as Oapparent since Ao is

negligible for the pH range studied.

Using Bolt's data it is possible to calculate at

what value of total surface area significant error would

have occurred for any given pH value. Figure 8 shows

Aa as a function of total surface area for pH = 7, 8, 9,
5 2
10. For pH 9 and total surface area of 10 cm 10% error
7 2
would have occurred. Bolt used 5.4 x 10 cm total





Figure 7.

Surface charge density error (Ao) vs. pH in
10-2, 10-3 M/L NaCI solution for vitreous
silica using Bolt's experimental conditions.

o -2
_- 10 -

0 10

b -3

10 09


105 106 107 10

Total surface area (cm2)

Figure 8. Surface charge density error (AF) vs. total
surface area for amorphous silica at pH = 7,
8, 9 and 10.

surface area in his studies. Therefore, even at pH = 10,

less than 1% error should be expected based on his

experimental values of surface charge density. This

calculation quantitatively demonstrates the importance of

the experimental condition of surface area in a

titration experiment if the formation of complex

species is to be neglected. To achieve greatest accuracy

a high surface area to volume ratio must be used.

Tadros and Lyklema (17) studied the surface charge

density as a function of pH for precipitated silica.

The absolute values range from 15 to 200 micro coul/cm2

in the pH range 7-10. Figure 9 shows that values of

Aa calculated in the present work reach 30 micro coul/cm2

which suggests that at pH = 10, 10-15% error is incurred

by Tadros and Lyklema by neglecting the complex species
6 2
formation. Tadros and Lyklema used 8 x 10 cm surface

area in their experiments (20g of 40 m2/g per 100 cc).

Figure 10 shows a plot of Aa versus total surface area.

At a surface area of 8 x 106 for pH = 10, 10% error should

be expected based on their value of a. Tadros and
8 2
Lyklema therefore should have used 10 cm of total sur-

face area to achieve 1% error or less at pH = 10 if they

wanted to neglect the complex species formation.

From data such as presented in Figures 8 and 10,

the amount of error in surface charge density incurred as



F <

Figure 9.

Surface charge density error (Aa) vs. pH in 10
10- 10-3 M7L NaCI solutions for hydrated
silica using Tadros's and Lyklema's experi-
mental conditions.

6 7
10 10

Total surface area, (cm2)

Figure 10.

Surface charge density error (Ao) vs. total
surface area for precipitated silica at
pH = 7, 8, 9 and 10.




a function of pH for any given surface area used can be

determined. The error becomes more significant as the

pH increases and surface area decreases. This should be

expected since soluble complex species form in greater

quantities at higher pH's due to increased ionization of

the neutral soluble silica species.

According to the calculations, based on the data

of Tadros and Lyklema, significant error should be

expected by using 2g of 40 m2 /g material in 100 cc of

solution. However, these authors claim no significant

difference in the amount of OH/g SiO2 adsorbed when 2, 10

or 20 grams of solid were used in the same volume of

solution. To observe this, they must have used much

less volume than 100 cc, which was the solution

volume V used in the calculation in the present work.

However, no indication of solids concentration was given.


These calculations demonstrate the importance of

knowing the amount of each phase present in an aqueous

silica system. In an experiment using the same total

surface area for all five phases less error in surface

charge density will occur for quartz than any other solid

silica phase. The order of least to most significant

error is the same as the most to least thermodynamically

stable. The calculations show that in a mixed phase

solid of 99.99% quartz and 0.01% hydrated silica, the

amount of soluble complex species formed from the hydrated

phase will be the same as the amount formed from the


One occurrence of a mixed phase solid has been

addressed by van Lier et al. (22). In their quartz

solubility studies, they confirmed the existence of a

disturbed layer on ground quartz particles by dissolution
studies at high pH values. The thickness of 300 A which

they calculated from their results agreed well with

Gibb et al. (23). Van Lier et al. found that abnormally

high solubilities in water and high pH solutions of

quartz are obtained if the disturbed layer is not re-

moved. However, removal of this layer yielded normal

solubility data. This suggests that the layer is not

quartz but probably a more thermodynamically unstable


Van Lier et al. were not able to identify the

disturbed layer phase. Using their experimental condi-

tions and the free energy of formation values in Table I

of the present work, it is possible to calculate the

theoretical thickness of the layer assuming the silica

phase is each of the five solid phases considered.

To make the thickness calculations, dissolution

reactions must be written. These are

Si02 + OH = HSi03 [18]

SiO2 + 20H = SiO0 + H20 [19]

Using quartz as an example, the standard free energy of

reaction for Equation 18 is 1640 cal/mole and for Equa-

tion 19 it is -1100 cal/mole assuming the data in

Table I are correct for quartz phase. Neglecting activity

coefficient, the equilibrium constant expression for each

equation is

-AGo [HSiO3]
K18 = exp( RT = Sio2 OH- [20]

-AGo [H20][Si0=]
19 = exp(fT-) = 2o]' [21]
[SiO2] [OH]

K18 is calculated to be 0.061 and K19 is 6.53. Since K18

and Kl9 are known, the concentrations of HSiO3 and Si0O

can be calculated using pH = 12.30 from experimental

conditions of van Lier et al. These values are [HSiO3] =

1.2 x 10-3 M/L and [Si0O] = 2.6 x 10-3 M/L. The equi-
valent concentrations of Si02 are 9.4 x 10 M/L and

2.05 x 10-3 M/L. Therefore the total concentration of

quartz (Si02) present in solution after dissolution is

2.99 x 10-3 M/L or 0.18 g/L. Since van Lier et al. used

0.030 liters, the final weight of powder present in the

system after dissolution was 0.3854g-0.0054g = 0.3800

grams, where 0.3854 is the initial weight of powder used

by van Lier et al. before dissolution.

To calculate the thickness of the disturbed layer,

two assumptions are needed. First, the particles are

assumed to be spheres. Secondly, the number of particles

in the system remains constant. The total initial volume

Vi of particles in the system is related to the initial

particle radius ri by

V. = Ni 4/3rr3 [22]
1 1 i

where Ni is the initial number of particles. The same

equation can be written relating the final total volume

Vf and final particle radius rf after dissolution. That


Vf = Nf 4/3rr3 [23]

where N is the final number of particles after dis-

solution. Since

P = V [24]


W. = N.p 4/3Br3 [25]
1 1 i


W = Nf p4/3irr3 [26]

where p is the density of the solid phase, Wi and Wf are

the initial and final total weight of powder respectively.

Since N. = Nf, then

1 f
W. Wf
-3 3 [27]
p4/3Trr~ p4/3irf3
A value for r. = 1.5 x 10 cm is chosen since it is mid-

range of the particle size range used by van Lier et al.


Wf 1/3
r = ( ) ri [28]
For quartz the value of rf is 1.493 x 10 The

thickness t of the disturbed layer would be t = ri rf =
-7 0
7 x 10 cm = 70 A.

Similar calculations for the four other silica

species were made. Table IV shows values of t and the

total concentration of soluble SiO2 at equilibrium for each

species. The values of t in Table IV represent the amount

of the particle which could be dissolved away if the

layer was each of the phases studied. They also represent

the thickness which the disturbed layer must have for the

solution to become saturated with respect to each phase

considered. On this basis, the disturbed layer cannot

Table IV. Calculated Values for Disturbed Layer
Thickness (t) and Soluble Si02 at
Equilibrium Using Data from van Lier
et al. (ref. 22)

2 o
Silica phase SiO2 x 10 (moles/liter) t (A)

Quartz 0.30 70

Cristobalite 0.50 119

Tridymite 0.70 167

Vitreous 3.88 968


*Calculations show total dissolution

have a quartz, cristobalite or tridymite structure since

the values of the thickness calculated are lower than
300 A. That is, the solution becomes saturated with

respect to quartz, cristobalite or tridymite when 70, 119
or 169 A of thickness are dissolved away. However, the

disturbed layer could have either an amorphous or hydrated

structure since saturation cannot occur in a system

containing silica particles with a disturbed layer thick-
ness of 300 A.


It has been shown that care must be taken in

adsorption density measurements using potentiometric

titrations to use high surface area powder to solution

volume ratio systems if one is to neglect complex species

formation. Guidelines for such experimental conditions

have been presented by using experimental data presently

available in the literature for the basis of the calcula-

tions. It is important to note that more experimental

error will be incurred by not considering complex species

formation for less stable silica phases especially at

higher pH values where the ionization of the neutral

soluble silica species occurs in greater quantity.

These calculations have also shown that as little

as 0.01% hydrated phase present in quartz forms equivalent

concentration of complex species as all of the quartz

itself. Hence the importance of careful consideration

of the solid phase under study cannot be over emphasized.

The thickness of the disturbed layer on quartz has

been calculated using data on the solubility of quartz

and thermodynamic data for solid silica species. The

calculations show that the layer cannot be quartz,

cristobalite or tridymite.



Different types of adsorption information can be

obtained for fine particle-aqueous systems depending on

the type of experimental method used. The method com-

monly used to calculate adsorption densities of oxide

systems from potentiometric titration using potential

determining ions to determine the zero point of charge of

the surface was described in Chapter 3. To determine

adsorption densities accurately for ions other than

potential determining ions, other techniques such as

solution depletion (as measured by atomic adsorption or

emission) must be used. These measurements require

synthesis of calibration curves which essentially doubles

the amount of work involved in determining an entire

adsorption isotherm.

The unique feature in the method presented in this

chapter for determining adsorption isotherms from electro-

phoresis measurements is the use of various solids

contents. By obtaining zeta potential behavior as a

function of apparent ion concentrations in solutions for

different solids concentrations, adsorption densities

and equilibrium solution concentration of the ions can be

calculated. The montmorillonite clay-aluminum ion solu-

tion system is used to obtain these experimental data.

These data are then plotted on the basis of their

relationship in the Stern equation (24) to determine if

they fit an analytical form of this equation.

Materials and Methods

Montmorillonite 1613 was used as received from the

supplier.* A stock suspension of 2% solids was prepared

by ultrasonic dispersion of the clay in water and was

equilibrated for one week at ambient temperature. After

equilibration, 0.5, 0.05, 0.005 and 0.0025% suspensions

were prepared by diluting the stock suspensions with

water described in Chapter 2.
-1 -2
Aluminum chloride solutions of 2 x 10 2 x 10

2 x 103, 2 x 10, 2 x 105 and 2 x 106 M/L were

prepared from one molar stock solution. Fifty ml of each

solution was adjusted to pH = 4.0. This was mixed with

50 ml of clay suspension previously adjusted to pH = 4.0.

This mixture constituted the electrophoresis solution.

*Georgia Kaolin, Inc., Elizabeth, N.J.

This procedure was carried out for each of the solids

suspensions listed above. Therefore, electrokinetic data

as a function of aluminum ion concentration for various

solids content could be obtained.

Electrokinetic data were generated using microelectro-

phoresis using the Riddick cell. The methods of

determining electrophoretic mobilities and calculating

zeta potentials are discussed elsewhere (25).

All chemicals used in this study were certified

reagent grade materials.

Results and Discussion

1. Calculation of C and F

Figure 11 shows the results of zeta potential as a

function of aluminum ion concentration C for various

solids contents. This concentration is the number of

moles of aluminum ions added to the system divided by

the solution volume. It is not the actual concentration

of aluminum ions in solution since adsorption occurs onto

the montmorillonite particle surfaces. Therefore, it is

the apparent concentration if no adsorption takes place.

It can be seen that zeta potential-concentration curves

for various solids contents differ except for very dilute

suspensions. This indicates that, at higher solids

Weight percent solids

* 1.0
o 0.25
o 0.025


pH =4.0

[AICI36H20] moles/liter
20 /liter

Figure 11.

Zeta potential vs. concentration of AlC13*6H20
for 0.00125, 0.0025, 0.025, 0.25 and 1.0%
weight percent montmorillonite.


oI 0-


- -


contents, the equilibrium concentration, C, of aluminum

ions in solution is significantly lower due to adsorption

onto the clay particle surfaces.

A mass balance equation can be written which de-

scribes the system at equilibrium as follows:

C = C + A(W) [29]

where Co is the apparent concentration of aluminum ions in

solution if no adsorption occurs, C is the concentration

of aluminum ions in solution after adsorption to the

particle surface, F is the adsorption density of aluminum

ions on the clay surface, VL is the solution volume, Ws

is the weight of solids in the system, and A is the

specific surface area of the clay. Ws/VL can be replaced

by the weight percent by the following argument: weight

percent (W/o) of solid is defined as s x 100 where W =
W +W L
s L

weight liquid and W = weight of solid. If WL > Ws then

W7/ = -x 100. Since WL = PL VL where pL = density of

liquid, when W1% s x 100. Since p = 1.00/cc, the

right side of Equation 29 must be multiplied by a factor

1000 cc/liter. Therefore

C = C + 10PL AF(W%) [30]

and the units of each term in Equation 30 are moles/liter.

If enough solid is present in the system, most of the

ions in solution will adsorb to the surface, and

C << 10PLArFW; therefore Equation 30 becomes

Co = 10pLAF(W/o) [31]

Taking the log of both sides of Equation 31 yields

log CO = log(lOpAr) + log(W%) [32]

Table V lists the values of C0 for zeta potentials =

15, 10, 5, 0, -5, -10 mV.

A plot of concentration C as a function of weight

percent using data at zeta potential equal to zero mV

from Figure 11 is shown in Figure 12. For high solid

contents the experimental points approach a straight line

whose slope is 1.0. At low solids contents, the points

approach a line of zero slope. The C value corresponding

to this line is C. Rearrangement of Equation 30 and

taking logarithm yields

log(C -C) = log(lOpLAr) + log(W7o) [33]

Therefore if Equation 32 correctly describes the system

under concentration,then plots of C -C versus W% should

yield straight lines for each zeta potential. As

Figure 13 shows, straight lines are obtained.

Table V. Values of Co for Various Zeta Potentials
and Weight Percent Solids

C(mV) 0.00125 0.0025 0.025 0.25 1.0
-10 3.5x10-5 4.5x10-5 1.1x10-4 7 x10-4 2.9x10-3

-5 8.5x10-5 9.3x10-5 2.1xl04 1.3x10-3 5.4xl0-3

0 2.2x10 2.3x104 4.5x104 2.3x10-3 10-2

+5 5 xl0-4 5.1x104 8.8x104 5.4xl0-3 1.8x10-2

+10 1.2xl0-3 1.2x103 1.8x103 8.5x103 3 x102

+15 2.7x10-3 2.8x10-3 3.6xl03 1.3x102 5 xl0-2

E pH=4.0

E -3


0.001 0.01 0.1 10

Weight percent solids

Figure 12. Apparent aluminum ion concentration Co vs.
weight percent montmorillonite for zeta
potential = 0 at pH = 4.0.

o0 I I 1 IIII I I I I l ll T111

-Zeta potential, (mV)
A 5

-3 a 10
10 -
o 15


pH 4.0

0.002 0.01 0.1 1.0

Weight percent solids

Figure 13. Concentration of aluminum Co minus the
equilibrium concentration (C) of aluminum in
solution vs. weight percent montmorillonite
for zeta potential = +15, +10, +5, 0, -5,
-10 mV at pH = 4.0.

From Equation 33, at 1% solids, C-C = 1pLAF.

Therefore, the adsorption density, r, can be determined if

the specific surface area of the solid is known. For

montmorillonite, a generally accepted theoretical specific

surface area of 800 m2/g was used in the calculations.

Table VI lists the values of C, F and log(T) for each

zeta potential considered. A plot of c and F vs. C is

shown in Figure 14.

2. Application of Experimental Data to the Stern

Stern (26) has derived an adsorption isotherm

relating the surface charge/cm2 (o) to the Stern potential

(4s). The Stern equation is
NN ve

1+Fr exp( kT-)

where 01 is the surface charge/cm2 associated with the

adsorbed ionic layer, N1 is the number of adsorption sites
per cm at surface, v is the zalence of the ion adsorbed,

e is the charge of the electron, N is Avogadro's number,

M is the molecular weight of the solvent, n is the number
of ions per cm far from the surface, s is the potential

at the Stern plane, C is the specific adsorption potential

of the adsorbed ion, k is Boltzmann's constant and T is

al MC
temperature. Since r and n = then
ve 100'

Table VI. Values of C, r and log (C/F) for Various
Zeta Potential Values

(mV) -10 -5 0 +5 +10 +15

C 3.5x 8x 2x 5x 1.2x 2.7x
(moles/1) 10-5 10-5 10-4 10-4 10-3 10-3

F 3.63x 6.75x 1.25x 2.25x 3.75x 6.25x
(moles/cm) 10-11 10-11 10-10 10-10 10-10 10-10

log 2.92 3.07 3.20 3.35 3.51 3.64


-9.00 E
-10 _

E -9.40

-9- --9.80
0 C
"S 10- ""\

S- -10.20 o

Zeta potential vs c o
a Log adsorption density vs c
I5 I04 2i3 i2
10 10 10 10

Concentration c, (m/,)

Figure 14. Zeta potential and log adsorption density (F)
vs. equilibrium concentration (C) of aluminum
in solution for montmorillonite at pH = 4.0.

r = [35]
1000 ve q) s
+MC exp( T )

Multiplying the numerator and denominator of the right

MC veis +
side of Equation 35 by TT exp-( kT ) yields

MC veqs +
nMj N1 exp-( kT- )
S MC s[36]
y-- exp-( kT )+i
es M -
If y = kT K = 100 exp(kT), and ys = C, then

N1 C K exp-(vy)
r = [37]
C K exp-(vy)+l [37]

Taking the inverse of both sides and rearranging Equa-

tion 37 yields

N -r
C( -) exp yv [38]

Since the total number of adsorption sites (NL) is much

greater than occupied adsorption sites, N1 >> F. There-


C exp(vy) [39]

Taking the natural log of both sides and converting to

common log,

log(C 1 v
log(-) = log N1K + 2 [40]
F N K' 2_.y3[40]

The experimental data (C and F) fit the Stern equa-

tion, since Figure 15 shows that a plot of log(y) vs. y

yields a straight line. According to Equation 40 the

slope should be 1.3 if v = +3. However, the straight line

obtained has a slope of 0.8.

A plausible explanation for the discrepancy between

the theoretically determined and experimentally obtained

slope must come from examining the factors involved in

v, the valence of the adsorbed ion. Errors in any other

term lead only to shifts in the position of the line but

no change in slope. In calculating the theoretical

slope, v was assumed to be +3 for the aluminic ion.

However, for the slope to decrease to the experimentally

determined value, v must be lower than three. That is,

the effective valence of the adsorbed ion must be smaller

than 3. This is possible only if aluminum ions are in-

volved in ion exchange for sodium ions (+1 valence, making

the effective vAl, +2) or calcium ions (+2 valence, making

the effective vAl, +1). Therefore the vAl would have a

value between 1 and 2. The effective valence calculated
from Figure 15 is 7-(3) = 1.8.

Energetically, the work required in bringing an

aluminum ion from solution to the surface is partially

supplied by the "negative" work involved in releasing a

Olk 3.2



2.4 1 1 1J11
-0.6 -0.4 -0.2 0 0.2 0.4 0.6


Figure 15. Log of the ratio of C and F vs. zeta potential
for montmorillonite for the experimental data.

sodium or calcium ion from the surface to the solution.

Therefore, v in the Stern equation is related not just

to the valence of the adsorbed ion, but also to the net

work involved in bringing the ion to the surface during

ion exchange reactions.

The intercept of the straight line with the y-axis

in Figure 15 yields the specific adsorption potential if

N1 is known. However, since the slope can vary due to

ion exchange processes, the specific adsorption potential

will also vary with the type and degree of ion exchange


The data treated in this chapter cover only one

pH value. Future work involving the entire pH range

should give more insight into ion exchange reactions of

aluminum for other ions in the clay. Also, the influence

on the Stern slope by aluminum ion hydrolysis as the pH

increases can be investigated. However, this type of

study would have to be performed using a material other

than montmorillonite so that the contribution by ion

exchange to Stern slope characteristics is minimal.


A new method for determining adsorption isotherms

from electrokinetic data of zeta potential vs. ion


concentration for various solids content has been

described and tested experimentally. It was found that

the experimental data fit an analytical form of the Stern

equation. However, the experimentally determined slope

was lower than the theoretical Stern slope. This was due

to the lowered effective valence of the aluminum ions

involved in ion exchange reactions.



Considerable interest has been generated in the past

ten years in the coagulation and flocculation behavior of

silica sols. It has been shown that of utmost importance

in this behavior is the hydration state of the surface.

It is now generally accepted that the surfaces of silica

sols are covered by silanol groups with a density of 5 OH/
100 A (27), whereas a quartz surface is relatively inert,

being covered by siloxane groups (28,29). However, some

silanol groups have been shown to exist on quartz,

especially on ground or milled material (22,30).

Lange (31) found that the amount of water hydrogen bonded

to precipitated silicas was about 38% of the total pos-

sible adsorption capacity. This meant that 62% of the

silanol groups were hydrogen bonded to each other. Using

an explanation involving ion exchange of hydrogen ions

for metallic ions, Allen and Matijevic (32) found that

the hydrophilic nature of silica decreased in the presence

of simple electrolytes.

Recently, Tschapek and Sanchez (33) studied the

amount of NaC1 required to coagulate different silica

sols as a function of suspension pH. It is clear from
their results that the absence of isolated silanol groups

lowered the amount of NaC1 required for coagulation at low

pH's. Also, the absence of hydrogen bonded silanol groups

did not lower the amount of NaC1 required for coagulation.

They concluded that isolated silanol groups were primarily

responsible for the stability of silica suspensions

calcined between temperatures 110C and 8000C. Con-

sequently, hydrogen bonding between silica particles

could not have been the mechanism of their coagulation.

Removal of silanol groups from the surface by heat

treatment also removes the mechanism of surface charge

development in solution. Their removal is highly

irreversible as shown by Rubio and Kitchener (34).

Removal of the charge development mechanism consequently

decreases the repulsive forces between particles to a

point which may allow close interaction of the primary

particles and ultimately their coagulation. This

coagulation would be facilitated by hydrophobic bonding.

If this is true then the results of Tschapek and

Sanchez (33) for untreated silica sol and silica sol

calcined at 800C suggest that only those silanol groups

which are not hydrogen bonded to each other can contribute

to the formation of adsorption sites for ionic species in

solution. Adsorption site formation occurs by ionization

of the silanol group. This could be the reason why

isolated silanol groups stabilized their suspensions,

i.e., the ionization of the isolated silanol group

created repulsion between the particles. Only in this

way could these two sols have identical coagulation


The previously mentioned mechanism for sol stability

demonstrated how thermal treatment rendering the silica

surface hydrophobic created the possibility of hydro-

phobic bonding between silica particles. Rubio and

Goldfarb (35) showed that chemical treatment can also

create hydrophobic surfaces. Their results suggest that

if a number of quaternary ammonium cations attach to the

surface by means of a nitrogen atom the surface will

become hydrophobic. The greater the surface coverage

the more hydrophobic the surface becomes. They state that

in this way, aggregation of the particles may be effected

by hydrophobic bonding. This type of mechanism is sup-

ported by the fact that quaternary ammonium salts have

been used as flotation collectors for quartz

particles (36).

Adsorption characteristics of species from solution

will be affected by surface hydration states of silica

since surface hydroxyls act as primary adsorption sites

for polar molecules (37). Conventional methods used to

study these phenomena require the use of finely divided

materials. In this chapter, a technique is introduced

which allows investigations of silica material which has

a size range of 300-800 microns. The results obtained

using this technique are interpreted in the same manner

as those obtained when fine material is studied.

Adsorption of aluminum ions from solutions by silica

surfaces is of particular interest to glass corrosion

studies. It has been shown that aluminum ions decrease

corrosion of silica when they are present either in the

glass itself (38) or in the corrosion medium in contact

with the glass (39). Weyl (40) proposed that a require-

ment for corrosion inhibition for the latter case is that

aluminum ions adsorb to the surface and not form a more

soluble compound than the glass itself. Lyon (41) found

that rinsing of a container glass with aluminum ion

solutions and subsequently rinsing with water did not

inhibit alkali extraction from the glass which suggests

that aluminum ion adsorption may not always be permanent.

Aluminum ion adsorption to a glass surface may

theoretically affect either ion exchange or network

breakdown. In his glass solubility work, Iler

investigated the effects of adsorbed aluminum ions on the

dissolution of fused silica. Using fused silica

eliminated the possibility of the ion exchange mechanism

since no alkali were present in the glass. Therefore only

network breakdown was involved. Iler's results showed

that as the amount of aluminum ions in solution increased,

the dissolution of silica decreased for a given period of

time. However, Iler exposed his samples to solutions

containing a 1:1 molar ratio of aluminum to citrate. He

presumed that negatively charged aluminosilicate sites

were created. The inhibition mechanism proposed was

repel of hydroxide ions from the negatively charged sites.

The present work will show that this presumption is

incorrect. Also since Lyon's work on container glass

suggests that aluminum ion adsorption is not permanent,

this chapter considers certain conditions under which

aluminum ion adsorption may be reversible or irreversible.

The conditions considered are those which are known to

alter the hydration state of the silica surface (42).


Silica used for this investigation was commercially

obtained.* Just before use in the experiments, 10 grams

*Vitreosil, Thermal American Fused Quartz Co., Montville,
New Jersey.

of a -20+45 mesh fraction (833-350 micron) was boiled in

100 ml concentrated hydrochloric acid until no discolora-

tion of the acid was observed. The sample was then rinsed

with conductivity water until all chloride ions were


All solutions used in these investigations were

prepared from Certified ACS reagent grade chemicals.

Water used to prepare the solutions was obtained from a

deionization system previously described in Chapter 2.


Electrokinetic theory can be applied to study

adsorption of electrolytes near the solid surface (12).

Theory predicts the existence of an electrical double

layer near the surface consisting of ions present in the

solution. The double layer contains an immobile (Stern

Layer) and mobile portion (Guoy-Chapman diffuse layer).

The electrical potential at the plane separating these

two layers is the zeta potential.

Usually zeta potential values are calculated from

the Smoluchowski equation (Equation 1, Chapter 2). As

shown in Chapter 2, to obtain zeta potential information

streaming potential experiments are performed. However,

modified streaming potential techniques can be used to

study not only adsorption but also desorption of aluminum

ions from the solid surface. Studying desorption

characteristics yields information about the reversibility

of adsorption. This is accomplished by using an apparatus

in Figure 16. Desorption studies were accomplished by

first exposing the particles in the cell (e) to a 10-3 M
NaC1 solution containing 10 M aluminum ions. Figure 17

shows that a positive zeta potential value resulted at

this aluminum ion concentration indicating specific

adsorption of aluminum ions to the surface. The solution

in reservoirs a and b was changed to a 103 M NaCI

solution containing no aluminum ions. This solution was

streamed through the cell at constant pressure [made

possible by maintaining a constant hydrostatic head in

(b)] for an extended period of time. Solution flowed

continually during the desorption experiment except when

data were taken. To obtain these data, the solution flow

valve (d) was turned off and on causing a deflection in

the electrometer (f) indicating the streaming potential

at that particular time. This operation required only a

few seconds resulting in only momentary interruption of

the desorption phenomenon. The signal from the electro-

meter was recorded on a strip chart recorder (g) and the

time was noted. A special circuitry discussed in

Figure 16. Streaming potential apparatus modified to
maintain constant flow pressure; a) secondary
reservoir; b) primary reservoir; c) pump;
d) solution flow valve; e) cell; f) electro-
meter; g) recorder; h) solution head height.




80- ACIC'6H O


-6 -5 -4 -3
10 10 10 10

Concentration, (moles/tr

Figure 17. Zeta potential vs. concentration of sodium
citrate and aluminum chloride.

Chapter 2 was used to eliminate undesirable effects of

electrode polarization.
By using 10-3 M NaCI as a supporting electrolyte for

adsorption of aluminum ions, no significant change in

E/P due to varying solution conductivity as predicted by

Equation 1 should occur during the desorption experiment.

Also since desorption studies were performed under

conditions of constant pressure, no change in the

streaming potential due to varying pressures as predicted

by Equation 1 should be expected. Therefore, during the

desorption studies, the change in E/P should only be due

to change in the zeta potential (C) due to removal of

aluminum ions from the silica surface.

Desorption characteristics of several types of

silica surfaces were studied by this technique. One set

of samples remained untreated. A second set of samples

was heat treated at 8000C for 8 hrs. in vacuum.* A third

set was treated with 1 M NaOH solution for 24 hrs. at

220C. Finally, a combined thermal and chemical treatment

of the glass particles was performed using the agents

described above.

*Centorr hot press vacuum chamber, 10-5 Torr.

Results and Discussion

Figure 18 shows that major differences exist in the

desorption behavior of aluminum ions from silica after

various surface treatments. It can be seen that most of

the aluminum ions desorb from the untreated silica.

However, some aluminum ions remain on the surface even

after 60 minutes of streaming as shown by the fact that

the E/P value does not reach the E/P value for non-
aluminated silica streamed only with 10 M NaC1 solution.

Chemical treatment of the silica particles in one molar

NaOH solution caused most of the aluminum ions to desorb

after approximately 80 minutes of streaming.

For heat treated silica desorption of aluminum ions

was much less than for untreated or chemically treated

silica. Only a small amount of aluminum ions appeared to

reversibly absorb to the surface.

Chemical treatment of the heat treated samples with

one molar NaOH solution reestablished the reversible

adsorption capacity of the silica. It can be seen that

almost complete desorption of aluminum ions occurred since

the E/P value approaches the value for nonaluminated
silica streamed in 10-3 M NaCl solutions.

It should be noted that data for nonaluminated
silica samples streamed in 10-3 M NaC1 solution were in-

dependent of the various surface treatments.

cn 3'"\ Heat treated and
I base treated
E Non-aluminated SiO2
E I -in 13 M NaCI
- 0
wa- -



0 10 30 50 70 90

Stream time, (min.)
Figure 18. Streaming potential-pressure ratio vs. stream
time for untreated, heat treated, base treated,
heat treated and base treated and non-aluminated
fused silica.

The major difference between untreated and heat

treated silica samples is that the former is dominated

by adjacent silanol groups while the latter contains

isolated silanol groups (2). The data show that aluminum

ions reversibly adsorb to a silica surface containing

adjacent silanol groups and irreversibly adsorb to a

surface containing isolated silanol groups. However,

since initial E/P values for the two samples were the

same as seen in Figure 18 at stream time equal to zero,

there appears to be little difference in the amount of

aluminum initially adsorbed. If only those silanol

groups which are not hydrogen bonded to each other can

become adsorption sites, as suggested earlier, then only

those which are not hydrogen bonded to each other can

be true specific adsorption sites for aluminum ions. When

aluminum ions adsorb, the double layer characteristics

change as indicated by changes in zeta potential. If

zeta potential characteristics change in the same way (as

indicated at stream time = 0 for untreated and heat

treated silica) for these two surfaces having different

hydration states, then the aluminum ion adsorption site

must be present in the same amount on both surfaces. This

trend is supported by the results of Tschapek and

Sanchez (33) which showed identical coagulation

characteristics for untreated silica sols and silica sols

calcined at 8000C.

An atomistic view of the adsorption-desorption

process may occur as follows. The pH of the solution

containing 10- M aluminum ions was 4.2-4.5. In this

range the aluminum ions are mostly in the aluminic

(Al ) form with some probability that some Al8(OH)20

species exist. This complex species has been postulated

and shown to exist by several workers (43,44). When

desorption begins using a 10- M NaC1 solution whose

pH = 5.5-6.0, more formation of the complex aluminum

ion species near the surface is favored. The contribu-

tion of the double layer characteristics of this species

will be greater than Al since they are quatrivalently

charged. Hence, their contribution to the double layer

characteristics will be the same for both types of

surfaces at time = 0. However, if these ions are to

remain on the surface during the desorption experiment.

they must hydrogen bond to the surface. They can hydrogen

bond only to the surface containing isolated silanol

groups since the surface containing adjacent silanol

groups cannot hydrogen bond with solution species. There-

fore, during the desorption experiment those aluminum

species which cannot hydrogen bond are streamed away and

the E/P value with time decreases significantly.

Similar interpretations of adsorption from solution

of poly(ethylene oxide) (PEO), a neutral species, onto

silica has been reported by Rubio and Kitchener (34).

Their results clearly show that isolated silanol groups

provide the best adsorption sites for PEO. Adsorption

occurs by hydrogen bonding of the ether oxygen with the

hydrogen of the silanol group. They found that complete

dehydroxylation by heat treatment rendered the surface

incapable of adsorbing PEO. Also, the hydrated surface

could not adsorb PEO since its hydrogen bonding capacity

was exhausted in its effort to hydrogen bond with another

silanol group on the surface.

Iler (39) was not able to show whether or not

citrate ions subsequently adsorbed to negatively charged

aluminosilicate sites. In fact, he assumed they probably

did not since adsorption between two negatively charged

species should not be expected. However, since

adsorption of aluminum ions onto silica creates positively

charged sites as shown by the results in Figure 16,

citrate ion adsorption may be possible.

Figure 19 shows that the zeta potential is reversed

back to a negative value when aluminated silica is

exposed to increased concentrations of citrate ion

solutions. Also shown is the curve for citrate on






- ^Supporting electrolyte
a -3
80 @10 M NaCI
N 80-
10 M NaCI, 10 M AICI 6H20



-6 -5 -4 -3
10 10 10 10

Concentration No citrate 2H20, ( mole/liter)

Figure 19. Zeta potential vs. concentration of sodium
citrate for fused silica in supporting electro-
lytes of 10-3 M/L NaC1 and 10-3 M/L NaC1,
10-4 M/L A1C13-6H20.

nonaluminated silica. The negative zeta potential values
for concentrations greater than 3 x 10 M citrate for the

aluminated silica curve could be due to one of two pos-

sible effects. Either aluminum ions were removed from

the silica surface to form aluminum citrate complexes in

solution or citrate ions adsorbed to positively charged

aluminosilicate sites. Study of Figure 19 alone does not

help to eliminate either of the possible mechanisms.

However, if the desorption characteristics of the

system are studied, one mechanism can be eliminated.

Therefore, untreated silica samples were exposed to a

10- M NaC1 solution containing 10 M citrate ions and

10- M aluminum ions. Then a 10- M NaC1 solution void

of citrate and aluminum ions was streamed through the

cell. Figure 20 shows the results obtained. If aluminum

ions were removed from the surface to complex with citrate

ions in solution, then the E/P value should decrease to
E/P value for nonaluminated sample streamed with 10 M

NaC1 solution. However, the E/P increased at early

times indicating the removal of a negatively charged

specie from the surface. The only negatively charged

specie present in the system capable of specific

adsorption was citrate ions. These results indicate that

citrate ions first adsorbed to aluminosilicate sites and

oase TreaTea
E -

E 0


SiO ,10 M NaCI


0 10 30 50 70 90

Stream time, (min.)

Figure 20. Streaming potential-pressure ratio vs. stream
time for aluminated fused silica for untreated,
heat treated, base treated and heat treated and
base treated surfaces.

then desorbed upon streaming the particles with 10-3 M

NaCI solution. At longer streaming times a slight

decrease in E/P occurred indicating removal of aluminum

ions from the silica surface.


A method for studying the desorption behavior of

large particles has been presented. This method yielded

results which can be interpreted in terms of the hydration

state of the surface of the silica particles.

Aluminum ions reversibly adsorbed to surfaces con-

taining adjacent silanol groups and they irreversibly

adsorbed to surfaces containing isolated silanol groups.

The mechanism of irreversible adsorption proposed for

aluminum ions was hydrogen bonding of the complex aluminum

species to the isolated silanol surface groups.

Contrary to Iler's results (39) aluminum ions adsorb

to fused silica forming positively charged rather than

negatively charged sites. Also, aluminum ion adsorption

was shown to activate the silica surface for specific

adsorption of citrate ions.



Parks (45) has compiled, a considerable amount of

information from many different authors concerning the ZPC

of several oxides and hydroxides including those of

aluminum. Knowledge of the ZPC allows prediction of the

sign of the surface charge at any pH of a solution in

contact with the oxide. Also, knowledge of the ZPC is

important for understanding coagulation processes since

minimum surface charge (minimum interparticle repulsive

forces) is necessary for maximum coagulation as predicted

by DLVO theory.

The ZPC for aluminum oxides ranges from pH = 6.0 to

9.5. The scatter in the ZPC is indicative of varying

experimental conditions and solid phase used. Excluding

adsorption of impurities, Parks (45) and others (46) have

attributed most of the scatter to varying degrees of

surface hydration. Since the ZPC of the oxide depends

on the degree of surface hydration, aging phenomena of

alumina can be studied by observing changes in the ZPC

with aging time, and, in this way, surface hydration can

be studied. Robinson et al. (47), O'Connor et al. (46),

Johansen etal. (48,49), and Schuylenborgh et al.(50-53)

all agree that treatment which leads to bulk or surface

dehydration results in a more acid ZPC than for oxides

which are hydrated. Those treatments which dehydrate the

surface (e.g. heat) lower the ZPC whereas treatments

which increase hydration of the surface (e.g. grinding)

increase the ZPC.

Most of the ZPC information on aluminum oxide com-

piled by Parks was obtained on either naturally occurring

minerals or on synthetic materials prepared in the

author's laboratories. However, little ZPC information

can be found for commercially prepared aluminas.

Information of this nature could be beneficial to both

suppliers and users of commercial aluminas particularly

if the powders are subjected to aqueous environments

during processing. Parks (45) has shown that very small

levels of adsorbed impurities such as phosphate and

sulfate as well as certain organic greatly affect the

ZPC of the oxide. If the impurities were undesirable,

they could easily be detected by measuring the ZPC of

the oxide. Appropriate steps could then be taken to

eliminate the impurities during processing.