Vous êtes sur la page 1sur 790



- _.-.. - -






. ",; ......


k:;..: . ,~. :



j ttl'Ol


'~ ~:.. '




. ;" ..-,













o '




.. 'I.




. ~~ UP"!.'!)

. ' .



: ":', erizaiiOll:

"So.'0(.. -"~aiio"
, ',' .... Ii J~'-'

loe ,ennstl.V'


. .

. . . .



.': /
.~~ ::1

. ...:~;

' . ,' . '


.-' "

:.~ C.~ollO.IdaD(I

t..,:;:.. -:
.f . :~:~;.L .




; t::


, ii, ,~.r,:;.







.- ...

~.-.- .-.~:.~

Silica, the major component-Olthe
earth's solid surface and. the constituent of ordinary sand,'becomes involved
at some point in a great many phasesof .' .
modern technology and science. It is an
essential material in many, if not all,
. forms of life. Its role in human disease,
aging, and health is Just beginning to be
explored. Here is a comprehensive account of the basic chemistry invOWed in
a wide range of research and development activities, as well as a wealth of information on production and production control.
Beginning with the solubility of different.
. forms of silica and the factors that influence dissolution and deposition, the solution chemistry of silica Is Introduced.
The author also compares and recommends analytical methOds. The digest
of all currently available information
provides a solid background as to the
nature of soluble silicates and particularly the mechanism of polymerization
of sHicic acid and formation of colloid.
:.'For the first time, the mechanism by
which silica sots, powders and gels -are
formed and their properties controlled
Is clearly described. Next, the many
types and uses of commercial concentrated sols, gels, and u'trafine powders
are examined, fotl~ by a discussion
of the biochemical properties and many
applications of the surface chemistry of
silica. The finat chapter draws together
all aspects of the occurrence and importance of silica in different life forms.
Those engaged in research, development, and production in the many diverse fields and Industries in which silica plays a vital role-such as chemistry, biology, medicine, agriculture, metallurgy, and mining-will find THE
CHEMISTRY OF SILICA an indispensable reference.
. -... .'"


Solubility, Polymerization, Colloid and

Surface Properties, and Biochemistry












A Wiley-Interscience Publication
New York Chichester Brisbane Toronto


To my wife, Mary, with gratitude for her never-ending

patience during the seemiaaly interminable
writing of this book


This book was at first intended to be an updated second edition of my
earlier book, The Chemistry of Silica and Silicates (Cornell University
Press, 1955). It necessarily covers much of the same subject matter, but
with 2500 new references to consider, it had to be reorganized and expanded
to such an extent that it constitutes an almost entirely new work.
The purpose of the book is to present a complete and coherent account of
the chemistry of amorphous silica, including soluble silica and silicate
precursors of soluble silica, polymerization to polysilicic acids, colloidal sols
and gels, and the surface chemistry of silica. In discussing practical applications of sols and gels, emphasis is placed on the chemistry involved. The.last
chapter on silica in living organisms is especially important in view of the
growing recognition that silica is present in many biological systems and
can function as an essential trace element.
Since publication of my earlier book in 1955, the literature on colloidal
metal silicates, including minerals, and on silicic esters has grown
enormously. Consequently these areas had to be omitted.
The title, The Chemistry of Silica, may be misleadingly broad but is
offset by a more definitive subtitle, "Solubility, Polymerization, Colloid and
Surface Properties, and Biochemistry."
It is remarkable that silica, the major component of the earth's solid surface, has never become a separate branch of study or instruction. Science
students graudate with little or no knowledge of its properties or chemistry.
Yet sooner or later, in such diverse fields as industrial chemistry,
electronics, agriculture, mining, metallurgy, petroleum, power development,
and even biochemistry and medicine, problems arise involving this common
element oxide. This book is written not only for those already engaged in
these areas, who may find it a useful guide to the literature, but also for






those in other fields who need specific information not otherwise easily




Wilmington, De/aware
November /978



l .

........ ~












I am indebted to the Cornell University Press for permission to include
some of my earlier book* with the following credit:
Reprinted from Ralph K. lIer: THE COLLOID CHEMISTRY OF
1955 by Cornell University.
Used by permission of the publisher. Cornell University Press.
I am also grateful to John Wiley and Sons for permission to include in
Chapter 5 portions of my monograph on "Colloidal Silica" in Col/Did and
Surface Science. VoL 6. 1973. edited by Egon Matijevic.
This work would have been impossible without the generosity of E. I.
duPont de Nemours & Co. in making available to me. as a retiree. the
facilities of the Lavoisier Library at the duPont Experimental Station.
It is impossible to mention all those who have kindly reviewed drafts of
portions of the manuscript and given invaluable advice. My friend and
fellow scientist. Dr. Paul C. Yates has been very helpful with sound
technical counsel.
To the late Mildred Syvertsen. who played an indispensible role in all my
earlier publications. I remain grateful for help in collecting references and
typing much of the present manuscript. The assistance of Patricia Cullen in
final typing. of Joseph A. Pankowski. Jr. in preparing illustrations and of
Jennifer J. Stiles in assembling indexes. is sincerely appreciated.

Now out of print.



Previous Books and Reviews of Silica Chemistry

Selection of References

. 1 The Occurrence, Dissolution, and Deposition of Silica

The Silica-Water System

Thermodynamics of the System
Relating Particle Size and Composition
Energy Change with Changing Particle Size and Composition
Soluble Silica-Monosilicic Acid
Volatility in Steam
Soluble Silica in Nature
Phases of Silica
Anhydrous Crystalline Silicas
Relation between density
.and . refractive index
. Hydrated Crystalline Silicas
Amorphous Silicas
Microscopic sheet. ribbon, and fiberlike forms. Common
amorphous forms. Hydrated amorphous silica. Biogenic sili
The Solubility of Silica _ .
Solubility of Quartz at Ordinary Temperature
Cleaning the surface
Solubility of Quartz under Hydrothermal Conditions













,; ,

.. ",.,._ ,_",-",,-.._

..__ ,. __ ._ ..__ .. _



Solubility of Cristobalite and Tridymite

Solubility of Other Crystalline Forms of Silica
Adsorbed Silica on Crystalline Silica
Solubility of Amorphous Silica
Establishment of solubility equilibrium, Effect of heating,
Solubility in water: pH 0-8. Possible solubility minimum at
pH 7. Solubility in nitric acid. Solubility in NaCIO. solutions,
Effect of electrolytes. Solubility under hydrothermal conditions
Solubility of Hydrated Amorphous Silica
Apparent High Solubility at High pH
Calculation of solubility and dissociation constant
Effect of Particle Size on Solubility in Water
Interfacial Energy
Effect of Impurities on Solubility
Effect of Organic Compounds on Solubility
Catechol and Related Compounds
Polyhydroxy Organic Compounds
Organic Bases
Living Tissues
Solubility in Alcohols .
Methanol. Higher alcohols
Solubility in Molten Salts"
Rate of Dissolution of Silica'
Effect of pH on Rate
Relation Between Rate of Dissolution and Particle Size
Rate of Dissolution of Very Small Particles
Rate of solution as particle dissolves
Rate of Dissolution of Particles of Different Sizes
Dissolution of Crushed Powders
Neutral Solutions-Effect of Salts
Retardants of Dissolution
Rate of Dissolution in Presence of Catechol
Rate of Dissolution in Aqueous HF
. Comparative Rates of Dissolution
Removal and Deposition of Silica from Water








Removal of Silica from Water

Precipitation mechanisms, Nucleation of quartz, Adsorption and
precipitation by hydrous oxides, Removal by ion exchange
Deposition of Silica from Water
Rate of deposition of monomeric silica, Silicification of biogenic
materials, Rate of deposition of colloidal silica













Methods of Analysis
Atomic Absorption
Chemical Methods
Methods Involving Silicomolybdic Acid
The beta silicomolybdate method, A recom mended procedure,
Interfering substances, Molybdenum blue method, For
biological sample.
Methods of Concentrating Silica for Analysis
Depolym erizing Colloidal Silica before Analysis
Standard Silica Solutions
Miscellaneous Colorimetric Methods
Detection of Colloidal Silica on Surfaces
Rapid Titration of Total Silica as Fluosilicate
Titration as the Silicomolybdic Acid



Water-Soluble Silicates


Sodium and Potassium Silicates

Commercial Solutions
Soluble Crystalline Sodium and Potassium Silicates
Properties of Solutions
Fields of Use
The Nature of Silicate Solutions
Physical Studies :
Effects of diluting silicate solutions. Effect of alkali metal
salts and other coagulants
Conversion to Suicic Acids
Reaction with molybdic acid. Conversion to esters of silicic
acids. Conversion to trirnethylsilyl derivatives of silicic acid










~ .~



i:! :
:; ~

~~ ~

r .



, ....

: ,

! .






Silicates with Coordination Numbers Four and Six

Solutions of Polysilicates
Sodium Polysilicate .
Potassium Polysilicate
Lithium Silicates
Lithium Polysilicates
Uses for Lithium Silicates and Polysilicates
Organic Base Silicates
Mixed Organic Base-Alkali Metal Base Silicates
Other Organic Base Silicates
Complex Metal Ion Silicates
Organic Chelates of Silicon
Catechol Derivatives
Humic Acids
Other Organic Cornpounds :
Hydrated Crystalline Alkali Metal Polysilicates
Silicates Convertible to Crystalline Forms of (H 2Si 2Os),c
Precipitation of Insoluble Silicates
Soluble Silicate Glasses
Peroxy Silicates


Polymerization of Silica


General Theory of Polymerization

Overall Effect of pH on Gelling
Monosilicic Acid
Dissolving silica, Hydrolysis of monomeric silicon compounds,
Dissolving monomeric silicates in acid
Characteristics of Silicic Acid
Diffusion constant. Ionization constants. Increase in ionization
constant with polymerization, Isoelectric point. Point of zero
charge, Stability of monomeric silica
Reactions of Monosilicic Acid
Phosphoric andboric acids, Sulfuric acid, Iron and uranium,
Chromium, Aluminum, Divalent cations
Characterization of Silicic Acids



180 ,

189 .

195 \.......--














Reaction with Molybdic Acid

Alpha and beta silicic acids, Measurement of reaction rates,
Reaction rate constants, Composition of molybdic acid
reagents, Other observations
Separation of Silicic-Acids
Particle Size and Surface Area by Titration
Correction for soluble silica
Coagulation with Gelatin-Salt
Mechanism of Condensation and Hydrolysis
Catalytic Effect of HF
Polymerization: pH 2-7
Formation ofOligomers Oligomers as Particles
Nucleation Theory
Particle Growth in Acidic Solution
Depolyrnerization in Acidic Solution
Polymerization by Aggregation-Gel Formation
Molecular versus Particle Chains
Mechanism of Interparticle Bonding
Formation of Chains of Particles and Networks
Partial Coalescence of Particles in Chains
Development of Microgel and Viscosity
Isolating "gel phase" or "rnicrogel", Effect of electrolytes and
coagulants. Gel density and structure, Increase in viscosity
Formation of Larger Particles by Coacervation
Polymerization above pH 7
Spontaneous Growth of Particles
Final Size of Particles versus Temperature
Viscosity of Sols before Aggregation Begins
Viscosity of Sols of Very Small Particles at Low pH
. Decrease in Viscosity on Conversion of Microgel to Sol
Thermal Effects
Energy of activation. Heat of polymerization
Summaries of Investigations
Investigations at Low pH
Iler; Alexander, Heston, and Iler: Schwarz and Knauff:
Bechtold; Goto; Okkerse: Audsley and Avcston: Weitz, Franck,










... .

_ .......,-~




and Giller; Bechtold, Vest, and Plambeck; Acker: H~ebbel

. and Wieker
Investigations Through the Neutral pH Range
Merrill and Spencer; Ashley and Innes; Baumann; Coudurier,
Baudru, and Donnet; Marsh, Klein, and Vermeulen; Ginsberg
and Sheidina
Investigations Above pH 7
Greenberg and Sinclair; Greenberg; Goto; Tarutani; Iler: IIer
and Sears; Richardson and Waddams; Makrides et al.
Polysilicic Acids
Preparation of Polysilicic Acid
Hydrogen-Bonded Complexes with Polar Organic Compounds
Method of comparing hydrogen-bonding activity, Structure
versus activity, Liquid hydrogen bonded complexescoacervates, Complex of silicic acid with amine salt,
Interaction of silicic acid with phosphoric acid ester
Combinations with Organic Polymers
Prevention of hydrogen bonding by negative charge on silica,
Cationic organic compounds
Miscellaneous Interactions with Organic Materials
Interaction with proteins-e-tanning. Esterification of
polysilicic acid
Activated Silica Sols-Water Treatment
Reaction of Polysilicic Acid with Metal Cations





fI ..


~ .





Colloidai Silica~Concentrated Sols


Definition of Colloidal Silica and Historical Development

Growth and Stabilization of Discrete Particles
Increasing Particle Size by Adding ..Active" Silica
Methods of Making Particles Under 10 nm in Size
Stabilization Against Particle Growth
Stabilization Against Aggregation
.. , Stabilization by ionic charge, Addition of salt to lower
viscosity, Sterk stabilization .
Porous Particles


328.. . . :...












'- ..



Elongated Particles
Particles with Non-Siliceous Cores
Methods of Making Sols
Neutralizing Soluble Silicates With Acids
Ion Exchange
Peptizing Gels
Hydrolysis of Silicon Compounds
Dissolution of Elemental Silicon
Dispersion of Pyrogenic Silica
Purification. Concentration, Preservatives
Ion Exchange
Dialysis and Electrodialysis
Washing Procedures
Evaporation of water, Centrifugation, Ultrafiltration,
Characterizing Sols
Chemical Analysis
Measuring pH. Electrolyte concentration
Particle Characteristics
Particle size. Specific surface area
Ionic Charge on Particles
Nature of ionic charge. Counterions and double layer
Aggregation of Particles
Effect of pH. Effect of particle size and concentration.
Electrolytes and organic liquids. Temperature. Theory of
strength of gels
Mechanism. Coagulation by electrolytes. Monovalent cations
as bridging agents. Coagulation by divalent metal ions.
Coagulation by polyvalent cations-basic metal salts. Effect
of silica concentration and other factors. Effect of
particle' size. Partly dehydrated surface


33 I





--;.. ...



. i







\! :



! .


! '
i ,

i :


! .



-, :









"j .

lJ _

i :


t ...



Flocculation with cationic surfactants, Flocculation with
organic polymers
Silica spheres by coacervation
Aggregation into Ordered Structures-Precious Opal
Opal structure, Other ordered aggregates, Formation of
uniform inorganic particles, Synthesis of opal
Adsorption of Silica Particles on Surfaces
Sols of Silica Particles with Modified Surfaces
Negatively Charged Surfaces
Aluminosilicate ions, Other anions
Positively Charged Particles
Polyvalent metal oxide coatings, Polyvalent organic cations
Organic Modified Surfaces-Organosols
Organic ions, Esteri fication, Silylation
Commercial Colloidal Silicas
Uses of Colloidal Silicas
Making Catalysts, Gels, Adsorbents
Inorganic Binder, Stiffener
Molded refractory bodies, Binders for fibers, Refractory
coatings, Molds for casting metals
Frictionizing Effects
Fibers, Paper, Steel rails, Other surfaces
Antisoiling Surfaces
Hydrophilizing Surfaces
Modifying Adhesion
Increasing adhesion, Decreasing adhesion
Coating Compositions
Coatings on ships: tanks
Reinforcing organic polymers
Polishing Agent for Silicon Wafers
Surfactant Effects
Dispersing effects; Antifoaming effects
Modifying Viscosity-e-Gelling .
Miscellaneous Optical Effects. Color, Photography
Use in Biological Research-. Density Gradient






, -..

436 <..





Source of Chemically Reactive Silica

Soluble silicates, Silica bodies, GI3~\ compositions, Forming
solid silicates-cements, Other rea..' \ions and uses
Colloidal Silicates'







Types of Gels
Types of Powders
Physical Characterization ?f Gels and P'lwders
Ultimate Particle Size



5 Silica Gels and Powders



Characterizing Pores by Adsorption h'Jtherms






Particle size and packing, Loss of SUrface area by particle





Electron micrographs, Specific sur(:II,;C area, Low angle X-ray

Aggregate Size-Powder Particles, Gd Granules
Pore Characterization





Pore volume, Pore size and size dislt'ibution, Miscellaneous

effects in micropores,
Nature of Silica Surface
Aggregate Strength-Interparticle B'IIIIJing



Electron micrographs, Partial diss,'lltlion method, Mechanical

strength of the aggregate
Silica pels
Sources of Silica Gel
From soluble silicates andminernb,From colloidal silica,
From hydrolysis of silicon com pou lids
Factors Controlling' Gel Characterist it:;
Size of pri~'~~ypa~iicles::'-pHefft.''', Wet gel strength.,
Particle size and packing density ill dried gels, Increased
porosity with 'removable fillers
Forming and Shaping Gel Particles
Wet Gel Treatments
Gel reinforcement




. :iii"'::' ..:

-;fl '

" " _..._ - -_._ -... -

I~ ~ .~ ~ . . . . ~ "


~_.l . _

. -'~.-"--, ....... - - - -.. -

" ' - " - -~-_... _. __.. - ,---

















;. I'


Drying and Shrinkage-Xerogels

Drying from low surface tension liquids
Drying without Shrinkage-Aerogels
Hydrothermal Treatments
Liquid or vapor phase
Heating in Air, Vacuum
Sintering uniform structures, Alkali metals, impurities,
Special Gel Structures
Submicroporous gels-impervious silica, Porous glass,
Specific adsorbents
Precipitated Silica Powders
Silica Precipitated from Sodium Silicate Solution
Silica coagulated with sodium ions, Coagulation by adding
sodium or ammonium salts, Coagulation with calcium,
polyvalent metal ions, Coagulation with organic materials
Silica Precipitated from Fluoride Solution
Silica Precipitated from Organic Liquids
Silica Precipitated from Colloidal Silica Sols
Silica Precipitated from Vapor: Pyrogenic Silica
, "vaporized Si02, Oxidation of SiO vapor, Oxidation and
hydrolysis of SiCI. vapor, Oxidation and hydrolysis of silicon
esters vapors, Hydrolysis of SiF. vapor
Naturally Occurring Silica Powders
Microcrystalline Hydrated Silicas
Hydrophobic-Organophilic Silica Powders '
Adsorbed Organic Cations
Adsorbed Polyvalent Metal Cations with Organic Anions
Surface Esterification
Organosilicon Coatings
Organic Polymer Coatings
Silica Gels with Ion-Exchange Surfaces
Inorganic Ion-~xchang'e .Sites
Organic-Linked Ion-Exchange Sites
Commercial Silica Gels and Powders
Uses of Silica Gels and Powders








576 .






Reinforcing and Other Effects in Organic Solids

In rubber, In silicone elastomers, In various organic polymers
Reducing Adhesion
Increasing AdhesionIncreasing Viscosity,Thixotrophy
Mechanism, Lubricating grease, Paints, coatings, inks,
. Parmaceuticals and cosmetics, Miscellaneous compositions
Optical Effects-Flatting
Surfactant Effects
Stabilizing emulsions, Hydrophilic surface, Antifoam agent
Hydrophobing Effects
"Dry" water
Aerogels, Base for mitochondria, Spillover
Reactive Silica
Cloud Seeding
Chromatographic Column Packings
. References









6 The Surface Chemistry of Silica




Reviews and Summaries

Nature of the Silica Surface
Structure of the Underlying Silica
Definition of Surface
The Hydroxylated Surface
. State of water at th-.7 hydroxylated surface, Electrical
. conductivity ofthe surface, Distinguishing adsorbed water
from silanol groups, Internal hydroxyl groups and trapped
water, Hydroxyl~groups per square nanometer, Theoretical
. concentration of surface hydroxyl groups
Dehydration and Rehydration
Surface Energies
Heat of Wetting Silica Surface
Physical Adsorption of Non-Ionic Low Molecular Weight





".. . . r_.



. :'i;.
,~ s




'jj::1,; .











Adsorption of Vapors
Effect of dehydroxylation on adsorption
Adsorption from Solution-Nonionic
Nonaqueous solutions, Aqueous solutions, nonionic, hydrogen
Ionization and Surface Charge
The Hydroxylated Surface
The Dehydroxylated Surface
Nature of the Anionic Charge Sites
The "Site-Binding" theory
Forces Involved in Adsorption of Ions
Univalent cations: Metals and lower amines, Alkaline earth
metals and magnesium cations, Polyvalent metal cations
(Table 6.3)
Nonionic Reactions of the Silica Surface (Table 6.4)
Hydrophilic Coatings on Silica
Hydrophobic Silica Surface
Organic Cations and Bases
Hydrophobing effects, Cationic dyes, Aluminosilicate surfaces
Hydrocarbon Groups Attached through Polyvalent Metals
Surface Esters with Alcohols
Surface coverage, Reaction conditions, Methyl esterified silica,
Reaction of alcohols with dehydroxylated surface, Reaction of
hydrocarbons with dehydroxylated surface, Esterification in
micropores, Substituted alcohols, Hydrolysis of ester groups
Organic Groups attached by C-Si Bonds
Adsorption on Hydrophobic Surfaces
Adsorption ofwater, Adsorption of inert gases
Adsorption of Organic Polymers on the Silica Surface
From Aqueous Solution
Polyethylene oxide, Polyvinyl alcohol, Cationic polymers,
Proteins, Adsorption of polymers on dehydroxylated silicas,
Effect of salts
From Nonaqueous Solvents
Deposition of Multilayers of Charged Polyions and Particles
The Surface of Alumina-Silica






........;.... _-




Active Sites, Free Radicals, Active Oxygen, Ozone





7 Silica in Biology






Origin of Life
Earliest Life Forms
Biological Disintegration of Rocks
Association with Primitive Organisms
Fungi and Lichens
Algae and Diatoms
Gastropods, Sea Cucumbers, Limpets
Nature of Silica Deposits in Plants
Strengthening Plant Parts
Equisetum, Bamboo, Grasses, Spiny plants, Job's Tears, Palms,
Mechanism ofAbsorption, Movement, and Deposition of Silica 747
Relation of Soluble Silica to Soil Fertility
Beneficial and Protective Effects of Silica
Fish, Amphibians, Reptiles, Birds
Mammals: Man
Essential Role of Silica in Mammals
Toxicity of Silica
Silica in Biochemical Combinations
Combination with polysaccharides, Combination with proteins.
Denaturation of proteins, coagulation of blood, Combination
with specific compounds, enzymes, Combination in
phosphorus compounds: Nucleic acids, DNA, RNA,
Mutations, atherosclerosis, cancer

i. 1











._t1t!'~A:.::9~n2"W~WCU"'~I_ aa,..,,"~-...,'_.

_ - - ' -


Silicosis-Pneum oconiosis- Fibrinogenesis

Mechanism of silicosis, Amorphous versus crystalline silica
and particle size, Susceptibility to silicosis, An unusual
compound in silicotic tissues, Solubility theory, Silica
antagonists to prevent silicosis
Asbestosis-M icroaciculosis
Beneficial Effects of Silica
Silicon Metabolism
Silica Gel as a Culture Medium
Organosilicon Compounds
Analytical Problems




Author Index


Subject Index






















Numerous outlines or surveys of the chemistry of silica. especially the water-silica

system. have appeared over the past half-century. Properties of silica have been
described by Sosman (la) and more recently by BrUchner (Ib), The state of
knowledge of soluble silicic acid and colloidal silica was summarized in 1937 by
Fricke and HUttig (2). and the general aspects of "silicic science" were reviewed by
Hauser in 1955 (3). At the same time. Iler exhaustively covered the rapidly developing theory and practice of the colloid chemistry of silica and silicates. including new
surface chemistry and the role of silica in biology (4). Following an initial book on
the physical chemistry of silicates in 1954 (5). Eitel authored a monumental series of
six volumes on silicate science from 1964 to 1975 (6). The colloidal aspects of silica
are summarized in Volume I and brought up to date in Volume IV. A review of the
solubility of silica in water. the nature of soluble silica." and its equilibrium with
polymer species was published by Stober (7). Similar reviews. each with a particular
emphasis. were written by Greenberg (8). Wittman (9). Coyle (10). Maher (I I). and
Kolthoff and Elving (12).
. Alexander has given an entertaining personal account of his researches in silica
chemistry (13). More recent summaries have been prepared by Kukolev (14) and



The major references used are those in which most variables have been deli ned. Thus
when pH has not been recorded in an aqueous system. or the source or characteristics of the silica used in experiments are not well deli ned. less attention is given,
There has been enormous duplication of experimental work. particularly in countries where earlier work was not recognized because of lack of availability to the
literature or language difficulties. In such cases original work has been emphasized.
and secondary work is cited only as conlirmation.
Patents are cited when the technical information is not otherwise available in the
literature. Where possible. United Stales patents arc cited because of general
availability throughout the vor ld upon request. But where equivalent U.S. patents
could not be located. the patents in other countries arc cited.



'7; . :;'

'The term "silanol" is used for any OH group attached to silicon and will be understood to include what has otherwise been referred to as "siloxanol" and "silicol"
groups. The term "polymerization," strictly speaking, means a linking together of
monomer units to form a polymer of the same composition, but, in the silica system
the monomer, Si(OH)., condenses to form polymer that ultimately has the composition (SiO z)II' However, the term "polymerization" has been so widely used for the
formation of condensation polymers that it will also be so used here.
"Silica" is used as a short convenient designation for "silicon dioxide" in all its
crystalline, amorphous, and hydrated or hydroxylated forms. This word does not
occur in many other languages, which generally use the term "silicon dioxide."
However, the latter implies only the composition SiO z In analysis, the term "silica"
indicates only that the silicon content is given in terms of weight' of SiO z regardless
of the form in which it is actually present.







iI ,


la. R. B. Sosrnan, The Phases of Silica, Rutgers University Press. New Brunswick, N.J .


,", ,

lb. R. Brilchner, "Properties and Structure of Vitreous Silica:' J. Non-Cryst, Solids. 5,

123, 177 (1911).





2. R. Fricke and G. F. HUttig. Handbuch der Allgemeine Chemle, Vol. 9. Hydroxides and
Oxyhydrates, Akademische Verlag. Leipzig, 1937, p. 146.


3. E. A. Hauser, Silicic Science, Van Nostrand. Princeton. N.J., 1955.


4. R. K. Iler, The Colloid Chemistry of Silica and Silicates. Cornell University Press.
Ithaca, N.Y. 1955.



1 I. P. K. Maher. Kirk-Othrner Encyclopedia of Chemical Technology. 2nd ed.. Vol. 18.

Wiley. New York. 1969. p. 61.
12. I. M. Kolthoff, P. J. Elving, and E. B. Sandell. "Analytical Chemistry of Silicon."
Treatise on Analytical Chemistry. Part II. Vol. 2. Wiley-Interscience, New York. 1962.
, p. 120.
13. G. B. Alexander. Silica and Me. Doubleday. Garden City. New York. 1967.







":{ f.

10. T. D. Coyle. Kirk-Othmer Encyclopedia of Chemical Technology. 2nd ed . Vol. 18.

Wiley. New York. 1969. p. 46.


7. W. Stober~ Kolloid Z. 147, 131 (1956).

,9. A': Wittman. Oesterr. Chem, Z . 62, 245. (1961).


6. W. E~tel. Silicates Science. Vols. I-VI. Academic. New York. 1964-1975.


8. S: (Greenberg.J. Chem, Educ. 36, 218 (1959).




5. W. Eitel. The Physical Chemistry of the Silicates. University of Chicago Press. Chicago.
III.. 1954.


14. G. V. Kukolev, Chemistry of Silicon and Physical Chemistry of the Silicates. Vols. 1-3.
translated from Russian by E. H. Murch. National Lending Library of Science and
Technology. Boston Spa. England. 19?1; reviewed inJ. Am. Ceram, Soc. 55,126 (1972).
15. W. Hinz, Silikate: Grundlagen der Sllikatwlssenschaft und Silikattechnlk, Vol. 2. Verlag ',-,
Bauwesen, East Berlin. 1971.



be underI "silicol"
agether of
ca system
composied for the
. in all its
does not
n "silica"



The Occurrence, Dissolution,

and Deposition of Silica

vick, N.J.

Solids. S,



sity Press



Silica is by far the major component of the earth's crust. yet much remains to be
learned of its chemistry and, in particular, its solubility behavior in water. The manner of its deposition to form such curiosities as quartz crystals containing inclusions
of mineral oil, mercury, or liquid carbon dioxide remains' a mystery (I). Flint, which'
our remote ancestors recognized as the strongest and toughest stone available, was
apparently formed in some instances from the siliceous skeletons of ancient sponges
by a mysterious process of solution transport. Within some plants and marine
organisms, soluble silica is transported and deposited in characteristic intricate patterns. Only recently has it been recognized that soluble silica, even in trace amounts.
plays a role in the development of mammals.



Vol. 18.
Vol. 18.
irk, 1962.

loIs. 1-3.
ence and

2..... clg

As water is a unique liquid, so is amorphous silica a unique solid. They are much
alike, both consisting mainly of oxygen atoms with the smaller hydrogen or silicon
atoms in the interstices. As pointed out by Weyl and Marboe (2), "Some properties
of water and silica are so similar that the transition between hydrated silicic acids
and the aqueous mat-ri is a gradual one." Washburn (3) noted that water and arnor. phous silica both have a temperature of minimum .volume. Ephraim (4) observed
another similarity between silica and water in that water is much less dense than
expected from close packing of the constituent atoms and from X-ray diffraction
studies. Bernal and Fowler (Sa) concluded that water molecules are arranged in a
rather open structure like quartz. and undcrcooled water has a still III are open structure,like tridymite. Another model has been proposed by Weres and Rice (5b).
These ideas lead to the suggestion that there is some relationship between the
density of water and the solubility of the various forms of silica. since both are
related to the close packing of oxygen atoms. Both silica and water consist. from the
stlWi4pQint of volume. largely of oxygen atoms. which are packed together with a
characteristic nnck inu density. The small hvdr oucn and silicon atoms lit between the

The Occurrence. Dissolution. and Deposition of Silica






oxygen atoms, contributing little to the volume. In pure orthosilicic acid, Si(OH). (if
it could be prepared). the small silicon and hydrogen atoms. lying in the interstices
between the large. oxygen atoms, would be more or less evenly distributed
throughout the mass. Polymerization of silicic acid to form solid silica and water
amounts to separation into two phases: in silica. the silicon atoms surround
themselves with oxygen atoms in a region of closer packing, and in water hydrogen
atoms surround themselves with oxygen in a region of more open packing. In amorphous SiO" there are 1.17 grams of oxygen per cubic centimeter; water of density
1.0, there is 0.89 gram of oxygen per cubic centimeter.
There is no evidence that silica is "soluble" to any appreciable degree in any
liquid other than water. However, that statement may depend on the definition of
"soluble." The dissolution of silica involves a chemical reaction or hydrolysis in an
excess of water:

Thus it is not a simple solution such as that of sugar in water, where the sugar
molecule exists intact in solution as in the crystal1ine state. Instead. it is analogous
to a hypothetical equilibrium of silica and ether in an excess of ether:

..i .




! .


Since low condensation polymers such as [(HO)2SiO]. appear to be clear watermiscible fluids resembling a poly hydroxy organic compound like glycerol (6). the
monomer, "soluble silica" or Si(OH). would probably be a clear liquid if it could be
isolated in anhydrous condition, In a pure state it might ~ven crystallize.
The unusual nature of the silica-water system has been noted by J. A. Kitchener
(7), who pointed out that the endless confusion in the literature concerning the
silica-water interface has arisen because the hydration and solubility characteristics
have not been understood. For example. there is the question as to why silica sols
are extraordinarily stable 'at pH 2 where the zeta potential is zero and become
increasingly sensitive to electrolytes at higher pH. where the potential is highest-in
contradiction to the generally accepted electrical double layer theory. Another
mystery is that crystalline quartz becomes coated with a film of amorphous silica
even though the solution is undersaturated with soluble silica with respect to a surface of amorphous silica.
The dissolution and deposition of silica in water involves hydration and dehydration reactions catalyzed by OH - ions:
(SiO z).. + 2 HzO



(Si0 2) x _1 + Si(OH).

For massive amorphous silica, the equilibrium concentration of Si(OH). at 25C

corresponds to 70 ppm as SiO:. This is the "solubility" of anhydrous nonporous -.
amorphous Si02 However, except for fused StO, glass, the common forms of arnor- '--""



m of Silica

The Silica-Water System

,i(OH). (if
and water
In amorof density

phous silica consist of extremely small particles of amorphous silica. or porous

aggregates, the surface of which is hydrated as SiOH groups. These exhibit a
somewhat higher solubility so that most powders and gels have a solubility of
100-130 ppm sto;
On the other hand, crystalline silica, such as quartz, almost universally present as
"sand," has a much lower solubility, of the order of 6 ppm SiO z.
Supersaturated solutions of monomeric Si{OH). are formed when silica is
dissolved in water at high temperature under pressure and then cooled, or when an
aqueous solution of soluble silicate is acidified:

ee in any
inition of
ysis in ~

the sugar

Supersaturated solutions of silicic acid in pure water are thermodynamically unstable because condensation polymerization through dehydration takes place. All
higher polymers of whatever size, molecular weight, or state of hydration can be
represented by a general formula containing n silicon atoms. The polymerization of
additional monomer 'molecules or the deposition of silica can be represented as
[SiIlO ZIl _C/UIZI(O H )lIz]

+ m Si(OH).


[Sill+mOzlI-clIZlzl+zmlz-p,(OH)IlZHm-p,] + 2pm HzO

:ar -vater>1,
t could be
rning the
silica sols
j become
ous silica
to a sur-

where n = number of silicon atoms in a polysilicic acid molecule or particle or

polymeric network
x = number of OH groups per silicon atom in the polymer, not exceeding 4
m = the number of monomeric silicic acid molecules added to the polymer
p = fraction of the hydroxyl groups per monomeric silicic acid molecule that
are converted to water during the polymerization reaction


Thus when p = 1, the monomer is converted to SiO z within the polymer molecule
without change in the number of OH groups in the polymer. There are, of course,
restrictions such as n and In having to be integers and the values of x and p being
limited by the possible structures of polymers and conditions of polymerization.
However, for the case where dense amorphous silica is being deposited on extensive, massive silica surfaces from slightly supersaturated monomer solution. especially at high temperature and neutral or alkaline pH. x is very small. p is unity. and
n is large. Thus the deposited silica may be essentially dense and anhydrous:

at 25C

Even vitreous or glassy silica contains some water. probably as SiOH groups. At a
given temperature and humidity there is an equilibrium "solubility" of water in vitreous silica, according to Hetherington and Jack (8). Flame-fused quartz contains
0.04 wt, % OH, whereas electrically fused material contains only 0.0003% as
detected byInfrared absorption at 2.73 micron wavelength. By extrapolation to'


The: Occurrence. Dissolution. and Deposition of Silica

30C. Moulson and Roberts (9) concluded the equilibrium concentration of water in
silica glass may be as high as 0.22% H 2 0 . probably present as internal SiOH .

Let us return to the behavior of soluble silica in water. When the solution is hfghly
supersaturated and insufficient solid silica surface is available to permit rapid
deposition of soluble silica. new small nuclei particles are formed by intercondensa_
tion on monomer and low polymers. Silica is also deposited on these until supersajjj,
ration is relieved.
It is in this manner that colloidal particles of silica are formed. These. in turn.
may be aggregated to form silica gel or may be laid down as opal. both of which are
highly porous with an extensive internal surface covered with SiOH groups. Thus
"hydrated" silicas are formed. Very slow deposition may produce quartz. .


Thermodynamics of the System


The heat of formation of silica by the reaction


Si(c) + O 2 (g)


SiC: (s)

was reported (10. II): ~H~ISOK ... - 217.5 == 0.5 kcal mole':' i for alpha quartz. and
-215.9 0.3 for amorphous silica.
Greenberg and Price (12) give somewhat different estimated values:





Alpha quartz (q)

Colloidal silica (cs)
Vitreous silica (vs)

'1 '


(cal mole:")

(cal rncle ")

'J. ",

f .:


- 1220 (~Fq - AFcs 200C)

- 550 (~Fvs - AFcs 25C)

for the overall equilibrium




Greenberg (13) calculated the following values for the thermodynamic functions:


AH (kcal mole-I)
~F~9S0K (kcal mole ' ')
.;lS~9"oK (cal deg" ' mole:")

Amorphous Silica


+2.65 == 0.28
+3.98 == 0.04
-2.82 == 0.50

+ 7.34 0.37
+5.20 0.04
+4.53 0.71

According to these data. the heat of formation of quartz from amorphous solid silica
is ~H = -4.69 kcal mole:", which is .....areatcr than the value -1.78 found bv
. W~'et

I1 .



and Deposition of Silica

mcenrration of water in
11 SiOH groups.
:n the solution is highly
lable to permit rapid
med by intercondensa1 these until supersatu-

irmed, These. in turn,

ipal, both of which are
h SiOH groups. Thus
rce quartz.


Wat.r System

a1. (14). The latter is closer to the value 0.54 :I: 0.2 more recently calculated by
Cochran and Foster (15).
Other reported values for the above hydration reaction of amorphous silica were
given by Morey, Fournier, and Rowe (16), who found

Kitahara (17a) measured the solubility of amorphous silica between 9 and 100C
and calculated AHa.o K .. 3.2 kcal mole:".
Walther and Helgeson (17b) calculated the thermodynamic properties of aqueous
silica and the solubility of quartz and its polymorphs over a wide range of temperatures and pressures. The thermodynamic constants derived from all available data
were evaluated as follows:
Entropy, So (cal deg " mole-I)
Volume, VO (em! rnole ")
Gibbs free energy, AG (cal mole:")
Enthalpy, AH (cal mole:")

Alpha Quartz

Amorphous Silica


. -214.57

'or alpha quartz, and

Coefficients were also given with equations for calculating the values over a wide
range of temperatures and pressures.


Relating Particle Size and Composition

. AFcs 200C)
. AF c 25C)

/narnic functions:
+ 7.34



.rphous solid silica

I f(\""d by Wise et

In most sols that consist of discrete spherical particles of amorphous silica, the
interior of the particles consists of anhydrous Si02 with a density of 2.2 gem -'. The
silicon atoms located at the surface bear OH groups which are not lost when the
silica is dried to remove free water."
The relation of particle composition to particle size can be calculated purely from
geometry and densities of the components. Let
n, = total number of silicon atoms in a particle
n, = number of silicon atoms at the particle surface
d . = diameter of particle on anhydrous basis (nm)
d" ~ diameter of hydroxylated particle (nm)
x = ratio of SiOH groups to total Si atoms
= ns/n t assuming one OH per surface silicon

w ... weight of one anhydrous Si0 2 particle (grams)


weight of one surface hydroxylatcd particle

= average number of silicon atoms across the diameter of a particle




The Occurrence. Dissolution. and Deposition of Silica

surface had decreased and the particles have grown to a certain size. Further spon-




taneous changes are unlikely to occur.

One point that has not been considered in the foregoing discussion is that the
energy values have been generally determined on types of silicas that have already
reached a relatively stabilized state of particle growth. On the other hand. for much
finer silica, for example, with a specific surface of more than 600 m Z g-l, the radius
of curvature of the surface is then less than 25 A, and the silanol groups must be
spread apart so that less hydrogen bonding can occur between neighboring hydroxyl
groups. In turn. it might be expected that this would increase the heat of "wetting,
decrease the heat of dehydration, and decrease particle density and surface energy.
Under these conditions, it is certain that particle growth occurs with decrease in the
radius of curvature, but energy data on such materials have not been obtained,
particularly in regard to surface energy of the silanol-water interface.

,J .





The soluble form of silica is monomeric, containing only one silicon atom and
generally formulated as Si(OH). This is often called monosilicic acid or orthosilicic
acid. The state of hydration is not known. although at high pressure there is some
indication that one water molecule is linked to each OH group, probably by
hydrogen bonding, so the hydrated molecule is represented by Willey (20) as
Si(OH: OH z).'
The structure of monosilicic acid is assumed to involve silicon coordinated with
four oxygen atoms as in amorphous vitreous silica and in crystalline quartz.
Although there are rare minerals such as the stishovite form of SiO z (21) or thaumasite (22), in which silicon is coordinated with six oxygen atoms, silicon in most
oxides and silicates is surrounded by only four oxygen atoms. If the monomer had
the structure HzSi(OH)., one would expect it to be a strong acid like the analogous
HzSiF., but in fact it is a very weak acid.
It is essentially non ionic in neutral and weakly acidic solution and is not
transported by electric current unless ionized in alkaline solution . .lt is not salted out
of water nor can it be extracted by neutral organic solvents.
It remains in the monomeric state for long periods in water at 25C, as long as
the concentration is less than about 2 x 10 -3 M. but polymerizes. usually rapidly. at
higher concentrations. initially forming polysilicic acids of low molecular weight and
then larger polymeric species recognizable as colloidal particles.
The question often arises as to whether the term "soluble silica" should include
the low polymers such as tetrarner or decarner, which are classed as "oligorners." It
becomes a matter of definition. "Soluble" materials have been recognized as those
that pass through a dialysis membrane. whereas colloids do not: but even though
membranes can now be made with pores sufficiently small to separate dextrose from
sucrose. we think of sucrose as being "soluble," On the other hand. sucrose is certainly not colloidal.
For the purpose of this book. the following terminology is used:


Deposition of Silica

Soluble Silica-Monosilicic Acid

size. Further spon-

Soluble silica (or monosilicic acid). Si(OH)4'

Polysilicic acid (oligomers). Polymers with molecular weights (as Si0 2) up to
about 100,000, whether consisting of highly hydrated "active" silica or dense
spherical particles less than about 50 A in diameter.
Colloidal silica. More highly polymerized species or particles larger than about 50
A, although sometimes down to 10-20 A.
Silica sol. May refer broadly either to polysilicic acid or colloidal silica.

cus.. is that the

that have already
ier hand, for much
m2 g the radius
01 groups must be
ghboring hydroxyl
ie heat of wetting,
nd surface energy.
lith decrease in the
lot been obtained,



The arbitrary borderline of 50 A or mol. wt. 100,000 is based on the general

observation that below this point the polymer species are generally unstable, having
only a transient existence owing to gelling or particle growth.
Also, as has already been shown, it is below this size range that less than half of
all the silicon atoms are present as Si02 that is, as "silica," whereas more than half
are each associated with at least one hydroxyl group. The term "silicic" acid is thus
The preparation and reactions. for example. polymerization. of dilute solutions of
monosilicic acid are further described in .Chapter 3. Meanwhile, some of its characteristics are noted, as follows, prior to discussing solubility:

silicon atom and

acid or orthosilicic
.sure there is some
'oup, probably by
Jy Willey (20) as

1. It is characterized by its rapid rate of reaction with molybdic acid to form the
yellow silicomolybdic acid.
2. It is generally inert in neutral solution if the concentration is below the saturation
level with respect to amorphous silica. Thus it is almost universally present at a
concentration of a few parts per million in most natural waters and in living
3. It combines with metal ions to an increasing degree with increasing pH, thus
reducing the concentration of free monosilicic acid. (Ferric and uranyl ions react
at a pH as low as 2, whereas most other metal ions combine only at higher pH.)
4. Above pH 9 it is ionized first to (HO)3SiO - or at still higher pH to (HO)2Si022-.
The first equilibrium constant (13.23) is approximately (25C)

c. dinated with
crystalline quartz.
Si02 (21) or thauns, silicon in most
the monomer had
like the analogous

lution and is not

.lt is not salted out

[OH-] [Si(OH)4]

t 25C. as long as
usually rapidly. at
ilecular weight and



. or
[(HO)3SiO-] [H -]

ca" should include

as "oligorners." It
ecognized as those
t: but even though
.rate dextrose from
LOd. sucrose is cer-

Even though the silica solution is neutral. if it is passed through a bed of strongbase cation-exchange resin in the free-base form. the soluble silica in contact with
the resin is ionized and is then held as silicate ions. In a mixture of Si(OH)4 in
equilibrium with colloidal silica particles at pH 7-8, the particles bear a negative
charge. According to Goto, Okura, and Kayarnu (24) electrophoresis and trans-




port studies show that the colloid. not the "molecularly soluble" Si(OH)., is the
charge carrier. When the mixture is passed through a mixture of strong-bas'
anion- and cation-exchange resin, monosilicic acid is removed. but not the
colloidal particles. [After a time the particles dissolve surticiently to reestablish
the equilibrium concentration of Si(OH).]
5. It is converted to H 2SiF. by reaction with HF in aqueous solution:




--The Occurrence. Dissolution. and Deposition of Silica


~ ...

. :.

!;.,. ~"

~ .

"1( :'


+ 6 HF

== 2 H+

+ SiFi- + 4 H20

6. It is converted to a complex anion by reaction with o-dihydroxy aromatic compounds, such as catechol, in neutral solution:

+ 3 o-C 2H.(OH)2 + 2 NH.+ + 20H-

(o-C 2H.02)3Sj2- + 2 NH. + + 8 H20

Volatility in Steam
Although Si(OH). is nonvolatile at ordinary temperature and polymerizes quickly
when heated, nevertheless at elevated temperature and pressure in'water its solubility
is greatly increased and it can exist in equilibrium as the vapor phase in the steam,
as shown by Kennedy (25). This is of importance in very high pressure boilers in
power plants where deposits build up on turbine blades unless all silica is remover
from the feed water. Brady (26) supposes the volatile species is Si(OH). 0,.
(HO)3SiOSi(OHh. Astrand (27) found that volatility increased with decreasing
alkalinity in experiments conducted up to 350C and 300 atm. This. of course. suggests that Si(OH). is more volatile than the silicate ion. Wendlandt and Glemser (28)
reviewed evidence from earlier workers and calculated the equilibrium constants
involved whence the species in the vapor were related to the density of the water

. ,

.. I.


Gas Density Range (g em -3)

Si0 2 + 2 H20 == Si(OH).

2 Si0 2 + 3 H20 == (HO)3SiOSi(OH)3
Si0 2 + H20
== OSi(OH)2

Up to 0.05
Up to 0.45
Above 0.65

Similarly. Martynova, Fursenko, and Popov (29) found that in a solution saturated with soluble silica at 263-364C about a third of the silica in the vapor was
present as disilicic acid, whereas in the range lSI-223C it was all monomeric.
Heitmann (30) concluded that deposition in turbines was minimal if the silica
concentration was less than 0.01 ppm. iron concentration less than 0.005 ppm. and
the conductivity less than 0.1 micromho cm -I. According to Heitmann's measurements (3 I). the silica concentration in the vapor phase ranges from O. I rng kg -I r
400C to 5 mg kg:" at 600C at a pressure of 0.3 kg em -2, but increases to mor.. _
than 100 mg kg -I under an applied pressure of 300 kg ern -2.

isition of Silica

,i(OH)., is the

.r ~"'ng-base

b, lot the
to reestablish

romatic com-

erizes quickly
r its solubility
in the steam,
ure boilers in
:a ie removed
H). or
th decreasing
,f course, sug. Glemser (28)
urn constants
of the water


' -.,

: (g cm -3)

~ -.

solution satu.he vapor was

I if the silica
~05 ppm, and
nn's measure.1 rn~. kg-I at
0 more

_ _ _ _ _ _t;;...'


Soluble Silica-Monosilicic Acid


Soluble Silica in Nature

Silica is constantly dissolving and precipitating over a large part of the earth's surface. The sedimentary cycles have been described in complex detail by Siever (32).
Soluble silica is mainly derived from the weathering of minerals which, in some
cases, results in amorphous silica residues that then dissolve. Very little can come
from the "sands of seashore," or quartz, which is soluble to only a few parts per
million; furthermore, the rate of dissolution is extremely slow. River waters range
from S to 3S ppm Si0 2, a few up to 7S ppm, and by the time they reach the sea may
range from 5 to 15 ppm. Seawater varies widely but the silica content may range
from 2 to 14 ppm (33). However, Lisitsyn and Bogdanov (34) report that surface
waters in the Pacific Ocean contain only 0.0001-0.3 ppm Si02 Plankton convert
6 X 10' tons Si0 2 from soluble to suspended form each year, but this is only 0.16%
of the available silica.
In addition to the silica carried into the sea by fresh water, additional soluble
silica comes from the suspended colloidal clays and related minerals. Tests show
that common colloidal silicates like clay will dissolve in seawater sufficiently to give
a silica concentration of 10 ppm (35).
Concentrations of silica of around 2 ppm were reached in dilute salt solution with
mica and kaolin and up to 15 ppm with montmorillonite (36). When seawater was
enriched with soluble silica to 25 ppm Si02 , it remained at that level for a year in the
absence of these minerals, but when the latter we~e then added, the silica was
removed from solution down to the 2-15 ppm level that was reached when the
minerals alone were added. Since many ocean waters contain 2-10 ppm Si0 2 , it is
possible that this value is reached as the equilibrium solubility of colloidal aluminosilicate in suspension. The above experiment is consistent with the fact that in pure
water, pure amorphous silica dissolves to give a concentration of monosilicic acid of
100-110 ppm, but in' the presence of polyvalent metal cations such as iron, aluminum, and other metals, colloidal silicates are formed with a much lower solubility
with respect to monosilicic acid. lIer (37) has shown that soluble aluminum reduces
the solubility of amorphous silica from about 110 to less than 10 ppm.
Willey (38, 39) has studied the natural interaction of soluble alumina and soluble
silica in 0.6 N sodium chloride solution. Addition of aluminum ion to 200 ppm
Si(OH). retards polymerization. Probably there is formed a colloidal complex which
reacts as monomer when put into the strongly acidic molybdate reagent. A very low
concentration of soluble silica also causes the precipitation of alumina.
Soluble silica as determined by the molybdate: method is not necessarily present as
Si(OH). Bogdanova (40) reported that in natural waters that contained only about 5
ppm total silica, 4-9% of the silica was polymeric but was converted to monomer by
acid. It is most likely that the "polymeric" silica was actually very small colloidal
particles of aluminum silicate that liberated monomer when acidified..
Silica is continuously removed from seawater by biochemical processes. Diatoms
and sponges as well as plants remove silica which is stored within the organism.
Although Calvert (41) believes that the concentration of silica in the sea is mainly
controlled by biological activity, Harder (42) reports that amorphous hydroxides of
AI, Fe, Mn, or Mg can react with and precipitate soluble silica, thus reducing the
concentration to as low as 3 ppm. Both processes are no doubt operative.

-'.;..... -

-The Occurrence. Dissolution. and Deposition of Silica


Amorphous silica is probably more soluble in seawater at great depths owing to

the higher pressure. Willey (20) and Jones and Pytkowicz (43) found that at about
OC, the solubility increased with pressure as follows:
Willey (OC)


Ib in. -2

ppm Si02




Jones (2C)
Ib in.- 2

ppm Si02







The salt at this concentration no doubt promoted more rapid establishment of equilibrium but would have little effect on solubility.
Hot springs in some areas produce a supersaturated solution of silica. Knowing
the solubility of quartz, Fournier and Rowe (44) have shown that the total silica
content of the water permits an estimate of the subterranean temperature at which
the water has become saturated with respect to quartz, which is the major phase that
usually determines the solubility. Typical results are as follows:


Dissolved Silica
660 .
425 .' .





j .

Temperature (OC)

Maximum Measured Temperature

in Drill Hole (0C)




The method is, of course, not dependable if the water encounters previously
deposited amorphous silica at any point. Also, the loss of water as steam must either
be prevented or taken into account.
. Major studies of silica in geothermal waters have been made in 1956 by White.
Brannock: and Murata (45) and in 1970 by Fournier (46). all of the united States
Geological Survey. Detailed analyses of waters from Yellowstone National Park,
Wyoming, have been published by Rowe, Fournier, and Morey (47).
Soluble silica is found in essentially all plants and animals. For example. human
blood contains I ppm. Ingested monosilicic acid as an undersaturated solution
rapidly penetrates all tissues and body fluids and is excreted apparently without any
effect (48). Plants, especially grasses. including the grains and rice. take up silica and
deposit it in the tissues as characteristic microscopic amorphous opaline particles.
which are later found in the soil and in the intestinal tracts of grazing animals (49).
The widespread occurrence and possible role of silica in living systems are more fully
discussed in Chapter 7.
In regard to the weathering of soils. it is noted that aluminosilicates (clays)
. undergo weathering in the tropical regions with dissolution of silica, leaving a .-'
residue high in alumina (bauxite) whereas in colder regions alumina seems to be
removed preferentially, leaving more highly siliceous residues (50).



Phases of Silica

sition of Silica

A possible explanation is that in the tropics the decomposing vegetation produces

tannins and other catechol-like materials that are known to dissolve silica in neutral
solution. In colder regions, .less organic matter is likely to be present and the pH
may be lower because more dissolved carbon dioxide is present, so that alumina is
preferentially dissolved.

ths owing to
hat '1bout

Since different phases of silica exhibit different solubility behavior, they are briefly
described. By far the commonest crystalline form is quartz, the main constituent of
common sand. However, under certain conditions in nature and in the laboratory,
other forms are produced. These forms in turn may be divided into the following
nent of equi-

1. Anhydrous crystalline Si02

2. Hydrated crystalline Si0 2xH 20.
3. Anhydrous amorphous silica of microporous anisotropic form such as fibers or
4. Anhydrous and hydrous amorphous silica of colloidally subdivided or
microporous isotropic form such as sols, gels, and fine powders.

:a. Knowing
e total silica
ure at which
or phase that

S. Massive dense amorphous silica glass.


Of these, 2, 3, and 4 exhibit extensive external or internal surfaces and are thus
pertinent to the present study.


Anhydrous Crystalline Silicas

Sosman (51) classified the more common phases as follows:

s previously
1 must either _

Stable at Atmospheric

;6 by White,
mited States
tional Park,

Quartz low
Quartz high
Tridymite S-1
Tridyrnite S-II
Tridyrnite S-lIl
Tridymite S-IV
Tridymitc S-V
Tridyrnite S- VI
Tridyrnite 1\1:1
Tridymitc M-I r
Tridymitc M-1I1
Cristobalite low
Cristobalitc high

nple, human
ted solution
without any
up silica and
ne particles.
mirnals (49).
-e more fully

cates (clays)
a, l-uving a
to be

_ _ _ _ _ _ _---..J:.


Stability Range (OC)
To 573
573-867 - tridyrnite

475-IHO - cristobalite
To 117
Above 163




The Occurrence. Dissolution. and Deposition of Silica


The different forms of quartz. tridyrnite, and cristobalite are transformed spontaneously with temperature so that from the standpoint of solubility there are only
the three phases to be considered.
The next group of three phases are those formed only under conditions of high
temperature and pressure. '
Thermodynamically Stable Range
Therrn odynam ically
Stable at High Pressure
:! .

't .
rI '


Temperature (OC)
From 300


Pressure (kilobars)




Surveys of these phases and their properties have been published by Fronde! (52).
Sosman (53). and Fldrke (54). Wells relates the structure of the different forms of
silica to various crystalline silicates (55).
Quartz, the commonest phase found in nature. ranges from huge crystals. to amorphous-looking powders a few microns in size. to shapeless masses of chalcedony agate
or flint consisting of densely packed. interlocked microscopic crystals. The
transformations between the three common forms and vitreous silica is as follows:












~ . .

j .;

i '




i .



The transformation to tridymite apparently requires traces of certain impurities or

The three phases metastable at ordinary pressure were recognized only recently.
Keatite was discovered by Paul Keat (56) in 1954. and its formation via cristobalite
and transformation to quartz were studied by Carr and Fyfe (57). Hoover (58), in a
patent filed in 1954. described the preparation of a very similar if not identical
material from "silicic acid". that is. hydrated amorphous silica powder. by heating it
in water at about 3000 atm pressure and 500-625C in the presence of about 1%
alkali based on silica.
Coesite was discovered by Coes, in 1953 (59). It is made from amorphous silica in
the same temperature range as for keatite, but at 10 times the pressure and with
weakly acidic catalysts such as boric acid or ammonium chloride (59). It was found
in nature in 1960 at Meteor Crater, Arizona. apparently formed under the high
temperature and pressure conditions of the im pact.
Similarly. stishovite was first made in the laboratory in 1961 by Stishov and
Popova (60) and discovered in Meteor Crater by Chao. Shoemaker. and Madsen. in
1962 (61). A most interesting story of the isolation of substantial amounts of coesite
and stishovite from the crater is told by Bohn and Stober (62.63).
There arc also some unusual anhydrous crystalline forms. as follows (64).


:ion of Silica

med spononly
ft - - :


of high



'ondel (52),
nt forms of
Is. to arnoredony agate
/stals. The


Phases of Silica

Silica W is a fibrous crystalline silica with a density of 1.97 g em -, described by

Weiss and Weiss (65), formed in the gas phase by oxidizing silicon monoxide vapor
at 1200-1400C and deposited as paperlike films. It is unstable above about 1400C.
It is fairly stable in dry air, but is converted by moisture to amorphous hydrated
silica, still retaining a swollen fibrous form. In this transformation. only about 0.08
mole of H 20 is taken up per mole of Si0 2, forming SiOH groups. Silica W can have
no true equilibrium solubility in water. Instead, it must decompose rapidly in water
to give monosilicic acid. When the powder is suspended in water and within 2 min
centrifuged to obtain a clear solution. then titrated with NaOH solution at pH
10.2-10.5 (thymolphthalein). 2 equivalents of base are required per mole of Si0 2 in
solution (66). After the solution has been aged for 1 hr, only 0.1 equivalent is
required .. Initially. therefore, the solution must have been supersaturated with
Si(OH)4, which when titrated with base, requires 2 equivalents of alkali. but after the
monomer has polymerized. much less alkali is required to neutralize the surface
acidity of the colloid. If the solution is mixed with silver salt an orange. light-sensitive AgSiO, is precipitated. In absolute methanol the fibers swell and form a
polymeric methyl ester containing one methoxy group per silicon atom which, when
heated in vacuum at 300-500C, yields cyclic methyl esters [(CH,OhSiO]II.'.u,
_When hydration of the fiber by water is followed by suitable technique under the
microscope it can be seen that the reaction starts at the end of the fiber and proceeds
rapidly along its length as the crystal swells and is converted to hydrated amorphous

vitreous -

npurities or
Iy recently.
er (58). in a
ot identical
'y heating it
f about 1%
ous silica in ...
re and with
t was found
er the high
Stishov and
Madsen. in
lS of cocsitc .

Anhydrous fibrous silicas formed in connection with high temperature metallurgical operations were noted as long ago as 1852 by Schnabel and 1859 by Rose.
Soft. silky fibers of more than 98% Si0 2 were classed as aphanitic (invisible) silica.
.' and also known as lussatite. Around 1910. in the mouths of electric furnaces making
silicon carbide. a sort spongy gray deposit called "elephant's ear" was identified as
microfibrous amorphous silica (67). It is likely that all of these were silica W.
. Melanophlogite, a long known but strange and little understood mineral. is found
in volcanic sulfur deposits in Sicily. Skinner and Appleman reviewed its history and
showed that it was a new cubic polymorph of silica (68). It has a cubic. very open
structure. containing 92.4% SiO; and about 5.7t:(. SO, (2.2SC;c as sulfur). 1.2%
carbon. and 0.81 % hydrogen. The density is 2.052 0.013 g COl -'. The initial
refractive index is 1.467. but when these volatile materials are driven off by heating.
the crystalline silica residue has a refractive index of 1.425 0.002 and a density of
1.99 g COl -'. which arc substantially lower than those of amorphous or glassy silica.
The silica crystal is stable up to about 900C. above which it changes into cristo. balite. However. ,vhen crystals arc subjected to grinding in a mortar at ordinary
temperalure. the open structure collapses to fine-grained quartz. Its solubility h~IS
not been measured. and it is doubtful if il can ha v c a true equilibrium solubility in





. '; .


- The Occurrence. Dissolution. and Deposition of Silica

water. The heat-purified material will probably react with water rapidly to give a
highly supersaturated solution of Si(OH)4' similar to the behavior of silica W.
The nature of the hydrocarbon and sulfur content is still not clear. However. calculations based on density data would seem to support earlier suggestions that the
sulfur must be present as S03 or H 2S0 4 within the silica lattice. The optical characteristics of the mineral show that the organic matter occurs in films between the
faces of the crystals. On the other hand, calculations based on the difference in
densities of the original mineral and the pyrolyzed silica crystals show that the sulfur
compounds at least must be within the crystal lattice. Kamb offers evidence (69) that
the silica structure is a clathrate with S02. H 20. and CH 4 in the lattice analogous to
the known 12 A gas hydrates of water, 6X .46H 20, where X is CH 4, H 2S, CO 2, S02.
C12 etc., and in fact the structure is the complete analogue of 6C1 2 H 20.
Silica 0 crystallizes from lithium silicate glasses during devitrification at low
temperature. It has a crystal lattice similar to quartz and may simply be "high
quartz" stabilized below 573C. the normal transition temperature, to low quartz
(53. 70) by inclusions of metal ion impurities. The only way pure material can be
obtained is by neutron bombardment of quartz.
Silica X is a microcrystalline form obtained as spherical aggregates of radial
fibers up to 12 microns in diameter by heating pure amorphous hydrated silica
("silicic acid") with 2% KOH solution in sealed tubes at 150C for a few weeks. The
refractive index of 1.484 0.004 is close to that of cristobalite. It is anhydrous.
maintaining its structure up to about 600C. above which it is converted to cristobalite (71a. 71b).
Silica lite. a very unusual new form of anhydrous crystalline silica hornogeneousl..
permeated by uniform pores 6 A in diameter and having a density of only 1.76 g
ern -3, has been described by FJanigen et al. (71 c) and patented by Grose and
Flanigen (71d). The pores constitute 33% of the volume of the crystal. A most
remarkable feature is that this silica is hydrophobic; the pores are lined with oxygen
atoms that are highly hydrophobic and organophilic or oleophilic. Thus the crystals
preferentially absorb hexane in the presence of water. which does not enter the pores
even at saturation pressure.
This type of silica is made first as a crystalline quaternary ammonium silicate. for
example, tetrapropylammonium silicate: (TPA).O .48Si0 2 H 20. It is then heated to
red heat to remove the organic matter and water. leaving uniform cylindrical channels throughout the three-dimensional crystalline framework of silica.
A similar but even more hydrophobic. anhydrous. microporous crystalline silica
was obtained by Flanigen and Patton (7le) by conducting the hydrothermal synthesis in the presence of some ammonium fluoride. which facilitated the formation of
crystals 2-15 microns in size at only 100C rather than at the higher temperature
and pressure required for making silicalitc. After crystallization from solution a
typical composition was about 88% by weight of silica. 11.0% tetrapropylarnmonium
oxide. and 0.9% fluorine, but after calcination at 600C the porous crystalline
product was essentially pure Si02 (containing less than 0.1 % fluorine) with a mean'
refractive index of 1.39 0.01 and a specific gravity of" 1.70 0.05. These value"
arc the same as those of silicalite and fall on the same curve with other forms 0
silica in Figure 1.1.

on of Silica

to give a
veve. callS

that the

.al characetween the

Terence in
: the sulfur
e (69) that
alogous to
CO2, S0 2.
on at low
, be "high
.ow quartz
-ial can be
; of radial
ated silica
veeks. The
I to cristoog,
mly 1.76 g

Grose and
J. A most
ith oxygen
he crystals
r the pores

Relation Between Density and Refractive Index

The anhydrous crystalline phases were arranged in order of increasing density and
refractive index, and found to have a linear relationship, by Skinner and Appleman
(68). Stishovite falls on the s~me line (72) (Figure 1.1).
It will be noted that the line for silica polymorphs has been extrapolated to meet
that of two forms of water. It seems odd that neither of these lines extrapolates to a
refractive index of 1.0 (for a vacuum) at zero density.
The analogous structures of Si0 2 and H 20 have been compared by Kamb (69),
who points out that the ratios of the densities of the various phases or polymorphs of
Si0 2 and water to those of the corresponding forms of Ice I and cristobalite are very
similar, and for each type of silica there is an ice counterpart with the same type of
crystal structure.
Hydrated Crystalline Silicas

Until the advent of X-ray diffraction, it was not clear whether solids containing only
silica and water were definite compounds. that is, had a definite stoichiometry or
structure. In 1905, Tschermak (73) believed that he had obtained definite hydrates
based on ratios of silica to water corresponding to Si0 2 : 2 H:O, 2 Si0 2 : 2 H20, 3
Si02 : 2 H20 , etc. by carefully leaching the metals out of certain crystalline silicates
and drying in air. Then Van Bemmelen (74) and Theile (75) gave evidence that no
definite silicic acids were thus produced.
Nevertheless, since then numerous instances have been found where definite
crystalline materials. having characteristic X-ray diffraction patterns and crystal
structures. have been made by extracting the cations from certain crystalline silicates
with a c i d . '


silicate, for

heated to
rical chan-


Phases of Silica



illine silica
nal syntherrnation of
solution .1
ith a mean
11:S" values
:r .
s of


~ 1.5






' - - - - - - CRIST06:'UTE




' - - - - - - MELANOPHLOGITE


Figure 1.1.

G -CM-)

Density versus refractive index of various forms of silica.







! :,;











-:-- -.-

Th~ Occurrence. Dissolution. and Deposition of Silica

There is no instance of a hydrated silica being crystallized directly from a solution

of silica and water. Yet there is the peculiarity that certain hydrates. once formed,
exhibit what seems to be a characteristic solubility. implying that since an equilibrium is reached, silica must pass from solution to the solid phase as well as the
reverse. However, the data are conflicting.
Sodium disilicate, Na 2Si20s , may be converted to hydrated silica. Thus !?y treating
the silicate with concentrated cold acid, washing out salts with water and water with
acetone, and drying in vacuum at 40C, a crystalline "disilicic acid" is obtained (76,
77). Such layerlike structures are termed "lepidoidal" (scalelike) or phylloidal (Ieaflike). Liebau distinguished two types of layer structures. They are internally
hydrogen bonded and exhibit only weak acidity (78).
Crystalline Na:Si0 34H20 gives another form of silica. Here treatment with dry
HCI gas at -25C gave definite silica hydrates containing 2.0. 1.5, and 1.0 moles of
H 20 per mole of Si02, as the product was dried under vacuum at progressively
increasing temperatures. The silica was anhydrous at 90C (79).
From gillispite, BaFeSi40 10 , Pabst (80) obtained flakes of crystalline hydrated silica consisting of Si 40 l o 4 - ions linked into sheets with the composition 4 H 4Si 40 l o +
2 H 20; density calculated 2: 15, found 2.05; refractive index 1.45 0.0 I.
From sheetlike KHSi 20S ' Le Behan, Kalt, and Wey (81) obtained by treatment
with a dilute acid a still different leaflike disilicic acid, H 2Si 20 s .
From the zeolite mordenite, Chen and Smith (82) obtained silica having the same
zeoli tic skeleton by repeatedly heating the hydrogen form in steam at 538C and
extracting the liberated aluminum with acid. The pores are apparently lined with
siloxane oxygen bonds and are hydrophobic, retaining high hydrocarbon sorption
Chrysotile asbestos, JHO).Mg, .Si40 10 , treated with acid yields fibrous hydrated
silica. By reacting the silica with (CH3)3SiCI as rapidly as it is liberated, it is
converted to the trimethylsilyl derivative which. in organic solvent. is a highly
swollen mass of ribbons of silica only a few angstroms thick. When solvent is
removed the ribbons roll up again into tubes or fibers with a structure similar to that
of the original fibrous mineral. The fibrous product by X-ray diffraction shows a
prominent 15 Ad-spacing (83).
Magadiite, NaSi 70 133H 20, is a mineral first found in Kenya by Eugster (84) and
later in Oregon and California. However, a similar structure had been synthesized
years before by McCulloch (85) by prolonged aging of a mixture of microamorphous
silica and alkali (Si0 2 : Na 20 = 4-5) in concentrated suspension at 100C. lIer (86,
87) prepared it in good yield from a mixture of sodium silicate and colloidal silica
(Si0 2 : Na 20 = 4.6) and made a number of derivatives by esterification and ion
exchange. McCulloch reported the product to have a Si0 2 : Na 20 ratio of 9.4, and
IIer found 8.0. whereas the ratio of magadiite was l-l.O. However. lIer found his
material. when treated with acid, to exchange all the sodium for hydrogen ions. and
the crystalline silicic acid had the same X-ray pattern (within 2C;C d-spacings) as
reported by Gude and Sheppard (88) for "silhydrite." 3Si0 2 H~O. which is a
crystalline silicic acid found in association with magadiite. The acid obtained by
leaching magadiite was reported to be somewhat di ffercnt. The structure of rnaga- '; .....~-



ion of Silica

Phases of Silica

a solution
ce r~"rned.
an _..ilibNell as the

diite has been further studied by Brindley (89). McAtee, House. andEugster (90).
Bricker (91a) measured the stability constant:

by treating
water with
tained (76.
oidal (leafinternally
it with dry

o moles of

/drated sil-{.Si.0 1o +






and also calculated the free energy of formation as j.Go = -1762.2 kcal mole",
where G = standard molal Gibbs free energy.
There is probably a family of similar high-silica crystalline silicates having characteristic crystal structures. and from which corresponding crystalline silicic acids
could be obtained. Thus, along with magadiite, is found kenyaite, with a Si02: Na20
ratio of 22; its formula is N aSi l1 0 2o.(OH). 3H 20 (84). lagaly and co-workers (91b,
91c) have further described these "phyllosilicic acids" in regard to structure and ionexchange properties (see Chapters 2 and 5).
Lepidoidal silica is a term that has been used in the past to describe silica in the
form of extremely thin flakes or scales. Such silicas are usually hydrated and may
have a crystalline structure or may be amorphous. depending on how they are made.
Three types that have been described are as follows:

g the same
D8C and
lined with


ated, it is

) a highly
solvent is
ilar to that
n shows a

er (84) and

:. lIer (86.
oidal silica
n and ion
)f 9.4. and
found his
1 ions. and
lacings) as
/hich IS a
otained by

1. A synthetic silica [(Si,O,)Oa](OH), prepared from siloxane [HO~i-~iH]n


obtained by the hydrolysis of calcium silicide.

2. Sheetlike, crystalline forms of silica obtained by extracting layer-type basic copper silicate with acid (92, 93).
3. Flakelike gels made by freezing sols of colloidal silica or sols of hydrolyzed
HSiCla and oxidizing the product (93).
These hydrated silicas are similar only in physical form and are entirely different
in structure. (The type made by freezing is amorphous to Xvrays and is considered in
the following section.) A comparison of the chemical behavior of types 1 and 3 has
been made by Kautsky and Reise (94).
Kautsky and associates (92. 93) have studied in some detail the lepidoidal silica
obtained from a synthetic layerlik e colloidal copper silicate which they formulate as
[(Si,O,)(OH)12][OCuOH]. It has a very high specific surface area similar to the copper mineral chrysocolla (95, 96). which has a specific surface area greater than 300
m2 g", The corresponding silica has ion-exchange properties. but all the original
copper cannot be put back into the structure once it has been removed.
Amorphous Silicas
Amorphous silica may be broadly divided into three types:
I. Vitreous silica. made by fusing quartz.

2. Silica M. an amorphous silica formed when either amorphous or crystalline

silicas arc irradiated with high speed neutrons. The density of amorphous silica




-The Occurrence. Dissolution. and Deposition of Silica

increases and that of crystalline silica decreases. It is thermally unstable and is

transformed to quartz at 930C in 16 hr. Actually the silica M from different
crystalline forms may differ slightly from each other. The density is 2.26. as compared with 2.20 for vitreous or microamorphous silicas (53).
3. Microamorphous silica; which includes sols, gels, powders. and porous glasses
which generally consist of ultimate particles less than a micron in size or have a .
specific surface area greater than about 3 m 2 g-l. (These are discussed in detail in
Chapters 4 and 5.)
Frondel (97) states that amorphous silica is not truly amorphous but consists of
regions of local atomic order, or crystals of extremely small size, which by careful
X-ray diffraction studies appear to have the cristobalite structure. Nevertheless. by
ordinary diffraction procedures this material gives only a broad band. with no
multiple peaks as are ordinarily obtained with macroscopic crystals. and is referred
to here as "amorphous."
In nature, microamorphous silicas have either been condensed from the vapor
.phase ejected in volcanic eruptions or deposited from supersaturated solutions in
natural waters and in living organisms. Except for silica deposited in plants or
diatoms, natural microamorphous silicas are generally too impure to permit a
study of solubility. (The formation and properties of natural opal .are discussed in
Chapter 4.)
Microamorphous silicas made in the laboratory can be divided into three classes:

I. Microscopic sheets, ribbons, and liberlike forms obtained by special processes.

2. Common amorphous forms consisting of ultimate spherical particles of Si0 2 less
than 1000 A in diameter, the surface of which consists of anhydrous Si0 2 or
SiOH groups. These particles may be separate or linked together in three-dimensional networks as shown in Figure 1.2:
(a) Discrete or separate particles as in sols.
(b) Coherent three-dimensional aggregates with siloxane bonding at the points of
contact, as in gels.
(c) Voluminous three-dimensional open networks of aggregated particles as in
aerogels, pyrogenic silicas, and certain dispersible silica powders.
3. Hydrated amorphous silica. here designated as a separate class in which most. if
not all, of the silicon atoms each retains one or more hydroxyl groups in the silica
structure. This type of polymeric structure is obtained if monosilicic acid or
oligosilicic acids in water are concentrated and polymerized at ordinary or low
temperature and in slightly acidic solution. It is now believed that under these
conditions the silica polymerizes to extremely small spherical units less than
20-30 A in diameter, which. when concentrated. link together into a three-dimensional gel mass, trapping water in the interstices. which are of molecular dimensions and retain water which can be desorbed only above about 60C.
Ordinarily such structures are not retained because in making sols and gels the pH is \._


:ion of Silica

able and is


Phases of Silica


26, i1~ comous glasses

e or have a
in detail in

consists of
by careful
rtheless, by
d, with no
i is referred

the vapor
ioluticns in
:t plants or
o permit a
liscussed in

ee classes:

Figure 1.2. Ultimate particles in common forms or colloidal silica. Aggregation is actually in
three dimensions but here represented only in two. A, sol; B, gel; C, powder.


of Si02 less
'us Si0 2 or

not kept sufficiently acidic nor is the temperature kept below 60C during processing to the final state.
Microscopic Sheer, Ribbon, and Fiberlike Forms

te points of

Sheet or flakelike amorphous silica particles are obtained in various ways:

tides as in

I. Formation at the gas-liquid interface: hydrolysis of gaseous SiF. and

simultaneous polymerization of the silicic acid in water at 100 or OC produced
. opaque "scales" of silica: so does hydrolysis of-SiCI. vapor at IOO"C (98). The
scales appear to be films of silica gel formed at the. gas interface when the
extremely reactive vapors meet the water surface. The remarkably "fluffy"
character of the powder made from SiF. by Jacobson (99) is shown by its very
low bulk density of 0.025 g ern -3 and the fact that it appears to now as water
does. The irregular gel flakes, a micron or so in diameter and perhaps a tenth as
thick, contain 92.86% Si0 2 and 7.14% H20.
2. Formation by freezing silica sols: When a solution of colloidal silica or polysilicic
acid is frozen, the growing ice crystals exclude the silica until it remains as a
concentrated sol between the ice crystals and then polymerizes and forms a dense
gel. When the ice is melted the silica is obtained usually as irregular flakes

ich most, if
in the silica
cic acid or
iary or low
rnder these
s less than
iree-dirnenliar dirnen:?

Is L

H is




The Occurrence. Dissolution. and Deposition of Silica

formed between the flat surfaces of the ice crystals. Lepidoidalsilica formed in
this way from polysilicic acid of low molecular weight has been extensivelv
studied by Kautsky and associates (92-94). The gel structure is of high densit,
owing to the high concentration of the silicic acid before it polymerizes to gel,
and the specific surface area of the microporous mass is around 900 m 2 g-1 so the
ultimate silica particles can only be 20-30 A in diameter. Kautsky reported that
the vacuum-dried powder contained about 10% H 20; if the water was all present
as SiOH groups and each OH group occupied 12.5 A on the surface of the ultimate particles, then the specific surface area is calculated to be 930 m 2 g-1 SiO z'
Kautsky investigated the capacity to exchange hydrogen for copper ions from
solution (0.1 M Cu(NH3)4S03 + 0.3 M NH 40H) in 200 hr. The product
contained one copper atom per silicon (-SiOCuOH) and had surface area (by
BET-method adsorption) of 870 m 2 g ",
3. Leaflets of silicon oxyhydride (HSi0 1. s ) up to 5 mm in diameter are formed when
HSiCl 3 in ether is hydrolyzed by very gradual addition of a theoretical amount of
water. The product appears to "crystallize" when the concentrated ether solution
is evaporated, It is possible the silica forms at the silica-water interface but
remains invisible because of the close refractive index until the ether "evaporates
4. Sheets of silica consisting of a single layer of colloidal particles are formed when
silica is coagulated under the influence of cationic surfactants (101). The
mechanism of formation is discussed in Chapter 4.
5. Highly elongated forms of silica, other than libers formed from glass, have bee'
obtained either by conversion from fibrous precursors or by unidirectional craz-
ing or cracking during drying of thin films of sols.
Silicon monoxide has long been known to oxidize to fibrous silica. Nemetschek and
Hofmann (102) investigated the material condensed from vapor from the reaction of
silica and silicon metal. Under the electron microscope it was found to consist of a
mat of remarkable hollow tubes and spiral fibers of amorphous silica less than 0.04
micron in diameter and many microns long. The morphology was compared with the
fibers of halloysite and chrysotile (103).
Silica W, an unstable crystalline silica fiber (65), as previously mentioned. is
converted to amorphous silica fibers by traces of moisture. These are very similar to
those obtained via silicon monoxide.
Fibers of amorphous anhydrous silica. I-50 microns in diameter. grow outward
from an electrically heated platinum surface exposed to nitrogen-diluted SiF 4 and
water vapor at 1100C, according to Haller (104). The mechanism of formation is
unknown. It may be that the silica is being deposited at the base of the fiber simply
because the temperature is highest at the platinum surface under the end of the liber
around which silica vapor is condensing. Once a nucleus of viscous silica is formed it
probably does not wet the platinum, so that surface tension pulls it up into a drop,
the outer, cooler portion of which solidifies and moves away. while more silica is
added at the hotter base. This suggested mechanism is consistent with the observe
I, .-








on of Silica

Phases of Silica

formed in
ext- .';vely
gh ~ . sity
zes to gel,
: g-1 so the
icrted that
all present
of the ultiz g-1 sio,
ions from
ie product
:e area (by

tion that there are often spherical beads on the end of the fiber and that the fibers
wave about as they are growing.

.med when
amount of
er solution
erface but
'med when
101). The

Common Amorphous Forms

The formation and properties of silicic acids. sols. gels. and powders are described in
subsequent chapters in detail, but a brief description of the nature of microarnor.
pho us silica should preface a discussion of solubility.
There are two broad classes, as follows.




h: ">een
iona .:raz-







tschek and
reaction of
onsist of a
j than 0.04
ed with the
ntioned, is
. similar to
w outward

SiF. and
rmation is
ber simply
)f the fiber
s formed it
Ito a drop.
re silica is
e r ':rvaj







1. Anhydrous amorphous silica particles formed at high .temperature, such as

"pyrogenic" or "fumed" silicas, are recovered from the gas phase as voluminous,
extremely finely divided powders. They are made in one of the following ways:
(a) Vaporizing silicon dioxide in an arc or plasma jet and condensing it in a
stream of dry inert gas.
(b) Oxidizing the more volatile silicon monoxide in the vapor phase with air and
condensing the SiO z
(c) Oxidizing silicon compounds in the vapor state, such as SiH SiCI or
HSiCI 3 , with dry oxygen or in a hydrocarbon flame.
When water vapor is present the surface of the particles may be partly hydrated
as SiOH groups.
amorphous silica particles are nucleated and grown from
aqueous solution supersaturated with monomer. Si(OH). The unique feature of
this system is that, unlike a solution of sugar from which sugar molecules
crystallize unchanged, Si(OH). does not crystal1ize as such. but must undergo a
dehydration reaction to form SiO z This reaction is slow at ordinary temperature,
so that Si(OH). can deposit SjO, on the surface of a growing quartz crystal only
from extremely dilute solution and at an extremely slow rate. If the concentration
of Si(OH). exceeds about 2 X 10- 3 M, condensation to polysilicic acids occurs. as
previously described. and colloidal particles are formed.
Ordway (105) has shown by molecular models that tetrahedral networks of silica
built up by the condensation process from Si(OH)4 are amorphous and are spherical
unless a nucleus of crystalline structure is initially present. However. it is obvious
that if the rate of condensation of Si(OH). onto the crystaline nuclei cannot keep up
with the rate of addition of Si(OH). to the system. the latter will accumulate until
amorphous networks are nucleated.
Under almost all conditions. soluble silica comes out of solution as spherical.
amorphous particles that. depending on concentration. temperature. and pH. remain
as a sol. are aggregated into a gel network. or arc coagulated as a precipitate.
As initially formed, the amorphous particles may contain some uncondensed
SiOH groups within the amorphous SiO z network. but in further processing these
are largely eliminated, leaving only SiOH groups on the surface.



The Occurrence. Dissolution. and Deposition of Silica




Microamorphous silica is not easily crystallized. When an ionic material such as a

salt is rapidly precipitated from a highly supersaturated solution. it may initially be
amorphous. but it rapidly rearranges to the ordered crystalline state. However. in
the case of silica in which bonds are largely covalent, such rearrangement can occur
only at elevated temperature or in the presence of a solvent such as water under
hydrothermal conditions. Silica. in effect, is a polymeric material. Walton (106) has
pointed out why an intermediate amorphous phase is likely to be precipitated if the
material is of high molecular weight or polymeric.
Carr and Fyfe (57) observed that amorphous silica in water crystallizes via cristobalite and keatite to quartz at 335C in 840 hr under 15.000 psi pressure. but in 18
hr at 45.000 psi.
Hydrated Amorphous Silica

. !

As will be seen, the highly hydrated silicas. which are generally stable up to 60C,
must be considered separately because they seem to differ in solubility from the
anhydrous or only surface-hydrated forms.
Two solid insoluble hydrated amorphous silicas were made by Signer and Gross
(107) by starting with cyclohexyl esters of the formulas



The first is a ring tetrarner: the second is a cage-like sphere of silica in which each
silicon atom is linked to one ester group. By removing the cyclohexyl groups with
anhydrous HI. they obtained white amorphous powders which were extremely hygroscopic, corresponding to
[(HOhSiO l,r


It can be seen that even if the low molecular weight silicic acid counterparts were
initially produced. condensation of silanol groups between adjacent molecules. forming siloxane bonds and water molecules. could occur rapidly. However. the water is
still held so tenaciously that the overall chemical analysis is not changed. while the
mass becomes cross-linked arid insoluble.
Natural opal may contain from 5.25 to 13.7% water. which is locked within the
structure and does not evaporate (108). Opal is generally nonporous in the sense that
it does not adsorb gases or liquids.
Highly hydrated silica gels which retain water tenaciously at ordinary temperature
can be obtained by reacting hydrated sodium silicate crystals with anhydrous acid so
that the silicic acid is liberated and simultaneously polymerized in a very dense state
Similar gels were obtained by hydrolyzing ethyl silicate in distilled water at 17C
to obtain a sol containing about 0.5% Si0 2 After a few days this became viscous as
aggregates were formed. which then formed a precipitate that was filtered off as a
wet gel and stored over water at 15-17C. Removal of all traces of HCI from this


)f Silica

ch as a
all" "~
:vel, .
1 occur
. under
)6) has
j if the
cristot in 18


Phases of Silica

ester retards hydrolysis and is not even desirable. Unfortunately, the pH" was not
recorded but was probably below 6 if traces of acid were present (110, Ill). When
the wet gel was dehydrated under vacuum at 11C, a plot of residual water versus
vapor pressure in the range from 7 to 1 torr revealed definite steps corresponding to
2.5, 2.0, 1.5, 1.0, and 0.5 molar ratios of water to silica.
From the vapor pressure of water at each stage of hydration, corresponding to
0.5, 1.0, 1.5, and 2.0 molar ratios, measured at a series of temperatures from 35 to
600C, Thiessen and Koerner found a linear relationship between log p and T.-I, and
calculated the average heat of hydration, h, per mole of water over the temperature
range TI - T2:

h = 2.3R
im the


where R
Pili' p/(.


gas constant
dissociation pressure of the hydrate at T T 2
vapor pressure of water at T h T2

h was 1.2-1.5 kcal mole-I H 20 .

It was mentioned earlier that Greenberg (13) found for the reaction



Si0 2(amorphous) + m H 20(l) = H.SiO.(aq)

j,H kcal mole:" = 2.65 0.28

is v..

ely hy-

s were
. form-ater is
rile the
lin the

se that
reid so
: 17C
ous as
ff ~ - "'
m '.

If one assumes the heat of solution of H.SiO. to be negligible and that m = 2, then
AH per mole of water would be 1.32 0.14, which agrees with the above value
obtained by quite a differentmethod by Thiessen and Koerner.
The formation of definite hydrates in a system now known to be amorphous is difficult to accept. The stepwise rem oval of rnonolayers of adsorbed water from a large
.fixed area of silica surface is more plausible if it can be assumed that the removal of
one layer, by coincidence. corresponds to about 0.5 mole of water per silicon atom.
If this is so. no formation of water by condensation of SiOH groups is involved, and
the above relation to Greenberg's data is only a coincidence.
as is discussed in Chapter 3, it is
Based on what is now known of
- this system.
likely that the sol made by hydrolyzing. ethyl silicate contains polysilicic acid or
silica particles that are so small that a substantial fraction of the silicon atoms are at
the surface of the particles and bear OH groups.
Initially, such particles aggregate into short chains to form a very open threedimensional network of gel filling the aqueous phase. As water is removed the structure gradually collapses. but at ordinary temperature there is a layer of water
probably several molecules deep associated by hydrogen bonding to the silanol surface. Thus a hydrated silica gel is obtained from which the vapor pressu re of water is
less than that of liquid water. The stepwise removal of water may correspond to the
removal of successive layers of water molecules from the silica surface. As this water
is removed from between the silica particles the gel structure shrinks and the parti-




The Occurrence. Dissolution. and Deposition of Silica

cles move closer together until the silanol surfaces corne into contact as alI free
water is removed.
This stage is probably reached in this particular type of gel at the point where
there are 2.75 moles water per SiO z, as shown in Figure 1.3. The water content of
the gel at 2 torr was plotted versus temperature (Ref. 110, p. 171). Since water
removal in the range 200-300C is known to involve dehydration of silanol groups,
the linear portion above 70C represents this process, whereas below this temperature it is only adsorbed water that is being removed. Therefore, the silanol content at
70 0 is about 0.27 moles HzO, or 0.54 OH groups per silicon atom.
From this the specific surface area of the silica can be calculated, since the area
per OH group is about 12.5 Az on a highly hydroxylated surface:

0.54 x 6 x 1023 x 12.5

60 X 10ZO

690 mZg- 1 Si0 2

When water is adsorbed on a silanol surface, there is one H 20 per SiOH group.
Hence a monolayer of water corresponds to 0.54 HzO per Si0 2 Inspection of the
steps in the adsorption curves (110) indeed indicates that each corresponds
approximately to 0.5 HzO per SiO z
Nevertheless, it must be admitted that by this theory it is difficult to explain
definite HzO: SiOz ratios of 1.0 and 0.5 identified in gels made by hydrolyzing
cyclohexyl esters of cyclic tetrasilicic acid, [(HOhSiO].. and of the polycyclic
decamer [(HO)zSi z0 3 ]a as described by Signer and Gross (107), or [(HOhSiO]z from
hydrolysis of SiSz by Schwarz (112).




"N 0.3





~ 0.1






Figure 1.3. Bound' water content of hydrated amorphous silica In vacuum at increasing
temperatures. [Data from Thiessen and Koerner (110).)

-~ ..


ion of Silica

Phases of Silica

as all free

On the other hand. removal of layers of water from an aluminum oxide surface in distinct steps has been noted by de Boer (113a). Thus the foregoing explanation cannot
be ruled out. since highly hydrated gels made by neutral or low pH and at room
temperature generally have specific surface areas in the range 600-800 m2 g-I.

oirn .. Ilere
content of
ince water
101 groups,
S temperacontent at

Biogenic Silicas

ce the area

JH group.
tion of the
to explain
iiC" 'rom



These are of many types found in widely different kinds of living organisms as
isolated particles. skeletal structures. and surface elements. In most cases, when
freed from organic matter, the silica exhibits characteristic patterns and shapes (see
Chapter 7). Essentially all biogenic silica is amorphous. It often has a substructure
of extremely small particles less than 50 A in diameter which have a surface of
SiOH groups. These are joined together in close-packed three-dimensional structures, some of which are isolated microscopic masses; others are solids permeated by
holes like Swiss cheese; and still others are like an interconnected mass of rods. The
ultimate particles in this size range can coalesce into denser structures as the
intervening pores become finer. Further deposition of silica can obliterate the
particulate appearance and lead to an impervious solid. Thus specific surface areas
observed in biogenic silicas vary widely from several hundred m2 g-1 to very low
values in those cases where the porosity has collapsed until the pores no longer
admit the nitrogen used for the measurement. These wide variations in structure and
porosity are also seen in synthetic silica gels but the latter do not exhibit the characteristic patterns usually seen in biogenic silicas.
The biogenic silica structure must be interpenetrated by organic tissue to effect the
deposition. Also, where the silica plays the role of a stiffening and strengthening
agent it is interlaminated or interpenetrated by biopolymer: in plants by cellulose. or
in diatoms by a protein, silicalemrna.
Many microcrystalline silica minerals such as flint, chert, and chalcedony can be
formed from biogenic silica by compaction and microcrystallization. Kieselguhr
(diatomite), which. was originally amorphous, is found in various stages of
transformation. The changes over millions of years can be noted, for example, by
the decrease in solubility and changes in other properties. Furthermore silica
undergoes continuous dissolution and biogenic redeposition in the oceans. Hurd and
associates (l13b-e) studied the properties of biogenic silica of ages up to 40 million
years, with emphasis on differences insolubility, density, and refractive index, comparing these with the values for known crystalline forms. which were thoroughly
Kastner, Keene. and Gieskes (1131) have examined the transformation of
amorphous biogenic silicas in the form of siliceous "oozes" on the ocean bottom to
partially crystalline geological deposits. They reviewed the complex background
literature and gave experimental data showing more rapid transformations when the
silica was embedded in carbonate (chalk). The remarkable 80 million year shrinkage
of the silica skeletons of large sponges to dense round boulders of flint in the chalk
beds of England is probably an example of this phenomenon.




Th_e Occurrence. Dissolution. and Deposition of Silica


Before the solubility of microamorphous silica, which is a basic factor in the
behavior of silica as a colloid, is considered in detail, the solubility of crystalline
phases of silica is briefly re~iewed.

Solubility of Quartz at Ordinary Temperature

Prior to the outstanding investigation by Van Lier, published in 1960 (114, 115).
there was speculation that quartz might exhibit no true equilibrium solubility in
water at ordinary temperature, although Gardner (116) reported 6 ppm in 1938.
Solubility data were very confusing until it was found that traces of certain metal
impurities and especially the presence of an amorphous or at least disturbed layer on
the crystal surface caused variable results, especially at temperatures below 150C.
In 1952, Dempster and Ritchie (117) reported that siliceous dusts have a layer of
high solubility that gradually blends into the solid core, which adsorbs basic dyestuffs (118). Alexanian (119) found by electron diffraction that quartz possesses a
surface layer of amorphous silica about 100 A thick. which is removed by HF but is
re-formed in ambient humidity. Waddams found that the quartz surface in water
released "mosaic" silica, presumably as particles of colloidal size, since they scattered light (120). This was confirmed by Sakabe et al. (121), who found that in neutral or alkaline aqueous suspension, quartz released both soluble silica and colloidal
particles of crystalline nature, 0.01-0.3 microns in size. Stober and Arnold (122)
found that the amount of silica released was much more than a monomolecular
layer, and that it decreased with successive changes of water. When quartz was
intensively pulverized in water, the disturbed surface layer can amount to as high as
35%, with a specific sur-face area of 70 m 2 g-l, and the solubility is increased from
less than 10 to 70 ppm at 25C (123). Paterson and Wheatley (124) made similar
The disturbed layer on ground quartz particles 1.5 microns in diameter was
examined by Koopmans and Rieck (125), by gradually dissolving it in dilute HF
while following the X-ray peak intensity. The layer thickness was 0.1-0.2 microns.
Similar studies by Lidstrom (126), who also used nmr, indicated a disturbed layer up
to 2 microns thick may exist, but upon aging in water and removal of the outer portion, the remaining underlying disturbed and strained layer returned to the normal
crystalline state.

Cleaning the Surface

Van Lier (l14) studied the dissolution of ground quartz of particle size 3-15 microns. He found a more soluble disturbed surface layer, 0.3 microns thick, which
could be removed by stirring 10 g of the powder into 50 ml 9-15% HF for 5 min to
dissolve 25% of the silica: then the residue was washed with 0.1 N NaOH and then
water to remove fluoride and alkali, and dried and stored in a desicator. When this
powder was suspended in 0.1 N NaOH at 26C, the rate of dissolution remained


of Silica


The Solubility of Silica


constant and was far less than for the untreated powder. More recently, Henderson.
Syers, and Jackson (127) have further studied the effects of HF treatment. Van Lier
found that cleaned quartz has a well defined solubility in water at autogenous
pressure, which can be ex.pressed by the equation:
log c

= 0.151 -


where c is the molar concentration of Si(OH).. and T is absolute temperature. Solubility in parts per million, as calculated from the equation, is shown as fine A in
Figure 1.4; at 25C it is II ppm.
From these data, the standard heat of dissolution is +5.32 kcal rnole", the free
enthalpy of dissolution is +5.11 kcal mole:", and the entropy change is zero according to Van Lier.
Morey, Fournier, and Rowe (128) obtained values somewhat lower than those of
Van Lier's. They found in one case that if quartz powder was tumbled in water at


Figure 104. Solubility of various forms of silica in water and salt solutions at .0-500C at
autogenous pressure unless otherwise noted. Quartz: A. Van Lier equation (dashed line) (114):
B, Morey (133): C. Morey-IOOO bars (133): D. Willey-in seawater (38): E. Mackenzie and
Gees (244b): F. Morey. Fournier, and Rowe (128). Cristobalite: G. Fournier and Rowe (137).
Amorphous: II. Stober (144): I, Elmer and Nordberg (153): J. Lagerstrom. in 0.5 M NaCIO.
(162): K. Willey (20): L. Jones and Pytk owicz (43); M, Goto (see also Figure 1.7) (167a); iV.
Okkersc (165); O. Jorgensen, in 1.0 M NaCIO. (161).




-The Occurrence, Dissolution, and Deposition of Silica

25C. the silica increased to 80 ppm after a year, and so was highly supersaturated.
and then dropped to 6 ppm, the true solubility. It is not known why the quartz surface already present could not accept the silica, but new nuclei had to be created to
reach equilibrium.
Siever (129) also investigated the solubility of quartz with results compatible with
those of Van Lier,

Solubility of Quartz Under Hydrothermal Conditions

It is impossible to review the extensive investigations in this highly specialized field.

Some of the publications are as follows.
Kennedy and associates developed data against which subsequent work was often
compared (25, 130, 131). In 1954, extensive studies were made by Wyart and
Sabatier (132), who measured the solubility of quartz, tridymite, cristobalite, and
vitreous silica in both water and steam phases at up to 480 bars and 470C.
Increased solubility of quartz with pressure had earlier been examined up to 600C
by Morey (133), and more recently by Heitmann (30); Anderson and Burnham (134)
examined solubility in water and salt and alkali solutions up to 900C and 6 kilobars. Solubility was only slightly reduced by the presence of salt, .but increased in
direct proportion to the base added. The solubility of quartz under hydrothermal
conditions is described in three papers by Kitahara (135) with special attention to
supercritical conditions up to -500C and 900 bars. The heat of solution of quartz
calculated from solubility data was 7.8 kcal mole:". In 1965, Heitmann (30) surveyed
the solubility of silica in water and stream and assembled data based on more than
1000 experiments up to 650C and 300 kg cm -2, on the basis of which a complete
solubility diagram was constructed. Verifying earlier work, the solubility of silica is
shown to increase with increasing density of steam or water, and reach a maximum
near the critical point of water. The thermodynamic properties and solubility of
quartz up to 600C and 5 kilobars pressure are being summarized by Hegelson (136).

. .~


Solubility of Cristobalite and Tridymite

There is some question whether true equilibrium can be established for cristobalite
and tridymite. There appears to be no instance where these crystals have been grown
under hydrothermal conditions in preference to quartz. Fournier and Rowe (137)
believed that precipitation of crystalline silica under hydrothermal conditions would
generally result in quartz. For this reason, in their study of the solubility of cristobalite, they left water in contact with the solid for long times without changing
temperature, so as not to supersaturate the solution any more than necessary. and
then suddenly quenched the system and measured the concentration of monomeric
silica, which was the only type of silica in the water. In Figure 1.4, line G strongly
suggests that a definite solubility value was established.


, ,
---. ..I


n of Silica

From the slope of the line,

rar: lJr
;re<t~_ ~


The Solubility of Silica



tible with

log S



zed field.
was often
'yart and
alite, and
i 470C.
to 600C
lam (134)
id 6 kiloreased in
ention to
of ouartz
lore than
)f silica is
ubility of
y Hegel-


where AH is the differential heat of solution of cristobalite. The solubility at 25C,

by extrapolation, is 27 ppm, or 0.45 millimolal.



From their corresponding data for the quartz system,



.en grown

= 4.58 kcal rnole "


= 5.46 kcal mole>'

Thus the free energy change going from cristobalite to quartz was estimated to be
-0.90 0.3 kcal rnole ", which is in reasonable agreement with 0.57 0.75 kcal
rnole" calculated from the heats of formation from elements:

196.582 0.30


iwe (137),

ins would
of cristochanging
sary, and
: strongly

Solubility of Other Crystalline Forms of Silica

For measuring the solubility of silica in water Stober (138) adopted a standard buffered salt solution (Ringer's solution) containing 0.9% NaCI. which expedites equilibration but does not affect solubility.ind 0.1 ~ NaHCO J which buffers the solution
at pH 8.4. where dissolution occurs 1110st rapidly without appreciable formation of
silicate ions that would occur at higher pH.

.':~'rl'; ,',

. ..;;... ...







Tire-Occurrence. Dissolution. and Deposition of Silica

Stober compared the solubilities of different crystalline forms using ground
powders of different particle sizes and treated with N aOH solutions to remove
disturbed material from the crystal surfaces. Dissolution behavior proved to be very
complex. Solubility in the sense of a dynamic equilibrium between silica in solution

and in the solid phase apparently exists only with crystalline quartz, and amorphous
or vitreous silica (139).



The dissolution of the other crystalline phases, crystobalite, coesite, tridyrnite, and
stishovite, with increasing concentration of Si(OH), is accompanied by adsorption of
Si(OH), on the crystal surface, which inhibits further dissolution. Thus the final
concentration of soluble silica depends on relative rates of dissolution and adsorption, and may reach the limiting solubility of amorphous silica.
Stishovite, which differs from the other forms by having a more dense octahedral
structure, with a higher surface concentration of SiOH groups. behaved most peculiarly. In 0.1 % HF solution it was completely insoluble, coesite dissolved slowly, and
quartz dissolved most rapidly. Yet in Ringer's solution quartz reached its equilibrium solubility of about 12 ppm in a few days, but coesite continued to dissolve for a
month as the silica concentration reached more than 25 ppm, and stishovite
dissolved rapidly until the soluble silica reached 190 ppm, at which point amorphous
silica was apparently nucleated and the silica concentration began to decrease (140).
Extensive further studies were summarized by Stober (141), and a theory of solubility was formulated.

Adsorbed Silica on Crystalline Silica

Holt and King discovered in 1955 that when powdered quartz was added to water,
the concentration of' dissolved silica, Si(OH)" increased rapidly at first but
thereafter very slowly. This was traced to a form of silica that amounted to 16% ora
monolayer on the crystal surface and that dissolved rapidly. Furthermore, when this
material was removed by treatment with alkali. the cleaned surface then adsorbed
soluble silica from solution, to reestablish the layer. Using Si(OH), containing
radioactive 31Si, they found that at pH 5 it did not exchange with the surface: at pH
9 it exchanged with adsorbed Si(Oti)" but not with the cleaned crystal surface
(142, 143).
Stober investigated the solubility behavior of several modifcations of silica and
developed a theory and equation for the behavior of adsorbed monosilicic acid in
retarding or preventing approach to true solubility equilibrium (144).
Stober believed that the interaction of Si(OH)~ and the silica surface involved the
following steps:

;. !

1. At the surface Si-O-Si bonds are split by hydrolysis. averaging two bondings per
tetrahedra. forming a hydrated silicic acid molecule adsorbed on the surface.
2. The Si(OH), is desorbed into solution.

3. Si(OH), is adsorbed onto the surface at equilibrium as the reverse process occurs.
followed by condensation and addition of Si0 2 to the surface.

C 1:




ion of Silica

The Solubility of Silica

.ng ground
to -ornove
I tc,

He formulated the following equations. For the adsorption of Si(OH). the equation


in solution


lyrnite, and .
'sorption of
lS the final
nd adsorp-

where nod. = surface concentration of adsorbed Si(OH).

no = surface concentration of complete mololayer
c = silicic acid concentration in solution
b = adsorption constant

most pecuslowly, and
its equilibssolve for a
ease (140).
iry of solu-

Adsorption occurs only on the bare surface not covered by Si(OH).:


where nH

d to water,
: first but
.016% ofa
. when this
n adsorbed
face: at pH
tal surface


1 + be

silicic acid released by hydrolysis per unit of surface area

kinetic constant for hydrolysis
activation energy of hydrolysis
gas constant
= temperature (0 K)

It is assumed adsorption equilibrium is established rapidly relative to the other steps.

Thus equation I can be inserted into equation 2:


- - - k exp
+ be



. Condensation involves adsorbed molecules only:


. silica and
cic acid in
ivolved the

JOdi ngs per


.ess occurs,

where n;

silicic acid condensed per unit of surface area

kinetic constant for condensation
activation energy for condensation

Since temperature is assumed constant and values of k, k', A. and A' do.not change.
new constants can be introduced:

k cxp



= k' cxp


-=fhe Occurrence, Dissolution. and Deposition of Silica

Combining equations gives






= -- +

-(K -K'bc)


1 + be

where n is the concentration of adsorbed monomer at time t.

The concentration of silicic acid, c, can be expressed in terms of volume of"
solvent, V, and total surface area of solid silica, F.


= Co -


+ bc)

+ -



where Co = initial concentration of Si(OH). in solution.

Differentiation gives the relation of changes in concentrations on the surface and
in the volume, respectively:


dt ==

l"F +

nob ] de
(I + be)2 dt

Substituting for dnjdt and integrating gives



[b InO + bc) _ Vbc _ bnoF + ,V(I - bctJ' In

I + be;
noF(l + bet:)

where c, is the equilibrium solubility concentration for the two-phase system soluble
silica plus amorphous silica. defined as c, = K] bK'.
The amorphous silica phase presumably might appear in solution as colloidal
particles or as a built-up layer at some points on the crystal surface.Btdber used the
experimental value of c, of liD ppm (micrograms of SiD, per milliliter). For no. the
value of 600 p.g m -, was calculated on the basis of 10 micromoles m -2 (l45) or 6
Si(OH). nm -2,
Experimental data were obtained by measuring. at many times over 3-4 weeks,
the amount of silica dissolved into 500 rnl of Ringer's solution at pH 8.2 at 25C
from IO m' of silica surface. Using particles of different sizes. Stober found a relatively constant rate of solution per unit of surface area.
For quartz. he used six different sizes of particles ranging from about 0.07 to 1.3
microns with specific surface areas of 33.6-1.7 m' g-I. From the curve of con centration of dissolved silica versus time. the constants for the above equation were
determined as


0.7 ml J.Lg- 1
1.7 day "

The final concentrations of soluble silica ranged from II to 13 ppm in excellent -",
agreement with Van Lier's value of It ppm (115) for the solubility of quartz.

.on of Silica


The Solubility 'of Silica

For vitreous silica, b = 0.7 ml J.Lg- 1 (as for quartz) but a much higher value for the
hydrolysis constant, K, indicated a much higher hydrolysis rate than for quartz.
With b = 0.7 ml J.Lg -1 assumed, the following values were found:

volume of _

urface and

em soluble
s colloidal
er used the
For no, the
(145) or 6

3-4 weeks,
.2 at 25C
und a rela-

0.07 to 1.3

arion were



. Stishovite





When the phases other than quartz and amorphous silica are suspended in water, the
final concentration of soluble silica is not a saturation concentration, but the result
of a competition between silica passing into solution as Si(OH). and Si(OH).
readsorbing on the surface to an ever-increasing extent, blocking further dissolution.
Thus a limiting concentration is reached that depends on the area of solid surface
exposed per unit volume of solution.
When the same sample of silica having an area of I m 2 is put each day into 50 ml
of fresh Ringer's solution (pH 8.4) for 24 hr, the amount of soluble silica dissolving
each day reaches a steady value, which is characteristic of the particular modification of silica used:

Vitreous (glass)
. Tridymite

Specific Surface
(rn" g-l)

Limiting Silica
. Concentration (ppm)



.. 8.5


If Stober's views are correct and cristobalite, for example, has no true equilibrium
solubility, then calculations based on published solubility cannot be valid. The need
for a more accurate definition of solubility has been pointed out by. Weill and Bottinga (146). Data presented as "solubility" of different forms of silica is often only
an indication of the rate of solution or the limiting concentration reached under
particular conditions. Thus a reportthat opal is 15-18 times as soluble as anhydrous
silica at pH 12.8 merely indicates the relative rate of dissolution (147).
Similarly, solubility data by Kopeikin and Mikhailov (148) for quartz (7 ppm),
cristobalite (12 ppm), tridymite (16 ppm), vitreous silica (88 ppm), and amorphous
. silica (120 ppm) are, in the cases of tridymitc and cristobalitc, higher than the limiting concentrations observed by Stober, no doubt because different techniques were



Jhe Occurrence. Dissolution. and Deposition of Silica

used. Of these. only the values for quartz and amorphous silica represent true solubility.
The relation of "rapidly soluble silica" on micron-range quartz particles to
"disturbed layer" was further investigated by Bergman. Cartwright. and Bentley
(149) with the conclusion that particles smaller than 1 micron in diameter contained
too much of the rapidly soiuble material to be explained by a monomolecular layer
of soluble silica. For example. 8% was rapidly soluble on 0.3 micron particles. It is.
clear that such "disturbed" material must be removed from quartz particles before
the behavior of adsorbed Si(OH). can be studied.
The dissolution and deposition of silica on quartz was examined by Beckwith and
Reeve (150), who noted an apparent equilibrium solubility of 18-20 ppm for HFcleaned quartz in 0.1 N sodium acetate solution at pH 7.7. The deposition of monosilicic acid was much faster at pH about 7 than in more acidic solution.
An unexplained phenomenon observed by Baumann (151) is that when cleaned
quartz powder was suspended repeatedly in a buffered pH 8.5 solution of 88 ppm
soluble silica, silica was adsorbed until the total far exceeded a monolayer coverage
(Figures l.Sa and I.Sb). The adsorption rate increased with increase of pH from 3 to
9 and was proportional to the second or third power of the initial concentration and
decreased with surface coverage. The kinetics show that the soluble silica is held not
by physical but chemical bonds. During seven repeated 40 day exposures to fresh 88
ppm solutions. the powder kept on adsorbing until four monolayers had been laid
down. The adsorbed silica was not ordinary crystalline quartz. because it could be
dissolved more rapidly than quartz. Yet it was much less soluble than amorphous
silica, because it continued to be deposited from a solution containing less than 30
. ppm of soluble silica, whereas the solubility of ordinary amorphous silica is 110- I20
Baumann proposed that the true solubility of quartz may be less than II ppm. He
thinks that quartz stops dissolving when the concentration of soluble silica reaches

















Figure 1.5a. Residual soluble silica in solution in contact with quartz surface after indicated
number of days at 25C. Curves are numbered according to the number or times the SOlution,.l
was changed. Initia! concentration was 80 ppm in all cases. [Data by Baumann {I 51 ).]

eposition of Silica

resent true solu-



rticles to
ht, and Bentley
meter contained
nnolecular layer
n particles. It is
particles before








>y Beckwith and

!O ppm for HFosition of mono-






at when cleaned

nion of 88 ppm
iolayer coverage
of pH from 3 to
incentration and
silicais held not
sures to fresh 88
rs had been laid
ause it could be
than amorphous
.in -ss than 30
silil._ .rs 110-120
han II ppm. He
lie silica reaches



z 0


Figure l.Sb. Number of layers of monomeric silica adsorbed on quartz versus the number of
40 day treatments at 25C. [Data from Baumann (151).]




















ice after indicated

tirr '. the solution

t, ,


Figure 1.5c. Concentration of soluble silica in solution in equilibrium with adsorbed

monolayers of silica at 25C: A. after 40 days: B. extrapolated to 100 days: C. solubility of
quartz. [Data from Baumann (151 ).1




The Occurrence. Dissolution. and Deposition of Silica

II ppm because at this concentration enough Si(O H). is adsorbed to stop further
dissolution at 25C. On the other hand, II ppm is the solubility value extrapolated
from solubility data at higher temperatures where it is known that true solubility
equilibrium is reached and that quartz crystals grow when the solution is supersaturated.
It is suggested that at 25C the adsorbed layers of silica are very slow to assume
the ordered structure of the underlying crystal lattice, but nevertheless crystallization
is slowly occurring just as it does more rapidly at higher temperatures.
A possible explanation for the above observations is as follows: the first adsorbed
layer of Si(OH)4 is held tirst by hydrogen bonds and then condensation by siloxane
bonds in a fairly regular arrangement almost matching the very regular array of
underlying silicon atoms. It is the regularity of the silanol groups on the quartz surface that permits almost perfect molecular matching and very strong adsorption of
the Si(OH)4' Each subsequent layer becomes less regular, and thus the adsorption
constant is lower and the apparent solubility of that layer becomes higher. Even in
the fourth layer, the regularity is still so pronounced that the solubility is still only'
about 25 ppm.
From a plot of Baumann's data for the concentration of silica in solution after 40
days versus the number of molecular layers of Si0 2 (Figure 1.5c) it appears that
after the second layer, the solubility increases with the number of layers as the
atomic arrangement becomes more disordered.
From a plot of the data in Figure 1.5a as parts per million in solution versus the
logarithm of time (days), the more nearly linear curves permit a reasonable
extrapolation to 100 days. As shown in Figure 1.5c the amount of silica in solution
continues to drop, suggesting that in all cases it is approaching the solubility of
quartz. It is therefore likely that if this quartz powder with multilayers of adsorbed
silica on the surface were left to age for a year or so instead of 40 days. or if the
suspension were heated to 100C, for example, the adsorbed material would become
further converted to the structure of quartz and its solubility decrease to that of

Solubility of Amorphous Silica

A given sample of amorphous silica exhibits a reproducible equilibrium solubility in
water. However, reported solubility values for amorphous silicas range from 70 to
more than 150 ppm at 25C. Such variations are apparently due to differences in
particle size, state of internal hydration, and the presence of traces of impurities in
the silica or absorbed on its surface during the measurements.
As a basis of comparison, it is of interest to note the solubility data for silica of
which the particles are large enough that size has no effect.
Based on data for smaller particles, an extrapolation of Alexander's (152) data
indicates the solubility of massive amorphous silica to be 90 ppm for silica from
pure sodium silicate. and 60-70 ppm for silica from commercial water glass. which
probably contains traces of impurities that reduce solubility. Similar extrapolated ~~'
values are described below (Figure 1. lOb).



.ion of Silica

The Solubility of Silica

top further
e ~. . Jility
; supersatu-

VitreouS silica has the same solubility as other amorphous silica. Because of the
mall specific surface area of powdered silica glass in comparison with that of
~icroamorphous or colloidal silicas, workers found it difficult to establish solubility
equilibrium. Stober (144) found that at pH 8.4 in Ringer's solution (0.9% NaC!,
0.1 % NaHC03 ) at 25C, at least 15 days was required to reach equilibrium when 20
m2 of silica surface was exposed per liter, regardless of particle size; Without the use
of this solution, which has an optimum catalytic effect, it would probably have been
impossible to establish equilibrium. The solubility was found to be about 100 ppm.
A highly porous, vitreous silica which equilibrated much more rapidly was used by
Elmer and Nordberg (153) in a study- of solubility in nitric acid solutions. Their
values for high dilution at pH 3 were 160 ppm at 36C, 260 ppm at 65C, and 400
ppm at 95C. They found the solubility was identical to that of a commercial dried
silica gel, provided care was taken not to abrade the gel by disturbing it during the
test. .
There have been several surveys of the solubility of amorphous silica. as follows.
In 1955. Baumann (154) concluded that amorphous silica had a uniform solubility
in the neutral pH region. that the dissolved species was almost entirely monomeric,
and that in the pH range from 3 to 6 the rate of dissolution increased linearly with
increasing pH. In the same year, Krauskopf (155) presented an excellent summary
of previous studies of the dissolution and precipitation of silica at modera,te temperatures, and emphasized the recognition of the differences between soluble. ionic, and
colloidal silica. relating these to the solubility behavior of silica in natural waters.

, to assume

st adsorbed
by siloxane
ar array of
quartz sursorption of
er. Even in
is still only'
ion after 40
ppears that
yers as the
versus the
olubiuty of
)f adsorbed
'S, or if the
uld become
: to that of

solubility in
from 70 to
fferences in
npurities in
for silica of
(152) data
silica from
;Iass. which
xt: 'Iated


Establishment ofSolubility Equilibrium

Equilibrium is established only very slowly. unless the amorphous silica is so finely
divided or microporous as to furnish an area of hundreds or thousands of square
meters of surface per liter of water. Stober (144) has shown that Si(OH). is adsorbed
on the surface of amorphous silica, retarding its dissolution. as has already been
described in the case of quartz. The surface of vitreous silica appears to consist of
SiOH groups. and if Si(OH). is absorbed by hydrogen bonding and then is partially
condensed the surface may then at least partly consist of =Si(OH)2 groups. referred
to as vicinal hydroxyl groups. Depending then on the state of the surface of amorphous silica before it is isolated and dried. there may be found different numbers of
hydroxyl groups per unit of surface area.
Likewise. the amount of silica passing into solution when a powder is first placed
in water may depend on its state before it was dried. However. this should not affect
the final equilibrium solubility of the silica.
According to Baumann (154). when amorphous silica powder (Aerosil) is placed
in water at 25C. the approach to equilibrium is different at high and low pH. Below
pH 7. the concentration of soluble silica increases for several days and approaches
the final solubility value asymptotically. Above pH 7. the silica concentration rises
rapidly in the first day to form a supersaturated solution containing. for instance.
155 ppm at pH 9. Then the concentration drops over a period of 3-4 days to the
solubility characteristic of the pH and type of silica. such as 125 ppm.
Baumann proposes that the initial rate of dissolution at this pH is greater than the


TheOccurrence. Dissolution. and Deposition of Silica


rate of deposition, thus leading to temporary supersaturation. and that there is an

oversupply of low molecular weight silicic acid, the excess of which then polymerizes
to form colloidal 'particles. This was shown by filtering the silica powder from the
. supersaturated solution and showing that the concentration of Si(OH)4 dropped
slowly to the equilibrium value at about the same rate as when the original solid was
Another possible explanation is that in the Aerosil powder, there might be some
extremely small particles only a few angstroms in diameter, which have a higher
solubility than the average larger particle.

Effect ofHearing

Silica gels made from sodium silicate and SiCI 4 respectively, were dried and samples
heated to a series of temperatures up to 900C. These all showed the same solubility
in water as reported by previous workers, according to Morachevsky and Piryutko
(156) (see Figure 1.4).

Solubility in Water: pH 0-8




In 1855, Struckman (157) concluded that amorphous silica dissolved in water at

ordinary temperature to the extent of 100-150 ppm. To this day, a single figure has
not been agreed upon because of the extreme variations in the form in which amorphous silica can occur. Curves of currently available data are shown in Figure 1.6.


























Figure 1.6. Solubility of amorphous silica versus pH: O. Alexander. 25C; ., Cherkinskii
and Knyaz'kova (160) 19C; Baumann. 20C; Baumann, 30C; Dashed line from Cherkinskii \ .
equation: log eM = -2.44 - 0.053 (pH).


.ition of Silica

The Solubility of Silica

t there is an
pc <erizes
ier ... In the
H). dropped
tal solid was

Alexander. Heston, and Her (158) were the first to show that microarnorphous
silica as powder made by collecting the smoke from burning SiH. or as colloidal
silica in sol form both exhibited about the same solubility. In retrospect, these silicas
probably had specific surface areas of 240 and 500 m 2 g " ', respectively, and the solubility at pH 7-8 was about 100 and 130 ppm, respectively. A silica gel, made under
conditions that we now know give a high specific surface area of more than 600 m 2
g-I, was soluble to the extent of 200 ppm at 25C. and the dissolved silica was
proved to be monomeric by chemical and physical methods.
Goto, Okura and Kayama (159) in 1953 showed that earlier data indicating a
minimum solubility at pH 3 were in error, and that solubility was constant from pH 2
to 8, then increased rapidly at higher pH.

ght be some
ive a higher

and samples
ne solubility
.nd Piryutko

in water at
le figure has
which amor'igu"~ 1.6.

Possible Solubility Minimum at pH 7

Baumann (154) collected data from other sources for comparison with the data of
Alexander plotted in Figure 1.6; these data seem to confirm that the solubility
reaches a minimum at pH 7-8, but the reason for the slightly higher solubility at
lower pH is not known. However. Cherkinskii and Knyaz'kova (160) verified the
data (see Figure 1.6) and proposed that silica is amphoteric and is cationic below pH
7. They give equations assuming that all the "soluble" silica is actually polymeric
and cationic at low pH. There is no experimental basis for such a theory: all of the
soluble silica has been shown to be monomeric, and there is no evidence for cationic
silica above the isoelectric point of pH 2.

Solubility in Nitric Acid

Elmer and Nordberg (153) have furnished data on solubility of porous vitreous silica
in nitric acid solutions up to 9 N:
log C = "";0.168N - 0.332
log C = -0.167N - 0.584
log C = ':'0. 184N - 0.796

at 95C
at 65C
at 36C

where C = concentration of SiO z (mg ml- I )

N = nitric acid normality
Thus in 7 N nitric acid at 95C the solubility was only about 30 ppm, as compared
to 400 ppm in water, whereas at 36C it was 8 versus 160 ppm. It was pointed out,
therefore, that to extract impurities from amorphous silica with nitric acid. strong
acid should
be used. The solubilities in water (N = 0) are included in Figure 1.4 at I.

Solubility in NaCiO. Solutions





A careful study by Jergenson (161) showed that there arc still mysteries to be
explained regarding the solubility of amorphous silica. ln spite of all previous work
indicating that the solubility in water is 100-120 ppm. and that solubility is not




TlreOccurrence, Dissolution. and Deposition of Silica

affected by the presence of salt, he reported 77.7 == 1.1 ppm at 25C in eM NaCIO.
solution. This value was reached only after 3-5 months, but was highly reproducible.
Jergensen suggested that in the presence of salt a uniform. reproducible. ordered
layer of silica hydrate is developed on the solid surface, and this is less soluble than
other forms of microamorphous silica: However, the lower solubility may be
explained instead by the larger radius of curvature of the surface of the final aged
amorphous silica.
Furthermore, he gave evidence that silica powders put into the perchlorate solution first formed a supersaturated solution in 10 days. The degree of supersaturation
depended on the amount of silica that had been added to the system. The supersaturation was higher with silica preheated to higher temperature (up to 700C).
However, silica that had previously been equilibrated with the solution did not cause
supersaturation and followed the dissolution rate




where C; = 77.7 ppm

C = concentration at time t
k = 1.2 x 10-12 mg ? hr " '
The solubility value of 77.7 ppm was obtained by approaching equilibrium from
the undersaturated and supersaturated sides.
Jergensen found that in water, equilibrium concentrations were variable. ranging
from 108 to 121 ppm. It was concluded that sodium perchlorate exerted a catalytic
effect in the formation of a uniform solubility-defining surface, regardless of the type
of silica used.
It would be interesting to determine if the silica that had been equilibrated with
perchlorate solution would then show the same low solubility when equilibrated in
water. Unfortunately, the effect of pH was not examined: raising the pH slightly
above neutrality should have reduced the time needed to reach equilibrium.
The mystery is deepened by the fact that in an investigation of the behavior of
silica and silicate ions in 0.5 M solutions of ,NaCIO Lagerstrom (162) concluded
that the solubility of amorphous silica is 0.002 M (I 20 ppm) at 25C and 0.0036 M
(216 ppm) at 50~C, which are the values obtained by other investigators (see Figure

. Effect of Electrolytes
The effect of electrolytes on the solubility of amorphous silica powder (750 m 2 g-l
over the pH range 7-11 was examined by Greenberg and Price (12). Up to 0.1 N,
NaCI or Na 2SO. had essentially no effect on the solubility or on the ionization
constant of silicic acid as measured around pH 10. Even the use of Ca(OH)2 solutions instead of NaOH, up to 0.04 N, did not change the solubility at a given pH in
the range from 8 to 10.





tion of Silica

The Solubility of Silica

M NaCl0 4
:prc- ~ib\e.
ole. .rered
soluble than
ity may be
e final aged

Silica gel dissolves in H:zSiF. in an amount to give a F: Si ratio around~5. Over an

acid concentration range from 0.03 to 3 M the ratio changes only from about 5.2 to
5.0, and is the same at 0 and 25C. A detailed study of the effect of acid strength on
this ratio was described by Ryss and Plakhotnik (163).
The solubility of amorphous silica at 40C in various strengths of solutions of
N a 2CO" K2CO" NaOH, and KOH ranging from 0.1 to 2.0 N as published by Drnitrevskii et al. (l64a) may prove useful in industrial research.

.lorate soluersaturation
e supersatuto 700C).
id not cause

brium from
ole,... .iging
I a catalytic

of the type

Solubility under Hydrothermal Conditions


As shown in Figure 1.4, the variation in the solubility of amorphous silica with
temperature is probably represented by a common slope for different samples under
the same experimental conditions. However, at a given temperature the values are
spread over a considerable range, for example, at 25C from 70 to 180 ppm.
The maximum temperature is around 200C since amorphous silica crystallizes at
. an increasing rate above this point, depending on what impurities are present.
Willey (20) measured the effect of pressure on the solubility of amorphous silica in
seawater at OC. The solubility increased from 65 to 71 ppm when ,the pressure was
increased from I to 150 atm and then linearly to 94 ppm at 1200 atm.
Pressure also has a marked effect on the rate of crystallization. Oehler (l64b)
found that silica gel crystallized to microcrystalline quartz (chert) even 'at 165C at
3000 bars.
Because of the rapid decrease in solubility upon cooling, sampling has been a
problem. Fournier and Rowe (l64c) devised improved means of quenching and dilut-. ing samples that had been equilibrated with silica at high ternperature and pressure
in the autoclave. Their data are expressed by the equation
log C = --" + 4.52

brated with

iilibrated in

p.H slightly
behavior of
) concluded
d 0.0036 M
(see Figure

where C = mg SiO, kg- 1 (or ppm) and T is the absolute temperature. According to
this equation the solubilities at the vapor pressure of the solution are as follows:
Ternperature (OC)

Solubility (mg kg " ')



Solubility under a constant pressure of 1039 bars is

:750 rn" g-'
p to 0.1 S.
: ionization

Jog C


+ 4.82


giwn pH in

The differential heat of solution, !},H. is 3.7\ 0.05 kcal rnole"' and the differential entropy. ;j,S, is 13.9 0.05 cal rnolc' '.


TOe Occurrence. Dissolution. and Deposition of Silica


Solubility of Hydrated Amorphous Silica



f. ~
;I .


, I



r: ~

There is a question whether silica in a dense but highly hydrated form is Jess soluble
than anhydrous amorphous silica. When silicic acid of low molecular weight. with a
particle size of only 10-20 A, is concentrated as soon as it is made at pH 2. at which
further polymerization is slowest, a very dense hydrated gel is obtained with pores
under 20 A diameter. The effect of pH in forming such gels has been discussed in
detail by Okkerse (165).
Gels can be made with pores so small that nitrogen molecules cannot enter. Iler
has prepared gels from polysilicic acid obtained by hydrolyzing ethyl silicate in an
alcohol-water mixture with HCI catalyst at 25C, and then diluting to 1% SiO z at
pH 2. The particle size was calculated to be 19 A based on the specific surface area
of 1405 m Z g ? by a titration procedure. Part of the sol was vacuum-dried and the
glass-clear silica gel had a specific surface area by nitrogen adsorption of only 45 m 2
Another part of the sol was adjusted to pH 6 and aged for I hr at 25C, then
readjusted to pH 2. The particle size had grown to 23 A as indicated by a surface
area of 1215 m 2 g-l. When this sol was dried in the same manner. the gel. which
looked the same as the previous one, had a specific surface by nitrogen adsorption of
626 m 2 g-l.
Thus when silica particles smaller than 20 A diameter are closely packed into a
gel, most of the pores between the particles are too small to be penetrated by
nitrogen. Such fine-pored gels are obtained only from very small particles that are
rapidly brought together before they can link together into a very open network.
At the point where such gels solidify, they still contain much water, but because
the pores are so small the vapor pressure is far less than that of free water. As
described in a previous section, these "hydrated" silicas sometimes appear to give up
water stepwise, as though definite hydrates were present.
When such gels are dried, the pores become so small that even nitrogen molecules
cannot penetrate, and the specific surface area by the BET method becomes very
small. It would therefore not be surprising if the rate of dissolution should also be
very low. However, there is no reason to believe that at equilibrium this form of
silica should be any less soluble than microamorphous or vitreous forms. In fact,
because of its hydration. since many of the silicon atoms are linked to hydroxyl
groups. the true solubility might be expected to be higher than the other forms.
On the other hand, Spychalski (166) reported that solubility decreased with'
increasing hydration. He gave the following tentative "solubilities" of hydrated
silicas at various stages of dehydration. as prepared by Thiessen and Koerner (I 10.
Ill) by hydrolyzing ethyl silicate:
Si0 2 2.5H 20
Si02 2.0 H20
Si0 21.5H 2 0
Si0 21.0H 20
Si0 2 O.5 H20

"Solubility" as ppm in
Water at 18-2rC




on of Silica

ess ~ .ubte
ght, with a
2. at which
with pores
iscussed in
enter. lIer
icate in an
% Si0 2 at
irface area
ed and the
only 45 m 2
!5C, then
I a surface
gel, which
sorption of
:ked into a
etrated by
es that are




The Solubility of Silica

However, there is no indication that solubility equilibrium had beef reached.

Nevertheless, there is a consistent relation in that the "solubility." as he measured it,
was inversely proportional to the state of hydration.
Possibly in microporous gels in which water is strongly hydrogen-bonded in the
pores at low temperature, the solubility would be much greater at higher temperature, where hydrogen bonds are less stable. Okkerse (165) measured the rates of
solution and solubility of microporous (717 m 2 g-l) and macroporous (293 m 2 g-I)
silicas in a veronal-NaCI-NaAc-HCI buffer at pH 2, 4, 6, and 8 at 60C. In every
case, the rate of dissolution of the microporous gel was at least twice as fast as the
other. After 200 hr at pH 4-8, both gels exhibited the same solubility of about 200
ppm, which is close to the value for other amorphous silicas in Figure 1.4.
Apparent High Solubility at High pH
From pH 9 to 10.7, there is an apparent increase in the solubility of amorphous
silica, owing to the formation of silicate ion in addition to the monomer which is in
equilibrium with the solid phase. Since the silicate ion is instantly converted to
monomer in acid solution, both monomer and silicate ion are included in the
determination as "soluble silica" by the molybdate reagent. In this range, amorphous silica is in solubility equilibrium with neutral monomer as well as silicate ions:
(Si0 2) % + 2 H 20 = Si(OH). + (Si0 2) %_ 1
Si(OH). + OH- = Si(OHh- (or HSiOi)


ut because
water. As
. to give up
: molecules
.ornes very
rld also be
.is form of
IS. I n fact.
J hydroxyl
eased with
f hydrated
erner (110.

Above pH 10.7, all the solid phase of amorphous silica dissolves to form soluble
silicate, since at higher pH the concentration of Si(OH). _is greatly lowered by
conversion to ionic species. so that no amorphous solid can remain in equilibrium.
Measurement by Alexander, Heston, and lIer (158) on the solubility of a very
pure amorphous silica (Linde silica) in water at different pH values (adjusted with
HCI or NaOH) show the following variations. as measured by the molybdate

Solubility of Amorphous
Silica (25C) (ppm)

The increase in the total "soluble" silica can be explained on the basis of the
following equilibrium. assuming that the concentration of Si(OH). in solution does
not change with pH:

+ 01-1-

H 3SiO; + '-1 20

[H 3SiO.1 = 1.85 x 10 [Si(OII).][OH-1



"Fhe Occurrence. Dissolution. and Deposition of Silica

The equilibrium constant calculated from these solubility data agrees reasonably
well with that found by Roller and Ervin (23) in a study of the association of silicate
ions in the CaO-Si0 2-H 20 system. These authors found at 30C a value of 1.5 x
10, Thus it is clear that the solubility of silica increases at high pH because of the
formation of silicate ion in addition to Si(OH). in solution.
Subsequent measurements have been made with additional precision.
At 90C. Van Lier (115) measured the total amount of silica dissolved at equilibrium from quartz in 10- 1 to 10- M NaOH and noted the change in pH. From these



Since the ionization constant for water at 90C is 12.42.

[OH] [Si(OH).]


103 2

1.6 X 103

The effect of pH on solubility of silica at temperatures from 0 to 200C was

measured by Goto (t67a) and is shown in Figure 1.7. At 22-100C the solubility was
about 30% higher than for most powders and gels. as shown in Figure 1.7. This is


iii 1000



















Figure I. 7. Solubility of amorphous

silica versus pH at di Iferent tern peratu res
[from Goto (167a)].

ion of Silica

Effect of Particle Size on Solubility in Watee., -

of ~;Iicate

undoubtedly because the colloidal particles were smaller than 50 A. since they were
prepared at room temperature from sodium silicate by ion exchange. However, at
155 and 200C in water, the particle size of the silica undoubtedly increased and the
solubility was the same as reported by others (see Figure 1.4).

ie I.


ause of the


Calculation 01 Solubility and Dissociation Constant

at equilib- _
From these

Where the ionization constant and solubility are not known, as for example at some
unusual pH. Van Lier (114) has shown how the ion concentration and solubility can
be calculated easily from data on pH and total soluble silica. which includes both
monomer and ionic silica as determined by the molybdic acid reagent. In general
terms their method is as follows.
Let S, = concentration of total soluble silica at pHI
S2 = concentration of total soluble silica at pH 2
So = concentration of Si(OH) which is the solubility of silica in neutral water
Then concentration of ionic silica at pH 1 = S1 - So
concentration of ionic silica at pH 2 = S2 - So
Since the decrease in H ion concentration is accompanied by a corresponding
increase in concentration of H 3SiO. -. then

20r - .... was

lub.. j was
1.7. This is
By rearrangement.

The concentrations of silicate ion can be obtained by difference and the ionization
constant calculated. using the appropriate value for the ionization constant of water:





The behavior of polysilicic acids. colloidal silica.ind silica gels cannot be understood without taking into account the fact that the solubility or silica is higher when

The-occurrence. Dissolution. and Deposition of Silica

the silica surface is convex. and lower when it is concave. It is a matter of'the radius
of curvature of the surface; the smaller the radius the greater the effect on solubility
As shown in Figure 1.8 smaller particles with a smaller positive radius of curvature have a higher equilibrium solubility. On the other hand. in a crevice. such as
where two particles are in contact, the radius of curvature is negative and the equilibrium solubility is low.
There are two important practical consequences:



2. If there has been an aggregation or flocculation of colloidal silica particles. so

that two or more particles are brought together, then at the point of contact the











Figure 1.8. Variation in solubility of silica with radius of curvature of surface. The positive
radii of curvature are shown in cross-section as particles and projections from a silica surface.
Negative radii are shown as depressions or holes in the silica surface. and in the crevice
between two particles.

1. When very small individual silica particles are brought into the same solution as
larger ones. especially at pH 9-10 where hydroxyl ions catalyze dissolution and
deposition of silica. the smaller ones dissolve and the larger ones grow.







Effect of Particle Size on Solubility in WatCI- -

on of Silica

radius of curvature is negative and extremely small, the solubility of silica in this
region is very low, and silica dissolves from the particle surfaces and is deposited
around the point of contact to minimize the negative radius of curvature, thus
forming a coalescence or neck between particles.

the radius
\ s('I-'l,ility
; of curva.e, such as
j the equi-

Even at low pH the same phenomenon occurs if the particles of silica are less than
about 5 nm in diameter, so that the solubility changes rapidly with radius of curvature.
The coalescence between adjacent silica particles in an aggregated gel structure by
this spontaneous process or by adding soluble silica to be deposited at the points of
contact between particles is described by Alexander, Broge, and lIer (168) and is
further discussed in Chapter 5.
Kitahara (169) shows that the effect of pH, salt, and temperature have the same
influence on the rates of polymerization of monosilicic acid as they have on the gelling of sol, showing that both phenomena, that is, the growth of particles and the
cementing together of particles once they are in contact, are influenced by the same
factors; that is, those jhat affect the rate of dissolution and deposition of monomeric
Greenberg (13) concluded that although theoretically solubility is a function of
particle size, there was no published data to bear this out. Alexander (152) was the
first to obtain data that showed that for a given type of silica, solubility increased
with decreasing size. Silica containing different impurities or having different
degrees of hydration within the particles cannot be used for comparison.
The. Ostwald-Freundlich equation, applied to solubility (known as the
Thompson-G,ibbs effect), is as follows:

.olution as
lution and
irticles, so
.ontact the

. 5,
= exp(2 EVR -IT-1r 1)
. 5,



where S, = solubility of particle or radius r

5, = solubility of a nat surface or particle of infinite radius (nm)
E = interfacial surface energy (ergs em -2)
V = molar volume = 27~2 ern" for amorphous silica
R= gas constant (8.3 X 107 ergs mole ? deg ")
T ~ temperature (degrees absolute)
." r = radius of curvature (cm) .
. d= particle diameter (nm)
E .: I04.6(J0 7 )(r )log(5 , / 5 ,) at 25C


The positive
lica surface.



. s:

Jog l o



= 2.85 x 10- 7













! ..

r '






i I

r I

i I




Theoccurrence. Dissolution. and Deposition of Silica

There have been past indications that very small silica particles are ibnormally
soluble, but no measurements were made.
In studying the solubility of finely ground quartz, Stober and Arnold (122) concluded that the silica which dissolves very rapidly at first when immersed in water is
much more than an adsorbed layer of Si(OH)., but instead is a minute fraction of
the powder which is finer than 0.1 micron and is thus, owing to the
Thompson-Gibbs effect, much more soluble.
The radius of curvature of the silica-water interface is of critical importance even
with a porous silica solid. Charles (170) found that the rate of dissolution of porous
high silica glass could be explained on the basis of a high local solubility of the silica
surface owing to its small radius of curvature.
Over millions of years the conversion of the amorphous siliceous remains of large
sponges to solid rounded flint boulders in chalk beds is an extreme example of the
conversion of a high surface area form of silica, through dissolution and deposition,
to a dense low area form. Examination of flint boulders with inclusions of oyster
shells and belemnites indicates that a once highly extended skeleton of a sponge had
been drawn together into a round black boulder (171). Between the flint and the
CaC03 there is a film 10 nm thick of intermediate hydrated calcium silicate, along
which the soluble silica must have been transported.
Another mysterious phenomenon is "ambient pyrite," Grains a f~w microns in
diameter have moved through solid chert (an extremely tine-grained quartz). leaving
a trail of coarse-grained quartz. This was described by Tyler, Knoll. and Barghoorn
(l72)(Figure 1.9).
One explanation is that there is a slow transport of silica from the more soluble
tine grains ahead of the crystal of pyrite to the growing quartz crystals behind the
pyrite. The pyrite is hydrophobic in nature and not chemically bonded to the surrounding silica. Thus it is possible that the pyrite grain is pushed ahead by, the growing quartz crystals. The resulting pressure ahead of the grain on the finer crystals of
quartz must also increase their already higher solubility. However, organic matter is
known to be present, and it is postulated that gas evolution and pressure buildup are
also involved..
The particle size effect is probably also involved in a phenomenon described by
Baumann (173). Amorphous silica powder condensed from a name consists of fine
spherical particles generally less than ISO A diameter. When such a pov..' der is placed
in water, a supersaturated solution of silica is obtained, no doubt owing to a very
small percentage of more soluble particles under 50 A diameter. The dissolved
monomeric silica then rapidly polymerizes to polysilicicacids. but these later disappear as supersaturation is relieved by deposition of silica on the larger amorphous
particles in suspension.
Agitation of granular silica in water may cause abrasion. liberating very fine
particles which then give erroneously high solubility data. Morey. Fournier. and
Rowe (16) measured the solubility of amorphous silica in the form of commercial
pure silica gel, crushed vitreous silica (fused quartz), colloidal silica formed by cooling a supersaturated solution (720 ppm) dissolved from quartz, and colloidal silica
from supersaturated hot spring waters. The silica gel and colloid from supersaturated solutions showed a reproducible solubility of 115 ppm at 25C. However. silica

\ .


--~ ..

:ion of Silica


I (I.., , conl in water is

fraction of
ng to the

rtance even
n of porous
of the silica
tins of large
mple of the
1S of oyster
sponge had
.int and the
icate, along
microns in
rtz), leaving
I Barghoorn
behind the
to the surly the growr crystals of
lie matter is
buildup are


lescribed by
isists of fine
Ier is placed
19 to a very
ie dissolved
later disapamorphous
19 very Ii ne

rurnicr. and
ned by coollloidal silica

-- .<. :.,:...



;,;-'.-;-' ~.



-. .




...... .









~ ... ~.!:. -.'.

..: .


.:,:;:'i-t~.~;,~:?,_:<;::':" ''C-'~';r :=J


(b) :


"; :;~,~;;



.~_~ __~:-'~-~~J

Figure 1.9. Ambient pyrite grains in chert. (a) Biwabik Formation. Mesabi Range. Finegrained dark material: organic matter. Black grains: pyrite. Trails: coarse-grained quartz.
Matrix: line-grained quartz (chert) (200x). (h) Magnified pyrite crystal and trail of coarse
quartz grains (2200x). (Courtesy of Professor E. S. Barghocrn, Harvard Univcrsity.)

rsatuwever, silica

1 ~


The Occ~rrence. Dissolution. and Deposition or Silica


glass evidently underwent self-abrasion during the continuous tumbling. generating

subcolloidal fragments of extremely high solubility, so that the monomer level
reached more than 300 ppm.

Interfacial Energy

II'1 "

i .!

I .

One problem in calculating the solubility of small particles is that the surface energy
is not accurately known. There are very few substances of which the solubility of
particles of uniform size and less than toO A diameter have been measured. According to Walton (174), a combination of these requirements has not previously been
achieved, but recently using radiotracer measurements and the electron microscope.
such measurements have been made on strontium sulfate particles, from which the
solid-water interfacial energy was calculated to be 84 8 ergs ern -2.
In the case of amorphous silica, Alexander's data gave E = I.l X 105 cal ern -2, or
46 ergs em:". For purpose of calculations, Iter (175) used both 80 ergs em -2, based
on some preliminary experimental data, and 133 ergs em -2, based on the surface
tension of glass extrapolated to zero alkali content (176), 275 ergs ern -2, minus the
energy of wetting of the siloxane surface by water, 142 ergs ern -'.
In the case of quartz, the interfacial energy was taken to be the surface energy of
the crystal, 980 ergs em -2 (177) minus the heat of wetting, 564 16 ergs em -2 (178),
leaving an interfacial energy of 416 ergs cm -'.
As discussed earlier under the topic of thermodynamics, calculations based on
data now available involve small differences between large numbers, and about all
that can be said is that the interfacial energy of the amorphous SiOH-H 2 0 surface
is low and probably of the order of 50-100 ergs crn "".
Some further preliminary experiments by lIer have given the results shown in
Figures 1.10a and b. On commercial silica sols made from the same type of sodium
silicate, but not all at the same temperature or necessarily containing the same trace
impurities, line A was obtained. On sols of very small particles made by polymerizing pure silicic acid for various times at pH 8 and 25C, then stabilizing at pH 2.2
and aging to a constant monomer concentration, line B was obtained.
The particle size was calculated" on the basis of specific surface areas determined
by a titration procedure, and assuming the particle is anhydrous SiO,. The values of
E, the interfacial energy calculated from the slopes of the lines A and B, using the

E = 0.1754Tdlog



are 54 and 46 ergs em -2.

In view of the scatter of data for line A, it is quite possible the slope could be the
same as that of line B, in which case both would give a value of E, the same as
reported earlier by Alexander, 46 ergs cm- I
The difference between the two groups of sols is that in making the commercial \"





tion of Silica

. generating





















Figure 1.10a. Relation between solubility of amorphous silica at 25C in water and particle
size at pH 8: A. particles made at 80-IOOC; B, particles made at 2S-S0C.
:E 1000






en .500

s::c: 400


~ 300



~ 200





Figure 1.10b. Relation between log solubility of amorphous silica at 25C in water and reciprocal of particle: size: at pH 8: A, particles made: at 80-IOOC: 8, particles made at 25-S0C.


The Occurrence, Dissolution. and Deposition of Silica

sols (line A). particle growth was carried out at temperatures above 80C. whereas
the particles made by aging silicic acid (line B) were all grown at lower temperatures
and probably are internally more hydrated with trapped silanol groups.
A dependable value for the interfacial energy remains to be determined for both
amorphous silica and quartz. '
Since the interfacial surface energy for the quartz-water surface may be as high as
416 ergs em:", 10 nm quartz particles may have a solubility of 28 ppm, and 3 om
particles. 93 ppm (18).






Certain impurities such as aluminum in minute amounts not only reduce the rate of
dissolution of silica, but by chemisorption on the surface of silica, even in amounts
less than a monomolecular layer. reduce the solubility of silica at equilibrium. Jephcott and Johnston (179a) have shown that the apparent solubility of amorphous finely
divided silica in water, which they lind to be 0.017% at 37C. is reduced to
0.003-0.0097%' when aluminum' oxide is added to the system and to less than
0.0001 % when powdered aluminum is present. Since it was also shown that the addition of silica to a suspension of alumina depressed the solubility of alumina. it seems
likely that a surface containing both Si0 2 and AI 203 is formed on both the silica and
alumina phases and has a lower solubility than either oxide. Earlier Denny. Robson.
and Irwin (I 79b) had shown similar effects of iron and aluminum.
This observation emphasizes the point that in attempting to measure the solubility
of silica in the form of coarse particles, where the surface area of the silica in the
system is rather low. it may be very difficult to obtain an accurate measure of solubility unless traces of aluminum and other metals forming insoluble silicates are
ridigly excluded from the system.
The fact that the amorphous silica on the ocean floor, found in enormous deposits
of skeletons of diatoms. does not dissolve in seawater. has been a mystery. Lewin
(180) found that as soon as these organisms died. silica began to dissolve. but only
slowly. and the concentration in solution reached only 30 ppm. Removal of organic
matter had no effect, but treatment with hot concentrated nitric acid greatly
increased the rate of dissolution. as did treatment with mixed oxalate and EDTA
solutions at pH 6.8. This clearly indicated certain metal ions were retarding the
dissolution of silica in tris buffer at pH 9.0.
The nitric acid-cleaned silica. which was free from organic material as well as
metal ions. was immersed in 4-10 mi\{solutions of selected metal salts at a series of
pH values and washed. and the rate of dissolution was measured in a standard manner at pH 9.0. Metals with no effect at pH 2-9 were La. Mo. and Cr. and at pH
4-11 were Ca and Mg: at pH 8. which is that of seawater. AI. Be. Fe. G~. Gd. and Y
all retarded dissolution. However. Al was unique in that when applied over the range
from 5 to 9. it rendered the silica completely insoluble at pH 9.
The amount of aluminum on the silica surface required to reduce the solubility of
silica has been measured by Iler (ISla). When only one aluminum atom was
absorbed on the surface as an anionic aluminosilicutc site per 2 nrn", at which point


.... .;..._n of Silica

i for both

high as
and 3 nm


he rate of
irn. Jephous finely
-duced to
less than
the addi it seems
silica and
e of solucates are
5 deposits.
'y. Lewin
but only
If organic
j greatly
rding the

s well as

series of
ard mannd at pH
id, and Y
the range

ubility of
tom was

Effect of Impurities on Solubility.


only 5% of the surface was occupied. the rate of dissolution as well as \he equilibrium solubility of the surface were drastically reduced. The surface of colloidal silica
in the presence of an aluminum citrate complex became saturated with aluminum at
pH 8 when 1.4 aluminosilicate sites nm-z were present (about 25% coverage) and the
solubility of silica was reduced to 10 ppm. The effect of extremely low concentrations of aluminum ion on the solubility of quartz was reported by Baumann (18Ib).
The quartz surface was first equilibrated with a buffer solution at pH 8.5 at 25C
until a concentration of about 6 ppm SiO z was reached. Then to different suspensions, different amounts of aluminum ions were added at concentrations of 3-60 J.l.M
(0.15-3.0 ppm equivalent of Al z0 3 ) . Over a 10 day period the amount of silica in
solution dropped drastically. Even with only 3 J.l.M aluminum present the soluble
silica decreased to 3 ppm SiO z, and with 12 J.l.M to zero. It is significant that similar
amounts of iron had no effect. Perhaps this should not be surprising when it is
recalled that quartz often is found in association with iron oxide.
Lieflander and Stober (182) found that when the surface of colloidal silica powder
was covered with an amount of aluminum corresponding to 3.6 aluminum atoms
nm -z, that is. about hal f the surface covered. the silica was essentially insoluble.
The effect of impurities AP+, Ca!", Mgz+. SO/-, and F- on the solubility of
silica gel in water and 0.1-2.0 N solutions of Na ZC0 3 , KZC0 3 NaOH, and KOH
were studied by Dmitrevskii (183). The ions F- and SO/- have no effect on solubility, but the polyvalent cations decreased solubility or led to formation of insoluble
The effect of impurities on the solubility of glass in alkaline solutions has been
discussed by Weil and Marboe (184). Calcium, zinc, and aluminum hydroxides at
low levels greatly retard attack by alkali. It is interesting that in strongly alkaline
solution. silica offsets the effect of these hydroxides, no doubt by removing them
from solution as insoluble silicate. Thus under certain conditions the addition of
silica to a solution of alkali can increase its rate of attack on glass. Similarly. traces
of iron or copper from screens used to sieve glass powders can reduce the rate at
which water attacks the glass and leaches alkali.
Boehm and Schneider (185) caused alumina to be adsorbed on the surface of a
pyrogenic silica by suspending it in a dilute solution of aluminum chloride and
adding enough NaOH to neutralize two-thirds of the chloride. thus forming basic
aluminum chloride. The silica surface adsorbed one Al atom per SiOH group and
the solubility of the silica dropped from 123 to 6 rpm.
A practical application of this phenomenon is to- insolubilize the silica in a porous
glass membrane at low pH by including 0.3 g I-I of AICI 3 6H zO in the feed solution.
or by treating the membrane every 100 hr with the aluminum salt. Iron or zirconyl
chloride was not as effective (186).
The dissolution of silica from glass in water (187). from biogenic amorphous silica
in seawater (18S). and from kaolin. talc. and mica dusts in Ringer's 'solution or
serum (189) were all accelerated by removing metal ions from the surface hy extraction with acid or cornplexing with citrate ion in neutral solution.
The reaction of-silica gel with MgCI 2 at p H 9A to form magnesium silicate evidently involves the dissolution or silica. Smirnovu. Dushina. and Alcskovskii (190)


The Occ~r~ncc:. Dissolution. and Deposition of Silica

found that introduction of aluminum into the gel (SIO z:AI 2 0 3 = 2: 1) permitted
only adsorption of magnesium onto the surface. but none reacted.
Magnesium ion insolubilizes amorphous silica only to the extent that it converts
silica to magnesium silicate. In the case of Quartz it combines with the amorphous
silica associated with the surface: and thus lowers the solubility at least to that of
pure quartz (191).
.Calcium ion has no effect on the solubility of silica up to about pH 9.S. where calcium silicate begins to form (192).
The effect of phosphoric acid on silica has been examined by Mitsyuk (l93), who
found that H,P04 is adsorbed on silica gel from water. The amount corresponds to
about one H 3P04 molecule per surface silicon atom, indicating formation of a surface compound of silicon phosphate. Once treated with as little as 0.06% H,P0 4 , the
rate of dissolution of the silica in water is greatly reduced. The addition of H 2 S 0 4 to
the dilute H,P0 4 solution prevents the formation of an adsorbed film and accelerates
dissolution. In strong H,P0 4 there is further reaction as silica is converted to SiP z0 1
in which silicon is coordinated with six oxygen atoms.
Lithium hydroxide differs from NaOH and KOH in the reaction of concentrated
solutions with silica (194). In 1 N solutions all three bases dissolve microamorphous
silica at about the same rates, but in 2.2 N solutions LiOH does not react. Also: the
addition of LiOH to strong solutions of the other alkalis retards the rate of dissolution of silica. apparently by the formation of an insoluble film of lithium silicate on
the silica surface. It should be pointed out that lithium silicate is soluble in water
below about 60C, and the system acts 'much like the other alkali silicates. but at
higher temperature lithium silicate is insoluble, and comes out of solution as a gel or
precipitate. It appears that the less hydrated form of the lithium ion behaves more
like calcium does.
Fluoride ion affects solubility only at low pH. and then only to the extent that it
- converts silica to SiF.z- ion.
Pressure increases the solubility of amorphous silica in seawater at oc. Willey
(20) reported the following: at 1 atm the solubility as SiO z was 65 ppm. This
increased to 71 ppm at 150 atm and thereafter linearly to 94 ppm at 1250 atm.










Broadly speaking, organic compounds can either retard dissolution by covering the
silica surface with a strongly sorbed liIm or accelerate dissolution by removing the
soluble Si(OH)4 that is in equilibrium with the surface by converting it to a soluble
The simplest example of the former is the effect of treating a pyrogenic silica with
trirnethylchlorosilane, which coats the surface with a chemisorbed monolayer of
SiOSi(CH 3)3 groups. In undisturbed suspensions of treated and untreated powders.
the rate of dissolution of the treated material in neutral 1% NaHCO J solution at
25C is only 15% as fast as the untreated. When 'shaken. however. the coated silica
aggregates are apparently broken apart. leaving bare spots on the ultimate particles
which then dissolve much faster. The difference is noted mainly below pH 7 (195).


, of Silica

Effects of Organic Compounds on Solubilit)'--


Catechol and Related Compounds

co., .ts
o that of
here cal93), who
ponds to
of a surjPO., the
12S0. to
o SiP 20 7
~Iso: the
, dissolulicate on
in water
s, hilt at
/es more
it that it

A 6-coordinated structure is suggested for the complex anion formed between catechol and silica, reported by Rosenheim, Raibmann, and Schendel (196). The ammonium salt, crystallized from, alcohol, has the composition (NH.)2Si(C,H.02)3 and
dissolves in water without decomposition of the anion. The compound is prepared by
boiling freshly precipitated silica in an ammoniacal solution of pyrocatechol out of
contact with air. Since silica is not soluble in ammonium hydroxide (because silica
does not form silicate ions below about pH 10.8). it is evident that the combined
action of ammonia and catechol converts silica to some soluble form other than a
simple silicate ion. The corresponding potassium, barium, guanidinium, and
pyridinium salts were prepared.
Numerous investigations have shown that o-dihydroxy aromatic compounds are
unique in forming complex ions with silicon that are not hydrolyzed in aqueous solution, but are destroyed by atmospheric oxygen, forming dark insoluble residues.
polynuclear complexes can also be formed and many salts, especially of organic
bases, have been prepared from a variety of silicon tricatecholate derivatives. Weiss,
Reiff, and Weiss (197) found that catechol solution will attack quartz and dissolve
amorphous silica at pH 7-8.' If the partly dissolved silica is washed, it retains
strongly adsorbed catechol.
Vlacil and Bock (198) have shown that silica can be extracted from the aqueous
. phase with nitrobenzene if the concentration of catechol in the water is 0.5 M. that
_of diphenylguanidine in the organic phase is more than 0.25 M. and the pH is
suitably adjusted. By this means silica can be concentrated for analysis.
, Her examined aqueous ammonium catecholate as an etching solution for different
forms of silica. In a mixture of 20% catechol, ~O% concentrated ammonium
hydroxide (28% NH 3). and 40% water. vitreous silica dissolved at a rate of 0.5
Quartz dissolved at about half that rate.
micron hr- I at 90C (0.2 mg cm -2 hr

:, Willey
m. This

:ring the
,ving the
1 soluble
lica with
layer of
ution at
ed silica





Polyhydroxy Organic Compounds

Since boric acid forms stable complexes with polyhydroxy aliphatic compounds such
as glycols and polyalcohols, these common compounds might be expected to have
some effect on silica: apparently they do. not. Richardson (199) examined nine sugars
and seven polyhydric alcohols. and found no influence on solubility. For some
unexplained reason he found glycine- to. retard the rate of dissolution of quartz by
. .

An apparently unique type of compound thal strongly interacts with silicic acid and
silica surfaces by hydrogen bonding is l-hydroxypyridine IV-oxide. It will be noted
that, as in catechol. there are two oxygen atoms closely spaced. so that they can
either chelate with .1 silicon atom or hydrogen-bond to adjacent SiOH groups on
polysilicic acid.

The Occu-rfence. Dissolution. and Deposition of Silica

Gardner and Katritzky (200) prepared and characterized about 24 amino- and
hydroxypyridine oxides. and showed that 2-hydroxypyridine l-oxide exists as the
strongly hydrogen-bonded l-hydroxypyrid-2-one. Weiss and Harvey (201) described
the water-soluble monovalent 6-coordinated silicon chloride salt. soluble in water
and alcohol. The tetrachloroferrate (III)

/C.. . . . .
HC' _ C-O






salt is soluble in tetrachlorethane.

Iler examined the pl-l-titration behavior of silicic acid in the presence of 2hydroxypyridine I-oxide by titrating 16 mM (1000 ppm Si02 ) silicic acid silica from
pH 10.5 to 3.0 in the presence and absence of a 43 mM concentration of the Noxide. At no point did the titration curves differ. indicating that no complex had
formed. In another experiment. a solution of Si(OH). containing 100 ppm as SiO:z
was mixed with a 200-fold excess of the above N-oxide at pH 6.15 and .aged for a
few hours. Tests with molybdic acid showed that the reaction rate with silica
monomer was the same as a control, indicating either that no complex was formed
at this pH or that it dissociated very rapidly. However. the rate of dissoiution of
monomer from colloidal silica particles at pH 1.4 was apparently doubled in the
presence of a 20 mM concentration of the N-oxide, indicating some type of interaction at low pH.
Schlipkoter and Brakchaus (202) investigated the interaction of polyvinylpyridine
in connection with the problem of silicosis. (The extensive literature on this general
subject is reviewed in Chapter 7.)

. Organic Bases
Organic bases in water interact with silica in a manner that would be expected from
the resulting pH. Strong quaternary ammonium bases like tetramethylammonium
hydroxide dissolve silica rapidly. forming the silicate salt.
In a search for an organic medium that would dissolve silica. Meyer and Yen
(203) found that quinoline plus KOH pellets. dissolved crushed quartz at a
substantial rate, in fact, faster than saturated aqueous KOH.



According to Foucalt and Col1ette (204), finely divided silica dissolves more rapidly
in the peritoneal cavity of a rat than at. 37C in a buffer solution at pH 7A. The

- !

.. ..=....-..

of Silica

ino- and
:s a'" the

Effects of Organic Compounds on SoJubiJi4'-_


reason is unknown. but it would not be surprising if there are compounds present
having a catechol-like structure that are responsible.

les, .d
in water
Solubility in Alcohols

At 25C, amorphous silica is essentially insoluble in methanol. Her' prepared
colloidal dispersions of 90 A silica particles at pH 9-9.5 in methanol-water mixtu res
and observed the approach to solubility equilibrium from the initially undersaturated
solution and from a heated and cooled supersaturated solution over a period of 2
Equilibrium solubilities were as follows:
ce of 2ica from
f the Nplex had
as SiD:
~ed for a
th silica
s formed
iu ti,.., of
:d. he
~ interacIpyridine
5 general

Wt. % Methanol

at 25C (mg 1-1)





Obviously,. the solubility is strongly related to .water content and is very low in
anhydrous methanol.
On the other hand. at 200C under pressure, silica "dissolves" to a higher extent
in anhydrous methanol than in water, as shown in a careful study by Kitahara and
Asano (205). They reported the solubility to be at a minimum at around 80%
methanol-20% water.
Just as silica reacts with water for form Si(OH)~. silica combines to some extent
with anhydrous methanol:

ted from

and Yen
'tz at a

e ra oidly

Ester form~tionis evidently Involved," since silica did not dissolve in acetone under
the same conditions. Also it was shown that the surface of the silica gel used as a
source of silica was fully esterilied. However. it is not certain that the dissolved
species was the ortho ester or esters ofpolysilicic acids. The polymerization of the
dissolved silica on cooling suggests that the dissolved species may have contained
some silanol groups either from traces of water in the system or from residual
silanol groups on the gel that was used. even though the latter had been dehydrated
at high temperature.
Kitahara and Asano found that at 500C. methanol dissolved up to 1.9 g 1- 1 SiD:.
maximum solubility being observed when the gel was dehydrated to a maximum

: y....

"the..Occurrence. Dissolution. and Deposition of Silica


extent at SOO-700C. Gels heated to still higher temperatures no doubt'sintered with

loss of surface area. accounting for lower solubility.
The maximum degree of esterification of the surface of the resulting silica gel was
5.0 methoxy groups nm -2. This means that the surface was covered with a closepacked layer of methyl gro,!ps. which explains its hydrophobic character.

Higher Alcohols
Using silica gels dehydrated under optimum conditions and suspended in anhydrous
alcohols, the following solubilities were observed by Kitahara (206) at 500C:

Solubility (ppm)

CH 30H


C2 H $O H
n-C 3 H 70 H

The surface of the resulting gels was fully esterified with ethoxy and propoxy groups,
there being 3.0 alkoxy groups nm -2 (205,207-209).


The solubilities of silica in molten NaCl and KCI under argon have been measured
(210) at 900-1100C. The values were 1.8 and 1.7 ppm at 900C, and only increased
by I.S-fold at 1100C. It was no doubt important to exclude oxygen and moisture,
which rapidly oxidize these salts to alkali, which then would combine with the silica.





The rate at which silica dissolves in water is influenced by many factors, but regardless of the type of silica involved, the dissolution process requires the presence of a
catalyst. The dissolution of silica in water is, in effect, a depolymerization through
hydrolysis, and the "solubility" is the concentration of Si(OH). reached as a steady
. state in the depolymerization-polymerization equilibrium. The "catalyst" is a
material that can be chemisorbed and increases the coordination number of a silicon
atom on the surface to more than four, thus weakening the oxygen bonds to the
underlying silicon atoms.
The hydroxyl ion is the unique catalyst in alkaline solutions and hydrofluoric acid
in acid solutions. The structure of amorphous silica is an even more open arrangement than that of cristobalite, to which it is closely related. On the surface there are
spaces between oxygen ions sufficiently large to accommodate hydroxyl ions. Such a

of Silica

ered with


a close-




ae ~ed
re silica.

regard.nce of a
a steady
it" is a
a silicon
Is to the


Rate of Dissolution of Silica

surface then bears an ionic charge and silica is constantly being exchanged in an
equilibrium between solution and surface, as indicated. The first step is the adsorption of OH- ion, after which a silicon atom goes into solution as a silicate ion. If the
pH is much below II, the silicate ion hydrolyzes to soluble silica, Si(OH)., and OHions and the process is repeated (Figure 1.11). Hydrofluoric acid no doubt acts in the
same way, the first step being chemisorption of a F- ion, which is about the same
size as OH- ion.
Catechol and related compounds dissolve silica in neutral solution, but it is not
certain that a catalytic effect is involved. It is likely that these materials simply combine with Si(OH). as rapidly as it is liberated from the surface by the catalytic action
of OH - ion and thus keep the solution from becoming saturated. Above pH II the
hydroxyl ions act in the same way, converting Si(OH). to silicate ions, thus keeping
the solution unsaturated so that silica continues to dissolve. But below pH 11, even
down to pH 3, the OH- ion is only the catalyst that controls the rate at which silica
dissolves until the solution reaches saturation.
The dissolution mechanism has been discussed in detail by Strelko (211), who
considers analogous mechanisms for dissolution of silica not only by catechols but
also HF, H 3PO., and possibly in acidic acetyl acetone, all of which are known to
form silicon compounds in which silicon is coordinated by six -surrounding fluorine
or oxygen atoms.
However, the catalytic effects of F- and OH- ion are not identical. as is obvious
from the fact that silica dissolves in NaOH solution but not NaF. It requires the
presence of H+ and F-. Stober made the important observation that stishovite,
which is the only form of silica in which silicon atoms are surrounded by six instead
of the usual four oxygen atoms. is insoluble in aqueous HF but dissolves even in
weak alkali (140). The first step in the action of HF on any type of silica is to
convert the SiOH surface groups to SiF groups. These two surfaces are physicalIy
very different. The SiOH surface is completely hydrophilic, whereas the SiF surface
is hydrophobic. since there are no hydrogen atoms to hydrogen-bond with water.
Consequently, in the case of stishovite, which is much denser than quartz. the surface is converted to a close-packed monolayer of hydrophobic fluorine atoms,
physically similar to a fluorocarbon surface. This. in effect, excludes water from the






oric acid
arrangehere arc



SI-O ::"S~,OH


Figure 1.11. Proposed mechanism of dissolution of silica in water in the presence of hydroxyl
ions. The dotted line represents the interface between silica on the left and water on the right.




The Occurrence. Dissolution. and Deposition of Silica

surface so that no dissolution can occur. On the more open quartz structure. the
fluorine atoms are not as densely packed and water is not excluded from the surface:
there is still room for F- ions to penetrate the surface and raise the coordination
number of silicon. so that dissolution can occur.
As for the effect of hydroxyl ion, it is not possible to see how it could catalyze the
dissolution of stishovite in which silicon has already reached its maximum coordination number. No data seem to be available on the effect of pH on the dissolution
rate of stishovite, but it is interesting that at pH 8.4 it dissolves about as fast as vitreous silica when compared on the basis of equal areas of surface being'exposed to
the solution. Furthermore, it continues to dissolve past the saturation level for vitreous or amorphous silica. The concentration of soluble silica can reach as high as
190 ppm, at which point colloidal particles are nucleated (139). It is likely that stishovite is hydrolytically unstable and would eventually decompose completely to
amorphous silica. Whether or not pH has an effect on the rate of hydrolysis is not
As for other forms of silica, equations for the polymerization-depolymerization
kinetics and equilibrium have been formulated by Stober (195) to explain solubility
The relative rates of dissolution of different forms of silica at pH 8.5 have been
reported by Stober ( 139):

Type of Silica


Fused glass

Rate of Dissolution
(10- 8 g rnl " day-I)



The measurements were made with powders having surface areas of I J 01 2 suspended
in 500 ml volumes.
In determining the rate of dissolution of amorphous silica powders. the possible
existence of a porous. rapidly soluble layer must be considered. Yates and Healy
(212a) have shown that BDH-precipitated silica. widely used as a standard for study.
has a surface layer that is impermeable to nitrogen but permeable to alkali. This
"gel" layer dissolves more rapidly than the remainder of the silica and must be
taken into account when the rate of dissolution is being measured.
After an initial rapid period the rate of dissolution. expressed as rnoles Si0 2 rn ?
hr- I becomes constant and is characteristic of the particular kind of silica. When
BDH silica was first heated to 800C for 36 hr, the characteristic rate decreased by
50% and approached that reported by Stober (139. 144) for vitreous silica. The
original SDH probably contained some internal uncondensed OH groups.






.;, ...

~ ..

.,";".'- .

ition of Silica

Rate of Dissolution of Silica

.ructure, the
th -..face;

Effect of pH on Rate

::00. ~_.Iation

Below pH 3, Elmer and Nordberg (153) found that the initial rate of dissolution
reached a maximum at 0.8 N HNO", corresponding to pH 0.1, and declined above
this pH. However, BaumannJl54) has shown that from pH 3 to about 6, the rate of
dissolution increased in proportion to the hydroxyl ion concentration. There may
therefore be a small minimum between pH 0.1 and 3.0 that has gone unnoticed.
Between pH 2 and 9, Baumann found a region from pH 3 to 6 where the rate of
dissolution was proportional to the OH- ion concentration, whereas outside the
range the pH depended less on pH (Figure 1.12). Because of the overall shape of the
curve, he examined the possibility that at low pH silica became a cation, but this
could not be supported by a pH titration of soluble silica with hydrochloric acid.
More than likely, there is a catalytic effect of H+ ion which becomes evident in the
range from pH 2 to 0; on the other hand, above pH 6 rate of diffusion or rate of
desorption of the silicic acid from the surface may limit the overall rate of dissolu-

catalyze the
m coordina~ dissolution
s fast as vit~ exposed to
level for vith as high as
:ely that stiimpletely to
olysis is not


in solubility


5 have been

Relation Between Rate of Dissolution and Particle Size


It is logical to assume that the rate of depolymerization or dissoluti~n of colloidal

silica particles would be proportional to the specific surface area of the silica particles. Thus different solvents such as dilute hydrofluoric acid, molybdic acid. or dilute
alkali, all of which convert the monomeric silica permanently to other silicon corn-










the possible
and Healy
'd for study,
alkali. This
nd must be






:s Si0 2 m :"

ilica. When
ecreased by
silica. The



Figure 1.12. Log rate of dissolution of amorphous silica versus pl-l. Slope of dotted line indicates that between pH J and .5 the rate is approximately proportional to the hydroxyl ion
concentration. [Baumann (I 54}]



TfiC Occurrence. Dissolution. and Deposition of Silica


pounds. may therefore be employed to measure surface area. The difficulty,

however, is that the types of silica particles studied have been so varied and illdefined that this approach has not been of any value until recently, when sols of uniform discrete solid particles have become available. Even then the value of the
method is in doubt unless one can be sure that the samples of silica under study all
have the same composition and structure.
In a patent issued to Balthis (212b) a method is described whereby a sample of
deionized silica sol is put into an excess of 0.0 I N NaOH solution at 30C and over
a period of 90 min samples of the solution are removed and the 'amount of
monomeric silica is determined by reaction with the molybdic acid reagent. The rate
of dissolution of the silica to form monomeric silicate in the alkaline solution was
related to A, the specific surface area of the silica (in .square meters per gram)
determined by other means, by the following equation:

where K D is a constant defined as the specific depolymerization rate for the

particular type of silica, dm/d, is the rate of monomer formation per 100 min, and m
is the fraction of original silica converted to monomer at time t, in minutes.
For the particular type of silica particles involved, which was highly porous. the
value of K was ISS at the point where 20% of the silica was converted to monomer.
For a dense silica particle prepared at elevated tern perature by the buildup process
according to Bechtold and Snyder (213). the value of K was about 8 at the same
stage of dissolution. The porous and dense particles com pared were of about the
same size, being 18 and 16 nm in diameter. respectively.
It is therefore obvious that the rate of dissolution in alkali cannot be used as a
dependable method for determining specific surface area unless it is certain that all
the samples being evaluated have the same particle structure. or at least have been
made under the same process conditions from the same materials.
The method is probably valid. for example. as em ployed by Goto (214). who used
it to compare the particle sizes of silica polymerized from monomer in a pH range
from 7 to 10. In this case the particles were depolymerized by diluting the sol a hundredfold with I g 1-1 of sodium carbonate solution. giving a pH of 10.8, the original
silica sols containing 2 g I-I Si0 2 Goto used an equation developed by Suite, Hirai.
and Taki (215), which assumes that the rate of dissolution of particles is proportional to their surface area.
Instead of an alkaline solution. a dilute solution of NaF and Hel to generate HF
was employed by Goto (216a) to measure the specific surface area of colloidal silica
from the rate of dissolution in this acid medium. The samples from the reaction mixture were removed. the reaction was stopped by adding aluminum salt to combine
with the fluoride, and the dissolved silica was then determined by the molybdic acid


In of Silica

Rate of Dissolution of Silica

:d ~ ~ illols _ .milue of the
r study all

Based on Goto's data, as summarized by Okura, Goto, and Murai (216b). the
initial rate of dissolution of colloidal silica particles of different sizes at a concentration of 20 mg I-I, at 25C, in 0.12 N HCI containing 300 mg I-I F- ion, was
expressed as ppm passing into solution per minute per milligrams of SjO, initially
present. The observed relation of rate of dissolution to the specific surface area of
the silica particles initially present was interpolated as follows:

sample of
: and over
.mount of
:. The rate
[ution was
per gram)

orous, the
rp I .:ess
the same
about the

who used
pH range
sol a hunie original
ito. Hirai.
is propornerate HF

)idal silica
ction mix) combine
ybdic acid

A, Specific Area

D, Particle Diameter

(rn" g ")


R, DissolutionRate
(mg mg -I min -I)




The particle diameter was calculated from specific' surface area assuming a density
of 2.2 g ern-3. The data are approximately represented by the empirical equation R
= 1.7(IO)-loA3. The data resemble similar data obtained by Her for particles of
similar size dissolving in dilute alkali, as will be discussed.
The many different factors that can influence the rate of dissolution of silica particles in addition to the surface area have been reviewed by O'Connor and Greenberg
(217). The temperature, degree of crystallinity, previous mechanical and heat treatment, and previous treatments with water. alkali. or acid all had their effects. They
found that the theoretical rate equations for dissolution hold only if the silica is completely dispersed as individual particles and that in aggregated or flocculated
material not all the surface' area is available to solution. They formulated the rate of
solution as follows:

.e for the
lin, and m

used as a
in that all
have been





where c = concentration of dissolved silica (moles 1-1)

. S = area of surface of solid silica phase (rn" I-I)
k 1 = rate constant for dissolution
k z = rate constant for deposition
t = time
At equilibrium. dcf dt = 0 and c;



k./k 2 from which

= kzS(c. - c)

where c, is the equilibrium concentration or "solubility" of silica.

.., ...:;- ...

-The Occurrence. Dissolution. and Deposition of Silica


In neutral solution. where a considerable excess of silica is' present. S is

approximately constant. and if c = 0 at t = O. then

c, - c


= k'St

where k' .. -k 2 O'Connor and Greenberg presented experimental data showing that
this expression held very well when an amount of silica was used so that S was
constant. But when different amounts of a given silica powder were suspended. the
times required to reach equilibrium were not proportional to the surface areas. In a
granule of silica powder consisting of aggregated ultimate silica particles, it is
obvious that the particles inside cannot dissolve as rapidly as the outer particles.
In alkaline solution above pH II, the rate of deposition is zero, so that




I ~





Letting Cp = moles per liter of amorphous silica in suspension consisting of n

particles of density 2.2 and diameter d,









. 1

; i;


,~ ,~




= s =

k: n 1!3C p 2 3

Integration gives

where CPo is concentration of suspended silica at time t = O. and k 1 * and k 3 are

Greenberg (218) found that at 40C above pH I I. the rate is independent of
hydroxyl ion concentration and proportional to the amount of silica introduced into
a fixed quantity of 0.025 N NaOH (pH 12.4) solution. where the Na 2 0 : S i 0 2 ratios
were I :0.66 to 1:2.
The energy of activation for dissolution in water was 17.8 and in alkali was 18
kcal mole:". Earlier. Greenberg had measured the rates of dissolution of both quartz
and amorphous silica powders in alkali, and concluded the energy of activation in
alkali was 21.5 kcal mole " '. also that above pH II the addition of Na 2SO. did not
affect the rate, and that the rate was profoundly influenced by the nature of the silica
A modification of the method of measuring dissolution rate was used by McNally
and Rosenberg (219) to determine the surface area of extremely small colloidal silicparticles which are below the practical resolution limits or the electron microscope. _

.:-, ...

on of Silica

Rate of Dissolution of Silica

sent, S is

om. They carried out the dissolution reaction using 1.6 g of colloidal Si0 2 in I liter
0.0025 N NaOH, and followed the rate of depolymerization by noting the change in
electrical conductivity of the solution with time. Where R, is the electrical resistance
after all the silica is dissolved and R o the resistance of the mixture before the silica
dissolves, a plot of (R, - R O} 1/ 3 versus time gave a constant slope which was related
to the particle diameter by measuring the slope for particles known to be 23 nm in
diameter, and assuming the slope is inversely proportional to the particle size as
small as 1.4 nm. The sol contained 14% silica, had a molecular ratio of Si0 2 : Na 2 0
of about 14: I, and the particles could not be resolved by electron microscopy.
However, the assumption that in such a solution all the silica is present as uniform
"particles is questionable, since a considerable portion of the silica must be present as
silicate ions.
In the method used by Balthis (2l2b) the assumption is made that the rate.
expressed in terms of the fraction of silica dissolved per 100 min, can be measured at
a given time, t. However, it is easier to use the integrated equation

owing that
hat S was
ended, the
areas. In a
icles, it is


(l - m}I!3 = 1 - k'Aol

isting of n

and plot the cube root of the fraction of silica remaining undissolved ".versus time to
obtain a line the slope of which is a combination of A o, the original specific surface
area of silica, and k', a constant proportional to the characteristic depolymerization
rate of the particular type of silica involved. The latter can be expressed in terms of
milligrams per square meter per minute, as measured, for example, at pH 12 and at
" Rate of Dissolution of Very' Small Particles

and k, are

Goto (216a) has shown that the rule that the rate of dissolution is proportional to the
surface area does not hold for particles of less than 5 nm diameter. As discussed
earlier, smaller particles exhibit much higher equilibrium solubility and this must be
"taken into account. Thus the rate of solution in alkali should be proportional to both
the surface area and the solubility of silica.

oendent of

duced into
5i02 ratios
ali was 18
oth quartz
tivation in
O. did not
.r the silica

rate of solution



where K = constant
A = area of silica
S = solubility of the particle. depending on the size
m = fraction of silica dissolved at time t
From the earlier discussed relation or solubility and particle size:

f McNally

:Jicl . silica
.ros __ ,IC. 5





5.7T- 1d-\


The-Occurrence. Dissolution. and Deposition of Silica

where Stj is the solubility of particle of diameter d. in nanometers, at time t; S, is the

solubility of a flat silica surface with infinite radius of curvature; is the interfacial
energy in ergs ern -2; and T is the temperature in degrees Kelvin.
Where two particle sizes are considered,

where Stt is the solubility of diameter d and So that of diameter do. Let B
l3.IEr- l


Rate of Dissolution as Particle Dissolves


If silica is assumed anhydrous with a density of 2.2, the specific surface area A
1373 d :',




particle at t = 0
= area at time t
do = diameter at time zero
d diameter at time I (nm)
fraction of silica dissolved

Let A o

= area of a

- ;i:
.~ i



(I - m)2/3

(I - m)I/3

. I




Sd == So [exp (-Bd;I)] [exp Bd;' (1 - m)-1/3]

- 1





KAoSo [exp (_Bd;I)] [1 - m ]-2/3 [exp Bd;l {I _ m)-lf3]

B ... 0.044

when T

= 298

For a given type and particle size of silica the rate decreases as the particles are
reduced in size because the surface area is reduced by the factor (1 - m )2f3 but as
the size becomes smaller it is increased by the increasing solubility, as expressed in
the final term in the above equation. Thus, since do is a constant in a given experiment, the relative rate of dissolution can be calculated from the equation



: ..J,

(I - m)2/3 [exp (O.044d;') (I _ m)-lf3J

. :


tion of Silica

Rate of Dissolution of Silica

e t; Sj is the
e iP"'rfacial

The relative rate can be plotted versus m, the fraction dissolved, as shown in Figure
1.13. It will be noted that, for particles of 3 nm diameter, there should be less than
:::30% variation in rate while up to half of the silica dissolves, as long as E is assumed
to be in the range of 20-100 ergs cm-t.
Apparently there are no published data on the rate of dissolution of particles
under 5 nm diameter. In a preliminary study, the writer has made a sol by hydrolyzing ethyl silicate in a water-alcohol mixture with HCI and then diluting and neutralizing to pH 2.1 to obtain a 1% SiOt starting sol of 1.9 nm particles, By raising
the pH to 6.0 for 2 min and reducing it again to 2.1 the particles increase in size to
2.1 nrn. If held at pH 6 for 1 hr the size reaches 2.8 nm.
Particle growth at pH 6 is more rapid at higher temperature; when the pH is
adjusted and the sol heated to 60C for 1 hr, cooled to 25C, and stabilized at pH

), Let B

e area A




















u, 0.5




iarticles are
n )2/3 but as
expressed in
iven experi-

.... _--



Figure 1.13. Calculated relative . rates of dissolution of amorphous silica particles with
assumed values for E and initial particle diameters:

E (ergs ern -2)

do (nm)








.The Occurrence. Dissolution. and Deposition of Silica


2.1, the particle size is 3.5 nm. The particle size is calculated from the specific surface area determined by a rapid titration of a sample containing 200 g I-I NaCI
from pH 4.0 to 9.0. corrected for the monomer present at the start which consumes
I equivalent of alkali per Si(OH). molecule.
The rate of dissolutionof silica was measured in an excess of 0.01 N NaOH solution at 25C, by following the amount of total soluble and ionic silica by the
molybdic acid method.
As shown in Figure 1.14 the experimentally determined rate of dissolution of
particles of a given size remained reasonably constant while 10-70% of the silica was
dissolving. According to the foregoing calculations for Figure 1.13 this could be true
for a 3 nm particle only if E is about 100 ergs cm ? and for a 2 nm particle if E is
even lower.

Rate of Dissolution of Particles of Different Sizes

Next must be considered the relation between the measured dissolution rates when
starting with particles of different sizes.
Since with very small particles the rate does not change much as the silica
dissolves, the average rate is about the same as the initial rate. where m = 0 and


d ;' [exp (0.044 Ed;I)]

This expression assumes that the rate of dissolution is proportional to the specific
surface area of the silica (proportional to do -I) and to the solubility as related to

o 1.00





.... 0.50








Figure 1.14.

Fraction of silica dissolved versus time for particles of indicated initial diameters

in nanometers,

on of Silica

Rate of Dissolution of Silica

oecific surI-I NaCl

I cc



aOH soluica by the






solution of
~ silica was
uld be true
:icle if E is









rates when



the silica
n = 0 and

.he specific
related to


Figure 1.15. Calculated versus experimental rates (circles)

of dissolution of silica of different particle diameters and
assumed values of E (ergs em -Z).

particle size. As-seen in Figure 1.15 it requires a value of E of 150-200 ergs em:" to
give a line parallel to the one from experimental data. (The data are empirically
plotted as log of the relative rate versus log of particle diameter in nanometers. and
only the slope is significant.)
There is therefore a discrepancy between theory and experiment. To explain the
constant rate of dissolution as particles dissolve requires a value for the interfacial
energy, E, in the range 50-100 ergs ern -:z. On the other hand. to explain the much
faster rate when starting with smaller particles requires a value for E in the range
150-200 ergs em -2.
In the case of particles approaching 2 nm in diameter, a larger fraction of the
silicon atoms is linked to OH groups on the surface. Thus the rate of dissolution
would be higher than anticipated because fewer siloxane bonds have to be broken
per molecule of Si(OH). that is liberated. In any case. the above considerations indicate that much remains to be learned about silica particles or "polymer molecules"
in the size range below 5 nrn, where the silica is labile and becomes unusually soluble.

Dissolution of Crushed Powders



The more rapidly dissolving layer on the outside of finely crushed quartz. tridyrnitc,
cristobalite, and vitreous silica was examined by Bergman and Paterson (220) by



"ftreOccurrencc:. Dissolution. and Deposition or Silica

measuring the rate of its removal in dilute alkali. It was concluded that the more
rapidly dissolved layer was from one to five monolayers of Si02 in average depth.
the finer particles having the greatest thickness. This was true of all the polymorphic
forms of silica. However. there is reason to believe that much of the "easily soluble"
material consisted of the corners and edges of the irregular fragments which have a
higher solubility owing to the very small radius of curvature, rather than a uniform
The dissolution rate of amorphous silica in 0.0 I N NaOH solution has been
measured at 23-88C by a rotating disk method by Anatskii and Ratinov (221, 222).
De Keyser and W olIast (223a) have been able to use the constant of the rate of
dissolution of a solid, obtained by the rotating disk method, to calculate the rate of
solution of powders of which the particle size has been measured by the cumulative
sedimentation method.
The nature of the "disturbed layer" on the surface of crushed quartz was
described at the beginning of the section on solubility, and its abnormal solubility is
discussed in Chapter 7 in relation to a theory of silicosis. When crushed powders are
to be used for solubility measurements it is essential that the highly soluble outer
layers and fine fractions be eliminated. The problems involved in obtaining quartz
with a reproducible dissolution behavior have been described by Baumann (223b),
who found that subsequent adsorption of silicic acid from solution or heating the
cleaned powder above 800C can reestablish a layer which is abnormally soluble.

Neutral Solutions-Effect of Salts





Rate measurements on fine amorphous powder and colloids have been made by
Doremus et aJ. (224) and by Friedberg (225). The effects of pH, temperature. and
presence of salts were similar to those reported by others. More than 50 years ago,
Dienert and Wandenbulcke (226) reported the basic facts that colloidal silica passed
into solution as soluble silica. which was detectable calorimetrically with molybdic
acid, and that alkalinity and salts were good catalysts for dissolution. They made an
observation which apparently has n~ver been followed up. They claimed that when
salt is present. the dissolution rate is faster in a quartz container than in platinum
. and that in the absence of added salt. colloidal silica would pass into the soluble
state when heated with water in quartz. but not in platinum. However. pH measurements were not made.
In a study of the attack of water on glass. Wiegel (227) found that glass was more
resistant to attack by water than by salt solution. Also when alkali was first
extracted from glass by acid. it was then more resistant to attack by water. no doubt
because there was no soluble electrolyte to build up in the water. Finally. he found
that copper. zinc. tin. and aluminum all reduced the amount of alkali extractable
from glass by water. while nickel. cobalt, and magnesium increased it.
It thus appears that in very pure water. in the absence of impurities which can
leach from silica into water. silica is remarkably inert (167a). remaining undissolved
for long periods of time. whereas it rapidly dissolves in seawater. Impurities have
such marked effects in neutral solution that it is doubtful if reproducible rates of';


of Silica

the more
ige ~epth,
lyr. ..Jhic
y soluble"
ich have a
a uniform


Rate of Dissolution of Silica

dissolution of amorphous or colloidal silica can be obtained unless unusual precautions are taken to start with very pure silica and avoid all possible contaminants.
Van Lier (114) found that NaCI increased the rate of solution of quartz in water
by factors of about 4, 14, and 67 in 10- 3 , 10- 2, and 10- 1 N solutions.

Retardants of Dissolution
has been
he rate of
:he rate of
uartz was
olubility is
iwders are
uble outer
ing quartz
nn (223b),
eating the

made by
ature, and
years ago,
ica passed
y made an
that when
, platinum
he soluble
measurewas more
was first
. no doubt
. he found

As discussed earlier there has been much study of the effect of impurities, especially
aluminum, on the solubility of silica, but only limited research on the effect on the
rate of dissolution.
The rate of dissolution of fused silica from a flask was measured by Sasaki (228);
the rate was proportional to the base concentration. Zinc and aluminum both greatly
diminished the rate of solution. Complete inhibition in 0.1 N NaOH was observed at
concentrations of 3.4 x 10- 3 N aluminum and 5 x 10- 3 N zinc.
Hudson and Bacon (229a) examined inhibitors to prevent the attack of 3% NaOH
solution on glass at 125F. By far the best was beryllium, which almost stopped
attack at a concentration of 10 mg atoms per kilogram of solution (about 2 g I-I
BeSO. 4H 20 ). Catechol. then zinc, were next best at this concentration. Surprisingly, aluminum in this strong alkali was little better than barium, lead, and other
heavy metals, none of which were very effective because they were precipitated from
Porous glass membranes used for hyper filtration were stabilized against dissolution by having 0.3 g 1-1 of AICI36H 2 0 in the feed solution (229b).

Rate of Dissolution in Presence of Catechol

It is interesting that in 3% NaOH solution catechol is a strong inhibitor of the dissolution of glass (224). However, in neutral and slightly alkaline solutions catechol is a
It was pointed out above that it strong ammoniacal solution of catechol dissolved
vitreous silica at a rate of 0.5 micron hr" ' at 90C and attacked quartz at half that
rate. This corresponds to a dissolution rate of about 1100 mg m -2 hr- I for vitreous
silica. and 550 mg m -2 hr " for quartz.
Measurements by Bach and Sticher (230) were made on HF-c1eaned quartz surfaces using 0.2 M catechol solution at pH 9.6. From 17 to 40C the rate of attack

log v. = 0.0295T - 20.87


which can
-it" 'lave
c r... ..:s of

At 25C the rate is 0.8 x 10- 12 g cm " sec-t, which is 2.9 x 10- 2 rng m ? hr- '. The
extrapolated rate at 95C is 3.6 mg m -2 hr " ',
Extensive data on the rates of dissolution of different forms of silica in dilute
alkali and in the presence of catechol were reported by Baumann (231 a). It was
noted that in dilute alkali the rates of solution declined as the solid phase was

- --.

TherOccurrence. Dissolution. and Deposition of Silica

consumed, but remained constant when catechol was used. Vitreous silica dissolved
much more rapidly than quartz. but stishovite was completely insoluble (231 b).
The declining rate in alkali may have been due to the accumulation of traces of
aluminum in the solution as the silica dissolved. When readsorbed on the silica surface as aluminosilicate ions t-hese would reduce the rate of solution. However. catechol can complex both silicon and aluminum so that the rate remained constant.
(See also discussion of chelates in Chapter 2.)


Rate of Dissolution in Aqueous HF

Bergman (232) conducted extensive measurements on the rates of solution of quartz
and tridymite in 0.1 and 1.0 M H F solution at 25C. Reasonably reproducible rates
expressed as milligrams per square meter per hour were obtained using particles in
the micron size range. Tridymite dissolved in 0.1 M HF about seven times faster
than quartz and vitreous silica about 45 times faster, but in 1.0 M HF there was little difference.
In 0.1 M HF, cristobalite and tridymite dissolved about 10 times faster than
quartz and vitreous silica 100 times faster. based on unit surface area (233).
There have been several studies of the etching rate of HF on silica in relation to
processes in the electronics industry. A patented mixture of HF and NH.F gives a
desirable etching rate for thin films (234).
. Judge (235) concludes that the rate of attack of HF on Si02 films is due to HF 2 and HF. the former acting four to five times faster than the latter. Pliskin and
Lehman (236) uses a solution of 2.26% HF and 2.15% HNO a to etch Si0 2 films on
silicon. The film formed by oxidation at 1000C. which is probably nonporous. is
etched at the rate of 2 A sec-t, which is 1600 mg rn ? hr ". Using vitreous silica
spheres. Leko measured the rates of dissolution in HF solutions up to 20%
concentration and at 20-80e. Conditions for removing a certain thickness of silica
can be calculated from the data (237).
A fundamental study of the attack of HF on vitreous silica, preceded by a literature review. was published by Blumberg and Stavrinou (238). They developed equations for the reaction rates and determined the velocity constants. At 32C. the
typical value was k , = 5 x 10- 8 g SjO, sec- 1 em -2 M H~'

Comparative Rates of Dissolution


Miscellaneous data have been collected in Table 1.1. and an attempt made to put the
rates on a comparative basis.


.. Removal" of silica from water is considered here from the standpoint of lowering
the silica content without regard to the means of disposal. This may occur by deposi-






In of Silica

Table 1.1

II b\


silica surever, cate, constant.

Type of





(mg m -2 hr " ')


4 x 10-

0.2 M catechol
(pH 9.6)
Ringer's solution
(0.9% NaCI.
0.1% NaHC0 3 )
NaOH (0.05-0.4 N)

of quartz
cible rates
iarticles in
mes faster
.re was lit-

relation to
.F gives a
e to HF 2 " and


)2 films on
.porous, is
eous silica
p to 20%
5S of silica

Iy a literaiped equa32C. the


Van Lier (115)

Stober (139)


0.13 a
0.5 a

0.015 b

1.00 M HF
Saturated catechol in
cone. NH.OH
0.8 M catechol
(pH 8.5)



Bergman (220)
Bergman (220)
Baumann. Klosterkotter, and Robock
Bergman (232)
Bergman (233)
Her (unpublished)



Baumann. Klosterkdtter. and Robock


Ringer's solution
NaOH (0.05-0.4 N)
1-5% HF
24-14% NH.F
J.l3 M HF
0.34 M HN0 3
1.0 M HF


0.6 a
2-10 x 10





0.1 MHF



0.1 MHF





Stober (139)
Bergman (220)
Strauss and Bauer
Pliskin and Lehman
Blumberg and
Stavrinou (238)
Blumberg and
Stavrinou (238)
Bergman (232. 233)
Baumann. Klosterkotter, and Robock
(23 Ib)
lIer (unpublished)

(pH 8.5)
Satd. catechol-NH.OH


0.1 MHF



Bergman (232. 233)


0.1 MHF



Bergman (232. 233)


0.8 M catechol



Baumann. Klosterkotter, and Robock


to put the



0.8 M catechol




0.1 NNaOH

aster than

.. ~

Rate of Dissolution of Silica



(pH 8.5)






Rate for "rapidly soluble" surface layer.

Rate after surface layer removed,


rl "ring
by L",Josi.








;c:. ,..~~;.;:; ....-;:



.The Occurrence. Dissolution. and Deposition of Silica

tion, adsorption. or precipitation. On the other hand. "deposition" of silica involves

the formation of solid silica either as a coating or a solid surface, or as crystals.
However. the deposition of soluble silica on particles that remain in water as a sol is
considered separately in Chapters 3 and 4; the biogenic removal of silica and deposition in living organisms are reviewed in Chapter 7.
Since silica exists in water both in soluble and colloidal forms, the mechanisms of
removal and deposition are quite different. and often must be considered separately.
Removal of Silica from Water
The removal of silica from water is involved in several types of practical problems.
Most important is removal from feedwater to high pressure boilers in steam power
plants, where even traces of silica can lead to formation of deposits on turbine blades
and scale formation on heat transfer surfaces. Another is removal from cooling
water where silica is deposited on equipment as the temperature is reduced rn:~ as
evaporation occurs. Of growing importance is the silica in geothermal steam and
water, where high silica levels lead to serious problems in -developing power from
these sources. An artifical but economically serious problem has also arisen in
industry from decrees that traces of colloidal silica must now be removed from some
industrial waste streams. even though the level is below that in some natural waters
which are not considered harmful (239). Glacial streams are often milky with suspended silica.
Since it is not possible here to review the large volume of literature on water treatment, only general principles are considered, along with some examples.

Precipitation Mechanisms.
In considering the phenomenon of precipitation it should be pointed out that two
entirely different mechanisms may be involved:
NUCLEATION. Most discussions of precipitation concentrate on the question of how
particles are initially formed. The mechanism has been considered in detail, for
example. by Nielsen (240) and Walton (174) in their treatises on precipitation. The
formation of amorphous silica particles by polymerization of Si(OH)4 involves some
aspects of the theory of nucleation. This is dealt with in Chapter 3.


COAGULATION. In this case, a precipitate is formed by coagulation of already existing colloidal particles. It does not involve nucleation and is often considered as a
separate subject.

Nucleation of Quartz
The formation of quartz must definitely involve nucleation. Although quartz is not a
colloidal form of silica. its genesis is discussed briefly.
Quartz crystals apparently never are precipitated in suspension. as, for example,
in the case of 8aS0 4 Instead of quartz, colloidal particles are nucleated in suspen-


n of Silica

; cr"..rals.
as ~ . is
id deposi1

anisms of

irn power
ne blades
n cooling
ed f;"~! as
team and
wer from
arisen in
rom some
'al waters
with susiter treat-

that two

of how
letail, for
.tion. The
lves some


ady existered as a

tz is not a



n Sl._tJcn-

Removal and Deposition of Silica from



sion in supersaturated solution from hot springs and also are formed in other natural
waters (32,241).
Solid amorphous silica (opal) and quartz crystals, as in geodes. are formed on
solid sufaces. In a comprehensive investigation of the behavior of silica in hotspring water, White Brannoc~, and Murata (45) pointed out that there was no evidence quartz could be formed within any reasonable time under ordinary conditions.
In nature, amorphous silica appears to be deposited from the highest concentration
of silica, then chalcedony (extremely tine-grained fibrous form of quartz) from a
lower concentration, and finally macroscopic quartz crystals from lowest concentrations just exceeding the saturation level for quartz. Over millions of years most
amorphous silicas in the presence of water are eventually transformed to chalcedony
or chert and quartz.
As early as 1909, Leitrneier (242) reported that quartz could be obtained in the
laboratory as microscopic spheroliths by heating 1% by weight of microamorphous
silica ("silicic acid") in an aqueous suspension of Al(OH)3 at 80C for 144 days.
Only recently was it shown by Harder and Flehmig (243) that even at 20C, quartz
crystals were nucleated in a suspension of Fe(OH)3 or AI(OHh in only 14 days. The
solutions were very dilute: 2 ppm Fe3+ or AI3+ and 0.4-5 ppm Si02 In some manner, amorphous hydroxides of Fe3+, AI, Mn, and Mg can absorb and hold up to 9
moles of Si02 per mole of hydroxide, The silica is converted to quartz crystals
10-100 microns in size, the quartz also being accompanied by "quartzin," a form' of
rodlike or fibrous quartz elongated along the z-axis.
Quartz crystals will not grow from solution saturated with amorphous silica,
containing around 80-110 ppm 0 r monomeric silica, because the latter is adsorbed
on the quartz surface in a disoriented' state, as discussed earlier in connection with
the solubility of quartz. The metal hydroxides combine with Si(OH). in some manner to repress its solubility to the point where quartz will nucleate and grow (244a).
A suitably low degree of supersaturation, without the presence of metal
hydroxides, was attained by Mackenzie and Gees (244b). who ground quartz crystal
to less than 6 microns in size and shook the powder in seawater at a concentration of
0.5% at 20C for up to 3 years. The concentration leveled off at 4.4 0.3 ppm and
new micron-sized quartz crystals were observed to have been formed. No supersaturation occured in sea water. The concentration of soluble silica rose to 3 ppm in I
month and 4.4 ppm in 1 year, but the pH remained at 8.1. It was speculated that the
silica was not deposited on the quartz surfaces already present. because they were
covered by organic or inorganic impurities such as iron oxide. The presence of traces
of iron or aluminum in the system might also explain why the observed solubility is
somewhat lower than the 9.5 ppm value that would be expected by extrapolating
Van Lier's data to 20C (see Figure 1.4). As shown by Willey (38) and by Iler (37).
the presence of traces of alumina reduces the equilibrium solubility of silica.
Adsorption and Precipitation by Hydrous Oxides

Silica, both soluble and colloidal. can be removed from water to varying low levels
in neutral or slightly alkaline solution by coprecipitation with insoluble metal








The Occurrence. Dissolution. and Deposition of Silica

hydroxides in situ or by adsorption upon freshly formed hydroxides added 'to the
lIer found at pH 3-4 Si(OH). is not adsorbed on the surface of crystalline
alpha alumina that has been cleaned by treatment with HF followed by ammonia to
remove fluoride ions and then thoroughly washed. However. such a surface does
strongly adsorb colloidal or polymeric silica at this pH (245).
On the other hand. at pH 7 and higher. Si(OH). begins to ionize to silicate ions in
increasing degree, and these are adsorbed and can form a silicate-type bond with a
variety of hydroxides that are in a highly dispersed state. For example. Wohlberg
and Bucholz (246) have reviewed the literature on the reduced solubility of silica in
the presence of metal salts and hydroxides. It was concluded that alumina should be
the best adsorbent. but aparently it is not widely used. Magnesium is the commonest
reagent, but requires a relatively high pH and recirculation of sludge. and has to be
operated hot (247a). Willey (247b) has shown that zinc ions at about pH 8.3 can
precipitate dissolved silica that is in excess of 2 ppm, and the precipitate has the
Si: Zn composition of 5: J.
In laboratory studies, silica was adsorbed and precipitated by hydroxides of aluminum, iron. manganese, and magnesium. according to Harder (42). After precipitating 15 ppm AI(OH)3 from a 3 ppm SjO, in solution. the residual silica was 0.8
ppm. whereas with 30 ppm aluminum hydroxide. no measurable silica remained.
Willey has given an excellent review of the extensive literature on the interaction of
silica and alumina in dilute solutions. The low levels of soluble silica reached (39)
depended on the particular solid aluminum silicate phase that was formed or
present. She studied the interaction at very low concentrations. generally less than
10 ppm, and found that only I ppm Si0 2 is required to initiate the precipitation of I
ppm of AI 203 from solution. and if more than about t ppm AI203 is in solution the
silica concentration becomes exceedingly low.
In further studies, Willey suspended separate mixtures of (I) microamorphous
silica ("silicic acid") and amorphous alumina, and (2) quartz and amorphous alumina in distilled water and 0.6 N NaCi solution as well as in seawater (which
behaved the same) at 4C. pH 7-8, for up to 38 days. The results can probably be
explained as follows.
The amorphous silica powder consists of aggregates of particles of colloidal size.
Consequently. during agitation of the suspension in water or salt solution. a small
nonreproducible fraction of the particles becomes detached and remains in colloidal
suspension. When salt is present. the colloidal particles are adsorbed at once by the
suspended alumina which does not disintegrate. Only soluble silica is left in solution.
However. in pure water. in the absence of coagulating effect of salt. coagulation is
less complete. Once alumina is transferred to the silica surface and silica to the alumina surface. both surfaces become equally negatively charged. and there is no
further tendency for mutual flocculation.
The solubility of amorphous silica was 82 ppm in perfect agreement with other
data (see Figure 1.4). but when alumina was present (0.04-0.10 ppm in solution),
enough was adsorbed on the silica to reduce the solubility to around 67 2 ppm.
The effect of soluble silica on the solubility of alumina was only slight. since the alu-


- I



--, of Silica

Removal and Deposition of Silica from Water,

ed to the

mina alone dissolved only to the extent of 0.06-0.3 ppm. some part of ~'hich could
have been colloidal.
Healy et al. (248) arrived at a similar conclusion in a study of the mutual or
heterocoagulation of mixed colloidal oxide dispersions. When two oxides with different isoelectric points are mixed at a pH midway between. they immediately coagulate. But if the coagulate is repeatedly aged. redispersed, and allowed to recoagulate, the coagulation time increases until finally the mixture forms a stable dispersion. This is because mutual transfer of oxides has occurred until all particle surfaces
become alike and bear the same charge.
Because of the great variation in experimental conditions. results of different
investigators are usually not comparable and generalizations are difficult to make.
Accordingly. all that can be done is to compare individual observations.
The effect of aluminum ion combining with soluble silica was found by Goto (249)
to be optimum at pH 8-9. To a solution initially containing 35 ppm Si(OH) aluminum ion was added as chloride or sulfate and then the pH raised to 8-9. Twenty
ppm AI reduced the soluble silica to 15 ppm. 50 ppm reduced it to ~ ppm, and 100
ppm to about 3 ppm. When each molecule of precipitated silica is accompanied by
one Al atom. then the molar concentrations of Si0 2 and Al remaining in solution
appear to be approximately related as follows:



monia to
'ace does
te ions in
id with a
~ silica in
.hould be
has to be
I 8.3 can
e has the

es of alur precipi1 was 0.8

action of
:hed (39)
irrned or
itiou of t
ution the
hous aluer (which
ibably be
iidal size.
. a small
ce by the
ulation is
a the aluere is no
lith other
:l:: 2 ppm.


M S IOt MAl = (0.1


0.05) x 10- 12

Colloidal silica can exist in water far below the equilibrium solubility of arnorphOUS silica. especially at low pH. Goto found that in water containing 40 ppm silica
present as colloidal particles, the addition of. I ppm Al as aluminum salt coagulated
90% of the silica, but only in the narrow pH range 4.1-4.7. There was essentially no
effect at higher or lower -pH, probably because below 4.1' the negative charge on
silica becomes vanishingly small. and above 4.7 the charge becomes too large to be
neutralized by such a small quantity of aluminum ion. Similarly. Okamoto. Okura,
and Goto (250a) found that the AI3+ ion precipitates colloidal silica only in the pH
range 4-5. It requires only I part AI3+ to precipitate 40 parts of Si0 2 On the other
hand. AP+ precipitates soluble silica in the pH range 5-11, but at least 4 parts Al
are required for I part Si0 2
From the widely diverse observations it is obvious that the alumina-silica system
is very complex. Under the right condition's alumina can remove silica from solution
down to a very low level. but this is difficult to accomplish under practical conditions.
Huang (250b) has made a detailed study of the removal of silica from dilute
aqueous solution by adsorption on gamma alumina, Adsorption isotherms and
kinetics were elucidated. Under suitable conditions the concentration of soluble
silica was reduced to the range of 2-4 x 10- M or 12-24 ppm SiO t ,
It is probably for this reason that lime and magnesia are commonly used for
removing silica from water. They flocculate colloidal silica along with other suspended or precipitated matter and combine with and adsorb soluble silica. A suitable
granular magnesia adsorbent for silica. made from magnesite. worked best at pH



The Occurrence. Dissolution. and Deposition of Silica

8.5-9. according to Shemyakina (251). The water, previously limed to reduce silica
to 2-4 ppm, was passed through a I m deep bed of adsorbent on a filter. which
reduced silica to 0.5 ppm initially, and 1.5 ppm later when the adsorbent bed had
taken up 9% by weight of silica. However, the linear rate of flow was only 6 m hr
Chugunov (252) reported that the resulting silica level was as low as 0.3 ppm, but
pointed out that some magnesium then had to be removed from the water by ion
exchange. Other investigations of the use of magnesia as a silica precipitant or
adsorbent have been described by Nikolaev, Samusenko, and Tush (253), Schumann
(254). Tkachenko, Chernova, and Shuryshkina (255). and Straub (256).
Demineralization of boiler water with calcium under optimum conditions can
reduce the silica level to 0.4 ppm, according to Sramik (257). This involves preheating the water to 80C, adding lime, filtering. saturating with CO2 , but not below pH
9, and refiltering. The recovered CaC03 is chemically pure. The nature of the
deposits on turbine blades and effect on efficiency have been discussed by Kot and
Lobanov (258).
'In some low pressure plants (200 psi) water containing as much as 60 ppm Si0 2
can be fed directly to the boiler and treated in situ with sodium tripolyphosphate,
NaOH, sodium sulfite. and Na 2SO. (400 ppm) to maintain a pH of 10.5-10.8 (259).





Removal by Ion Exchange


. For feedwater for high pressure stearn power plants. precipitation processes are used
only as a first step to reduce silica concentration to a few parts per million. Then ion
exchangers must be used to further remove silica to less than 1 ppm. In oncethrough boilers operating at supercritical pressure (3200 psi) and corresponding
temperature, where Si(OH}. becomes completely volatile, the silica content must be
kept as low as 0.01 ppm. The amount of silica carried mechanically is proportional
to concentration in the boiler water and varies with steam pressure (260).
Combinations of processes used for silica removal have been reviewed by Applebaum (261). Silica concentration is reduced to 2-3 ppm by a cold process (70 0 F)
using dolomitic lime (32% MgO). By operating hot (over 120F), only half as much
lime is needed. The most efficient is an activated MgO. of which only 12 ppm is
needed in a hot process to reduce silica from 5 to 3 ppm. Further reduction of silica
is accomplished by ion exchange on a strong-base resin. According to Stassart (262)
much of the silica can be removed by adsorption on Fe(OHh and the remainder
down to 0.3 ppm by a strong-base resin.
It is apparent that the lime or magnesia treatment reduces silica to 3 ppm under
rather simple process conditions. but is capable of reducing it to below 0.5 ppm if a
more complex process is used. If silica must be reduced eventually by ion-exchange
resin, it is likely that the more efficient but complex pretreatments are not justified.
Ion-exchange resins of the strong-base type adsorb only the soluble silica as silicate ions, but special wide-pored resins have been developed to permit also the
adsorption of colloidal silica (263).
The colloid is desorbed from the resin with acid. and the resin regenerated with
alkali. According to Ryabin, Novoselov, and Lazareva (264) after silica has been






Removal and Deposition of Silica from Water_

ce silica
, \11'- ;,.h
oec . j
m hr".
ern, but
. by ion
.tant or

removed from water to a low level by a strong-base ion-exchange resin. further

reduction is achieved by passing it through a zirconium hydroxide-loaded
macroporous anion-exchange resin. A totally different type of exchange resin for
silica is a catechol-formaldehyde polymer in which the two phenolic groups in ortho
position can strongly adsorb sjlica by chelation (265).
Removal of colloidal silica by flocculation with cationic organic polymers is
involved in other aspects of water treatment. The chemistry of flocculation is
covered in Chapter 4.


:>ns can
oreheat:low pH
of the
(ot and

Deposition of Silica from Water

The deposition of monomeric silica involves a very different mechanism from the
deposition of colloidal particles. Monomeric silica forms an impervious glass-like
film, whereas colloidal particles form a porous film, often white and opaque when
dried. In an intermediate range, where particles are smaller than about 50 A in
diameter, that is, approaching molecular dimensions, the coating is clear and the
pores in the close-packed mass of particles may be so small that only water or
certain ions can penetrate. This borderline region has been little explored.
Silica can be deposited molecularly from su persaturated aqueous solution.
Supersaturation can be brought about by one of the following processes: .

.rn Si0 2

; (259).

He used
"hen ion



must be
. Apple; (70F)
:IS much
ppm is
of silica
irt (262)







ted with

1. Concentrating an undersaturated solution by evaporating water.

2. Cooling a hot saturated solution in water. '

3. Lowering the pH of an aqueous solution of a soluble silicate below about 10.7.
4. Generating Si(OH). in water by hydrolyzing a monomeric silicon compound such
as an ester, Si(OR)., hydride SiH., sullide SiS 2 halide SiCI. or even linely
divided elemental silicon.
Monomeric silica comes out of supersaturated solution in three ways:




oprn if a
1 as silialso the





1. As a deposit on a solid surface: Si(OH). condenses with any preexisting solid surface that bears OH groups with which it can react. namely. SiOH, or any MOH
surface, where M is a metal that will form a silicate at the pH involved. Once a
receptive surface is covered by the reaction









the further deposition is silica on silica, thus building up a lilm.

2. As colloidal particles remaining in suspension: if an insufficient area of a receptive solid surface is available to accept silica rapidly. and if the concentration of



The-Occurrence. Dissolution. and Deposition of Silica



Si(OH). is greater than 200-300 ppm (depending on pH), polymerization occurs

with formation first of low polymers such as the cyclic tetramer; then these
further condense to form small three-dimensional polymers which are colloidal
3. As biogenic amorphous silica: by mechanisms still unknown. living organisms can
remove silica from extremely dilute solutions and deposit solid silica within
themselves in precisely controlled structures of intricate design, perfect almost to
molecular dimensions (see Chapter 7).
The mechanism of molecular deposition of SiO: from Si(OH). is apparently the
reverse of dissolution of solid silica. It involves a condensation reaction catalyzed by
hydroxyl ions and accelerated by the presence of salts. The process therefore occurs
principally above pH 7, since it is catalyzed by hydroxyl ion, but obviously not
above pH II where silica dissolves as silicate ion. Deposition is more rapid and
condensation and dehydration of the silica are more complete in hot solution.
The silica deposit, when laid down monomolecularly, is nonporous and probably
as impervious as vitreous silica. For example, such a film only 20-30 A thick
deposited on the surface of a nickel powder renders the metal insoluble in acid.
However, it must be emphasized that special conditions are required for this to
occur. The degree of supersaturation must be sufficient for deposition, to occur, but
must not be so great as to allow the formation of colloidal particles. or the process
becomes very inefficient (266).
Deposition of colloidal silica occurs by a process entirely different from the above.
In this case. deposits on solid surfaces are formed by either or both of the following




1. Alternate wetting of dilute sol and drying can build' up an adherent coating.
Once the particles dry on a surface they are irreversibly bonded. Hard films can
be built up. especially if each additional layer of silica is only a few particles
thick. Much of the siliceous deposit at the waterline around hot springs is formed
in this way.
2. Deposition of colloidal particles on a surface from a dilute sol may occur when
conditions of pH and salt concentration are close to those causing coagulation or
precipitation. Deposition occurs when the colloidal particles collides and combines with the solid silica surface. The concentration of colloid must be suitably
low because under these conditions a concentrated sol would be rapidly coagulated or gelled. and no coating obtained. This is somewhat analogous to
molecular deposition, in that the concentration of silica must not be so high that
the particles collide and adhere to each other rather than to the solid surface.
Deposition of colloid is greatly accelerated and the deposits are much more dense
and hard when soluble silica is deposited along with the colloidal particles. Some
hot-spring waters contain soluble silica in equilibrium with the colloidal silica. and
some soluble salts. When the solution cools. the colloidal particles are deposited on
solid surfaces at the same time the cooled solution becomes supersaturated with

Removal and Deposition of Silica from Water.

of Silica

en these

silica, which then is deposited on the layer of deposited colloidal particles, cementing
them together.
Most silica deposits are probably built up in this way. They may be hard and very
adherent, yet are microporous, since usually there is not enough molecular or soluble
silica present to fill the pores between the colloidal particles. In nature. almost any
imaginable set of conditions can occur at one time or another. Further deposition of
molecular silica can occur until the siliceous deposit becomes completely impervious,
This effect is seen in some opal and in the walls of geodes.
In the laboratory, air-drying a sol of silica particles 1.5-2.0 nm in diameter on a
surface gives a glass-clear layer of gel in which the pores are too small to admit
nitrogen molecules and the BET surface area is very low. By first growing the particles to 3-4 nm diameter and then drying the sol in the same manner, a similar glassclear gel is obtained, but the BET area is 800-900 m 2/g- 1 However. the first gel is
still porous to water molecules, and so is not equivalent to vitreous silica.
On the other hand, when silica is deposited molecularly, the resulting solid silica
contains little or no adsorbed water and few internal silanol groups. It'is very similar
. to vitreous silica or "fused quartz," as it is commonly but erroneously called.


sms can
1 within
lmost to

:ntly the
lyzed by
e occurs
rsly not
pid and
A thick
in acid.
this to
.cur, but

Rate oj Deposition oj Monomeric Silica

e above.

ilrns can
; formed
ur when
ation or
ld comsuitably
I coagugous to
.igh that









re dense
s. Some
ica, and
lsi ted on









At the outset it must be emphasized that regardless of the type of material on which
silica is. deposited, after more than a monomolecular layer of silica has been
adsorbed, further deposition of silica is always on a silica surface. It is obvious,
therefore, that at the beginning the substrate must be receptive to monomeric silica
at pH 8-10, where deposition is carried out. Hydrophobic surfaces are obviously not
receptive, and must be rendered hydrophilic by adsorption of a cationic surfactant or
polymer or other means. 'Negatively charged surfaces such' as anionic ion-exchange
films, carboxymethyl cellulose. or carboxylated surfaces on surface-oxidized
polymers are not receptive. but can be made so by the well-known methods used for
preparing surfaces for the deposition of metals. such as treatment with stannous
chloride solution. Such surfaces can also be treated with polybasic metal salts of iron
or aluminum. which are known to reverse the charge on negative surfaces (mordants).
Neutral hydrophilic surfaces such as cellulose are not very receptive. but can be
made so by treatment with conventional inorganic mordants. All insoluble metal
oxides and silicates including most refractory materials and glass appear to be receptive to silica after the traces of organic fatty acids that invariably contaminate such
surfaces are removed. As soon as the receptive surface is immersed in a solution of
supersaturated silica. a monolayer of silica is adsorbed. probably as silicate ions.
It is obvious that silica will be deposited from supersaturated solution onto a solid
surface at a rate that increases with degree of supersaturation. Since the condensation reaction is catalyzed by hydroxyl ions. deposition is always carried out above
pH 8 and it progresses faster at high pH. Also. the presence of a monovalent electrolyte such as sodium chloride will no doubt accelerate deposition as it docs dissolution, and for the same reasons.


The Occurrence. Dissolution. and Deposition or Silica

On the other hand, there are upper limits to all these factors which place an upper
practical limit on the rate of deposition on a given substrate. Thus if the concentration of monomeric silica is too high, it polymerizes with itself, forming silica particles or nuclei on which silica will also deposit. Since these particles have a very high
specific surface area, much sf the soluble silica thereafter is deposited on these
colloidal particles, as well as on the substrate.
The pH cannot exceed about II or all the silica will be dissolved as soluble silicate, but it should not drop below about 8, since the rate of deposition will be slow.
Also, if the electrolyte concentration is too high, for example, 0.2 N, adsorbed
hydrated metal ions will remain in the deposited silica, leaving micropores when
removed at low pH. However, a 0.2 N concentration of sodium salt accelerates the
Although monosilicic acid, Si(OH)., is believed to be the material actually
deposited, it is possible to use "active silica" in the form of low molecular weight
'polysilicic acids (including extremely small colloidal particles) as a source of silica.
Such small particles are highly soluble and are in equilibrium with a concentration
of Si(OH). that is highly supersaturated with respect to larger particles or a flat surface.
Deposition of an impervious film of silica on a wide variety of surfaces has been
described in detail by Iler (266).
The active silica is defined as one that will depolymerize completely to soluble silicate in 100 min at 30C in an excess of 10- 2 N NaOH solution (pH 12). Such silica
is obtained when a dilute solution of sodium silicate (3.25 Si0 2 : 1.0 Na 20) is
deionized at ordinary temperature and contains monomeric' silica as well as particles
up to 10-20 A in diameter.
The rate of addition of active: silica to the reaction solution from which silica is
being deposited is preferably, but not necessarily, sufficiently slow for all of the silica
to be deposited on the substrate and not form colloid. This can be judged by determining the yield of silica on the substrate. However, in coating coarse powders or
other substrates of relatively low surface area, the silica yield is not as important as
long as the colloid formed can be separated from the substrate.
As an example, 400 g of powdered ferromagnetic nickel flake, cleaned by extraction with chloroform and alcoholic alkali, was suspended in 4124 g H 20 containing
71 g Na 2SO. and sodium silicate equivalent to 120 g Si0 2 and 37.5 g Na20 and
heated to 95C. Over a period of 3.5 hr 0.67 N H 2SO. was added to the well agitated mixture, and the pH dropped from 11.0 to 9.87. The metal powder contained
8.13% silica as a continuous film. corresponding to a 26% yield of silica. When compacted at 70.000 psi the billet had a resistance of more than 30 megohms. as compared to 30 ohms for the untreated powder. Also, the coated powder was much less
readily attacked by nitric acid.
Another method of operation is to put the substrate in a solution containing some
of the silicate and then adding sodium silicate and acid simultaneously to maintain
the pH around 10. When the same nickel flake was coated in this way the silica yield
was 80%. The coating thickness was estimated to be about 60 A.






.:..,. .. -

of Silica

Removal and Deposition of Silica from WateL_

nce-vra;a to
:ry high
in these

A convenient source of active silica is a 3% solution of silicic acid obtained by

passing sodium silicate solution through a column of sulfonic acid type ion-exchange
resin at such a rate as to collect an effluent with a pH less than 4.
Other materials coated in this manner were clays. asbestos fibers. aluminum flake
pigment. iron powder. and !l titanium dioxide pigment which showed tenfold
improvement in photodurability in an organic polymer. By using mordants. the individual fibers in cotton and nylon fabrics were coated with silica which amounted to
over 5% by weight.
The maximum rate at which active silica can be added to the system 'without
forming colloidal silica is expressed as follows:


ible silibe slow.

es when
ates the

. weight
If silica.
flat sur-




where n = (T - 90)/10 and S = amount of Si0 2 added (g hr- I g-I of substrate). A

= specific surface area of substrate (rn" g " '), and T = temperature (OC). Thus at
90(, silica can be added at a rate of 5 grams per 1000 m2 hr
Extrapolated to
30(, the rate would be only 0.08 g per 1000 m 2 hr- I.
In depositing silica on colloidal silica nuclei to increase the size of the particles,
Mindick and Vassos (267) set a maximum rate of addition of silica' (molecular
weight less than 90,000) to avoid forming new nuclei by a formula that gives 5.0
grams silica per 1000 m 2 hr " '. In this case no sodium salt is present in the system, so
that the maximum rate is only half of what could be used when salt was present.
To determine more accurately the maximum rate of deposition at ordinary
temperature, lIer carried out the following experiment.
To 1 liter of solution containing 350 ppm monomeric silica at pH 9. at 22C. was
added a small amount of a concentrated silica sol of 8 nm particles (specific surface
area 350 m2 g-I) to give a concentration of 3.0 g I-I colloidal silica. At once the
monomer content was followed as it dropped over a period of 12 hr. In the first 12
min the concentration dropped from 350 to 250 ppm, or a rate of 500 ppm hr- I.

las been
ible sili:h silica
'a2O) is
silica is
he silica
y deter/ders or
rtant as

0 and







en com3S comuch less


19 some



ca yield





Thus 0.500 g was deposited on 1050 m 2 of silica surface. The rate was thus about 0.5
grams per 1000 m2 hr ",
In the absence of the colloidal nuclei the monomer level in the solution dropped
only from 350 to 320 in 2 hr, indicating only minor self-nucleation. The equilibrium
solubility of silica in the system was about 100 ppm. so the average supersaturation
was about 200 ppm or three-fold saturation.
When the supersaturation was only 50 ppm, of 1.5-fold saturation. the deposition
rate was 33 mg hr ", or 0.03 gram per 1000 rn" hr- I, which is below the maximum
because there was no measurable rate of self-nucleation even at two fold saturation .
The maximum rate determined experimentally. therefore. lies between 0.03 and 0.5
g, whereas that calculated from the above formula was 0.15 gram per 1000 m 2 hr ' '.
The maximum deposition rate can be put in terms of coating thickness deposited
per hour. Film thickness for I gram per 1000 m', based on a silica density of 2.2



The Occurrence. Dissolution. and Deposition of Silica

cm -3, is 0.45 nm, or 4.5 A hr- I Thus at 90C, the maximum rate is 4.5'A hr and
at ordinary temperature is about 0.7 A br ", or I mm in 1600 years.
A silica coating was applied to particles of colloidal thoria by Barrett et al. (268)
by a simple heating process. A 5-10% deionized thoria sol was first added to a
strongly agitated 1-2% acidic sol of "active" silica. This coats the thoria particles
with a monolayer of silica. Then the pH of the mixture was suddenly raised to 10 to
stabilize the active silica which is converted to very small colloidal particles. Because
the silica-coated thoria particles are greater than 50 nm in diameter, silica dissolves
from the "active" silica particles and is deposited upon the thoria. The 'process is
completed by heating the sol mixture in an autoclave at 250C to obtain thoria
particles heavily coated with dense silica.
Also, a silica coating can be applied by the hot hydrolysis of a high ratio soluble
silicate. When sodium silicate with a ratio of 3.25 SiO~: I Na 20 is diluted from its
typical commodity strength of about 28.4% Si0 2 and 8.7% Na 20 to 2% Si0 2and
0.61% Na 20, the alkali concentration is not longer sufficient to keep at least some of
the silica from polymerizing. When such a dilute solution is heated to 80-100C, as
much as 20% of the silica is deposited on any receptive substrate surface, or
otherwise forms of colloidal particles. According to Azarian (269a) the presence of
sodium acetate is particularly benefical when apply a little coating in this manner to
bronze and other metal powders such as copper or aluminum. As discussed in a
following section, lithium polysilicate appears to deposit colloidal silica.
It should be mentioned that alumina can also be codeposited with silica as aluminosilicate ions in a silica matrix. Alexander (169b) patented a 'process for coating
particles up to 5 microns in size, including titania pigment with a skin from 3 to 50
nm thick composed of Si0 2 and Al 203 codeposited from sodium silicate and
aluminate solutions. The AI0 2 - is accompanied by Na + which can be ion exchanged
from the coating. This coating is much less soluble in water than pure silica. A very
similar process was later patented by Foss (269c) with an optional undercoat of
"sponge" alumina on the titania.


Silicification of Biogenic Materials

Our knowledge of life forms of the remote past is based on examination of

specimens preserved as silica pseudomorphs. Best known of these is petrified wood,
but of even greater fundamental importance are the fossilized microorganisms discovered by Barghoorn and Schopf, which are as much as 3 billion years old (270).
The latter were apparently coated with, then later impregnated with and enveloped
in, amorphous silica, which later was converted to a very fine-grained black chert
(submicroscopic quartz), which preserved the form of the filamentous algae and bacteria-like organisms. Small amounts of organic material are still present.
The most perfect replication of the organic forms is by silica, which is the one
mineral that is transported in solution in water and deposited in amorphous form.
All other minerals are deposited as crystals which, in most cases, are larger than the
detailed structure of the organism. Even after amorphous silica has crystallized in







,t .



Removal and Deposition of Silica from Water

n of Silica


situ to chert, the crystals are so small that even submicroscopic structures are
retained, visible only by electron microscopic techniques.
The unusually perfect preservation of the original structure of plant tissue is exemplitied in Figure 1.16 (271).
The natural process by which organic material is so faithfully reproduced as a
pseudomorph of silica is not yet understood and has not yet been duplicated in the
laboratory. It should be possible to generate silica gel within the spaces in plant
tissues which are normally" tilled with water, but the subsequent replacement of the

hr" and
at. >8)
lded to a
j to 10 to
. Because
orocess is
in thoria
o soluble
from its
Si0 2and
t some of
-OOC, as
rface, or
esence of
tanner to
ssed in a
as alumir coating

a. A very
ercoat of

ration of
ed wood,
isms disild (270).

ick chert
and bac; the one


than the
rllized in






Figure 1.16. Photom icrograph of a section of silicified wood-Pt(!rocaryoxl'/oll Know/toll

(65 x). (Courtesy or Elisabeth Wheeler. North Carolina State College. E. S. Barghoorn. Harvard University, and Richard Scott. U.S. Geological Survey.)


The Dccurrence, Dissolution, and Deposition of Silica

organic material with more silica would be difficult, and in nature is undoubtedly an
extremely slow process. The first stage of the process can be observed by placing
specimens of wood, for example, in hot mineral springs rich in silica, such as the hot
springs of Yellowstone Park, which contain up to 717 ppm of silica (272,273).
Silicification of wood is commonly associated with volcanic ash, which is a rich
source of readily available soluble silica (274). Correns (275) suggests that the silica
may be precipitated from alkaline natural waters by the carbon dioxide evolved during decomposition of the wood. In this way, silica would be deposited immediately
at the surface of the organic material, and as the organic portion dissolved 'away, it
would be replaced by silica. This presupposes that the silica initially formed is amorphous and porous, permitting diffusion of solution through the specimen: Since plant
tissues contain membranes that can be penetrated by soluble silicic acid but not by
colloidal particles of silica, Hellmers (276) believes that silicification occurs
immediately after the soluble silica is liberated by decomposition of silicate minerals
and before it can polymerize.
Barghoorn (277) has commented as follows on the mechanism of silicification:
Silicified wood, regardless of geologic age, contains unusual amounts of organic residues,
occasionally present to such an extent that the demineralized wood may be embedded, sectioned, and stained, much as living tissue.

However, in other cases, there is no cellulose remaining.

It seems likely that silicification took place at a time when degradation of the cell wall had
not yet passed the stage of the partial retention of a cellulosic structural residue. Thus, the
cellulosic framework of tissues was retained for sufficient time to allow preservation of structure before silicification. After"or during silicification, the remaining cellulose was lost from the
tissue, leaving a modified but coherent lignin residue.


The precipitation of minerals such as silica therefore fixes the structure in a

mineral form which retains a large measure of biological detail.
As pointed out by Correns (275), the liberation of acidic material, such as carbon
dioxide, by the decomposing vegetable matter may also lead to the formation of a
type of silica different from that of the original embedded matrix. It would be
expected that cellulose will decompose by hydrolysis and oxidation at a much faster
rate" than the lignin, in view of the known differences in chemical stability of these
materials. Thus the cellulose is likely to be replaced by silica before the lignin is
The final stage in the formation of an organic-free dense fossil involves the final
filling in of the pores in the original porous silica deposit and the gradual conversion
of part or all of the specimen to crystalline silica. Because of the original variations
in the silica which replaced the different types of organic material, the final
crystallization will be varied and the pseudomorph will be visible even when it
contains few impurities to give it color. Ordinarily, iron is present, and the resulting
variations in color make the original organic structures plainly visible.


of Silica

.tedly an

s t.


, ).

is a rich
he silica
ved duriediately
away, it
is amorice plant
t not by
ded, sec-



rhus, the
of strucfrom the

ire in a

; carbon
ion of a
ould be
.h faster
of these
lignin is
he final
ie final
when it

Removal and Deposition of Silica from Watsr,


The silicification of various types of biogenic materials was reviewed' by Siever

and Scott (278). Possible mechanisms are discussed, but they believed that exact
duplication of the silicification process in the laboratory is impossible, because of the
time required. The youngest silicified wood is of the early Pleistocene age.
It is not only a matter of how rapidly silica is deposited, but also how rapidly the
organic material inside a partially silicified specimen decomposes in the absence of
microorganisms within the structure. Chemical degradation must occur before the
resulting space can be filled by further silica deposition. This fact, furthermore, precludes deposition of colloidal silica except initially in the open pores, since' colloidal
particles cannot pass through the cell walls. For this reason, silicification for the
most part involves diffusion of a supersaturated solution of soluble Si(OH)4 through
the structure. If the process occurs at ordinary temperature, the silica concentration
is not likely to be more than 200 ppm and the rate of deposition not more than 1
mm in 1600 years. Since rate of diffusion, even through micropores, is relatively
rapid, it means that silica deposition can continue to occur even in cavities to which
there is access only through a few pores. Thus dense, almost impervious masses of
silicified structures can be formed. It is difficult to v isualize how this could be
accelerated, say, a thousandfold, in the laboratory.
The most successful approach to duplicating fossilized wood has been made by
Leo and Barghoorn (279), who also presented a survey of the subject since the
attempts of Basil Valentine, the alchemist, in 1520. The wood specimen is boiled in
water until free of gas and then alternately immersed in separate sealed jars containing water and ethyl orthosilicate, stored at 70C.
Exposure time in each fluid may vary from a few days to a month and the
alternate immersions are continued for us to 1 year or more. Several times during
the ethyl silicate immersion vacuum is applied to remove water vapor from the wood
to facilitate further penetration by the ester. The ethyl silicate is replaced with fresh
liquid whenever it becomes viscous or cloudy. During the final cycle in the ester the
specimen is soaked in 0.004% HN0 3 to hydrolyze the residual ester in situ.
The silica-filled specimen still contains all the original organic matter and closely
resembles natural petrefaction of geologically young age. After organic matter has
been removed, as shown by a fragment not turning dark when it is place in
concentrated H 2S0 4, the silica Iithomorph is found to faithfully reproduce the
original organic structure. Unlike previous methods, this technique gives a strong
coherent lithomorph.
Since the organic matter can be removed by oxidizing agents it is obvious that the
lithomorph is highly porous. Presumably, this would be an ideal substrate for further
deposition of silica, perhaps colored with a little iron, to more closely approach the
appearance of a natural, highly petrified specimen.
Duplication of chert containing microorganisms is feasible in the laboratory
because diffusion distances are so small. Oehler and Schopf (280) made specimens
by embedding filamentous algae in silica gel and then autoclaving at 2-4 kilobars for
2-4 weeks at 150C. Under these conditons, the gel undergoes syneresis until completely solid and is converted to the microcrystalline state, as further described by
Oehler (l64b).

. The Oeeurrence, Dissolution. and Deposition of Silica


Rare of Deposition of Colloidal Silica

There are some situations, as in equipment handling hot geothermal brines, where
the cooled liquid deposits silica orders of magnitude faster than is possible by the
deposition of molecular silica. Such deposits invariably are hydrated. and although
often very hard, have a microporous structure. A high buildup rate is to be expected,
since particles corresponding to molecular weights of thousands (several nanometers
in diameter) are being deposited instead of single SiO, units. Also porosity, corresponding to the spaces between the colloidal particles, is to be expected. . .
There is one major difference in mechanism: deposition of colloidal particles
requires the presence of a potential coagulating agent, usually small concentrations
of polyvalent metal ions, although monovalent ions such as sodium have a similar
effect above about 0.3 N concentration. In the absence of flocculating ions, in neutral or alkaline solution, a colloidal particle of pure silica bears a negative charge
and so does the silica surface of the substrate. There is mutual repulsion so that the
collision rate is low. However, in this pH range metal ions are adsorbed on the silica
surfaces in some degree and upon collision, adhesion occurs. For deposition on a
surface to occur, the metal ion should not be present at such concentration as to coagulate the colloidal particles in suspension (Figure 1.17).





MilI :.


Figure 1.17. Flocculating action of calcium ion followed by spontaneous cementing effect
through dissolution and redeposition of soluble silica.



.. -

-- .:-



Removal and Deposition of Silica from WateL_

n of Silica

ile by the
sity, corparticles

a similar

s, In neuIe charge
J' that the
the silica
tion on a
1S to coa- .


ing effect



Deposition is likely to be unusually rapid when the colloidal particles are less than
5 nm in diameter. Such small particles are in equilibrium with a concentration of
monomer that is supersaturated with respect to a flat surface. Hence, as shown in
Figure 1.17, the surface of deposited silica rapidly fills in so that the radius of curvature is much greater than that of the colloidal particles. In effect, the surface tends
to be filled in and flattened out. This is an important factor in rapid desposition
because the small colloidal particles are then much more likely to adhere to the surface than to each other in suspension.
Also, the presence of small colloidal particles, especially in a hot sol that' is being
cooled, furnishes a high degree of supersaturation that causes rapid polymerization
of monomer at the moment of contact between particle and surface (Figure 1.17),
.thus increasing the rate of deposition beyond that initiated by the divalent cations.
As pointed out by lIer (281) in the slightly alkaline pH range, a concentration of
100 mM calcium ion is required to flocculate 4 nm particles (700 m 2j g- l ) whereas
for very large particles only abou t 10 mM is required. A flat surface corresponds to
the radius of curvature of a very large particle. Hence when 4 nm particles are in the
neighborhood of a relatively smooth surface, there is an intermediate calcium
concentration that will promote adhesion of particles to the surface, yet not cause
flocculation in solution.
It is probable that deposition of more than a monolayer of colloidal particles
larger than about 10 nm by the described local coagulation mechanism is unlikely to
occur. Much more monomeric silica would have to be deposited to smooth out the
surface. It is conceivable that by careful addition of monomeric silica to the sol to
maintain an optimum degree of supersaturation along with careful control of the
concentration of coagulent, continuous deposition could be maintained without coagulating the sol.
The importance of flocculating ions is borne out by Midkiff (282, 283) who found
that cooling water containing more than 300 ppm silica did not deposit a scale if cal. cium ions were first converted to soluble chelates: otherwise, a deposit of colloidal
silica. associated with calcium carbonate, was formed. At the concentration
involved, over half of the silica in solution must have been colloidal and the particles
quite small. The rate of deposition was over 100 times faster than would be possible
if monomeric silica alone were involved. Wohlberg and Bucholz (246) reported that
if the silica concentration exceeded about 240 ppm (at which point colloidal silica
would be expected to be nucleated). scale formation occurred when calcium was
The deposits from hot deep-well brine near the Salton Sea in Cali fornia build up
very rapidly as the brine cools while going through pipes. The silica content is 400
ppm in a solution containing up to 15% NaCl as well as,a few percent of CaCl 2 and
KCI. The brine is slightly acidic. so there is no interaction of silica with calcium ion.
but iron. which is present at only 0.2%. is adsorbed on silica at this pH and is a
major component of the scale. More striking is that up to 20% copper and 6% silver
are found in the scale as sulfides. The deposit is amorphous to X-rays and consists of
a hydrated silica. classed as opal, but is actually a microporous silica gel. under the
coagulating influence of the metal ions. Since the brine contains 1-2 ppm H 2S, the

- _.-.

The Occurrence. Dissolution. and Deposition of Silica

adsorbed metal ions eventually migrate and nucleate as fine sulfide crystals embedded in the silica.
The difference between depositing material molecularly and as particles has been
noted in other systems. Thus Howard and Parfitt noted that in depositing silica on
titania pigment either a layer, could be deposited as polysilicate ions by an
isothermal, pH-dependent process or colloidal silica particles could be deposited by
a coagulation mechanism (284a).
Silica is rapidly deposited on receptive surfaces from a hot lithium silicate solution
of 2 Si02 : I Li~O ratio. Sams (284b) describes the character of the coatings obtained
on glass from solutions of different ratios of silicate, from a solution containing
12.5% SrO, heated at 95C. The opaque white color of coating deposited during 1~
hr indicates that it is porous and that the silica was deposited as colloidal particles.
Even after only 60 sec the appearance of a hazy blue coating (after being washed and
dried) indicated that colloidal rather than monomeric silica was being deposited. It
was essential that the substrate be immersed in the solution as it is heated. Merely
drying the solution on a surface gave no adherent film. This behavior of 2: I lithium
silicate is associated with its unusual property of becoming insoluble and forming a
precipitate, apparently amorphous, when the solution is heated above about 48C,
yet redissolving slowly after the solution is cooled to ordinary temperature.
However. at 100C, the precipitate apparently disproportionates to insoluble silica,
probably colloidal in size. Further discussion of forming films from colloidal silica is
reserved for Chapter 4.

Various methods of analysis are involved in every aspect of silica chemistry, For
convenience, sources of information are assembled here, along with a few methods
particularly useful for research purposes.

Atomic Absorption



Analysis for silicon in the larger laboratories has been revolutionized by the atomic
adsorption method. Although equipment represents a large investment, the method
can be used for a wide variety of elements and, once samples have been prepared in
solution, permits dozens of samples to be run in a few hours. Usually the instrument
manufacturers can recommend suitable methods for preparing solutions for analysis.
Bowman and Wills (285) have recommended specific procedures for silicon.
Dissolution of solid samples and preparation of suitable solutions have been
described by Terashima (286). Dissolution of mineral samples in H3PO~ is a convenient method, according to Horton and Baines (287), especially since it eliminates
the background correction when the silicon is dissolved after alkali fusion. Spectral
interference of vanadium can be a problem (288). The method is ideal for determining total silicon.

of Silica

Methods of Analysis




:- .. -


A wide range of chemical methods and procedures is found in the treatise of

Kolthoff and Elving (289). They also give a summary of the chemistry and solubility
of silica. Preparation of solutions for analysis, the silicomolybdic acid yellow and
blue colorimetric methods including interferences, gravimetric procedures, and special procedures for biological materials' are discussed in concise detail.
Procedures especially suited to silicate rocks, minerals, and refractory silicates
and aluminosilicates have been described by Bennett and Reed (290). A history of
analytical methods was published by Andersson (291), and three new spectrophotometric procedures were developed. Available chemical methods listed by
Meites (292), in addition to the conventional gravimetric and colorimetric methods,
also include precipitation of the silicate ion as the cobalt salt, which is then
determined by chelometric titration, and as a nitrogen base salt which is titrated
with perchloric acid.
It has been my experience that for research purposes, in addition to the atomic
absorption method for total silica, the alkali titration of silica as SiF.2- is most useful for concentrations greater than 0.1 %, and the yellow and blue silicomolybdate
methods for concentrations down to I and 0.1 ppm, respectively.


silica on
, by an
.sited by
uring I ~

.hed and
sited. It
rming a
It 48C,

Ie silica,
silica is

Methods Involving Silicomolybdic Acid

The reaction of molybdic acid with monomeric Si(OH). to give the yellow
silicomolybdic acid is indispensable in investigating the behavior of soluble and
colloidal silica. The literature is too voluminous to cover here. Morozyuk (293)
made a chronological list of the literature to 1971.
This section deals with the chemistry of the reaction and gives some recommended
analytical procedures. Application of the reaction to determining the nature of polysilicic acids and colloidal particles is reserved for Chapter 3.
It is obvious that only Si(OH)., but not polymers thereof, can react directly with
acidified ammonium heptarnolybdate to form the yellow silicomolybdic acid, since
the latter molecule contains only one silicon atom:

try. For

; atomic
Jared in
le been
; a conrninates


Since the molecular weight of ammonium molybdate is 1235.9. I g SrO, consumes

35.3 g of the ammonium molybdate.
Polysilicic acid de polymerizes slowly enough that it is possible to determine the
monomer in the presence of polymers by noting the rate of color development



., ..
, It. ;j



The-Occurrence. Dissolution, and Deposition of Silica

Although this reaction was discovered in 1898 by Jolles and Neurath (298) and
used for many years. it was only in 1952 that Strickland (299) showed that variations in the extinction coefficient under some conditions were due to the existence of
two forms of molybdic acid. alpha and beta. that react with silica to give yellow
silicomolybdic acids having, different extinction coefficients and absorption peaks.
The beta form is obtained at lower pH and is used in the "yellow" method. but
unless conditions are optim urn, it changes with time to the alpha form, which is less
colored but more stable. The beta form is the only one that is used for determining
monosilicic acid in the presence of polymers because it is formed at once. and its
color is more intense.
To eliminate the problem of beta changing slowly to alpha, Garrett and Walker.
(300) proposed measuring the color at 335 nm, where the alpha and beta forms have
the same absorption coefficient. They also studied the kinetics of formation of
silicomolybdic acids in very dilute molybdic acid solutions (0.0025 M) and concluded
the rate was proportional both to silica concentration and molybdate ion concentration; the alpha .forrn had a constant optical density between pH 2 and 4.5. Alcohol
catalyzes the color formation, especially when present at about 30 vol, %. even at
pH 5.
In contrast to the above. Andersson (30 I) reported that at a wavelength of 325
nrn, at pH 1.5. alpha silicomolybdate had a very low extinction coefficient as compared with the molybdic acid blanks, and for this reason he used 400 om. However,
his method involves a heating step to convert the beta to alpha form. and so is timeconsuming and inconvenient for studying silica polymerization. It is useful for
accurate determination of total silica with a standard deviation of less than 0.5%.
The Beta Silicomolybdate M ethod
This method was used long before Strickland's (299) discovery of the existence of
the alpha form. Thompson and Houlton (302) had used conditions that resulted in
the beta form and later Alexander (303) used it in his study of monosilicic
. Govett (304) determined the critical factors in obtaining the beta form:



I. Adjust the stock ammonium molybdate solution to a pH of 7.5 with NaOH to

ensure that only MOO.2- ions are present.

2. Use an amount of molybdate in the reaction solution to give a 0.06 M concentration of MOO.2-.
3. After the reaction with silica there should still remain a 0.05 M concentration of
M00 4 2- ; that is. a fivefold excess should be used.
4. About 2.7-5.0 equivalents of H 2S0 4 should be added per mole of MoO/-.
Guignard and Hazebrouck (305) further examined this method and found that
over 3-4 hr the absorbance increased and then decreased as alpha was formed. In
20-50% alcohol. the beta form was more stable.



Methods of Analysis

of Silica

A Recommended Procedure

298) and
at v-riaste.

For most purposes I have found, it is possible to make up a dilute molybdic acid
reagent solution that is stable for a week. Thus to determine silica it is necessary
only to add the sample solutio~. Standards should be run each day.

e yellow
n peaks.
hod, but
ch is less
. and its



Add 41.0 011 95.5% H 2SO. to 800 ml water and dilute to 1 liter
(1.5 N) .


'ms have
at ion of



even at

Dissolve 100 grams (NH.)6M0102. 4H 20 (mol. wt, 1235.9) in

900 ml H 20, add 47 ml concentrated NH.OH solution (28%
NH 3 ) , and dilute to 1 liter (0.566 M MOO.2- and 1.18 N NH.';'
To 500 011 H 20 add 200 ml solution A and 100 011 solution B
(800011 total). This is 0.0707 JI;[ with respect to MoO/-, 0.148 N
in NH.';' ion, and 0.375 N in SO/- and has a pH of about 1.2.

To 40 ml reagent mixture C add up to 10 ml of sample solution

containing not more than 2000 p.g Si0 2, and adjust the volume to 50 011. Measure
absorption at 410 nm wavelength.

h of 325

as comIowever,
is time.eful for

The number of H.;. ions added per MOO.2- ion is 5.3. which is slightly more than
recommended by Govett. If 2000 p.g of Si0 2 is present. it will consume 0.2 millimols
MoO/-. The mixture contains 40 x 0.0707, or 2.83 millimoles MOO.2-, so there is
a sevenfold excess.
This composition is compared with that of Govett, Alexander, and Kautsky et al.
(306) as follows:

tence of
suited in
cic acid.

Composition of Color-Forming Mixture

aOH to

Kautsky et al.


mg Si0 2/ 50 01 I

Molar MoO/0.0566



ation of
Her found that with a H": M 00. 2- ratio of 5.3. the p H was 1.2 and monomer
reacted fully in 2 min. and the color was stable for 2 hr. At 3.7 ratio. the pH was 1.5
and the monomer reacted fully in I min. but the color began to fade in less than 100
min. At a ratio of 10.6. at a pH of 0.8. the monomer did not completely react in 100
min. The reaction rate appears to be proportional to the hydroxyl ion concentration.
even at this low pH.

md that

med. In



The-Occurrence, Dissolution. and Deposition of Silica

Interfering Substances
Iwasaki and Tarutani (307) found that in concentrated salt solutions, the
silicomolybdic colorimetric method gave low values. (Presumably the method can be
standardized using salt solution.)
Fluoride ion interferes if it is present before the molybdic acid is added (308). It
can be masked by Al H
Tannins, especially in natural waters, interfere with the yelIow silicomolybdate
method because of their yellow color. Thus 6.8 ppm of pyrocatechol tannins or 2.7
ppm of pyrogallol tannins give a color equal to 1,0 ppm of silica. Reduction of the
yellow to molybdenum blue by sulfite at low acidity (pH 2.5) and measurement of
light absorption at 620 nm avoid the problem, and at the same time elirnihate
interference by phosphate ion (309).
Traces of ferric iron may contribute to the yellow color and for this reason
Kenyon and Berwick (310) prefer to have tartaric acid present to form a less colored
complex with iron. In analyzing caustic liquors they added varying amounts of
sodium or potassium chloride, even though color is suppressed, because this permitted them to maintain a constant salt level when the alkali samples were neutralized
with hydrochloric acid.
Interference by phosphate ion is an especially common problem, Since the phosphate ion reacts like silica to form a yellow phosphomolybdic acid, its' interference
must be eliminated. Numerous techniques have been proposed. either for separating
the silica and phosphorus before analysis or preferentially reducing silicomolybdic
acid to molybdenum blue in the presence of the phosphomolybdic acid (311-313).
Snell and Snell (314) summarized the possible procedures: (a) precipitating and
removing phosphate as the calcium salt. (b) adjusting pH so only silica will form the
yellow color, (c) destroying the yellow phosphate complex with citric. oxalic, or tartaric acids, and (d) preferentially reducing the silicomolybdic acid to molybdenum

Molybdenum Blue Method

:; i

f t


j I

:i'I I'

:; I

" I


~ I




When the silica concentration is only a few parts per million, the absorbance of the
yellow complex is insufficient for accurate measurement. If the yellow complex is
reduced to molybdenum blue, and a reagent is used that will not reduce the residual
excess of molybdic acid, the much greater absorbance increases sensitivity perhaps
The best reducing agent, according to a study by Mullen and Riley (315). is a mixture of rrietol, sulfite. and oxalic acid which gives complete reduction in 90 min at
20C and stable color for 48 hr. Their method measured silica at a concentration of
0.4 J.Lg of SrO, in 20 ml of sample (0.02 ppm) with a standard deviation of I % and of
about 0.3% at concentrations up to 6 J.Lg (0.3 ppm). Possible interferences by all
common metals and anions were checked. eerie ion, fluoride, germanium, and vanadium gave serious interference; interference by phosphate was eliminated by the
oxalic acid. Their method was further developed by Yolk and Weintraub (3 J 6), expc-

of Silica


Methods of Analysis

cially for analyzing plant tissues. which were first ashed and fused with l'::lzC0 3 The
following reagent solutions are prepared in plastic bottles.


d can be



(308). It


1S or 2.7
>n of the
-rnent of


ounts of
; permitrtralized

he phos.rference


:ing and
form the
:. or tar'bdenum


Dilute a 1-20 ml sample containing 10-50 J.Lg Si0 2 to 20 mJ. Add 3
ml solution A and let stand 10 0.5 min at 25C. Add 15 ml reducing solution E
and dilute to 50 mJ. Wait 3 hr and measure absorbance at 810 nm wavelength.

A similar method is described by Jarabin, Vajda. and Szarvas, except that the
color is measured at 660 nm (317). Another modification of the molybdenum blue
method is to develop the color in perchloric acid medium, by reduction with stannous ion and ascorbic acid (318), which eliminates interferences and gives a stable
color. Kahler, Betz, and Betz (309) and Milton (319) favor reduction with sulfite ion
for water analysis.
For Biological Samples

For analyzing biological materials the molybdenum blue method is invariably

used because of the very low concentrations of silica. For this purpose a method of
analyzing for traces of silica in the presence of iron. phosphorus. arsenic, and reducing substances was developed by Baumann (320).

:e of the

nplex is
s a mi:<-

) min at
'ation of
7cJ and of
;s by all
rd vana-

I by the


20 g I-I (NH~)dMo702~4H20; 60 ml I-I concentrated HCl

20 g I-I 'S-amino-2-hydroxytoluene sulfate (metal); 12 g I-I
Na2S0~ (discard when colored).
100 g I-I oxalic acid.
9 N H 2S04 (250 mil-I 95% H2S0~).
Reducing solution: 100 ml solution B; 60 ml solution C: 120 ml
solution D; 20 ml H 20 (store refrigerated).


Acetic acid.
17.46 .grams NaHS0 44H zO + 23.71 grams Na2S0~.IOH20
made up to 100011 (2 M SO/-).
(C) 10% ammonium molybdate solution.
(D) 42 grams citric acid + 200112 N HCI made up to 100 m1.
(E) 0.20 gram l-amino-J-naphthol-ri-sulfonic acid dissolved in 100
ml solution with 2.4% Na 2SO . 7H 20 and 12% Na2S20~. Before
each use, 0.2 gram ascorbic acid is added per 10 011.
(F) 0.1 N KMnO~ solution. freshly prepared.
(G) 0.5 M K2C0 3 + 0.5 M Na ZC0 3 solution.
(H) 2 N HCI made by distilling HCI. titrated against solution G.

In a 25 011 flask put 5 011 acetic acid. 10 - y 011 H20. and y 011 of
sample. and add 2 1111 of sulfate solution B. Add pcrrnangunate solution F to a violet
color and 1-2 ml of molybdate solution C. Heat 7 min in water bath at 60C. Cool 5




. :!:


The.Occurrence, Dissolution. and Deposition of Silica

min in water at room temperature. Add 5 ml citric acid solution D. Wait 20 min for
the phosphorous complex to be destroyed and add 0.5 ml of reducing solution E and,
after another 20 min, measure the extinction coefticient with light of 800 nm. As a
blank use a mixture of 5 ml acetic acid, 18 ml H 20, and 2 ml sulfate mixture B. The
color solution must be kept out of sunlight. Results by this method show that earlier
data on the silica content of blood, for example, were too high; by this method ox
blood contained only 2.2 ppm Si0 2, and this was present only as soluble Si(OH)4'
Proteins are removed from samples earlier by combustion or ultratiltration.'
A very small amount of silicon in the presence of a hundredfold amount of phosphorus was determined by Austin, Rinehart, and Ball (321a), who used a combination of H 2SO., preferred by King et a1. (321b), along with tartaric acid, preferred by
others (32Ic).

Methods of Concentrating Silica for Analysis

For samples containing less than 0.1 ppm soluble Si0 2 it may be necessary to
concentrate the silica before analysis. Some methods are as follows:
I. Add NaF and H2S0 4 to the sample in a lead still and distill SiF 4, up to 140C,
into H 2 0 (314).

2. Accumulate soluble silica as the silicate ion on a strong-base ion-exchange resin

and elutriate in more concentrated condition with base (322). This method was
used on samples containing 10 ppb silica.
3. Silicomolybdic acid can be concentrated by extraction with n-butvl alcohol
followed by reduction to molybdenum blue with ascorbic acid. The method is said
to be useful for determining silica in aluminum oxide (323).
4. Convert silica to SiF 6 2 - and absorb on ion-exchange resin. then elutriate. Useful
down to I ppb (324, 325).
5. Convert the silica to an amine salt of the silicornolybdenurn blue complex and
extract with chloroform. Especially useful in separation from metal salts (326).
The reaction product of laurylarnine with ethylene oxide was used.
6. Convert the silica to the Rhodamine B salt of silicornolybdic acid which is
extracted with isopropyl ether and measured at 590 nm wavelength. Arsenic and
phosphorus do not interfere in hundredfold excess over silicon (327).
7. Convert silica to catechol complex sail of an organic base and extract with
nitrobenzene ( 98).
To concentrate colloidal silica for analysis. coagulating procedures can be used:
I. Freeze the solution to coagulate the colloid so it can be filtered off (328).

2. Coagulate with polyvinyl alcohol at low pH (329). Gelatin has been used in the

rf Silica

Methods of Analysis

min for

Depolymerizing Colloidal Silica Before Analysis

E ~--i ,
n.. .t
B. The

.hod ox

f phos-

rnbinarred by


Clear solutions of silica may contain polysilicic acids or small colloidal particles that
will not react completely with molybdic acid. Thus before determining total silica,
depolymerization to monomer is necessary.
Okura (296) recommended the following procedure: to 50 ml of the hot sample in
a platinum dish add 4 drops concentrated HzSO., 1 ml 10% ammonium molybdate
solution, and 20 mg NaF, in that order; cool and measure the color. On the other
hand, Tarutani (308) stated that F- ion interferes with the silicomolybdic method
and must be removed by cornplexing with boric acid or AP+. The difference
probably is that Okura added fluoride only after adding the molybdate. Shapiro
recommended a method of depolymerizing the silica with F- for the molybdate
method, rather than using alkali (330).
Govett (304) recommended that solid samples be fused with NaOH (3 g per 0.1 g
sample) in a nickel crucible for 10 min, leached into a plastic beaker, and acidified
with 1 N H 2SO. to pH 1.5. (This should be done quickly.) Aqueous samples should
likewise be boiled with solid NaOH in a nickel beaker and then diluted.

sary to
Standard Silica Solutions

:e ~- ''1
od . _.,;

is said
ex and
hich is
tic and

:t with

Neutral solutions of silica in seawater containing about I ppm silica can lose silica
on standing in polyethylene bottles (316). but there was no loss at low pH. It was
thought that this could have been caused by adsorption of silica by the plastic at
neutral p'H, but it is more likely caused by aggregation and settling owing to traces
of metals such as aluminum in seawater. No such interaction occurs at low pH.
Solutions appear to be quite.stable at pH 12. but the pH may drop in storage owing
to absorption of atmospheric carbon dioxide. Solutions containing up to 80 ppm
should be stable in plastic at pH 2. Standard solutions can be prepared from pure
Na 2Si039H 20, but fresh material must be used since the water content can change
in storage.
The simplest procedure is to dissolve 1.00 gram of pure microamorphous silica
(so-called c.P. Silicic acid) that has just been heated to 800C in about 4 grams of
fused NaOH in a nickel crucible, cool. then add 20 ml water and heat until all is
. dissolved. dilute to I liter. and store in a cleaned. scaled plastic bottle. Then dilute
40.0 ml of this solution ( 1000 'ppm Si0 2 0.1 N alkali) by pouring it into stirred 600 ml 0.1 N H 2SO. and make up to I liter to give a stable 40 ppm standard
solution of Si0 2 Store in a polyethylene bottle that has been previously cleaned by
filling it with hot 5% NaOH solution. letting it stand a d3Y. then rinsing it with
water and 0.\ N H2S0~ solution.

Miscellaneous Colorimetric Methods

i r,

Snell and Snell (31,,0 recommended a simple colorimetric test de veloped by Berg and
Teitelbaum ill which 1 rnl of sample solution containing no more than 4 pg of silicon


The.Dccurrence, Dissolution, and Deposition of Silica


(2 ppm) is mixed with 1 ml of 5% ferric chloride solution and 0.5 ml of a I % alcoholic solution of pyrrole and the mixture diluted to 10 ml. The color is compared
against standards. Sensitivity is 1 ppm Si02 The only interfering substance is selenium.
A turbidometric method based on the extreme insolubility of the quinine salt of
silicomolybdic acid was developed by Defasse (331). The reagent is a mixture of quinine sulfate, ammonium molybdate, H2S0~, and HNO~, and owing to the large mass
of precipitate' formed by a small amount of silica, the turbidity permits rapid
determination of silica down to 1 ppm in water as long as phosphorus is not present
at concentration greater than 1 ppm.
After converting silica to alpha-12 molybdosilicic acid, the latter can be titrated
with ferrous sulfate in the presence of EDTA, tartaric acid, and chloroacetic acid
buffer at pH 2.5. The end point is determined amperometrically with two platinum
electrodes (332).

Detection of Colloidal Silica on Surfaces

To detect whether paper or a textile material has been treated with colloidal silica,
Turner (333) sprayed the surface with a colorless. 1-2% solution in a hydrocarbon
solvent of a lower-alkyl amino-substituted triphenylmethane lactone. In contact with
the dry silica surface the silanol groups open the lactone ring, which converts' the
compound to a deep blue color. When the surface is then wetted with water or
alcohol, the color disappears, but returns when the surface is dried.

Rapid Titration of Total Silica as Fluosilicate

To determine silica in solutions containing more than about 0.1 % Si0 2, a convenient
and rapid method developed by Konig (334) is based on the conversion of Si0 2 at
ordinary temperature to H 2SiF. with excess HF; the excess is neutralized with
NaOH without decomposing the SiF.2- ion. Then in hot solution, SiF.2- can be
hydrolyzed quantitatively by titration with NaOH. The solution must be free of

CO 2 ,
Halfter (335) had already applied a similar but less versatile method for determining silica in soluble silicates. Another variation was adapted to determining silica in
fluosilicic acid (336).



Indicator mixture: 0.2% bromothymol blue and 0.1 % phenolphthalein.

48% HF reagent grade.
Powdered NaF.
3.0 N NaOH standard solution.
Boiling distilled water.




Methods of Analysis

o alco-

PROCEDURE. (I) In a clear 500 ml wide-mouth plastic bottle place the sample, liquid
or solid, containing about 0.1-0.3 gram Si0 2 , and add water to give a total volume
of usually no more than about 20 ml.
(2) Add about 1.5 ml 48% HF, preferably dispensed from a plastic pipette using a
bulb or plastic syringe for suction (see Warning below).
(3) After stirring the mixture gently and letting it stand for a minute or so, add 3
grams of powdered NaF and stir until dissolved.
(4) Add about I ml of the indicator mixture and, if necessary, add concentrated
HCI dropwise to turn the solution yellow, avoiding excess.
(5) Add 3 N NaOH from a burette dropwise until the color changes from yellow
to blue-green.
(6) Add 200 ml boiling distilled water (which has been freed of CO 2 by boiling).
(7) Note burette volume and titrate dropwise with swirling or stirring until the
color changes to a deep blue.


is ~
salt of
of quie mass
ic acid

ct with
rts the
ite- -r

;i0 2 at
j with
can be
ree of


NOTE. Preferably a size of sample should be taken to give a titration of 3-10 ml.
Silica content of sample is calculated on the basis that 1 ml 3 IV NaOH = 0.045
gram Si0 2
Stirring is most safely accomplished with a Teflon's-coated magnetic stirring bar.
Although less convenient, it probably would be just as effective to use NaF and
50% H2S04 instead of the 48% HF solution.
WARNING. HF in any concentration is a dangerous reagent and rubber gloves
and face shield should be worn. For additional safety, the upper part of the pipette
and bulb or plastic syringe can be encased in an inverted plastic bag closed around
the pipette and manipulated through the bag, to avoid any possibility of contact of
HF with the fingers. The HF- reagent bottle should be set in a large plastic or metal
beaker to catch droppings.
Another method, described by Ackerman and Lange (337), depends on titrating
silica in 3 N HCI with NaF solution until the silica is all converted to SiF.2-, after
which the presence of excess F- ion is detected potentiometrically with a titanium
electrode. An automatic titrator is preferred, since the reaction is slow. Also, AI, Fe,
N0 3 -, SOl-, and P0 42 - all interfere.

erminlica in
Titration as the Silicomolybdic ACid
, phe-

A method not requiring measurement of color and permitting use of larger samples
was described by Funk and Frydrych (338). It involves stopping the reaction between
silica and molybdic acid by adding an- excess of He\, precipitating and recovering
the silicomolybdic acid as the insoluble salt of quinoline. dissolving this in standard
alkali. and back-titrating with standard acid to the mixed end points of cresol red
(pH 7.2-8.8) and thymol blue (pH 8.0-9.6).

- --"' ;

The..Qc<;;unence. Dissolution. and Deposition of Silica



32% HCI (d = 1.156).

20 ml quinoline in 900 ml of acid solution A. made to 1000 ml.
(C) 0.1 gram cresol red and 5.3 ml 0.1 N NaOH. diluted to 100 ml; 0.1
gram thymol blue dissolved in 20 ml ethanol plus 2.1 ml 0.1 N
NaOH. diluted to 100 mi. and the two indicators mixed.
(D) 0.5 N HC\.(E) 1.0 N NaOH.
(F) 70 grams (NH4)eM07024 .4H 20 dissolved to I liter.
(G) Molybdic acid reagent:30 ml solution F, 8.4 ml solution D. and
1.6 ml H 20 mixed to give 40 ml reagent for each determination. A
larger quantity can be made up for use on anyone day. A water
bath at 20C or other suitable temperature is needed for rate

SAMPLE. Up to 10 ml volume containing up to 10 mg Si0 2 in the pH range of

about 1-7.






PROCEDURE. Place 40 ml reagent G in the thermostat at 20C and adjust the

sample to the same temperature before mixing. After the desired reaction time stop
the reaction by adding 25 ml of acid solution A. After 15 sec add 25 ml of solution B
from a burette. Cool the mixture to 10-15C and filter in a fritted glass crucible of
porosity G4; wash free of HCI. Disperse the precipitate in water: dissolve the residue
from the filter with 10 ml I N NaOH. and wash it into the precipitate suspension,
which dissolves. Back-titrate with solution Dusing 5 drops of reagent C.
One milliliter of I N NaOH consumed by the precipitate corresponds to 2.504 mg
Si0 2
1. H. C. Dake. F. L. Fleener. and B. H. Wilson. Quartz Family Minerals, McGraw-Hill.
New York. 1938; E. N. Cameron, R. B. Rowand P. l. Weis, Am. Mineral. 38, 218
( 1953).
2. W. A. Weyl and E. C. Marboe, The Constitution of Glasses. Vol. 2. Wiley.
1967, p. 1480.



3. E. W. Washburn,J. Am. Ceram, Soc., 29, 1049 (1933) [discussion]

4. F. Ephraim, Inorganic Chemistry. 4th ed . Nordemann, New York. 1943, p. 399.
Sa. J. D. Bernal and R. H. Fowler. Trans. Faraday Soc.. 29, 1049 (1933).
5b. O. Weres and S. A. Rice, J. Am. Chern, Soc .. 9.4. 8983 (1972).
6. J. W. Robinson. U.S. Pat. 2,392,767 (Du Pont). 19-16.


7. J. A. Kitchener, Faraday Disc. Chem, Soc . 52. 379 (1972).


8. G. Hetherington and K. H. Jack, Phys. Chem . Glasses, 3, 129 (1962).

9. A. J. Moulson and J. P. Roberts, Nature. 182,200 (1958).


10. S. S. Wise, J. L. Margrave, H. M. Feder, and W. N. Hubbard. J. Phys. Chern .. 66,381

( 1962).

of Silica


11. W. D. Good. J. Phys, Chem., 66,380 (1962).


11 0.1 N

12. S. A. Greenberg and E. W. Price,l. Phys. Chem .. 61,1539 (1957).

13. S. A. Greenberg,l. Phys. Chem . 61,196 (1957).
14. S. S. Wise. J. L. Margrave. H. M. Feder, and W. N. Hubbard. J. Phys, Chem . 67, 815.
( 1963).
15. C. N. Cochran and L. M. Foster,J. Phys, Chem., 66, 380 (1962).
16. G. W. Morey. R. O. Fournier, and

D, and
ation. A
A water
for rate

J. J. Rowe,l. Geophys. Res.. 69, 1995 (1964).

17a. S. Kitahara, Rev. Phys, Chem, Japan. 30, 131 (1960).

I7b. J. V. Walther and H. C. Helgeson. Am. J. Sci. 277, 1315 (1977).
18. R. K. IIer, Colloid Chemistry of Silica and Silicates. Cornell University Press. Ithaca.
N.Y., 1955.
19. S. Brunauer, D. L. Kantro, and C. H. Weise. Can.J. Chem . 3-4,1490 (1956).
20. J. D. WiJ.ley. Mar. Chem., 2, 239 (1974).

range of

21. R. J. P. Lyon. Nature. 196,266 (1962).

22. B. M. Mitsyuk, A. K. Dorosh, A. FcSkryshevskii, and Z. Z. Vysotsk ii, Ukr. Khim. Zh .
32, 833 (1966).

!just the
ime stop
.lution B
.icible of
e residue
~.5v ..ig

23. P. S. Roller and G. Ervin, Jr.,J. Am. Chem. Soc. 62,461 (l9-tO).
24. K. Goto, T. Okura. and I. Kayama, Kagaku Tokyo, 23,426 (1953).
25. G. C. Kennedy. Econ. Geol., 45, 629 (1950).
26. E. L. Brady,l. Phys. Chem., 57, 706 (1953).
27. L. Astrand, Mitt. Ver. Grosskesselbesitzer, MJ, 14 (1956).
28. H. G. Wendlandt and O. Glemser, Naturwissenschaften, 9, 352 (1963).
29. 0.1. Martynova, V. F. Fursenko, and A. S. Popov. Teploenerg. (12). 51 (1972).
30. H. G. Heitmann. Chem. Ztg., 88, 891 (1964).
31. H-G Heitmann. Glastech, Ber., 38, 41 (1965).
32. R. Siever. Am. Mineral., 42, 82 (1957).
33. E. A. Rabin, Wiss. Meeresunters. Kiel, 11,311 (1910).


34. A. P. Lisitsyn and Y. A. Bogdanov, Okeanol. Issled, Sb. Statei, 18,5 (1968).

38, 218

35. F. T. Mackenzie and R. M. Garrels, Science, ISO, 57 (1965).

ew York.

36. Fr. T. MacKenzie, R. M. Garrels. O. P. Bricker. and Fr. Bickley. Science. ISS, 1404
( 1967).
37. R. K. lIer, J. Colloid Interface

s. 43, 399 (1973).

38. J. D. Willey. Mar. Chem., 3, 241 (1975).

'39. J. D. Willey, Mar. Chem . 3, 227 (1975).
40. V. I. Bogdanova, Eksp. lssled. Mineral. 1972-3, 183 [Chem. Abstr. 82, 160059:(j.
41. S. E. Calvert. Nature. 219,919 (1967).
42. H. Harder, Geochim. Cosmochim, Acta. 29,429 (1965).
43. M. M. Jones and R. M. Pytkowicz, Bull. Soc. R. Sci. Liege. 42 (3-4).118 (1973).
4-t. R. O. Fournier and J. J. Rowe. Am.J. Sci.. 264, 685 (1966).
.. 66,381

45. D. E. White. W. W. Brannock. and K. J. Murata. Geochim . Cosmochim . Acta. 10, 27

( 1956).
46. R. O. Fournier. "Silica in Thermal Waters. Laboratory and Field Investigations." Proc.


The.Occurrence. Dissolution. and Deposition of Silica

Symp, Hydrogeochem . Biogeochem . 1970. Vol. I. Clarke Co.. Washingtonr DiC; 1973.
p. 122.
47. J. J. Rowe. R. O. Fournier. and G. W. Morey. U.S. Geol. Surv, Bul/. 1303 (1973).
48. H. Baumann. Hoppe-Seyler's Z. Physiol. Chern . 320, II (1960).
49. F. Smithson. Nature. 178, t07 (1956).
50. J. D. Miller. State College. Pa. private communication.

S I. R. B. Sosman, Ceram, Bu//. 43, 213 (1964).

52. C. Frondel, The System oj Mineralogy oj DANA. 7th ed. Vol. 3. Silica Minerals, Wiley.
New York. 1962.
53. R. B. Sosman, The Phases oj Silica. Rutgers University Press. New Brunswick. N.J .
54. O. W. FlOrke. Fortschr. Mineral. 35, 35 (1957).
55. A. F. Wells. Structural Inorganic Chemistry. 4th ed . Clarendon Press. Oxford, 1975.
56. P. P. Keat. Science. 120,328 (1954).
57. R. M. Carr. W. S. Fyfe, Am. Mineral. 43,908 (1958).
58. C. L. Hoover. U.S. Pat. 2.965,448 (Du Pont). 1960.
59. L. Coes, Jr. U.S. Pat. 2.876.072 (Norton Co.). 1959.
60. S. M. Stischov and S. V. Popova, Geokhimiya, 10,837 (1961).
61. E. C. T. Chao. E. M. Shoemaker. and B. M. Madsen. Science. 132,220 (1960).
62. E. Bohn and W. Stober. Beitr. Silikose-Forsch, Sonderb . 6. 27 (1965).
63. E. Bohn and W. Stober. Neues Jahrb. Mineral. 3,89 (1966).
64. I. S. Kainarskii, Zh. Vses. K him. O-va. 10,519 (1965); Chern, Abstr. 64, 3165h.
65. A. Weiss and A. Weiss. Naturwissenschaften, 41, 12 (1954); Z. Anorg, Chern. 276, 95
66. A. Weiss and A. Weiss. Sonderdr.IUPAC Col/oq . Altinster/West!. 2-6,41 (1954).


67. R. B. Sosman, The Properties oj Silica. ACS Monograph 37. Chemical Catalog Co ..
New York. 1927.
68. B. J. Skinner and D. E. Appleman. Am. Mineral . 48. 854-867 (1963).
69. B. Kamb, Science. 148,232 (1965).
70. R. Roy. Z. Kristallogr. 11], 185 (1959).
71a. R. Greenwood. Am. Mineral. 52, 1662 (1967).
71 b. A. Heydernann, Beitr. Miner. Petrogr, 10 (2). 242 (1964).
71c. E. M. Flanigen et al., Nature, 271. 512 (J 978).
71d. R. W. Grose and E. M. Flanigen, U.S. Pat. 4.061.724 (Union Carbide Corp.), 1977.
71e. E. M. Flanigen and R. L. Patton. U.S. Pat. 4.073.865 (U nion Carbide Corp.). 1977.
72. I. S. Kainaraskii, Zh. Vses. Khim, O-va.. 10,519 (1965).
73. G. Tschermak, Z. Phys. Chem .. 53, 349 (1905).
74. J. M. Van Bemmelen, Z. Anorg, AI/g. Chem .. 59. 225 (1908).
75. M. Theile. Silik. Z. 2. 33 (1914).
76. R. Schwartz and H. Richter. Chem . Her.. 6OB. I II (1927).
77. R. Schwartz and H. W. Hennisk c, Z. Anorg. AI/g. Chern .. 283. 346 (1956).
78. F. Liebau, Z. Kristallogr. ]20.427 (1964).


<, ,

of Silica

.. -



: . 1973,

79. H. Simon, P. Rath, Z. Anorg, AI/g. Chern . 202,191 (1931).


82. M.-T. Le Bihan, A. Kalt. and R. Wey. Bul/. Soc. Fr. Mineral. Cristallogr., 9~, IS (1971).

80. A. Pabst. Am. Mineral. 43, 971 (1958).

83. N. Y. Chen and F. A. Smith,/norg. Chem . 15,295 (1976).
83. S. E. Frazier. J. A. Bedford, J.. Hower. and M. E. Kenney./norg. Ch<?T1I . 6,1693 (1967).
84. H. P. Eugster, Science. 57, 1177 (1967).
85. L. McCulloch,J. Am. Chem, Soc. 74, 2453 (1952).

r, Wiley.

86. R. K. Iler,). Col/oid Sci. 19,648 (1964)..

87. R. K. Iler, U.S. Pat. 3,492.137 (Du Pont). 1970.

.k, N.J .

88. A. J. Gude, R. A. Sheppard. Am. Mineral. 57, 1053 (1972).

89. G. W. Brindley, Am. Mineral. 54, 1583 (1964).


90. J. L. McAtee, L. R. House, and H. P. Eugster, Am. Mineral. 53. 2061 (1968).
91a. O. P. Bricker, Am. Mineral. 54, 1026 (1969).
91b. G. Lagaly, K. Beneke. and A. Weiss, Z. Naturforsch . 28b. 234 (1973); Amer. Mineral.
60,642.650 (1975).
91c. J. Voitlander, E. K. H. Wittich. and G. Lagaly, Z. Naturforsch . 309.1330 (1975).
92. H. Kautsky and R. Irnich, Z. A norg. AI/g. Chem., 295, 193. 206 (1958).
93. H. Kautsky, W. Otto. and

!"'. Pfleger, Z. Naturforsch, 19, 102 (1964).

94. H. Kautsky and R. Reise. Z. Naturforsch . 20b, 93 (1965).

95. S. L. Pohlman and F. A. Olson. Extractive Metallurgy oj Copper. Vol. 2. Metallurgical
Society of AIME. New York. 1976. p. 943.
96. M. Van Oosterwyck-Gastuche, C. R. Acad. Sci. Ser. D. 271, 1837 (19iO).

97. C. Frondel, Ref. 52. p. 154.

98. R. Schwartz and O. Liede, Chem, Ber . 538, 1680 (1920).


og Co .

99. C. A. Jacobson, J. Phys. Chem . 40, 413 (1936).

100. E. Wiberg and W. Sirnrnler, Z. Anorg. AI/g. Chem . 283.401 (1956).
101. G. B. Alexander and R. K. lIer, U.S. Pat. 2.801,902 (Du Pont). 1957.
102. Th. Nemetschek and U. Hofmann. Z. Naturforsch, 8b (8). 410 (1953).
103. Th. Nemetschek and U. Hofmann, Z. Naturforsch . 9b, 166 (1954).
104. W. Haller. Nature. 191,662 (1961).
105. F. Ordway. Science.


800 (1964).

106. A. G. Walton. The Formation and Properties of Precipitates, Wiley-Interscience. New

York, 1967. p. 155.



107. R. Signer and H. Gross'. A,;n.

cil~;lI., 488, 56(i931).

108. W. H. Huang and D. L. Vogler, Nature. Phys. Sci.. 235. 157 (1972).
109. A. Simon and P. Rath, Z. Anorg, AI/g. Chem .. 202,191 (1931).
110. P. A. Thiessen and O. Koerner. Z. A norg, AI/g. Chern .. 189, 163. 174 (1930).
III. P. A. Thiessen and O. Koerner. Z. Anorg. AUg. Chcm .. 197,307 (1931).
112. R. Schwarz. A ngcw. Chcnt . 63, 333 ( 1951).
II Ja. J. H. de Boer. in D. H. Everett and F. S. Stone, Eds .. The Structure and Properties of
Porous Materials, Tenth Symposium or the Colston Research Society (Uni v, Bristol).
Academic Press. New York. and Buttcrworths. London. 1958. p, 243.

The Occurrence. Dissolution. and Deposition of Silica


113b. D. C. Hurd. Geochim, Cosmochim. Acta. 37, 2257 (1973).

IDe. D. C. Hurd and F. Theyer, Analytical Methods in Oceanography (Ad~. Chern. Ser . 47).
American Chemical Society. Washington. D.C., 1975. p. 211.
113d. D. C. Hurd and F. Theyer, Am. J. Sci., 277, 1168 (1977).
113e. D. S. lawson and D. C. Huro. Am. J. Sci. 1977 [in press].
113f. M. Kastner, J. B. Keene. and J. M. Gieskes, Geochim, Cosmochim, Acta. 41, 1041


114. J. A. Van tier, The Solubility a/Quartz. Drukkerji en Uitgeversrnij v/h Kemink en Zoon
n.v, Dom Plein. Utrecht, Netherlands, 1965.
115. J. A. Van tier, P. L. de Bruyn, and Th. G. Ovcrbeek,J. Phys. Chern . ~, 1675 (1960).
116. L. U. Gardner. Am. Inst, Min. Met. Eng. Tech. Publ.. 929, 7 (1938).


117. P. B. Dempster and P. D. Ritchie. Nature, 196,538 (1952).

118. J. G. Gibb and P. D. Ritchie,l. Appl. Chem., 4, 483 (1954).
119. C. Alexanian, C. R. A cad. Sci . 242, 2145 (1956).


120. J. A. Waddarns, Research London. 10,410 (1957).

121. H. Sakabe, K. Koshi, K. Matsushima, and I. Shima. Bull. Sat. Inst, Ind. Health. 1,3
122. W. Stober and M. Arnold. Beitr, Silikose-Forsch, Sonderb .. 4. 73 (1960).
123. L. G. Ganischenko, M. M. Egorov, V. F. Kiselev, K. G. Krasil'nikov, and G. S.
Khodakova, Dokl. Akad. Nauk SSSR. 131,597 (1960) [Engl. transl. p. 275]'.
124. M. S. Paterson and K. Wheatley. J. Appl. Chern. 9, 231 (1959).
125. M. Koopmans and G. D. Rieck. Br.J. Appl. Phys. Res. Notes, 16.1913 (1965).
126. L. Lidstrorn, Acta. Polytech. Scand. Chern, Met. Ser., 75 (1968).
127. J. H. Henderson, J. K. Syers. and M. l. Jackson. Isr. J. Chern . 8. 357 (1970).
128. G. W. Morey. R. O. Fournier. and J. J. Rowe. Geochim, Cosmochim, Acta. 26, 1029
129. R. Siever.v. Geol. 70,127 (1962).
130. G. C. Kennedy. Econ.

c-ot: 46, 821 (1951).

131. G. C. Kennedy. G. J. Wasserberg, H. C. Heard. and R. C. Newton. Inst . Geophys.. Univ,

Calif. Los A ng. Publ.. 150 (1960).



132. J. Wyart and G. Sabatier, C. R. A cad.

s-i.. 238. 702


133. G. W. Morey.v. Am. Ceram . Soc .. 36. 279 (1953).

134. G. M. Anderson and C. W. Burnham. Am. J. Sci.. 265, 12 (1967).
135. S. Kitahara, Rev. Phvs. Chemi Jap . 30, 109. 115. 122 (1960).
136. H. C. Hegelson. private communication [see also Ref. 17b J.
137. R. O. Fournier and J. J. Rowe. Am. Mineral.. -17. 897 (1962).
133. W. Stober. ,\-Ied.

w.. 43.

I (1966).

139. W. Stober. Beirr. Silikose-Forsch, 1.89. 1 (J 966).

140. W. Stober, Bcitr. Silik ose-Forsch, Sondcrb .. 6. 35 (1964).
141. W. Stober.

s-u-. Silikosc-Forsch, 89.113 (1966).

142. P. F. Holt and D. T. King, Nature, Phys, Sci.. 175.514 (1955).




)f Silica


143. P. F. Holt and D. T. King. J. Chern. Soc. 1955, 773.

.er. ~7).

144. W. Stober. Adv. Chern . Ser. 67, 161 (1967).

145. W. Stober. Kolloid-Z. 145, 17 (1956).
146. D. F. Weill and Y. Bottinga, Contrib. Mineral. Petro/. 25.125 (1970).


147. V. A. Avakov, Strait. Mater: Silik. Prom . 11,35 (1972) [Chem. Abstr.. 78, 47018n].

en Zoon

148. V. A. Kopeiken and A. S. Mikhailov, Dokl. Akad. Nauk SSSR. 191,917 (1970).
149. I. Bergman. J. Cartwright. and R. A. Bentley. Nature. 196,248 (1962).
150. R. S. Beckwith and R. Reeve. Geochim, Cosmochim . Acta. 33, 745 (1969).
151. H. Baumann. Natur.....issenschaften, 53, 177 (1966); Fortschrittsber, Kolloide Polym . 55,
37 (1971).
152. G. B. Alexander, J. Phys. Chern . 61, 1563 (1957).
153. T. H. Elmer and M. E. Nordberg,J. Am. Ceram, Soc .. 41,517 (1958).
154. H. Baumann. Beitr. Silikose-Forsch . 37, p. 47 (1955).
155. K. B. Krauskopf. Geochim, Cosmochim . Acta. 10, I (1956).

!th. I, 3

156. Y. U. Morachevsky and M. M. Piryutko, Bull. Acad. Sci. USSR. Div. Chern. Sci. (Eng/.
Trans/.). 1956,917.
157. C. Struckman. Ann. Chern . 94,341 (1855).

i G. S.

158. G. B. Alexander, W. M. Heston. and R. K. IIer,J. Phys. Chern .. 58,153 (1954).

159. K. Goto, T. Okura. and I. Kayarna. Kagaku Tokyo. 23, 426 (1953).
160. Y. S. Cherkinskii and I. S. Knyaz'kova, Dokl. Akad, Nauk SSSR. 198,358 (1971) [Eng!.
trans!. p. 415].
161. S. S. Jergensen, Acta. Chern. Scand. 22, 335 (1968).
161. G. Lagerstrom. Acta. Chem . Scand.. 13,722 (1959).

:6, 1029

163. I. G. Ryss and V. N. Plakhotnik, Russ.Ji Inorg, Chem . 15,




G. E. Drnitrevskii, L. G. Martynova, E. I." Antoshenko, and Z. T. Lukornskaya, Zh.

Prikl. Khim, Moscow. 44, 2451 [Engl. transl.] (1969).

164b. J. H. Oehler. Geol. Soc. Am. Bull.. 87, 1/43. (1976).

". Univ,

164c. R. O. Fournier and J. J. Rowe. Am. Mineral.. 62, 1052 (1977).

165. C. Okkerse, Thesis. Tech. H.S. Delft. Delftische Uitgevers, Maatschappij, N.V. Delft.
166. R. Spychalski. Z. Anorg, AI/g. Chern .. 239, 317 (1938).
167a. K. Goto.J. Chern, Soc. Jap, Pure Chern, Sect . 76.



167b. G. Hulett. "Solubility and Size of Particles,"in Jerome Alexander. Ed .. Colloid

Chemistry, Chemical Catalog Co .. New York. 1926. p. 637.
168. G. B. Alexander. E. C. Brege. and R. K. ller, U.S. Pat.

2.765.2~2 (Du

Pont). 1956.

169. S. Kitahara.J. Chern, Soc. Jap, Pure Chern, Sect .. 8i. 618 (1966).
170. R. J. Charles.J. Am:..Ceram . Soc.. 47.154 (1964).
171. P. Buurman and L. Van der Plus, Geol. Mijnbouw, 50. 9 (1971): Sew Sci.. Sci. J . 1971
(June 3) 557.
172. A. H. Knoll and E. S. Barghoorn. Proc, Natl, Acad, Sci.. C.S.. 71. 2329 (1974): S. A.
Tyler and E. S. Barghoorn, Alii. J. Sci.. 261.414 (1963).

The Occurrence, Dissolution. and Deposition of Silica


173. H. Baumann. Beitr. Silikose-Forsch. Sonderb . 1,31 (1955).

174. A. G. Walton. The Formation and Properties of Precipitates. Wiley-Interscience, 1967. p.
17S. R. K. Iler, "Colloidal Silica." in E. Matijevic, Ed.: Surface and Col/oid Science. Vol. 6.
Wiley, 1973.
176. L. Shartsis and S. Skinner:J. Res. Natl. Bur. Stand. %,385 (1951).
177. J. W. Axelson and S. L. Piret,/nd. Eng.

cso. 42, 665 (1950).

178. A. C. Zettlemoyer. G. J. Young. J. J. Chessick, and F. H. Healey, J. Phys. Chem., 57,

649 (1953).
179a. C. M. Jephcott and J. H. Johnston, Arch. Ind. Hyg, Occup, Med., 1,323 (1950).
179b. J. J. Denny. W. D. Robson, and D. A. Irwin. Can. Med. Assoc. J . 40, 213 (1939).
180. J. C. Lewin, Geochim, Cosmochim, Acta, 21, 182 (1961).
181a. R. K. Iler,}. Colloid.Tnterface Sci. 43, 399 (1973).
181b. H. Baumann, Fortschr. Staublungenforsch., B2 51 (1967) [Niederrhein. Druckerei
GmbH, Dinslaken].
182. M. Liel1ander and W. Stober, Z. N aturforsch., ISh, 411 (1960).
183. G. E. Dmitrevskii, L. G. Martynova, E. I. Antoshenko, and Z. T. Lukornskaya, Zh.
Prikl, Khim, Leningrad. 44, 2381 (1971).
184. W. A. Weil and E. C. Marboe, The Constitution of Glasses. Vol. 2, Wiley. New York.
1967, pp. 1119-1120.
185. H. P. Boehm and M. Schneider. Z. Anorg, Allg, Chem., 316, 128 (1962); Adv. Catal . 16,
226 (1966).
186. E. V. Ballou, M. I. Leban, and T. Wydeven, J. Appl. Chem. Biotechnol. 23 (2). 119
187. F. R. Bacon and F. C. Raggon,J. Am. Ceram, Soc. 42,199 (J959).
188. D. C. Hurd. Geochim ..Cosmochim, Acta, 37. 2257 (1973). .
189. Q. Rahman, M. U. Beg. P. N. Viswanatham, and S. H. Zaidi, Environ. Physiol.
Biochem lndia, 3 (6),286 (1973); [Chem. Abstr.. 82, 93776p.]
190. M. F. Smirnova, A. P. Dushina, and V. B. Aleskovskii, Izv, Akad. Nauk USSR. Neorg,
Mater. 2, 284 (1966).



N. Hast. Ark. Kemi, 9. 343 (1956).

S. A. Greenberg and T. N. Chang. J. Phys, Chern . 69, 182 (1965).
B. M. Mitsyuk, Zh. Neorg, Khim., 17.903 (t972) [Eng\. transl. p. 471].
M. Lawrence and H. E. Vivian. Aust, J. Appl. Sci. 12,96 (1961).
W. Stober, Kol/oid-Z. i47. 131 (1956).
A. Rosenheirn, B. Raibrnann, and G. Schendel, Z. Anorg, A//g. Chern .. 196, 160-176
A. Weiss. G. Reiff, and A. Weiss. Z. Anorg. A//g. Chem., 311. 142, 151 (1961).
F. Vlacil, and R. Bock, Z. Anal. Chern . 249, 19 (1970).
E. Richardson.t/, Appl. Chem . London, 9. 371 (1959).
J. N. Gardner and A. R. Katritzky, J. Chem, Soc.. 1957,4375.
A. Weiss and D. R. Harvey. Angew. Chem., 76, 81S (1964).
H. W. Schlipkdter and A. Brakchaus, DIsch. Med. Wochenschr.. 85.920.933 (1960).

-of Silica



203. W. C. Meyer and T. F. Yen, Arn. Chern. Soc., Div, Fuel Chern. Prepr. 19,242 (1974).



e. Vol, 6,

204. G. Foucalt and A. Collette. C. R. Soc. BioI .. 149,507 (1955).

205. S. Kitahara arid T. Asano, Bull. Fukuoka Univ. Educ. 23. III 53 (1973) [Chern. Abstr..
81, 8290Id].
206. S. Kitahara, Nippon Kagaku.Zasshi, 90, 237 (1969) [Chern. Abstr., 70, 11846Iz].
207. S. Kitahara and A. Yonekura, Fukuoka Univ, Educ. 22. III 121 (1972).

hem., 57,

208. T. Asano and S. Kitahara, J, Chern. Soc. lap. Pure Chern. Sect., 91, 109 (1970) [Chern.
Abstr.72, 136792d].
209. S. Kitahara, T. Asano, and T. Hirowatari, Nippon Kagaku Zasshi, 92, 377 (1971).


210. Z. S. Barsukova and I. D. Ivanova, Miner. Syr'e, 19 (1970) [Chern. Abstr., 74, 11657w.]


211. V. V. Strelko, Teor. Eksp, Khim., 10,359 (1973) [Eng!. transl. p. 277].
212a. D. E. Yates and T. W. Healy,J. Colloid Interface Sci.. 55, 9 (1976).


212b. J. H. Balthis, U.S. Pats 2,614,994 and 2,614,995 (Du Pont), 1952.
213. M. F. Bechtold and O. E. Snyder, U.S. Pat. 2.574.902 (Du Pont), 1951.
214. K. Goto, r. Phys, Chem., 60, 1007 (1956).

caya, Zh.

215. E. Suito, N. Hirai, and K. Taki,J. Chern. Soc. lap. Pure Chern. Sect., 72, 713 (1951).
216a. K. Goto, Bull. Chern, Soc. Jap., 31, 900 (1958).

ew York,

216b. T. Okura, K. ocrc, and M. Murai, Mem, Fac. Eng., Hokkaido Univ., 11 (I) (No. 47).
25 (1960).


217. T. L. O'Connor and S. A. Greenberg,J. Phys. Chem . 62.43 (1958).


218. S. A. Greenberg,J. Phys. Chem., 61, 960 (1957).

C' 119

219. P. H. McNally and N. W. Rosenberg, U.S. Pat. 3,113,112 (Diamond Alkali), 1963.
220. I. Bergman and M. S. Paterson,J. Appl. Chem., 11,369 (1961).
221. F. I. Anatskii and V. B. Ratinov, Zavod. Lab., 36. 283 (1970) [Chem. Abstr., 73,



222. F. I. Anatskii and V. B. Ratinov, Dokl. Akad. Nauk, SSSR (Phys. Chern.). 186. 1341
(1969) [Chern. Abstr., 71, 95380b].
223a. W. L. De Keyser and R. Wollast, Sci. Ceram., 3, 237 (1965); Chem . Abstr., 70.
223b. H. Baumann. Beitr. Silikose-Forsch., H85, 27 (1965).
224. R. H. Doremus, F. Alim-Marvasti, E. K. Pavelchek , and R. H. Doremus. U.S. N. T.l.S ..
AD Rep . 762767 (1973) [available from NTIS], Washington D.C.
225. K. D. Friedberg, Beitr. Silikose-Forsch, Sonderb 1,49 (1955). .
226. F. Dienert and W. Wandenbulcke, C. R. Acad. Sci.. 176, 1478 (.1923).

, 160-176

227. E. Wiegel, Glastech, Ber., 34, 259 (196\); 34, 141 (1964); 38. 166 (1965).
228. B. Sasaki. Bull. Chem, Res.Tnst., Non-aqueous Solutions, Tohoku Univ., 2,113 (1952).
229a. G. A. Hudson and F. R. Bacon, Alii. Ceram, Soc. Bull.. 37. 185 (1958).
229b. E. V. Ballou. M. I. Leban, and T. Wydcvcn.J. Appl. Chern, Biotech., 23 119(\973).
230. R. Bach and H. Stieber, Experientia, 22,515 (1966).
231a. H. Baumann. Beitr. Silikose-Forsch . 85, I, 15.27 (1965); Sonderb .. 649 (1963) [Chcm.
Abstr. 64, 58\2. 5813. 5666).


231b. H. Baumann. W. Klosterkottcr. and K. Robeck. Silikosebcr. Nordrhein- Wcstfalen 6, 33

(1967) [Hermann-Bosman GmbH. Dctmold, Germany).




The Occurrence. Dissolution. and Deposition of Silica

232. 1. Bergman. J. Appl. Chem . 13,356 (1963).

233. 1. Bergman. J. Appl. Chem . 12,336 (1962).
234. C. M. Strauss and J. M. Bauer, U.S. Pat. 3.650.960 (Allied Chemical Corp.). 1971.
235. J. S. Judge, J. Electrochem, Soc. 118, 1772 (1971).
236. W. A. Pliskin and H. S. Lehman, J. Electrochem, Soc., 112, 1013 (1965).
237. V. K. Leko and L. A. Komarova, Steklo Keram . 11, 15 (1973); Chem, Abstr. 80,
238. A. A. Blumberg and S. C. Stavrinou, J. Phys. Chem . 64, 1438 (1960).
239. D. E. White. W. W. Brannock. and K. J. Murata. Geochim, Cosmochim. Acta. 10,27
240. A. E. Nielsen Kinetics of Precipitation. Macmillan. New York, 1964.
241. R. O. Fournier, Proc, Symp, Hydrogeochem, Biogeochem . 1. 122 (1970).
242. H. Leitrneier, Neues Jahrb. Mineral. Geol. Palaeonto/. 27, 244 (1909).
243. H. Harder and W. Flehmig, Naturwissenschaften, 54, 40 (1967).
244a. H. Harder and W. Flehrnig, Geochim, Cosmochim, Acta, 34, 295 (1970).
244b. F. T. MacKenzie and R. Gees. Science. 173,533 (1971).
245. R. K. Iler, unpublished observation.
246. C. Wohlberg and J. R. Bucholz, "Silica in Relation to Cooling Tower Operation." Paper
143. International Corrosion Forum. April /4-18. 1975. National Association of Corrosion Engineers. Houston, Texas.
247a. E. Nardell, Water Treatment for Industrial and Other Uses. 2nd ed., Reinhold. New
York. 1961.
247b. J. D. Willey. Mar. Chem., 5267 (1977).
248. T. W. Healy. G. R. Wiese. D. E. Yates. and B. U. Kavanagh, J. Colloid Interface Sci.
~2, 647 (1973).
249. K. Goto, personal communication.







250a. G. Okamoto. T. Okura. and K. Goto, Geochim, Cosmochim. Acta. 12, 123 (1957).
250b. C. P. Huang. Earth Planet. Sci. Leu., 27, 265 (1975).



251. O. N. Shemyakina,/ssled. Vodopod. (Moscow) Sb . 1959, 190 [Chem. Abstr.. 56, 5770c].
252. V. S. Chugunov and O. N. Shemyakina, Elektr. Stn . 27, 13 (1956); Chern, Abstr. SO,
253. A. V. Nikolaev, A. P. Samusenko, and K. N. Tush. Elektr. si. 29, 26 (1958); Chern,
Abstr. 53, 13460h.
254. E. Schumann. Milt Ver. Grosskesselbesitzer, 40, 30 (1956) [Chem. Abstr., SO. 8949].



255. V. K. Tkachenko, L. A. Chernova, and G. K. Shuryshkina, eta. Sin . 22, (1951).

256. F. G. Straub, Ill. Univ, Eng. Exp, Sin. Bull. Sa.. 364, 92 pp. (1946) [Chem. Abstr. 41,
257. J. Srarnik, Milt Ver. Grosskesselbesitzer, 49, 275 (1957).
258. A. A. Kot and V. P. Lobanov. Trans. Wortd Power Can! sih, Vienna. /956 [Chem.
Abstr. 50, 13343b].
259. D. F. McDonald, Appita, 24 (2), 270 (197\).
260. A:A. Kot, Elektr. Stn . 22 (12).24 (1951).
26L S. B. Applebaum. Demineralization by Ion Exchange, Academic. New York. 1968.


lf Silica


262. M. Stassart, Bull. Cent. Belge Etude Doc. Eaux Liege, 6, 371 (1949).
263. K. A. Kun, R. Kunin. Ind. Water Eng . 3,16 (1966).


264. A. I. Ryabin, A. A. Novoselov, and E. A. Lazareva, USSR Pat. 401.385 (CI Bolj. C
02b), October 12, 1973; Pat. Appl. 1,734.816. Jan. 4. 1972 [Chern. Abstr. 71, 9-54252z].
265. H. Baumann. Ger. Pat. 1,l..6I,867. B 65,735; (Bergwerksberbar.d G.M.B.H.). 1964.

str., SO,

266. R. K. Her, U.S. Pat. 2,885,336 (Du Pont), 1959.

267. M. Mindick and P. Vassos, U.S. Pat. 3,538,015 (Nalco Chemical Co.). 1970.

10, 27


268. W. T. Barrett, Arnold. M. G. Sanchez, G. Burnie, and M. C. Yanik. U.S. Pat: 3.097,175
(W. R. Grace & Co.). 1963.
269a. M. Y. Azarian, U.S. Pat. 3,454.513, 1969.
269b. G. B. Alexander. U.S. Pat. 2.913,419 (Du Pont), 1959.
269c. W. M. Foss. U.S. Pat. 3.649.322, 1972.
270. E. S. Barghoorn and J. W. Schopf. Science. ISO. 337 (1965); 152.758 (1966).
271. E. S. Barghoorn,J. Sediment. Petrol., 22, 34-41 (1952).
272. E. T. Allen, Am.J. Sci. 28, 373 (1934); Econ. Geol. 30, I (1935).
273. J. G. Vail, Soluble Silicates (ACS Monograph Series). Yol. I. Reinhold, New York.
1952, 160.

," Paper
f CerroId, New

ace sci.,

274. K. J. Murata, Am. J. Sci., 238, 586-596 (1940).

275. C. W. Correns, Geochim. Cosmochim, Acta, 1,49-54 (1950).
276. J. H. Hellmers, Abh, Geol. Dienstes (D.D.R.). 218. 1-15 (1949).
277. E. S. Barghoom. Bot. Mus. Leafl. Harv. Univ. 14, (l), 1 (1949).
278. R. Siever and R. A. Scott, Organic Geochemistry. Pergamon Press, Oxford, 1964.
Chapter 14.
279. R. F. Leo and E. S. Barghoom, Bot. Mus. Leafl. Harv. Univ. 25 (1). 1 (1976).
280. J. H. Oehler and J. W. Schopf, Science. 174, 1229 (1971).


281. R. K. Her. J. Colloid Interface Sci. 53,476 (1975).

282. W. S. Midkiff, "Amorphous Silica in Cooling Waters." Annual Meeting. Cooling Tower
Institute. Houston. Texas. January, /976 [LA-UR-75-2313. Los Alamos, N.M.] .

. 5770c].

istr., SO,

283. W. S. Midkiff and H. P. Foyt, "Silica Scale Technology and Water Conservation."
National Association Corrosion Engineers. March /4-/8. /977. San Francisco. Calif.

,; Chern,

284a. P. B. Howard and G. D. Parfitt, 49th Natl. Colloid Symp . Clarkson College of
Technology. Potsdam. N. Y. June /4-/6. /975..
284b. R. H. Sarns, U.S. Pat. 3.565,675 (Philadelphia Quartz Co.). 1971.

t ).

istr. 41,




i. A. Bowma~ and J.

B. Wills. Anal. Chern .. 39, 1210 (1967).

286. S. Terashirna, Chishitsu Chosajo Geppo, 23 (5). 287 (1972).

287. M. E. Hofton and S. Baines. Open Report GX/EX/43/72/C. British Steel Corp. 1975.7
pp. England.
288. H. Urbain. and G. Carret, Analusis (Soc. Chilli. Fr.). 3 (2). 110 (1975).
289. I. 1\1. Kolthoff and P. J. Elving, Treatise on Analytical Chemistry: Part II. Vol. 2. Wiley.
New York, 1962.
290. H. Bennett and R. A. Reed. Chemical, Methods of Silicate Analysis. Academic. New
York. 1971.
291. L. H. Andersson. Ark. Kem. 19 (21). 257 (1963).



T..he Occurrence. Dissolution. and Deposition of Silica


292. Louis Meites, Handbook of Analytical Chemistry. 1st ed., McGraw-Hitl. New York.
293. I.G. Morozyuk, Metody Khim, Anal. Miner. Syr'ya, 255 (1971) [Chem. Abstr. 77,
294. F. Dienert and F. Wandenbulcke, C. R. Acad. Sci. 116,1461478 (1923).
295. R. W. Harman,J. Phys, Chem., 31,616 (1927).
296. T. Okura,J. Chern. Soc. Jap. Pure Chem, Sect. 72, 927 (1951); Chern, Abstr. 46, 6995.
297. E. Weitz. F. Heinz. and M. Schuchard. Chem, Ztg . 74, 256 (1950).
298. H. Jolles and F. Z. Neurath, Angew, Chem . 11,893 (1898).
299. J. D. H. Strickland.J. Am. Chem, Soc. 74, 862.868,872 (1952).
300. H. E. Garrett and A. J. Walker. Analyst, 89,642 (1964).
301. L. H. Andersson. Acta Chern. Scand., 12,495 (1958).
303. T. G. Thompson and H. G. Houlton. Ind. Eng. Chem. Anal. Ed . 5, 417 (1933).
303. G. B. Alexander, J. A m, Chem, Soc. 75, 5655 (1953).
304. G. J.S. Govett, Anal. Chim, Acta, 25, 69 (1961).
305. J. Guignard and G. Hazebrouck, J. Chem, Piiys . 63 (10). 1351 (1966).
306. H. Kautsky, H. Pfleger, R. Reise. and W. VogeII. Z. Naturforsch.. 17B. 491 (1962).
307. 1. Iwasaki and T. Tarutani, Bull. Chern, Soc. Japan. 32, 32 (1959).
308. T. Tarutani, Nippon Kagaku Zasshi, 77, 1292 (1956).
309. H. L. Kahler. W. H. Betz, and L. D. Betz.lnd. Eng. Chern. Anal. Ed .. 13,536 (1941).
310. O. A. Kenyon and H. A. Berwick. Anal. Chem . 25,145 (1953).
311. L. Isaacs, Bull. Soc. Chim, Bioi., 6, 157 (1924).
312. J. H. Foulger,J. Am. Chern, Soc. 49, 429 (1927).
313. H. Gohr and O. Scholl, Beitr, Klin. Tuberk . 8102, 29 (1949).
314. F. D. Snell
C. T, Snell. Colorimetric Methods of Analysis. 3rd ed. Vol. 2. Van Nostrand, New York, 1949, p. 692.
315. J. P. Mullen and J. P. Riley. Anal. Chlm . Acta. 12, 162 (1955).
316. R. J. Volk and R. L. Weintraub. Anal. Chem . 30, lOll (1958).
317. A. Jarabin, M. Vajda. and P. Szarvas, Acta Phys. Chint . Debrecina, 1910, (15/16).201.
318. L. H. Andersson, Ark. Kern .. 19,223 (1962).
319. R. Milton,J. App/. Chern. 1 (Suppl. 2), S-IJO (1951).
320. H. Baumann. Hoppe-Seyler's Z. Physiol. Chem .. 319. 33 (1960).
321a. J. H. Austin, R. W. Rinehart Sr., and E. Ball. Microchem, J., 17,670 (1972).
321b. E. J. King, B. D. neu, D. M. Yates, and D. Pickles. Analyst. 80.441 (1955).
321c. N. H. Furman. Ed. Scott's Standard Methods of Chemical Analysis. 6th ed. Vol. I. Van
Nostrand, New York. 1962. p. 962.





O. I. Martynova, V. F. Fursenko, and A. S. Popov. Teploenerg., 6, 86 (1972).

Z. G. Szabo. E. E. Zapp, and S. Perczel, Mikrochim . Acta. 2. 167 (197.t);
R. Wick bold. Z. Anal. cs-.: 171.81 (1959).
S. Fisher, R. Kunin. Nature, 177, 1125 (1956).
W. Sonnenschein. Z. Anal. Chern . 168. 18 (1959).
A. Gclkowska, Chern. Anal. Warsaw. 1~ (4), 803 (1969); 15 (1),59 (1970).




of Silica


;w York.

328. H. J. Thorner. Tech. Wiss. Abh, Osram Ges . 9, 241 (1967) [Chem. Abstr.. 69:797865j.
329. N. S. Tkachenko and A. V. Dobrzhanskii, Zavod. Lab . 35 (6). 666 (1969).
330. L. Shapiro,J. Res. U.S. Geol. Surv. 2 (3). 357 (1974).
331. C. Defasse, Trib. CIBEDEA U, 25 (342) 267 (1972).
332. D. Burne! and L. Malaprade, Ghim, Anal. Paris. 49 (5).253 (1967) [Chern. Abstr. 67,


333. V. L. Turner. Jr. U.S. Pat. 3.032.401 (Du Pont). 1962.

334. H. Konig. Z. Anal. Chern. 196,401 (1963).
335. G. Halfter, Angew. Chem . 61,413 (1949).
336. S. M. Thomsen. Anal. Chern . 23,973 (1951).
337. G. Ackerman and J. Lange. Talanta, 17,693.701 (1970).
338. H. Funk and R. Frydrych, Nail. A cad. Sci., Nat, Res. COline. Pub!.. 1389,284 (1966).


Ian Nos-

i), 201.

II. I. Van

- ---.


Water-Soluble Silicates


Vail's two-volume monograph on soluble silicates (1) after a quarter of a century is

still the primary source of technical and industrial information. It is pertinent to
review some of the advances made since then in the chemistry of soluble silicates,
from which silica colloids and gels are largely derived.
Silicate and polysilicate ions are still not well understood. The main distinction
between polysilicate ions and very small negatively charged colloidal silica particles
is probably one of terminology. In the past, solutions of alkali. metal silicates
obtained by dissolving sodium or potassium silicate glasses in water seemed to be
clearly distinct from sols of colloidal silica stabilized with a minor amount of alkali.
Silicate solutions were made by dissolving glasses with Si0 2 : Na 20 ratios less than
4: 1. Since these glasses with more silica and less alkali do not dissolve, silicate solutions of higher ratios were not available.
On the other hand, when it was learned how to make silica particles of different
sizes it was found that smaller particles could be stabilized as sols by adding more
alkali. Thus 5 mm particles were stabilized with a ratio of 25: 1. However. attempts
to obtain concentrated solutions with still smaller particles and lower ratios of 25: 1
to 4: I led to gelled mixtures when sodium hydroxide was used as the alkali. Then it
was found that such solutions could be made using other alkalis so that solutions
containing particles smaller than 5 nm, on down to molecular dimensions, were
Solutions in this range are referred to here as "polysilicates,' with the understanding that these polyanions may at the same time be treated as very small charged
Further study of this question was simplified when it was recognized that the
sodium ion was unusual in its influence on the gelling behavior of the system. Solutions in this intermediate ratio range made with a potassium, lithium, or
tetramethylammonium base proved to be more stable in regard to thickening and
gelling. Other stabilizing cations were guanidinium and ethylenediamine copper.
Over the past 50 years there have been major changes in theories of the nature of
the silica in the conventional alkali silicate solutions with Si0 2 : Na 20 ratios from
2: 1 to 4: 1. The physical similarity of viscous. "stringy" solutions of soluble silicates
to solutions of linear organic polymers led in some cases to the supposition that vis116


- .:...--Sodium and Potassium Silicates


cous silicate solutions must contain high molecular weight linear polymers. (This
was in spite of the obvious fact that concentrated solutions or syrups of low
molecular weight sugars are also very viscous.) Likewise, the rapid gelling of a silicate solution upon adding acid appeared to support the idea that the gelling
mechanism involved cross-linking of preexisting linear polymers, by analogy to
organic polymer systems. Thus developments in the theory of organic polymers,
improperly applied to soluble silicates, may have delayed a correct interpretation of
this inorganic system. Polymeric silicate species certainly occur in solution, but they
are not highly linear in form and the polymerization mechanism is entirely 'different
from that of any organic system.


.tury is
rent to


i to be
ss than

It has been known since the seventeenth century that sand and sodium or potassium
carbonate react at red heat to form a water-soluble glass called "water glass." As
noted by Vail (I), Johann Nepomuk von Fuchs was the first to investigate alkali silicates systematically and even before 1850 proposed their uses as adhesives, cements,
and fireproof paints. By 1855 water glass was being made commercially, both in
Europe and America.
Manufacture has generally been carried out in large open-hearth furnaces above
1300C by the following reactions:


s more
if 25: 1
Then it
;, were
iat the
. Solu.m , or

ng and
ture of
s from


3 Si0 2 + Na 2SO.

+ - C

Except under unusual circumstances, the latter reaction using "soda ash" is always
The manufacture of soluble silicates was reviewed by Williams (2). Improvements
in reaction furnaces have been proposed. Burke (3) discloses coating sand particles
with ,a mixture of sodium silicate and NaOH to give a Si0 2: Na20 ratio of 2.5: 1 to
4: 1 and heating the particles in suspension in a flame to more than 3500F, at which
temperature uniform particles of soluble glass an: formed. Other proposals are a
cyclone type furnace (4), a vertical column furnace in which reactants pass downward into a heated reactor and the product moves upward around the column as a
heat exchanger (5), anda submerged combustion furnace in which a 70CC reduction
in fuel cost and 80% reduction in size can be realized (6)~ Quantitative data on the
rate of dissolution of sand (quartz) in molten Na 2C0 3 or Si0 2-Na 20 glass have
become available. Terai, Sugac. and Hayarni (7) found that at atmospheric pressure
the reaction product at the Si0 2-mollen Na2C03 boundary is rnctasilicate, whereas
under vacuum the reaction rate is slower and mctasilicatc is not detected. Diffusion





Water-Soluble Silicates

of oxygen ions through the silicate boundary layers control the reaction rate. which
is not limited by the diffusion of sodium ions. Truhlarova and Veprek (8) found the
rate of dissolution of fused silica in molten SiOz-NazO had an activation energy of
29.8 kcal rnole " at 1050-1250C. At l200C. the rate depended on the diffusion
constant, which increased from 1.86 to 64.5 x 10- 8 ern" sec- 1 as Na.O content
increased from 22.5 to 50.8" wt, %. Kreider and Cooper (9) measured the rate at
which 1 mm quartz spheres dissolved in sodium silicate glass at 950C and concluded
it was diffusion controlled and the coefficient was 4.4 x 10 -8 em? sec:" when the Na 20
content was ~O%. Similar results were reported by Hlavac and Nadelrnynska (10).
Minor improvements in the clarity and purity of solutions continue to be made.
For example, Tauch found that if sodium silicate glass of 3.25 ratio is dissolved in
water in an autoclave to too high a concentration, the resulting solution becomes
turbid. By stopping at a concentration corresponding to 38Be and then vacuum
concentrating to 41Be, a clear concentrated product is obtained (11). To obtain a
clearer solution of sodium silicate, Rathmell disclosed adding TjO, to the glassbatch to give a concentration in the melt of 400-3000 ppm (12). Apparently the
presence of 4 moles of Ti0 2 per gram-atom of calcium prevents the later formation
of insoluble calcium silicates.
The use of sodium salts other than carbonate is still of limited interest. In 1941,
Iler and Tauch (13) studied the equilibrium
SiO:z(s) + 2 NaCI(g) + H:zO(g) = 2 HCI(g) + Na:zSi03(1)
and found K = [HCl]2[NaCI]-2 [H:zO]-l = +2.9 at \OOOe.
Calculations showed the impossibility of producing HCI at a practical concentration
while obtaining a reasonable conversion based on NaC!. The equilibrium is more
favorable under 1000 o e, but the reaction rate is very low. In 19i2, Hanf and Sole (14)
reexamined the reaction at lower temperatures down to 700C. and concluded that the
equilibrium constant was favorable to using the reaction for processing certain copper
ores containing silica.
Sodium sulfate reacts with sand even in an oxidizing atmosphere at 1400C if
reactants are spread in a very thin layer (0.5 mm). However, under reducing conditions (presumably CH 4 ) The reaction proceeds at 900C with production of
Sodium sulfate in a closed system reacts with silica at 700-900C as follows (16):


When the cost of NaOH relative to Na 2CO a is favorable. silicate solutions with
SiO z: Na 20 ratios up to 2.0 or 2.5 can be made directly by autoclaving sand with
strong N aOH solution. According to Ordiales (17) if hydrated or microcrystalline
silicas such as opal. chalcedony, diatomite, or tripolite are available. it is possible to
produce a solution with ratio as high as 3.22. The system water-silica-sodium
monoxide has been examined by Morey and Hesselgesser (18) at pressures up to
2500 bars and 400C. The solid phases are Na 2SiO a NazSizOs. and quartz.



: Silicates

e, which
ound the




Sodium and Potassium Silicates

Commercial Solutions

Properties and uses of soluble silicates have been summarized by Weldes and Lange
(19). Typical commercial (American) solutions are listed in Table 2.1. The key


variables in silicate solutions are the following:

= rate at

I. The alkali metal.

2. The ratio of Si02 to alkali metal oxide expressed by either weight or molar ratio.
In the case of sodium silicate the two ratios are nearly identical.


he Na 2 0
ie made.
.olved in
obtain a
ie glassently the

3. The concentration of solids.

Sodium silicates are produced as glasses having Si0 2 : Na 20 molar ratios of

1.6-3.9. These are sold as lump or pulverized form, partly hydrated powders, and
concentrated solutions. Potassium silicate glasses have Si0 2 : K 2 0 molar ratios of
2.83-3.92 and are sold in pulverized, flake or solution form.
For the production of colloidal and finely divided silica powders the 3.25 ratio
sodium silicate solution is almost universally used.
The maximum practical silica concentration in commercial sodium silicate solutions is limited by the viscosity which must be low enough for the solutions to be
pumped under ordinary conditions. The higher the Si0 2 : Na 20 ratio, the lower the
maximum silica concentration in commercial products, as shown in Figure 2.1.

In 1941,


Table 2.1.


is more
Sole (14)
n copper


Wt. Ratio
Si0 2:M 2O

Si0 2

Na 20












Ig condiction of


vs (16):


ons with
and with
issible to
-sodiu rn
cs up to






Composition of Typical American Commercial Alkali Metal Silicate






--::-.,.-.__ .


Water-Soluble Silicates


en 40









O""----J._ _-L.._ _




__'__ __'




Figure 2.1. Maximum concentration of

SiO, for a pourable viscosity of silicate
solution versus molar SiO,: M,O ratio. 0,
M - Na; ., M "" Li.

Soluble Crystalline Sodium and Potassium Silicates


Sodium metasilicate nonahydrate Na 2Si03 9 H 20 (m.p. 47C) is the only form made
chemically pure for use as a standard. For industrial uses, both anhydrous Na 2Si0 3
"and Na2Si03' 5H 2 0 are produced along with other more highly alkaline grades.
The following anhydrous crystalline forms, obtained from the melt, have a
SiO,: M 2 0 ratio of 1 or higher:



Na 2Si0 3
N a 2Si 20 s
K2 S i0 3
K2 Si 2 0 s
K2S i.O g


There are numerous hydrates of meta- and disilicates, as listed by Vail (1), but
Na,Si0 3. 5H 20 (m.p. 72.2C) is the principal commercial product.

Properties of Solution
The properties of sodium silicate solutions at all ratios and concentrations have been
described in detail by Vail (1) in 1952. A very convenient summary of data on key
properties such as phase diagrams, pH, density, viscosity, and solubility has been

. i




.. _.

Sodium and Potassium Silicates



assembled by We1des and Lange (19). A few further observations are here noted as
follows. although a thorough literature search was not made.
The specific gravity of a silicate solution is used to determine concentration, if the
ratio is known. Otherwise, analysis for silica or alkali is required. Shtyrenkov et al.
(20) propose titrating the normality of the alkali, from which the molar ratio of
Si0 2 : Na 20 can be determined- as 55.16(d - I)N - 2.28 where d is specific gravity
and N is alkali normality.
Measurements have been made of certain physical properties of sodium silicate
solutions over a wide range of ratios and concentrations. Accurate pH data have
been published by Bacon and Wills (21), who used specially designed electrode cells.
Densities of the solutions were also reported and an empirical equation given:
d = 1 + aA

where d

A =

+ bA2 + cA

specific gravity, 20C/20C

wt, % Na 20 in the solution

and a, b, and c are constants that vary with the molar ratio of Si0 2 : Na 20 which,
however, covered only the range from 1.3 to 0.5.
Electrical conductivity was measured by Ukihashi (22) on solutions ranging in
Si0 2 : Na 2 0 ratio from 1.0 to 3.95 and concentrations from 10- 3 N to maximum viscosity.
The viscosity of 3.41 ratio sodium silicate solutions was measured by Grant and
Masson (23) over the concentration range of 0.005-0.3253 g ml " and the intrinsic
viscosity (specific viscosity divided by concentration) was found to be independent of
the shear rate. At a concentration of 0.325 g ml- 1 the intrinsic viscosity was 16 ml
g-l; at 0.02 g ml " it was 3.2 ml g-I, and at zero concentration the extrapolated
value was 3.1 ml g-l. Thisdemonstrated that the silicate ions were of low molecular
weight and lacked chainlike character even in dilute solution.


a/" .....
lave a

Fields of Use

I), bu t



e been
In key
s b-r-n


There are three major types of uses for soluble silicates, particularly sodium
1. Cleaners and detergents that depend on controlled alkalinity are made generally
from silicates with molar SiO,: Na,O ratios less than 2.5. especially meta- and

2. Adhesive, binder. and dcflocculant applications. which depend on the presence of
polysilicatc ions. require silicates generally with ratios from 2.5 to 3.8. For
maximum adhesive strength the lower ratio types are used because they can be
obtained with a higher concentration of solids. For more water resistant bonds.
the higher ratios are needed.


.. ..... .....



Water-Soluble Silicates

3. For the production of precipitated silicas, sols, and gels, sodium silicate of ratio
about 3.3 is usually used rather than lower ratios, since less acid is required for
neutralization of alkali per unit of silica, and this ratio is available at low cost
because such large volumes are produced. However, the 3.8 ratio may be used to
some further advantage if acid consumption is an important factor in cost.
For each particular application, specific grades of carefully selected ratio and
concentration (viscosity) are recommended by the manufacturers.
Trends in uses will no doubt change as organic binders and adhesives become
more expensive relative to sodium silicate. Over the last 40 years, starch replaced
much of the silicate adhesives used in making corrugated boxboard, but the trend
may be reversed. Resins used as binders in foundry molds and cores become less
economic with increasing raw material costs and the necessity to reduce atmospheric
pollution. Replacement by silicate apparently is becoming attractive as new means
of rapid set are devised (24). Fast setting by carbon dioxide, used in a limited way
for many years, may become universal (25). The use of silicate as a binder in cores
has posed problems not encountered with resins, which burn out easily. The addition
of amorphous silica along with silicate binder permits the development of a strong
cristobalite-rich bond at high temperature, which disintegrates when cooled below
200C, according to Yates (26).
Sodium silicates used as adhesives have Si0 2:Na 20 ratios in the range I to 4: 1.
A ratio of 3.3 is most commonly used on products of paper or asbestos. At ratios of
2 or less, adhesive bonds are weaker and very water sensitive. At ratios above 4.0
(polysilicates) the adhesive strength is less, although water resistance is better. Since
at a workable viscosity, the solids content increases with decreasing ratio. silicate of
ratio 2 and of high concentration is used in dense products such as grinding wheels,
where water is hard to remove and highest bond strength is needed.
As is discussed later, the silicate adhesive solution of ratio 3.3 may be thought of
as a solution of ratio 2 consisting of N a + and HSi0 3 - ions in which the remaining
silica is suspended as exceedingly small colloidal particles 1-2 nm in diameter. At
higher ratios, there is a decreasing proportion of ionic silica.
It is observed that when silicate solutions of increasing ratios are spread as uniform thin films on glass and dried, the films crack or craze beyond a certain ratio.
This assumes uniform film-forming and drying conditions, since lilm thickness,
rapidity of drying, and temperature can all influence the degree of crazing. The
greater the degree of crazing or micro-crack ing, the less strength the silicate exhibits
as an adhesive.
It is probable that crazing occurs when the volume of dried NaHSi0 3 is insufficient to till the spaces or pores between the spherical colloidal particles as they
become packed together upon drying. If the residual solid ionic silicate does not fill
, the pores, then the gel-like' network of particles shrinks. and cracks develop in the
film. Thus the greater the tendency to micro-crack. the lower the strength of the
dried film.
It is possible to calculate the relative volumes of NaHSi0 3 and colloid present in
dried silicates of different ratios. It will be assumed that NaHSi0 3 solution dries to









-- -=-.-.



The Nature of Silicate Solutions

:d f"'"

its indicated composition, in which case its density is about 2.0 g em -3. "The silica
particles of about II A diameter contain 23% H 20 and 77% Si0 2 and have a density
of 1.72 (Chapter I). It is assumed that when the silicate is dried. the sodium ions are
coordinated with the oxygen atoms and silanol groups in the system and there is no
water bonded to the sodium ions. When a suspension of small spheres is dried, the
particles are randomly packed with a packing density about 50% by volume.
To meet the above postulated requirement that the pores must be filled to avoid
crazing, the final volume of dried NaHSi0 3 must equal that of the colloidal particles. This in turn requires, for 100 em? of dried adhesive, 85 grams of the colloid
(19.5 grams H 20 and 64.5 grams Si02~ and lOa grams of NaHSi0 3 (31 grams
Na20, 60 grams Si0 2, 9 grams H 20). This overall composition has a Si0 2: Na 20
weight ratio of 4.0: 1, or a molar ratio of 4.13. This is in agreement with the
observation that sodium or lithium silicate solutions of ratio higher than 4, generally
called "polysilicates" in commerce, do not dry to clear, continuous films as do solutions of lower ratio because micro-cracking occurs.
It should be pointed out that such a composition of 50 vol. % solids dried to an
extremely brittle state before the last of the water is removed. In considering the
behavior of fillers in plastics, for example, more than about 40SC by volume gives a
hard, relatively brittle mass. Likewise, less than 40% by volume of colloid in silicate
should give a less brittle film as the last of the water is removed. In this case, by
similar calculations it is found that this corresponds to a SiO:t:Na 20 ratio of 3.5.
Perhaps by coincidence this is near the common ratio used in adhesives.
It seems. therefore, that at least some of the behavior of soluble silicates with
ratios greater than 2 is consistent with the theory that the solutions consist of
N aHSi0 3 in .solution along with very small particles of colloidal silica with silanol

.ed to

, and
e less
I way

) 4: 1.
ios of
'e 4.0



-r. At
. The

su ffithey
ot fill
n the
.r the
cnt in


The fundamental equilibria involved have in the past been listed as follows:

For the above equations at 25C, taking the ionization constant of water as 10-U:
[H"j [HSiOJ-j

I 0 -~,~


[OH -j [Si(OH).j




~~~.. ~~


~ ..

~:~ ~

Water-Soluble Silicates


[H+] [SiOi-]


[H+] [SizO:-]
[H+] [HSizOl-]


10- u

10- 12.8





[OH -] [HSiO;]
[OH-] [HSi:zOi]

[OH] [SizOi-]

1O- 1. z0


The above constants were measured by Roller and Ervin (27) in a system involving calcium as the base. There is now some doubt whether disilicate ions are formed
in sodium silicate solution in accordance with the above equations. My observations
indicate that in the presence of colloidal particles and at very low sodium ion
concentration, only HSi03 - and Si0 3 z - are in equilibrium with Si(OH). and OHions. A I % sol of very small particles of colloidal silica was made by polymerizing
monosilicic acid made from ethyl silicate, for 24 hr at 25C, at pH 7.8. There was
some increase in pH as the specific surface area, determined by the Sears method
(28), dropped to 1040 m Z g-1 as the silica particles grew to about 2.6 nm in diameter.
To portions of this sol NaOH solution was added to adjust pH to various values
between 8.5 and 10.5. These sols were then aged at 25C while small samples were
taken repeatedly to measure the rate of color development with molybdic acid
reagent. After 2 hr aging, no further changes were noted in the reaction rates in the
different samples. The amounts of silica reacted at indicated times are shown in
Figure 2.2. After about 5 min, the slopes of all the curves are alike and correspond
to the rate of reaction of the colloidal particles with molybdic acid. As a matter of
interest, the rate of reaction was about 300 mg SiO z hr- 1 per 10 mg of total silica in
the sample, which had a total surface area of 10,400 m", The specific rate was 3 x
. 10- 5 g m -2 hr ".
By extrapolating the linear portions of curves in Figure 2.2 to zero time. the total
concentrations of soluble. that is, monomeric and ionic. silica in the original sol
samples were determined:



Monomeric and Ionic Si0 2





pH of Sol












- ---.


The Nature of Silicate Solutions










urn ion
d OHierizing
ere was
es ,- 're
s in the
own in

OL.....-_---'-_--'_ _...J...._ _...l...-_---J_ _- '





Figure 2.2. Reaction of molybdic acid with samples of a silica sol adjusted to various pH,
indicated on each curve.


atter of
.ilica in
'as 3 x
re total
nal sol

From Figure 1.l0b, line B, the equilibrium solubility of 2.6 nm particles at pH 8 is

indicated to be about 150 ppm or 2.5 mM; this is reasonably close to the value
extrapolated to pH 8 from the above data (see Figure 2.3).
Taking this value of 2.5 mM for the concentration of Si(OH)., and Roller's and
Ervin's equilibrium constants, it is possible to calculate the concentration of the
various species at equilibrium.
[SiO~- ]

[Si2 0 ; - ]
[HSi 20i]


[HSiO~-][H-]-1 X

10- 9 . 8
10- 12.15


[Si20;-J[H-] x lO. u

Ca leuluted concentrations of Si(O H). HSi0 3 ". and Si0 32 - arc plotted cumulatively
for various pH values in Figure 2.3. Although the curves arc somewhat different at


~ ...

Water-Soluble Silicates
















- -~


Figure 2.3. Concentration of soluble silica species in equilibrium with 2.6 nm diameter silica
particles at pH 8.5-10.5: A, Si(OH). in equilibrium with 2.6 nm particles: B, calculated total
concentrations of Si(OH). and HSiO~ ": C, calculated total concentrations of Si(OH)., HSiO~-,
and SiOa2 - ; D, observed total soluble silica.

pH 9, it is clear that at pH 10.5, the total concentration of soluble silica can be

accounted for by Si(OH). and HSi0 3 - , with possibly a small amount of Si03 2 - .
Thus no other ionic species can be present in appreciable amounts.
According to the above equations, the disilicate ions Si 20$2- and HSi 20$ - should
also be present in much higher concentrations than HSi0 3 -. Since it is well known
that disilicic acid reacts completely with molybdic acid in less than 5 min, the total
silica reacting in 5 min should have been much higher than observed if the equations
relating to disilicate ions were applicable to the present system.
It is therefore concluded that Roller's and Ervin's (27) observations regarding the
formation of disilicate ions were valid only in the calcium silicate system in which
their studies were made. It would not be surprising to find that the divalent calcium
ion promotes the formation of complex disilicate ions in which Si-O-Ca linkages
are involved.


A concentrated solution of sodium or potassium silicate with Si0 2 : M 20 molar

ratios of I: I to 1:2 consists mainly of Si03 2 - and HSiO- ions. However, in solutions of higher ratios it is proposed that an increasing proportion of the silica forms
extremely small three-dimensional, internally condensed silicate polymer-ions or
charged particles. These are generally equiaxed or spheroidal in shape. The silicon








The Nature of Silicate Solutions

atoms in the interior are linked only to other silicon atoms by oxygen atoms, but
each silicon atom on the exterior bears at least one OH group. These particles are in
solubility equilibrium with soluble silica monomer, SiC0 H)., which in turn is in equilibrium with monomeric silicate ions. The particles are also negatively charged by
partial ionization of the surface OH groups to SiO- anionic sites.

+ OH-

= (HOhSiO-


+ OH-

= (HO)2SiOi-

+ H 20
+ H 20
4n - nx

where x is the ratio OH: Si that decreases as the degree of polymerization, n,
increases. Upon raising the pH, ionization occurs:

r silica
d total
Si0 3 -,

: total

19 the

-ns or
.ilir '.

where z is the number of charges on a polymer containing n silicon atoms.

Although it has long been known that polymeric silicate ions are present in such
silicate solutions, it has not previously been suggested that such polymeric species be
.considered as silica particles in solubility equilibrium with Si(OH).,' in 'accordance
with the Ostwald-Freundlich equation relating diminishing particle size to increasing
solubility. It is suggested that the equation may be at least approximately followed
even by spherical particles as small as 10-20 A in diameter, where a large fraction of
the silicon atoms are on the surface.
The existence of such small dense charged silica particles would appear improbable were it not for the fact- that particles of this type are known to exist as discrete
ions containing as few as eight silicon atoms. As discussed later in connection with
the structure of certain crystalline silicates, it is definitely possible for silica to form
a compact octameric structure (HOSiOI.5)s in which the silicon atoms are at the
eight corners of a cube, each being linked through oxygen atoms to three other
silicon atoms and each having one SiOH group which may ionize to give the polyion
(Si s020)8- (Figure 2.6). This unit is the beginning of a colloidal "particle" that is
about I nm in diameter. However it is only after it is surrounded' by further
condensation of monomer that it becomes a "silica particle" having a core of Si0 2.
From the various cyclic polysilicate ions that exist in small amounts in solution as
the Si0 2: Na 20 ratio exceeds 2: I, similar polycyclic oligomers appear as the first
three-dimensional particulate "colloid" species. These then serve as nuclei as still
larger colloid-ion species are formed at higher Si0 2: Na 20 ratios.
I n my earlier book (29), I suggested possible structures of polysilicate ions on the
assumption that the silicon might have a coordination number of six. Baker, Jue, and
Wills had earlier proposed that polysilicate ions might thus be analogous to isopolyor heteropolyacid anions (30). However, current evidence indicates that silicon still
has a coordination number of four toward oxygen in silicate ions, as will be further

Water-Soluble Silicates


discussed. Accordingly, the present theory does not involve 6-coordinated silicon in
any way.
The size of the polymeric or colloidal species in sodium silicate solutions can be
estimated from various available data according to the foregoing theory, and can
also be determined directly by several means including ultrafiltration, rate of reaction, and conversion to stable species which can be isolated.
According to Bacon and Wills (21) accurate pH values were measured on a silicate of Si0 2:Na20 molar ratio 3.36 and were reported at different concentrations of
Na 20 in solution. Since the ratio of Si0 2 to Na 20 was constant, the silica concentrations are also known. Using the foregoing theory it is possible to estimate the
particle size of the polymeric silica from its solubility, that is, from the concentration of Si(OH). present.
Referring to symbols defined in Table 2.2, and assuming that the only silicate ion
present is HSi0 3 - t
A = H +M'




10-4.2 (value listed earlier)


M =




For solutions containing 0.7-0.10% Na20 the value of HI(A - H) is relatively

constant and averages 0.008. This also indicates that all the silica can be accounted
for as Si(OH). and HSi0 3 .: and that no disilicate is present. From this the
concentration of Si(OH). can be calculated:





whence M is 10- 2. 1 or 0.0079 M, or 474 ppm as Si0 2.

This value remains relatively constant over a sevenfold range of concentration. It
is assumed that 474 ppm is the equilibrium solubility of the postulated colloidal
silica component in silicate of molar ratio 3.36. From the formulas in Chapter 1
relating particle diameter to solubility (Figure 1.10b), the solubility is that of silica
particles about II A in diameter. From equations relating size to molecular weight,
this corresponds to a molecular weight of 1.15 x 10- 21 d 3 X 6 X 1023 , or 920, or
about 15 silicon atoms per particle.
It is known that such polymeric species are present only when the molar ratio
exceeds about 2 Si0 2: Na 20. Hence in a solution of ratio 3.3, about (3.3 2.0)(3.3)-1, or 39%, of the silica is polymeric. whereas 61 % is mostly monomeric. If
the degree of polymerization of the higher molecular weight fraction is about 15,


.... .=-...


ilicon in


The Nature of Silicate Solutions

Table 2.2.

pH of Sodium Silicate Solutions"

5 C.

and can
of reac>n a sili-

uions of
ncentrarate the
ncentracate ion

this the

uion. It

iapter I
of silica
920. or
ar ratio
(3.3 reric, If
lout 15.

% NazO














Data from Bacon and Wills (21).

a A = normality of total alkali in the solution; H = normality of OH - = lOU-PH:

S/ .. total silica concentration as moles SlO, per liter; R = molar .ratio of
. SiO z : NasO = 3.36; R = 2St! A and S/ .. 1.68 A; M' = normality of HSi0 3 - ;
M = molarity of Si(OH). = K I (A - H)/H. where K I = 10 -<.2.

then one can calculate the number-average and weight-average m.olecular weights,
which are 180 and 284. These values are at least of the same order of magnitude as
the number-average molecular weight 280 found by Her (63) by a cryoscopic
method, and the weight-average molecular weight 325 found by Debye and Nauman
(37) by light scattering. Aveston's (31) value of900 by equilibrium centrifugation is
probably higher because he measured the molecular weight in sodium chloride solutions. Extrapolation of his data at the lowest salt concentration (0.08 M) suggests a
molecular weight of about 600 for this Si0 2:Na20 ratio; in the absence of salt it
would be still lower.
The question might be raised why the polysilicate fraction should not consist of a
wide range of polyion or particle sizes. However, if this were the case, since the
smaller particles are more soluble than the larger, rapid redistribution of silica
would occur until all the charged particIeswere within a relatively uniform size
Ultrafiltration is another method by which at least an approximate idea of
polymer size can be obtained. In an unpublished preliminary study, lIer used
ultrafilter membranes having poresof known diameters.
The silicate solution is somewhat diluted. aged. clarified. and ultrafiltered in a
rapidly stirred pressure cell at 50 psi. By following the concentration of SiO z and
Na 20 in the filtrate one can calculate by difference how much of the silicate is
. polymerized to a size that will not go through the pores. Also. it can be shown that .
Si(OH). and HSi0 3 - pass freely through the membrane so that the initial concentration of these species in the filtrate is the same as in the initial solution. The commercial sodium silicate solutions were diluted with an equl volume of water. aged
about a week. filtered through fine filter paper to remove traces of suspended matter, and then ultrafiltered.

., ..... -..

Water-Soluble Silicates





With sodium silicate of 3.75 ratio the following observations were made:
Nominal Pore
Diameter of Membrane


% of Silica too


Large to Pass






In another series of tests using a single type of membrane with 18 A pores, solutions
of different ratios were tested:
Molar Ratio
Si0 2:Na20

% of Silica too Large

to Pass Pores of 18 A Diameter

Si0 2 : K 20






In a further study by Iler a 3.25 ratio sodium silicate solution was diluted to 6%
Si0 2 and diluted suddenly to 3% by addition to an equal volume of acid, giving a pH
of 2. Ultrafiltration showed that about half of the silica would not pass a 24 A filter.
Only 7% of the silica remained as monomer.
As will be shown later, esterification of such silicic acid with n-BuOH gave an
organosol that from the composition was calculated to consist of particles about 23
A in diameter. Apparently when the silicate was acidified the monomer added
rapidly to the colloidal nuclei already present. Since 14% of the silica had been particles of II A size they would grow by addition of monomer to a calculated size of
11 x (93/14)1/3 or 21 A diameter, which is in reasonable agreement with the size calculated from the organosol.
The main conclusion is that colloidal species are present in solutions of sodium silicates having Si0 2 : Na 20 ratios greater than 2: I and that at higher ratios a greater
proportion of the silica is present as polysilicate ions or particles generally smaller
than 30 nm diameter.
These observations are consistent with the foregoing theory that the particles are
in solubility equilibrium with Si(OH)4, which in turn is in equilibrium with HSiO J - .
Physical Studies

It is true that the results of some investigations in the past led to the conclusion that
dimeric species were present in sodium silicate solutions. Thus measurements of dif-


... .=:...---



to 6%
: a.
.ve an
out 23
parti.ize of
re calsilireater


es are

0 3- ,

n that
of "


The Nature or Silicate Solutions

fusion rates of silicate ions by J ander and J ahr (32) in 1940 and still earlier by
Jander and Heukeshoven (33) indicated Si03:- is the principal ion present above pH
13.6, but between 13.6 and 10.9 the main ion present is disilicate, Si:O/-; below pH
10.9 polymeric ions are also present, whereas below about 9.0 the silica exists mainly
as colloidal particles in equilibrium with Si(OH).
However, it is possible thatthe lower diffusion rate ascribed to Si 20s 2 - might be
due to HSi03 - hydrogen-bonded to water molecules, or to a loosely hydrogenbonded [03Si~~ Si0 3]2- complex and not to the formation of an oxygen-bonded
dimeric ion. In any case, the exact nature of the dimer ions in silicate solutions, if
present, remains to be determined.
Most investigators, especially since 1950, indicate that in aqueous 3.3: 1 ratio
sodium silicate solutions there are few dimer and trimer ions and the silica is distributed between monomer ions, HSi0 3- or SiO/-, and three-dimensional polymer
.ions or charged particles. There is some evidence that a minor amount of cyclic
tetramer ion may also be present.
Even in 1928, Harman (34) concluded from conductivity, transfer numbers,
activity coefficients, hydrolysis, osmotic activity, freezing point data, phase relations, and diffusion experiments that there are only two simple silicates, Na 2Si0 3
and NaHSi03, and that silicates in the Si0 2 : Na 20 ratio range of 2: 1 to 4: I become
increasingly "colloidal."
Viscosity studies by Main (35) led to the conclusion that above a ratio 'of2 Si0 2 : 1
Na 20 , colloidal aggregates of some type were present, along with simple silicate
ions. Hagg (36) reached similar conclusions from potentiometric measurements.
Nauman and Debye (37) carried out light scattering studies on carefully filtered
solutions of different Si0 2 : Na 2 0 ratios and measured the following molecular
- weights (expressed as anhydrous Si0 2 ) . They concluded that at ratios less than 2.0
. there are no polymers or particles, but at higher ratios increasing amounts are present.
Molar Ratio
Si0 2:Na 2 0

Average Mol. Wt.



In a potassium silicate solution of 3.8 molar ratio, Brady, Brown. and Huff (38a)
found an average molecular weight of 2000.
Using 29Si nrnr, Marsm an (38b) examined' solutions of sodium potassium and
tetraethylammonium silicate with molar ratios of base i Si or 4: I to 0.02: I. The
following species were detected: tetrafunctional branching groups. Si(OSi}4: trifunctional (HO)Si(OSi)3: middle groups, (HO)2Si(OSi)2: end groups. (HO)3Si(OSi): and
orthosilicate, (HO)3SiO -. These are in equilibrium. and equilibrium constants were
derived. With lower ratios of base to alkali. tctrafunctional groups predominate.
indicating the presence or colloidal particles.



. ,

Water-Soluble Silicates

Engelhardt et al. (38c) reviewed similar studies reported up to 1975 and made
more detailed measurements by which they identified different cyclic tetramers and
double-ring trimer (a prism-like hexarner). Many types of ions are in equilibrium
and as the ratio ofNa:Si falls below 1.0 (SiOz:NazO over 2.0), branching and crosslinking groups begin to predominate. This is in agreement with other types of evidence that as the SiOz:NazO ratio exceeds 2: 1, colloidal species begin to be present.
Infrared measurements can be used to identify simple silicate ions (absorption
band at 950 ern -1) and distinguish from polymeric species of molecular weight up to
10' (absorption at 1120 cm " ') (39).
In similar studies, Borisov and Ryzhenko (40) observed the absorption and reflection spectra (500-1700 em:") to vary with the SiO z:Na 20 ratio. In concentrated 1.5
ratio solution of (HOhSiO zz-, and (HO).{SiO),O:- the cyclic tetramer ion predominated, but disappeared when more alkali was added or when the solution was
diluted, No dimer or trimer was mentioned.
Dilution with attendant depolymerization probably explains why Greenberg (41)
concluded from a study of the literature and his own emf and conductance measurements that the data do not require postulating polymeric species even in solutions of
3 SiO z: Na-O ratio. However, careful measurements at different concentrations gave
valuable information as' to the variation of pK 1> the first ionization constant of
Si(OH), to HSi0 3 - , at different ionic strengths, the value being 9.85 at zero ionic
By chronopotentiometric and thermogravimetric methods, Franks and Mukherjee
(42) identified in a 2.5Si0 2:Na20 ratio melt the ions SiO:-, Si 20 7 ' - , and Si 30 1o aby their electroreduction potentials. However, there is no evidence that these ions
retain their identity when the glass dissolves in water.
Using a technique developed by Sillen and co-workers, Lagerstrom (43) studied
sodium silicate in 0.5 and 3.0 M solutions of sodium perchlorate to maintain
constant ionic strength, measuring the emf values with a hydrogen electrode. For
solutions corresponding to molar Si0 2: Na 20 ratios of 2: 1 to 2.8: 1 (1-0.7 negative
charges per Si atom), the data can be explained by assuming that three monomeric
species, Si(OH)" SiO(OHh -, Si0 2(OH)/-, and the tetramer Si,06(OH)..2- were
present. In 3 M NaCIO, solution some additional ions, Si 20 3(OH),2-, and
Si,Oa(OH):-, were postulated. Different SiO:NazO ratios were obtained by starting
with various proportions of colloidal silicic acid, NaOH, HClO., deaerated water,
and NaClO, solution with special precautions to exclude CO 2 , Equilibrium constants
were calculated for all species. As long as clear solutions (free from visible colloid)
were involved, equilibria were attained rapidly.
In solutions where there were less than 0.75 negative charges per silicon atom
(Si0 2:Na 20 = 2.67: I), the concentration of Si(OH), reached the approximate solubility of colloidal silica (120 ppm at 25C and 216 at 50 C). and colloidat species
were formed. Thus sodium silicates with ratios higher than 2.6 contained not only
monomeric ion species and tetra mer. but also higher polysilicates or very small
colloidal particles bearing negative charges.
Ingri (44a) carried out a similar series of pH titrations in a 0.5 M NaCI solution
with simiar results. and also gave further support to the existence of the tctrarner


- =-. ...


ers -.,d
libi ,
j crossof evij

it up to
l reflecited 1.5
on preon was

:rg (41)
easuretions of
ns gave
tant of
'0 ionic
;i 3 0 Il\8 .se
de. For

:- were
-, and

n atom
te soluspecies
ot only
{ small


The Nature of Silicate Solutions

Si 4 0 s ( O H ) s 2 - . The following constants were reported:

log 3:.: at 25 "C
Si(OH)4 + OH- _ SiO(OH); + H 20
Si(OH)4 + 20H- _ Si0 2(OHH- + 2 H 20
4 Si(OH)4 + 20H- _ Si 40S(OHn- + 6 H 20

4.:9 == 0.05
5.:3 == 0.15
15.03 == 0.20

Following Ingri, Busey and Mesmer (44b) extended potentiometric studies tc 290C
in 1 m NaCl solutions at silica concentrations of 0.005-0.05 m. The equilibrium
quotients for the ionization of Si(OH)4 to SiO(OH)3 - and SiO:(OH)22- were
determined in. 0.1-5 m NaCI up to 300C. In the more concentrated (0.02 m)
solutions at around pH 8-10 some of the silica was present as the polyanion
Si 40 s(OH)s2-, but the amount of this form decreased at higher temperature.
A more detailed picture emerges from ultracentrifuge Sl udies of sodium silicate in
0.5-2 M NaCl solutions, by Aveston (31). He studied solutions in which each silicon
atom carried from 0.5 to 1.0 charges, corresponding to SiO:: Na.O ratios of 2: 1 to
4: 1. The data suggest that although the cyclic tetramer may be an important
component, there must exist a series of higher polymers in equilibrium. the average
molecular weight increasing first gradually and then more rapidly as the Si0 2 : Na 20
ratio is increased from 2: 1 to 4: 1. Converting the "hydroxyl number" (charges per
Si atom) to Si0 2:Na20 molar ratio. the following approximate average degrees of
polymerization, hence average molecular weights. were found in 0.5 ~l NaCI solution. The molecular weight increased with the strength of the solutio:::
Molar Ratio
Si0 2:Na20
4.0 (extrapolated)

Degree of

Mol. Wt.
(as Si0 2)





In addition to monomers and tetramer, but no dimer or trimer. the polymers were
assumed to consist of units containing q silicon atoms with p charges to which
"monomer and hydroxyl ions are added stepwise: .

The authors did not assume any condensation of SiOH groups as polymerization
progresses. However. the equation might be rewritten on the assumption that the
particles contain anhydrous Si0 2 cores and a surface of Si(OH): with a certain portion of the hydroxyl groups ionized. Based on the maximum surface charge on larger

Water-Soluble Silicates


particles, about 40% of the SiOH groups on the surface are probably ionized. In this
case, there would be a decreasing fraction of the total silica bearing negative
charges, that is, the ratio of p/q would decrease with molecular weight once the
degree of polymerization reached about 10-12, above which all silicon atoms cannot
be on the surface of roughly .spherical particles.
For p = q = I:
log (31.1


[Si(OH).] [OH-]


This value com pares with 4.29 from Ingri's data (44a).
Electrical conductivity measurements on potassium silicate of SiO::: K:rO molar
ratios from 0.39 to 4.78 were made by Ivanov and Gur'eva (45). Anion mobilities
were determined and interpreted on the basis of the usual assumption that
(HO)2SiO:r:r- is in equilibrium with [(HO)2SiOSi(OH):r]:r- and (HO)JSiO-. Dissocia-


tion constants of the silicate ion in dilute sodium silicate solution from 0 to 250C
were calculated from conductivity measurements by Ryzhenko (46).
The nature of silicate ions in molten Na 2SO.IOH 20 (m.p. 32.38 ClC) was extensively studied by Thilo and associates, who used the lowering of the melting point to
establish the number-average molecular weight of the silicate ions (47-49). The
degree of hydrolysis depended on the Si0 2 : Na 20 ratio and not on concentration. At
extreme dilutions the species Si(OH)., (HOhSi0 22 - , and (HOhSiO- are present. At
higher concentrations equilibria set in between these monomers and higher molecular
weight anions or particles.
At equilibrium, the average molecular weight of the silica varied inversely with
N a 20 content (SO).
Higher degrees. of polymerization were visualized as occurring stepwise. This
could, of course, also be interpreted as a change in particle size:

A general equation was developed and an equilibrium constant calculated for

Si0 2: Na 20 ratios from 2.0 to 3.34. However, the relation between these observations in molten Glauber's salt to the molecular weight distribution in water is not
Another approach by Thilo and associates (51) was to precipitate silicates from
solution as the silver salts. Starting with Si0 2 : Na 20 ratios of 1 and 2. the degree of
polymerization of silica in the silver salt varied inversely with the silica concentration and content of neutral salt. It was suggested that the silver salt. containing one
silver atom per silicon atom, was a linear polymer, although no viscosity or
molecular weight measurements were made.

Effects of Diluting Silicate Solutions

Since the concentration of hydroxyl ions and pH are reduced when concentrated
alkali metal silicate solutions are greatly diluted, it is not surprising that hydrolysis









The Nature of Silicate Solutions

. In this
.ncc ~
, cannot

of silicate ions occurs and a large part of the silica is converted to higher polymeric
species or colloid along with a residual silicate of lower SiO z: Na.O ratio.
The electron microscope was used by McGarry and Hazel (52) to characterize the
colloidal particles in 3.45 SiO z : I KzO potassium silicate solutions which had been
diluted out from 10 to 0.03% SrO, in various ways. The samples were further diluted
to 0.015% SiOzjust before being examined. When diluted directly to 0.03%,89% of
the particles were from I to 5 nm in diameter. When diluted to 0.3SC silica and aged
for an hour and then diluted to 0.03%, the number-average diameter was 8 nm.
The results indicate that when the solution is aged at a concentration of 0.3% the
1-2 nm colloidal particles that were in the original 10% silicate act as nuclei upon
which the monomeric silica, liberated by hydrolysis of silicate ions. is deposited. On
the other hand, when the 10% solution is diluted directly to 0.03% SiO z or 5 x 10- 3
molar, the pH is 10.26 and all the silica passes into solution as Si(OH)4 and HSi0 3 ions, and even the original 1-2 nm polysilicate ions or particles slowly dissolve. As
seen in Chapter I, at pH .10.26 the total solubility of silica including Si(OH)4 and
HSi0 3 - is around 0.05%. The authors used pincyanol chloride to detect the presence
of colloid species. The dye micelles formed on the colloid by absorption shifted the
color from blue to red (600 to 500 nm).
The further polymerization of the colloidal species owing to the lower pH when a
3.3 SiO z: NajO silicate solution is diluted was also studied by Lehrman and
ShuJdener (53). They measured the amount of molybdate-reactive silicain solution
over a period of days and found that at a concentration of about 0.14% SiO z, the
amount of reactive silica, presumably monomeric species, decreased over a period of
2 days, although some further drop was noted over 1 week. The same occurred at
0.07% silica but not at 0.03%, where complete depolymerization occurred and all the
silica was molybdate-reactive within a few hours.
When 3.25 ratio sodium silicate solution is diluted to around 2st SjO, and heated
for 1 hr, there is disproportionation whereby colloidal particles grow in size, while
the SiO z : Na-O ratio of the so-called crystalloidal or dialyzable soluble silicate
approaches 2.6: 1 or 2.0: I, depending on dilution. It is my experience that during
this time the solution is supersaturated with respect to amorphous silica so that any
clean surface of glass or metal oxide becomes coated with a film of silica to a thickness of a few nanometers.

) molar
on that

extenpoint to
9). The
tion. At
sent. At

ely with
.e, This

ted for


r is not
es from

cgree of
icentra.ing one
isity or

Effect 01 Alkali AIetal Salts and Other Coagulants

The addition of a salt of a polyvalent metal to a solution of a soluble silicate results

in formation of a precipitate of a metal silicate. In this precipitate. silicate ions are
presumably linked together by the metal ions to form an insoluble network. On the
other hand. the addition of a salt of a monovalent metal. for example. sodium
chloride. to a solution of sodium silicate yields a precipitate that will redissolve if the
system is diluted with water.
The alkali metal silicate is salted out especially easily when the ratio of
Si0 2 : Na 20 is greater than 2: I. The fact that the precipitate can be dissolved in
water suggests that it may consist of polysilicatc ions linked together by coordina-






Water-Soluble Silicates

tion through sodium ions; such linkages would persist only in the presence of a large
excess of sodium ions.
Further details of the effects of a sodium salt such as the chloride or sulfate on a
solution of sodium silicate have been summarized by Vail (I). Thus sodium chloride
causes separation of a precipitate of sodium silicate which is more siliceous than the
original solution. In smaller amounts, sodium chloride and also the sulfate and carbonate increase the viscosity of sodium silicate solutions. It appears that the added
sodium salt coagulates the colloidal or high molecular weight polysilicate ions but
has less effect on the metasilicate ions.
The effect of salts on the equilibria in a silicate solution was studied by Pevzner et
al. (54) up to a NaCI concentration of 5 N, with silicon concentrations of 0.001-0.3
g-ion 1-1 in the pH range 10.2-14.0. Equilibrium was appraoched from lower and
higher pH. The dissociation constants (pK) for water at 25C in 0.5, 3, and 5 N salt
solutions are 13.7. 14.1, and 15.2, respectively. The tetrarneric silicate ion was
detected, as had also been reported by others (44). It is noteworthy that no dimer or
trimer was mentioned. The data are summarized in Tables 2.3 and 2.4.
The effect of salt on the degree of polymerization of the polysilicate ions in
aqueous potassium silicate was examined by Brady, Brown, and Huff (55), using the
light-scattering method. The molar ratio of Si0 2 : Na 20 was 3.8: I and in freshly
diluted solutions the weight-average molecular weight of the Si0 2 was about 2000.
When neutral salts were added, the apparent molecular weight increased in proportion to the fourth power of the salt concentration and in inverse proportion to the
square root of silica concentration. In alkaline solution, the molecular weight varied
with the temperature. The salt-induced polymerization at high pH was reversible
upon dilution, but the acid-induced polymerization at neutral or lower pH was irreversible.
It seems likely that. at high salt concentrations the. cations act as cross-links
between polyions through coordination with polysilicate oxygen atoms which replace
coordinated water on the cations. This would be reversed when water is added. On
the other hand, acid-induced polymerization leads to formation of siloxane
(Si-O-Si) bridges between the polysilicate ions which are not broken upon dilution.
Table 2.3. Composition and Stability of Silicare Ions at 25 C in and 5.0 1"1 Aqueous
Sodium Chloride Solution
Logarithm or the Equilibrium
Constants in Solution with
Ionic Strengths

Si(OH). + OH - ;:= Si(OHh - + H 20

Si(OH). + 2 OH ;:= Si0 2(OHh 2 - + 2 H 20
4Si(OH). + 2 OH-;= Si.O.(OH)S2- + 6 H 20
4Si(OH). + 4 OH-;= Si.O.(OH):- + 8 H 20

Data by Pevzner et al. (54).








4.3 == 0.1
5.3 == 0.1
15.1 == 0.3

4.6 == 0.1
6.0 == 0.1
11.0 ::!: 1.0

5.6 ::!: 0.2

7.\ 0.1
12.0::!: 1.0




ile Silicates

of a large

=..... ..-

The Nature of Silicate Solutions


Table 2.4. Changes in Free Energy, Enthalpy. and Entropy for the
Formation of Silicate ions in 5.0 M Aqueous Sodium Chloride Solutions at

I a
n chloride
s than the
e and carthe added
e ions but


Pevzner et

Source. Data by Pevzner et al. (54).

The types of equilibria are indicated by the numbers given in Table 2.3.

lower and
d 5 N salt
~ ion was
) dimer or
:e ions in
using the
in freshly
lout 2000.
.n proporion to the
ght varied
re- 'ible

:h replace
idded. On


. with

5.6 0.2
7.1 0.2
12.0 1.0




(kcal g-ion "')

7.6 :: 0.3
9.7:: 0.3'
16.4 :: 1.5

-J.H (kcal g-ion ")

j.S (eu)

1.0 :: 0.5
1.5 :: 0.5

22 :: 3
28 3
45 :: 8

According to Prikhid'k o and Molchanov (56), the addition of alcohol or acetone

to a solution of sodium metasilicate causes the formation of two liquid layers, the
silicate accumulating in the lower layer. It seems probable that this is a simple
dehydration of the silicate solution, the organic solvent removing water from the silicate solution and being immiscible because of the "salting-out" effect of the ions in
the lower layer. Ordway (57) studied silicate of other ratios and concluded that
organic solvents always precipitate a liquid or a solid phase containing a higher ratio
of SiO z : Na.O than the original aqueous solution of silicate. This indicates that,
whereas the silicate separates from a fluid layer rich in organic solvent, the sodium
hydroxide is soluble in this organic layer and thus leaves a silica-rich aqueous phase.
Ammonium hydroxide is similar to alcohol in its effect. Sodium hydroxide is soluble in strong ammonium hydroxide or liquid ammonia. but sodium silicate is not;
a process of purification of sodium hydroxide from small amounts of sodium silicate
has been patented by Muskat and Ayres (58a) in which, at certain high concentrations of ammonia and water, the ammonia-rich layer dissolves the sodium
hydroxide, while the sodium silicate in the system accumulates in an immiscible
lower layer.
Effects of this type are apparently involved in a process for depositing silica coatings on glass from a solution of sodium silicate containing a critical amount of a
"clustering" or "coacervating" agent such as NH.OH, NazSO., or CHJOH at
ordinary temperature (58b).

Conversion to Silicic Acids

The nature of the silica in silicate ions in any alkaline solution cannot be determined
by a chemical measurement that involves any change in the concentration of silica or
alkali. electrolyte content, or temperature because these all shift the equilibrium
between monomeric and various polymeric ion species. However, if a sample is
simultaneously and instantaneously diluted and acidified to pH 2 at less than 30C.
the resulting silicic acid is sufficiently stable to permit characterization. The problem
is to ensure that acidification is so sudden that the various silica species do not have
time to polymerize or dcpolyrncrize as the pH is dropped from the usual region of

-- _



Water-Soluble Silicates

11-13 to 1.5-2.0. At the same time the silica must be diluted to a concentration of
1% or less to minimize polymerization even at this pH where the rate is at a
minimum. Also the temperature should be preferably at a-5C if the sample is to be
stored for more than a few minutes.
As will be discussed in detail in Chapter 3, this technique has been shown by
Alexander and others to al1o~ the formation of Si(OH). from N azSiOJ 9H:O. Saltfree solutions can be obtained by ion-exchange techniques. Since conditions have
been found for converting monomeric silicate ions to monosilicic acid, which is
extremely prone to polymerize, it is evident that higher polysilicates can likewise be
converted to the more stable polysilicic acids with even less difficulty.
By this means solutions of sodium metasilicate, Na:Si0 3 9H 20, especially when
previously mixed with excess alkali, are shown by molecular weight measurements
to contain monomeric ions, since only Si(OH). is obtained by sudden acidification
(59-61). Alexander (62) and Her (63) described making a 0.1 ,H solution of monosilicic acid, Si(OH)., by stirring finely crystalline N a 2SiO J 9H:O into a suspension
of a sulfonic acid type of ion-exchange resin in the hydrogen ion form at OC while
maintaining the pH at about 3.0 with 10- 3 N HC!. The degree of polymerization
was 1.1 by the freezing point method. Finely powdered Na.SiO. \vas also successfully used.
Similarly, Iler (64) found that a 3.3: 1.0 ratio solution of sodium silicate containing 12% Si0 2 could be added to the acid resin slurry to obtain silicic acid with a
number-average molecular weight of 200, as SiO:. A more dilute solution containing
2% silica gave product with molecular weight of 180 (63). This value compares with
325 found by Nauman and Debye (37) in a more dilute solution of sodium silicate
where the lower pH permits some further polymerization of the polysilicate ions.
Reaction with Molybdic ACid
The history and use of the silicomolybdate method for analyzing for soluble silica
has been discussed in Chapter 1. It is sufficient here to point out that molybdic acid
reacts only with monomeric silica. to form the yellow silicomolybdic acid. It is
fortunate that the reaction with molybdic acid occurs at pH 1-2, where silicic acid
polymerizes least rapidly. Thus polymeric silica must first depolyrnerize before it
can react; hence the higher the degree of polymerization. the longer the time
required for depolymerization and color development. This is reviewed in detail in
Chapter 3.
This method of characterization of soluble silicates is complicated by the fact that
a given type of silicate can give different results if not allowed to age under fixed
conditions of concentration and temperature until equilibrium has been established.
For example, samples of 3.25 ratio sodium silicate were diluted by Iler from a silica
concentration of 400 gl " to 200 and 100 gl-', and each was divided into two parts:
(a) aged at 30C for 6 hr, and (b) held at 100C for 6 hr, then cooled and aged at 30C
for 6 hr.
A 50 .ul sample of each relatively concentrated silicate solution was injected suddenly from the tip of a microsyringe reaching to within a few millimeters of a high

..:- ...


The Nature of Silicate Solutions

:ion of
s at Ii
s tL

speed stirrer in 50 ml of the molybdic acid solution. Ionic species react in' less than
10 min, whereas most of the colloidal or polysilicate species require more than 20
Owing to depolymerization upon dilution, the proportion of colloid or polysilicate
decreased from about 70% at a silica concentration of 400 g 1-1 to 45% at 100 g 1-1.
However, heating a solution at a given concentration and then cooling had little or
no permanent effect at these high concentrations, since equilibrium was reestablished
at 30C in a few hours. This conclusion was also reached by O'Connor (65), who
carried out similar experiments on concentrated silicates of different ratios.' He
found that solutions with Si0 2 : Na 20 ratios of 0.5-2.4 contain only a minor proportion of polysilicate ions, but there is a marked increase in polysilicate (or colloid)
content as the ratio increases from 2.4 to 2.8 and higher ratios. He proposes a rate
equation whereby the molybdic acid reaction rate constant can be correlated with
molecular weight as determined by light scattering. He showed that the distribution
of the ionic species depended mainly on concentration and that equilibrium was
reached quickly even when the solution was diluted from 400 g I-I to 20 g I-I of
Si0 2 and did not change further in 2 weeks at 25C.
Reaction rates of specific polysilicic acids with molybdic acid have also been
measured by Wieker and Hoebbel (66), whose data indicated the time required for
95% of the silica to react was about 1.5 min for monosilicic, 2.5 min for disilicic
(almost indistinguishable in a mixture), and 9.5 'min for the cubic octasilicic acid.
The authors later expressed the rates as reaction velocity constants, k n .l I , for higher
molecular weight species. The rate constant was approximately inversely proportional to the molecular weight of the polysilicic acid (67). From their data the
average degree of polymerization nh the number of silicon atoms per polysilicic acid
molecule, can be related to the reaction rate constant:

wn by
'. Salts have
rich is
vise be
, when
: while
recesslntainwith a
:s with

: silica
ic acid
. It is
c acid
fore it
: time
tail in
ct that
. fixed
l silica
d suda1


Coudurier, (68b) further investigated these reactions and proved that mono- and
disilicic acids both react directly with molybdic acid, the latter only half as rapidly
as the former. Higher polymers all must depolymerize to the monomer before reaction can occur. In this wayan empirical relation between molecular weight of the
polymer and the rate of formation of silicomolybdic acid was worked out (see
Chapter 3).
Thilo, Wieker, and Stade (68a) standardized the molybdate reaction with silicate
species and calculated the characteristic reaction rate constants:


silicornolybdate in [. min and r 2 in 12

where '1 is the fraction of silica converted
min. Plotting -Iog( I - r) versus t. usually gives a linear section with a constant
slope between times 11 and 12 from which the constant can be calculated.
Certain anhydrous crystalline silicates known to contain SiO. 2 - ions, such as
NaH 2SiO.8H 20 (Na 2Si0 3 9H 20). Ca 2SiO . Li.SiO., and i\lgK 2SiO. all gave the


Water-Soluble Silicates

same constant, which was therefore characteristic. of monomeric silica. Other

crystals known to contain the disilicate ion such as Ca~Na2Si207 and N a.Si 20 7
served as standards for dimeric silica.
For monomer, k = 1.7 min:" and for dirner, k = 0.90 min-I. The constant for
cyclic tetramer was about half that-of the dimer, using (KH)4Si 4012 as the standard.
A solution of SiOz:NazO ratio of 2: 1 was shown to depolymerize with increasina
dilution. Thus in a solution 1.0 M in SiOz, the molybdate reaction curve was very
similar to that of cyclic tetrarner, However, upon dilution to 0.1 M it acted like
dimer and at 0.01 M had hydrolyzed to monomer.

Conversion to Esters ofSilicic Acids

Some indication of at least the maximum size of the silicate species in a solution of
sodium silicate of 3.3 ratio SiO z : NajO has been obtained from the nature of the corresponding silicic acid ester (69). Silicic acid of low molecular weight is so unstable
in aqueous solutions that any attempt to isolate it by evaporation of water, even at
ordinary temperature, results in rapid polymerization to a gel. The direct esterification of silicic acid therefore remained impractical until the discovery of Kirk (70) of
a method for transferring silicic acid of low molecular weight from aqueous solution
to solution in an alcohol. This transfer is accomplished by extraction of the acid with
a suitable polar organic solvent, simultaneously saturating the aqueous phase with
sodium chloride in order to salt the silicic acid into the organic phase. An alcohol
such as n.butyl alcohol is then added, and esterification is effected by azeotropic distillation of water from the alcohol solution (29).
The polysilicic ester, which is a tacky resin after the butanol is removed under
vacuum, typically contained around 25% carbon and 62% SiO z. The composition
can be expressed in terms of the molar ratios (C 4H9 )20 : Si0 2 =' 0.25 and H 20: SiO z
= 0.22.
Viscosity studies showed that polymer consisted of units of spherical rather than
elongated shape.. From the composition it can be calculated that the ratio of
hydroxyl plus butoxy groups to silicon is 0.95, which would be the OH:Si ratio in
the corresponding silicic acid. Referring to the formula in Chapter 1 relating particle
size to composition of surface-hydroxylated silica particles. OH: Si = 24.6cF-/l1.5d3,
the particle diameter is estimated to be 2.3 nm.
The monomeric and low molecular weight species in the original silicate had
polymerized onto the higher molecular weight polysilicate or colloid fraction. which
has been shown to consist of particles about 1.0-1.5 nm in diameter, so that the final
particle diameter in the ester is somewhat larger.

Conversion to Trimethylsilyl Derivatives ojSilicic Acids




The various ion species which might be found in sodium silicate solutions can be
found in pure form only in certain crystalline silicates. Lentz (71) first announced in
1963 that the silica in selected minerals in which the ionic structure is known from
X-ray diffraction data can be converted to the corresponding trirnethylsilyl deriva-

-- ---: Silicates


-.r a, C': f)1

stant for

vas very
cted like

lution of
the corunstable
, even at
sterifica( (70) of
reid with
ase with
opic dis:d under

her than
ratio' of
ratio in
:; 11.5ef3.
:ate had
n. which
the final

; can be
unced in
. . . n from

The Nature of Silicate Solutions


tives. Thus olivine. (MgFehSiO., contains monomeric silica which can be isolated as
[(CH3hSiO].Si. Hemimorphite, Zn.Si z0 1(OHh 2H zO. contains dimeric silica.
isolated as [(CH 3)SiOhSiOSi[SiO(CH 3hh. Similarly, natrolite, Naz(AlzSi301o)
2H zO, furnishes the linear trimer, and laumontite, CaAl zSi.0 1z4H zO, gives cyclic
rerramer, which is isolated as [(CH3hSiO]aSi.O.
Lentz then obtained the most direct and convincing evidence of the true nature of
the silica in sodium silicate solutions (72a) by developing a technique for converting
the silicate species to the corresponding organosilyl derivatives and isolating them by
distillation and gas chromatography. Conversions and yields were less than perfect
and the study was made on I M silicate solutions where some hydrolysis and
polymerization is known to occur, as compared with the concentrated solutions of
commerce. Also, it must be pointed out that since the technique involves the
transient formation of silicic acid, some polymerization may occur under the locally
acidic conditions to form dimer and trimer that may not be present in the original
alkali silicate.
The results showed that there is always less dimer than monomer and less trimer
than dimer, but a significant quantity of tetramer in the diluted solution with
SiO z: Na-O ratio of 2: 1. Colloidal species comprised an increasing fraction of the
total silica with increasing SiOz:NazO ratio. At 2: I ratio (Na.Si ratio of I) it
amounted to 45% of the silica and at 3.3: 1 ratio it was about 70%.
The colloidal species were converted to silyl derivatives having (CH3}3Si0o.s: SiO z
molar ratios of 1.32-1.05. Lentz concluded that these were spherical particles. on the
basis of which the diameter was deduced to be 10-15 A.
The fact that the solutions were somewhat dilute. containing only 6% by weight of
SiO z (l M), no doubt permitted some hydrolysis and polymerization to occur. As
shown by Thilo, Wieker, and Stade (68a) still further dilution results in depolyrnerization to dimer and then .monorner. In 1 M metasilicate solution (SiOz:NazO =
I: 1), 22% of the silica was colloidal, 10% was cyclic tetramer, 7% linear trimer, 12%
disilicate, and only 26% monomer. (Total recovery was 76%.) However. in
concentrated metasilicate solution and in N a zSi03 9H zO it has been shown that
under carefully regulated conditions. most of the silica is monomeric by reaction
with molybdic acid.
Glasser and Sharma (72b) compared different procedures for trirnethylsilylation
of the silica in sodium rnetasilicate solution (NazO:SiO z = 1: 1). The proportion of
polysilicate ions in solution decreased with dilution from 1 to 0.001 M, where only
monomer was found. Even in I M solution only a third of the silicate was
Conditions for reacting (CH 3hSiCI with minerals in various reaction solvents were
investigated by Gdtz and Masson (73). who found that a small amount of water
must be present. hexarnethyldisiloxane and isopropyl alcohol are beneficial, and conditions must be carefully controlled to minimize degradation or polymerization. to
obtain an optimum yield or the trirncthylsilyl derivative or the original silicate ion.
Subsequently it was found that sodium silicate was especially difficult to react
without polymerizing the silica species to some extent (74).
The structure of ions in numerous crystalline silicates has been further elucidated

Water-Soluble Silicates


by coverting the individual silicate ions directly to trirnethylsilyl derivatives. which

were identified after separation by thin-layer chromatography. This is possible if the
derivatives are soluble, that is, if the silicate ion exists as isolated units and not as
infinite chains or three-dimensional siloxane networks. Thus Wieker and Hoebbel
(66) showed that when Li:H 2SiO. solution is concentrated over H 2SO., the ions are
largely present as cyclic tetrasilicate. Also, crystalline N a 20 2eaO 3Si0 2 actually
. contains cyclohexasilicate ions and should be formulated Na.Ca.(Si.O u ) .
Similarly, these authors (75) coverted the following silicates to the corresponding
trimethylsilyl (T) derivatives which are used as standards for the identification of
individual silicic acids:



N a2 H 2 SiO. H 20
Na 6(Si 20 7 )
Na.Cd 2(Si30 lo)
Si.0 3C1 lo
(CaSi0 3)3
[(CH3)3N+],Sia02o' 69H 20
[Cu(enh].Si a020 30H 20

T aS i207
T aS i30 10
T 1oSi.O I3
T 6S i309
T aS i.0 12
T aS ia0 20
T aSi,02o

Linear trimer
Linear tetramer
Cyclic trimer
Cyclic trimer
Tricyclic heptamer
Cubic octamer
Cubic octarner

In this octamer, referred to as a "double-four-ring" structure there are eight silicon

atoms at the corners of a cube.

Silicates with Coordination Numbers Four and Six


It has been previously theorized that in aqueous silicate solution silicon is coordinated with six oxygen atoms. Such views of Weyl (76) and Treadwell and Wieland
(77) were discussed by Iler (78). However, more recent spectroscopic studies of silicate solutions seem to indicate that the silica is 4-coordinated with oxygen and that
the Si(OH)62 - ion, for example, does not exist, at least in solution. Fortnum and
Edwards (79) concluded that Si(OH)i- was not present in solution because Raman
lines analogous to those of SiF.2-, PF6-, and Te(OH)6 were not found. The four low
intensity lines found at 448, 607, 777, and 935 cm -I suggested that the ion is
(HO)2Si0 22- since it is similar to (HO)2P02 ". which gives lines at 40 I. 511, 885, and
1050 em -I.
Similar conclusions were reached by Early et al. (80). who also noted the similarity of Raman spectra of (HOhSi0 22- and (H0>zS02' Using HzO and DzO, Freund
(81) examined silicate solutions with Raman-laser spectra and identified (HO)3SiO-,
(HO)2SiOl-, and (HO)3Si023-, the latter only at very high alkali concentrations.
Equilibrium was obtained rapidly and completely. There was no evidence for the
presence of dimer species.








Solutions of Polysilicates

: if the
ns are

On the other hand, it is likely that most reactions of silica and silicates-in aqueous
solutions do involve a temporary expansion of the coordination number from 4 to 5
or 6. Also in certain crystalline silicates, the silicon atom is coordinated with six
oxygen atoms, for example, in thaumasite (82). This was also confirmed by Edge
and Taylor (83), who worked out the structure of this calcium silicate sulfate carbonate crystal and demonstrated the presence of Si(OH)e 2- ions.
It is in the hydrated metasilicate crystal that one might expect the silicon atom
most likely to be surrounded by six hydroxyl ions, but Jamieson and Glasser have
shown that this is not the case (84). In Na 2Si03 9H 20 they find by X-ray structure
analysis that silicon is surrounded by only four oxygen atoms. The sodium ions are
surrounded by six oxygen atoms and the more appropriate constitutional formu la is
Na2[(HO)2Si02]8H20. By similar procedures, Jost and Hilmer (85) found a similar
structure and concluded that isolated SiO. tetrahedra are linked through hydrogen
bonds into layers, separated by layers of sodium and water. The same formula was
arrived at by quite a different method. The nature of Na 2Si03.xH 20, where x is 9, 6.
or 5, was investigated by Thilo and Miedreich (86) by hydrolyzing the salt in wet
alcohol. From analysis of residue it was concluded that the salt should be formulated
as Na 2(HOhSi02yH 20, where y is 8, 5, or 4. A hydrated so-called "disilicate," is
formulated as Na(HO)3SiO, which has the same structure as Li(HOhP0 2, and is
actually monomeric.
From infrared spectra, Borisov and Ryzhenko (87) conclude that in 'solutions of
sodium metasilicate (Si0 2: Na 20 ratio of 1: 1) there are mainly ions of (HOhSiO/with some cyclic tetramer (HOMSi.O.)O.-, which increases in concentration with
increasing ratio, and also that silicon is coordinated with four oxygen atoms.
Thus, except in very rare instances, the silicate ions in solution or in crystals
appear to contain silicon with a coordination number of four.

ion of


of silid that
.ir low
ion is
5. and
simi'reu nd
)r .'

For many years it was supposed that alkali silicate solutions with Si0 2: M 20 ratios
higher than 4: 1 could not be made. The corresponding glasses are not homogeneously soluble in water. Attempts to dissolve quartz or amorphous silica powders
in a minimum of NaOH solution to achieve the desired higher ratios were not successful. Liquid silicates with ratios higher than 4: I were sought for use in adhesives
because the lower alkali content was anticipated to furnish bonds that are more
water resistant. Also. in refractory compositions. loweralk ali was needed to raise
the softening point and improve refractory character.
As discussed at the beginning of this chapter. silicates with Si0 2: Na 20 ratios
of 4: I to about 25: I are usually referred to as "polysilicates." They do not have the
high content of sodium and silicate ions as do the conventional soluble silicates.
which have ratios less than 4: I. However. they contain much more alkali than
necessary to stabilize sols of smallest particle size manufactured and sold as colloids.
The borderline between a polysilicate solution and an "alkali-stabilized sol of
colloidal silica" can logically be drawn at the particle size of the silica below which
the increased solubility of the silica as Si(OH). at ordinary temperature causes the

Water-Soluble Silicates


particles to grow spontaneously unless more alkali is added. However," when particles less than 5 nm in size are stabilized against growth by adjusting the ratio to less
than 25: 1, an increasing and substantial fraction of the silica is converted to silicate
ions. Thus a "polysilicate" solution may be considered to be an equilibrium mixture
of ionic and negatively charged particles of colloidal silica of such extremely small
size that they can be classed also as polyanions.
Because lithium silicates are so different in behavior from sodium and potassium
.silicates, lithium polysilicate is discussed separately.

Sodium PoJysiJicate
The advantage of a binder having a Si0 2 : Na 20 ratio of 10: I, for example, is
demonstrated by the fact that such mixtures of an . alkali metal silicate and a
colloidal silica sol are proposed for making waterproof fireproof coating mixtures
even though the mixtures have a short shelf-life (88).
During the development of concentrated silica sols stabilized with NaOH around
1950, it was realized that the smaller the particle size, the more alkali and the lower
the Si0 2 : Na 20 ratio required for stabilization. However, in the ratio range of about
4: 1 to 25: 1 the concentrated com positions were generally found to be unstable, and
eventually gelled. When a 3.25 ratio sodium silicate solution was added to a
concentrated silica sol to reduce the Si0 2 : N a 20 ratio from 100: I to 5: 1, for
example, a gel immediately formed. However, lIer (89) found that by aging or
warming the mixture, a stable liquid was again obtained. Thus solutions with ratios
from 4.2: 1 to 6: 1 containing 10-30%Si02 could be prepared from sols originally
containing typically 5-25 nm particles. However, longer heating destabilizes the
mixture and an insoluble crystalline polysilicate or gel appears. the polysilicate solution is a clear, transparent fluid of low viscosity in which around half of the silica is
present as soluble or ionic silicate, as shown by its reactivity with the molybdic acid
reagent. Thus much of the silica that was introduced as colloid remains as such, but
the particle size is smaller owing to some depolymerization at equilibrium.
The unique nature of the product is shown by the fact that when it is mixed with a
silica sol containing an equal weight of 7 nm colloidal silica along with a sol containing an equal weight of colloidal silica as 100 nrn particles, there is formed a mixture
which, when cast as a film on glass, dries at 150C to clear, hard, adherent, waterresistant coating. The spaces between the 100 nm particles are filled by 7 nm panicles and the spaces between these by the still smaller colloidal com ponent of the
polysilicate and, in turn, the remaining spaces of molecular size by the soluble ionic
silicate species. Once dried, the sodium apparently is adsorbed on the extensive silica
surface so that no soluble silica remains and the film is impervious.
Instead of starting with a sol of preformed particles, Oken (90) claimed that by
starting with a low molecular weight solution of silicic acid, temporarily stable
liquids could be obtained with Si02:Na20 ratios in the range 4: 1 to 6: I, as in Iler's
com position, and also from 8: 1 to 40: I.
More recently, compositions with ratios higher than 4: I have been prepared by
heating an amorphous silica powder with 3.22 ratio sodium silicate to above 100~C ,






Lithium Silicates

partito less
. small

to cause intumescence, and then the partly dried mass is further heated "to 315C,
after which the mass can be dissolved to give a polysilicate solution (91). The higher
temperature apparently partly dissolves the agglomerated silica powder, bringing
about its disaggregation. Probably the amorphous powder employed must consist of
ultimate silica particles of the same size range as in the sols em ployed in earlier
processes. According to a German patent, if waste silicic acid powder from the AIF 3
process is used, it is so lightly agglomerated that it can be dispersed by alkali alone
to give a solution of 4.3: I ratio of Si02 : Na 20 containing 13.7% Si0 2 (92).



Potassium Polysilicate
ple, is

and a

~ lower
~ about
le, and
j to a
: I, for
~ing or


gin.... y
zes the
.e solusilica is
lie acid
ch, but
with a
water1 partiof the
Ie ionic
e silica

It has not generally been recognized that in the polysilicate system, the potassium
ion behaves very differently from sodium. Although the coagulating effect of these
ions is generally similar in most colloid systems and even in collloidal silica at pH
7-10, the potassium loses its coagulating effect at higher pH but sodium does not.
(This is discussed in further detail in Chapter 4.)
When concentrated 3.25 ratio sodium silicate is mixed with an alkali-stabilized sol
containing 30% silica, coagulation and gelling occur at once, but they do not occur
if potassium silicate is added. This is apparently the basis for the disclosure of Waltersdorf (93). Thus stable mixtures of colloidal silica and potassium silicate can be
prepared with a silica concentration of 15-30 wt. %, with Si0 2 : K 2 0 molar ratios of
11: 1 to 24: 1. Although it is not so stated, such compositions are relatively stable
toward prolonged heating in comparison with similar sodium based mixtures, which
tend to crystallize.
Appreciation of this effect has led to the development of a number of practical
applications. Thus amorphous silicic acid powder is heated with 2.4: 1 ratio
potassium silicate solution to give a viscous solution that forms. a binder for a pigmented 'coating on asbestos-cement board (94). The final molar ratio of Si0 2 : K 2 0 is
5.7: 1. A composition with similar use is described by Yano (95).
Another unusual application of the compatibility of potassium silicate with
sodium-free colloidal silica is described by Shoup (96). By mixing concentrated
potassium silicate solution with a concentrated silica sol and adding an ester,
aldehyde, or amide which liberates acid as a homogeneous gelling agent, very strong,
dense, uniform silica gels can be made in cast shapes. When washed with acid to
. remove the potassium salt. these gels are so strong and coherent that they dry in the
shape in which they are cast. Because the silica is so concentrated, cracking is
minimized as the gel shrinks as it is dried. When fired at 900C it gives a strong
body with 70% by volume porosity. It can then be tired at 1400C and further
shrunk to an object of clear fused silica without losing its relative shape (97).

that by
n Iler's



The lithium ion is very different from sodium or potassium in its behavior in silica
systems. The lithium ion rescrnbles these clements to some extent at ordinary



Water-Soluble Silicates

temperature but above about 60 C it assumes some of the characteristics of calcium.

Thus, although a solution of lithium silicate can be made by dissolving amorphous
silica in LiOH solution at 20-40C, the silicate comes out of solution when heated.
The effect is reversible; it is remarkable to see the solution solidify to a white mass
when hot, then reliquefy to a clear homogeneous state after being cooled to 25C for
a few hours.
Ordway (57) in 1907 studied the properties of lithium silicate with molar ratios of
Si02 : Li 20 of 1: 1 to 2.3: 1 and noted the peculiar reversible insolubility when the
solution was heated and cooled. Even as late as 1952, Vail (1) stated that the possible concentrations and ranges of composition of lithium silicate were much more
restricted than in the case of sodium. In 1969, Weldes and Lange summarized the
properties of available aqueous lithium silicate systems (19).
It is apparently only since 1950 that it has been appreciated that stable solutions
of lithium silicates and polysilicates could be prepared with any desired Si0 2 : Li 20
ratio and that these have useful properties.
The difference from sodium is especially apparent in concentrated solutions with
molar ratios of Si0 2 : Li 2 0 from 4: 1 to 15: I or higher. These are clear and quite stable at 25C whereas, as previously discussed. most of the analogous sodium silicate
compositions either gel or crystallize. Furthermore, the properties of concentrated
sols of colloidal silica are appreciably different from those stabilized by sodium base,
especially in regard to much greater compatibility with water-miscible organic
Lithium silicate solution cannot be made by dissolving the glass. so is prepared by
dissolving very finely divided amorphous silica in LiOH solution at 20-50C.
However, the reaction apparently proceeds more rapidly at 75-250C in an
autoclave, whereby one obtains a solution of LiSi0 3 plus a precipitate of higher ratio
silicates which dissolves when the solution is aged at 20-30C (98). A solution of 3.8
molar ratio can be prepared by adding 200 mesh silicon powder to a stirred 12%
solution of LiOH H 20, from which hydrogen is evolved, while the temperature is
held at 38-45C. The slightly viscous filtered soltuion contained 19% Si0 2 (99).
Another approach is to prepare a cation-exchange resin in lithium form and pass a
solution of sodium silicate through it to exchange sodium for lithium (100).

... ;c-





Lithium Polysilicates
. In 1954 it was discovered by lIer (101) that concentrated stable solutions containing
Si0 2 : Li 2 0 molar ratios from 4: 1 to 25: I could be obtained by adding LiOH to a
solution of polysilicic acid. to a suspension of silica gel. or to a silica sol free from
alkali metal or other cations. Since the mixtures thickened or immediately set to a
gel, this approach appeared useless until it was found that after a few hours or a day
or so at ordinary temperature. the mass spontaneously liquefied. When such mixtures were heated to 80- J OOC to accelerate liquefaction. they remained gelled. The
liquid compositions contained both ionic and colloidal silica.


.". =-.--icates

Lithium Silicates


Commercial solutions of lithium polysilicate are reported with the- following

properties (102):


C for


os of
n the
d the

Si0 2( %) .
Molar ratio Si0 2 : Li20
Viscosity (cps)
Specific gravity
Shelf-life at 75 F

e sta-

ed by


of 3.8
ure IS

lass a

4.8: 1

Type 85
8.5: 1


Since compositions of all Si0 2 : Li20 ratios from 4: 1 to 25: 1 are soluble and stable, this system offered lIer an opportunity to examine the relation between the silica
to alkali ratio and the nature of the colloid species pressent. Lithium polysilicate
solutions were made by mixing solutions of polysilicic acid, obtained by ion
exchange from 3.25 ratio sodium silicate, and lithium hydroxide and aging the mixtures at 25C for a week until clear liquids were obtained at equilibrium. The solutions contained 10% of Si0 2 and the Si0 2 : Li20 ratios ranged from 3: I to 10: 1.
These solutions were compared by measuring the rate of formation of the yellow
silicomolybdic acid when microsamples were suddenly mixed with molybdic acid
solution at low pH. For each sample. the percent of the total silica, P, which had
reacted at time t was noted at 2.5, 5. 10, 15, and 20 min at 25C. A solution of
sodium silicate of 3.3: 1.0 M ratio was also included and found to react in about the
same way as the lithium silicate
solution of corresponding... ratio.
It was assumed that the solutions all consisted, for the most part, of monomeric
silicate ions and higher molecular weight polysilicate ions in the form of very small
charged colloidal particles. Since the monomeric ions react with molybdate in about
2.5 min. the percent of monomeric silica. Pm. was noted in each case.
Then P - P trI' or p, was the percent silica in the polymeric state that had reacted
at time. i. This was found to follow the equation

1 to a
. from
t to a
a day
i nuxj, Tht:

Type 48

100 - P


(see Figure 2.4)

The values of k; the relative reaction rate. wen: plotted against (R - 2)-1 since
colloidal species are present only at a ratio greater than 2: I. This gave the relation
shown in Figure 2.5. Since. as has been elsewhere noted. the reaction rate is an
inverse function of particle size of the colloid. it is concluded that the size of the
colloid species must increase regularly and spontaneously in proportion to the degree
to which the ratio exceeds about 2:.1. This is at least true in the ratio range 3: I 10
10: I. A: higher ratios. the particle size is less closely related to ratio. since it is well

... =-- -j


Water-Soluble Silicates











_ _-'-_ _.L-_----L_ _----L_ _-.I-.......J





Figure 2.4. Reaction of colloid species of silica in lithium polysilicate solutions of different
ratios of Si0 3 : Li 30 (indicated on curves) with molybdic acid reagent.





known that at ratios higher than 25: 1. sols of different particle sizes can be stabilized at the same ratio without much. if any. difference in particle size.
Thus the polysilicate, solution spontaneously equilibrates to a mixture of
monomeric species and colloidal species. the size of which increases with increasing
ratio overZ: I.
Numerous procedures for manufacture have been suggested. Cuneo (103) prepared a lithium polysilicate from a 2.6% silica sol obtained from sodium silicate by
ion exchange. by adding LiOH to attain ratios of 2.5: I to 8: I and then concentrating the solution by vacuum evaporation to a silica content of 21 %. More
"crystalloidal." as opposed to colloidal. silica was said to be present because of the
nature of the starting sol which probably contained silica particles 1-2 nm in
A lithium polysilicate solution of 3.5 M ratio can also be made by heating finely
divided amorphous silica with LiOH solution at a molar ratio of 1.5: I. separating
the precipitate that forms at elevated temperature. and letting it cool. whereupon it
liquefies to a polysilicate solution containing 23.4% Si0 2 Apparently the silicate
that separates from hot solution is of higher ratio than that which remains (104).
According to Patton and Cox (105). lithium silicates of molar ratios from 2: I to
6.3: I can be made in a similar manner at 75-150C by leaving the precipitate in
suspension and cooling to effect total solution. The polysilicate can also be made by
dissolving silicon powder in LiOH solution at 50-80C with Cu(OHh present as a
catalyst (106).


Lithium Silicates



a:: 0.010







( R-2f '



Figure 2.5. Relative rates of reaction of colloid species in lithium polysilicates with molybdic
acid reagent versus (R - 2)-1, where R is the molar ratio of Si0 2 : Li 2 0 .

. st.

The presence of some lithium silicate in sodium or potassium silicate solutions is

apparently of some advantage (107, 108).

ure of
Uses for Lithium Silicates and Polysilicates
3) pre-

.ate by
of the
nm in

ipon it

2: I to
tate in
ade by
1t .1" a

A major use of polysilicates is in anti-corrosion coatings containing finely divided

zinc, where the silica acts as an inorganic binder (109, 110). The addition of an
organosiliconate improves the water resistance of the coating (III). A similar composition is said to provide a binder for brake linings (112). For a simliar reason, a
minor amount of a styrene-acrylic copolymer emulsion may be added. which also
gives im proved adhesion to steel (113). Another additive to improve saltwater
resistance is a small percentage of barium hydroxide (114). Barium silicate is more
soluble than the calcium or strontium salt and liberates enough silicate ion into solution to inhibit the corrosion of aluminum by alkali. according to Dupre and Boornan
(115). The adhesion and weldability of zinc-rich primers bonded with lithium polysilicate were improved by substituting some di-iron phosphide or cadmium for some
of the zinc (J (6).
Some of the difficulties involved in using zinc-rich paints bonded with polysilicates
were reviewed by Wright (117). who pointed out that in storage. slow reaction of
zinc with the vehicle on aging reduced the quality of the final coating.
Several types of soluble silicates were evaluated. but a lithium polysilicate with





Water-Soluble Silicates

Si0 2 : Li 2 0 ratio about 5: 1 was satisfactory. using zinc dust of 2-3 micron particle
size. The advantage of lithium was mainly that the mixtures are more stable (118).
The binder properties are useful in other pigmented coatings such as clay-based
paints (119). The zinc-lithium silicate coatings can be applied by anodic electrodeposition on steel (120). In another-type of use, lithium polysilicate provides an intermediate bond between the fluorocarbon polymer coating and metal on antisticking
cook ware (121).
The precipitation of higher-ratio insoluble silicates from hot solution may be used
to deposit siliceous coatings on hydrophilic surfaces. Sams (122) discloses that silicarich coatings for optical or protective effects are deposited on surfaces such as glass
from 2: 1 ratio lithium silicate solution when the surface is immersed in the solution
as it is heated to 95C. Even after 10 sec, there is formed a colored film showing
light interference, which must be of the order of 0.3 microns in thickness. Solutions
with ratios from 1.3: 1 to 10.7: 1 may be used.




Since silica dissolves above about pH 10.7-11.0, silicates can be prepared with
organic bases having dissociation constants greater than about 10- 3 , although some
silica will dissolve in aqueous solutions of weaker bases with constants as low 'as 10-
or 10- 5 (123-125). Merrill and Spencer (124) reported the preparation of a number
of water-soluble quaternary ammonium silicates by grinding silica gel with a solution of the free base. However, the compounds all appear to have a ratio of 2: 1
when expressed as SiO:z:(NR.hO by analogy with the alkali metal system.
The unusual behavior of the (CH 3).N+ ion in silica systems is probably due to its
inability to form coordinate bonds with the oxygen of SiOH groups as sodium does
at high concentration by exchanging one or more water molecules of the (HzO)sNa"
ion for silanol oxygens. Hence the quaternary ammonium ion cannot act as a coordinative crosslinking agent to promote precipitation and gelling. Furthermore. there
is some evidence that the quaternary ammonium ions are adsorbed on the surface of
the silica, thus interfering with the dissolution and deposition of silica that is
involved in particle growth.
In the case of the interaction of tetramethylammonium hydroxide with colloidal
silica, there is evidence that. unlike NaOH. this quaternary ammonium base does
not allow the colloidal and ionic species of silica to equilibrate readily. This is
probably because the (CH 3).N" ions are strongly adsorbed on the surface of the
colloidal particles and retard dissolution of Si(OH). This was described by Beard
(39). who examined mixtures of colloidal silicas with (CHJ).N"OH-(TMA) or
NaOH with equivalent molar SiO z: Na.O ratios from 0.5 to 3.25 by infrared
spectroscopy. This method does not require diluting the sample or otherwise 'disturbing the equilibrium. Silica concentrations were 13.3, 15. and 20 wt. %. Mixtures
were aged for up to 6 days to reach equilibrium.
Polymeric or colloidal species could be detected in sodium silicate solutions by an
absorption band at about 1120 em - I . At a ratio of 2.0 a peak at about 1025 cm - I is
assumed to be due to dirner, but is more likely HSi0 3 -. At a ratio of 0.67, where







: used

; 10-'
solu'f '

to its
)e Na +
Ice of
iat is
his is
Jf the
. . ) or
rar ed
by an
1 -,


Organic Base Silicates


silica must be present largely as Si0 32 - ions. the characteristic peak is at about 980
ern -1, whereas at a ratio of 1.0 it is not as sharp, suggesting the possible presence of
some HSi0 3-. With this interpretation, the results indicate that 3.25 or 3.0 ratio
concentrated sodium silicate solutions contain only colloid and HSi03-.
However, (CH 3),N+ silicate solutions differ from the sodium silicate solutions in
that even at 1.0 ratio, colloid still persists along with HSi0 3-. Likewise, in 3.3 ratio
solutions, there is much more colloid in the (CH 3),N+ based solutions than with
sodium base. The authors conclude that TMA cation "shifts the equilibrium to a
mixture of low and high species." It is likely that the colloid phase is stabilized by an
adsorbed monolayer of (CH 3),N+.
It will be noted that in these experiments the solutions were made from TMA
hydroxide and colloidal silica. It is possible that if TMA had been added to a dilute
solution of polysilicic acid of low molecular weight at a ratio of 3.3 and then vacuum
evaporated to 10% silica, solubility equilibrium would become established and the
distribution of silica species would be like that in sodium silicate solutions.
The tetramethylammonium salt, CH 3hN>s[Si s0 2oJ. was prepared and studied by
Hoebbel and Wieker (126a), who converted the polysilicate ion to the trimethylsilyl
derivative [CH3)3SiJs[Sia02o] characterized by analysis and mass spectroscopy. This
polysilicate ion is in effect a cube of Si0 2 indicating the possibility of small compact
particles existing in sodium silicate solutions of Si0 2: N a 20 ratio between 2: I and
4: 1. The copper-ammonium salt is discussed further in this chapter.
In crystalline organic base silicates, the size of the organic cation 'apparently controls the structure of the polysilicate anion. Hoebbel and Wieker (I26b) prepared a
crystalline tetra{n-butyl) ammonium silicate hydrate having the empirical composition 0.8 I N(n-C,Hgk Si0 2 26.5 H 20. The silica is present as two parallel fivemembered siloxane rings held together by five oxygen bridges. Each silicon bears
one ionizable OH group. However there are only four anionic charges on each fivemembered siloxane ring; possibly there is a proton, H 30+, at the center of each ring.
The corresponding N(t-CsH u ) , + silicate has a similar structure.
A series of organic nitrogen base silicates, formulated as (NRR'R"R"'hO
xSi0 2,yH 20, were patented in 1954 (127), where R, R', R", are H, alkyl, aryl,
aralkyl, hydroxylamine. and amino-substituted alkyl radicals, and R'" is any of the
aforementioned groups (other than H. alkyl, or R") or CH 2 groups forming part of
a heterocyclic ring structure which includes the nitrogen. The value x varied from 0.5
to 4.0. and y from 0 to 15. Typical examples' were (HOC 2H,),N+OH -,
CeHsN(CH3)3+0H-, and guanidines. A crystalline compound with analysis corresponding to the following formula was isolated:

Quaternary alkanolammonium bases are desirable for making solutions of silicate

free of alkali metals. since there is generally no amine odor and the compounds are
very water soluble. Wcldes patented a series of these materials along with a process
for making them using ethylene oxide as follows. in water solution (128):



Water-Soluble Silicates

When vacuum evaporated, the resulting syrup contained 39.3% SiO: and the composition corresponded to [(HOC2H.).N]20. 5.9Si0 2 8.3H 20 (129).
Another way of making this type of organic silicate is to add (HOC 2H.).N "OHto a solution of 3.75: 1.0 ratio sodium silicate and then remove the sodium by ion
exchange with (HOC 2H.).N+ ions on a cation-exchange resin. The mixture at first
gels, but then forms a viscous solution. An alternative is to prepare a dilute solution'
of ethanolamine and silica in this way and then form the quaternary ammonium salt
by reaction with ethylene oxide (130). Diethanol ammonium and morpholinium silicate and related salts have also been prepared (131).
Some of the available quaternary ammonium silicates have been listed by Weldes
and Lange (19), although the nature of the NR. + ion was not disclosed. Malar ratios
of Si0 2: (NR.)20 range from 1.2: 1 to 30: 1 and silica contents of the solutions from
10 to 45%. Hydrated amorphous powders are also listed with ratios of 15.6: I and
19.7: 1. These dissolve rapidly in water and are remarkably soluble in mixtures of
water and water-miscible organic solvents such as acetone and the lower alcohols.
The following suggests that even with a ratio as low as 7.5: 1 the composition may
be largely a colloidal solution rather than a soluble silicate. Wolter (132) first discovered that when enough tetramethylammonium hydroxide is present in a sodiumfree silica sol to have present four or five (CH 3).N+ ions per square nanometer of
silica surface, the sol can be dried to a gummy solid which can be redissolved in
water to the original sol state. In the sol of 1.6 nm particles the required
Si02:[(CH3).N]20 ratio was}.5: 1.
Guanidinium silicate has been prepared from the free base, readily prepared from
the carbonate and Ca(OHh and colloidal silica (133). The solutions had molar ratios
of Si0 2:G 20 of up to 3.1: 1.0 where G is guanidinium ion. The solution dries to a
clear water-soluble film: When alcohol is added, the salt is precipitated as an amorphous powder useful as a binder for refractory oxides (134, 135). The powder is stable and may be used for making up solutions which may be used as a binder in sand
cores, set by carbon dioxide, which are readily removed after casting metals. The
solution may be used to deposit silica when heated, owing to the slow decomposition
of the strong base. It is viscous and may be spun and heated to produce silica fiber.
According to Nakajima, Yamaguchi, and Tahara a mixture of guanidinium silicate
and sodium silicate is an especially useful binder for sand molds in metal casting,
since it can be set by carbon dioxide and gives excellent mold strength (136).
A variety of uses have been suggested for the quaternary ammonium silicates. As
shown by Weldes and Lange (19) these compositions supply silica as a binder which
loses most of its organic content at 300-400c. They may be used also to fill pores,
form coatings, and are of potential use for treating inorganic fibers, reinforcing
polymers, and as vehicles for coatings and paints for use at high tern perature. A
slowly dissolving tetraalkyl silicate acts as a setting agent in silicate-bonded cement
(137). Properties and suggested applications have been described for a series of quaternary ammonium silicates of undisclosed ion type, available under the trademark
QURAM:~ (138). Molar ratios of Si0 2 : (NR.hO range from 2.25 to 11.3, and silica
contents of liquids from 34 to 45%; a soluble powder contains 71 % SiO:.

., =-.-


Organic Base Silicates

ie com-

Mixed Organic Base-Alkali Metal Base Silicates


by ion
at first
urn salt
urn siliWeldes
.r ratios
ns from
5: I and
tures of
on may
irst dissodium.ieter of
olved in
ed from
'ies tu a
n amor.r is stain sand
als, The
ca fiber.
ates. As
:r which
II pores,


Double silicates of quaternary ammonium with Li, Na, or K silicates have been prepared. Hydroxy alkylammonium salt mixtures were described by WeIdes (139-141).
Sodium polysilicate solutions which are unstable in certain ratio ranges can be stabilized by adding quaternary ammonium base, even in small amounts. Of course large
amounts will reduce the Si02 : M 2 0 ratio (M being cation) to form soluble silicates,
but the effect of small amounts has not been explained. For example, solutions
containing 21.8% Si0 2 with Si0 2:Na20 ratios of 4: 1 to 12: I are stable ii the quaternary ammonium base N-hexamethylhexamethylenediammonium hydroxide is
present to give a mole ratio of Si02 to quaternary ammonium of only 850: I (142).
In a similar patent, Freyhold and Wehle (143) disclose that the Si0 2:(R.NhO
ratio may range from 35: I to 1000: 1. Possibly strong adsorption of relatively few
large organic cations on the surfaces of the colloidal particles serves to keep them
apart, preventing the aggregation that occurs when only sodium ions are present.
Organic bases, some of which are not as strong as the quaternary ammonium
type, when present along with an alkali metal hydroxide, have a different type of stabilizing effect on silica, namely, maintaining the colloid in the form of extremely
small particles. Yates (144) maintains silica sols with a specific surface area of
950-1800 m2 s" corresponding to particles only 2-3 nm in diameter with a
Si0 2:Na 20 ratio of 7: I to 20: 1 by having an organic base present with a basic
dissociation constant of at least 10- 2 at a concentration of 0.1-1.5 molal. This
includes bases such as morpholine or cyclohexylamine as well as tertiary and quaternary ammonium types. The small particle size is not maintained in the presence
of either the alkali metal hydroxide.or the organic base alone. The compositions are
prepared by adding the amine to a solution of sodium silicate and then removing
sodium to the desired degree with an ion-exchange resin. In the absence of the
amine, the specific surface area of the resulting colloidal silica spontaneously
decreases and the particles grow, but when amine is present it is adsorbed in proportion to the specific surface area of the silica and stabilizes the silica in the form of
extremely small particles or polyions.
That the effect is not that of pH or overall base strength alone is shown by Yates
in another case (145) where polyhydroxy organic compounds such as glycerol or
sugars, in conjunction with alkali metal base, have a similar effect of preventing
particle growth and maintaining stable particles of high specific surface area.
Adsorption of the polyhydroxy or alcohol-ether type compounds on the silica surface is probably involved, since calculations show that at least : one alcoholic
hydroxyl group must be present per square nanometer of silica surface.

iture, A

of qua-

Other Organic Base Silicates


.id silica

In the field of pharmaceuticals. an unusual type of organic base silicate is

methionine mcthylsulfonium silicate, which has the advantage of being nonhygro-


Water-Soluble Silicates

scopic and odorless (146). A wide variety of precipitated organic base silicates has
also been prepared for use as intermediates in making pharmaceuticals. In view of
the method of preparation and weak strength of some of the bases, it is possible that
the compounds are adsorbed on precipitated silicas of very high surface area (147).
Aluminosilicate is formed as a. stable sol when tetramethylammonium silicate and
tetramethylammonium aluminate solutions are mixed, whereas the sodium salts
immediately form precipitates unless the solutions are very dilute. When sodium
chloride is added to the sol, a gel or precipitate is formed which then develops zeolite
crystals, the size of which increases from 25 to 300 nm with NaCl: Al 20 3 ratios' from
0.1 to 1.0 (148).
Long-chain alkyl quaternary ammonium silicates are presumably formed when
cationic surfactants of this class are added to sodium silicate solutions and whipped
to form stable foams. Thus the silicate containing a gelling agent is foamed and set
to a lightweight gel (149). Similarly, a mixture of potassium silicate and colloidal
silica, which forms a very strong gel when acidified, is used as a binder for inorganic
fiber and expanded perlite. The mass is foamed with a long-chain quaternary ammonium salt before it sets to form a low-density thermal insulating material with a
density of 10-20 lb ft- 3 (150). Also, 3.25 ratio sodium silicate alone can be mixed
with perlite, foamed with hexadecyltrimethylammonium ion, and set with carbon
dioxide gas.
Reference also should be made here to the unique, highly siliceous crystalline
zeolite-like insoluble quaternary ammonium silicates from which the organic ions
can be removed, leaving' anhydrous, hydrophobic, microcrystalline Si0 2 of an
entirely new type discovered and described by Flanigen and Grose (see Chapter I,
refs. 71c, 71d).



Werner complexes of metals such as cobalt with amines yield strong free bases
which can form silicates. Schwartz and Bausch (151) reacted Na 2Si0 3 with
[Co(NH3)~Cl}Cl to obtain Co(NH 3hCISi.O g and with [CO(NH 3)e}Cl to obtain
[Co(NH 3).Si3 0,]Si0 3, although there was no proof that silica was present both in
the anion and cation. The constitution of the silicate ions of ethylenediamine copper
silicate has been worked out by Hoebbel and Wieker (152a). In the compound having the empirical formula CuO 2en. 2Si0 2 7.4H 20, where en is NH 2C 2H.NH 2 , the
anion was shown to consist of a cube formed by two Si.O. rings, the corners of
which are linked by oxygen atoms. This probably corresponds to the smallest
"particle" of internally condensed silica that can be formed. It is a polysilicic acid
octamer which contains only one hydroxyl group per silicon atom and can be formulated [HOSiO I ~]8' The copper salt has the composition [Cu(en 2].[Si e0 2o ] 30H 20
(Figure 2.6). The compound was studied by paper chromatography, by the reactivity
with molybdate, and also by silylation to form the derivative [(CH3)3SiOSiOI' ~18'
which had also been made previously from the corresponding tetramethylammonium
silicate (126, 153). This octasilicic acid, which is called the "double-four-ring" silicic
acid, reacts with molbdic acid at a characteristic rate, 95% reacting in about 8.5




Organic Chelates of Silicon


.tes has
viev of

ate and
n salts
rs from
d when
and set
ammowith a
. mixed

ric ions
of an

e bases
)3 with
both in

rd havH 2 the
ners of
:ic acid
formu30H 2 0
au' )

Figure 1.6. The double-four ring, or cubic

octasilicate ion, Sis0 2os-.

These authors (152b) later showed that the analogous cobaltic complex silicate
had a similar composition, [Co(en)3MH 2Si,02o], with 16-28 water molecules. Here
there are only three negative charges per ring of four silicons. There may be a
proton, H 30 -, at the center of each ring.
The ethylenediamine copper silicate was obtained in the form of a slightly different hydrate [Cu(en)2].[Si a0 2o] 38H 20 and the structure determined by' X-ray diffraction by Smolin, Shepelev, and Butikova (153), who confirmed the existence of
the anion lSi a0 2o]a- which is a "fourfold two-story ring." The anions and cations are
bound into a structure through hydrogen-bonded water molecules.
The precursor of the "double-four-ring" ion is probably the "single-four-ring"
cyclic tetramer ion which was found to be the anion in crystalline lead orthosiJicate,
Si0 22PbO. In the lead silicate glass, Gotz, Masson, and Castelliz (154) showed that
the silica is present as the orthosilicate ion, (SiO.)-, but as devitrification proceeds,
higher polymer species (Si 20 7 ) S - and (Si.o 12 )a- are formed until the product is
crystalline lead cyclic tetrasilicate, having the structu re Pb.(Si.0 12 )S- .
Yates called attention to the numerous metal polyamine silicate compositions that
can be formed with amines H 2N C"H 2,,(NH C"H2,,).rNH2, where n = 2-6 and x =
0-3. Metals include Cd2+, Cr 3+ , Fe2+, l-.!n 2 " , NiH, and Zn 2+ . The silicate solutions
are film-forming and give insoluble metal silicates when heated (155).
In another type of silicate solution containing metal cations which normally result
in water-insoluble .silicates. the metal ions are kept from reacting by cornplexing
them with chelating agents (156).ORGANIC CHELATE5 OF 51 LICO:"

It is not widely recognized that silicon can enter into chelate-type bond formation
with some oxygen- and nitrogen-containing organic compounds such that silicon is
coordinated with six of these polar atoms. The overall structure may be negatively
or positively charged and could possibly be neutral. depending on the organic
molecule involved. Such hcxacoordinatcd compou nds of silicon arc relatively

=-. ..


Water-Soluble Silicates

uncommon because of the strict requirements of spacing and orientation of the COordinating atoms.
One necessary structure is that in catechol, where adjacent hydroxyl groups are
held in planar relation by the unsaturated carbon ring and the oxygen-oxygen
distance is exactly that required. between oxygen atoms in octahedral coordination
with a silicon atom. If the ring is saturated, as in 1.2-cyclohexanediol, the hydroxyl
groups are not in the plane and are farther apart. Another structure is that found in
l-hydroxypyridine N-oxide, where again the ring is unsaturated and there are two
oxygen atoms attached to adjacent carbon and nitrogen atoms in a six-membered
ring. It also appears that two oxygen atoms or hydroxyl groups on adjacent carbon
atoms in other than a six-membered ring can chelate with silicon if they are held the
correct distance apart (see also Chapter 7).
Chelates of silicon have already been discussed in some detail in Chapter 1 in connection with the effects of organic compounds on the solubility of silica and also on
the rate of dissolution of silica. Further discussion will be limited to some of the
compounds that have been formed.
Catechol Derivatives
The hexacoordinate silicon dianions in pyridinium tris(o-phenylenedioxy)siliconate
have been examined by X-ray diffraction by Flynn and Boer (157).
The oxygen atoms are only slightly distorted from octahedral symmetry. The correct formulation is [CaHsNH]2%+[(CaH402)3Si]2-.
The reaction of silica with catechol, pyrocatechol. and 2,3-naphthalenediol has
been studied by several investigators (158-162), but Bartels and Erlenmeyer appear
to have been the only ones to use this reaction to characterize the rate of
depolymerization of silica (l63a). For example, monomeric silicic acid from ethyl
silicate in a standard solution of catechol in 0.8 N HCl reacted rapidly and could be
titrated to a constant pH of 8.5 with an equivalent amount of standard NaOH solution in a few minutes. An equivalent amount of silica gel required 2.5 hr, but ignited
gel reacted only slightly in 5 hr. The rate of reaction, followed by a constant pH titration, provides a way to estimate the relative degree of polymerization of silica or
possibly the specific surface area.
Baumann (163b) examined the relative stabilities of the six-coordinated silicon
complex anions of some catechol derivatives. He measured the equilibrium constant
K and also the pH at which 99% of the silicon in a solution containing 60 ppm silica
as monomer is complexed when the concentration of the agent is 0.1 M.
Agent X


2.1 x 1011
2.4 x 1010
5.7 x 109




- --.157


Organic Chelates of Silicon

re coor-

Baumann also examined the details of the reaction of silicic acid with catechol
(163c) as well as with a catechol-formaldehyde condensate which acted as a polyphenol (163d). The latter resin. when not too highly cross-linked. removes soluble
silica from slightly alkaline water.

are two
aeld the
in conalso on
: of the


rate of
ould be
H soluignited
pH tit.ilica or
m silica

Humic Acids
The presence of catechol-like materials such as tannic acid in decaying 'vegetable
matter and their solubilizing effect on silica has led to the interesting theory that aluminum ore, bauxite, occurs predominantly in the tropics because of the preferential
leaching of silica from clays under conditions of heavy rainfall and large quantities
of decomposing vegetation. On the other hand, in colder climates the residues are
more siliceous because alumina is more soluble in water with a lower pH and higher
carbonic acid content in the absence of vegetation (164).
Humic acid from decaying vegetation is believed to have a structure related to
fulvic acid, which has adjacent hydroxyl groups on an unsaturated six-membered
carbon ring and forms chelates with silicon (165). The soluble chelate with ammonium humate has been used to supply silicon as a nutrient to rice plants, which
require silicon in their structure (166).
The formation of humic acid in soil may involve oxidation of phenolic vegetable
substances such as tannins while adsorbed on silica. A solution of pyrogallol passed
rapidly over silica gel produces brown oxidation products. The silica surface plays a
role since it is inactivated by being heated to high temperature or by absorption of
AI, Ca, or Mg ions (167). The mechanism of the reaction was examined by
Zeichmann (168).
Pentacoordinated organosilicon derivatives of RSi(OHh, where R is a
hydrocarbon group. formed by reaction with catechol have been described by Frye

Other Organic Compounds

According to Muetterties and Wright. tropolone (T). 1,2-diketocycloheptane. can

form a cationic chelate with hexacoordinated silicon, T 3Si +CI-, in acid solution
(171). It can be crystallized from water but is unstable toward alkali.
A cationic silicon chelate is also formed with l-hydroxypyridine N-o:dde. Gardner
and Katrizky prepared 24 derivatives of this and the l-arnino analogue (172), and
Weiss and Harvey made salts with different anions (173, 174). However, as
described in Chapter I these chelates, which are apparently stable in water at low
pH, decompose rapidly at higher pH.
A polymer containing this N-oxide configuration. poly(2-vinylpyridine N-oxide),
appears to combine with rnonosilicic acid according to Holt and Nasrallah (175). but
this may involve hydrogen bonding.

-- ---158

Water-Soluble Silicates

Chelates of silicon with triethanolamine are classed as "atranes" by Veronk ov

(176). These are essentially esters with an additional R group attached to the silicon
atom which is coordinated with the nitrogen atom, as shown by the high dipole
moment infrared and nmr spectra: N(CzHsOhSi. R.
Hydroxy acids, such as tartaric, do not form complexes with silicon as they do
with the larger germanium atom, according to Clark and Waddams (177), who
examined a large number of mono- and poly hydroxy aliphatic acids. including those
derived from sugar.


There are certain crystalline silicates of Li, Na, and K in which the silicate ions are
polymeric and thus are not directly soluble in water. The conversion of some of these
to hydrated crystalline silicas has been discussed in Chapter I. Since some of these
can occur or can be formed in silicate solutions and represent different polysilicate
structures, some further discussion is justified.
Until Wegst and Wills isolated crystalline hydrated sodium "tetrasilicate" (178)
it was not realized that alkali silicates of higher SiO z : NajO ratio than 2: I could be
crystallized. Subsequent investigation (179a) showed that this water-insolu hie silicate
was better represented by the formula 13SiO z3NazOllH zO. It is probably similar
to or the same as the mineral kanemite later described by Johan and Maglione
(179b) and shown by Benke and Lagaly (179c) to have a layer structure in which the
sodium can be exchanged with a variety of organic cations. A simple synthesis and a
variety of properties and transformations to other sodium silicates and crystalline
hydrated silicas were described. The idealized formula is NaHSizOs3HzO.
McCulloch (180) discovered two more hydrated crystalline silicates of higher
ratios, 9.4: I and 11.1: 1. Also, he found another tetrasilicate different from the one
found by Wegst and Wills.
All the sodium ions in these sodium polysilicates were exchangeable for hydrogen
and other metal ions, as Wills discovered in the case of the "tetrasilicate" (181). He
was thus probably the first to synthesize a crystalline hydrated form of silica. A later
study by Her (182) showed that a microcrystalline sodium polysilicate hydrate having the formula 8SiO zNazO9HzO had an X-ray diffraction pattern similar to that
of McCulloch's compound 9.4SiO z NazOxHzO. and demonstrated that since the
crystals had ion-exchange properties, there could be varying degrees of substitution
of Na" for H+. Also, it was concluded that the cyrstals have a layerlike structure,
since sodium could be substituted by C16H33N(CH3)3 + ions, widening the interlayer
spacing from around 11.2 to 13.6 A. The interior surfaces of the silica layers could
even be esterified with n-butyl alcohol with little change in the spacing.
Mineralogists, apparently quite independently of the above publications, became
interested in silicates of this type with the annou ncement in 1967 by Eugster (183) of
the minerals magadiite and kenyaite from Lake Magadi in Kenya, having the composition 14 SiO z Na zO9H 2 0 and 22SiO z ' NazO 10H 2 0 . The corresponding acid,





: silicon
1 l

they do
7), who
19 those

.ons are
of these
of these
ould be
ric' '-e
is ah~ a
. higher
the one
81). He
A later
ite havto that
nee the
s could
183) of
re comg


Hydrated Crystalline Alkali Metal

:- ...



6Si0 2 H 20, was claimed as the first known crystalline silica hydrate, indicating that
the mineralogists were not aware of the earlier investigations by McCulloch and
following workers. Magadiite was soon also discovered in California (184) and
Oregon (185).
In spite of the marked similarity in dehydration and ion-exchange properties to
the synthetic sodium polysilicate first made by McCulloch, magadiite differs in having a layer spacing of about 15.4-15.8 A instead of 11.2 A This is probably related
to the difference in Si0 2:Na 20 ratios of 14: 1 versus 8: 1. Conditions for the formation of magadiite were examined by Mariner and Surdam (186), who concluded that
the mineral must crystallize from a solution that is rich in silica and low in aluminum, since the brines of Lake Magadi range up to 1900 ppm Si0 2 but contain
only 1 ppm A1 203 Absence of alumina is undoubtedly of critical importance. .
The stability constants of both magadiite and kenyaite have been measured by
Bricker (187), who concluded that the minerals are stable in alkaline waters rich in
sodium and soluble silica. It appears that at pH 9 these crystalline sodium silicates
are actually less soluble in water than amorphous silica. The Gibbs molar free
energies of formation were also calculated as - 1,762.2 kcal for magadiite and
- 2,603.5 kcal for kenyaite.
Changes in the lattice spacing of magadiite when dehydrated and when Na + was
replaced by H+ were examined by Brindley (188). Lagaly and Beneke (189) further
examined the hydrogen form, or silica hydrate, which they formulated as
H 2Si 1402 9 5.4H 20, and found that it still retained water at 100C to the remarkable
extent of 14 Si0 2 : 1.5 H 20. The hydrogen form also absorbed various organic liquids
which became intercalated between the silica layers. For example, dimethyl sulfoxide
and various N-oxides and phosphine oxides gave different layer spacings with different arrangements of aromatic rings, A wide variety of heterocyclic nitrogen bases
of different ring sizes was examined and the I~yer spacings correlated with molecular
structure. Acid amides, urea and its derivatives, and alkyl amines received similar
Other workers (190a, 190b) studied rnagadiite in which sodium ions were
exchanged with long-chain and other quaternary ammonium ions, with special attention to the arrangement of the chains in the intercalated layers. A similar compound
had earlier been made by lIer (182) by replacing the sodium ion in McCulloch's
crystalline polysilicate with the cetyltrimethylarnmoniurn ion. Replacement of the
sodium in magadiite with other metal ions has not been examined, analogous to the
lithium, magnesium, copper, and nickel salts made by Iler (183) from McCulloch's
8: 1 ratio silicate.
It is now apparent that there can exist a series of similar polysilicate structures
depending on the ions present and conditions of formation. It is interesting that
m agadiite was synthesized by Lagaly and Beneke from a solution containing 27.4%
by weight of Si0 2 and 4.06 Na 20 with a molar ratio of Si0 2 : Na 20 of 7: lto obtain
a product with a ratio of 14: l , and under otherwise essentially identical conditions
Iler (182) used a ratio of 4.6: 1 to obtain McCulloch's product with a ratio of 8: I.
Ilcr also reported that starting with ratios of 5: I to 8: 1 a crystalline product with a
ratio about 20: 1 was obtained, but no X-ray pattern was reported.



Water-Soluble Silicates

Russian workers, Il'in et al. (191), have extended knowledge of this field by similarly preparing the following:
4-5Si0 2 Na 20. 30-60H 20
19.5Si02 Na 20 22H 20
12.5Si02 Na 2023H 20
These were apparently different from the previously reported crystalline sodium
polysilicates, but had similar ion-exchange properties. Systems containing Li + and
K + with different silica-alkali ratios were also studied. Crystalline silicates in the
lithium system can be expected, but my experience with the potassium system indicated that insoluble polysilicates did not form under conditions where the sodium
compounds were obtained.
Il'in et a1. (192) further examined the sodium silicate system and reported two
structural types of sodium polysilicates. One consists of rounded aggregates, 1-6
microns in size, of the general composition 9-l6Si0 2 Na 20 9-l2H 20. The second
type made at higher alkali ratio consisted of rectangular plates 2-6 microns across
with the general formula 8-l0Si0 2Na 2 0 9 - I I H 20. A third type,
2.5Si02 Na20 .xH 20, was reported also to be insoluble in water. Previously known
polysilicates appear to fall into one of the first two classes; except for the tetrasilicate, which was not reported. All the polysilicates were reported to have ionexchange properties like those of zeolites.
Mixtures or solutions with Si0 2:Na 20 ratios greater than 4: 1 are not necessary
for the formation of insoluble crystalline species, as indicated by Wegst and Wills
(178), who first made the "tetrasilicate." Apparently this species in one or more
crystalline forms can eventually crystallize out of cornrnercialvwater glass" solution
of 3.3: 1 ratio at 25C, as mentioned by McCulloch (180). What nucleates such
crystals is unknown, but on very rare occasions the whole contents of a container
partly filled with concentrated sodium silicate solution of this type has been known
to crystallize to a solid mass. This usually occurs after the solution has been aged for
several years and indicates the advisability of cleaning out storage tanks occasionally.





-~ i





Certain anhydrous crystalline silicates yield layerlike crystalline hydrated silicas with
the approximate formula (HzSi 20s)z. These are cited to illustrate the com plexity of
such apparently simple compounds.
Anhydrous crystalline K 2Si03 and KzSi z0 5 are hydrolyzed in methanol at 200C
to give two types of crystalline KHSi z0 5 With KOH present, crystalline (KHSi0 3 ).I
is obtained (193). The KHSi 20 s has a sheetlike silica structure from which the corresponding H 2Si z0 5 can be prepared (194), Similar crystalline silica hydrates can be
obtained by reacting acid with K zSi 20 5 , BaSi z0 5 , and beta-Na-Si.O, (195). From

Precipitation of Insoluble Silicates


. simi-

." and
in the
1 indiodium
:d two
.s, 1-6
trasili'e ion;e~

I WiIls
. more
s such
ged for

as with
.xity of
ISi03 )%
he corcan be



alpha-Na:Si 2 0 $ two sheetlike phyllosilicic acids. H2Si 2 0 $, were obtained, "depending

on whether a strong or a weak acid in methanol was used to exchange sodium ions.
Both silicas had structures similar to that of Li2Si 2 0 $, which proved impossible to
leach with acid, according to Le Bihan, Kalt, and Wey (194).
According to Hubert, Guth, and Wey (196), the beta-H 2Si 2 0 $ obtained by reacting beta-NajSi.O, gives a new'hydrated silica that gives a single DTA peak at 220C
where hydroxyl groups are lost. Still another form of H 2Si2 0 $ was obtained by
-extracting basic copper silicate with acid, having a two-dimensional (lepidoidal)
structure (197) of which the surface area was estimated by gas adsorption. .
Few if any of these various "disilicic acids," which are highly polymerized in two
dimensions, are identical. The relation of these to the previously discussed hydrated
crystalline silicas obtained from hydrated sodium polysilicates is not known. It is
evident that a large number of crystalline hydrated silicas may exist, some more stable than others, but all obtained from crystalline silicates by ion exchange. No
crystalline hydrate silica is likely to be formed directly in the silica-water system in
the absence of cations to bring about the organization of a regular polysilicate structure.


When a soluble silicate is mixed with solutions of salts of metals other. than the
alkali metal group, insoluble amorphous metal silicates are precipitated. However,
the nature of the precipitate can very widely, even with the same starting solutions,
depending on the intensity of agitation at the .point of mixing and whether one solution or the other is present locally in excess. The mechanism of formation of the
precipitate in the absence of mixing is visible in the making of the so-called
"chemical garden" in which crystals of metal salts are dropped into a relatively
concentrated solution of sodium silicate. The metal salt dissolves and at once a
mem brane is formed between the two solutions, consisting of amorphous metal silicate. However, since hydrogen and hydroxyl ions diffuse rapidly, silica gel is formed
on the silicate side and metal hydroxide on the metal-salt side. On the other hand,
when the two solutions are brought together in a zone of high shear, as by intensive
stirring, a more homogeneous, gelatinous amorphous metal silicate is precipitated.
In view of the fact that essentially all natural silicates are crystalline, it is of
. interest to consider why most synthetic silicates are amorphous when precipitated
from aqueous solution. In a crystalline metal silicate, metal ions and silicate ions of
uniform size and shape are arranged in a regular manner in a solid lattice or framework characteristic of the ionic sizes and charges. However. most solutions of soluble silicates contain not silicate ions of uniform size but. instead. a mixture of polysilicate ions, some containing up to 5-10 SiO: units each. Even in the case of
highly alkaline metasilicates which are known to contain monomeric silicate ions.
lowering the pH upon addition of a solution of a metal salt causes rapid polymerization to polysilicate ions. Since the polysilicatc ions are not of uniform size. they cannot arrange themselves along with the metal ions into a regular crystal lattice. Thus

., =-.---


.... ; .,.'




Water-Soluble Silicates

when polysilicate ions combine with metal ions, the resulting insoluble precipitate is
almost always amorphous. In order to obtain a crystalline silicate, the polysilicate
ions must be depolymerized to smaller silicate ions of uniform size in order to rearrange themselves into a regular lattice.
Polymerization of basic metal ions is another factor which may interfere with
crystallization. When the metal salt is mixed with silicate, the increase in pH in the
environment of the metal ions at some point causes the formation of polymeric basic
metal ions or colloidal metal hydroxide, which likewise cannot fit into the silicate
For these reasons, the precipitation of a metal silicate from aqueous solution at
ordinary temperature does not yield crystalline silicate, but instead usually results
in a colloidal mixture which may be visualized as an adsorption of metal ions on
gelatinous silica or as a mutual coagulation of the positively charged colloidal metal
hydroxide and negatively charged colloidal silica.
In dilute solutions, colloidal metal silicates are precipitated at a pH slightly below
that at which the metal hydroxide alone would be precipitated, according to Britton
(198, 199). This suggests that no definite compounds are formed and that the
hydroxides are precipitated along with silica by mutual coagulation.
From the above, it will be evident that "colloidal silicates" may vary from rather
homogeneous colloidal aggregates of extremely small' ultimate units of polysilicic
acids and metal hydroxide, to heterogeneous masses in which either 'silica or the
metal hydroxide is present as discrete colloidal units, held together by the other
Vail (I) has extensively reviewed the practical aspects of precipitated metal silicates. In the case of magnesium silicates, which are used as adsorbents, decolorizing
agents, catalysts, and fillers in paints and also for treating stomach ulcers, the composition may be varied within wide limits. The strong tendency for magnesium
hydroxide to react with silica (even in the monomeric state) is shown by the fact that
the addition of 300 ppm Mg(OHh to water will reduce the soluble silica content
from 42 to 0.1 ppm.
In the case of precipitated calcium silicates made from commercial sodium silicate
at ordinary temperature (Si0 2 : Na 20 = 2.4: 1.0). there was no evidence of
crystallinity (200).
Precipitates from 3.3 ratio silicate (201) were found to consist of gels along with
excess Ca(OHh. Uses for soluble silicates, depending on the formation of insoluble
silicates. are extremely varied, ranging from surface coatings. pigments, adsorbents,
and fillers to use in cements and oil-well drilling muds. The precipitation of zinc
salts with sodium silicates at ordinary temperature appears to involve a mutual COJgulation of hydrated zinc oxide and silica, no crystalline compounds being formed
Wells (203) investigated the precipitation of a number of mctal salts with 3.2 ratio
silicate using insufficient silicate to react with the metal ions and found that by using
pairs of metal salts the relative ease of precipitation could be observed. The order of
precipitability, using mostly metal sulfates and starting with the most readily



Soluble Silicate Glasses

.tate is

precipitated metal, is copper, zinc, manganese, cadm iurn, lead, nickel, silver: magnesium, and calcium.
Under certain circumstances, where the metal hydroxide is formed in such a way
that it may be surrounded by hydrated silica, "metal silicate" sols, stabilized by
alkali, can be obtained. Vail states that those metal salts that are acid reacting, such
as FeCh, or acidified solutions of C 0 2 + , Zn 2 + , or Cu 2 + salts react with the silicate to
liberate colloidal silica, which then acts as a protective colloid, that is, forms a protective layer on the colloidal metal hydroxide. This being the case, it is not surprising
that high-ratio silicates favor the formation of such dispersions by supplying more
Precipitated metal silicates are difficult to characterize physically because, as Vail
points out, they are usually not homogeneous in composition. Upon the addition of
one solution to the other, it is obvious that, at the point of mixing, there is no control of the ratios of silicate and metal ions which are brought together: yet the composition of the precipitate as it is formed depends on the ratio of the components in
the reaction medium. Thus the composition of the first part of the precipitate which
is formed differs from that of the last. Homogeneous gels can be prepared by keeping the pH low, for example, until all the metal salt and silicate solution are mixed.
However, when the pH is then raised to gel the mass, it is still not truly homogeneous, since the silica is polymerizing continuously in such a mixture so that the
proportion of metal ions bound to the silica particles depends on how far' the silica
has polymerized before the pH is raised.
Metal silicates may crystallize in the amorphous mass after aging, especially if
components are present in suitable stoichiometric proportions. Also in some
instances, microcrystalline metal silicates may be formed immediately if the solutions of metal salt and silicate are brought together at high temperature or in very
dilute solution. It is unlikely, in any case, that macrocrystalline metal silicates
similar to those found in minerals will be formed, except at high tem perature under
pressure. However, when aluminum is present for form aluminosilicate ions, fairly
rapid crystallization of certain aluminosilicates such as the zeolites can occur at

iilir- ...

) rv.
e with
in the
:: basic
:ion at
ons on
at the
or the
, other
e comresium
ct that
tee of

g with
of zinc
al coa'orrned
2 ratio
y using
rder of


For a further discussion of colloidal silicates, see references listed in the introduction, the chapter on colloidal silicates in Colloid Chemistry of Silica and Silicates
(29), and in Chapter 6, the section on polyvalent metal cations.


Since most soluble silicates are made by dissolving the corresponding sodium or
potassium silicate glasses, it seems appropriate to review some of the investigations
of glasses that have been made since Vail's comprehensive survey in 1952 (I).
However, it should be kept in mind that the structures occurring in these glasses
bear little or no relation to the nature of silica in the derived aqueous solutions
beyond the effect of the Si0 2 : Na 10 ratio.







Water-Soluble Silicates

A theory as to why silicate glasses do not readily devitrify involves the "presence of
small polymeric silica units of pentagonal symmetry which cannot fit into any
regular crystal lattice. Tilton (204) proposes that this "vitron" unit is built up of
pentagonal rings of SiO. tetrahedra, with the silicon atoms lying in a plane with all
Si-O-Si bond angles close 1{> 1800). As further tetrahedra are added, a stressed network cluster of 20 SiO: units is built up in such a way that further growth cannot
In sodium (or potassium) silicate glasses of SiO:: Na20 molar ratios of2: I to 4: 1
such "vitron" units, if actually present, might disperse without complete depolyrneri,
zation, giving the colloidal species known to exist in solution. However, since a 3.3: I
ratio solution appears to have the same properties whether made from glass or by
dissolving amorphous fine silica in alkali solution, it seems unlikely that the same
"vitrons" are formed spontaneously in solution. Possibly a detailed examination of
3.3 or 3.8 ratio solutions of sodium silicate made by the two methods might still reveal
a persistent difference.
The "vitron" theory is probably related to the later "iceberg" theory, namely, the
tendency of silica to form small immiscible clusters in melts of lower alkali content
(205). In sodium silicate glass with a SiO::Na 20 ratio of 7:1, the model predicts
collidal silica units or particles about 1.9 nm in diameter.
According to Tran (206), electron microscopic examination of sodium silicate
glasses shows two principal structural elements, the silica and 3Si0 2 Na 20 groups.
Masson has applied polymer theory to silicate melts to predict the distribution of
ionic species over the entire range of compositions up to the maximum polymerization allowed by the theory. Results were compared with experimental values for
polymer distribution determined by converting the ionic species in the glasses to trimethylsilyl derivatives (207-212).
Similar calculations of size distribution of ions in silicate melts were made by
Hess (213). Structural models of binary silicate melts have also been proposed by
Gaskell (214), who discussed experimental evidence for and against the validity of
the models.



The only mention by Vail (1) of peroxide structures in relation to silicates is the stabilizing effect of sodium silicate in combination with magnesium silicate in stabilizing peroxide bleaching baths, probably by adsorbing and inactivating traces of
catalytically active metals such as copper.
Rietz and Kopp (215) have found that dissolution of sodium metasilicate,
NazH2SiO~ 8H:O. in 30% H 20: gave crystalline NazSiOJ" 3H zO z . However,
this peroxide decomposes at room temperature via. the intermediate
Na:Si0 3 H 20 2 2H 20 to Na 2Si0 3 3H 20. which is dehydrated to Na zSi0 3 H 20,
which is stable up to 480C, then decomposes exothermally to (Na:SiOJ)r'
Hydrogen peroxide is useful as a blowing agent, along with MgO and paraforrnaldehyde, in making low density foams from a solution of sodium silicate (216).




ice of
) a-'

th all
j net1000t

men3.3: 1
or by
on of

y, the
ion of
to trr-

de by
.ed by
lity of



1. J. G. Vail, Soluble Silicates (ACS Monograph Series), Vols.

and 2, Reinhold, New

York, (1952).
2. J. R. Williams, Ind. Chem., 37,419 (1961).
3. O. W. Burke, Jr., U.S. Pat. 3,537,815, (1967).
4. V. A. Chubinidze, V. V. Pollyak, and Y. W. Troyankin, Ref Zh. Khim, 1972, Abst.
5. F. Spalink, Ger. Pat. 2,116,661 (Cl. COlb), Nov. 2, 1972; Pat. Appl. 21-18-66f.8, April
17, 1971.
6. G. Y. Anders (Selas Corp., Dresher, Pa.), Proc, Fall Meeting, Material Equipment White
Wares Divn., Amer. Ceram, Soc. 1970, p. 94.
7. R. Terai, 1. Sugae, and R. Hayami, Zairyo, 17 (177),527 (1968).
8. M. Truhlarova and O. Veprek, Sb. Vys. Sk. Chern, Technol. Praze, Anorg. Chem,
Technol., 13, 89 (1969) [Chem. Abstr., 73, 123152r]; Glastech. Ber., 40 (7),257 (1967)
[Chem. Abstr., 67, 102416r].
9. K. G. Kreider and A. R. Cooper, Glass. Technol. 8 (3), 71 (1967).
10. J. Hlavac and H. Nadelmynska, Glass. Technol. 10 (2), 54 (1969) [Chem. Abstr., 70,
II. E. J. Tauch, U.S. Pat. 3,090,679 (Du Pont), 1963.
12. R. K. Rathmell, U.S. Pat. 3,203,755 (Du Pont), 1965.
13. R. K. Her and E. J. Tauch, A.I.Ch.E. Trans. 37, 853 (1941).
14. N. W. Hanfand M. J. Sole, Inst, Min. Metall. Trans. Sect. C, 81, 97 (1972).
15. P. Beyersdorfer, Silikat tech, 3, 11 (1952); Ceram, Abstr. J. Am. Ceram, Soc., 36 (9), 159
16. S. B. Holmquist, Glass. Ind., 47 (10),558 (1966).
17. E. Ordiales, Ger. Pat. 1,900,066 (Cl. COlb), Aug. 28, 1969 [Chem. and Abstr., 71,
18. G. W. Morey and J. M. Hesselgesser, Am.J. Sci. Bowen Vol. Pt. 2, 343 (1952).
19. H. H ..Weldes and K. R. Lange, Ind. Eng. Chem., 61, 29 (1969).
20. E. V. Shtyrenkov, V. M. Yaushev, V. 1. Gorshkov, N. S. Yadrintsev, and Y. P. Yakunin,
Khim, Neft. Mashlnostr., 1975 (3),40.

re sta-

abilizces of

21. L. R. Bacon and J. H. Wills, J. Franklin Inst .. 258 (5), 347 (1954); Soap Chem, Spec., 30
85,86,89, 105 (1954).
22. H. Ukihashi, Bull. Chern. Soc, Jap., 29,537 (1956).
23. R. Grant and C. R. Masson,J. Colloid Interface Sci., 41, 606 (1972).

. H 20,

24. D. Warren, Fonderia, 21 (11),379 (1972).

25. K. E. L. Nicholas, The Carbon Dioxide-Silicate Process in Foundries. (British Cast Iron
Research Association Alvechurch). 1972.
26. P. C. Yates, U.S. Pat. 3,540,519 (Du Pont), \970.
27. P. S. Roller and G. Ervin, Jr.,J. Am. Chem, Soc., 62,46\ (\940).


28. G. W. Scars. Anal. Chem., 28,1981 (\956).


Water-Soluble Silicates

29. R. K. Iler, Colloid Chemistry of Silica and Silicates. Cornell University' Press. Ithaca.
N.Y . (1955).
30. C. L. Baker, L. R. Jue, and J. H. Wills. 1. Arn. Chern. Soc. 72,5369 (1950).
31. J. Aveston, J. Chern. Soc., 1965,4444.
32. G. Jander and K. F. Jahx Kolloid-Beih., 41,48 (1934).
33. G. Jander and W. Heukeshoven, Z. Anorg, AUg. Chem., 201, 301 (1931).
34. R. W. Harman, J. Phys. Chem., 32,44 (1928).
35. V. R. Main,J. Phys, Chem., 30,535 (1926).
36. G. Hagg, Z. Anorg: Ailg. Chem., 155,21 (1926).
37. R. V. Nauman and P. Debye,]. Phys. Chem., 55, 1 (1951).

39. W. C. Beard. Adv. Chern. Ser., 121 (Mol. Sieves. Int. Conf. 3rd). 162 (1973).


40. M. V. Borisov and B. N. Ryzhenko, Geokhlmlya, 1974 (9). 1367.



38a. A. B. Brady, A. G. Brown, and H. Huff, J. Colloid Sci., 8, 256 (1953).


38b. H. C. Marsman, Z. Naturforsch . 29B. 495 (1974).


38c. G. Engelhardt et aI., Z. Anorg Allgem, Chem., 418, 17 (1975).

41. S. A. Greenberg,]. Am, Chern, Soc . 80,6508 (1958).

42. E. Franks and A. Mukherjee, J. App/. Electrochem., 3 (3), 179 (1973).
43. G. Lagerstrom, Acta Chern. Scand. 13,722 (1959); Rec. Trav, Chim . 75, 721 (1956).
44a. N. Ingri, Acta. Chern. Scand., 13,758 (1959); Sven. Kern. Tidskr., 75 (n. 199 (1963).
44b. R. H. Busey and R. E. Mesmer,lnorg. Chem., 16 (10),2444 (1977).
45. N. K. Ivanov and T: G. Gur'eva, Zh. Obshch. Khim., 44 (5). 964 (1974).
46. B. N. Ryzhenko, Geokhimiya, 1%7 (2). 161.
47. E. Thilo and G. Kruger, Z. Elektrochem., Ber. Bunsenges Phys. Chern. 61, 24 (1957).


48. E. Thilo, Angew, Chem., 70,107 (1958).

49. E. Thilo and G. Kruger, Angew. Chem., 68, 154 (1956).

50. E. Thilo, G. Kruger, et aI., Z. Elektrochem . Ber, Bunsenges, Phys. Chem., 61, 1305

51. E. Thilo, F. Wodtcke, and H. Funk. Z. Anorg, AUg. Chern. 278, (5-6) 225 (1955).
52. M. A. McGarry and J. E. Hazel, J. Colloid sa; 20, 72 (1965).
53. L. Lehrman and H. L. Shuldener, J. Am. Water Works Assoc. 43, 175 (1951).
54. I. Z. Pevzner, N. I. Erernin, N. N. Knyazeva, Y. B. Rozen, and V. E. Mironov, Zh.
Neorg. Khim., 18 (4). 1I29 (1973).
55. A. P. Brady. A. G. Brown, and H. Huff. J. Col/Did

sa.. 8, 252 (1953).

56. N. E. Prikhid'ko and V. S. Molchanov, Kolloidn. Zh., 13,450 (1951) [Chern. Abstr.. 46,
2377 (1952)].
57. J. M. Ordway. Arn.J. Sci. 40 (Ser. 2).173.190 (1865); 24 (Ser, 4). 473 (1907).
58a. I. E. Muskat and F. D. Ayres. U.S. Pat. 2,285.299, (1942).
58b. C. L. Baker and J. F. Austin, U.S. Pat. 3,301.701 (Philadelphia Quartz Co.). 1967.
59. F. Mylius, E. Groschuff, Ber. Dtsch. Chern. Ges., 39, 116 (1906).
60. H. Kraut, Ber. Disch, Chern. Ges., 64B, 1709 (193 [).



61. E. Weitz, H. Franck, and M. Schuchard. Chem, Zbg., 74, 256 (1950).
62. G. B. Alexander, J. Arn. Chern. Soc . 75, 2887 (1953).





-- =-....

: Silicates
s, Ithaca.



63. R. K. Iler, U.S. Pat. 2.588.389 (Du Pont). 1952.

64. R. K. Iler,}. Phys. Chern. 57, 654 (1953).
65. T. L. O'Connor.v. Phys. Chem., 65, 1 (1961).
{)6. W. Wicker and D. Hoebbe1, Z. Anorg, Chem., 366, 139 (1969).
67. D. Hoebbel and W. Wieker, Z. Anorg, AI/g. Chem., 400, 148 (1973).
68a. E. Thilo, W. Wieker, and H. Stade, Z. Anorg, AI/g. Chem., 340, 261 (1965).
68b. M. Coudurier, R. Baudru, and J.-B. Don~et, Bull, Soc. Chim, France (9) 3147 (1971).
69. R. K. lIer and P. S. Pinkney, Ind. Eng. Chem., 39, 1379 (1947).
70. J. S. Kirk, U.S. Pat. 2,408,656 (Du Pont), 1946.
71. C. W. Lentz, Chern. Eng. News, 44 (Sept. 23, 1963).
72a. C. W. Lentz, Inorg. Chem., 3,574 (1964).
72b. L. S. D. Glasser and S. K. Sharma, Br. Polym, J., 6,283 (1974).
73. J. Gdtz and C. R. Masson,J. Chern. Soc. A, 1970, 2683;J. Chern. Soc. A, 1971,686.
74. S. K. Sharma, L. S. D. Glasser, and C. R. Masson,J. Chern. Soc. A, 1973, 1324.
75. D. Hoebbel and W. Wieker, Z. Anorg. Allg. Chem., 405, 163 (1974).
76. W. A. Weyl, "A New Approach to Surface Chemistry and to Heterogeneous Catalysis,"
Miner. Ind. Exp, Stn. Bull., 57, 46-48 (1951) [State College, Pa.].


77. W. B. Treadwell and W. Wieland, He/v. Chim, Acta, 13,842 (1930).

78. R. K. Iler, Ref. 29, p. 37.
79. D. Fortnum and J. O. Edwards,J./norg. Nucl. Chem., 2, 264 (1956).
80. J. E. Early, D. Fortnum, A. Wojicicki, and J. O. Edwards, J. Arn. Chern. Soc., 81, 1295
( 1959).
81. E. Freund Bull. Soc. Chim, r-; /973 (7-8), 2238,2244.
82. B. M. Mitsyuk et al., Ukr. Khim, Zh., 32, 833 (1966).
83. R. A. Edge and H. F. W.,Taylor, Nature, 224, 363 (1969).

, 61, 1305

84. P. B. Jamieson and L. S. Dent Glasser, Acta Crystallogr., 20, 688 (1966); 22, 507 (1967).
85. K. H. Jost and W. Hilmer, Acta Crystallogr., 21, 583 (1966).
86. E. Thilo and W. Miedreich, Z. Anorg, AI/g. Chem., 267, 76. (1951).
87. M. V. Borisov and B. N. Ryzhenko, Geokhimiya, 9, 1367 (1974).
88. R. H. Schneider and J. B. Schutt, U.S. Pat. 3,615.781,1971.

.onov. Zh.

89. R. K. Iler, U.S. Pat. 3,492,137 (Du Pont), 1970.

90. A. Oken, U.S. Pat. 3,533,816 (Du Pont), 1965.

Abstr. 46,

91.C. B. Meyers. U.S. Pat. 3,712,941 (Diamond Shamrock Corp.), 1973.

92. H. Biegler and J. Wiljes. Ger. Pat. 2.219,503. Appl, p-22-19-503.5 (Degussa), Apr. 21.
1975 [Chern. Abstr. 80, 38891f].
93. W. J. Waltersdorf, Def. Pubi. U.S. Pat.


orr. 728.926. Jan.

7, 1969.

94. R. Sano et al; Ger. Pat. 2.060.155 (Kansai Paint Co.) June 16. 1971 (Chem. Abstr., 75,
95. T. Yano, Jap. Pat. 74-34529 (NGK Insulators) Chern, Abstr. 81, 79489r].
96. R. D. Shoup, Ger. Pat. 2.129,029. 1972; U.S. Pat. 3.678.144 (Corning Glass Works).
1972; R. D. Shoup and W. J. Wcing, U.S. Pat. 4,059.658.1977.


Water-Soluble Silicates

97. R. D. Shoup "Controlled Pore Silica Bodies Gelled from Silica Sol-Alkali Silicate-Mtr:
tures;" in Colloid and Interface Science. Vol. 3. Academic Press. New York. 1976. p. 63.
98. R. H. Patton and W. K. Stich, U.S. Pat. 3.180,746 (Esso Research and Engineering),
99. J. S. S. Bobb and H. H. Weides, Brit. Pat. 1,198.313 (Philadelphia Quartz Co.). 1970
[Chern. Abstr., 81, 93572y].
100. M. A. Segura and E. A. Hunter, Brit. Pat. 1,263,984 (Esso Research and Engineering)
101. R. K. Iler, U.S. Pat. 2,668,149 (Du Pont) 1954.
102. Polysilicate 48 and 85, E. I. du Pont de Nemours & Co., Wilmington, Del. [technical
103. F. L. Cuneo, U.S. Pat. 3,392,039 (Philadelphia Quartz Co.) 1968).

104. H. V. Freyhold and V. Wehle. U.S. Pat. 3.579,597 (Philadelphia Quartz Co.), 197\ [see
also Chern. Abstr., 72, 47972z].
105. R. H. Patton and J. B. Cox, U.S. Pat. 3.180,747 (Esso Research and Development),
106. K. Ohsawa and M. Tanaka. Jap. Pats. 7315160. 73-32796 (Furukawa Electric Co.), 1973
[Chern. Abstr., 79, 46131c].
107. M. W. Forsyth, U.S. Pat. 3,522,066 (Lithium Corp. Amer.), 1970.
108. S. Isohata and K. Sasaki, Jap. Pat. 74-18920 (Asaki Chern. Ind.), 1974 [Chem. Abstr.,
81, 16098w].
109. W. M. McMahon and C. B. Abda, U.S. Pal. 3,130.061 (American Pipe & Construction
Co.). 1964.
110. G. W. Sears, U.S. Pal. 3,455,709 (Du Pont). 1969.
Ill. G. W. Sears and D. R. Smith, U.S. Pat. 3,549,395 (Du Pont), 1970.
112. R. E. Steck, U.S. Pat. 3,624,234, 1971.
113. K. Suzuki et al., Jap. Pat. 7415721 (Nissan Chern. Ind.), 1974.







..' .'







f ;

114. K. Suzuki et al., Jap. Pat. 74-22427 (Nissan Chern. Ind.), 1974.
115. J. Dupre and K. A. Boornan, U.S. Pat. 3,655.582 (Rohm & Haas), 1972.
116. V. P. Simpson and F. A. Simko,J. Oil Colour Chem, Assoc., 56 (10). 491 (1973).
117. M. G. B. Wright. Paint Varn. Prod., 61 (10). 65 (1971).

118. Zincolite Zinc Dust. Supp!. 2, March, 1964, Amalgamated Oxides 1939 Limited, Victoria
Works, Dartford Kent [Richmond Towers Publishers. London].
119. M. Shimano, Jap. Pat. 74-44564 (Kubota. Ltd.). 1974.
120. C. Y. Man, U.S. Pat. 3,630,869 (Du Pont). 1971.
121. E. R. Werner. Jr., U.S. Pats. 3,692,558. 3.697.309, 3.694.392 (Du Pont). 1972.
122. R. H. Sarns, U.S. Pat. 3.565,675 (Philadelphia Quartz Co.), 1971 [Chem. Abstr. 82.
123. R. Schwarz. Ber, Disch, Chern. Ges . 49, 2358 (1916).
124. R. C. Merrill and R. W. Spencer,J. Phys, Colloid Chem .. 55,187 ([951).
125. R. C. Merrill. U.S. Pat. 2.689.245 (National Aluminate. Ltd.). 1954.
12641. D. Hoebbel and W. Wieker, Z. Anorg: Allg. Chem.; 38-1 (1).43 (1971).
126b. D. Hoebbel and W. Wicker. Z. Anorg, Allgem. Chern . 418, 35 (1975).


-- =---Silicates


ate Mix5, p. 63.

127. Joseph Crosfield and Sons, Ltd., Brit. Pat. 709,634, 1954 [Chern. Abstr.. 48. 124'61f].


128. H. H. \Veldes. Neth. Pat. Appl. 6,611,654, 1967 [Chern. Abstr. 67, 90385q].
129. H. H. Weldes, U.S. Pat. 3.301,853 (Philadelphia Quartz Co.), 1967.

0.), 1970


130. H. H. Weldes, U.S. Pat. 3,326,910 (Philadelphia Quartz Co.), 1967.

131. H. H. Weldes, U.S. Pat. 3,239,521 (Philadelphia Quartz Co.), 1966.
132. F. J. Wolter, U.S. Pat. 2,601,352 (Du Pont), 1952.
133. P. C. Yates, U.S. Pat. 3,597.248 (Du Pont), 1971.
134. P. C. Yates, U.S. Pat. 3,592,832 (Du Pont). 1971.


135. P. C. Yates, U.S. Pat. 3,475.375 (Du Pont). 1969.

136. M. Nakajima, A. Yamaguchi, and M. Tahara, Jap. Pat. 73-71728 (Nippon Carbide
Indust. Co.), 1973.

1971 [see

137. W. M. Lee, U.S. Pat. 3,024,125 (Pennsalt Chem. Corp.), 1962.

138. QURA},[i) Bulletin 20-H, Philadelphia Quartz Co . Philadelphia, Pa., 1975.


139. H. H. Weides, Nuova Chim . 47 (8),37 (1971) [Chern. Abstr. 75, 147244q].
140. H. H. Weides, Ind. Eng. Chern. Prod. Res. Div. 9. 243 (1970).

'0.), 1973

141. H. H. Weldes, U.S. Pat. 3.769,309 (Philadelphia Quartz Co.), 1973.

142. Henkel, et Cie, Fr. Pat. 1.551.442. Dec. 27, 1968 [Chern. Abstr., 71, 114765z].
143. H. V. Freyhold and V. Wehle, U.S. Pat. 3.625.722 (Philadelphia Quartz Co.), 1972.




144. P. C. Yates, U.S. Pat. 3.630,954 (Du Pont), 1971.

145. P. C. Yates, U.S. Pat. 3,655,578 (Du Pont). 1972.
146. H. Sawai, Jap, Pat. 71-38590 (Sawai Pharmaceutical Co.), 1971.
147. W. Mengel. H. Leuchs, K. Wolfgang, and R. Strufe, Belg. Pat. 626,545, 1963 [Chern.
Abstr. W. 8038a].
148. W. M. Meier and J. B. Uytterhoeven, Molecular Sieves. (Adv. Chern. Ser., 121), 132
149. S. A. Gerow, U.S. Pat. 3,661.602 (Du Pont). 1972.
150. V. W. Weidman and P. C. Yates, U.S. Pat. 3,725,095 (Du Pont), 1973.
lSI. R. Schwarz and H. Bausch. Ber. Dtsch. Chern. Ges., 54. 802 (1921).


152a. D. Hoebbel and W. Wicker, Z. Chem., 12 (8), 295 (1972).

152b. D. Hoebbel and W. Wieker, Z. Anorg. Allg. Chem. 423. 225 (19~6) .

. Victoria

153. Y. I. Smolin. Y. F. Shepelev, and I. K. Butik ova, Kristallografiya, 17 (1),15 (1972).

154. J. Gotz, C. R. Masson, and L. M. Castelliz, Amorphous Materials, Wiley-Interscience.
New York, 1972. p. 317.
155. P. C. Yates, U.S. Pat. 3,637,775 (Du Pont), 1972.
156. L. E. Campbell. U.S. Pat. 3,715,224 (Corning Glass Works). 1973.

ibstr. 82,

157. J. J. Flynn and F. P. Boer, J. Am. Chem, Soc.. 91. 5756 (1969).
158. R. Hess, R. Bach, et al., Experientia, 16, 38 (1960); 22. 515 (1966).
159. R. Schwarz and W. Kuchen, Z. Anorg. Allg. Chem .. 266,185 (1951):
160. H. Bauman. Beitr. Silikose-Forsch, Sonderb., 5, II (1962) 6. 49 (1963).
161. H. Bauman, Beitr. Silikose-Forsch., 85, I. 15,27 (1965).
162. A. Weiss. G. Reiff and Weiss, Z. Anorg. Allg. Chem .. 311,142.151 (1961).




Water-Soluble Silicates

163a. H. Bartels and H. Erlenmeyer, Helv, Chim . Acta. 47,7 (1964).

163b. H. Baumann. Beitr, Silikose Forch, Sonderb . 4,43 (1960).
163c. H. Baumann, Beitr. Silikose-Forsch; Sonderb. 5, 11 (1963).
163d. H Baumann. Kolloid-Z. Z. Polym., 188 (1),23 (1963).
164. J. P. Miller, personal communication.
165. A. V. Fotiev, Dok/. Akad. Nauk, SSSR, ]99 (1), 198 (1971) [Chem. Abstr., 75,120230v].
166. Roger W. McCormack, U.S. Pat. 3,552,943 (Scientism Labs.), 1971.
167. S. M. Siegel., Experientia, ]6,358 (1960).
168. W. Ziechrnann, Z. Pflantzenernaehr. Dueng; Bodenkd. 84, 155 (1959).
169. C. L. Frye, J. Am. Chern. Soc., 86, 3170 (1964).

. :.:

170. C. L. Frye, U.S. Pat, 3,455,980 (Dow Corning Corp.), 1969; U.S. Pat. 3,576,025 (Dow
Corning Corp.), 1971.
171. E. L. Muetterties and C. M. Wright,J. Am. Chern. Soc., 86, 5132 (1964); 87, 21 (1965).
172. J. N. Gardner and A. R. Katrizky.v. Chem, Soc., 1975,4375.
173. A Weiss and D. R. Harvey. Angew, Chem., 76, 818 (1964); 77, 347 (1965).
174. A. Weiss and D. R. Harvey. Agnew, Chem, Int. Ed. 3 (10) 698 (1964).
175. P. F. Holt and E. T. Nasrallah, Nature, 211, 878 (1966).
. 176. M. G. Voronkov, Khim, Geterosikl. Soedin., 511 (4) (1966) [Chern. Abstr., 65,8915,66,
177. E. R. Clark and J. A. Waddams, Nature, 180,904 (1957).
178. W. F. Wegst, J. H. Wills, U.S. Pat. 2,179,806 (Philadelphia Quartz Co.), 1939.
179a. C. L. Baker, L. R. Jue, and J. H. Wills, J. Am. Chem, Soc., 72,5369 (1950).
179b. Z. Johan and G. F. Maglione, Bull. Soc. Fr. Mineral. Crystallogr., 95, 371 (1972).
179c. K. Benke, G. lagaly, Amer. Mineral. 62, 763 (1977).
180. L. McCulloch,J. Am. Chem, Soc., 74, 2453 (1952).
181. J. H. Wills, U.S. Pal. 2,640,756 (Philadelphia Quartz Co.). 1953.
182. R. K. I1er,J. Colloid s, 19 (7), 648 (1964).
183. H. P. Eugster, Science, 157, 1177 (1967).
184. J. L. McAtee, L. R. House, and H. P. Eugster, Am. Mineral., 53, 2061 (1968).
185. T. P. Rooney, B. F. Jones, and J. T. Neal, Arner. Mineral., 54, 1034 (1969).
186. R. H. Mariner and R. C. Surdam, Science, 170,977 (1970).
187. O. P. Bricker, Amer, Mineral.. 54, 1026 (1969).
" ;.

188. G. W. Brindley, Am. Mineral., 54,1583 (1969).

189. G. Lagaly and K. Beneke, Am. Mineral., 60, 642,650 (1975).
190a. G. Lagaly, K. Beneke, and A. Weiss. Z. Naturforsch . 28b, 234 (1973).
190b. J. Voitlader. E. K. H. Wittich, and G, Lagaly, Z. Naturforsch . 303.1330 (1975).
191. V. G. Il'in, L. F. Kirichenk o, et a!', Dokl. Phys. Chern, USSR, Engl. Transl.. 174.408
192. V. G. Il'in, N. V. Turutina, K. P. Kazakov, and V. M. Behezovsk ii, Dok/. Akad. Nauk:
SSSR, 209 (5),1102 (1973).

193. H. Funk and H. Stade. Z. Anorg: AUg. Chem . 315,80 (l962).

194. M. T. Lc Bihan, A. Kalt. and R. Wey. Bull. Soc. Fr. Mineral. Cristallogr. 94. 15 (1971).








195. F. Wodtcke and F. Liebau, Z. Anorg, Allg. Chem . 335, 178 (1965).
196. Y. Hubert, J. L. Guth, and R. Wey, C. R. Acad. Sci. Ser. D 278 (II) 1453 (1974).
197. H. Kautsky and R. Irnich, Z. Anorg, Allg. Chem., 295, 193.206 (1958).
198. H. T. S. Britton, J. Chem. Soc., 1927,425.


199. H. T. S. Britton, Hydrogen Ions, 3rd ed, Vol. 2 Chapman and Hill, London, 1942.
200. V. A. Kargin, S. A. Kats. and A. F. Kornovskii, J. Appl. Chem, (USSR), 10,82 (1937).
201. G. L. Kalousek,J. Res. Natl. Bur. Stand., 32, 285 (1944).
202. J. G. Vail, Ref. 1, Vol. 1, pp. 154,234,241.
203. R. C. Wells, U.S. Geo/. Surv. Bull., 609 (1915).
204. L. W. Tilton,J. Res. Natl. Bur. Stand., 59 (2),139 (1957).

5 (Dow

205. H. Bloom and J. O'M. Bockris, in B. R. Sundheim, Ed., Fused Salts, McGraw Hill. New
York, 1964, p. 56.


206. T. L. Tran, C. R. Acad. s. 258, 4234 (1964).

207. C. R. Masson, Proc, R. Soc. London, Ser. A., 287, 201 (1965).
208. C. R. Masson,J. Am. Ceram, Soc., 51, 134 (1968).
209. C. R. Masson, Can. J. Chem., 48 (1), 20 I (1970).
210. C. R. Masson. Chemtech, 56 January (1974).

)[5, 66,

211. C. R. Masson, Nuova Chim., 50 (3),39 (1974).

212. C. R. Masson, Chemical Metallurgy of Iron and Steel, Iron Steel Inst. London. Publ.
1973, p. 3.
213. P. C. Hess, Geochim, Cosmochim, Acta. 35 (3),289 (1971).
214. D. R. Gaskell. Metall. J., 19, 14 (1969).
215. G. Rietz and H. Kopp, Z. Anorg, Allg. Chem., 382 (1), 31 (1971); 384 (1),19 (1971).


216. W. Von Bonin. Ger. Offen Pat. 2,227,640 (Bayer A.G.), 1973.





Polymerization of Silica
From the time of Graham (1), who made an intensive study of sols and gels, many
attempts have been made to explain the behavior of silicic acid. When freshly made
by acidifying a soluble silicate or hydrolyzing the ester, silicic acid is not "colloidal,"
since it diffuses easily through parchment or animal membranes and has a molecular
weight by freezing point depression corresponding to monomer. Soon the molecular
units become larger and pass through membranes only slowly and then not at all (2).
This could be because the monomer or other small primary particles form
aggregates, or because the individual particles increase in size and decrease in
Freundlich appeared to recognize these alternatives when he wrote:
Whether it is rather a matter of polysilicic acids. which give larger micellae. being formed
from simple silicic acid. or whether the crystalloid particles originally present already consist of
polysilicic acids. but are exceedingly fine amicrons which continually increase in size-cannot
yet be said with certainty.

In his terminology, a "micella" is a colloidal particle in which foreign substances

(ions, water) are present in its structure, that is. a porous aggregate. whereas the
"arnicron" is a discrete particle too small to be seen with the ultramicroscope. He
recognized that such particles in a colloidal solution could "consist of one very large
molecule," in other words. a single unit. not an aggregate.
Because the most obvious behavior of a silicic acid solution is that it increases in
viscosity and finally forms a gel. its polymerization was generally assumed to be an
aggregation process or a polymerization by which smaller molecular units linked
together into larger ones. The nucleation and growth of discrete particles prior to the
stage where aggregation begins have not been recognized by many workers. who
held to the idea that Si(OH). polymerized into siloxane chains which. then were
branched and cross-linked as in many organic polymers. Even now attempts are still
made to apply the idea of monomer functionality and condensation polymerization
theory of organic chemistry to the silica system. In fact, there is no relation or
analogy between silicic acid polymerized in an aqueous system and condensationtype organic polymers. [However. the behavior of silica in molten glasses is quite a



Polymerization of Silica

Is, many
11y made
illoidal, "
at all (2).
les form
crease In

consist of


ereas the
cope. He
(ery large
creases in
; to be an
its linked
'ior to the
cers, who
hen were
.s are still
elation or
lens-vionis -, : a


different matter, and in that system the conventional polymer theory has been shown
by Masson (3) to be applicable.]
In 1925, Kruyt and Postma (4a) pointed out that there are two groups of silicic
acid sols. The first group has a pH of 4.5 or less, and the viscosity of the sol
increases with time. On the other hand, pure silica sols, having a pH of 7 or higher,
are relatively stable, the viscosity either remaining the same or decreasing with time.
This difference in behavior is explained as follows. The more alkaline sols bear a
negative charge and are thereby stabilized. However, the addition of soluble salts
lowers the charge of the particles and causes gelation or flocculation. On the acid
side, where there is essentially no charge, aggregation or flocculation occurs, causing
an increase in viscosity, and eventually gelation. Tourky (4b) also discussed the
structural differences between silicic acids in acidic and basic solutions; in acidic
solutions, fibrillar or network structures arise through the formation of oxygen
bridges between silicic acid units.
It was Carmen (4c) who first clearly stated that silicic acid polymerizes to discrete
particles which then aggregate into chains and networks.
The formation of silica gel can be regarded as taking place in two stages. In the first, initially
formed Si(OH). condenses to form colloidal particles. In dilute solution, a further slow
increase in particle size is the'only subsequent change, but at a concentration of about 1 percent
silica, these primary particles are able to condense together to give a very open but continuous
structure, extending throughout the medium, thus bestowing a certain degree.of rigidity upon
it. In both stages of polymerization, the mechanism is the same, that is, condensation to form
Si-O-Si links, but in the first stage, condensation leads to particles of massive silica, while in
the second, since it is not possible to fit two particles accurately together overa common face.
the number of Si-O-Si linkages between particles is fewer in number than those within the
particles themselves. They are merely sufficient to bind adjacent particles together, in a fixed
position relative to one another, and thereby lead to a rigid, highly porous. tangled network of
branching chains....
Thus three stages are actually recognized:
1. Polymerization of monomer to form particles.

2. Growth of particles.

3. Linking of particles together into branched chains, then networks, finally extending throughout the liquid medium, thickening it to a gel.

Since Carmen published this in 1940. further experimental data continue to confirm his point of view. There is general agreement that polymerization, that is, the
reactions that result in an increase in molecular weight of the silica. involves the
condensation of silanol groups:

+ HOSi-


+ H20

The term "polymerization" is used in its broadest sense, the mutual condensation of
Si(OH) to give molecularly coherent units of increasing size, whether these are

Polymerization of Silica


spherical particles of increasing diameter or aggregates of an increasing numb-er of

constituent particles. Formation and growth of spherical particles is one kind of
polymerization that takes place under certain conditions. Aggregation of particles to
form viscous sols and gels is another kind of polymerization occurring under other
conditions. Both types of polymerization may occur at once.


The general theory of polymerization is first outlined. Then the details of each step
are reviewed and finally the more recent work of a number of investigators is discussed. Succeeding steps in polymerization from monomer to large particles and gels
or powders have been represented schematically by IIer (5) as in Figure 3.1. This
applies to aqueous systems, in which silica is somewhat soluble. Very little is known
about the polymerization when Si(OH)~ is formed in nonaqueous solutions.
The individual steps are as follows. Later each step is considered in detail in the
light of individual investigations.
(a) Monosilicic acid is soluble and stable in water at 25C for long periods of
time if the concentration is less than about 100 ppm as Si0 2 When a solution of





pH< 7



pH 7-10 WITH




pH 7-10 WITH



c$ ~30nm






Figure 3.1. Polymerization behavior of silica. In basic solution (B) particles in sol grow in size
with decrease in numbers; in acid solution or in presence of flocculating salts (A). particles
aggregate into three-dimensional networks and form gels.


General Theory of Polymerization





monomer, Si(OH)., is formed at a concentration greater than about 100-200 ppm as

Si0 2 , that is, greater than the solubility of the solid phase of amorphous silica, and
in the absence of solid phase on which the soluble silica might be deposited, then the
monomer polymerizes by condensation to form dimer and higher molecular weight
species of silicic acid.

(b) The condensation polymerization involves an ionic mechanism. Above pH 2

the rate is proportional to the concentration of 0 H - ion and below 2 to the H + ion .
(c) Silicic acid has a strong tendency to polymerize in such a way that in the
polymer there is a maximum of siloxane (Si-O-Si) bonds and a minimum of
uncondensed SiOH groups. Thus at the earliest stage of polymerization, condensation quickly leads to ring structures, for example, the cyclic tetramer, followed by
addition of monomer to these and linking together of the cyclic polymers to larger
three-dimensional molecules. These condense internally to the most compact state
with SiOH groups remaining on the outside.
(d) The resulting spherical units are, in effect, the nuclei that develop into larger
particles. The solubility of these very small particles depends on the particle size,
that is, the radius of curvature of the surface. It also depends on the completeness of
the dehydration of the internal solid phase. If the latter is formed at ordinary
temperature it may contain uncondensed OH groups but if formed above 80C, and
especially above pH 7, it is almost anhydrous.
(e) Because small particles are more soluble than larger ones (Chapter 1) and
since not all the small three-dimensional particles are the same size, the particles
grow in average size and diminish in numbers as the smaller ones dissolve and the
silica is deposited upon the larger ones (Ostwald ripening). However, the higher solubility of smaller particles is pronounced only when the particle size is smaller than
about 5 nm and very pronounced when it is less than 3 nm. Hence above pH 7,
where the rate of dissolution and deposition of silica is .high, particle growth
continues at ordinary temperature until the particles are 5-10 nm in diameter, after
which growth is slow. However, at low pH, where the rate of polymerization and
de polymerization is slower, particle growth becomes negligible after a size of 2-4 nm
is reached. At higher temperatures, growth continues to larger sizes, especially above
pH 7. .
The very early formation of particles was also proposed by Vysotskii et al. (6),
who studied the early stages of polymerization and similarly stated that there are
two basic processes of particle growth of silica in the aqueous system:
I. Growth of particles .at the expense of silicic acid in the solution from the moment
of its preparation.

2. Further growth of larger particles by deposition of silicic acid dissolving from the
smaller particles. This is a slower process and may be negligible at low pH after
the monomer has been used up.
(I) Above pH 6 or 7, and up to 10.5, where silica begins to dissolve as silicate, the

silica particles are negatively charged and repel each other. Therefore they do not
collide, so that particle growth continues without aggregation. However, if salt is





Polymerization of Silica


present at a concentration greater than 0.2-0.3 N, as when sodium silicate is neutralized with acid, the charge repulsion is reduced and aggregation and gelling occur.
It is paradoxical that under some conditions, precipitation or gelling is prevented
by raising the temperature. In this pH range a sol of 2-3 % silica with a borderline
salt concentration of 0.2-0.3 IV gels if aged at ordinary temperature. However, if the
sol is first heated to 80-100 0 the particles grow in size and decrease in number so
that aggregation and gelling are greatly retarded or even prevented permanently.
(g) At low pH the silica particles bear very little ionic charge and thus can collide
and aggregate into chains and then gel networks. If the concentration of Si0 2 is
m ore than 1% such aggregation may beginas soon as the first small particles are
formed. However, at lower concentrations and at pH around 2, the monomer is
converted largely to discrete particles before they begin to aggregate. On the other
hand, at pH 5-6, monomer is converted rapidly to particles which simultaneously
aggregate and gel so that it is not possible to separate two processes. The rate of
aggregation increases rapidly with concentration so that in any case above 1% silica,
aggregation probably involves not only particles but also oligomers.
The process of aggregation and gelling in the silica system is unique because,
unlike other metal oxides, the solid phase remains completely amorphous and appreciably soluble in water and is generally in solubility equilibrium with the monomer.
It is essential to understand that while sol is being converted to gel. the growing
aggregates contain the same concentration of silica and water as in the surrounding
sol regions. These aggregates or "gel phase" cannot be seen because the density and
refractive index of the gel phase are the same as those of the remaining sol. Thus
before the sol solidifies only a slow increase in viscosity can be noted, with little
change in other properties, up to a point where the viscosity begins to increase
rapidly and solidification occurs at the "gel point." The most common way of determining the "gel point" is .to observe when the meniscus of-a sol in a container no
longer remains horizontal when the container is tilted.
It may be difficult to visualize how particles in a suspension can rearrange
themselves into three-dimensional networks without changing the silica concentration. In certain microscopic regions in the sol the particles arrange themselves in
chains, and these in turn branch and form networks. These can be isolated in sols
gelling at low pH by adding an inert miscible fluid such as water or alcohol. With a
twofold dilution, all the particles not attached to a three-dimensional network move
apart, but the rigid networks retain their structure and thus are more dense than the
medium, and so can be separated by centrifuging. In this way the percentage of
silica that has been converted to "gel phase" can be measured. Gelling occurs when
about half of the silica has entered the gel phase, which can be thought of as
spherical solidified regions in suspension which cause a rapid increase in viscosity
when the "volume fraction" reaches about 0.5.
After the gel network has been formed, the structure becomes stronger as the
necks between particles become thicker owing to solution and deposition of silica.
In this chapter, attention is concentrated on the details of the various processes of
polymerization Lip to the point of gel formation. The completion of gel formation
and subsequent changes in structure arc dealt with in Chapter 6.


Monosilicic Acid


Overall Effect of pH on Gelling

Whether polysilicic acids or larger particles of colloidal silica are involved, the
general effects of pH are generally as indicated schematically in Figure 3.2. Curve
ABC represents the behavior of silica in the absence of salts. Sols have a maximum
temporary stability with longest gel time around pH 1.5-3, and a minimum stability
with rapid gelling around pH 5-6. Above about pH 7, no gel is formed since the
particles are charged and only particle growth occurs. Curve DEF represents the
general behavior when an electrolyte such as NaCI or Na 2SO. is present at a
concentration above about 0.2-0.3 N. The salt lowers the ionic charge on particles.
At low pH both sols gel, and salt has little effect. In the neutral region the pH of
minimum stability is higher (E) when salt is present.

Monomeric silicic acid, Si(OH). has never been isolated. It is a very weak acid and
exists only in dilute aqueous solution, since it polymerizes when it is concentrated. It



Figure 3.2. The effect or pH on the gelling of silica sols. Curves .-I-C-sols in the absence of"
sodium salts: D-I-": in thc presence of sodium salts.






Polymerization of Silica

is the soluble form of silica that is in equilibrium with the solid phases. As described
in Chapter 1, it is a neutral highly hydrophilic, essentially nonionized substance that
cannot be isolated from water. In pure form, if it could be prevented from polyme.
rizing, it might be expected to be a clear liquid resembling glycerin. The basis for
this speculation is that a very low molecular weight polysilicic acid was isolated as a
clear viscous anhydrous liquid by Robinson (7), who found that it polymerized
immediately to clear, hard silica gel when exposed to a trace of atmospheric moisture. It was highly hygroscopic, and soluble in polar organic solvents such as
'alcohol, but insoluble in hydrocarbons. In view of its unique character, a brief
description of its preparation is justified.
To a violently agitated mixture of 83.2 grams tetraethylene glycol dimethyl ether,
52 ml H 2 0 , and 40 ml 20% H 2S O ., was added, in a thin stream, 125 ml of a solution
of sodium silicate containing 122 g 1-1 Si0 2 and 38 g 1-1 Na 20. After IO min, 50 g
anhydrous Na 2SO. was added and the mixture stirred, then allowed to stand. The
supernatant clear liquid layer of the ether contained silicic acid weighing 186 grams.
This was separated and at once subjected to vacuum distillation at ordinary
temperature to remove the water. Then the high-boiling ether was extracted with an
equal volume of benzene, leaving a viscous, water-clear, anhydrous polysilicic acid.
Sufficient H 2 SO. was present to make the liquid acidic, which gave it temporary
stability. It soon set to a clear hard gel, especiaIly when exposed to moisture or
Monomer can be removed from solution by strong-base ion-exchange resin, presumably because it is ionized to HSi0 3 - by the OH - ions at the resin surface, and is
then adsorbed.

The various methods of preparing monosilicic acid may be summarized as follows.
A saturated solution of monosilicic acid, Si(OH)., containing about 0.01 ~ Si0 2 , is
obtained when pure amorphous silica is equilibrated with water at room temperature. A more concentrated (supersaturated) solution can be obtained only indirectly
by liberating monosilicic acid from its compounds under carefully controlled conditions; at low temperature and low pH. dilute solutions remain supersaturated with
respect to amorphous silica for appreciable periods. For example, at pH 3 and OC,
solutions of monosilicic acid up to 0.1 M (0.6% Si0 2 ) can be prepared by spontaneous hydrolysis of monomeric silicon compounds, sich as silicon tetrachloride or
methyl orthosilicate, and also by reacting monomeric silicates. such as sodium or
magnesium orthosilicates or hydrated crystalline sodium metasilicatc, with dilute

Dissolving Silica
Jander and Heuk eshoven (8) reported that amorphous silica gel gives a true solution
of silica in water; later Jander and Jahr (9) found that the silica in solution had a diffusion coefficient of 0.53, indicating a molecular size about equivalent to Si(OH).



Monosilicic Acid


Subsequently Alexander, Heston, and Iler (10) verified that amorphous ~ilica
exhibits a definite equilibrium solubility in water, amounting to about 0.01-0.012%
Si0 2 in the saturated solution. Kitahara and Oshimo (II, 12) dissolved quartz under
high pressure at 400C to obtain a solution containing 350-400 ppm which was
quenched to obtain a solution of monomer at pH 6 which then polymerized. A 400
ppm solution of monomer was obtained by saturating water with silica gel at
95-100C (13). Egorova (14) reports that in the pH range 1.2-3.7 it remains
unpolymerized even after 2 he.
This appears to be a convenient way to make and store a 400 ppm solutiori of
monomer for polymerization or deposition studies at 25C, where the solubility of
gel is around 100-120 ppm.
Hydrolysis oj Monomeric Silicon Compounds

Dilute solutions of monomer can be obtained by hydrolysis of halides, esters, or acyl

derivatives such as silicon tetraacetate: SiCI. requires the later removal of HCl. Thus
Willstatter, Kraut, and Lobinger (15) led SiCI. vapor into water atOC while adding
silver oxide to maintain pH 3 and to precipitate the chloride. This method was also
investigated by Gruner and Eldd (16). The very rapid hydrolysis of tetraacetate gave
monosilicic acid, whereas (CH 3SOOhSiOSi(CH3COOh gave disilicic acid according
to Schott and Fischer (17). Using a cryoscopic method, they demonstrated 'that in
this system at 25C Si(OH). was most stable at pH 2.8 and the disilicic acid at 3.1.
The relative rates of polymerization of monomer and dimer, each at its most stable
pH, were compared at unspecified concentrations. The monomer was said to be
more stable than the dimer but this probably depends on the pH and the method of
Ethyl orthosilicate has been -used to prepare silicic acid that, is not com pletely
monomeric because the two-phase hydrolysis is not instantaneous. Disilicic acid was
prepared from tetraethyl orthosilicate by the Brintzingers (18).
Methyl orthosilicate hydrolyzes rapidly t~ monomer. Brintzinger and Troemer
(19) obtained essentially monosilicic acid by hydrolyzing methyl orthosilicate in
0.001-0.01 Nl HC/. Weitz, Franck, and Schuchard (20) prepared monosilicic acid by
hydrolyzing tetrarnethyl silicate in 0.002 Nl HCI at room temperature, checking the
molecular weight by the freezing-point method. The methyl ester gives monomer
when hydrolyzed in about 10 sec in water at pH 3. Thus monosilicic acid can be
obtained as a 0.13 M solution (0.8% Si0 2 ) by hydrolyzing methyl orthosilicate in
\0-3 IV H2SO. or HClsolution at 25C, and it remains practically unchanged for
2-3 hr as described by Schwarz and Knau ff (21). They hydrolyzed the ester in an
apparatus from which the methanol and some water could be vacuum distilled and
the molecular weight of the silicic acid was determined by the freezing-point method.
After 24 hr. the molecular weight corresponded to that of dirner. but there was no
indication that it was more stable at this point.
At silica concentrations around 0.5-1 % Si0 2 it is impossible to preserve the
monomeric state at 25C, even at the pH of optimum stability. Five minutes after
the preparation of a 0.083 M solution of Si(Ol-!). (0.5% 5iO)2) at pH 2. some
disilicic acid is present (22).



Polymerization of Silica

Dissolving Monomeric Silicates in Acid




Monomeric crystalline silicates dissolve and are neutralized to liberate monosilicic

acid at about pH 2. Kraut (23) prepared monosilicic acid by dissolving sodium metasilicate hexahydrate in various acidic solutions at low temperature. He reported that
monosilicic acid is most stable at around pH 2-3. Weitz, Franck, and Schuchard
(20) demonstrated that when Na 2Si0 39H 20 was reacted with acetic acid it liberated
Si(OH). Also olivine (magnesium orthosilicate, Mg 2SiO.) dissolved in 1.0 N J:lCI to
give a practically 100% yield of monosilicic acid, the solution containing 0.04%
Si0 2 Thus monosilicic acid may be liberated from silicates which contain SiO.2 ions separated by cations, such as are present in anhydrous orthosilicates.
Alexander (24a) found that sodium metasilicate hydrolyzed to disilicate when
dissolved in water unless NaOH is added to form orthosilicate, Na.SiO . However,
if crystalline Na 2SiO.9H 20 and strong-acid H+ ion-exchange resin are added
simultaneously to a 10- 3 N solution of H 2SO. at OC to maintain pH 3, a 0.1 I'd
solution of monomer can be obtained.
Thilo, Wicker, and Stadt (24b) made monomer by dissolving Na.SiO. glass in
cold dilute acid. With glass of composition I Si0 2 : 1.5 Na 20, disilicic acid was
obtained. Contrary to Schott and Fischer (17), they found disilicic acid is much
more stable than the monomer and can be made reproducibly at higher concentrations.
Coudurier, Baudru, and Donnet (25) modified Alexander's method by maintaining
the pH at 2.5, and Okkerse (29) preferred 0.01 N HCl at pH 2 as the reaction
Silicic acid was prepared by Funk (26) by dissolving certain mineral orthosilicates
("monosilicates" in German) in a solution of HCI in anhydrous methanol. The calcium, barium, and magnesium chlorides were soluble in methanol and the resulting
monosilicic acid was much more stable than in water. Soluble minerals were beta
and gamma Ca-SrO, (dicalcium silicate); Ca 2 0 H ( H O S i 0 3 ) (dicalcium silicate alpha
hydrate); Ca 3 (HOSi0 3 ) 2 ' 2H 20 (synthetic aufwillite); BaOH(H 3SiO.) 4H 20;
Mg 2SiO. (synthetic forsterite); and Mg 3AI 2(SiO.)3 (garnet). The monosilicic acid
polymerizes and is precipitated when water is added. It would seem likely that it is
present at least partially as a methyl ester. Funk and Frydrych (27) prepared solutions in which 90% of the dissolved silica was Si(OH). at concentrations up to 1%.
by dissolving anhydrous Ca 2SiO. in dry methanolic HC!. Polymerization was very
rapid unless the solution was diluted at once to 0,1 % Si0 2 By adding acetone, the
CaCI 2 was precipitated, leaving a relatively pure solution of silicic acid (28).

Characteristics of Silicic Acid


Since Si(OH). has never been isolated or even obtained in a concentrated solution
without considerable polymerization, very little is known about its physical or
chemical properties. Most measurements have therefore been made in very dilute
solutions .





Monosilicic Acid


Diffusion Constant

This was measured in seawater by Wollast and Garrels (30) and found to be 1.0
0.05 X 10- 5 em? sec:".


Ionization Constants






The ionization constant of monosilicic acid has been evaluated in many ways. As
already discussed in Chapters 1 and 2, the pK a appears to be about 9.8 at i5C
determined by Roller and Ervin (31) in a system involving calcium oxide, silica, and
water. More recently, careful measurements by Marsh, Klein, and Vermeulen (32)
of the equilibrium between Si(OH). and HSi(OHh - over a range of pH led to a
value of pKa = 9.9.
Measurements on a system containing extremely pure silica by Schwartz and
Muller (33) give a still more precise value of pKa = 9.91 0.04. They hydrolyzed
extremely pure methyl orthosilicate in water in a system which rigorously excluded
atmospheric impurities, to obtain solutions containing from 12.4 to 155 ppm Si02.
The conductivity and pH were measured with precision at 25C using low frequency
alternating current. From the initial conductivities the value of the acidity constant
was calculated.
In 0.5 M NaCIO. solution Bilinski and Ingri (34a) found that rnonosilicic acid,
Si(OH)., had a first dissociation constant at 25C corresponding to pK a = 9.46
The values of the ionization constants according to Scherban (34b) are as follows:


[H+] [H3SiOi]







[H+] [HSi0 3-]







[H 3SiOi]


2 x 10- 10


[H+] [SiO-]


2 x 10- 12
2 x 10- 12
2 x 10- 12

The increase in the first ionization constant of monosilicic acid with temperature
was measured by Seward (34c). The data were obtained in the presence of borax buffer at 0.1-0.6 M concentrations. The value of pK 1 ranged from 8.88 0.15 at 130C
to 10.0 0.2 at 350C.
The ionization behavior of Si(OH). and the formation of polysilicate ions in 1-5
M NaCI solutions at silica concentrations of 0.005-0.05 m have been measured with
precision by Busey and Mesmer (34d). They found negligible formation of any complexes between monomer and the sodium ion to form NaO(OH)J in solution. (The
cornplcxing behavior of polysilicic acid is of course quite different.)


Polymerization of Silica

Increase in Ionization Constant with Polymerization

The ionization constants of disilicicic and polysilicic acids, colloidal silicas, and gels
are pertinent to the polymerization of monomer and so are considered here.
The increasing acidity of silicic acid upon polymerization was reported by Belyakov et a1. (35). As the monomer polymerized. the pK a was determined by titration
and the degree of polymerization by the cryoscopic method. The maximum pK a was
reported as 10.7 for H 2 S i2 0 1 but then decreases to 6.5 for high polymers. However,
it is not known whether the decrease occurs upon the formation of ring compounds
with =Si(OHh groups or when three-dimensional particles with =SiOH groups on
their surfaces have been formed.
The most doubtful point is the reported pK" of 10.7 for disilicic acid. which would
mean it is a weaker acid than monosilicic acid, of which the pK" is 9.8. Analogy
with other inorganic acids would suggest that disilicic acid should be a stronger acid
than monomer. Unfortunately the dimer is difficult if not impossible to prepare and
keep in sufficiently pure state for strength measurements, although a solution in
which probably at least 50% of the silica was dimeric was prepared by Coudurier,
Baudru, and Donnet (36).
Another indication that polysilicic acid is a stronger acid than Si(O H). is
furnished by ion-exchange studies by Strazhesko and others (37, 38). The ionization
constants of acid centers on the polysilicic acid surface are at least two or three
orders of magnitude higher than the constant of monomeric Si(OH). Dugger et al.
(39a) estimated the acidity of the silanol groups on silica by measuring the ion
exchange of H + with 20 metal ions. By this means they showed that the first
hydrogen to leave the pure silanol surface must have a dissociation constant k" of
10-'-10- 5 This is much more acidic than 5i(OH) which has a k" of 10- 9. 8
The variation of pKa of the silica surface with degree of neutralization was found
to be as shown in Figure 3.3 by Strazhesk o et al. (39b), who carried out studies of
the mechanism of ion exchange on silica gel using Na ", Ca ", Cs", Ca 2 + , 5r2+, and
BaH, and also on gel in the form of divalent and trivalent metal salts.
Duffy and Ingram (39c) have been able to estimate the ionization constants of a
wide variety of acids from the electro-negativity of the constituent elements and also
from the Lewis basicity or "optical basicity" from spectroscopic data. Although this
has failed when applied to monosilicic acid it might give useful data for polysilicic
acids (39d).
Allen, Matijevic, and Meites (3ge) also developed an equation relating surface
change on particles of colloidal silica and pH which indicated that the pK" for the
surface entirely in the hydrogen form is 6.4 and completely in the sodium form is

The acidity of silanol groups on the surface of silica or polysilicic acid has been
examined by Schindler and Kamber (40), who calculated the intrinsic acidity
constant, K l n l , from results of titrating silica gel at 25C in 0.\ M NaCIO~ solution:


log K i n l





Monosilicic Acid
u 10.0






















Figure 3.3 Relation between pK a of silanol groups on the surface of amorphous silica and the
degree of neutralization according to Strazhesko (39b).

With regard to the equation

+ H+



log K

log [H+]

+ log

I - a

where a is the degree of neutralization or the fraction of silanol groups that are
ionized at a given pH.
The intrinsic acidity constant K i n t is defined as equal to K when a approaches
zero. From the data they developed the following equation:

log K

where log KIM

= -

log K i n t

+ a 2)
. 0.039 + a

6.81. Then

+ a2 )
- log
0.039 + a
I - a


log [H +]

log K l n t

To obtain this equation it was necessary to take into account the silicate species in
solution at equilibrium. The following ionization constants for these species in 0.5 M
NaClO. solution at 25C were reported by Bilinski and Ingri (34a).
K, = Si(OH). = 10- 2 . 7

- ~,~<,,:

.. . . . :;..
::t- x,

Polymerization of Silica





This assumes the solubility of silica gel is 120 ppm.


[H+] [(HOhSiO-]

10- 9 . 46




. [(HO)3 SiO-J

[H+]2 [(HO)6 Si.

10- 12.5 7


[H+J [OH-]


10- 13 . 8




The total exchange capacity, C, defined as the maximum number of ionizable groups
under the conditions of titration, was found to be 2.43 OH groups nm -z in 0.1 N
N aCI0 4 solution. As a matter of interest, the authors report 3.4 OH nm -2 in I M
NaCIO. and 5.83 in 3 M NaCIO. solution.
This approach to the acidity of the silica surface is quite different from that of
Yates (41, 5). which was developed as a modification of the conventional equation
applicable to organic polyacids: .

~ pK - k, log" aN -

k, log"

~~ - ~

where R is the ratio of the molar concentrations CSlo,/CNa:o (where Na-O is the titratable alkali), A is the specific surface area of silica in square meters per gram, and
a andN are the activity and normality of sodium salt in the system. Based on Bolt's
data (42), the constants were found to be pK = 12.08, k l = 0.74, k z ;; 3.47, and k 3
= 2430.
In this formula there is no assumption made as to the fraction of silanol groups
that are ionizable, but the number of charges per unit area can be calculated from
C N a : O and A. As an example, a point was taken on Schindler and Kamber's curve
for silica gel of specific surface area of 372 rn" g-I. At a = 0.1, log K was -8.3.
From their equations a pH of 7.35 was calculated.
Then taking the Yates equation with pH = 7.35 and A = 372, a value of R was
found to be 240 Si0 2 : Na-O. Thus for 60 x 240 grams of Si0 2 there are present 2 X
6 X 1023 Na" counterions and ionized OH groups. Then the number of ionized sites
per square nanometer is


60 x 240

372 x


Schindler assumed there are 2.43 ionizable OH groups nm -2 so a "" 0.22/2.43 =

0.09. This is very close to the value of 0.10 for the point originally selected. In other
words, the two approaches appear to lead to similar results.

:~.I ; -~










Monosilicic Acid


The acid dissociation constant of the OH groups on polymeric silica has also been
shown to be about 10- 7 . 1 by an entirely different method. Hair and Hertl (43)
measured the frequency shift of the infrared absorption band of phenolic hydroxyl
groups when adsorbed on the surface of silica and compared these with the shifts of
phenol in the presence of alcohols of known acidity constants. Marshall et a!. (44)
concluded from a similar study that the pKa of some SiOH groups on the silica surface may be about 7.2.
Isoelectric Point

The isoelectric point (iep) of Si(OH), in solution in the absence of colloid or solid
phase has apparently not been measured but presumably it would be between pKa
and pKo, where these are the negative logarithms of the equilibrium constants for:





(HO)3Si"" + OH-

+ H+

The latter equation might also be written

Most measurements have involved solutions in which both Si(OH), and polymeric
silica or colloid or solid phase were present. An exception is the case where the
initial rate of polymerization of Si(OH), has been measured at different pH values.
Here the initial step is one of the following:

+ -OSi(OH)3 = (HOhSiOSi(OH)3 + OH-


+ +Si(OH)3 = (HO)3SiOSi(OH)3 + H +

. Presumably, then, the pH at which monomer reacts most slowly with itself to
form dimer might correspond to the iep of Si(OH). Okkerse (29) measured the rate
of disappearance of molybdate-reactive silica from solution and found it to be at a
minimum between pH 2 and 3. In a study by Goto (45) on the rate of disappearance
of monomer from a solution of 2400 ppm Si0 2 at 25C, a minimum at pH 2.0-2.2
was found.
CATlO;';IC SILlCO;';. The existence of a cationic form of monomeric silica is of
course implied in the assumption that Si(OH). has an iep, Colloidal particles of
silica have been shown to carry a positive charge at low pH. but direct proof that
silicon can exist as a cation has not been available. It is therefore interesting that in
very dilute solution (66 ppm) monomeric silica has been shown to react with HCI to
form the ion (H 20 3Si(OH)3+C!- according to Cherkinsk i and Knyaz'k ova (46). This
was determined by the difference in precise conductivity measurements of 0.0025 N


Polymerization of Silica

NaOH solution titrated with 0.005 N HCI, with and without the presence 'of 0.0011
M Si(OH). The chloride compound exists only in very dilute solution.

Point 01 Zero Charge

The point of zero charge (pzc) where the surface charge is zero and the isoelectric
point where the electrical mobility of silica particles is zero have been measured by
many methods. De Bassetti, Tschadek, and Helmy (47) measured the pzc for silica
gel by a calorimetric method, from which they concluded the value must be 'between
2.5 and 3. However, the data may not preclude a value as low as pH 2 since the heat
of neutralization becomes exceedingly small below pH 3.
In an extensive study of silica polymerization, De Boer, Linsen, and Okkerse (48)
found the iep to be between pH I and 1.5, and that condensation was slowest there,
as shown by several means including viscosity studies.
Vysotskii and Strazhesko (49) have pointed out that there has been relatively little
attention paid to the pzc or iep of silica in spite of the fact that in other colloid
systems they are key factors. These authors recalled the observation of Freundlich
(2) that whereas lyophobic colloids are least stable at the iep, the lyophilic colloid,
silica, appeared to be the most stable. This is not quite true because colloidal silica is
permanently stable when it is negatively charged at pH 9-10, but there is, as Freundlich recognized, a marked temporary stability maximum at the iep around pH 2
(see Figure 3.2).
Vysotskii and Strazhesko show that in the presence of a given acid such as
sulfuric, the iep is not only the point of minimum rate of gelling but also of syneresis
and is also the point at which gels of maximum strength and maximum specific surface area are obtained. All these characteristics result not only because the rate of
aggregation is at a minimum at the iep, but also because the rate of growth of the
ultimate particles from monomer is at a minimum, so that the ultimate particles are
smallest as they form the gel.
These authors noted the relation between the pH of slowest gelling and the pK a of
the acid used. Their data are plotted in Figures 3.4 and 3.5. Their pH values for
HN0 3, H 2SO., and H 2CrO. are similar to those reported by Iler (50), who also
reported a number of other very strong acids which gave maximum gel times at
about the same pH as for HCl and HN0 3 , for example. NH 2S03H, HCIO., and
CH(S03H)3' However, the point was not brought out that with weaker acids such as
acetic, although the sol may be most stable at pH 3.5, it is nevertheless far less stable than the sols made with stronger acids at pH 1.5-2.0.
Similar results were reported by Tai and Kiang (51). hydrochloric. sulfuric, and
nitric acid giving a maximum gel time at pH 2, phosphoric at 2.5, and acetic at 4.0.
It was proposed that the polymerization rate is. proportional to

The iep appears to be at about pH 1.5, according to ion-exchange studies by

Vysotskii and Strazhesko (52) and Kirichenk o and Vysotskii (53) of ion-exchange














ed by
e heat


e (48)
lica is


Figure 3.4. Effect of pH on gel time at 25C of silicic acid sols of different concentrations
made from H2S0~ and sodium silicate: Curve 1: 1.09 M. 65.4 gl-I Si0 2. Curve 2: 1.33 M. 79.8
gl " SiOz. Curve 3: 1.78 M, 106.8 gl-I SiOz. [From Vysotsk ii and Strazhesko ('+9).]


pH 2
.ch as


ic sur'ate of

of the
!es are


JKa of
.es for

o also
nes at
'~, and
uch as
ss sta-






c, and
at 4.0.


Figure 3.5. Relation between the pH or slowest gelling rate and PA"'I of the acid used
trulizing the sodium silicate. [From Vysotsk ii and St ruzhcsk o (-1<)).]






Polymerization of Silica


sorption of rubidium ions from 0.1 N RbN0 3 solution on silica gels pretreated at
temperatures up to lOOOe. All curves in Figure 3.6 converge to this pH at zero
adsorption. This general approach is summarized by Klirnentova, Kirichenko, and
Vysotskii (54).
In summary, iep and pzc of silica have been variously reported to be from pH 0.5
to 3.7 according to a review of the literature on this point by Parks (55), who cited
12 references. However, a pH of around 2 0.5 appeared to be an average for
various types of silica ranging from purified ground quartz to colloidal silica. Some
variation may be expected, depending on whether the surface is crystalline or amorphous, possibly on particle size, and especially on the presence of impurities. The
question remains how the iep determined from maximum gel time or minimum rate
of disappearance of monomer relates to the polymerization mechanisms involved.
The relation between the isoelectric point of polysilicic acid and the stability of
sols, rate of gelling, and properties of resulting gels has been summarized by Klimentova, Kirichenko, and Vysotskii (54). This behavior can be summed up by saying that all the phenomena observed involve the formation and hydrolysis of
Si-O-Si bonds, and that the rates of these reactions depend on a catalytic effect
which is at a minimum at pH 1.5-2.0 in the presence of anions of strong acids and
the minimum becomes greater at higher pH in the presence of anions of weaker acids.
From the fact that the rate of disappearance of monomer by polymerization is
second order above pH 2 and third order below 2, Okkerse (29) concluded that an
anionic form of silica was involved above pH 2 and a cationic form below 2. Thus
the isoelectric point must be at pH 2. Similarly, De Boer, Linsen, and Okkerse (56)
considered that the isoelectric point is around pH 2, since the polymerization rate is



~ 90












Figure 3.6. The isoelectric point. Adsorption of

rubidium ions versus pH on silica gels preheated to
various temperatures. Curves 1-5. temperatures 300.
500. 700. 900. and 1000C, respectively. [From
Kirichenk o and Vysotskii (53).]

=-. ..-


Monosilicic Acid


a function of H+ and OH - on each side of this point. In further work (S7) they
found by electrophoresis studies that the iep was between pH 1.0 and 1.5 in a 0.5%
Si0 2 sol and at pH 2 when the sol was diluted to 0.26%. It was also shown that there
was a sharp minimum in the viscosity at pH 1.9.
Similar observations were made by Tai An-Pang (58), who related gel time to the
ionization constant of silica.
The significance of the iep of silica in the silica-water system involving Si(OH).
and polymerized or solid silica surfaces is still not clear, but the preponderance of
evidence suggests that for monomeric Si(OH). the iep may be between pH 2 and 3,
and for polymeric forms between 1.5 and 2.

Stability of Monomeric Silica

As long as the concentration of Si(OH). is below the equilibrium solubility of amorphous silica, usually assumed to be about 120 ppm for silica gel but around 70-80
ppm for vitreous silica, it has been assumed that monomer would remain, as such, in
water solution at 25C. However, such a solution is supersaturated with respect to
quartz and probably to other crystalline species (Chapter 1). There is also possibility
that a solution of monomer at a concentration of 100-150 ppm might nucleate a
particular less soluble polymeric species of lower solubility.
Such a case may be involved in the observations of Schwartz and Muller'(33), who
made a highly purified solution of silicic acid from methyl orthosilicate at concentrations up to 150 ppm. Initially, conductivity measurements indicated that the silica
was monomeric, but after half an hour the conductivity, at all concentrations, slowly
decreased to about half the original value. This happened even though in half the
samples the concentrations were less than the solubility of amorphous silica. It was
assumed that the monomer polymerized slowly at pH 7 to a polymer species that is
smaller than usual colloidal dimensions, since it passed through an ultrafiler, yet it
must be more insoluble than amorphous silica.
Unfortunately, this change was not followed by means of the molybdic acid
method to see whether it involved simple dimerization at pH 7, which might have
escaped the notice of previous investigators. However, if this were the case, and if
disilicic acid has a pK of -10.7 as reported by Belyakov et a!' (35), then the conductivity would have decreased by much more than 50%. (This is discussed later in
further detail.)

Reactions of Monosilicic Acid

In view of the relatively neutral character of Si(OH), with its physical resemblance
to an organic polyol, it is not surprising that at pH 2, where it is not ionized, few if
any interactions with other substances have been observed. Its most obvious reaction
is self-polymerization to higher molecular weight polysilicic acids which are more
reactive. The interaction of polysilicic acids with other substances is considered later
in this chapter. However, there are a few reactions in which Si(OH). may take part.


Polymerization of Silica

These are interactions either with other acids to form anhydrides or with" a few
extremely weakly basic metal cations.

Phosphoric and Boric Acids

Silica has long been known to react with anhydrous H3PO~ but the wide variety of
possible compounds has not been investigated. The reaction is, in effect, a condensation, with water eliminated. For example, by heating amorphous silica with H3PO~
at a molar ratio of 1:2 for a week at 80-180C, silicon phosphate is formed. Excess
H3PO~ is removed with dioxane and the product is dried at IOOC. A 10% solution
can be made in water, giving a 2.7% concentration of silica (59a). Silicon phosphate
has long been known but this example of a water-soluble material is mentioned
because it probably hydrolyzes to Si(OH)~.
The reactions of boric acid with silica appear to parallel those of phosphoric acid
since in dilute solution there appears to be no interaction between the acids, but on
dehydration at high temperature, Si-O-B bonds are formed in the resulting mixedoxide glass. The Si-O-P and Si-O-B bonds are hydrolyzed in aqueous solution.

Sulfuric Acid
The issuance of a series of patents involving silicon salts of sulfuric acid is surprising
since it is unexpected that a reaction product of two of the oldest known chemicals
should have escaped attention for so long. However, the existence of silicon phosphate suggests that the sulfate might also exist. Blount (59b) has disclosed the compound "silicodihydrogen sulfate," SiO(HSO~), which was obtained by dehydrating
"dihydroxy silicon dihydrogen sulfate," (HO)2Si(HSO')2' with concentrated sulfuric
acid. It is claimed that these solid compounds are obtained 'by stirring powdered
N a 2Si0 3 5H 20 for several hours in an excess of concentrated H2S0~. Finally the
sulfate salt is hydrolyzed in water giving a white granular "silico-Iorrnic acid" or
"monosilanol," HSi(O)OH, and "rnonosilandiol." H 2Si(OH h(?). However, no
further information about properties or analysis is given. I f the products exhibited a
characteristic x-ray diffraction pattern or other identifying features their existence as
compounds would be less equivocal. If a crystalline character is retained the compounds might be clathrates with H2S0~ within the lattice or exist as a different crystal
structure, as in the case of the phosphates. On the other hand. if the powders are
amorphous then they may be microporous silica gels with pores filled with acid; if
anhydrous, internal surface groups of =SiOS0 3H may be present.

Iron and Uranium

Monomeric silica does not react with most metal ions in water at low pH where
can exist, since for reaction to occur it is probable that some hydrolysis to a
basic metal ion must first take place.



Monosilicic Acid


However, very few metal ions form basic ions at the pH of 2, where monomeric
Si(OH). is most stable. Iron and uranium are the only ones which have so far been
reported. Monomeric silica reacts with uranyl ion as follows, according to Porter
and Weber (60):

The equilibrium constant for monomer concentrations in the range 0.024-0.035 AI

is 0.01 0.001.
As evidence of some chemical combination, certain forms of natural hydrated
silica gels and also laboratory-prepared gels impregnated with uranyl salts have been
observed by Iler to fluoresce with a strong greenish yellow color under ultraviolet
The other known reaction of monomer with a metal cation is the case of ferric
iron, reported by Weber and Stumm (61) and further examined by Porter and
Weber in regard to the effect of the degree of polymerization of silica. They
polymerized the silica at a concentration of 2280 ppm at pH 9-10 for various lengths
of time, conditions that are known to give very small spherical particles. With:
increasing polymerization of silica with formation of adjacent SiOH groups that can
combine with iron, at pH 2, the number of SiOH groups combined per iron ion
increases from one on the monomer to two or three as the particles become larger,
the radius of curvature larger, and the SiGH groups closer together. The following
equation suggested by the authors does not indicate the degree of polymerization of
silica, but only the number of SiOH groups that can react with Fe3+, liberating the
corresponding number of H + ions:

Following the absorption characteristics of the iron as it complexes, the following

values were obtained:

Mol. Wt.




Diameter (nm)





Qn = equilibrium constant
D.P. = degree of polymerization

In the stock solutions containing 2280 ppm of silica at pH 9-10 there must have
been an appreciable concentration of monomer in equilibrium with the polymer.
Based on the calculated particle sizes this would amount to at least 2.6, 2.2. and 1.7

.' ~:~~~.:


Polymerization of Silica


mM or 156, 132, and 102 ppm as monomer, which undoubtedly also combined with
the iron but was not taken into account.
The interaction of Si(OH)~ with ferric iron is evidenced by the fact that concentrations of 10-.-10- 3 M SiO:: in water catalyze the oxidation of Fe 2+ to Fe H Schenk
(62) has derived a quantitative relation between the rate of oxidation and the
concentration of monomeric silica. Below pH 3.5 a soluble complex between
Si(OH). and Fe3+ exists.
At pH 6-8, a ratio of 3 Si(OH). to 1 Fe3+ prevents precipitation of Fe(OH)3'
However, in the case of AIJ+, a fivefold excess of Si(OH). is required to prevent





It is peculiar that in view of the similarities of AI3+ and Cr3+ in their precipitation
behavior as hydrous oxides these elements are widely different in their interaction
with silica. One reason is that the chromite ion is not formed as easily as the
aluminate ion, AlO:: -. The Cr3+ ion is much larger than AI3+ and cannot fit into the
SiO:: lattice to give stable anions like (SiAlO.) -.
Thus the Cr3+ ions show a peculiar inertness relative to monomeric, silica, in
marked contrast to the behavior of AP+. When amorphous silica was heated under
pressure with a mixture of Cr(OH)3 and AI(OH)3 for 2 days at 300C, only the alumina combined with silica (63). This behavior of chromium probably explains the
rarity of chromium silicate minerals.
Hexavalent chromium as H::CrO. appears to form a complex with Si(OH)~,
according to Iler (50). The chromate ion is unique among inorganic anions in that it
retards the polymerization of Si(OH)~ in the pH range from about 0.5 to 3.0. At pH
1.7, where Si(OH). is most stable, the increase in gel time of a I M silica sol was
linear with Cr03 : SiO:: ratio. It ranged from 69 hr with no Cr0 3 to 270 hr at a
Cr0 3 : SiO:: ratio of 0.75, and at higher ratios was then constant at 270 hr. The latter
is the gel time of a 0.5 M Si0 2sol in the absence of H::CrO.
Thus the system behaved as though H2CrO~ dimerized the silica quantitatively and
the excess had no further effect:



It appears that this "dirner" then gelled at a rate as though the concentration of
silica was only half the original. Unfortunately, the gel obtained was not examined
to see if the CrO~::- ion was actually bound in the structure.
No compound such as silicon chromate has been reported, but evidently silicon
can be linked through oxygen to hexavalent chromium. A chromic acid ester of a
silanol group was made by Schmidt and Schrnidbaur (64), who prepared the tri-



Monosilicic Acid


rnethylsilyl ester:

(CH 3hSiOCrOSi(CH 3 ) 3

, !

As discussed in other chapters relating to the effect of aluminum ions. there is a
peculiar affinity between the oxides of aluminum and silicon. At this point only a
few observations regarding the interaction with monomeric Si(OH). are noted.
Aluminum oxide is far less soluble than silica in water at 25C. pH 5-8. as evidenced by early data by Okura, Goto, and Murai (65), shown in Figure 3.7.
Monomeric silica reacts with AP+ ions and is precipitated most effectively at pH
9, according to Goto (66). Thus with a solution containing initially 35 ppm
monomeric Si0 2 at pH 9. the addition of 20-100 ppm of Al as AP+ ions reduced the
silica concentration to a value C. such that A C = 300. where A and C are ppm of
AP- and Si0 2 respectively. However. this probably did not represent true equilibrium.
Over a long period of time monomeric silica. (SiOH) reacts with AP+ ion at
25C to form colloidal aluminum silicate of the halloysite composition:

By reacting soluble silica and alumina at various pH values for periods up to 4 years
and measuring the concentrations of residual Si(OH). and AP+. Hem et al. (67)









Figure 3.7.

Solubility or aluminum oxide in water versus pII [From Okura. GOlO. and Mur ai


(65) ).






Polymerization of Silica

measured the following constants:

[AP+]Z [Si(OH).]Z
[Al(OH)iP [Si(OH).]Z [H]Z
The standard free energy of the colloidal aluminum silicate was - 897 1 kcal
Monomeric silica is strongly adsorbed onto the surface of hydrous aluminum
oxides. There is a reaction between Si(OH). and crystalline AI(OHh by which
several reaction layers of Si0 2 are built up, with simultaneous decrease in pH of the
suspension (68a). Formation of the first layer is rapid, but the second and third
layers form progressively much more slowly. It would seem that diffusion of AP+ Or
AIO2 - from the surface of the crystal must be involved. with the formation of a
silica-rich aluminosilicate. A relatively low content of aluminum ion in the SjO,
layer greatly reduces its solubility, thus explaining the deposition of Si0 2 from a
solution unsaturated with respect to pure amorphous silica.
Baumann (68b) found that when different amounts of aluminum ion were added
to a solution of monomer (420 ppm Si0 2 ) , more silica remained in the molybdate
reactive state than when no aluminum was present. With no aluminum' present, after
4 days there remained 130 ppm of molybdate-reactive silica as monomer in equilibrium with 290 ppm of relatively inactive high polymer. But when aluminum was
present in the AI: Si atomic ratio of 1: 7, there remained about 200 ppm of molybdate-reactive silica. It can be interpreted that the alumina had combined with silica
to form an aluminosilicate that later was decomposed by the strongly acidic molybdate reagent liberating additional active silica that appeared as monomer.
However, when the silica concentration was only 60 ppm. and thus below the solubility of amorphous silica, no polymerization occurs except when alumina is added.
In this case when the AI: Si ratio is 1: 1 to I: 10 the aluminum ion brings together
monomer to form a silica-rich complex in which some of the silica is also linked
together into a state that is later less molybdate-reactive. Baumann's extensive data
deserve detailed study.
The final reaction product at the alumina surface is halloysite. When a dilute
solution of monomeric silica is brought in contact with gamma alumina, it is
adsorbed at a rate strongly dependent on pH and area A. In a mediu m of constant
ionic strength (0.1 N NaC!) and at silica concentrations of 10- 3 - 10 - < M, Huang
(69a) found the initial adsorption to be rapid. When below pH 9 the rate is proportional to A Z [Si0zl U [H+]-o.5, whereas above pH 9 it is proportional to A 2 [Si0 2 JU
[H +]. Huang proposed that HSi0 3 - is the major reacting species.
It is possible that with a quaternary ammonium base and in the absence of metal
cations, aluminosilicate anions may remain in solution, for example,
(HOhSiOAI(OH)20Si(OH)J1-. Flanigcn (69b) reported that quaternary ammonium
silicate and aluminate remained in solution until a sodium salt was added.
In the case of pure alpha alumina, the writer has found that there is no interaction
with monomeric silica. Colloidal alpha alumina free from other forms of alumina or




Characterization of Silicic Acids


AP+ ions has been prepared by treating the particles with 24% HF solution for 2~ hr
to remove all other types of alumina and silica impurities, and washing with water,
then NH.OH to remove all F- ions from the alumina surface (70a). At pH 7-8,
monomeric silica is not adsorbed, nor does it react with this form of alumina even
though the specific surface area.is 24 m 2 g-t. Presumably AP+ or polybasic Al ions
are required for reaction with Si(OH).

Divalent Cations
It is known that the ligand properties of deprotonated Si(OH). with a polyvalent
metal cation can lead to a stable complex such as [FeOSi(OHh]3+, as reported by
Weber and Stumm (61); however, much less is known about complexes of divalent
cations. Santschi and Schindler (70b) measured the stability of complexes involving
Ca2+ and Mg2+ at around pH 8-9 in 1 M sodium perchlorate solution at 25C.
Complex formation was weak and occurs only in the presence of excess salts. In
natural waters, such complexes are not formed.


It is not possible to discuss all the techniques used for measuring or characterizing
silicic and polysilicic acids and small colloidal particles, but some of the methods,
especially applicable to following the polymerization, are reviewed.
Reaction with Molybdic Acid
The history and use of this reaction in analyzing for silica is discussed in detail in
Chapter 1 and its application in characterizing silicate ions in Chapter 2.
Further refinements and use of this indispensable reaction for studying the
polymerization of silica are now described. Most of these involve following the
course of polymerization by measuring the rate at which the monomer, or the
monomer and dimer, disappears. This in turn involves distinguishing monomer and
dimer, which react rapidly with molybdic acid. from higher polymers that react
more slowly. The method is particularly useful because, as described earlier, the
color-forming reaction is carried out at a 10:" pH where the polymerization or
depolymerization of silica is at a minimum. Hence samples taken from rapidly
polymerizing or depolymerizing solutions at higher or lower pH are "frozen" at the
moment they are added to the molybdic acid reagent. The monomer and dimer react
very quickly, whereas each higher polymeric species dcpolymerizes at a slower,
characteristic rate.
The structure of the silicornolybdic acid is such that within the molecule there is a
tetrahedron of four oxygen atoms in which only one silicon atom can fit (20, 24).
Thus only monosilicic acid. Si(OH). can react directly. All polymeric species must
first dcpolyrncrizc to monomer. The silicomolybdatc anion Sii\10 12 0 .o - apparently
has a compact structure similar to that established for basic aluminum chloride in


Polymerization of Silica


Figure 3.8. Structure of silicomolybdic acid. All corners of octahedra are occupied by
oxygen atoms. A molybdenum atom is at the center of each octahedron. (see text).

which the polybasic aluminum ion is A 113 0 .(O H h .(H 20 )121 + , as established by
Johansson (71). The analogous structure for the silicomolybdate ion requires that all
the oxygen sites be filled by oxygen atoms (including those that in the aluminum
complex are filled by OH groups and coordinating water molecules): SiMo I2O:o- .
The structure of the silicomolybdate heteropolyion is shown in Figure 3.8. The
details of the sharing of the oxygen atoms (or ions) between the McO, octahedra
and the central SiO. tetrahedron have been described clearly by Cotton and
Wilkinson (72a).

Alpha and Bela Silicic Acids

Polymers of silica were classified first by Goto (45) into two types. "A" reacts
rapidly with molybdic acid and has a low degree of polymerization of less than four,
whereas "B" reacts more slowly with increasing molecular weight. It appears that
the difference is the size of the ultimate particles and thus the reaction ratevaries in
proportion to the specific surface area. Other workers have variously defined the
increasing degree of polymerization, as evidenced by decreasing rate of reaction with
molybdic acid, as alpha, beta, and gamma. Usually alpha is defined as silica that
reacts almost completely in less than 5 min. Beta reacts completely in 10-30 min.



Characterization of Silicic Acids


and has been classed as an oligomer or oligosilicic acid by Baumann (72b)~ Gamma
then is the higher polymers that do not react after 10-30 min; it is often referred to
simply as "higher polymers."
Goto and Okura (72c) proposed that the monomer and dimer species which
reacted in 5 min be classed as type A. These could be removed from solution by a
strong-base anion-exchange resin. They recognized that there are different types of B
type polymer, since those formed in an acid medium depolymerized more rapidly
than those formed in a basic medium.
A method that is said to distinguish alpha and beta from gamma silicic acid was
developed by Nemodruk and Bezrogova (73a), who defined the gamma silicic acid as
that which did not react with molybdic acid reagent at 100C in 20 min, whereas
beta reacted com pletely.

Measurements of Reaction Rates

ied by

ed by
hat all

L The
;"1 and

, four,
's that
ries in
ed the
n with
a that
) r

Anum ber of investigators began to use the procedure developed by Alexander (24a)
to measure the rate of reaction of specific polysilicic acids with molybdic acid. This,
in effect, was a measurement of the rate of depolymerization in the colorimetric
reagent. It was hoped that once the reaction rates of individual polyacids were
known, the more complex reaction rate of a mixture of polymers could be
interpreted as a distribution of molecular weights.
The depolymerization of a particular species of silicic acid is a first-order reaction
so that the species can be characterized by a specific reaction rate constant.
Since in most solutions monomer is already present along with a higher polymer
or colloid, the following equations will hold:
. Cr



fraction of total silica reacted with silico-rnolybdate at time t

total silica in the system at t = 0
total monomer in system at t = 0
total polymer in system at t = 0


where k ; and k p are the reaction velocity constants for monomer and polymer. Taking a hypothetical case where 735 ppm of monomer is in equilibrium with 7265 ppm
of cubic octarner, or 9.2% of the silica is monomeric, with the known values k m =
1.5 and k p = 0.45, the color development curves are calculated from the equation. In
Figure 3.9, curve A is the curve that results when all the silica is monomeric. B is the
curve for the above mixture, and C is the curve for higher polymer alone. It will be
noted that the amount of monomer would be difficult to estimate from this plot.
However, by plotting the log of the fraction of silica not yet reacted at time t against
time. as in Figure 3.10, lines are obtained for A and C and. at longer times. also for
the mixture B. The linear part of B extrapolates at zero time to the fraction of
higher polymer (90.8%) in the mixture.




iii 0.5






Figure 3.9. Reaction of silica with molybdic acid. Calculated curves: A. monomeric silica, B,
a mixture of 9.2% monomer and 90.8% cubic octarner; C. cubic octamer alone.


u 0.5






i= 0.1






Figure 3.10. Reaction of silica with molybdic acid: A. monomer: B. 9.1 % monomer and
90.0% cubic octarner: C. cubic octarner alone.

:.... .


Characterization of Silicic Acids

: silica, B,

Because of rather low precision the method is of value only for distinguishing
monomer and very low polymers from relatively high polymers, not for following
the early stages of polymerization. However, Baumann (72b) studied the early stages
by stopping the reaction by adding citric acid and reducing the yellow complex to
the more sensitive molybdenum blue.
Alexander's method (24a) was used by Thilo (73b) and several other investigators
to characterize polysilicic acids by the rates of reaction with molybdic acid, each
having a characteristic reaction velocity constant k. Their procedure, in slightly
modified form for convenience, is given in detail in Chapter 1 as a recommended
In some cases the polysilicic acid acid must be liberated from a crystalline silicate
in acid at 2C, or even in methanol-HCl, to obtain a solution stabilized long
enough to take a sample for the molybdate test. The reaction of molybdic acid with
disilicic or linear trisilicic 'acid is rapid because these depolymerize to monomer
within a few minutes at pH 3. Schwartz and Knauf (21) prepared the pure methyl
esters of these acids and found that by the time they had completely hydrolyzed in
water in 4 and 10 min, respectively, only monomer was present in solution.
The molybdic acid was somewhat modified by Coudurier, Baudru, and Donnet (36)
for their extensive study of polymerization of disilicic acid. Two solutions of molybdic
acid were used containing 4 and 6 g I-I ammonium molybdate, respectively, both at
pH 104. These contain 0.0235 and 0.0352 g-atorns 1-1 molybdenum. When they were
reacted with monomer at 25C the reaction rate constants were 2.1 min -I for the
more dilute and 2.6 min-I for the more concentrated solution. However, with higher
polymers the reaction rates were the same, thus indicating that the slow step is the
depolymerization to monomer:

orner and


) Si(OH).


) silicomolybdic acid

However, disilicic acid also reacted at different rates with the two different
concentrations of molybdic acid, indicating that it dissociates very rapidly to
monomer. Equations were developed on the basis that polymer must first
depolymerize before reaction. Using these equations, experimental data plotted as
logarithm of unreacted silica versus time can be resolved to give the relative proportions of monomer, dimer, and polymer.
The reaction rate of molybdic acid with specific polysilicate anions has been
measured after obtaining a solution of the free polysilicic acid by dissolving waterinsoluble, but acid-soluble, crystalline silicates of known crystal structure. Wieker
(74) applied this method to a number of calcium silicates. Four different types of
silicic acid were characterized by their rates of reaction with molybdic acid, by Funk
and Frydrich (75). However, they did not use the method of Alexander. Instead, the
reagent was more concentrated containing 0.28 g-atorns I-I Mo with a H"':Mo ratio
of 1.5. The high concentration of molybdic acid and the relatively low acidity
accounts for the rapid reaction of this reagent with monomer and also its promotion
of the de polymerization of polymers more than twice as fast as Alexander's reagent
(75). It will even gradually attack quartz. The reaction was followed not

Polymerization of Silica


colorirnetrically, but by precipitating the silicomolybdate as quinoline salt- and titrating the latter with base. Thus the method has the advantage of not requiring a spectrophotometer or colorimeter (for details see Chapter I).
As sources of the silicic acids, crystalline acid-soluble satls of monosilicic,
disilicic, and cyclic trio, tetra-, and hexasilicic acids were dissolved rapidly in rnethanolic HC I, in which the silicic acids are more rapidly dissolved yet are more stable
against further polymerization than in water. The liberated silicic acids were reacted
at once with molybdic acid reagent,at 20e.
For each silicic acid the reaction is first order and the constant is calculated:




where C is the fraction of unreacted silica at time t, and K is the rate constant (C =
LO at t = 0). Then k (sec- 1) = 0.693 (th)-\ where t h = half-life.
Reaction Rate Constants

Values of constants for silicic acids from known crystalline silicates are given in
Table 3.1. It is emphasized that these apply only when Funk and Frydrych's type of
reagent is used. The reaction rate decreases more rapidly than the increase in
number of siloxane bonds that must be hydrolyzed to depolymerize the polysilicic
acid to monomer. This is probably because of the greater stability of the ring structures as compared to corresponding chain polymers.
Since several investigators have used nearly the same molybdic acid reagent solution as used by Alexander (24), a number of values for the constants can be compared for monomer and polymers, excluding those of Funk-and Frydrych, who used
other reaction conditions. Each polysilicic acid in Table 3.2 was prepared from a
particular crystalline silicate known to contain that polysilicate anion, by dissolving
it under conditions that avoided changing the structure.
. The linear polysilicic acids hydrolyze rapidly to monomer according to O'Connor
(77), and the linear pentamer should have a rate constant of 0.66. that is, 90%
reacted in 3.5 min. It appears that in the case of all linear, cyclic, or polycyclic silicic
acids where all siloxane bonds are exposed to the solution, the rate of depo/ymeriza"


Table 3.1. Reaction Rate Constants of Silicic Acids with Funk and
Frydrych's Molybdic Acid Reagent

Starting Silicate

Ca 2Na 2Si 20 ,
K4H 4Si,0I2
ce.st.o 11.6 H 20

Silicic Acid
[(HO)2SiO b





;.....- ..


Characterization of Silicic Acids


Table 3.2.


Reaction Rate Constants of Silicic Acids with Alexander's Molybdic Acid'Reagent



Silicic Acid
Degree of

Type of






Double 4-ring, cubic

r =

n in
ie of
e in

Single chain
Double chain

High mol. wt.

High mol. wt.

K (min-I)


2.05 .

Thilo et al.
Marsh et al.
Hoebbel et al.
Coudurier et al.
Thilo et al.
Hcebbel et al.
Cordurier et al.
Thilo et al.
Hoebbel et al.
Hoebbel et al.
Wieker et al.
Hoebbel et al.
Hoebbel et al.
Hoebbel et al.

.olu- .



tion is so rapid that the rate of reaction with molybdate does not increase very
greatly with the degree of polymerization. However, with Frydrych's faster reacting
reagent, differences in depolymerization rates are more apparent.
Composition of Molybdic Acid Reagents
The compositions of the solution in which the color was actually developed are summarized in Table 3.3. One group of investigators used the Alexander composition
essentially unchanged. Others modified this for specific reasons. Except for the compositions of Funk and Frydrych and of Nemodruk, the reaction rate constants with
the various silicic acids are all about the same. The more concentrated reagents,
such as Iler's, permit the use of higher concentrations of silica. The indicated ratio
H ": Mo is not based on the H + ion concentration in the solution, but is the ratio of
acid to ammonium molybdenum used in making up the mixture. Actually, a ratio of
H ": M 0 of 0.86 is required to neutralize the N H. + ion.
Other Observations

The composition of silicomolybdic acid, determined by Khornchcnko et al. (78), corresponded to H .[SiM OIZO.O] 29H zO.
Complete conversion of silicornolybdic acid from the beta to the less intense
yellow alpha form was observed at a H +: MOO.2- of 1.66: 1.0 according to Mars




.. ". :... """


# . . .,~


Polymerization of Silica


Table 3.3.

Composition of Molybdic Acid Reagents

Thilo et a!.
Marsh et al.
Coudurier et al.
Funk and Frydrych


Mo (g-atorns I-I)



(pH 1.4)





H 2SO.
H 2SO..
H 2SO.
H 2SO.
H 2SO.

(79). It is for this reason that a H ": Mo ratio greater than 4 is generally used to
develop the beta form.
Sugars and other polyhydroxy organic compounds interfere with the reaction of
molybdic acid with monomeric silica. This is believed to be due to the formation of
stable complexes with the molybdic acid (80).
Goto and Okura (81) were the first to recognize that the depolymerization of
silicic acid is catalyzed by the presence of molybdic acid. Thus at pH 1-2 in the
presence of He 1 alone, polysilicic acid formed monomer only very slowly, as shown
by adding molybdic acid after 50 min. The rate of formation of silicomolybdate was
then the same as when molybdate was added at the start. However, it is not known
whether the molybdic acid 'is actually involved as a catalyst by direct interaction
with the polymer or whether it simply reduces the concentration of monomer in
solution to such a low level that an equilibrium between polymer and monomer is
A peculiar phenomenon has been noted by IIer. When a small amount of NaF is
added to a polysilicic acid solution at pH 2 it converts an equivalent amount of the
silica to SiFez-, which, when molybdic acid is then added reacts us though it were
monomer. However, if the same amount of NaF is added with or after the addition
of molybdic acid reagent it does not depolyrnerize an equivalent amount of silica,
but instead acts as a catalyst for the depolymerization of polysilicic acids. When
NaF is added before the molybdic aeid so that it is converted to SiFez-, then when
the latter reacts with molybdic acid, the fluoride ion combines irreversibly with
molybdenum so that is is no longer free in the system. When added later, the
molybdic acid reacts with monomer as it is developed, but does not inactivate the
fluoride, which at the low pH is probably present as HF.

Separation of Silicic Acids

Although the rate of reaction of molybdic acid with individual polysilicic acid
species obtained from crystalline silicates can be measured. the results are of no

f Silica

.sed to
tion of
tion of
:ion of
in the
.te '" ...s

mer in
.mer is
NaF is
of the
.t were
. silica,
1 when
y with
er, the
ate the

c.. .
of no

Characterization of Silicic Acids


value in studying the polymerization reaction unless it can be shown which polyacids
are actually present in the polymerizing mixture. For this reason, methods of
separating the oligomers or low molecular weight species are essential. A few examples follow.
Chromatography can be ,used, provided conditions are chosed to minimize
polymerization or depolymerization during the procedure. Wieker and Hoebbel (22)
found that by working rapidly, monomer, dimer, and higher species can be separated
by paper chromatography in 3-4 hr using dioxane containing (a) 1.6 g
I-ICCl 3COOH and 30 g 1-1 H 2 0 to separate monomer and lower polymers, or (b)
8.0 g I-I CCl 3COOH and 90 g L 1 H 20 to separate higher cyclic polymers. The paper
is dried and the separated spots are developed by spraying with 0.1 N NaOH and
aged wet for 10 min to depolymerize the silica, then with 2% ammonium molybdate
in 0.3 N Hel and aged wet horizontally for 30 min, then the yellow spots are
reduced to blue with 0.1 N ascorbic acid and bleached with ammonia gas to destroy
molybdenum blue, thus leaving the spots of blue silicomolybdate.
Low molecular weight silicic acids were separated by Baumann (82) with paper
chromatography using a mixture of isopropyl alcohol, water, and acetic acid as the
moving liquid and the molybdic acid reaction to locate the separate species.
Polysilicic acids of different molecular weights can be separated and molecular
weights estimated by gel chromatography on Sephadex columns, using 0.1 M NaCl
solution adjusted to pH 2 with HCl as the eluent. A blue dextran 2000 in 0.2% solution was used as a standard. Tarutani (83) made silicic acid at a concentration of 500
ppm by neutralizing the monomeric solution of sodium metasilicate with acid to pH
7. This solution was aged for various lengths of time and then acidified to pH 2 to
stop polymerization.
Polysilicic acids of low molecular weight have been isolated as trimethylsilyl esters
and separated by thin layer and gas chromatography by Hoebbel et al. (84). Specific
polysilicate ions known to exist in certain crystals were used to make the corresponding trimethylsilyl derivatives to use as standards. This method makes it
possible to separate these derivatives and characterize them further by gas
chromatography and mass spectroscopy. The sources of individual silicic acids and
their chromatographic constants are listed in Table 3.4. The derivatives were
separated, using a mixture of Merck alumina G and Merck silica gel G as adsorbent
and n-heptane as solvent. Programmed temperature chromatography was also-used

Particle Size and Surface Area by Titration

At a relatively early stage in the polymerization it is possible to characterize the

polymeric silica, or silica particles in terms of the specific area of the silica-water
interface. This is done by measuring the adsorption of hydroxyl ions in the pH range
4.00-9.00 (Beckman Typc E electrode) in a nearly saturated salt solution which
permits the surface charge denstiy to approach a maximum. This method was
developed by Sears (85) to determine the specific surface areas of colloidal particles
and gels. Then it was found that if carried out rapidly it could give reproducible

... ,.

-" ..,... -- ,,-- ...

....- -~.


Polymerization of Silica

Table 3.4. Sources of Individual Silicic Acids and Chromatographic Constants of their
Trimethylsilyl Deriratires

Silicic Acid

Ca 2SiO.

















[(HOhSiOh (cyclic trimer)


.0.31 "

[(HOhSiO]. (cyclic tetramer)

Tricycloheptasilicic acid
Cubic octasilicic acid
Cubic octasilicic acid







(CaSi0 3h
(KHSi0 3).
[(CH3).N + ].o[Si 70 18]( aq)10[(CH 3).N +]a[Si a0 20 ] S -

" The constants for the "cyclic trimer," as compared to those of the cyclic tetramer, suggest
that it is more stable and less reactive even though the trimer ring should be under greater
Source. Hoebbel et al. (84).

results on sols of particles only 3-4 nm in diameter with a specific surface approaching 1000 m 2 g-l.
However, in sols of such small particle size. there is an appreciable concentration
of monomer at equilibrium. Also, in alkaline sols at pH 9-10.5, there is an appreciable amount of ionic silica which is converted to monomer before the titration. Since
monomer reacts with base at pH 9 it is therefore necessary to correct the titration
for the effect of soluble silica in order to obtain a reliable value for the specific surface area of the polymer. The term "soluble silica" is used to include the ionic silica
and dimer which react with alkali-like monomer.
The soluble silica can, of course, be removed at pH 2 either by washing the silica
in a filter or ultrafilter or by centrifuging. Also the sample can be adjusted to pH 8
and let stand a few hours until the soluble silica has been polymerized upon the
colloidal material.

.;.. .J

. ~l,



Characterization of Silicic Acids

However. for colloids of extremely small particle size or dilute sols .....'here an
appreciable fraction of the silica is present as soluble silica. it is preferable to
determine the amount of monomer by suddenly acidifying a sample and determining
it by the molybdate method, after which the corresponding correction can be made.


Correction/or Soluble Silica







In the Sears procedure, titration with 0.1 N NaOH is begun with a silica sample in
150 ml of solution containing 30 g NaCl at pH 4.0.
As a source of soluble silica a 0.1 % solution of Si0 2 from Na 2SiOJ 9H 20 in 0.1 N
NaOH was used. Different volumes (x ml) of this solution were added with rapid
stirring to 135 - x ml of water containing 1.25 ml 4 N HC!. Then 40 g NaCI was
added and the pH raised with 1.0 N NaOH to 3, and to 4.0 with 0.1 N NaOH, as
ra pidly as possible.
The mixture was immediately titrated with 0.1 N NaOH solution rapidly to pH
9.0 and the volume noted. Samples of the final solution were then analyzed by the
molybdate method to determine how much silica was still present as
monomer-dimer (reaction time 2-3 min). When no more than 200 ppm of monomer
was added at the beginning, essentially all of it remained, showing .that it had not
polymerized. Blank titrations were also run on x ml of 0.1 N NaOH containing no
silica. This blank titer was 0.3 m 1.
With samples ranging from 3.8 to 45 mg Si0 2 , a net titer of 0.135 0.005 ml 0.1
N NaOH mg " ' was observed. To convert the Si(OH)4 completely to HSi0 3 - would
require 0.166 ml mg- 1 Hence the soluble silica is 81.3% ionized at pH 9 in the NaCI
solution. [From this it can be calculated that in this 3.4 N NaCl solution, pK a for
Si(OH)4 is 8.37.] Then in titrating samples for surface area, for each 100 ppm soluble silica (15 mg) in the 150" ml of mixture being titrated, a correction of 2.08 ml 0.1
N NaOH should be substracted.
As an example in which this correction is significant, consider the surface titration
of a 4 nm sol containing 5% Si0 2 at pH 10 that has stood until the total soluble
silica (154 ppm) plus ionic silica is 439 ppm. Usually a 1.5 g sample of Si0 2 is used
for surface titration. Thus 30 g of sol is required and this contains 439 ppm of soluble silica. This will be diluted to 150 ml for titration, resulting in a concentration of
88 ppm of monomer. This then requires a titration correction of 1.83 ml 0.1 N
NaOH. The total titration volume is 25.9, which would give an uncorrected area of
734 rn' g-I. After substracting the correction, the surface area of the colloid is 686
m 2 g-I, a difference of 6.5%.
Determining the surface area of silica by the Sears method has been automated by
Meffert and Langenfeld (86a). They used a saturated solution of sodium chloride
containing suspended fine salt crystals which seemed to scour the silica off the surface of the glass electrode where it otherwise was deposited and gradually changed
the electrode characteristics. An automatic titrator was used so as not to exceed pH
9 at any time. Using the formula

'....: ;', :.



Polymerization of Silica

specific surface area (rn" g-l)

ml 0.1 N NaOH required from pH 4.0 to 9.0 for 1.5 g SiO z
blank titer (ml)
Meffert found K = 25.2 when using a solution volume of 200 ml saturated NaCl
solution. This compares with K = 26.5 usually found with the Sears method using
150 ml of 200 g 1-1 NaCl. Vb is 1.3 ml for Meffert's and Langenfeld's method and
0.75 ml for Sears's.

Coagulation by Gelatin-Salt


As the ultimate particles of polysilicic acid begin to aggreate into chains and small
three-dimensional networks, the increase in size of the polymeric species can be
followed by an empirical procedure involving the precipitation of a complex of the
silicic acid with a given type of unflavored "edible gelatin" (86b). For a given sol the
method can be standardized by measuring the molecular weight by the freezing point
The method is based on the fact that at pH 1.5-3.5, gelatin alone is coagulated
from dilute solution above a critical concentration of sodium chloride. This critical
salt concentration decreases with increasing molecular weight of the gelatin. When
polysilicic acid is present it hydrogen-bonds and cross-links a given type of gelatin,
thus in effect increasing its molecular weight. Then a lower salt concentration is
required for precipitation.
However, as the molecular weight of the polysilicic acid is increased, it soon
reached the point where, with the particular gelatin used, precipitation occurs even
without adding salt. To increase the range of the titration; a competitive low
molecular weight hydrogen-bonding agent is used, specifically the diethyl ether of
diethylene glycol (formerly known as diethyl Carbitol, or DEC). As shown in Figure
3.11, with increasing concentrations of DEC more salt is required in the system to
cause precipitation of the gelatin.
Since the tendency of silicic acid to be precipitated by gelatin increases as the
silicic acid polymerizes, all experiments are carried out near pH 2, where polymerization is least rapid. With a sol of a given age and at constant concentration of electrolyte, the precipitation characteristics are independent of pH in the range 1.5-3.5
and relatively independent of the concentrations of silicic acid and gelatin. When
other factors are held constant, precipitation occurs as the concentration of salt
exceeds a critical value. With the concentration of salt held constant, solubilization
of the precipitate occurs (or precipitation is inhibited) when the concentration of the
hydrogen-bonding agent exceeds a certain critical level. The coagulating effect of
salt is therefore offset by the solubilizing effect of the hydrogen-bonding agent on the
gelatin-silicic acid complex.
The general behavior of the system is exemplified in Figure 3. I I, which shows the
effects of concentration of sodium chloride, concentration of DEC, and the age of
the silicic acid solon the formation of a gelatin-silicic acid precipitate in a system
containing 0.5-2% gelatin and 1-5% Si0 2 at pH 2.5 and 27C. Each of the lines,



Characterization of Silicic Acids






d and



In be
)f the
01 the
.on is

: low
ier of
ern to
s the
f salt
if the
.ct of
n the

's the
ge of








Figure 3.11 Precipitation of gelation-silica complex in the presence of a given concentration

of diethyl ether of diethylene glycol (DEC) occurs at concentrations of sodium chloride above
the indicated lines: A, gelatin without silicic acid; B, gelatin with low molecular weight polysilicic acid; C, D, E, with sols of intermediate ages and increasing sizes of microgel or
molecular weight: F, G, with sol approaching the gel stage. Scale on abscissa can be used as an
empirical measure of the degree of aggregation-polymerization of the silica. [From IIer and
Pinkney (98).]

representing a particular sol, is the borderline above which a precipitate is formed. It

will be noted that gelatin alone is salted out of water at pH 3 when the N aCl
concentration exceeds about 8 gj 100 ml. However, when DEC is present, a higher
salt concentration is required to cause precipitation, as shown by Line A of Figure
3.11. When silicic acid sols of increasing age are also present in the system, lines B
to G are obtained. An extrapolation of these lines to the DEC abscissa provides a
convenient scale for characterizing the state of polymerization of the polysilicic acid.
Before the effect of silicic acid in the system is discussed, the behavior of gelatin
alone should be considered. The hypothesis is advanced that gelatin molecules in
solution are associated with a sheath of water molecules, probably held by hydrogen
bonds. As salt is added, water is gradually removed from the gelatin molecules at an
increasing number of points. These dehydrated sites on different molecules then
associate with each other to form the cross-linked, gel network which is the
precipitate. Shorter gelatin molecules of lower molecular weight thus' require a
greater number of cross-links and therefore a higher concentration of salt for coagulation. The solubilizing effect of the hydrogen-bonding agent is visualized as follows.
When DEC, for example, is present in the system, the addition of salt brings about
partial removal of water from the sheath surrounding the gelatin molecules as
before, but these sites are immediately covered by adsorption of DEC molecules, so





Polymerization of Silica

that cross-linking of gelatin molecules through hydrogen bonding at these' sites is

The effect of silica might be explained as follows. The polysilicic acid molecules
may adsorb or otherwise hold together a number of gelatin molecules through
hydrogen bonding. Furthermore, the larger polysilicic acid molecules can adsorb a
larger number of gelatin molecules and thus form a larger complex unit. However,
this larger unit then behaves the same way as a single gelatin molecule of higher
molecular weight, requiring less salt for precipitation. The mechanism of aggregation of the gelatin-silica complex is therefore the same as that of gelatin alone, as
indicated by the parallel slopes of the lines.
A titration procedure is as follows. A 10 ern" sample of a solution of silicic acid,
free from organic solvents and containing 4-6% Si0 2 at a pH of 1.5-2.5, is pipetted
into a I x 8 in. test tube. To this is added 2-3 cm' of an aqueous solution of dicthyl
ether of. diethylene glycol (diethyl Carbitol) containing 500 g I-I, the pH being
adjusted to 2.5 with HC\. To this is added 5 em? of a 2% solution of high grade edible gelatin. Sufficient diethyl Carbitol solution is used to prevent the formation of a
precipitate when the gelatin solution is added. This mixture' is then titrated with a
standard solution of NaCl containing 300 g I-I (previously adjusted to pH 2.5 with
HCl) until the solution becomes turbid. This is observed by holding the tube against
an aperture % in. in diameter illuminated from behind by a 60-watt lamp. The end
point is arbitrarily taken when the clear outline of the aperture can no longer be
discerned. This relatively crude method is satisfactory, since the end point is quite
sharp; the mixture usually changes from perfect transparency to extreme turbidity
within 0.2-0.3 em", The mixture is shaken vigorously for about 10 sec after each
addition of salt solution before the turbidity is observed. The turbid solution is then
cleared by adding another 1-2 ern" of diethyl Carbitol solution and again titrated to
the end point with NaCI solution. At each end point the total concentrations, in
grams per 100 ern", of NaCI and of diethyl Carbitol are calculated, with the total
volume at each end point taken into account. These values are plotted as in Figure
The position of the lines obtained with sols of increasing molecular weight move
progressively to the right-that is, higher concentrations of diethyl Carbitol are
required to prevent precipitation. The intercept on the diethyl Carbitol axis therefore
served to indicate the relative molecular weight of the silicic acid.
In order to have an arbitrary scale with positive values, a function X was adopted
such that

where C = intercept on diethyl Carbitol axis. This "X value" then ranges from zero
for gelatin solutions alone (or for silicic acid sols of extremely low molecular weight)
to about 11 or 12 for silicic acid solutions which have polymerized to the gel stage
and which, therefore, contain material of high molecular weight.
In the case of polysilicic acid obtained by adding a 3.25 ratio sodium silicate solution to a rapidly agitated solution of acid to obtain a final pH of about 2. the relation between X value and the degree of polymerization as the silicic acid polymerizes



Mechanism of Condensation and Hydrolysis









Figure 3.12. Relation between "X value" and degree of polymerization, N determined cryroscopically.

is shown in Figure 3.12. In effect. the method indicates the degree of aggregation of
discrete particles of silica. In the present instance the initial discrete polysilicic acid
particles are about 2 nm in diameter. If one starts with particles 5 nm in diameter and
follows the aggregation process as the sol is being transformed to gel. the titration
procedure gives a series of lines similar to those of Figure 3.11. but ofa different slope.


As discussed in connection with the isoelectric point of silica the rate of polymerization of monomer and formation of silica gel is at a minimum at around pH 2. It has
been assumed that since this is the isoelectric point of silica, the catalyst below pH 2
is the H+ ion which forms an active cationic complex. Also, above pH 2 the OHion is the catalyst in that active anionic silica is generated.
However, lIer (87) discovered that the pH at which the polymerization rate is at a
minimum is .strongly influenced by traces of fluoride ion, as little as 10- 4 M
concentration (1.9 ppm F) having a marked effect. Since traces of fluorine are difficult to avoid it was suspected that the catalytic effect of H + ion might depend
entirely on fluoride impurity. However. in view of the observations by various
workers on highly purified silica systems, it is now apparent that the effect of
fluorine is merely superimposed on the effect of H ion alone. The self-condensation
of monomer catalyzed by OH - ion is commonly written:
2 Si(OH)4








:u:;~~ .~

Polymerization of Silica


Treadwell and Wieland (88) advanced the following hypothesis:


In order to explain the mechanism of the polymerization of silicic acid, we would have to
start with the fact that the tetravalent silicon is still unsaturated coordinatively. Both of the
strongly defined secondary valences which are active in the fluoride complexes of silicon must
also playa role in the hydrated OXIde.

Iler (89) represented an intermediat-e 6-covalent stage shown in Figure 3.13, and
A. Weyl (90) postulated even more extensive complexes involving 6-coordinated
There appears to be a basic difference in the polymerization mechanism below
and above pH 2. (See also foregoing section on the isoelectric point.) Above pH 2,
the rate of disappearance of monomer is a second order reaction; below 2 it is third
order. The order of the reaction has been explained by Okkerse (91) on the basis
that silicon increases its coordination number to 6, as a three-silicon intermediate is
formed below pH 2, shown at B in Figure 3.13.

/ ,I,


Figure 3.13. Possible intermediates in polymerization of silica: A, two-silicon anionic complcx involving OH - ion; B, three-silicon cationic complex involving H - ion [Ok kcrsc (29. 91)]:
C and D, suggested alternates to A: E, intermediate at silica surface.

Mechanism of Condensation and Hydrolysis


Above pH 2 a proposed alternative to A involves only two silicon atoms, a's at C

in Figure 3.13. Another structure for the intermediate was suggested by Dalton (92).
shown at D. If one or both of the silicon atoms should be part of a polysilicic acid or
surface of a silica particle. only a coordination number of 5 might be involved. as at

In any case it is supposed that when silicon becomes coordinated with five or six
oxygen atoms all the silicon-oxygen bonds are weakened so that rearrangements can
occur. This has been considered in considerable detail by Strelko (93). who proposed
the following rate equation for the pH range from 2 to 10:

where K D = M-[H+]/M, M and M- are the concentrations of Si(OH). and

HSi0 3 -. and C is the total concentration of Si0 2 Here

=C- M


This equation corresponds to a maximum in the rate of polymerization somewhere

in the mid-pH range.~,,,,
A general theory of polycondensation equilibrium in silicic acid solutions was
proposed by Stober (94). From it he deduced the concentration of monomer in equilibrium with polymers of different degrees of condensation. Further data are needed
to check the validity of the complex equations that were developed.

Catalytic Effect of HF

Traces of fluoride ion have a marked catalytic effect at low pH where it is mainly
present as HF. Her (87) found that. whereas the polymerization reaction is catalyzed
above about pH 2 by hydroxyl ions as reported by others. at lower pH the polymerization is catalyzed by hydrogen fluoride. or more specifically the rate is proportional
to the concentrations of hydrogen ions and fluoride ions (Figure 3.14).
The value of the pH for optimum stability of silicic acid depends on what
impurities are present in the solution. Traces of aluminum ions and to a lesser
extent. iron. thorium. and beryllium ions tend to offset the effect of fluoride ion by
forming complex fluorides and thus retard polymerization in this pH range. Depending on the purity of the solution. the pH of optimum stability may range from I to
as high as 3-3.5. In silicic acid solutions free from aluminum impurity. as little as I
ppm of fluorine has a marked effect on the rate of polymerization in acid solu tion.
Iler advanced the hypothesis that the polymerization mechanism may involve the
temporary expansion of the coordination number of silicon from four to five or six,
just as in the case of OH- ion. It is not known whether the Hf'-catalyzcd reaction is
third order as in the H +-catalyzed reaction in the absence of fluoride below pH 2. or














SOL-I.O r.lOLAR Si0 2






Figure 3.14. Effect of fluoride ion on gel time of silicic acid sols; line H represents effect of
hydroxyl ion (Iler). Below the iscelectric point. pH 2. the gelling rate (inverse of gel time) is
proportional to H' ion concentration and also to F- ion concentration. provided the
concentrations of AP- and Fe3 - impurities remain constant. Thus the catalyst is HF. [From
lIer (89).]




Polymerization: pH 1-7

whether it is second order and therefore analogous to the OH -catalyZed reaction

above pH 2. Strelko also recognized the possibility of alternate reaction mechanisms
but there are insufficient experimental data to make a choice.
Whereas lIer showed the effect of fluoride on rate of gelling which is the final
stage in polymerization, Tarutani (95) demonstrated that it accelerated the
polymerization of monomer which disappeared at a rate proportional to F- ion up
to 19 ppm fluorine in 3 N HCI. Fluoride had no effect at pH 7. Tai and Chen (96)
found that when silica gel was formed at low pH in the presence of fluoride ion, the
latter was held on the gel and none appeared in the syneresis liquid. In an alkaline
medium no fluoride ion was held.
The removal of F- ion by AI3+ can be estimated from the stability constant for
AIF,,_t + F = AIF", as measured by Agarwal and Moreno (97a). The values at
25C of K" for n = 1,2,3,4, were 6.69,5.35,3.68, and 2.75 (all less than O.04).
The equilibrium constant involving formation of SiF/- in the presence of excess
Si0 2 is as follows, according to Scherban (34b):

+ 2 H 20

[F-]6 [H+]4


Si0 2

+ 6 F- + 4 H+

10- 17

It is necessary to consider the polymerization of silica in solution below pH 7
separately from that above pH 7 for the following reasons:

I. It is only at low pH that the polymerization is sufficiently slow to allow the early
stages to be followed. Changes that occur in hours at pH 2 are completed in
minutes or seconds at pH 8-9.
2. The polymer units or particles formed at low pH bear no charge and unless the
silica concentration is very low, aggregation begins to occur soon after polymeric
particles are formed, not only because of the lack of charge on the particles, but
also because the particles are extremely small and cease to grow after they reach
a diameter of 2-3 nm. Since the rate of aggregation depends mainly on the
number of particles per unit volume and less on their size, aggregation occurs
even at very low silica concentrations.
:ct of
ne) is
j the


3. Above pH 7 particle growth continues up to a larger diameter, depending mainly

on the temperature, unless aggregation or gelling takes place when an electrolyte
is present in an amount approaching the critical coagulation concentration at the
pH of the sol.
It is therefore only at low pH that the progressive polymerization of monomer can be
followed and the initial formation of three-dimensional condensed silica polymers or
nuclei particles can be observed. The general observations of investigations are summarized here, but detailed discussions are included later.

Polymerization of Silica


Formation of Oligomers

By "cligorners" is generally meant the low molecular weight polysilicic acids from
trimer to perhaps octamer or dodecamer which react within 10-20 min with the
usual molybdic acid reagent. This includes linear, cyclic, and polycyclic polysilicic
acids. Their completely trimethylsilylated esters are generally distillable under high
vacuum and, except for monomer, are solids. No clear-cut distinction from
somewhat higher polymers appears to have been made. However, there may be good
reason for this. Once this degree of polymerization has been attained, further
increase in particle size of molecular weight occurs rapidly if monomer is still available or the overall silica concentration sufficiently high.
The peculiar polymerization behavior in this system appears to be explained by
the following:
1. Monomer, Si(OH)., has a pK a of about 9.9.
2. Dirner pKa is reported higher but is likely lower than that of monomer.
3. Higher polymerized species have a much lower pK a approaching 6.7 and hence
are more highly ionized than dimer or monomer.
This implies that the greater the number of siloxane linkages and the fewer OH
groups on a silicon atom, the stronger the acidity. Although the dimer was reported
as an exception in the acidity series, this seems doubtful.








'\Si(OHh < =SiO-SiOH



As previously mentioned, polymerization involves intermediate ionization to

=SiO- or to =Si+ below or above pH 2, respectively, but to simplify discussion only
==SiO- is used as the example.
It should be understood that in the following formulations the ions are represented
as anhydrous whereas they probably are associated with water in the form of
hydroxyl groups and in hydrogen-bonded forms. In the absence of specific
knowledge the anhydrous forms of the ions are used.
The key point is that condensation involves the reaction of an =SiO- ion with a
nonionized silanol group:




+ OH-

Then the following is evident:

I. At low pH where Si(OH)4 is ionized only to a very small degree and no polymeric
species is present, dimerization is slow.



Polymerization: pH 2-7


2. Similarly, if dimer is a stronger acid than monomer, the main reaction-is with
ionized dimer:

.h the
," high



ed by


r OH


In to
1 only

'm of



+ Si(OH).


+ OH-

However, the existence of linear trimer is transitory and concentration probably

remains low.
Owing to the ionization of dimer it can combine also with its own ionized species
to form linear tetramer.
Owing to the proximity of SiOH end groups in a given linear polymer, for
example, tetramer, and low concentration of monomer, ring closure is rapid.
Cyclic tetramer is mainly formed (84) since bond angles make ring closure of
linear trimer unlikely.
Once the cyclic species predominate, monomer and dimer react preferentially
with these more highly ionized species, increasing their molecular weight.
At the same time, the adjacent SiOH groups on the polymers condense wherever
further ring closures are possible, leading to more compact, three-dimensional
species, the precursors of colloidal particles.

The simple polysilicic acids can be represented by conventional formulas as in

Figure 3.15. As the, polymers become three-dimensional they can be shown instead
with spheres representing oxygen atoms arranged tetrahedrally about silicon atoms
which cannot be seen, as in Figures 3.16 and 3.17. In Figure 3.17 the symmetrical
structures of the double cyclic trimer and double cyclic tetramer (cubic octamer) are
shown as models of the smallest three-dimensional, fully condensed species.
However, these have never been isolated from polymerizing. solutions since they
would form only a small fraction of all the three-dimensional species that can be
formed by random combinations of monomer and various lower oligomers present.
After the three-dimensional species have been formed with their more highly
ionized silanol groups, the remaining monomer and dimer react preferentially with
them and through ring closure build up larger spheroidal particles as in Figure 3.17.
These roughly spherical polymers undergo further internal condensation and rearrangement to a more compact state. This change is rapid at pH above 7 and especially at elevated temperature, forming colloidal particles with a core of Si0 2 and a
surface of SiOH groups.

'lith a
Oligomers as Particles


It has been generally known that when monomer polymerizes in alkaline solution
in the pH range 8-10 colloidal silica particles form quickly and grow spontaneously
to recognizable size. The possibility that three-dimensional or particulate polymers
of "subcolloidal" size form during the polymerization of monomer soon after cyclic
oligomers appear has not been generally recognized. However, particles about 20 A
in diameter corresponding to a degree of polymerization around 100 wert: shown to

::-#.- --,

...... __ ...




HOSiO- +

HO~iO~iOH + OHaH H



o 0 0
HOSiaSiOSiaH + OHa 0 0


a a




o 0 a 0


+ ow









a 0
a a


a a
a 0
a 0



Figure 3.15. Conceivable steps in the polymerization of silica to cyclic species. However.
consideration of bond angles makes the formation of cyclic trimer much less likely than cyclic
tetramer. [From Iler (97b), permission of Plenum Press.]

Figure 3.16. Molecular models of silicic acids of Figure 3.15. Spheres represent oxygen
atoms; black dots, hydrogen atoms. Silicon atoms within oxygen tetrahedra are not visible.
Not all hydrogen and oxygen atoms are visible: (I) Si(OH ).: (2) (H OhO SiOSi(O H),: (3)
(OHh SiOSi(OHhOSi(OHh: (4) [(OHl,SiOh: (5) [(OH>,SiOj. The existence of cyclic trimer
is questionable. (From Iler (97b). by permission of Plenum Prcss.]



polymerization: pH 2-7



Figure 3.17. Models of (A) cyclic trisilicic, (B) cubic octasilicic acids, and (C) and (D) the
corresponding theoretical colloidal particles formed by condensing monomer to form closed
rings until the original species is completely surrounded by one layer of deposited silica bearing
silanol groups. When formed above pH 7 the inner silica contains few silanol groups. Different
kinds of incompletely condensed oligomers could form the cores of colloidal particles. There is
no evidence that A and B are specifically involved. Spheres. oxygen atoms; black dots.
hydrogen atoms. Silicon atoms are not visible. [From lIer (9ib) by permission of Plenum

n cyclic

H),; (3)

: trimer

be present in polysilicic acid formed and aged at pH 2 (98). The possibility of

smaller spherical polymers with even lower degrees of polymerization was not
recognized until corresponding siloxane structures were shown to be formed by the
hydrolysis and polymerization of compounds of the type RSiCI 3. Scott (99) first
isolated the compound (CH 3SiOI . s)n, which sublimed without melting, and concluded the molecular weight was low but n was not determined.. A spherical siloxane
polymer was made by Olsson (100) in the form of the compound (C2HsSi)a0I2 or
(C 2HsSiO I . s)a. This molecule was reportedly so inert that it was not attacked by
refluxing nitric acid, aqua regia. bromine, perchloric acid, or 60% aqueous KOH.
Evidently the ethyl groups are so closely packed around the spherical core of silica
that only the very inert CH 3 ends are exposed at the surface of the molecule.
The corresponding siloxane (HSiO I . 5 )a was obtained as a crystalline compound
(m.p. 250C) by MUller, Kohne, and Sliwinski (10 I). If this were oxidized in dilute
solution it might yield the pure octasilicic acid.
When R is a bulky alkyl group, the condensation of RSi(OH)3 is forced into the
smallest possible three-dimensional unit, polycyclic tetramer, (RSiO I . s ). (102).
However, it is unlikely that this tetrahedral structure involving only four silicon
atoms can be formed by polymerization of monomer in water, since it was not
formed from RSi(OH)3 unless the R groups were bulky.
Further examples of small, generally spherical siloxanc structures of the type
(RSiO I ' 5 ) 1l were furnished by Barry et al. (103), who prepared the cubic octarner and
also a hexaprisrn 12-mer consisting of two parallel hexagonal rings linked together
through oxygen atoms. where R was methyl. ethyl. II-propyl, II-butyl, cyclohcxyl,
and phenyl. The phenyl group was sufficiently bulky to cause the formation of the
hexarncr in which two parallel trimer rings were linked together through oxygen



Polymerization of Silica

atoms. Thus from a structural standpoint corresponding polysilicic acids should be

capable of existing as the smallest possible "particles" of silica.
However, in water, condensation of Si(OH)., dimer, and trimer with cyclic
species, for example, tetra mer, must occur randomly. Thus it is not likely that any
regular three-dimensional fully internally condensed species such as cubic octasilicic
acid, (HOSiO ,.5 )s, will be formed in appreciable amount. Instead, less regular,
incompletely condensed three-dimensional oligomers must appear. Later, after
higher molecular weight polymers equivalent to very small particles have appeared,
the more soluble oligomers depolymerize or dissolve and the dissolved silica' is added
to the growing particles.
Nucleation Theory

It should be noted that in this system, polymerization, in terms of formation of

siloxane bonds, is a reversible process. The concentration of OH - ion that promotes
ionization and condensation equally promotes hydrolysis and depolymerization.
Thus each polymer species is in equilibrium with a certain concentration of
monomer in solution. Three-dimensional polymer units larger than the double-threering hexamer or similar species can be considered to be the core of a particle. A
polycyclic octamer or decamer is not actually a silica particle because there is no
anhydrous Si0 2 "core." As additional monomer condenses with SiOH groups on
these oligomers they also condense with themselves until the oligomer remains as an
anhydrous Si0 2 core surrounded by a layer of silica in which the silicon atoms retain
outwardly disposed silanol groups as shown in Figure 3.17.
In studying models it is found that when an oligomer such as octamer, or
decamer, is the starting point for further addition of monomer, the geometry is such
that by the time the original oligomer has become fully condensed to a nearly anhydrous "core" of Si0 2 surrounded by silicon atoms forming the outer hydroxylated
surface of the particle, the latter must contain 40-50 silicon atoms.
It is interesting that the formulas calculated for the composition and size of silica
particles in Chapter 1 give values in this range. Thus a 48-mer would have an OH: Si
ratio of 0.8-0.5 by calculation, whereas on the model it appears to be about 0.7. The
calculated hydroxylated diameter is 1.6 nm and from the model it appears to be
about 1.3 nm. The equivalent diameter of an anhydrous Si0 2 particle is 1.09 nm.
Even a "40-mer" particle is probably in solubility equilibrium with monomer and
should be large enough to follow the equation relating solubility and particle size, at
least approximately, although the surface energy must change somewhat as the
radius of curvature approaches atomic dimensions. Once such nuclei have been
formed, they grow by the addition of whatever monomer remains in solution and are
further supplied by dissolution of oligomers.
The theory of homogeneous nucleation has apparently not yet been developed on
a quantitative basis, but some relationships have been considered between degree of
supersaturation, interfacial energy of silica to water, and the critical size of nuclei.
A major difference between the silica-water system and other aqueous solutions
of inorganic compounds is that in the case of silica a catalyzed making and breaking
of siloxane bonds occurs. whereas no such requirement for a catalyst seems to be


,f Silica

uld be

at any

ion of

ion of
-threecleo A
: is no
IpS on
; as an
er, or
s such

. silica
7. The
to be


ize, at
IS the
ad are
ree of
to ~_

Polymerization: pH 2-7


involved in the nucleation and growth of molecular or ionic crystals, for example,
sugar or BaSO~. In most systems, heterogeneous nucleation is difficult to avoid
owing to the presence of minute impurities in suspension, but this is not the case
with silica. It is, of course, possible to supply particles with a large receptive surface
areas, such as a finely divided metal oxide or silica itself to a solution of monomer
and thus prevent self-nucleation. However, unless colloidal material is deliberately
added, a supersaturated solution of silica undergoes spontaneous homogeneous nucleation that seems to be unaffected by small amounts of foreign matter. The homogeneous nuclei are formed by condensation of monomeric Si(OH)~ when the
supersaturation ratio is sufficiently high.
The homogeneous nuclei are apparently in the same size range as those of ionic
materials, which, according to Walton (104), contain about 10-100 ions. In the case
of silica, 10-100 Si0 2 units corresponds to particles about 1-2 nm in diameter.
Experimentally, Nielsen (l05) found that with BaSO~ having an interfacial energy of
90 ergs cm -2, a supersaturation ratio of 21.5 was required for nucleation. If the
supersaturation ratio is the same function of interfacial energy as in the size-solubility equation, so that log R = kE, where R is the supersaturation ratio, then for
BaSO~, log 21.5 = k(90), whence k = 0.0148. Since the solubility of massive silica is
70 ppm and E is 48 ergs ern -2, log R = 0.0148(48) = 0.71, whence R = 5.12 and the
critical concentration for nucleation would be 5.12 x 70 or 358 ppm.
Marsh (32) followed the disappearance of monomer at pH 8 where therelatively
high concentration of hydroxyl ions permits dissolution-deposition to occur rapidly.
He observed that there was an induction period as though nuclei were being formed.
The period ranged from about I min at a monomer concentration of 697 ppm, to
1000 min at 359 ppm, indicating possibly that the critical supersaturation ratio may
be of the order of 5-10. This is within the range of R calculated above.
Dunning et al. (106a, 106p) have developed equations relating the rate of formation of nuclei to the supersaturation ratio and surface energy of the nuclei. Assuming
that nucleation became apparent when new nuclei appear at a rate of 100 ern -3 sec- 1
and taking the interfacial tension as 48 dynes em -1 and the solubility of massive
amorphous silica as 70 ppm, a supersaturation ratio of about 7 or a critical nucleation concentration of 500 ppm was calculated. This appears to be of a reasonable
magnitude (I06c).
Definitive data on nucleation of colloidal silica particles in brine solutions at pH
4.5-5.5 and 95C have been obtained by Makrides and associates in a study related
to the deposition of silica from hot geothermal waters (106d). Their work showed
conclusively that a solution of monosilicic acid requires an induction period for the
formation of nuclei that strongly depends on the degree of supersaturation. Under
these conditions appreciable time is required for the early stages of polymerization
to produce three-dimensional polymer particles of the type that can function as
nuclei. With a supersaturation ratio of 2-3. the nucleation time ranged from a few
minutes to several hours.
A theory of nucleation was developed. based on an interfacial surface energy of
the silica-water interface of about 45 ergs ern - I in good agreement with the values
obtained from solubility studies (Chapter I). Fluoride ion at 10-100 ppm accelerated
nucleation and particle growth.




0"-.- ,
" i


Polymerization of Silica

Particle Growth in Acidic Solution

Since the growth of amorphous silica particles involves the random aggregation of
SiO. tetrahedra, a model developed by Ordway is of interest (107). Thus on a growing sphere, Si(OH). will condense preferentially at a site where the silicon atom on
the surface bears only one OH group and is therefore the most acidic and readily
ionized. The model was built up to a "degree of polymerization" of 300 in the form
of a sphere free from voids. As soon as a certain concentration and surface area of
nuclei have been formed, the remaining monomer and dimer react preferentially
with the nuclei because of the greater acidity and degree of ionization of the polymer
surface. As the concentration of monomer and dimer decreases, the oligomers,
which have been in solubility equilibrium with the higher monomer concentration,
depolymerize and the dissolved silica adds to the growing nuclei.
As will be seen in experimental studies to be discussed later, the general course of
the polymerization to form high polymer is typified by the results of Weitz, Franck,
and Giller (l08). To make it possible to follow the polymerization, the initial
monomer concentration was 4000 ppm and the pH was 3 so that the changes were
relatively slow. The oligomer consisted of polymers which were low enough in
molecular weight to depolymerize and react with molybdate reagent, although the
reaction rate dropped as the average molecular weight increased. The concentration
of oligomers reached a maximum when about half the monomer 'had been
polymerized. At this point colloidal particles of higher molecular weight began to be
formed at the expense of oligorners, which then decreased in amount.
In the early stages of conversion of silica to particles, where the concentrations of
oligomers are decreasing, it is probable that once a certain concentration of nuclei
has been reached, they remain constant in number as they grow in size. When
oligomers have been consumed, the particles then grow in size at the expense of the
smaller particles, and thus decrease in number.
In acidic solution at 25C these changes can be followed only under conditions
where the solution is so dilute that the rate of aggregation is slow in spite of the lack
of charge stabilization. The low pH also minimizes siloxane bond formation upon
collision. However, aggregation inevitably occurs at low pH although only a viscous
suspension, rather than a gel, is formed in very dilute solutions. In any case, because
at low pH particles do not grow much larger than 2-3 nm in diameter and because
the minimum concentration for gelling varies inversely with particle size (Chapter 4),
gelling will occur even at a concentration of 1% Si0 2 or less.
In alkaline solution above pH 7, all these early stages of polymerization occur
very rapidly and particle growth by "Ostwald ripening" continues as the charged
particles remain as sol.

Depolymerization in Acidic Solution

A study of depolymerization of low polymers suggests that the degree of internal

condensation within three-dimensional polymers depends on the temperature at
which they were made or to which they have been heated. Even more importantly it


f Silica


growern on
; form
.rea of

irse of
) were
igh in
gh the
I to be

polymerization: pH 2-7


depends on the pH since the rate of internal condensation within the three-dirnensional polymer units is controlled by the OH - ion concentration. Even at pH 2-3
there is a marked difference and at pH 8 the rate is many orders of magnitude
greater. As a result the rate at which polymer units of a given size depolymerize in
the molybdate reagent is much less when the polymer has been exposed to higher
temperature or to higher pH. .
Depolymerization of polysilicic acid was studied by Stade and Wieker (l09), who
prepared a solution of polysilicic acid containing a minimum of monomer by
dissolving (Li 2Si0 3)n powder in dilute acid at OC. Samples were then diluted to 12
ppm Si0 2 in water at pH 1.3-5.5 and the depolymerization to monomer followed at
different pH values and temperatures. The rate of depolymerization was followed by
removing samples periodically and analyzing for monomer in solution by the molybdate method of Alexander (24a). At the same time the rate at which the residual
polymer reacted with molybdate was also measured.
Figure 3.18 was drawn from data given by Stade and Wieker (109) and shows that
the half-life of the polymer at 12 ppm in water is at a maximum, that is, the rate of
depolymerization is at a mimurnun, at about pH 2.5. However, the effects of





of the



ter 4),



~ 10








ntly ,


Figure 3.18. The effect of pilon the rate of dcpulyrncrization of silica at

12 ppm SiO~. (From Stade and Wicker (109).J

a to 65C diluted to





............ r


Polymerization of Silica

temperature and pH are quite complex. It is probable that this is because ~\'O dif.
ferent reactions are going on simultaneously.
When polysilicic acid is formed at low pH at OC, polymer particles are formed,
but the inner SiOH groups are not all fully condensed. However, when such a sol is
warmed and/or the pH is raised. above 2, further internal condensation occurs. It is
therefore logical to expect that if the sol is diluted at OC, only depolymerization
will occur. However, if the temperature is raised, or if the pH is also raised, two
processes occur. The particle starts to depolymerize or dissolve, but also it condenses
internally with formation or more siloxane bonds that must later be broken to form
monomer for reaction with molybdic acid. Thus the reaction rate with molybdic acid
is greatly reduced.
Evidence for this is that the reaction velocity constant for the polymer made and
diluted at OC was 0.046 min-I, but when it was diluted at 25C the constant dropped to 0.03 min ? after 5 min. Additional evidence is that when depolyrnerization
was followed at OC where no internal changes occur in the polymer and only
depolymerization occurs, the rate of depolymerization in the range from 3 to 5 was
exactly proportional to the OH - ion concentration.


As already briefly described in the general theory of mechanisms and shown in
Figure 3.1, when silica particles are not charged, they link together into branched
chains to form three-dimensional network regions in the sol. These microgel regions
grow at the expense of the sol regions until the solid microgel occupies about half the
total volume, at which point the viscosity becomes infinite and the sol has reached
the "gel point."
A more detailed discussion of gel formation is reserved for a later chapter on
silica gels and powders. Meanwhile, to understand the role of aggregation in the
polymerization of polysilicic acid it is necessary to discuss the mechanism by which
the molecular weight of silica increases by aggregation of particles or small spherical

Molecular Versus Particle Chains

In the silicic acid system, the formation of gel has usually been ascribed to the
condensation of Si(OH). into siloxane chains, then branching and cross-linking to
form a three-dimensional molecular network. Measurement of gel time has often
been used as an indication of rate of polymerization (90, 110-113).
Such a siloxane gel network might be obtained under conditions where
depolymerization is least likely to occur so that the condensation is irreversible and
siloxane bonds can not be hydrolyzed once they are formed. Thus Si(OH). made by
dissolving an anhydrous orthosilicate in anhydrous methanol-hydrochloric acid, or
by hydrolysis of methyl silicate with a theoretical amount of water in methanol, will
slowly form a gel. Owing to the insolubility of silica in the system (which is another

of Silica

Polymerization by Aggregation-i-GelFormatipn _

.wo dif-

way of saying that siloxane bonds are not readily broken) the condensation 'polymer
of siloxane chains cannot undergo rearrangement into particles. Even if some
extremely small particulate units were formed, they could not undergo Ostwald
ripening. Under these special conditions perhaps the polymerization of Si(OH).
might closely resemble the polymerization of a polyfunctional organic monomer and
may follow the corresponding theories which have been developed in organic
polymer chemistry.
However, the transformation of silica sols of relatively large discrete spherical
particles of known uniform size into firm homogeneous gels suggests that most, if
not all, silica gel networks may be made up to discrete particles rather than chains of
individual SiO. tetrahedra. Furthermore, it is becoming increasingly OO\10US that a
very early stage in the polymerization of Si(OH). involves the formation of cyclic
and then three-dimensional polycyclic polymer units that are, in effect, very small
Si0 2 particles bearing surface SiOH groups.
Also a study of the structure of even the finest-pore silica gels has led to the conclusion that they have a particulate structure. Vail (114) has pointed out that in silica
gel the particles grow in size, become more anisotropic, and coagulate by aggregation; that is, in the same volume of gel the individual particles may decrease in
number and increase in size. Signer and Egli (115) also concluded that the framework of silica gel was not made up of simple chains of Si-O-Si linkages. Even in the
case of silica gels made from 3.25 Si0 2 : Na 20 ratio sodium silicate. it is probable
that the network gel structure is composed initially of polysilicic acid uaits containing an average of 3-5 Si0 2 units each, since, as shown by Iler (116). tbe silicic acid
liberated from this silicate by acid is already polymerized at least to this degree.
On the basis of studies of the thermal aging of silica gels, Shapiro azd Kolthoff
(117) agreed with Elkins, Shull, and Roess (118) that the structure of silica gel could
best be visualized as being .made up of discrete particles. The latter authors estimated, from small angle X-ray scattering, that the average size of the primary particles, assumed to be spherical, in freshly prepared silica gel was around .:O-tO A.

a sol is
rs, It is
ed, two
to form
die acid
ide and
it droprization
rd only
05 was

.own in
half me
pter on
I in the
y which


Mechanism of Interparticle Bonding

to the
king to
.s often
ble and
lade by
reid, or

The conversion of a sol of spherical particles to a uniform gel containing all the
liquid in the sol is not easily understood. When particles collide it is assumed that
adhesion can occur but in the case of silica particles there is reason to celieve that
the attachment is through the formation of Si-O-Si bonds. One reason fer thinking
so is that the same factors that promote polymerization of monomer and low
molecular weight silicic acids also promote the conversion of a sol oi colloidal silica
particles to a gel. Thus sols consisting of well-defined spherical particles form gel
least rapidly at about pH 2 and the process is accelerated by fluoride ices atlow pH.
One can only conclude that when particles collide, there are r.e:.:t::l.1 =SiOH
groups as well as =SiO- ionized groups on the surface of the particles which
condense to form Si-O-Si linkages by the same mechanism involved in the
polymerization of low molecular weight species. However, the presence of soluble
silica or monomer then plays a role of further cementing the particles together.

Polymerization of Silica


... ...






. ;



, ....----~

"'0 '





o ,.., T' 0

, 0",

Y', ...



i ,.





Figure 3.19. 'Bond formation between silica particles. With little or no charge repulsion. collision results in formation of interparticle siloxane bonds. catalyzed by base. Once bonded. the
panicles grow together.


Indeed it is even possible that the presence of Si(OH). at the point of contact
between colliding particles may playa role in promoting the formation of the initial
As shown in Chapter I. the solubility of silica at a negative radius of curvature is
less than that of massive silica. When two spherical particles come into contact there
is at that point an infinitely small negative radius of curvature and the solubility is
zero. Monomer will therefore be instantly deposited around that point (Figure 3.19).






polymerization by Aggregation-e-Gel Formation


Formation of Chains of Particles and Networks



e is
y is

A silica sol of very small particle size which contains only 1% Si0 2, or about 0.5%
by volume of particles, will nevertheless solidify to a coherent gel containing all the
water of the sol. It is obvious that this can happen only if the particles can link
together into chains. As will be shown, these chains can develop into rods or
"fibers." It is therefore not surprising that there has been debate in the past whether
gels are made up of particles or have a fibrous structure or a solid structure full of
interconnecting pores like a very porous but uniform sponge. As is often the case in
such scientific arguments, each point of view may be essentially correct under
certain conditions. All such types of structures can be developed in the silica-water
system, depending on the original particle size and conditions of formation and
An electron microscopic study of fine-pored silica gel by Sugar and Guba (119)
revealed that the structure indeed was made up of a threadlike or fiberlike network,
but the fibers were made up of chains of spheres. The pore volume determined by
adsorption could be correlated with the observed size and spacing of the chains of
particles on the assumption that these were on the average cylindrical and that the
pore space was the complementary space of a skeleton built up of the network of
cylindrical elements.
The view that the structure of gel made from dilute hydrosol is better described as
"fibrillar matter" rather than as a series of independent particles was expresed by
Scott, Hockey, and Barby (120). The fact that the sol gelled at a concentration of
only 1-3% Si0 2 and was made at pH 3-7 by deionizing sodium silicate indicates that
the particles were less than 3 nm in size and that these must have chained together
into rather long chain segments between branch points. Evidently such particles
formed chains and coalesced into rods or fibers until they essentially lost their
particulate identity.
The aggregation of particles into chains was first explained by Usher (121) on the
basis that a single particle would preferentially aggregate with a neighboring pair of
particles to form a linear rather than triangular arrangement, since. the linear
arrangement requires the approaching particle to overcome the repulsive force of
only one of the particles of the pair. Rees (122) further analyzed the potential energy
contours around a pair of spheres in contact under different conditions of repulsion
and attraction. He showed that under conditions of high attraction, where presumably rapid coagulation occurs, compact three-dimensional aggregates occur, but
under conditions of low attraction chainlike aggregates arc formed. In the case of
silica. in the neutral pH range with coagulating salt present, a precipitate is formed;
when little or no salt is present chainlike structures art: formed leading to gel.
A study by Thomas and McCorkle (123) of the interaction of a linked pair of
spheres with a third approaching sphere showed that under certain conditions 'the
third particle "flocculates" or adds to the end of the pair and subsequent particles
add to the end rather than to the side of the growing chain. Applying the
Dcryaguin-Vcrwcy-Overbcek theory (124, 125). they developed equal-energy
contours around two spheres in contact. assuming certain zeta potentials and clcc-

.... y


... -f





Polymerization of Silica

trolyte concentration. The contours show that a particle approaching" a pair or a

short chain of particles will have a greater probability of surmounting the charge
barrier if the approach is along the longitudinal axis. Thus the particle adds to the
end of the chain.
The question remains why silica sols will still form gels at pH 2, the isoelectric
point at which the charge' on the particles is presumably zero. It is true that gel
formation is slowest at this point, but this can be ascribed to the slowness with which
siloxane bonds are formed between particles at this pH. Nevertheless, silica particles
do chain together into a uniform gel structure at this pH and the structure is not any
different from that formed more rapidly at pH 3-5, for example. Also at the
isoelectric point no extreme retardation of gelling has been noted.
Since the formation of an Si-O-Si bond between two colliding particles requires
not only an SiOH group on one surface but also an ionized SiO- on the other surface, it appears that some ionization must occur even at the isoelectric point, pH 2.
But because the overall net charge is zero, there must also be present an equal
number of positive charges of the kind present at lower pH. However, the number of
both charged sites per unit area of silica surface must be exceedingly small, and
coverage sparse.
If this is so, then on a pair or on a chain of particles, one may speculate as to how
such sparse charges might be distributed. If such charges tended to be located at the
ends of chains as rather widely separated pairs with their accompanying counterions,
then this would explain why further particles would be added at the ends so that
chains would be formed.

Figure 3.20.

Computer projection of a chuinlik e aggregate by purely random aggregation of

uncharged particles which adhere rigidly together upon collision [According to D.N. Suther-

land (126:1).]


on of Silica

Polymerization by Aggregation-Gel Forrnatiop__

. pair or a
the' "~rge
ide, ,the

Another mechanism by which aggregation of particles can result in a somewhat

fibrillar structure is proposed by Sutherland (126). He points out that directional
ionic forces have been previously thought necessary to account for chain formation.
However, his computer simulation of flocculation, which requires only that a rigid
bond be formed when particles and/or aggregates collide on a random basis, can
result in highly extended brandied chainlike aggregates. These consist not only of
short chains of single particles, but also of random multi-particle width as shown in
Figure 3.20. This may be the most reasonable explanation of how very small
uncharged polysilicic acid particles at pH 2 can still aggregate in such a open network, which can occupy and solidify the whole volume of a solution with less than
I % by volume of particles. However, the silica system differs from the simulated
flocculation in that, at least at a given low pH, only a certain small fraction of the
collisions result in formation of a rigid bond. Also on the periphery of a given growing network (microgel particle) there will be extended chains, between which unattached particles will be found, but these later attach themselves to the network so
that the concentration of silica within the microgel is the same as in the original sol
on a macroscopic basis.
A comprehensive review of computer simulation of colloidal systems has been prepared by Medalia (126b).

ue that gel
with which
ca particles
~ is not any
.lso at the
les requires
other suroint, pH 2.
it an equal
~ number of
small, and

:e as to how
cated at the
nds so that

199regation of
D.N. Suther-

Partial Coalescence of Particles in Chains

An important step in the formation of a rigid gel is the strengthening of the interparticle bonds beyond the mere formation of a few siloxane bridges at the points of
contact, as shown in Figure 3.19.
When two silica particles unite in water, they grow together because the solubility
of silica in the crevice at the point of contact is less than that of the silica over the
surface of the particle. The neck joining the particles therefore increases in diameter,
until the difference in solubility becomes small. There is, of course, no true equilibrium, since the equilibrium condition would be represented when the two particles
became fused together, first into an oval and finally into-one large spherical particle.
The rate of thickening of the neck between the two particles will become very slow
after a certain point. This point can be estimated by the following reasoning:
As shown in Figure 1.8, the solubility, Sn, of a surface with a negative radius of
curvature is less than that of a -flat surface. Later in this chapter it is shown that
under given aging conditions very small discrete particles grow spontaneously until a
certain size is reached, after which growth becomes slow. Let this aged size correspond to a radius r a with a solubility of Sa' If two particles of radius R come into
contact as in Figure 3.21 the interparticle neck will grow to a radius r.,
Under the given conditions, the rate of growth of the neck will become very slow
when the effective negative radius oj curvature, r n, at the neck, is numerically
greater than a certain value. The solubility at the neck S; will be less than Stnt since
Sn/S/nt = exp(K/r,,) and r has a negative value.



Polymerization of Silica





.... _--""", "

Figure 3.21. Solubility at the negative radius of curvature of the neck between spheres is less
than elsewhere on the spheres: R. radius of spheres: rio positive radius of curvature of the neck:
r2. negative radius of curvature of the neck.

The effective radius,


in Figure 3.2\ is given by



where T" is the "effective radius of curvature," T 1 is the maximum positive radius of
curvature, and T 2 is the minimum (negative) radius of curvature.
For triangle ABC:

from which



+ 4Rr n (r n
(r n


ri -



+ 2r l

It seems logical to believe that if the growth of discrete particles becomes slow
when Sa - S, reaches a low value, then under the same aging conditions the neck
between two particles will likewise grow only slowly when Si - Sn reaches the same
low value. Thus

on of Silica

polymerization by Aggregation-Gel



Using the equation given later in this chapter for relating solubility to. particle
log S/


log - =

and eliminating S/:


heres is less
of the neck:


radius of

ornes slow
i the neck
; the same


Therefore when a gel is formed under certain conditions the degree of coalescence
or neck growth can be estimated if one knows the size to which very small, separate,
and discrete particles will grow under these same conditions.
As an example, assume that at 25C at low pH, silicic acid particles will grow to a
radius, r a, of 2 nm, If this aged sol then is gelled and further aged under the same
conditions, the radius of the necks, r h can be estimated as follows:
Taking the formula developed later for 25C, where K = 0.918 and To = R = 2,
and substituting in equation 4,
IOo.918/2 r n =


10.91814 =


whence r" = -0.89 and r 1 = 0.84 nm. Thus the neck diameter. 1.68 nrn, is 84% of
the original particle diameter. The particles are strongly coalesced.
On the other hand, if particles forming the gel are larger initially. for example, R
= 10 nrn, much less coalescence occurs when the gel is similarly aged. The values of
r a and r remain the same but the calculated neck radius becomes 2.2 nm. The neck
diameter of 4.4 nm is thus only 22% of the original particle diameter so the particles
are grown together to a much less extent.
In a gel, formed initially of discrete particles chained together, coalescence will
similarly convert the particulate chain structure into a fibrillar network, especially if
the particles are smaller than about 10 nm. As shown in Figure 3.22, the chain of
spherical particles (dashed lines) is thus converted into an irregular rod or fibril
(solid lines). If the particles are large as in one case in Figure 3.23. much less coalescence occurs under the same aging conditions where the smaller particles
approach a fibrillar shape.
The conversion of a particulate gel to one with a fibrillar structure of very high
specific surface area is represented two-dimensionally in Figure 3.24.
Although never demonstrated, the solubility of a freshly formed gel of very small
particles should be less than that of the sol particles from which it was formed. even
though there is little decrease in specific surface area. The solubility also should
decrease measurably during aging.


~ ...






/ r.'"












' ............. _-



f. : ,:


Figure 3.22. A chain of small particles is converted to a fiber or rod by the laws of solubility:
R. original diameter of spherical particles in a chain before formation of necks: R h final radius
of curvature of particles after neck growth: rio positive radius of curvature of neck: r2. negative
radius of curvature of neck; Rio radius of a rod formed by complete smoothing of chain (R, ..


,---:-7 II
". ~
.. -.1-----1----------



g .


j. .








Figure 3.23. Gels of equal silica

concentration, However, smaller particles
develop greater coalescence and strength
and the structure becomes more fibrillar.



polymerization by Aggregation-Gel Fcrmation,


al radius

in (R t =

Figure 3.24. Two-dimensional representation of

a dilute gel converted to "fibrils" by silica rearrangment.

Development of Microgel and Viscosity


str th

. The difference between formation of microgel and formation of a precipitate is that

in a microgel region "the concentration of silica particles in the three-dimensional
network is the same as in the surrounding sol but in the floes or aggregates of a
precipitate the concentration is higher than in the surrounding liquid. In the rnic. rogel the refractive index is the same as in the surrounding sol so the microgel cannot be seen. Likewise the density is the same so the microgel does not settle out even
when-centrifuged. On the other hand in the precipitate the refractive index and
density are both higher so that the 'precipitate is visible and settles out. The difference is represented in two dimensions in Figure 3.25.
The initial step toward .Iormation of microgel is the aggregation ~f some of the
particles into short chains as shown in Figure 3.26. As the chain length is increased,
particles begin to be added to the sides as well as the ends of the chains. Thus a
region of branched chains is created. Free particles within this region are added to
the network which soon contains all the particles within its domain, forming a "microgel region" of what has been called "gel phase." Each region of gel phase
increases in size by addition of sol particles in the surrounding fluid region until gel






Figure 3.25. Two-dimensional representation of the difference between a gel and a






ogo aoo


00 c::::::>

Figure 3.26. The gelling process involves formation of an increasing volume-fraction of the
sol being converted to microgcl (darker lines) with increasing viscosity until solidification
occurs. The three-dimensional process is repesentcd only in two dimensions.'


Polymerization by Aggregation-s-Gel Formati.on

resenta:1 and a


phase occupies much of the total volume. At that point the gel regions grow together
into a continuous network of coherent gel.
Gel phase in invisible and the sol remains clear. Within the microgel regions the
refractive index is essentially the same as that of the surrounding fluid sol. The silica
concentration is the same and thus there is no difference in density so that the gel
phase does not settle out.

However, there is an increase in viscosity as the gel phase develops. The increase
is at first small but detectable. Also the presence of even a fraction of a percent of
gel phase can be detected by the rapid plugging of a filter through which the original
sol, free of rnicrogel, passed freely. Also when a sol is being washed or concentrated
by being circulated through an ultrafilter, the membrane becomes blinded rapidly if
even I % of the colloidal particles have been converted to microgel.

Isolating "Gel Phase" or "Microgel"

The following method is so simple that it is surprising it has not been used previously. A sample of sol that is in the process of gelling but has not reached the gel
point need only be diluted, for example, threefold, and centrifuged under a gravity
just insufficient to settle single particles. Since the microgel does not expand, it
retains its original specific gravity whereas the ungelled sol medium is diluted out
and has a lowered specific gravity. The gel phase is thus easily separated as a solid
cake. The proportion of the silica found in the cake is a measure of the fraction of
silica in the microgel at the moment of dilution.
In carrying out this test, it is best to use water containing the same pH and
concentration of salts (if any) as in the original sol. To ensure that no further gel
formation occurs during the centrifuging it may also be necessary to quickly adjust
the pH of the sample sol. to 1.5-2.0 with HCI or HN0 3 before the dilution. This stabilizes the system long enough for the test.

Effect of Electrolytes and Coagulants

In ofthc.:

This discussion refers to homogeneous gels formed by uniform solidification of a sol.

Such gels are formed below about pH 6 whether or not a salt such as Na 2S0 4 is
present, and above pH 6 only if the salt concentration (sodium) is not too high. In
hot solution containing more than about 0.3 N salt, silica is precipitated rather than
Colloidal particles join into chains and thence into three-dimensional gel networks'
only if there are no forces present to cause particles to coagulate or come together
into aggregates which contain a higher concentration of silica than in the original
sol. Coagulants such as metal cations. especially polyvalent ones. or some organic
bases. especially cationic polymers, lead to precipitation rather than gelling.
Thus if a silica sol contains substantial amounts of salts and is neutral or alkaline,
a homogeneous gel is not obtained: instead. the silica usually appears as a white
precipitate, or in some cases as a white opaque gel owing to partial precipitation
before gel formation. Similar effects arc observed when small amounts of coagulants

:- ...

Polymerization of Silica

or flocculants are present as, for example, water-miscible organic liquids .polymers,
or surfactants.

Gel Density and Structure

Manegold (127) has considered what kinds and packings in regular network arrangements result when the "coordination number" of particles is varied, that is, the
number of particles linked to a given particle. In most densely packed gels, each
sphere is touched by 12 surrounding spheres, and the volume-fraction of spheres, S,
is 0.745. In a more open regular network, each sphere is connected to three others
(Le., a coordination number of 3) and S is only 0.05, as in Figure 3.27.
However, there is a tendency also for particles to form short chains in which the
coordination number is 2. Thus, along any series of particles forming a network, the
coordination numbers could be 3, 2, 2, 3, 2, 2, etc., in which case S = 0.0083. This
would correspond to a gel formed from a sol containing only 0.83% by volume, or
1.8% by weight of Si0 2, as in Figure 3.28. However, if the particles are small enough,
a silica sol can form a weak but continuous gel network at even 1% Si0 2 or less. In
this case, the coordination numbers may be 3, 2, 2, 2, 3. A model of such a very
open network in which the average coordination number is of this type is shown in
Figure 3.29.

! ;: ~ i



, . ;.

. ..,--~~


..... :'



." . i-~;;... ;':;~:~~< . ;..:.:.~.-~<~::.(} ... ;..~-;':2.-:-":; .: , ..: .


: .:

Figure 3.27. Packing of spheres with
coordination number of three. [From Ilcr
. (89). by permission of Cornell University
Press. J

i .






i, S,
:, or

Figure 3.28. Packing of spheres with

coordination numbers of 3, 2, 2. 3. Black
spheres touch three others.

n in


'sit ,.


Figure 3.29.

Gel with particles with coordination numbers of 3. 2. 2. 2. 3.


Polymerization of Silica




Increase in Viscosity

The question arises whether in the sol-phase regions when microgel is already
present, new aggregates continue to be developed initially as twins or triplet chains.
These could not contribute appreciably to viscosity until, through network formation, they enclose water in new microgel regions. Within any microgel network trapped particles, whether single or short chains, will soon join the network. Thus within
each microgel region the final gel structure is rapidly attained.
The viscosity during the process of gel formation increases in proportion to the
volume fraction of gel phase present. It will be .assumed that the diverse microgel
regions can be considered as having an average size which increases as the remaining
sol particles are added to the periphery of each microgel aggregate region. Once a
substantial volume fraction of the sol has been converted to suspended microgel
regions, the free single, doublet, or triplet particles then add to the periphery of each
microgel region, increasing its volume at the expense of the still fluid medium
without changing the silica concentration in either fluid or gel regions. We know that
this is so because at any point during the process of gel formation as evidenced by
the increasing viscosity, there is no large increase in turbidity and no gel phase can
be separated by centrifuging unless the sol is diluted.
The volume of microgel phase can be estimated from viscosity on the assumption
that the microgel regions are spherical in shape and of approximately uniform' size.
The volume fraction of microgel phase (which is also the fraction of silica immobilized in gel) can be calculated from the Mooney equation (128a):
. . 2.5C
In n, =
1 - k 1C

where k, = 1.43 and n, is the ratioof the viscosity of the gelling sol to the viscosity
of the sol before the gelling process began, and C is the volume fraction of microgel
present. For convenience, values from this equation are plotted as log(n,. - 1) versus
log C in Figure 3.30.
As the volume fraction of microgel .approaches about 0.5, the viscosity rises
rapidly because the microgel particles approach random 'close packing. However,
since the microgel regions undoubtedlyvary in size. so that smaller gel fragments fit
between larger ones: actual solidification may not occur until the microgel volume
fraction. determined directly, is greaterthan 0.5.
As mentioned earlier. after an early stage the rate of increase in microgel may be
proportional to theamount already present. growth of each gel region being due to
addition of single. particles .or short chains. which have enough kinetic energy owing
to Brownian motion to collide with the microgel regions and add on. Then



-. Silica

Polymerization by Aggregation-Gel




where t is time and C is volume of dispersed rnicrogel phase, and so




c trapwithin
to the
)nce a
If each
w that
:ed by
se can
n size.





where Co is the amount of microgel present at time zero.

To investigate the formation of microgel, a 15% silica sol of 6 nm particles was
deionized by Iler to pH 5.55 and aged at 30C. Samples were taken at different
times, diluted to 4% Si0 2, and deionized to pH 3.5 (to leave little or no charge on
particles), and the viscosity was determined. From this the volume percent of "dispersed phase" in the 4% sol was calculated from Figure 3.30 and then multiplied by
15/4 to find the volume fraction in the original sol before dilution.

Time (hrs)

Volume Fraction
C in 4% Sol

Volume Fraction
C in 15% Sol


k = 0.055

____0 _. __
I 2.5

When these data are plotted in accordance with the above equation. as in Figure
3.31, it is seen that the equation is followed up to 0.50 volume fraction of dispersed
phase. Above this point, the equations relating viscosity to volume of dispersed
phase are not valid. Also the gel network becomes so extensive that i~ is irreversibly
disturbed by dilution and measurement.
The activation energy for gelling varied with pH:


nts fit
lay be

Iue to


kcal mole-I



Also. the use of a hydrophobic cation base (CHJ).NOH- instead of Na Ol-l ' at
pH 5.5 reduced the activation energy to about 8.0 kcal rnole " '. When this cation is
adsorbed on silica there is hydrophobic bonding between the silica particles and gel.
thus holding the surfaces 'together for longer times upon collision. and promoting
siloxane bond formation .. :;i



( n, -( )

(n, -I)











. i






. 0.05






Figure 3.30. Plot of the Mooney equation relating the volume-fraction of uniform spheres in
suspensions to viscosity.





z 0.5

~ .....


~ 0.2




Figure 3.31. The vol~~e-fractio~'o("dispersed phase" or microgel increases as a sol is

transformed lO gel. Sol: t 5% Si0 2 of 6 nm particles at pH 5.55 and 30C [Iler, unpublished.]






Polymerization Above pH 7


Formation of Larger Particles by Coacervation

When the concentration of silica is low, generally under I %, and under certain conditions of pH and low salt concentration, the primary particles form small spherical
aggregates or secondary colloidal particles. In aqueous solution this is much more
likely to occur at ordinary temperature than in hot solution where growth of primary
particles with decrease in their numbers reduce the probability of aggregation.
. Such a mechanism must have been involved in the formation of 200 nm spheres in
a solution of pure silica sol prepared by hydrolyzing SrCl, and removing HCl by
electrodialysis, as reported by Radczewski and Richter (128b). The purified clear sol
contained about 0.5% SiOz and the pH was 6.8. Similarly, uniform porous spherical
silica particles up to I micron in diameter are formed by the aggregation of primary
particles less than 5 nm in size formed by the hydrolysis of ethyl silicate in a
water-alcohol-ammonia system as developed by Stober and Funk (128c) .

. . . . ." , ' ,

.., .. '

. . . ',




pheres in

In the foregoing section it has been described in detail how at low pH silica
polymerizes in stages to nuclei of silica which then increase in size to 2-3 nm. These
particles then aggregate into chains, forming higher molecular weight polymers.
Above pH 6-7 the behavior is entirely different. The ionization of polymer species is
much higheriso 'that"monomer polymerizes and decreases in concentration very
rapidly, even in a matter of minutes at 25C. At the same time, the particles grow
rapidly toa size that depends mainly on temperature., However, no aggregation or
chaining together of particles occurs because they are charged and mutually repel
each other:
All this is true provided the electrolyte concentration is low, as when silicic acid is
prepared by. hydrolyzing an 'ester or when the sodium' or potassium cation is
. .. :' :':"
removed frorna silicate solution to give pH 6-10. ,.
Only above' pH 7 can silica particles of suitable size be prepared and concentrated
, .. for industrial use,a;; discussed in Chapter 4:

Spontaneous Growth of Particles

* ",

a sol is

- .

; .:.


Almost all the initial stages in thepolymerization or monomer to oligorners and threedimensional particles 1-2 nm in diameter, occur in only a few minutes at above pH 7.
Thereafter the polymerization involvesonly the increase in size and decrease in num. bers of silica particles. At the same time, the concentration of monomer in equilibrium with these particles decreases as the average particle diameter increases.
Under conditions where the relative rate or aggregation of silica particles is low;
the particles increase in size and decrease in numbers owing to "Ostwald ripening,"
a term' commonly applied to the coarsening or crystals in a precipitate such as
8aSO.; As described in Chapter I, small particles, especially smaller than about 5
' .l


.. ~ .




Polymerization of Silica

nm, are much more soluble than larger ones. According to available data, particles
of Si0 2 formed at 25C and not subsequently heated have the solubility of Curve B
in Figure 3.32, whereasparticlesmade by hot processing (80-100C) appear to
follow Curve A. ' , ';
The rate of growth depends on the distribution of particle sizes since the growth
occurs by the more soluble smaller particles dissolving and the silica being deposited
upon the larger ones. Generally, the distribution of particle sizes is not known but,
.with Gaussian distribution assumed, the increase in average particle size continues at
a given temperature until the size is so large that the difference in solubility of
, smaller, versus larger particles becomes negligible. The final size depends on the
, ,:,:,'. ,_
The kinetics of Ostwald ripening has been examined mathematically by Sugimoto
(128d), who derived equations for the change in particle size, distribution with time.
lIer has found that with a fixed Si02 : Na 20 ratio the size, to which particles grow is
independent of the silica concentration. Silicic acid solution (3.7% Si0 2 ) was made
by passing silicic acid through a H+ ion-exchange resin, adjusted with NaOH to a
Si0 2:Na 20 ratio of 190: I and vacuum concentrated at 35C,to 7.7%. Samples were
diluted to lower silica concentrations and heated at lOOC for 48 hr and the particle
sizes then determined bysurface titration:

:: .....

% Si0 2 '


Initial pH, : : Final pH"

Diameter (nm)

7.75 '
8.59 , .':
.,3.0 ~; -~" ' .: 8.91
1.0".:, /1; , 9.26

9.9 ,

13.5 '

, 13.1,

II! ;,a given, ;~o~" ~i :;~-:~i~e'n' temperature, ' particle -size' appears to approach
asymptotically a final value that depends on the temperature, as shown in Curve A
'.' ,ofFigure,3.33. It might be supposed that particle growth would continue over long
periods of time until the specific surface area was reduced, to zero," However, this
does not occur. Sols of 15 nm particles have been stored at ordinary temperature at
pH 9-10 for more than 20 years with only minor increase in particle size.
Theoretically, if all the particles were exactly the same size, no growth would occur,
but such was not the case, since some particles were at least three times as large as

others. :,'






, Experience has shown that alkali-stabilized sols of silicic add polymerize during' a
few months 'at' 30Cuntil'thesurfacearea slowly drops to about 500m 2 g-I, corresponding to 5 nm: particles, after which there is little change.' At 90~C, the particles grow in a few days to around 7-8 nrn, after which spontaneous growth is very
", - :."
,. :


Table 3.5 includesdata by Alexander and McWhorter (129) and Broge and lIer
, (130),' who 'studied the' particle growth 'during relatively long heating times at high
"-temperaiures"uri'der' superatrnospheric' pressure. The latter authorsfound that when
the alkali-stabilized sols were heated to above 300C, quartz crystals were produced




__ .

f Silica

urve It.

LOG 10

ss /s, =0.0191


in but,
nues at
.lity of
on the


A: Sj = 85 E = 53.5

c. 300

B:Sj =70. E=48.0







.1 time.
~row is
; made
H to a
:s were




2 3 4

Figure 3.32" . Effect oi'particle siz~ on solubility of amorphous silica in water at 25C: A,
particles heated at 8~-:loooC: B, particles formed at 25C but not heated [Iler, unpublished.]
: ,

, Table 3.5. Growth of Silica Particles by Heating a 4% Sol of Silicic Acid at

,pH 8-10

, v.;

Mole Ratio
Si02:Na20 ,':Time


urve A


er long
sr, this
ture at
e size.
irge as

iring a

corpartiis very



. : -.:


;100.1 hr . 8 0

: ;-::'i~

~c ::,'.~.~~ ;.~. ::,,;:;:~;' ~.:;



6 hr
80 .'- ': 30 min
85'> '{3 hr :
: ' :,270
85 .<:. "'3.25 mfn
0.9 min
'. 250
.. ' 200
.: , 9 0 , ' 3.1 min
10 min
10 min
30 min
Very high"
3 hr
Very high"
6 hr" ... ~. ~'-" 340


td Her
tt high
t w' 1

,', ~. - i.(OC)

',Very high",. 3 hr
- ;....




Specific Surface Area

, (m" g~l) .






:J'CI'Traces 'of sodium ions remaining in the starting particles after deionization of
the sol before autoclaving resulted in a pH of about 8 in the /inal sol..


.,. .........



Polymerization of Silica










if 5



8 0







Figure 3.33. Growth of silica particles at

pH 8.5: concentration of SiO:. 3%; SiO::
Na 20 . 102. (A) 90C; (8) 50C. [Iler,

instead of stable colloids. Rule (131) found that if a so] consisting of particles from 8
. , to 15 mu in diameter is first thoroughly deionized and then autoclaved at 300-350C
under autogenous pressure, there. is insufficient alkali present to cause conversion to
quartz, yet traces of sodium and hydroxyl ions occluded in the original particles are
released during the growth process in amounts sufficient to raise the pH and stabilize the sol. In this manner, particles in the diameter range 88-150 m,u diameter were
produced, and such products could be concentrated to a stable sol containing more
. than 60% by weight of silica. It is pointed out by Biage and lIer (130), that if the
starting so] consists of discrete particles they grow as discrete spheres, but if they
start as aggregates or as a gel structure, the final sol particles are of irregular shape,
since they are formed by the densification of irregular aggregations.
particlegrowth Isnot proportional to concentration of OH- ion in the
,- -- 'The rate
pH range 8-10, but is only moderately influenced by it. Since the silica surface is a
poor buffer, the pH rises as Jhe particles g-row and specific surface area reduced.
Shown in Figure 3.34 is a sol starting with particles 3.5 nm in size made by deionizing a solution of sodium silicate and adjusting the Si0 2 : Na 20 ratio to various
values. When held at SO or 90C the particles grow very rapidly in the first 5 hr from
the 3.6 nm size they had reached at 25C. At 50C in 100 hr they grew to around 5
om and at 90~C to' 7-8 nrn, after which further growth was very slow when the pH
was 7-8, but continued somewhat further when more alkali was present to give a pH

Final Size of Particles Versus Temperature

- Obviously the rate of particle growth at a given tern peratu re becomes exceedingly
'slow when the difference in solubility of the smallest versus the largest particles has
i; , :..
, .
become negligible.


. .1

.: - I







f Silica

icles at
; Si02:
. [Iler,

Polymerization Above pH 7


Thus at 90C, for example, the solubility of massive silica formed is ~bout 350
ppm, and the size above which growth is slow at 90C is about 8 nm. If we assume
growth occurs until all the particles are within the size range of == 10% or 7.2-8.8
nrn, then using the expression 101.03ID, the range of solubility isbetween 1.39 x 350
and 1.31 x 350, or 487 and 45~, that is, a solubility difference of 28 ppm.
At 30C the solubility of massive silica is 70 ppm (Si0 2 made at 85C) and
growth becomes slow at 3.5 nm diameter. A similar calculation, assuming the
expression lOu 2l u , gives for a == 10% spread in particle sizes a range in solubilities of
137-121, or a difference of 16 ppm.
It is therefore clear that the initial distribution of particle sizes about the mean in
the initial sol will have marked effect on the final size reached when the sol is aged
. at some higher temperature. Aggregation may occur when a 2-4% silicic acid sol is
made at pH 2-4 and then alkalized. At the low pH, and during the change of pH
. through the range around 5, colloidal aggregates or microgel are likely to be formed
unless the operation is conducted rapidly. The aggregates may then act as larger
particles or nuclei and the heat-aged sol will have a larger final particle size.
. For a quantitative study of spontaneous growth. data are needed on the particle

'rom g.

.ion to
les are
stabim .


if the
f they

in the
ce is a



. from
lund 5
he pH
:a pH

...... :. 3

c- -::-,

es h"<;




Figure 3.3t

Rate of growth of silica particles at 50



Cur~c :



.' 100 - 200.




'Temperature (OC)




., ..... - .


.. "


.... -

e and 90 e and different pH


sio.. NazO






POlymerization of Silica

size distribution in alkali-stabilized sols of particles 3-15 nm diameter. Eventually

such data may be obtained by stabilizing such sols at pH 2, diluted to around 1%
concentration, and measuring the size distribution by ultracentrifugation or liquid


: 1

This is of relatively minor importance in the overall view of silica behavior, yet must
be considered in interpreting viscosity data obtained during early stages of polymerization.
When silica first polymerizes to very small discrete particles at .low pH, the
increase in viscosity can be appreciable because so much water is associated with the
surface of the particles. As polymerization progresses with increase in particle size,
the viscosity' decreases, which may seem surprising. Thereafter the viscosity
.increases again when the particles begin to aggregate. This initial viscosity of the sol
of discrete particles must be taken into account in calculating the development of the
gel phase, which has already been discussed.

Viscosity of Sols of Very Small Particles at Low pH

At the point where most of the monomer has been converted to particles but where
these have not yet begun to aggregate, the viscosity of a sol may be calculated from
the volume fraction of the dispersed phase, which is defined as the silica particles
along with the water that adheres to the surface by hydrogen bonding.
As shown by Dalton and Iler (132), there is a monolayer of water molecules
hydrogen-bonded to the SiOH surface. From the standpoint of viscosity behavior
this layer acts as part of the dispersed phase.
Based on the formula for thecomposition of a particle developed in Chapter 1. the
formula for a particle of diameter d nm is

where the OH is presentassilanol groups on the surface of the particles. For each
OH there is' a hydrogen-bonded H 20 so that from the standpoint of viscosity
behavior, the composition of a particle of the dispersed phase is

where d is the diameter of an equivalent anhydrous Si0 2 particle in nanometers.

More simply, the formula may be written

From this formula the following can be estimated:

Viscosity of Sols Before Aggregation Begins


equivalent molecular weight as anhydrous (Si0 2 )z = 690 d 3


664.2d 2

weight ratio H 20: Si0 2

690d 3

0.963 d :'

volume of dispersed phase per gram of SiO z .;, 0.45 + 0.93d- 1

Knowing the silica content ofa sol, and the particle size one can calculate the
volume fraction of the dispersed phase. for example, 1 ml of a 2% SiO z sol of 1.5
nm particles contains 0.022 ml of dispersed phase. From the Mooney equation
(Figure 3.30) the value of n, - I is 0.055.
As an indication of the degree to which the viscosity of a sol at pH 1.5-3 can be
affected by particle size at different concentrations, typical calculated values are
shown in Figure 3.35.
It is obvious that a few percent of monomer in water would not contribute appreciably to viscosity, for example, any more than a small molecule like glycerol would.
It is probable that below a certain particle size, probably between I and 1.5 nrn, the
polymer no longer acts as a particle, but rather like liquid oligomers. Thus lIer (116)
found that the viscosity of polysilicic acidsolution containing 6.34% SiO z at pH 1.7,
derived from 3.25oratio sodium silicate, had a viscosity of 1.11 relative to the sodium
sulfate solution medium. This increased to l.15 as the degree of polymerization
(cryoscopic method) increased from o~ to 23, the latter corresponding to a 1.25 nm
in Figure .3.35, for the 6.34 g per 100 ml
particle of anhydrous SiO z As can be seen













Figure 3.35. Calculated decrease in viscosity of a sol of very small discrete particles owing to
" particle'growth:n;; relati~c viscosity: Concentration ofSi0 2 g perIOO,oml: A,'IS: B, 6.34: C.
2.0. L.large particles (sec text).

0 '

0 '


Polymerization or Silica

sol, the calculated value (solid line) for a 1.25 nm particle would be considerably
higher than this observed value, indicating that when particles are smaller than 2 nrn
diameter they do not have as much effect on viscosity as larger particles do. When
the particle size exceeds about 2 nm the contribution of such small silica particles to
viscosity is considerably greater than that of larger particles.
An indication of the type of situation that could arise if careful viscosity studies
were made in the initial stages of polymerization at low pH and at a concentration
where particle development occurred before aggregation began is shown in Figure
3.36. Here the logarithm of the reduced viscosity is plotted against the logarithm of
molecular weight, that is,"a Staudinger equation plot. The calculated values for the
viscosity of small discrete particles are shown as A. The values observed by Acker
(133a) for the aggregation stage are shown as B. There must be a transition region
as shown.
" .
To calculate the volume fraction of microgel in the course of a sol setting to a gel,
it is necessary to express the viscosity relative to that of the sol at zero time rather
than to the aqueous medium alone. Thus Acker (133a) presents data on a sol of
silicic acid made by neutralizing a 3.25 ratio sodium silicate solution at pH 1.5. It
contained 15% Si0 2 and gelled in 110 min. The initial viscosity was 5.4 cp. The
relative viscosity "at subsequent times can thus be expressed as the ratio of viscosity
at time 1 to that at time zero. Taking these values for TJtlTJo the Mooney equation can
be used to calculateC, the volume fraction of the dispersed phase (rnicrogel). This
interpretation of Acker's data is shown in Figure 3.37.
It is noted that after about 30 min. at which Acker reports a molecular weight by
light scattering ~f 14,000, there is ~.!it relation betwe~n log ~ and I. This relation
was observed with the J5% sot ~erete 6 nm particles (FIgure 3.31) whereby
the' large growing aggregates as a first-order
particles or small aggregates add
reaction. Acker similarly proposed that the larger molecular weight species bond
with the low molecular weight species, but assumed polymerization was linear based





- ....




;: 0.10.
"- - - t::-- .en






Figure).36.' :Cha'nges in reduced viscosity of polysilicic acid sol due to particle growth and
aggregation: A/single particles. diameter in nanometers: B, aggregates, 'molecular weight in
daltons, [From Acker (1333).]

: Silica


Viscosity of Sols Before Aggregation Begins





'des to




thm of
for the





LL. 0.05




o a gel,
_sol of
1.5. It
:p. The
ion can
I}. This

.ight by
'el' '0
:s bond
r based

> 0.02




..': ;.







Figure 3.37. Volume fraction of microgel versus age of sol during the process of gelling. [Calculated from data by Acker (133a).]

on viscosity behavior.. .However, as shown above, the viscosity can be equally well
accounted for on _the basis of growth of rnicrogel regions and the solidification to a
continuous gel network when the. microgel reached a' volume fraction of 0.5 under
. ,
,- these particularconditions. ' "
: _

. --"




Decrease in Viscosity on Conversion of Microgel to Sol

iwth and
veirl- in

When microgel is formed from extremely small particles at low concentration at Jow
pH, it may disperse and form a sol when the pH is raised above 7. This occurs only
if the salt concentration in the system is less than about 0.2-0.3 N, that is, the
critical coagulation' concentration. Otherwise the growing particles remain aggregated and become further cemented together as a precipitate.
When ~the pH is 'raised,there is particle growth as chain segments dissolve and
silica is deposited 'around branch points, eventually forming irregularly shaped,
separate particles. "At the same time the specific surface area of the silica is greatly
, .Pfleger and Kautsky (133b) observed this phenomenon in silicic acid sols that had
been aged and developed .high viscosity as they approached the gel point a't low pH.
When the pH was raised to 8-9 the viscosity dropped to a low value as the sol was
.aged -fora month at room temperature. The particles thus formed at ordinary
temperature were not fully internallycondensed since they were porous enough to
still exchange H - for CuOH - ions. As described in Chapter 4, there is a commercial


Polymerization of Silica

process for making sols by washing the salt out of gel made from acid and sodium
silicate at low pH and then autoclaving the gel at pH 8-10.



i '

; ,

Energy ofActivation
In view of the different polymerization processes occurring simultaneously, it is
generally not possible to link a temperature coefficient to a specific reaction.
Flemming (134) observed that the temperature coefficients of polymerization were
different in acidic and slightly basic solution. This was verified by Hurd and co
workers (135-138), who found that the temperature coefficients were less in very
strongly acidic solutions than in weakly acidic solutions, namely, 9-11 versus 24 kcal
mole:". More recently, Penner (139) found that in the pH range from -0.441 to
0.863, where it is known that the reaction rate is proportional to the concentration of
H+ ion, the activation energy was about 9.5 kcal mole-', whereas at pH 4.64, where
the rate is proportional to OH- ion, it was 16.1 kcal mole';'. This strongly suggests
that two different mechanisms might be involved in these different pH regions.
Brady, Brown, and Huff (140) found that the activation energy of polymerization is
15.5 ::I:: 0.5 at pH 5.5, 9.6 ::I:: 0.3 at pH 8.5, and 14.6 ::I:: 0.5 at pH 10.5 in the presence
of potassium chloride.
Coudurier, Badru, a'nd Donnet(36) obtained separate values for the energies of
activation er the condensation 'process (disappearance of monomer), 14.6 kcal
, mcle:", and of theaggregation pr'ocess,IS.I kcal mole:", at pH 4. ;
Bishop and Bear (141) followed the polyrnerlzation of monomeric silica at pH 8.5
at 25:-45C by measuring the decline in unreacted monomer using the molybdate
method. The initial rate constant, assuming a second-order reaction, showed a
peculiar variation with. ternperature; .'

',i,"' ' " " ,


- ; .. '



,.:; ,'" (OC) , '

',' 25
" , ,30
, 40
',45 '


,., ...



<s, X



._. ':

10',; r - : . " '

,(I molet-sec")

38::1:: 4





, Th~ drop iriactivatione~ergy,'in the, 25-35C region implied that. there was a
"preequilibrium" step in the polymerization. It is likely this involved the same type
of induction 'period observed by other workers under similar conditions at 25C and
the 'first step is; the' formation of some small polymeric species with which the
monomer' thenreacts preferentially.


..-. :..; .-





of Silica

Summaries of Investigations


Heat of Polymerization

Iy, it is
on were
and coin very
24 kcal
).441 to
ation of
~, where

.ation is

rgies of.


pH 8.5
owed a



Since solutions of silicic acid gel suddenly under certain circumstances, it might be
expected that a considerable heat of reaction might be involved. This is not the case.
Tourk.y (l42) found that when silicic acid was made from sodium silicate and acid
and polymerized in neutral solution. the heat of reaction, excluding the heat of neutralization ofthe acid by the base, was about 148 cal g-1 of Si02, corresponding to
8000 car mole:", In this case. the starting material was sodium 'metasilicate; hence
this probably represents the overall heat evolved in the transformation from monosilicic acid, to high molecular weight polymer.
': .
On the ather hand, Treadwell and Wieland (l43) found that during gelatinization
or a sol there was practically no thermal effect. no change in the refractive index,
and no change in electrical conductivity.' They concluded that the gelling process
cannot be a, normal change in liquid viscosity, since it did not cause a change in conductivity. The lack of a heat effect at the point wherethe viscosity increased rapidly
just before the solution gelled indicated to the authors that the polymerization of
silicic acid first goes from the molecularly dissolved oligomers to colloidal particles
of high molecular weight. Then, if these particles are not stabilized by sufficient
ionic charge, they forf!} themselves into ,"threads" or chains and by floc formation
occlude thedispersing medium (water) whereby .theviscosity is increased. At the
point of solidification or gelling, the solvent is completely occluded. :;
" ' However, in Treadwell, and Wieland's experiment, the' sol was made by elec"
trolysis on the alkaline side and we know now the 'particles were made larger than in
".: :.Tourky's sols" so that far fewer'siloxane (SiOSi) bonds 'were' formed during gelling,
and thus much less heat evolved.":_';','
.'_ _; -,:' '.~',',-,~ .'::
;, ..,. ' No accurate data exist for the heat evolved in the reaction in solution: . '


. r .... ..,..

, >

~ '":' ; ~ '.., _~ _

,,- " : ::.,



~,;. '.,



:>. :.-:;.~..


t. '.







.' ,


:. .






~ i-

__. ;

,," . .

_, _,,,

+ ~20.:

. ; ~ >-.: ". ~"-. ; ~ r..-";l~_"_:-.-)

. ' ,:'." ".'
: :'Howe~er, -the e~terit"of condensation in a silicic acid system has been followed by
":'"Tourky (144) by noting the-change in volume by a dilatorneter. ~A combination of
' ; ..>this techniquewith calorimetrymight permit an estimation of the heat of reaction if
one were to start with a solution of monosilicic acid.





was a
ne type
C and
ich the

The foregoing description of the polymerization of silicic acid was based on a uniform interpretation of the data of about 20 investigators who studied polymerization
~ under different conditionsby different methods and interpreted their results in differentways.- From the following separate discussions of each investigation it will be
seen thai thedata are consistent with the view that at an early stage in the polymerization, the sllicic acid is converted to spheroidal particles about 1-2 nm (10-20 A) in
diameter. The data of severalearlier investigators confirm the most recent work that
. : ....: -:.
.. . ' . .





Polymerization of Silica


shows clearly that nucleation plays a role in the formation of amorphous silica particles as it does in the formation of crystalline precipitates.
The amorphous particles' then grow' in size and decrease in numbers by the
phenomenon of Ostwald ripening. as has long been observed with very fine
crystalline particles. Under mosj conditions further increase in "molecular weight"
involves the chaining together of the particles into growing microgel networks until a
gel is formed. However, insalt-free alkaline solution. discrete particles continue to
grow; forming stable 'sols. ,': ,
The problem has been that the different processes have overlapped and so were
not easily recognized.
' ' .:
'" '
The lnvestigationsarediscussed in the or~er listed in Table 3.6.






:~.:. J"~







Alexander' '. .,;,:,:,~ ;.', ""~:'

0': -',.
:..", 2-~=: ....~ .."._ ..
.~.i .. Alexander (24a) was the first .to follow the polymerization of monomer in the
absence of salts. At pH ,2, the number-average molecular weight increased linearly
with the square root of time. asreportedby Her (116).At pH 3.2 and 3.8 it increased
"in proportion to time. At pH4.36 it increased linearly with the square of the time, as
inasecond-orderreaction.. :~,.~>.~:;; :"'.'-.:1~~~~:>~, .. '-.
,a, ... <.~.,
However. at a given degree of polymerization of 3.5, the polymers formed at dir<

. ,:-;






Iler (116) studie~fthepoly;nerizatiori ~r~~ess at silica concentrations sufficiently high

, that the increaseinmolecular weight mainly involved the aggregation of small parti'cles. He prepared 2':6% 'sols' from 3.25 ratio sodium silicate by acidification at pH
.. , 1.7, where salts have negligible effect on polymerization. ",'
The degree,' of 'polymerization; calculated from the number-average molecular
weight. increased linearly with' the square root of the time. Also. the time required to
reach' a given deg~~e of 'polymerization was inversely proportional to thesquare of
the silica concentration.. Gelling occurred when the number-averagem.olecular
weightapproach'cci:a'-degree :ofpolymerization of 100' regardless' of the' silica
concentration. The viscosity increased rapidly after about 30 hr,: at which point the
relative viscosity indicated that the volume of the dispersed gel phase had increased
to occupy one-third of the total volume.
" These' observations" are -consistent with..the view that discrete silica "particles"
containing 10-20 silicon atoms were formed early in the process, as evidenced from
the low. constant viscosity which did not -begin to increase until the number-average
degree of polymerization reached about 16. Thereafter. aggregation proceeded until
the volume fraction of the microgel phase approached 0.50. Even at that point. half
the silica was still present as low molecular weight species. so that the num beraverage molecular weight was still low..-: , "









Investigations at Low pH

':F' .,'
.. ~'! ,;



: !'







Summaries or Investigations


ferent pH values were not alike. The higher the pH at which they were formed, the
more slowly they reacted with molybdic acid:

'y "- ..





2.1 .

mtil a
iue to



Reaction Rate Constant


This is now interpreted to mean that the polymer particles initially formed were
internally more completely condensed and crosslinked at the higher pH. Polymerization involves three simultaneous process:s:
1. Reaction of monomer and dimer with themselves and higher polymers.
2. Ring closure and further condensation of silanol groups within the polymer, leading to particles.

3. Aggregation of particles into chains and microgel.



Because temperature can affect each of these mechanisms differently, it is not

possible to interpret the temperature- coefficient as an "energy of activation" since
two or more mechanisms operate simultaneously.
Some idea of the size of the discrete silica particles in the system is obtained as
follows. Alexander reported that when the 0.1 M solution of monomer at pH 2.2
was heated 30 min at 90C and cooled, the overall degree of polymerization (D.P.),
calculated from the number-average molecular weight, was 'only 2.4. This was
-because 40% _~of the silica was still monomeric. The remainder was polymeric
. (colloidal). The monomer was' present at a concentration of 2400 ppm, whereas the
solubility of massive amorphous silica is 'only 380 ppm at 90C. This can be
explained by assuming that the monomer was in equilibrium with small colloidal
":. particles. The. supersaturation ratio relative to massive silica was 6.3. For silica
formed at this temperature the ratio is 1OD.92IO. where D is the particle diameter in
nanometers. According to Figure 3.32, the particle diameter is!.3 nm.



red to

at the
, half



-Schwarz and Knauff

-. -

. _Schwarz' arid Knauff (21), noted a strange ternporarily limited polymerization of

--. monomer. 'After about 150 hr at pH 3. 75 hr at pH 2. at a concentration of 0.8%
Si02 the number-average molecular weight became essentially constant at about
:-- 300 corresponding to a D.P. of 5 (Figure 3.38). The authors concluded that the
: system was stabilized by ring closures forming cyclic pentarners or hexarners.
.~ One can calculate the size of the particles, assuming the concentration of
. monomer corresponds to their solubility.
-- Let it be assumed that at this point there are n polymeric or colloidal particles of
'; diameter, l!. _nanometers having a solubility of S/) ppm of monomeric or soluble
. '.' ~- silica. From Chapter 1 itIs calculated that each particle contains 11.5 D3 silicon,

n the
ne, as
,t c'

. . .&




.... ;


.. ,.

. ...





..". ,
':'. :" r.: ::, .::......~:.;':t:',:



.. _....








Table 3.6.



'So~e InYesti~~tions of Polym'~rization of Siiica 1950-1975' ,



, '.




~ .o'



Methods Employed

' .

;. Conditions .

Mol. Wt.

~. " . ~

I nvestigators

At LowpH .
lIer (116)
Alexander (24a)
Schwarz and Knauff (21)
Bechtold (145)
Goto (45)
Okkerse (29)
Audsley and Avestori (147)
Weitz, Franck, and Giller (148)
Bechtold, Vest, and
Plambeck (149)
Acker (133)
Hoebbel and Wieker (84)


_~ ..





Plambeck (149)
Hoebbe . .J Wicker (84)





0.05 N







Greenberg and Sinclair (155)

Greenberg (156)
Goto (157)




Tarutani (83)
Iler (Unpublished)



Through Neutral pl/ Range

Merrill and Spencer (150)
Ashley and Innes (151)
Baumann (152)
Coudurier, Badru, and
Donner (36)
Marsh. Klein. and
Vermeulen (32)






Gel chara


Polarized light

Above pH 7

Iler and Sears (158)

Richardson and Waddams (159)
Makrides et al. (106d)







0.5 (?)


95C nucleation







Polymerization or Silica















Figure 3.38. limited polymerization of monosilicic acid (0.8% SiO z) from hydrolyzed methyl
orthosilicate in different hydrolysis and aging media at 25C. Molecular weight: number
average as sio, Curve I. 10- 2 N HCI: Curve 2. 10 -~ N HCI: Curve 5. 10-' N HzSO.o [From
Schwarz and Knauff(21).]

atoms so that in 1 liter containing n particles there are 11.5 nD3 silicon atoms. In 1
liter there are also So ppm of monomeric Si0 2 or 1010 So silicon atoms. The total is
then 11.5 nD3 + 1019 So Si-atorns per liter. Since the total concentration is 8000
ppm. or 8 X 1022 atoms 1-1,

11.5 nD3 + 10111 So = 8


The relation of solubility to diameter (Curve B. Figure 3.32) is

From the D.P. of 5.


10 So

+n =


1.6 X 10 22

Solving the three equations gives D = 0.9 nm.

This particle size corresponds approximately to the diameter of the cubic octamer
and a monomer concentration of 735 ppm. Cubic octamer would be more stable
than the suggested cyclic species.


Summaries of Investigations


Bechtold (145) studied the polymerization of silicic acid made by removing sodium
from a solution of sodium silicate, leaving small amounts to adjust the pH. The
number-average molecular weight of the freshly made polysilicic acid was around
660, or a degree of polymerization of about II == 2. Thereafter, the molecular
weight increased over a period of days until the gel point was reached. Most rapid
gelling occurred at pH 5.05. An inflection in the pH titration was observed at pH
4.65. The molecular weight data were interpreted on the basis that S:(OH).
condenses with a functionality of slightly over 2.0, forming linear and then crosslinked siloxane polymers, eventually creating a gel. However, the intermediate
formation of particulate units, which then aggregate into chains, is now a preferable .



s. In

otal is
) 8000

Goto (45) measured the disappearance of monomer as it polymerized in a 0.24%

Si02 solution at pH 2.1 at 25C. Over a period of 60 days the monomer concentration dropped and the particle size increased, judging from the decreasing rate at
which the particles reacted with molybdate. Assuming that the monomer was in
solubility equilibrium with the particles, and using the formula' relating size and
solubility, the particle size was calculated, as shown in Table 3.7.
When the pH of the fresh solution was suddently adjusted to 8.8 the monomer
decreased to 20% of the original value in only I min. Assuming this is the solubility
of the resulting particles, the latter are calculated to be about 1.1 nm in size. This
particle growth at pH 8.8 was 3 x 10 times as rapid as at pH 2.
The relative rate of reaction with molybdic acid varied with the calculated particle
size. A log-log plot of the data showed the same slope as found by Her (Figure 1.15)
for the rate of dissolution of particles in this size range in 0.1 N NaOH solution.
This shows that the rate of depolymerization or dissolution in both systems depends

Table 3.7. Growth of Polymer from Monomer in Dilute Solution of Silicic Acid (0.24% Si0 2,
pH 2.1, 25C)

Age of Solution
time at 25C

-IS min
2.5 hr
4 days
9 days
20 days
60 days

Remaining as
Monomer (%)

Particle Size of
Polymer (nm)
(From Solubility)

Relative Rate
of Reaction with
Molybdic Acid




Calculated from data by Goto (45).


Polymerization of Silica

on the solubility of the particles. which in turn is an inverse log function of particle

Okkerse (29) determined the apparent "order" of the polymerization by measuring
the rate of disappearance of monomer, at different silica concentrations and pH.
The equations for second and third order reactions are

Ci l = k 2 t + COl
Ct-Z = 2 k 3 t + C0 2
where k z and k 3 are the rate constants at a given pH and Ct and Co are the
concentrations of monomer at times t and zero. His values for the constants at different pH's are shown in Figure 3.39.
It is significant that the rate constant of the "third"-order reaction below pH 2 did
not change with silica concentration, but the constant for the "second" order reaction above pH 2 decreased when the silica concentration was less than about 0.06 M
(360 ppm). The authors pointed out that pH affects the rate, not by the direct
LOGlo k3






Figure 3.39. Rate of disappearance or monomer by polymerization at different low pl-l's and
Si0 2 concentrations. molar: Circles - 0.0128: Triangles - 0.0417: Squares - 0.0641 and
0.834: k 3 k 2 - polymerization rate constants below and above pH 2.2. [From Ok kerse (29).]


of Silica

Summaries of Investigations


catalytic effect of H+ or OH-, but by their effect in generating positive or negative

ions on reacting species, which then react with nonionized species. However, this
does not explain why it was only in the case of the "second"-order reaction involving
negative ions above pH 2 that silica concentration had any effect.
The reason may be related to the phenomenon of the induction period noted at
low silica concentration, to be described later. The rate of disappearance of
monomer is very slow when the initial reaction can occur only with ionized
monomer or dimer. However, after some polymeric species with more acidic SiOH
groups have been formed, these ionize to a greater extent than monom.er and then
more rapid polymerization occurs, giving a higher rate constant. At low silica
concentrations, this stage is reached more slowly.
It was previously suggested by lIer (87) that polymerization below pH 2 was not
catalyzed by H+ ion, but by HF. However, Okkerse doubted that fluorine was
present in his system, since "no trace" of fluorine could be found in the sodium silicate used. Although Her noted that as little as 2 ppm of fluorine ion had an effect, it
now seems unlikely that this amount could be present in all the systems studied.
In a possibly related effect at low pH, Bogdanova (146) observed that below pH 3
the rate of depolymerization of polysilicic acid when greatly diluted is acid
catalyzed; that is, the rate increases with decreasing pH. The rate is independent of
the diluted Si0 2 concentration from 30 to 560 ppm at 20C. Traces of fluorine were
probably present since the polysilicic acid was derived from natural natrolite. As
shown by Iler, fluorine raises the pH of minimum polymerization rate from 2 to 3.


are the
s at difH 2 did
er reac0.06M
e direct

Audsley and Aveston

Audsley and Aveston (147) made very careful measurements of molecular weight by
light scattering, ultracentrifugation, and viscosity, on a 2.19% solution of silicic acid
at pH 2.1 as it polymerized at 24.5C for 29 days. Since the sol was made on the
alkaline side and purified by ion exchange, the silica must have polymerized to small
spherical particles in the first day or so. Further polymerization during the next 28
days then involved aggregation. The viscosity and molecular weight data can be
interpreted in terms of the growing volume fraction of microgel and the size of the
microgel regions which corresponds to molecular weight. This is consistent with the
view of the authors that polysilicic acid does not exist in solution as an extended


.H's and
6.tI '''d
(2lj .

Figures 3.40 and 3.41, based on the investigators' data, show the increase in limiting viscosity n,P/C, from which the fraction of water immobilized within the
polymer or microgel was calculated. From the Mooney equation. viscosity increases
rapidly as the volume fraction of dispersed phase approaches 0.5, that is, where half
of the liquid volume is occupied by microgel. Then in their 2.19% silica sol. where
the overall molar ratio H 20: Si0 2 is 148: 1, gelling should occur when the ratio in
the gel phase reaches 74: I. From Figure 3.41 the weight-average molecular weight
at this ratio is about 4 x 107 On the other hand from Figure 3.40 at the gel point
the limiting viscosity is 0.75 and from the authors' curve (Ref. 147, Figure 7. not
reproduced here) this corresponds to a molecular weight of lOS, which is a reasonable agreement.



~- ~


Polymerization of Silica





















Figure 3.40. Increase in limiting viscosity during aggregation polymerization: C. silica

concentration. 2.19%: pH. 2.1; temperature. 24.5C: n. p specific viscosity. [From Audsley and
Aveston (147).]

This molecular weight reflects the size of the microgel units as the gel point is
reached. A molecular weight of lOs corresponds to a microgel particle containing 1.7
X 10- 18 grams of Si0 2 Since the microgel contains 2.19% silica and has a density of
1.0 I g ern -3, the diameter of the average microgel particle is calculated to be about
0.27 microns.
The size of the initial silica particles can be estimated from the initial "hydration"
of the silica, reported as 1.8 H 20 per Si0 2 In the foregoing section on viscosity of
sols of very small particles, it was shown that the total bound water on discrete
particles (as silanol groups and hydrogen-bonded water) was related to particle size:
H 2 0 : S i 02 = - - = - 11.5d3
where d is the particle diameter in nanometers.
From the hydration ratio of 1.8, dis 1.8 nm. The degree of polymerization is 11.5
d", or 65, corresponding to a molecular weight of 3900. This compares with 5000 by
direct measurement.
The limiting viscosity number, nsp/C, of the sol was related to weight-average
molecular weight, M w' of the aggregating polymer, by the relation

n sp

I. 74 x 10 -3 M u.'



.... - - -

of Silica

Summaries of Investigations


This and the remarkable linear increase in log of limiting viscosity with time cannot at present be explained in terms of a satisfactory theory of aggregation of particles into growing microgel regions. Obviously, such microgel particles are not uniform in size, as assumed above, but instead cover a range of sizes as shown by the
centrifugation study by the authors. Thus the mathematical implications are very
com plica ted.

Weitz, Franck, and Giller

-::. silica

Weitz, Franck and Giller (148) carried out one of the few studies at low pH in which
the polymerization of polysilicic acid was followed by measuring the decrease in
monomer while at the same time measuring the number-average molecular weight
and finally noting the time at which the monomer plus oligomer reacted completely
before any non-molybdate-reactive high polymer was formed. This was done at pH 3
and 20C in a solution containing 4000 ppm Si0 2
Their data were used to calculate the specific reaction rate constant of the
oligomer with molybdate by assuming that the time required to reach maximum
color was when 99% had reacted and by taking into account the amount of
monomer originally present. It is assumed that at anyone time the oligomer
molecules are all about the same size, since otherwise spontaneous particle growth

sley and

ooint is
ling 1.7
e about

C) 100








osity of
e size:










is 11.5
iOOO by















Figure 3.41. Immobilization of water in microgel as aggregation polymerization proceeds:

Si0 1 concentration. 1.19'J.: pl-l, 1.1; tern perutu rc, 24.5 C. [From Audsley and Aveston (147).1



Polymerization of Silica


by "Ostwald ripening" would occur. The data are shown in Figure 3.4~ and are
interpreted as follows.
From the fraction. f of silicon that is present as monomer and the number-average
molecular weight. n one can estimate the average degree of polymerization of the
oligorners, P:

f + (1

- j)(P)-1

= 60n- 1

Thus when 43% of the silica is still monomeric and n is 118 then P is 7.3. Between
1.67 and 3.75 days P remained relatively constant between 8 and 11. Thus the
oligomers are in the range of octamers and dodecimers and their numbers increase
and then decrease. This relatively constant size also is indicated by the reaction rate
with molybdate which decreased only from 0.4 to 0.2. (Note that it is the inverse of
the rate that is shown in Figure 3.42.)
Oligomers decline when inert polymer begins to appear. The reaction rate with
molybdate decreases rapidly to 0.1 between 3.75 and 4.75 days. It is probably at this
stage that nucleation of colloidal particles begins. The oligomers dissolve and the
silica is deposited upon the nuclei. The nuclei grow and the concentration of
monomer in equilibrium decreases. The particle size can then be calculated from the
concentration of monomer. The proportion of monomer drops to 5% after 90 days.
which is 5% of 4000. or 200 ppm. This corresponds to the solubility of'2 om particles.

k- I





; .










Figure 3.42. The relative proportions (%) of silica present .IS monomer. oligomer. and high
polymer (colloidal particles) as a 400 ppm solution of Si(OIl). polymerizes at :woC at pl l J: A.
monomer: B. oligomer: C "high polymer" or colloidal particles: D. reciprocal of the reaction
velocity constant. k, of oligomer with molybdic acid. [From Weitz. Franck. and Giller (I-lS).]

of Silica

Summaries of Investigations


2.0 ...--......-...,...-----,---'T-----r--....-------,

and are
m of the



o 1.5

rhus the
rion rate
nverse of





"ate with
ly at this
and the
ation of
from the
90 days,
nm par-












Figure 3.43. Relation between the degree of polymerization and the ratio of shared oxygen to
silicon atoms: Circles. Hoebbel and Wicker (84): squares. Bechtold. Vest. and Plambeck (149):
solid line. theoretical relation for silica particles.

Bechtold, Vest, and Plambeck

Bechtold, Vest, and Plambeck (149) prepared polysilicic acids by hydrolyzing ethyl
silicate at low pH and then at various stages of polymerization the silanol groups
were esterified or "end capped" with trimethylsilyl groups. Products were
characterized by chemical analysis and measurement of molecular weight, from
which the proportion of shared oxygen (oxygen atom connecting two silicon atoms)
in the polysilicic acid was calculated. They showed that the polymer was not a linear
siloxane chain, which would have one shared oxygen per silicon atom. Instead they
found that after the degree of polymerization reached about IS the shared 0: Si
ratio increased only slowly and was in the region of 1.5.
This observation led to the postulate that the growing polymer consisted of a ribbon of three parallel siloxane chains. However, similar work later by Hoebbel and
Wieker (84) showed that at higher degrees of polymerization from 100 to 1000, the
ratio changes from 1.7 to 1.8. As shown in Figure 3.43 a composite plot of data of
both investigators indicates that the two sets of observations are reasonably compatible.
The results are consistent with the model that the silica consists of very small
spheres and that "polymerization" in the intermediate stage involves an increase in
size of these spheres. with decrease in their number. As shown in Chapter I. for surface-hydroxylated silica particles the ratio of shared 0: Si atoms is related [0 11, the
number ofSi atoms in a sphere:

and high
pH 3: A.
: rc~"'tion


O:Si = 2.0 - 2.42n- 1 3 + 3.911- 2

2.09n- 1

The O:Si ratio according to this formula is shown as the solid line in Figure 3.43.



.-'~ :~:-';';'-:"-'


-'-Polymerization of Silica


However, there is evidence that after a certain point the polymer does not consist of
single particles, but chains and networks of particles, as will be discussed in connection with Hoebbel and Wieker's work (84).


i ;

Acker (l33a) carried out a study of aggregation very similar to that of Audsley and
Aveston. The initial polysilicic acid had a molecular weight of 2000 and reached
about 3 x 10' at the gel point. The reduced viscosity, (71,. - l)/c, or [17], where 17,. is
the viscosity relative to water, was related to the weight-average molecular weight by
the formula

This is of similar form but very different in value from the equation found by
Audsley and Aveston, no doubt owing to the great difference in silica concentration.
The continued increase in gel strength after the gel point was shown to continue as
an extension of the viscosity before gelling.

Hoebbel and Wleker

Hoebbel and Wieker and co-workers (84) published a series of papers describing by
far the most intensive and thorough study of polymerization, simultaneously using
several methods of analysis and characterization to follow the process from the
earliest stages. Individual polymer species were isolated as trimethylsilyl derivatives
and characterized as to molecular weight and composition whence the characteristics
of the original silicic acid were calculated.
The earliest stages in the polymerization of monosilicic acid occur so rapidly that
it was necessary to study the reaction at -2 and even at -13C in order to follow
the formation and disappearance of low molecular weight species in 0.5 and 1.56 M
solutions. The use of ztSi-nmr spectroscopy was essential to clearly distinguish
monomer from dimer-trimer mixtures and from cyclic tetramer and higher cyclic
species. They observed the following negative displacements of the signal (ppm)
toward a higher field as measured on standard compounds of known structures:
Silicic Acid
Dimer plus linear trimer
Cyclic tetramer
Higher cyclic
Branching groups-

99 - 101

Silicon linked through oxygen to three other silicons.

This was combined with improved methods of making trimethylsilyl derivatives of


.n of Silica

Summaries of Investigations

consist of
n c""nec-

these intermediates and clean separations by temperature-programmed gas

chromatography. As shown in Figure 3.44, the most significant conclusion was that
first there is a buildup of dimer, linear trimer followed by cyclic tetramer, and then
polycyclic polymers and three-dimensional polymers. The polycyclic compounds are
detected by the branching groups.
About 90% of the monomer had disappeared by the time the small cyclic
polymers had reached their peak. At pH I even at -15C the reactions went more
rapidly and higher polycyclic acids accumulated and then went on to higher
branched polymers.
The transformation of the latter, by further condensation with dimer, trimer, and
tetramer, to silica particles large enough to act as nuclei for particle growth, must
occur at some point. However, the mechanism, including internal rearrangements to
form a dense silica core, is not yet known. It probably involves breaking and reforming new siloxane bonds within the initial three-dimensional polymer in such a way as
to allow maximum formation of siloxane groups and elimination of silanol groups
with some shrinkage in particle size. However, the main effect will be to create more
SiOH and fewer Si(OH)z groups on the surface. The single silanol groups are more
highly ionized and thus condensation of additional monomer occurs much more
Once these nuclei have consumed all the smaller oligomer species, the resulting
colloidal particles remain in solubility equilibrium with monomer. However, if the
concentration is more than a few thousand parts per million (0.6% or 0.1 M SiO z)
the particles soon aggregate into strings or branched networks of microgel.
In earlier studies at 25C, Hoebbel and Wieker obtained basic data on the
polymerization of a 0.4 M solution of monosilicic acid (2.4% SiO z) at the most stable pH of 2.0 as the solution aged from 5 min to 24 days. Measurements were' also

::1 reached
there '1,. is
weight by

found by
mtinue as

cribing by
usly using
.pidly that
to follow
id 1.56 M
her cyclic
aal (ppm)



















Figure 3.4-t


; of

Earliest stages in the polymerization of 0.501 monosilicic acid at - 2C and pH

2.0: A, monomer: B. probably cyclic trimer, may also be dimcr and linear trimer: C. cyclic
tetrarner; D. higher polysilicic acids. [Hocbbel and Wicker (84).1

. ':r"'.""::~-=t1'
. . . r..:...- :~.::~



Polymerization of Silica



- -:-









. -,












~ 40










Figure 3.45.

Reaction of polysilicic acids at various stages of polymerization with molybdic

2SC. A. polymerized for 5 minutes: B. I
hour; C.4 hours: D, I day: E, 3 days: F. 5 days: G, 8 days: H. 2.J days. For comparison: I monomer, 2 - dimer, 3 - three-dimensional polymers. 4 - same curve as D double: chain
polymer. [From Hoebbel and Wieker.1973 (84).]

acidr SfO, concentration - O.4m: pH -2: temp. -

'. i

made at 0.084 and 0.97 M concentration. The following data were presented: as the
polymerization progressed, the reactivity of silica to form yellow silicomolybdic acid
decreased. In each sample taken at different aging times, the percent of silica
remaining unreacted with molybdic acid was measured versus the reaction time up
to 80 min. For each age of silica sol, that is, stage of polymerization, the reaction
rate of the high molecular weight fraction was calculated as k H.\f min -I. Figu re 3.45
is based on Figure 1 of Hoebbe1 and Wieker (Ref. 84, 1973).
In 5 min, 40% of the silica had polymerized to dimer. By chromatography it was
shown that after I hr, perhaps 40% of the monomer and dimer had polymerized to
form cyclotetrasilicic acid as well as some hexarneric silicic acid and higher
polymeric acid. After 4 hr, about 50% of the polymers up to tetramer were
converted to cyc1opolysilicic acids with an average molybdate reaction rate constant
of 0.103 min - I .
As shown in Figure 3.46, there is a discontinuity in the relation between molecular
weight and the molybdic acid reaction rate constant that distinguishes oligomers
from higher polymers. As will be shown, this change occurs at the point where the
"high molecular weight," that is, colloidal, species begin to be the main constituent.
After 1 day, 15% of the polymers up to tetramer remained with a higher polymer
that had a reaction rate constant of 0.025 min - I and appeared in chromatograms as

. f

" I,

:- .. ~



:5: B. I

Summaries of Investigations


a band. the position of which changed with molecular size. The molecular ~eight of
the polymeric species. as determined on the trimethylsilyl derivative. was 3440. corresponding to a degree of polymerization of about 52 silicon atoms per molecule or
equivalent to an anhydrous SiO z particle 1.65 nm in diameter.
Thereafter. as the solution aged 1-8 days. the polysilicic acids were converted to
trimethylsilyl ester species in which SiOH groups of the acids were converted to
SiOSi(CH 3)3 groups. After removal of volatile oligomer species by evaporation
and/or solvent extraction. the high molecular weight polysilicic acid derivatives were
analyzed for C, H, and Si, and the molecular weight in benzene determined by a
sensitive vapor pressure osmometer. The molecular weight of the polysilicic acid was
calculated from that of the trimethylsilyl ester and correlated with the rate constant
of the reaction with molybdic acid. From 8 days to the gel point at 42 days, the
molecular weight was estimated only from the molybdate reaction constant.
The authors showed the absence of polymers consisting of linear multiple chains
but left open the question of the nature of the high polymer species. It is here
proposed that in the I day old sol, when the degree of polymerization reached 52
silicon atoms per molecule, the polymer existed as spherical silica particles. In the
period of 1-8 days, these particles increased in size and decreased in numbers by
Ostwald ripening. Then between 8 and 42 days. the particles continued to increase in
size as they aggregated into chains and networks until gel was formed.
During the period of Ostwald ripening the molecular weight was estimated from
the rate constant of the reaction with molybdate. This constant decreased as the

In: 1-





as the
ic acid
. silica
me up
'e 3.45
it was
zed to
. were




u.. 10


I Day




,-" - -- .........,


, \





Cubic Oclamer-0.52
Cyclic Hexamer-0.58
" Cyclic Telromer - 0.67

Monomer- 1.7


.re the

IlylT' ..r





Figure 3A6. Relation between molecu lar weight of (SiO z. :< HzO)" and rate constant of reaction with molybdic acid k ll M It may be significant that the cubic octarncr and cyclic hcxamcr,
which arc not on the linear plot. arc also not round in the course of aqueous polymerization of
monosilicic acid. [From Hoebbcl, Wicker ct ~II.. 1973 (84).J

Polymerization of Silica




particles grew in size. However, during the aggregation process the rate constant did
not decrease much further. When particles associate into chains, only a small part of
the surface area is lost at the points of contact.
Thus even though the effective molecular weight may have increased to millions as
the aggregates increased in size, .the molecular weight calculated from the molybdate
reaction constant corresponded only to the size of the particles making up the chains
and networks. Hence even though the molecular weight of the aggregates
approached infinity at the gel point, that of the equivalent constituent particles only
reached around 100,000, or a degree of polymerization of 1800.
The authors measured the molecular weight of the isolated trimethylsilyl derivatives of the high-polymer fraction during the first eight days of polymerization.
These were calculated from the formula

where x is the ratio of silyl groups to silicon atoms of the polysilicic acid and n is the
degree of polymerization as determined by osmometry.
The authors give evidence that all silanol groups were converted to trimethylsiloxy
groups. The molecular weight of the 'polysiltcic acid was calculated according to the

Values for x and n at different polymerization times are shown in Table 3.8. The
number of shared oxygen atoms per silicon atom are calculated from x, As shown in

Table 3.8.

Values of n and x Calculated from Hoebbel and Wieker's Data

(days) k H .., (min-I)





10- 3
10- 3
10- 3
10- 3
10- 3

Mol. Wt.



(80.000 C )

Degree of
Polymerization. n

( 1260 C )
( 1800C )

2 - 0.5x



(from n)

Source. Hoebbel, Wieker, et a1. (84).

Extrapolated values.
a Assuming polymer is a spherical silica particle.
From the molybdate reaction constant. However. aggregation had begun and actual
molecular weight was much higher.




Summaries of Investigations

it did

Ins as
. only


Figure 3.44, the ratio of shared oxygen to silicon atoms is a function of the degree of
polymerization and the values are reasonably in line with those of Bechtold, Vest,
and Plambeck (149).
According to the formulas developed in Chapter 1.

x = 2.15d- 1

1.53d- z + 0.36d- 3

where d is the equivalent diameter in nanometers of an anhydrous SrO, particle

containing n silicon atoms.


= 11.Stfl

d = 0.443n 1 / 3



x = 4.8Sn- 1 / 3

7.80n- Z / 3

+ 4.2n- 1

is the

o the



Values for x were calculated for different values of n after the indicated numbers
of days of aging as shown in Table 3.8.
At the low degree of polymerization of 52, there is some discrepancy between the
o H: Si ratio of 0.83 calculated from the formula for a spherical particle and the
value 0.65 calculated from experimental data. However, in the discussion of nucleation theory it will be noted that. as shown in the model in Figure 3.17, at the stage
where the degree of polymerization is 48 (that is, a particle with a SiO z core has
been formed), the model shows a OH:Si ratio of0.66 rather than 0.8 from the formula for a spherical particle! This may be purely fortuitous but it seems to support
the view that the first particles of silica are formed with a nucleus of a three-dimensional octamer or decamer. (see Figure 3.17.)
Thereafter as the particles increase in size, their stoichiometry more closely
approaches that for a sphere, as postulated. Thus above a degree of polymerization
of 500 (particle size as anhydrous SiO z of 3.5 nrn), the composition of the polymer is
very close to that calculated for a spherical particle.
RATE OF POL'i:'tIERIZATION. The increase in degree of polymerization with time is
reasonably well represented by the equation


3000 - n




or dnfdt = 0.023(3000 - 11), where n is the degree of polymerization and t is time in

days up to about 8 days (Figure 3.46). At 8 days the degree of polymerization is
about 460 corresponding to a particle diameter of about 3.4 nm. Thereafter aggregation begins, and as it progresses. the microgel network continues to coarsen and
reacts more slowly with molybdate. The extrapolated limit of 11 = 3000 corresponds
to the size of discrete particles (6.4 nm) that would react at the same rate as the rnicrogel polymer docs: this size corresponds to a specific surface area of 400-500 m Z


Polymerization of Silica


g-1 which is reasonable for a gel. Also the size is about that at which the solubility of
silica drops to 100-120 ppm.
It is observed that the constant k H ." of the reaction rate of high polymer with
molybdate is close to being inversely proportional to the degree of polymerization, n,
The approximate relation is .

where n ranges from 121 to 1260.
It might be supposed that the reaction rate would be proportional to the surface
area of the silica, and also to the solubility difference between the very small particles and massive silica. It can be shown that by coincidence this difference varies
inversely with the square of the particle radius in this size range. Since the specific surface area varies inversely with radius, the rate should then vary inversely with cube of
the radius which is proportional to n, the degree of polymerization as observed above.
Investigations Through the Neutral pH Range

M erri// and Spencer

Merrill and Spencer (150) showed that in the presence of the electrolytes when a 3.3
ratio sodium silicate solution is neutralized with ammonium sulfate, the minimum
gel time is no longer about 5.5, as for salt-free sols, but moves to pH 7 and higher,
depending on the concentration (Figure 3.47). The gel time decreases, or the gelling
rate increases, not only with increasing silica concentration, but also the increasing
concentration of electrolyte- that accompanies the silica. (See Chapter 4, regarding
the effect of salts on the gelling of colloidal silica at different pH values.)
The very great decrease in gel time with increase in concentration at a given pH in
the range 7-9 is due to the great increase in coagulation effect as the electrolyte
concentration exceeds about 0.1-0.2 N, corresponding to silica concentrations here
of 1-2%.

Ashley and Innes

Ashley and Innes (151) in 1952 pointed out that the low viscosity of colloidal silica
at high concentrations clearly showed that the particles were spherical in shape,
since any other shape, chainlike or even acicular, would cause a much higher viscosity. The conclusion is that silica in some way polymerizes to dense particles of
more or less spherical shape, at least above neutrality.

Baumann (152) studied polymerization under conditions where nucleation and
growth of particles could be followed. Wide ranges of concentrations and buffered

-:-. ..f Silica

Summaries of Investigations

ility of


er w.
:ion, n.




l partivaries
fie surcube of

1.5 25












Figure 3.47. Gel time of sols from mixtures of sodium silicate and ammonium sulfate. Numbers on curves: % Si0 2 [Data by Merrill and Spencer (150).)
n a 3.3

pH in
IS here

I silica
er visdes of

Iffe" -l

pH were employed to follow the polymerization of monomeric silicic acid, the

concentration of which was followed by the molybdate method. Baumann (72b) also
followed the early stages of polymerization at pH 3 and 6. I by stopping the molybdate reaction in samples by adding citric acid and reducing. the yellow complex to
molybdenum blue for greater sensitivity.
In the alkaline pH range 5.8-9.0 the polymerization, which we now know consists
mainly of growth of very small spherical particles, showed an induction period, if the
concentration was low enough, after which monomer disappeared at an increasing
rate and then at a negligible rate as monomer concentration approached 150 ppm,
corresponding to the solubility of particles about 3 nm in diameter.
For Baumann's Figure I, the values were recalculated to obtain the amount of the
silica in excess of ISO ppm (the final value at longer times) that remained as
monomer at time t (minutes). This was then plotted against log t, as in Figure 3.48.
It is noted that the curves appear to have an inflection point at 50% remaining
monomer, at a time designated t l
When log t l is plotted against the reciprocal of the initial concentration of
polymerizable silica (parts per million in excess of 150). straight lines are obtained
as in Figure 3.49. In Figure 3.48 the shape of the curves strongly suggests that the
reaction rate is a function of monomer concentration times "polymer" concentration. The curves appear to be symmetrical about the point where half of the
monomer has disappeared.
The following relationship may be adduced: let C = fraction of polyrncrizable

~ -,-:.


Polymerization of Silica



























Figure 3.48.

Unreacted monosilicic acid versus time in polymerizing solutions:


Concentration (ppm)









P .. polyrnerizable silica - (total - 150) ppm. [Data from Baumann (152).)

monomer at time I. Then Cis 0.5 at time t; Assuming

d log I


= kC(l - C)

1- C
= k log


From the data, k = 32 C-0.5.

The relation between I, and the initial concentration Co follows the empirical

f Silica

Summaries of Investigations


log II

+ 800C;! - 4.02

,;. - -

where Co is the initial concentration of polymerizable SiO z.

It is likely that there is a similar relationship over the range of pH and concentration, since the data can be expressed by a single empirical equation except at high
concentration at pH 9, where no correction was made for the presence of Si0 3 - ions
which reduce the amount of polymerizable silica.
A possible explanation of the rate equation may be that a small portion of the
monomer polymerizes first to a three-dimensional nucleus such as polycyclic
. octamer or decamer, and that the main reaction is the addition of monomer to this
polymer, as will be discussed in relation to the work of Marsh, Klein, and Vermeulen (32). From solubility the final particle size is about 3 nm. If the initial nuclei
have an average diameter of around 1.5 nrn, then only (1.5/3.0)3 or 12.5% of the
silica initially is needed to serve as nuclei, after which the addition of monomer
proceeds by reaction with the particle surface until the monomer level approaches
the solubility of the particle.
Coudurier, Baudru, and Donnet

Coudurier, Baudru, and Donnet (36) made an extensive study of the polymerization
of "disilicic acid" in nearly neutral solution. By removing Na + from Na zSi0 3 9H zO


::J 200






















Figure 3.49. Relation between half-life of monomer and reciprocal of

polymerizablc silica at different pH"s. [Data from Baumann (152).)

concentration of


:.... ...

Polymerization of Silica


in solution at pH 2 they obtained more than 50% dimer plus monomer. but' since
both react so rapidly with molybdic acid both are called "monomer:'
Their data indicate that below pH 5-6 silicic acid first polymerizes to discrete
particles less than 4 nm in diameter (less than 50,000 molecular weight) and then if
the silica concentration is high enough, these particles aggregate by linking together
into chains which branch into three-dimensional networks, each of which is a
"polymer" molecule of much higher molecular weight according to light scattering.
Eventually all the discrete particles join the networks. which finally fill the aqueous
medium forming a gel.
The solution containing only 0.6% Si0 2 polymerizes rapidly to discrete particles
before the latter begin to aggregate (Figure 3.50). The concentration of "monomer"
decreases rapidly, whereas increasing amounts of oligomers and then higher
molecular weight polymers are found. Then the oligomers decrease as the polymers
become colloidal particles. The size of the colloidal particles must be less than 4 nrn,
judging from the concentration of monomer in solubility equilibrium. They increase
in size only slowly thereafter, as evidenced by the small decrease in their equilibrium
solubility, indicated by the monomer concentration.
The increase in molecular weight by light scattering is entirely due to aggregation
of the preformed primary colloidal Si02 particles into networks (Figure 3.51).
Donnet and co-workers recognized several early stages in the polymerization and
classified the corresponding polymers as oligomers. As shown in Table 3.9.'the reactions are 'only some of the possible ways in which oligomers are formed. Oligorner-I,






















Figure 3.50. The relative proportions of "monomer:' oligomer and colloidal species at 6000
ppm of monomer as Si02 polymerizes at pH 5. Open circles. "monomer": squares. oligomer:
solid circles. polymer (nonreactive with molybdate). [Data from Coudurier, Budru, and Donnet

,r Silica

Summaries of Investigations

It since

en 20 .....------.----,---.,...-----.

then if
.h is a
:14 nm,
ion and
te reacarner-I,








which includes cyclic tetrarner, is of such low molecular weight that it can still
depolymerize and react with molybdic acid at a relatively rapid rate. Still higher
molecular weight oligomer-Z de polymerizes to monomer only slowly, the rate
diminishing with the molecular (particle) size.
The effect of pH on the relative rate of disappearance of monomer and formation
of high polymer is shown in Figures 3.52 and 3.53.
At pH about 2 the initial disappearance of dimer is due to condensation with
itself, since they found the reaction is "second order," giving a linear plot of the
reciprocal of concentration of unreacted "monomer" versus time. On the other
hand, at pH 7.2, the reaction is "first order," the "monomer" adding to polymer
particles formed almost instantly after the pH was raised. In this case a plot of low
concentration of unreacted "monomer" versus time is linear with a negative slope.
From the curves it is apparent that the rate of disappearance of "monomer" and
of appearance of polymer decreases with change in pH from 1.2 to about 3, then
increases to pH 5 and 7.
Table 3.9.

Polymerization of Silicic Acid at pH 1-5-0Iigomer Species

Stages of Polymerization


Probable Structure

(I) 2 Si(OH). '" (HOhSiOSi(OHh

Dimer :

Monomer, dirner, some trimer

"Oligomer- I"
"Oligorner-I "

Cyclic trimer (?)

Cyclic tetramer
3Dimensional polymers
Particles 2-4 nm in size

H 20
(2) Si(OH). + dirner .. [(HOhSiOh
(3) Dimer + dimer '" [(HO)2Si-O].
(4) Trimer and tetramer plus
(5) Si(Of:I). + nuclei


. at 6000
I Dt'-"ct

Figure 3.51. Molecular weight of polymer by

light scattering as monomer polymerizes at a
concentration of 6000 ppm Si0 2 at pH 5. [Data
from Coudurier, Badru, and Donner (36).]


W1. Jess



(6) Oligorner-J + oligorncr-Z

Higher polymer
Aggregates of particles
(mol. W1. up to millions)

Refer 10 Coudurier, Baudru, and Donner (36)


~ ...

Polymerization of Silica














. t,





Figure 3.52. Effect of pH on course of polymerization at 6000 ppm SiO z: M. "monomer"; P.
polymer. [Data from Coudurier, Badru, and Donnet (36).)

Thus there are two entirely different mechanisms involved in the conversion of
monomer to polymer. Hydrogen ion must be the catalyst in this system below pH 3,
although with higher silica concentration this is true below about pH 2. Above pH 3,
the OH- ion is the catalyst for the addition of monomer or dimer to the growing
polymer. This addition is believed to be because the oligomer-2 particles have SiOH
surface groups that are more acidic and thus more highly ionized than those of
monomer or dimer. Thus monomer and dimer condense with these surface groups
. more rapidly than with lower molecular weight species.
The overall rate of disappearance of monomer-dimer (M) indicates that it reacts
simultaneously with itself. with oligorner-I (0 1) and with cligcrner-Z, which was







Figure 3.53. Effect of pH on V. the

initial rate of disappearance of monomer
at 6000 ppm SiOz [Data from Coudurier,
Badru, and Donnct (36).)

Summaries of Investigations


later called S; (polymerized silica). At pH 4, the rate V was found to be


= K [Mp14 exp (0.135[0 1 + Sp])

The effect of temperature on the rate of disappearance of dimer was also

measured. In the above equation only K varied with temperature. The energy of
activation was 14.6 kcal mole- 1 at pH 4 for the process of polycondensation.
Aggregation of the silica particles referred to as oligomer-2, or Sp, begins at about
the time where the monomer or dimer has been mostly consumed and very little
oligorner-I remains (Figure 3.52). The rate of aggregation in anyone sample in the
pH range 1-6 follows the equation

: P.

I 3,
I 3,
. of



where M w is the molecular weight of the aggregates at time t, and M o is the initial
molecular weight of the discrete particles. The size and number of the discrete particles remains constant in any particular sample below pH 6.
As shown in Figure 3.54, linear plots are observed except above pH 6, where it is
known that the particles grow by Ostwald ripening. Also the particles begin to have
an increasing ionic charge above this pH, sufficient to prevent collision and aggregation. At higher pH the increase in molecular weight is then identical with the
increase in particle size, since no aggregation-polymerization occurs. The curve of
Figure 3.54 at pH 6.3 is in a transition region where aggregation and particle growth
are both occurring.
It is of incidental interest that at pH 4, curves 1 and 4 of Figure 3.54 differ only in
silica concentration. The slope, which is the rate of increase in molecular weight, is
proportional to the square root of the silica concentration, and in this case to the
square root of the number of particles per unit volume.
Donnet and associates examined the solutions at the point where monomer-dimer
had been largely consumed using polarized light. This showed that at pH 5-6 the
polymer consisted of extremely small spherical particles, whereas below pH 5 the
polymer was apparently more anisotropic. The fact that the particles had begun to
link together into chains probably explains the observation.
The rate of aggregation was shown in general to be related to the total silica
concentration as follows:

The overall temperature coefficient indicated an energy of activation of 15.1 kcal

mole:" for the process of aggregation at pH 4.
Gels obtained from some of the polymerized sols were isolated and characterized.
The specific surface area of the gels by nitrogen adsorption was related to the pH of
the preparation as shown in Figure 3.55. However, it is known that gels with specific




Polymerization of Silica









Figure 3.54.

Molecular weight of polymer versus time.







Temp. (0C)

Si0 2 (gl-I)




[Data from Courdurier, Badru. and Donnet (36).J


surface areas higher than 800 m Z g-1 shrinkupon drying so that pores become too
small to admit entry of nitrogen molecules to the whole surface.
It is also known that the specific surface area of a gel is not much less than the
ultimate particles of which it is composed. The silica made at pH 5 with 800 m 2 g-I
corresponds to ultimate spherical particles about 3.5 nm in diameter, and by
extrapolation the particles of the silica made at pH 4 would be about 2.3 nm in
diameter. As will be seen from consideration of other investigations, this is close to
the size of particles made in a similar manner, but characterized by other means.

Marsh, Klein. and Vermeulen

Marsh, Klein. and Mermeulen (32) prepared a comprehensive review of the literature on polymerization kinetics and equilibria and conducted a detailed study to
establish the polymerization mechanism in the range of silica concentrations of
200-1800 ppm and pH 4-10 at 25C. Starting with Si(OH). solution prepared from
Na 2H zSi0 4xH zO by ion exchange at about pH 2.5. mixtures buffered with acetate
were made at various concentrations and pH values and the rates of polymerization

f Silica

Summaries of Investigations



and disappearance of molybdate-reactive silica were measured. The "reactive silica"

was mainly Si(OH). and for convenience is referred to as "monomer:' Strong similarities in results can be noted between data presented here and in Figures 3.48-3.52,
both of which suggest nucleation and particle growth.
Valuable new data were developed, but the authors did not consider all the
implications of the fact that at least under the alkaline conditions studied, the
"polymerization" must have consisted of the nucleation and growth of dense amorphous particles of colloidal silica which were in solubility equilibrium with
Fortunately in many of the experiments, the ionic strength and silica concentration were sufficiently low that little or no aggregation or gelling occurred so that
"polymerization" involved only nucleation and particle growth. The ionic strength
of the buffer was varied but not taken into account as a factor that could affect the
rate of polymerization.
Several interesting points can be deduced from the data on
polymerization at pH 8, which is shown in Figure 3.56. This confirms the earlier
observation of Baumann (152), who was apparently the first to observe that under
some conditions of pH and concentration there is an "induction period" during
which there is little or no polymerization of monomer. Marsh, Klein, and Vermeulen
also found that when the monomer solution, in the concentration range' of less than
500 ppm, is suddenly alkalinized from pH 2, where it is relatively stable, to pH
around 8, the monomer does not begin to disappear at once, but only after an indueINDUCTION PERIOD.

ne too
an the
m2 g-1

nd by
nm in
lose to

Iiteraidy to
JOS of
j from

Figure 3.55. Effect of pH onthe specific

surface area of silica gel. Dottcd line:
probable values if area had been
determined in solution to include smaller
polymeric species. [Data from Coudurier,
Badru, and Donnet (36).]



... ".



Polymerization of Silica








z 0.8



u 0.7







1000 10,000

Figure 3.56. Disappearance of monosilicic

acid versus time at different concentrutions of
359-697 ppm sio, Temperature. 25C: pH.
8.0. A-C. induction times at which polyme;i.
zation began: -H. times at which "nuclei"
may have appeared: I-L. points at which
monomer concentration became the controlling factor: -H. approxirnate inflection
points. [Data from Marsh. Klein, and Yermeulen (32). Circles: data by Ginsburg and
Sheidina (154) for 400 ppm concentration
interpolated between pH 7 and 9.J

lion period. The induction time increases enormously with only a relatively' small









decrease in total silica concentration. as can be seen at points A. B. and C in Figure

3.56. Then the monomer concentration drops at an increasing rate to points E-H.
suggesting that the initial formation of polymer is autocatalytic. Here also a small
decrease in the total silica concentration causes a large decrease in the rate. Based
on the relation between silica concentration and rate of the reaction. a "reaction
order" of 6 or 7 is indicated.
The authors concluded that polymerization of monomer involves a reaction
between Si(OH)4 and =SiO- ions, whether the latter are in solution as HSi03 - or
on a polymer as an ionized =SiO- group. This is the mechanism suggested by
Ashley and Innes (lSI). Marsh. Klein. and Vermeulen similarly explained the autocatalytic effect on the basis that the ionization constant of SiOH groups on the
polymer increased with the molecular size. and thus as more polymer formed and
grew in size the total concentration of SiO- ions in the system at a given pH also
increased by orders of magnitude. Thus the rate increased until there was an offsetting decrease in the concentration of Si(OH)4'
This may well be a valid explanation and is also consistent with the fact that the
rate of disappearance of monomer is approximately proportional to OH - Ion
concentration. which in turn determines the concentration of SiO- ions in the range
from 6 to 8. and thus the reaction rate in a given system.
NUCLEATION. One can also consider this induction phenomenon in terms of nucleation of colloidal particles in the 10-20 A size range, which then grow by further
deposition of silica from solution. The extreme effect of concentration on the induction period might be expected from nucleation theory. According to Nielsen (153).



Summaries of Investigations

ns of






;- or
j by




the apparent kinetic order of nucleation may be as high as 10. As shown in 'Figure
3.56, one may relate the times at the inflection points with the concentrations and it
will be seen that a plot of log time versus log concentration corresponds to a "reaction order" of 7 or 8. In this regard, the phenomenon resembles nucleation in other
systems. In Figure 3.56, points E-H probably represent the points after which no
additional nuclei are formed.' They then grow at the expense of monomer and

oligorners to points I-L, after which "Ostwald ripening" occurs and they are iifsolubility equilibrium with monomer as they grow.
PARTICLE GROWTH. The monomer concentrations after points I-L can be used to
calculate the size of the colloidal particles versus time. In the molybdate test, the
"unreactive fraction" corresponds to colloidal silica, and the "reactive fraction" is
monomer or "soluble silica." Using the graph of Figure 3.57 relating particle size
and solubility (taken from Figure 3.32) the calculated particle size versus time is
shown in Figure 3.58.
The linear relationship over such a wide range and the fact that the lines are
parallel over a certain range of size and silica concentrations indicate that the same
growth mechanism is involved at all four concentrations. It is also noted empirically
that for a given particle size, log time is a linear function of the reciprocal of the
silica concentration (Figure 3.59).

Ginsberg and Sheidina

Ginsberg and Sheidina (154) in a similar study gave some data similar to that of
Marsh, Klein, and Vermeulen, but observations were not reported for short enough
times to note the induction phenomenon. However, the final monomer concentration
in a 400 ppm solution at pH i was 120 ppm after 30 days at 25C, corresponding to





Tsett the
















5 6 7


Figure 3.57. Solubility as a function of the

particle size of amorphous silica formed at
25C. [Data from Chapter I.]

5 X 10 r------,r----r--...,---..,....----,






.: i














: I




Figure 3.58. Time required at pH 7.

25C. for silica to polymerize to particles
of a given size. Silica concentrations indicated parts per million. [Data from
Marsh. Klein. and Vermeulen (32).] .

Figure 3.59. Time required at pH 7. 25C.

for silica to attain the indicated size versus
inverse of silica concentration (rpm). [Data
from Marsh. Klein. and Vermeulen (32).)

.. -



Summaries of Investigations


the solubility of particles 3.7 nm in diameter. This compares with 3.5 mn

extrapolated from Figure 3.58.

Investigations Aboye pH 7
Greenberg and Sinclair

Greenberg and Sinclair (ISS) examined the polymerization of silica in the pH range
of 7-12 in mixed solutions of ammonium acetate and sodium metasilicate by means
of light scattering. Unfortunately, the pH and electrolyte concentration (sodium
acetate) were not controlled separately, so their effects could not be distinguished.
However, this is one of the few studies in which the main reaction involved growth of
porous microge/ particles 20-120 nm diameter containing water, by aggregation of
primary particles of the order of 3-4 nm diameter, followed by a secondary aggregation of the microgel particles to form a solid gel.
In the absence of salts in this pH range it is known that polymerization of
monomer at these concentrations forms particles about 2 nm in diameter in a few
minutes. However, in the presence of ~0.3 N ammonium acetate. which is above the
critical coagulation concentration for 2 nm SiO z particles. these particles then
Since the silica concentrations were generally less than I %, the sol did not form a
coherent gel. Instead the aggregates formed secondary, roughly spherical gel particles which grew in size to 150 nm and then these in turn coagulated. These aggregate
gel particles are highly porous and, as the authors emphasize. are in a "highly solvated state" since, of course, they- include much water. It was shown that up to a
diameter of 120 nm, the diameter of the secondary particles increased linearly with
time. The explanation might be that after a certain number of secondary particles
have been formed to act as nuclei early in the process, when most of the primary
particles are still nonaggregated, no additional aggregate nuclei are formed, so that
the rate of accretion will be proportional to the surface area of the growing particles
and the concentration of primary particles

where D is the diameter of the growing secondary particles and C is the concentration of primary particles. Early in the process, C will be nearly constant so that D =
k't, as observed. However, there is no way of knowing whether in fact only a small
part of the total silica had formed secondary particles by the time they reached 120
nm in diameter.
The kinetics of formation of primary aggregates and combination of these into
larger secondary aggregates remain to be determined.

Polymerization of Silica



Greenberg (156) also studied polymerization of silica in a mixture of sodium metasilicate and ammonium acetate which served as an acid to lower the pH below about
10.7 so that the silicic acid was liberated and polymerized. He followed polymerization by small changes in the refractive index of the solution which bear a significant
but not entirely understood relation to the condensation of SiOH groups to siloxane
(Si-O-Si) linkages. No polymerization occurred at pH 11.4 but rates were measured
at pH 10.0 and 10.8. It was concluded that the rate of disappearance of SiOH
groups in the system followed the first-order equation:



where C is the concentration of silanol groups, t is time, and k is a constant. In this

region the rate increases with decreasing pH, no doubt owing to increasing conver-sion of HSi03 - to Si(OH). which polymerizes. The rate is also proportional to the
concentration of uncondensed silanol groups.

Goto (157) was one of the first to examine the polymerization of silicic acid in the
absence of salts and in the pH range 7-10, where colloidal particles are formed and
no aggregation or gelling can occur as at lower pH. He used the equation



! .


= k(C-


to calculate the rate constant k, where C is concentration of monomer at time t, and

S is the solubility of "amorphous" silica. He found log k to be an increasing linear
function of the pH or log [OH-] such that

He also aged a silicic acid sol containing 2000 ppm Si0 2 for 6 days at different pH
values from 7 to 10 and then measured the rate at which the part