Silica in Acid Leaching

Silica in Acid Leaching; David Parris; Parris Consulting Limited, Melbourne, Australia
Parris Consulting Pty Ltd (ABN 87 112 023 917)
6/2 Manningham Street, Parkville, Victoria 3052 Australia
Phone 61 3 9328 24622 Mob 61 4 0000 9377
The original document has been edited to remove references to commercial products
and processes and to make the document more generally applicable and generic.

Introduction
Silicate minerals are ubiquitous in the upper layers of the Earth’s Crust, and there would be few deposits
without significant levels of gangue silicates. Silicate minerals include the pure silica’s such as quartz,
cristobalite and opal, the clays, chlorites and feldspars, and a vast number of less common silicates such as
zircon and the uranium mineral coffinite, and numerous gangue minerals.
During the acid leaching of ores, two types of fine silica particles are released into solution. The first consists
of partly depleted silicates, and the second is colloidal silica, which grows out of solution. Acid leached
silicates, sometimes called ghosts because they retain some of their original shape, are chemically quite
different from colloidal silica. They can be quite small, typically one to twenty microns, because they are
soft and mill readily, and are further reduced by acid leaching. They have a mixture of positive and negative
surface charges and are quite often hydrophobic because the hydrophilic portions have dissolved. They can
make leached slurries difficult to flocculate and settle, and reduce settled densities. The focus of this training
is on colloidal silica that grows out of solution.
This document has a two tier structure. In the overview the occurrence and impact of colloidal silica in acid
leach liquors will be covered completely but briefly, with the second tier intended as a series of “how to”
sections, covering how to recognize issues, how to carry out appropriate tests, and how to apply the results
to the elimination, control or at least amelioration of the issue.

Origins of Colloidal Silica

Many of the more basic silicates will have all of the metals, typically sodium, potassium, iron, magnesium
and aluminum, extracted during leaching, causing the crystal structure to collapse and release silicic acid
into solution. The solubility limit for silicic acid is about 130 ppm. Above this concentration, the silicic
acid polymerizes to form condensed polymers less than 10 nm in size. Around pH 2, these polymers
can remain stable for extended periods of time, and at high concentrations; in one case a solution containing
3.5 g/L silica remained crystal clear and stable for over twelve months. Above pH 2 the polymers grow by
Ostwald ripening: silicic acid depolymerizes from small polymers and condenses back onto large
polymers, which grow into dense, spherical mono-sized colloidal particles between 0.01 and 0.5 microns.
This is the mechanism of opal formation in nature, where the mono-sized particles are laid down in ordered
layers that diffract light and give opals their fire and color.
Colloidal silica is covered in hydroxyl groups and is highly hydrophilic. It is negatively charged above pH
2, its iso-electric point, and positively charged below that point, although it is amphoteric, and always has
both charges on the surface. It is hydrogen bonded into the surrounding water, stabilizing the Ice II structure,
and this water layer is responsible for its stability and resistance to coagulation.
Below pH 2, colloidal silica grows by coalescence. This is a chaotic process, and can be very fast. This growth
results in a wide range of particle sizes, as well as precipitated silica and in special cases, silica gel. The
particles as formed are open and irregular in shape, but over time they undergo syneresis, where they
become rounder and denser by breaking and reforming of bonds. However they never become as round
and dense as those formed at high pH. The surface is still hydrophilic, and covered in hydroxyl groups.
The term “Soluble Silica” is frequently used to refer to the combination of silicic acid, silicic acid polymers,
and colloidal silica in solution. Colloidal silica is readily identified, however, because it is removed by
filtration through a 0.1 micron syringe filter.
The description above is deliberately simplified. Like all real systems, the details can be quite different from
case to case, and so can the consequences.

Consequences
The presence of amorphous colloidal silica particles in a leach liquor can have a range of consequences
from having no noticeable impact at all through to shutting down a process. There are three main areas
where colloidal silica can have an impact; gelling during leach, flocculation and clarification, and in
solvent extraction.

Gelling
If significant amounts of silica are released into solution rapidly during leaching, polymerization can occur
so rapidly that the slurry gels. Even if this only occurs locally in the leach, the impact on slurry viscosity can
be disastrous. Gelled slurries are often confused with over-thickened slurries, but they are quite distinct in
both underlying cause and in behavior.
Thickening events are caused by the volume fraction of solids approaching ΦCRITICAL – the volume fraction
where all of the particles are in contact and viscosity rises rapidly to infinity. Shear has little effect, but
dilution rapidly reduces viscosity of the slurry, and behavior returns to normal. Thickening events are
often caused by a drop in the specific gravity of the ore caused by an increase in clay content. By trying
to maintain mass density in the slurry, the volume fraction rises and approaches ΦCRITICAL.
The presence of clays contributes because plate-like clays pack badly and can cause anomalous thickening
as a result. The role of clays in both events can lead to them being seen as closely related. ΦCRITICAL is also
often i n correctly referred to as the “gel point” for the slurry, which is not helpful in avoiding ambiguity.
Gelling is the formation of a continuous reticulated structure throughout the slurry, in this case silica gel. The
viscosity can be high, but is not reduced rapidly by dilution. Shear will however break down the gel over
time, and can prevent the gel from forming in the first place. Gelling most often occurs in a second or later
leach vessel, where acid is added to drop the pH. If the pH in the first vessel is close to pH2, then the silica
polymers can build up in solution, and then crash out as the pH suddenly drops.

Flocculation and Clarification

Colloidal silica can have a dramatic effect on both flocculant demand, and the clarification of Pregnant Liquors
aka Pregnant Leach Solutions (PLS).The mechanism by which colloidal silica blocks clarification is called the
protective colloid effect. Each of the silica particles is sterically stabilised by two or more layers of hydrogen
bonded water with the Ice II structure. If there is 0.5 g/L of silica in solution, about 0.2 g/L will be colloidal
silica with an average diameter of 0.05 microns, then there will be eight hundred trillion particles per litre.
This will far exceed the number of all other particles put together, even though silica is a minor component
by mass. These particles are separated by less than 1.5 microns from each other. This means that other, larger,
fine particles are supported by the network of colloidal silica particles, and have to “force” their way
through in order to settle, and so are held in suspension.

Put the above together with the fact that many flocculants do not bind to colloidal silica, and you have a
second mechanism: the physical blocking of the flocculant polymer chains by the silica particles that prevents
them finding and bridging gangue particles. While there is not a lot of direct scientific evidence for either of
these effects, they are useful explanations for what is observed in practice, the presence of colloidal silica in
solution can greatly increase flocculant demand and still result in cloudy PLS with high total suspended
solids.

Solvent Extraction
Colloidal silica can lead to emulsification, CRUD formation, and high entrainment losses in SX. On the other
hand it can behave like just another anionic species in solution and pass right through.
Silicon is one of a number of metals and “metalloids” that form amphoteric or weakly acidic oxides. This
group includes bismuth, zirconium, uranium, and many of the Rare Earths and Actinides. This means that
the extractants used for uranium will also bind to the silica surface, making it hydrophobic. If the colloidal
silica is of the right size, typically around 0.05 microns, this will stabilize Pickering emulsions. Again
depending on the size of the silica particles, this can be a gel-like emulsion, or CRUD, a fluid emulsion, or
isolated regions of what I call micro-CRUD, that cause high levels of organic entrainment in raffinate or
spent liquor, and high levels of aqueous entrainment in loaded organic. Any of these can be evidence for the
presence of colloidal silica.

The “How To’s” of Colloidal Silica

Analysis
Colloidal silica is notoriously difficult to analyze. On one nickel laterite mine, the mine analyst showed that
silica levels were low in solution, and did not change through the process. However, analysis of the
voluminous CRUD formed in the nickel and cobalt extraction showed it to be silica based. The reason for
the error was that all samples were carefully filtered through a 0.45 micron filter before analysis.
Modern syringe filters are primarily used in medical applications, and in these applications they must meet
two stringent criteria, they must not pass any of the bacteria or viruses being filtered out, and they must
not be blocked by the cell debris present. This has led to a change in their design over time, from membrane
filters to deep bed filters, to ensure high capacity, and from close tolerances to guaranteed retentions. So a
0.45 micron syringe filter guarantees to remove all particles above 0.45 microns, but does not guarantee to
pass particles below 0.45 microns. In fact the d50 is likely to be well below 0.045 microns. On top of that, if
a filter cake develops, the filter will retain particles that a fresh filter would pass. The consequence is that a
0.45 micron filter will remove most if not all of the colloidal silica present, and the analytical laboratory will
return low, typically about 250 ppm, and consistent silica levels.
The opti mum answer is not filtration, but to immediately dilute samples by about 10:1 with 0.01M
hydrochloric acid to stabilize them. The following must be taken into account:
1. Hydrochloric acid dissolves many fine particles because the chloride ion is a good complexing
agent. Oxalic acid, citric acid, EDTA or DTPA (di-ethylene diamine pentacetic acid) can be
added as additional complexing agent.
2. Silica is stabilized at pH 2, so the diluted samples should be checked for pH when
developing the method and the hydrochloric acid concentration varied to account for acid in the
sample. Any pH between 1.8 and 2.4 is acceptable.
3. The dilution will also protect the silica from polymerizing. An 800 ppm sample will become
an 80 ppm sample, i.e. below the free silicic acid solubility. The dilution ratio should be chosen to
bring the silica levels into the range appropriate for the analytical method.
4. The stabilized samples can be allowed to stand so that fine particles such as quartz, hematite
or sericite can settle out.

If the analytical method is accurate and sensitive enough, then a second, filtered, sample can be analysed
and the colloidal silica estimated by difference. In this case a 0.1 or even 0.05 micron syringe filter can be
used. Acid resistant filters are commercially available.
XRF is relatively insensitive for silica, and liquid cells must be used. The ideal analytical method is
ICP/MS, followed by AA, ICP with internal standard, and XRF last.
Silicomolybdic acid is very accurate and reliable, but is complex and requires considerable practice and
skill, and allowance made for the many possible interferences, and is not recommended for routine work.
Stabilized samples can be stored and sent off site for analysis, but for accuracy and reliability it is best to
always analyze samples fresh. Even stabilized samples can drift over time, and it uncertain whether this is
due to the silica absorbing onto plastic surfaces, reacting with metals present, or some unidentified reason.

The solids from CRUD can be analyzed by conventional methods, as the silica has polymerized within the
CRUD and can be recovered by filtration. Everyone has their favorite methods, but I recommend breaking
the CRUD with isopropanol, followed by filtration through a membrane filter (such as Millipore), careful
washing with water followed by dry isopropanol, and drying. A final acetone wash assists drying of the
sample.
A good light microscope will show the CRUD structure, and the droplets of organic should be clearly visible
at high magnification; CRUD is almost always aqueous continuous. SEM is needed to look at the solids.
Note that CRUD collects suspended solids very efficiently, and large particles contribute to mass. The fact
that hematite, for example, is the major component of a CRUD solid sample does not mean that it is a hematite
CRUD. If 5% of those solids are pure silica, I would strongly suspect that it is a silica CRUD.

Clarification
For Solvent Extraction, SX, a Total Suspended Solids, TSS, of less than 20 ppm is considered the minimum
target, although many plant operate with average TSS figures above 40 ppm. The cost of TSS in the feed to
SX is high. Generally, SX, including all designs of mixer settlers and pulsed columns, is efficient at removing
solids. This means that at a minimum, every cubic meter of PLS with 20 ppm solids generates 40 grams of
sludge, since the sludge also contains organic and aqueous entrainment. This is a direct cost in terms of lost
reagent and the physical removal of sludge from equipment. In addition, the load on CRUD sucking and
filtration, and entrainment losses into the barren aqueous, or raffinate, can also be high. Aqueous entrainment
can have up to 100 parts of organic to every part of fine solids, especially if the solids are colloidal silica.
As mentioned in the introduction, apparently low levels of silica, of which as little as 50 ppm is present
as colloidal silica, can interfere with clarification. Coagulation of the silica with a coagulant can have a
dramatic impact on clarification, dropping TSS from above 150 to below 40 ppm, and halving flocculant
usage. The mechanism is simple; the impact of colloidal silica is dependent on the number of particles per
unit volume in the solution. If coagulation only doubles the diameter of the particles, their number drops by
2-3 or to an eighth of the initial number. In fact as the particles grow, they become easier to coagulate further,
so if the correct dose rate and dosing regime is used, essentially all of the colloidal silica can be removed.
Colloidal silica, even below its iso-electric point of around pH 2, still has negative charges on the surface.
Coagulants bind mainly by hydrogen bonding, but are also weakly positively charged. Coagulation is poor
until electrical neutrality is achieved, and overdosing may cause the silica to re-disperse.

Field Testing for Clarification, removal of silica and associated suspended solids
A complete guide to the field testing methods is available from suppliers of coagulants, and from consultants
experienced in the field of solid/liquid separation, and it is generally advised that these are consulted.

Gelling
With some reactive ores, silicates dissolve quickly enough to reach the concentrations where gelling can
occur. Typically this must be above 3 g/L. Adding a product such as POLYSIL® RM1250 to the leach can
be effective in some cases, because it “catalyses” silica growth to the point where it precipitates
harmlessly. Gelling is caused by the rapid polymerisation of very small colloidal silica particles to form
open networks. It causes a dramatic increase in both plastic viscosity and yield stress that is not reversed
by dilution. Current understanding of the gelling process is given in the Appendix on Scientific background.
Should gelling events occur, it is best to consult with an expert in the area or talk to your supplier.

Solvent Extraction
For colloidal silica to have any impact on Solvent Extraction, its surface must be modified to make it partially
hydrophobic. Amine extractants can bind to the acid sites on the silica surface, and phosphates,
phosphonates and phosphites can displace the surface hydroxyls to form direct ether linkages to the silicon
atoms. Degradation compounds from solvents, extractants, modifiers, flocculants and microbes can also
act as surface active agents and make the colloidal silica surface partially hydrophobic.
This complexity explains why plants can run for some time, even months, before the problems are
encountered.
Colloidal silica can have a number of serious impacts on SX, but these are easy to identify and correct. These
impacts include emulsification, CRUD formation, reduced transfer kinetics and entrainment. The ways to
measure these and confirm the contribution of silica are as follows:
Emulsification is measured by determining the Phase Disengagement Time, PDT. The routine measurement
is carried out in a graduated vessel, typically a glass cylindrical vessel holding one litre of liquid. Generally,
measuring cylinders can be used, but more sophisticated glass vessels with baffles and agitators have been
used to to derive more repeatability and accuracy, as well as generating more data. Use can be made of digital
cameras and/or videos to record the tests and show the data visually. Noting the continuity is also important
as the PDT’s are often quite different for Organic Continuous (OC) and Aqueous Continuous (AC) tests.
Notes
1. Phase continuity is important. It can be judged from experience by looking at the interface,
remembering that in Organic Continuous the aqueous droplets are falling out of the organic phase,
and vice versa. Organic Continuous will have a ragged upper surface and a flat lower surface.
Aqueous Continuous will have a ragged lower surface and a flat upper surface.
2. An extremely useful tool is to measure the conductivity (or resistivity) of the emulsion/mixture prior
to the test – an Aqueous Continuous mix will be conductive, whilst an Organic Continuous mix will
almost totally non-conductive.
3. Under unusual conditions, bi-continuity can occur.
4. Aqueous Continuous PDT’s are usually much longer than Organic Continuous.
5. Silica generally favours Aqueous Continuity, and can cause flipping from OC to AC on the Plant.
This is because if a zone of AC is formed, its lifetime is greater and so can build up at the expense of
OC regions.
6. CRUD is defined in the laboratory test as any layer with a lifetime >> 2 minutes. The volume of
CRUD is best determined using an Imhoff Cone.
7. Almost any hydrophilic substance will lengthen AC PDT’s and shorten OC PDT’s. Fine particles
can significantly increase both.
8. If testing fresh organic liquors with Plant aqueous, several contacts may be required, as the colloidal
silica can build up in the organic interface, and even in the organic in some cases.
9. Using freshly made up organic and synthetic aqueous phases will give the best (shortest) possible
PDT. A good PDT for a one litre sample is between 30 and 60 s.
10. Filtering the Plant Pregnant Liquor through a fine (< 0.1 microns) membrane filter before adding it to
the organic will show whether fine particles are responsible for a long PDT.
11. Pre-treating a small aliquot of PLS with a coagulant and filtering through a standard (Whatman 541)
filtration paper prior to a PDT will give a good indication if colloidal silica is inducing CRUD
formation.
 12. Synthetic aqueous and Plant organic will show whether the degradation products responsible are held
in the organic phase. If so, some form of regeneration should be considered.
13. Plant aqueous and synthetic organic will show whether that something in the PLS is responsible for
the long PDT’s; refer to the note above about the need for multiple contacts with the organic to
“condition” it to the impurity responsible.

Phase Transfer Kinetics

Standard tests are available, and you should contact your reagent provider for their recommendations.
If colloidal silica coats the droplets during SX, it can physically block the transfer of metal to the organic.
This can show up on the Plant and not in the laboratory for reasons explained in the Appendix, but is
due the additional energy put into mixing in the laboratory.
As with PDT measurements, filtering the aqueous or adding coagulant could eliminate the effect of colloidal
silica on transfer kinetics.

Entrainment
Organic entrainment in aqueous can result in a loss of reagent, and the presence of solvent downstream can
have serious consequences, such as attack on rubber lined steel equipment. Aqueous entrainment can carry
over undesirable elements, such as chloride and iron, and often carries over particulates as well. Micro-
CRUD particles containing solids (such as colloidal silica), aqueous and entrained bubbles, sometimes called
fish-eyes, can have the same density as the organic and will not settle out. Similarly, organic and solid
containing micro-CRUD can have neutral density in the barren aqueous liquor.
Organic and aqueous entrainments are best measured directly on fresh Plant samples. To measure aqueous
entrainment, a sample of Loaded Organic is collected, placed in an Imhoff Cone and allowed to settle.
Organic entrainment in barren aqueous is much harder to measure. A recommended method is given here.

Organic Entrainment
This method uses glass one-litre UN bottles. The bottles must be carefully washed to remove any trace of
organic. A useful method is to wash with detergent and alcohol, rinse well with demineralized water , and
finally rinse with acetone and air dry.
The bottles are weighed. Permanent (engraved) labels identifying the bottles are useful and reduce the
chance of error.
Samples are collected from a suitable sampling point, taking all of the precautions necessary when sampling
two-phase streams to ensure that the organic content is neither increased nor decreased over the true value.
The bottles are general ¾ filled in one pass. They must not be rinsed out, nor must they be collected in two
“goes”. Filling accuracy is not important, as weights not volumes will be used.
The bottles are returned to the laboratory, the outside wiped dry with a clean, lint-free cloth and weighed. A
fixed volume of ethylene dichloride with an internal standard is added to each bottle, and the bottles placed
in a shaker or rotary mixer for an hour. The ethylene dichloride is recovered and either injected directly
into a GC, or allowed to evaporate down and then injected.
Appendix: Some of the relevant science

Silicate Minerals
Silicate minerals fall into layer silicates and three dimensional silicates, or non-layer silicates. Layer silicates
include minerals that at first glance do not fit the definition, but are included in layer silicates because of
strong structural similarities. Two examples are asbestos, in which the layers are rolled up like a Swiss Roll,
and Attapulgite, where the layers are so disordered that they behave more like fibres than layers.
The two main classes of layer silicates are clays and chlorites. Chlorites include minerals such as
vermiculite. They are similar to clays in many respects, but have lower silica contents and are much more
soluble in acids. For this reason they are more problematic gangue minerals in acid leaching than the clays.
Clays
Clays have incomplete close packing oxygen layers. Silicon fits into the tetrahedral interstices of two layers,
and M2+ or M3+ into the octahedral interstices in the next layer down. This can be followed by a second
tetrahedral layer, or a layer of hydroxyl groups, and so on.
The view down onto the top of a tetrahedral layer is given in Figure 1.

Figure 1: Top down view of tetrahedral silica layer on a clay surface.

The Octahedral layer is shown in Figure 2. Note that all of the octahedral sites are filled by magnesium
ions, but only two thirds by aluminium ions, which are also smaller. For this reason, magnesium based
clays are called Trioctahedral, three out of three are filled, and those based on aluminium are called
Dioctahedral, two out of three are filled.
Not shown in the figures are the many opportunities for replacement of ions by similar ions. These can be
neutral in their effect, or create a residual charge, that must be balanced by a complementary replacement or
by interstitial ions, usually alkali or alkaline earth metals, or ammonia.
Common replacements are:
Parent Replacement
Ion Ions
Mg2+ Fe2+, Zn2+ In the octahedral layer, neutral
Al3+ Fe3+ In the octahedral layer, neutral
Si4+ Al3+ In the tetrahedral layer, negative charge
Mg2+ Fe3+ In the octahedral layer, positive charge
Al3+ Fe2+ In the octahedral layer, negative charge

Figure 2: View into Octahedral Layer, metal ions between two close packing layers of oxygen ions.
Because of the complexity of the structures, the simpler, useful diagrams shown in Figure 3 are
commonly used.

Figure 3: Simplified representation of layer silicates.

Chlorites
‘Chlorites’ get their name from their green color, due to ferrous ions, not from chloride or chlorine. The
structure of a chlorite is given at the bottom of Figure 3.
The octahedral layers that lie between the 2:1 layers are pure metal oxyhydroxides, and extremely
vulnerable to acid attack.

Acid attack on silicate minerals

Acids do not readily attack acidic silicates, such as quartz. Silicates are attacked by the dissolution of the
metals present in the structure (occlusions and inclusions).
As can be seen from the structure of layer silicates, the removal of the metals in the octahedral and the
tetrahedral layers will effectively destroy the structure. The absence of a three dimensional silicate
structure causes the silica to fall apart and essentially depolymerize to silicic acid.
The rate of acid attack will vary significantly from mineral to mineral, but even within one class of minerals
the rate of attack will vary depending on particle size and crystallinity, and on the reactivity of isomorphic
replacement elements. This can vary even within a single deposit.
Silicic Acid Polymerisation
Silicic acid is a weak acid, and is amphoteric, accepting protons at low pH. Its solubility is low, with some
debate about its true solubility, but in equilibrium with amorphous silica it is about 130 ppm. The solubility
is a strong function of temperature and of pH, as at higher pH values it is ionized.

Silicic acid can readily dimerize with the loss of water to form a siloxane link:
Si(OH)4 ≡ (OH)3.Si.OH
(OH)3.Si.OH + HO.Si.(OH)3 ⇒ (OH)3.Si.O.Si.(OH)3 + H2O

Silicic acid can also ionize up to four times:

Si(OH)4 ⇒ SiO(OH)31― ⇒ SiO2(OH)22― ⇒ SiO3OH3― ⇒ SiO44―

And in strong acid it can protonate:

Si(OH)4 ⇒ Si(OH)3.(OH2)1+

Under all pH conditions, silicic acid polymerises to form complex structures up to 5 nm in size. These contain
three, four, five and six membered rings; different shapes are preferred at different pH values. These
polymers have been isolated and studied as their trimethyl silyl derivatives. Typical shapes are given in
Figure 4.

Figure 4: Typical silica polymers in solution

Generally four and six membered rings are favoured at higher pH, and five membered rings at lower pH.
Formation of Colloidal Silica
At the isoelectric point, pH 1.8 in demineralized water, the polymers form, but do not grow beyond about
5 nm. The isoelectric, or maximum stability point, for silica varies in real systems. It is a function of many
factors, of which the main ones identified are fluoride, which replaces hydroxide groups on the surface
and pushes the IEP down, and metal cations such as aluminum, molybdenum, manganese, rare earths and
actinides, which bind to the surface and push the IEP up. Fluoride also acts as a “hard” version of
hydroxide, and can catalyze silica growth even at the IEP. The point of maximum stability can lie between
pH 1.8 and pH 2.4.

Most colloidal particles have their lowest stability at the IEP, and rely on charge stabilization away from
that pH. Silica is anomalous because it has its highest stability at its IEP.

The current theories for the anomalous stability of silica put it down to steric stabilization by a layer of
hydrogen bonded water held in the ice II structure by the spacing of the surface hydroxyl groups. Close
approach of two particles is prevented as it would disrupt this structure.

Another anomalous behavior of colloidal silica comes from the molecular bonding. In the simplest Si-O-Si
bond, the silicon atoms are 3sp3 and the oxygen 2sp. The sp bonds on the oxygen are σ bonded with
the sp3 bonds of the silicon. If this were all that was happening, we would expect all Si-O-Si bonds to be
linear, and all O-Si-O bonds to have the tetrahedral angle. However the lone pairs of electrons in the oxygen
2p orbitals form π bonds with the empty d-orbitals on the silicon. This double bonding contributes to
the stability of silica, but has far reaching implications for silicate structures. Firstly, as the Si-O-Si
bond bends, the σ bond is weakened, but the π bonds are strengthened. Secondly, the σ bonds favor
a tetrahedral arrangement of silicon atoms while the π bonds favor an octahedral arrangement.

One consequence of this tension is that the bonds in silicates are very flexible, another is that five membered
rings are the most stable arrangement. Five membered rings are not compatible with ordered crystalline
structures unless they are in symmetrical pairs, and although they are common in silicate crystals such as
zeolites and feldspars, the consequence is that colloidal silica is one of the most truly amorphous structures
in nature. The structure has short range order, but essentially no long range order at all.

Because of its amorphous nature, colloidal silica behaves like crystalline colloids in some ways, but quite
differently in others. One of the major differences is in seeding. Seeding a supersaturated solution of a
crystal will cause a mass of crystallites to crash out of solution, especially if the solution is vigorously agitated.
This does not happen with colloidal silica, and the major determinant of the nature and characteristics of
a colloidal silica system is its history.

Particle growth
Colloidal silica grows from the initial polymers, as shown schematically in Figure 5. The mechanism
depends on whether the pH is above or below the IEP.

Colloidal silica above the IEP grows by Ostwald Ripening. Silicic acid and its anions in solution condense
onto the particle surface, and are hydrolyzed away from the surface by hydroxyl ions. Larger particles
grow at the expense of smaller particles because of the reduced curvature of their surface. As the hydroxide
ion is the catalyst, the rate of growth rises steadily with pH, although at pH values above 8 the negative
surface charges reduce the surface energy on the larger particles and resist further growth.
This monomeric dissolution/condensation mechanism gives rise to almost perfectly spherical, mono-sized
particles that are free of voids and are nearly fully dense. Colloidal silica has an estimated density of 2.5
g/mL.

Figure 5: Mechanism of colloidal silica growth

Below the IEP the primary mechanism of growth is coalescence. This mechanism can be very rapid away
from the IEP due to the breakdown of the stabilising water structure. This leads to high surface area particles
irregular in shape and with a wide range of particle surfaces. When initially formed, they have high
internal voidage and low density. Due to this low density their refractive index can match that of leach
liquors so that even quite large particles still give apparently clear solutions.

Over time these particles will rearrange to more spherical shapes and higher density. This occurs by internal
bond breaking and reforming without the silica ever detaching from the particle (what I think of as the
Tarzan effect, where the silica lets go of one bond and swings by the other two to a new position and re-
attaches).

As a rough guide, when colloidal silica grows below its IEP about 70% ends up as precipitated silica and
only 30% as colloidal.

Although coalescence is the main mechanism of growth, if colloidal silica is in contact with slowly leaching
ore, for example in a CCD circuit, or the solution cools slowly, the additional silica does appear to come
out of solution onto the surface of existing particles; in other words the rate of nucleation of new particles
is low compared to the growth of existing particles.

In a process with recirculating liquor, such as a CCD/SX circuit, it is possible for the colloidal silica to grow
slightly at each pass, and so take many passes to reach the size where it causes problems in SX. Combined
with the different shear fields and rates of temperature and composition change between plant and
laboratory, this is why laboratory work is a poor guide to colloidal silica behaviour on a full- scale plant. In
addition, mine plans are designed around recovering consistent levels of the target mineral, and are not
designed to give consistent levels of gangues. Pockets of gangue giving high silica in solution are almost
bound to occur and be hard to predict in advance.
Silica gelling
Gelling due to high silica in solution can be dramatic, with gels of 3% silica being strong enough to hold
the weight of a man. In the presence of a slurry, the resulting mess can be difficult to pump but too soft to
shovel, and can shut a leach circuit down for days while it is cleaned out. Only alkali redissolves the gel
and this is generally not practical and very expensive.

Despite its dramatic difference from the low viscosity colloidal silica suspensions, the underlying
mechanisms of gel formation are no different.

If the pH of a concentrated (>3 g/L) solution of silica is dropped suddenly, the silica polymers are
suddenly destabilised. As they collide by Brownian motion they immediately “stick in place” and form
covalent bridging bonds. This leads to open structures, as the extremities of the clusters are more likely to
collide with others than the internals. If these structures form so that a path can be followed through the
structure from any part to any other part, then the structure is a gel, if not, it becomes colloidal silica, or
precipitated silica where the particles are too big to stay suspended.

The above mechanism shows quite clearly why gelling is such a strong function of silica concentration. The
formation of a gel is a statistical process; the probability of cross-linking must exceed the probability of
termination of a chain. Less obvious is the importance of particle size. The bigger the colloidal silica particles
the shorter distance that they reach for the same mass. Finer particles, which mean younger particles, are
more likely to gel.
There is an excellent description of the gelling process with micrographs and drawings in Ileri.

Measuring particle size

It is possible to obtain particle size distributions for these colloids using modern laser scattering particle
size instruments. Care must be taken not to destabilise the silica, and the machine programs must be allowed
to fit the optimum refractive index to the silica as it can vary widely, and literature values are not a good
guide.

If diluted, solutions must be diluted with water and nitric or hydrochloric acid to give a final pH close to 2,
the stable pH. Solutions must be measured quickly. As the refractive index can be low, and the colloidal
silica only 50 ppm of the silica in the solution, dilution may put the concentration of particles outside the
range of the instrument. Some dilution is recommended to protect the instrument from scale formation such
as jarosite precipitation.

Silicomolybdic acid gives an intense blue colour with silicic acid. This can be used to estimate the size of
colloidal particles since only the surface silica reacts, and the comparison between total silica and surface
silica can be used to infer the size of the silica particles from geometric considerations. However this
method is demanding and subject to a large number of assumptions and interferences. It was at one time the
only reliable method.

Filtration through membranes has been used effectively and with simple precautions is straightforward
and reliable. It is analogous to using screens to determine particle size in milled slurries. The difficulty is
obtaining absolute membranes.

If a set of suitable membranes is available, the solution is filtered through each of the membranes in turn,
starting with the largest. One of the precautions is to ensure that the membrane is never blinded, as a
blinded membrane will remove particles smaller than its nominal size.

When absolute membranes were first available, they were mainly used in research where the accuracy of
the holes in the membrane was critical, and ease of use was not a consideration. Membranes are now used
in large quantities in a wide range of routine biomedical applications. Ease of use is important, and the
absolute removal of particles above a certain size is critical, for example viruses. Allowing smaller particles
to pass is of no interest to these users, and holding them back of little consequence. As a result, most
filters have been redeveloped as deep bed filters, that resist blinding and have large capacities, and even
membrane filters have been redeveloped from ones with precise hole drilled using ion beams to
structured membranes manufactured using surfactant self- assembly techniques. As a result, the retention
of membranes has changed as shown schematically in Figure 6.

Figure 6: Schematic representation of modern vs original syringe filters.

Figure 6 is for syringe filters, but the same is true in principal for many membrane filters as well.
Manufacturers are coy about the shape of the above curves, which is why I have used made up curves,
although the figures for the original membrane filters are roughly correct, with the 1.45 micron
filters passing <1% of 0.45 micron particles, and 99% of 0.225 micron particles. In practice, probably only
a few % of 0.225 micron particles would be passed by a modern syringe filter.

Solvent Extraction
Colloidal silica is covered in hydroxyl groups and is strongly hydrophilic. Thermally produced colloidal silica
such as fumed silica is covered not in hydroxyl groups but in siloxane groups formed by the dehydration
of adjacent hydroxyl groups, and these are strongly hydrophobic. However even these hydrolyze back to
hydroxyl groups on extended contact with water, and become hydrophilic.

However, colloidal silica is made hydrophobic and attaches to organic droplets because it absorbs
hydrophobic molecules. Once this happens, colloidal silica can have dramatic and disastrous effects on
solvent extraction.
The sort of molecules that absorb to the surface of colloidal silica are extractants, and the degradation products
of microbes, solvent, modifiers, flocculants, organic material in the ore, or other process additives. Amine
extractants bind as salts to the silica surface, phosphorus based extractants bind via ether links to the silica,
as shown in Figure 7.

Figure 7: Extractants binding to a colloidal silica surface.

The consequence of hydrophobic molecules binding to the silica surface is to make the particles partially
hydrophobic. This causes them to be attracted to the organic phase and attach to the interface. The way
they attach, and the consequences, are a function of the contact angle, itself a measure of the balance
between hydrophilic and hydrophobic properties. The driving force is the surface energy, or surface tension.

The position of attachment is shown schematically in Figure 8.

Figure 8: Attachment of a spherical particle to an organic droplet

Particles with contact angles ≥90° will coat the surface of larger oil droplets as shown in Figure 9. However,
if the droplet size is smaller, it will require many more particles to coat the surface, as shown on the right
of Figure 9.

Figure 9: Coating of oil droplets with particles

Particles covering an oil droplet as in Figure 9 will act as a physical barrier to the transfer of metal from
the aqueous into the organic phase and vice versa. Increasing the energy and reducing the droplet size will
increase the rate of transfer provided that there are not enough particles to cover the new surface area
generated.This can explain why mass transfer can occur in the laboratory tests, but not on the Plant, as it
is quite normal for a higher mixing energy to be used in the laboratory.

Wetting and coating of particles is only one potential effect of colloidal silica or other fine particles. The
more serious consequence is the stabilization of emulsions. Emulsions stabilised by solid particles are called
Pickering emulsionsii,iii and these can be stable or metastable. Examples of Pickering emulsions are
whipped egg white, mayonnaise and margarine.

In Solvent Extraction stable Pickering emulsions are commonly called CRUDiv. If these are fluid they are
also called (incorrectly) a third phase. An extended Pickering emulsion stabilised by solids is shown
schematically in Figure 10.

Figure 10: Diagram of extended Pickering emulsion and importance of particle size

Figure 10 raises a number of issues that are best addressed in point form.
1 Emulsions are stabilized by fine particles because the particles block the draining of the continuous
phase from between the droplets of the discontinuous phase.
2 Large particles, like particle A in the figure, do not stabilize emulsions. If the film is of the same
dimensions of the particle, it is too thick to be rigid and stable, and if it drains to the right thickness,
the contact angle required is so close to 90° that the particle is not held stably in the film.
3 Small particles, like particle C, stabilize the film against draining, but play no other part in
stabilization.
4 Particles of the correct size, like particle B, with the correct contact angle, strongly stabilize the film
against draining, are held firmly in the film, and resist distortion of the emulsion, making it a
pseudo-solid structure. Typically these particles are between 0.01 and 0.1 micron in diameter. This is
true for spherical particles; non-spherical particles are far more complex.
5 Part of the stabilization is due to boundary pinning, as shown for particles D in the picture of an
extended emulsion. The effect is similar to grain boundary pinning as occurs with steel, where carbon
and carbides prevent the grains from growing or distorting.
6 Colloidal silica formed at low pH has a wide range of particle sizes and so there are likely to be
particles of the right sizes to prevent draining and to pin boundaries.
7 Note that in Pickering emulsions, the continuous phase has the smaller volume, that is, they have high
internal phase ratios. For silica, the aqueous phase appears to be the continuous phase, and so water
drains out of the emulsion, increasing the A to O (Aqueous to Organic) ratio elsewhere, and along
with the effect on PDT’s this is a major cause of flipping from Organic Continuous to Aqueous
Continuous. Difficulty achieving OC operation is an early sign of the presence of colloidal silica in the
liquors.

The above drawings are stylised, and are intended to demonstrate concepts. In Figure 11 are actual
micrographs of oil droplets covered in colloidal silica. Note how the silica clusters around the neck between
the two droplets, and also how in reality the coverage is not uniform (ignore the finest, uniform dots,
they are an artefact of the electron microscope).Once the silica coats the droplets, the particles are brought
into close contact and will tend to polymerize or cement together, making a hard, continuous coating. If
colloidal silica CRUD forms, it will become more rigid and intractable over time, and the enclosed organic
cannot be recovered economically. Prevention is definitely better than cure.

Figure 11: Electron micrographs of oil droplets covered in colloidal silica

Wetters
This theoretical background would be incomplete without a discussion of wetters. In the late 1970’s and
early 1980’s work started on the development of wetters o r w et ti n g agen ts for Solvent Extraction. These
were molecules that would absorb onto the solid surfaces and change the contact angle so that they were
no longer stabilized in the organic/aqueous interface. Many compounds achieve this with the particles, the
challenge is to find molecules that work well but do not interfere with the solvent extraction process
itself by acting as emulsifiers, extractants for undesirable species (such as chloride), reduce the kinetics of
phase transfer, and so on. What occurred historically is that as these molecules were being developed, the
mining industry went into one of its longest and deepest recessions in history, and development stopped.

There are currently no commercial wetters in use in Solvent Extraction, and it is a field that warrants further
research and development, as it may prevent the uptake of amine and keep the silica out of the interface,
reducing entrainment losses and improving extraction kinetics.

***********************************************************************************************************

References

i Ralph K. Iler, “The Chemistry of Silica: Solubility, Polymerization, Colloid and Surface Properties and
Biochemistry of Silica”, John Wiley & Sons, Inc (1979).
ii S. U. Pickering, J. Chem. Soc. 9, 2001 (1907).
iii S. Komura, Y. Hirose and Y. Nonomura, THE JOURNAL OF CHEMICAL PHYSICS 124, 241104 (2006).
iv CRUD is a clever play on words, standing for the old English word crud meaning a messy, dirty, or
sticky substance, and Chalk River Unidentified Deposits from the Canadian nuclear facility in Canada.