You are on page 1of 16

Hydrometallurgy 96 (2009) 199–214

Contents lists available at ScienceDirect

Hydrometallurgy
j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / h yd r o m e t

Aqueous precipitation and crystallization for the production of particulate solids with
desired properties
G.P. Demopoulos
Department of Mining and Materials Engineering, McGill University, Montreal, Quebec, Canada H3A 2B2

a r t i c l e i n f o a b s t r a c t

Article history: It is the scope of this paper to review and apply the theory of crystallization kinetics to the analysis and
Received 9 October 2008 design of aqueous inorganic hydrometallurgical precipitation systems. In particular the critical role of
Accepted 9 October 2008 supersaturation in controlling the properties (crystallinity–stability, particle size and cleanliness) of the
Available online 28 October 2008
precipitated compounds is emphasized. The approach taken is of generic nature and not system-specific;
however, reference is made to particular precipitation systems such as production of very fine
Keywords:
Aqueous
monodispersed particles, hydrolysis/neutralization of acidic ferric sulphate liquors, iron precipitation from
Processing zinc process solutions, immobilization of arsenic in the form of crystalline scorodite, and gypsum
Precipitation crystallization in waste water treatment. Reference to these systems is not made with the purpose of
Crystallization describing the respective processes but rather to reinforce the validity of the supersaturation-controlled
Hydrometallurgy approach to industrial inorganic precipitation processes.
Supersaturation © 2008 Elsevier B.V. All rights reserved.
Powders
Non-ferrous metals
Inorganic materials

1. Introduction (Dutrizac and Monhemius, 1986; Dutrizac and Harris, 1996; Dutrizac
and Riveros, 2006), to the production of alumina, and from the
Aqueous precipitation is one of the most important and often most production of ultra fine monodispersed powders of metals (Sugimoto,
neglected unit operations in various processes involving aqueous 1987) like Cu, Ag, Pd etc. used in the electronics industry and the
media for the purpose of inorganic materials production (e.g. synthesis of precursor compounds for production of high temperature
hydrometallurgy) or immobilization of toxic inorganic wastes (e.g. superconductors (copper, barium and yttrium oxalates) (Dirksen and
industrial effluent and waste water treatment). Traditionally pre- Ring, 1991) to the treatment of effluents for the immobilization of
cipitation has been viewed (i) from a chemical equilibrium (solubility contaminants like As and P (Krause and Ettel, 1989; Ramalho, 1983). In
product) standpoint and (ii) from an inorganic cation or anion removal this paper reference to a number of specific precipitation systems is
standpoint, i.e. the emphasis was on the solubility (=aqueous species made because on one hand this helps to reinforce crystallization
concentration) rather than on the characteristics and quality of the theory with specific examples and on the other because this reflects
precipitated compound. The latter, however, becomes now increas- current research interests of this author.
ingly more critical as process chemists and engineers are seeking to
prepare/synthesize materials with specific properties (“value adding 2. Crystallization theory of precipitation
processing”) and to design effluent treatment processes that can
immobilize contaminants in compact and stable solids the disposal of According to crystallization theory, precipitation is defined as
which can avoid or minimize adverse environmental effects (“clean reactive crystallization. This definition is preferred as it emphasizes
technology”). This shift of focus from the solubility to the quality of the the formation of the solid product via a chemical reaction instead of
precipitated compound calls for the study and application of the the classical definition which focuses on the metal ion removal aspects
theory of crystallization to aqueous precipitation systems. This is of the process. In the analysis that follows the emphasis is given to the
indeed the scope and content of this paper. factors which control the properties of the solid product, namely
The material covered here is of generic nature and can be applied composition, crystallinity, particle size, morphology, and cleanliness.
to practically any aqueous precipitation system spanning from the The correlation of the precipitation processing conditions to product
removal and rejection of iron from hydrometallurgical solutions properties is determined via the study and control of the following
aspects of the process: (a) solid–liquid equilibria; (b) crystallization
kinetics, i.e. supersaturation, nucleation and growth; (c) colloid-
E-mail address: george.demopoulos@mcgill.ca. surface chemistry, i.e. the aggregation of particles and the adsorption

0304-386X/$ – see front matter © 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.hydromet.2008.10.004
200 G.P. Demopoulos / Hydrometallurgy 96 (2009) 199–214

such as hematite, α-goethite, basic ferric sulphate, jarosite, ferrihydrite


or totally amorphous hydrous ferric oxide (Posnjak and Merwin, 1922;
Voigt and Göbler, 1986; Jambor and Dutrizac,1998; Dzombak and Morel,
1990).
However, it is not only the possible formation of more than one
crystalline phase that complicates matters; there is also the possible
formation of metastable phases (i.e. phases enjoying a kinetic stability
despite their thermodynamic instability). In other words, in the
domain of thermodynamic stability of a compound, a less stable phase
may form due to favourable kinetics. Usually this metastable phase is a
precursor phase, which eventually, given the time, should convert to
Fig. 1. A new paradigm for aqueous precipitation research. the stable one. For example, hydroxycarbonate salts (i.e. M2(OH)2CO3)
tend to form ahead of their carbonate counterparts (i.e. MCO3) within
the domain of stability of the latter, or jarosite tends to form within the
of impurities; and last but not least (d) reactor selection and design. domain of Fe2O3, (see shaded area in Fig. 3) (Zerella et al., 1983) during
Full attention to all these issues is of critical importance in ensuring pressure leaching of sulphide feedstocks.
excellent results. This integrated approach to aqueous precipitation At this point it is worthy to remind ourselves that in general all
research is illustrated in Fig. 1. equilibrium diagrams like the ones mentioned here have to be treated
with caution not only because they neglect the kinetics (after all they
2.1. Precipitation phase equilibria are equilibrium diagrams) but because many of them are based on data
not collected with reliable experimental procedures or simply because
When we consider the precipitation of a compound we have to they were built on the basis of a number of simplifying assumptions.
examine first the phase equilibria of the system, i.e. to define the One such case was the early solubility data and diagrams that were
conditions, temperature, pressure and activities — concentrations published for FeAsO4·2H2O that were based on Fe as opposed to As
which favour the formation of one or the other phase from a solution concentration measurements (Krause and Ettel, 1989). In a
thermodynamic point of view. Precipitation or solubility diagrams review paper (Cheng and Demopoulos, 1997) dealing with the analysis
have to be consulted as part of this examination. As an example the of the hematite precipitation system it has been shown that significant
solubility diagram of goethite (α-FeOOH) at 25 °C is given in Fig. 2 discrepancies occur with a number of precipitation diagrams referring
(Baes and Mesmer, 1986). This diagram by no means, however, can be to the system Fe(III)–SO4(II)–H2O. In connection with the latter it might
interpreted as an indication of the feasibility of producing goethite (i.e. be added here that it is commonly accepted that hematite can be
the crystalline compound α-FeOOH) at 25 °C. This is so because produced only at temperatures above 130 °C (see also Fig. 3). However
precipitation–dissolution reactions are not necessarily reversible, the temperature at which hematite precipitation is favoured is highly
especially when it comes to the degree of crystallinity of the dependent on the concentration of the iron solutions, on the method of
compound and also because in this type of thermodynamic repre- precipitation, e.g. direct hydrolysis vs oxydrolysis or on the matrix of
sentation there are inevitably a number of assumptions and restric- the solution, i.e. sulphate vs. chloride vs. nitrate media. Refer for
tions associated with it. example to Voigt and Göbler (1986), Cheng and Demopoulos (1997)
Let us consider the general case of precipitation of a compound and Dutrizac and Riveros (1999). To explain these differences it is
AB(s): important to resort to the analysis of the crystallization kinetics of iron
precipitation and in particular the effect of supersaturation on the type
AðaqÞ + BðaqÞ + X−ðaqÞ + IðaqÞ + H2 OYABðsÞ + ð?ÞfABX and=or IB; ABOH etcg
and quality of the precipitated compound. Hence, published equili-
ð1Þ brium data and diagrams have to be interpreted with caution and only
− as providing general trends.
where X is the matrix anion and I represents other ionic species.
Despite the fact that a simple thermodynamic analysis might
indicate that the formation of the solid product AB is feasible, side
reactions may be occurring simultaneously leading to production of
additional phases. A good example in this respect is the Fe(III)–SO4(II)–
H2O system where a range of iron (III) — containing phases can form
upon hydrolysis, depending on temperature and solution composition;

Fig. 2. The solubility diagram of α-FeOOH (25 °C). The numbered lines refer to the Fe– Fig. 3. Goethite–jarosite–hematite equilibria as a function of temperature (adapted
OH aqueous complexes (reproduced from Baes and Mesmer (1986)). from Zerella et al. (1983)).
G.P. Demopoulos / Hydrometallurgy 96 (2009) 199–214 201

2.2. Supersaturation ing on the site on which nucleation occurs we have different nucleation
mechanisms (Garside,1985), three of which are the most important ones:
2.2.1. Saturation ratio
(i) homogeneous, i.e. primary production of nuclei in the absence
According to crystallization thermodynamics, a solid phase forms
of a surface
out of a solution when
(ii) heterogeneous, i.e. primary production of nuclei on a foreign
surface and
ΔG = −RTlnðα=α o Þb0 ð2Þ
(iii) surface, i.e. secondary production of nuclei on the surface of a
where ΔG: molar Gibbs free energy; R: the universal gas constnat; T: solid of the same kind with the one which precipitates.
the absolute temerature; α: activity of the solute in the initial
Other secondary nucleation mechanisms involve fragmentation of
solution; and αo: activity of the solute in the final solution (i.e. in
crystals (apparent nucleation) and deposition on reactor walls
equilibrium with the solid phase).
(contact nucleation) — the latter is discussed in Section 2.10.
We define the ratio α/αo as the saturation ratio, S. If we take (a
rather gross simplification) activities equal to concentrations we have:
2.3.1.1. Homogeneous nucleation. The molecules or ion-pairs of the
S = α=α o ≈C=Ceq ð3Þ supersaturated solution combine together to form clusters or embryos
consisting of 10–1000 molecules/cluster. These embryos constantly
where Ceq: equilibrium concentration; when S N 1 we have ΔG b 0. This form and disappear until they assume a critical size (rc) above which
is the condition of supersaturation. We define the supersaturation they are stable, i.e. they do not go back into solution. According to
ratio equal to crystallization thermodynamics (Dirksen and Ring,1991) the free energy
  change associated with the formation of the embryo is given by:
C−Ceq =Ceq = S−1 ð4Þ
ΔGðr Þ = ΔGvol −ΔGsurf ð7Þ
Therefore, supersaturation is the driving force for crystallization. where the first term is the free energy change associated with the
For the case of reactive crystallization, i.e. A(aq) + B(aq) → AB(s), the generation of volume and the second term is associated with the
saturation ratio is defined as; generation of surface. The first term, i.e. ΔGvol is negative but the second
one is positive. So, it is only after a certain size rc that |ΔGvol| N|ΔGsurf| and
S = ½A½B=Ksp ð5Þ ΔG(r) becomes negative. A graphical representation of the ΔG(r)
function is given in Fig. 4 (Mullin, 1993). The ΔGmax represents an
where Ksp = [A]eq[B]eq = exp(−ΔG°/RT).
activation energy barrier.
It must be added here that the solubility of a compound (“Ksp”
The critical nuclei size, rc is a function of the saturation ratio
value) is not only a function of its crystallinity (i.e. amorphous vs
(decreases as the latter increases) and may vary between 40 to 200 Å.
crystalline precipitates) but as well a function of the crystallite size in
In practical terms then a high degree of supersaturation is expected to
the case of nanocrystalline precipitates. This is described by the
yield a large population of ultra-fine particles.
Gibbs–Thomson equation (Sugimoto, 1987):
Another critical feature of the dependency of the nucleation rate
  on saturation ratio is the fact that nucleation starts only when a critical
2γVm
Ceq = C∞ exp ð6Þ S value (Scr,homo) is exceeded (Fig. 5). Beyond this Scr,homo value the
rRT
nucleation rate increases sharply, reaching quickly its maximum rate.
where A commonly used log–log plot of the dependency of the nucleation
rate on saturation ratio is shown in Fig. 5 (Sohnel and Garside, 1992).
Ceq solubility of crystalline particles The form of the nucleation rate equation resembles that of the
C∞ solubility of an infinite size crystal Arrhenius equation (Dirksen and Ring, 1991):
γ specific surface energy (J m− 2)  
Vm molar volume of the crystal (m3 mol− 1) J = 2D=d5 expð−ΔGmax =kB T Þ ð8Þ
r radius of the particle (m) where

Application of this equation to scorodite precipitates revealed that J is expressed in number of nuclei per unit volume (m3) per
the solubility increases at least 2 times when the crystallite size unit time(s),
decreases from 200 to 10 nm (Demopoulos et al, 1994). D is the diffusivity of the solute species (m2 s− 1)
d is the molecular diameter of the solute species (m),
2.2.2. Supersaturation state
The state of the supersaturated solution is not necessarily the same and
as that in a dilute ionic solution. In dilute solutions we have ionic
species (free or complexed) solvated with H2O. As the concentration of kB is the Boltzman constant; 1.38 × 10− 23 J·K− 1.
the solute increases extensive ion-association occurs (usually due to a
decrease in the activity of water) which leads to the formation of The maximum rate is a very large number of the order of ∼ 1030
electrically neutral ion-pairs (always solvated with water). When the nuclei cm− 3 s− 1.
solution becomes supersaturated we have aggregation–polymeriza- It appears, therefore, that as soon as homogeneous nucleation
tion of the solute species, which leads to the formation of clusters of starts, it propagates quickly and in a rather short time we have the
ion-pairs/molecules (10–1000 monomers per cluster). This cluster is formation of an ultra-fine colloidal precipitate.
the precursor of the particle (Flynn, 1984). Scr,homo is an intrinsic kinetic property of each precipitated
compound that may vary from as low as S = ∼ 1.1 to as high as more
2.3. Kinetics of crystallization than 10,000. It has been found that Scr,homo is a function of the size of
the precursor polymeric species; more specifically Scr,homo decreases
2.3.1. Nucleation with increasing nc, where nc = number of monomers that make up the
The formation of a solid phase out of a solution involves the critical nucleus. Conversely Scr,homo decreases as the specific surface
phenomena of nucleation, growth and aggregation of particles. Depend- energy of the precipitating solid increases (refer to Fig. 5).
202 G.P. Demopoulos / Hydrometallurgy 96 (2009) 199–214

2.3.2.3. Modes of growth. Depending on the relative magnitude of the


surface nucleation rate vs. the growth rate, two different modes of
growth arise (Randolph and Larsen, 1971). The first one is the layered or
mono-nuclear growth mode which occurs when the rate-limiting step is
surface nucleation; the second is the continuous or polynuclear growth
mode which occurs when growth is the rate-limiting step. In industrial
precipitation systems it is the latter that is most commonly observed.

2.3.2.4. Growth by aggregation. Growth units, i.e. nuclei of 0.01–0.1 µm


aggregate provided that colloid stability is suppressed.
The stability of a colloid suspension due to double layer interaction
can be described with the aid of the colloid stability factor (Dirksen
and Ring, 1991):

W = ða + bÞa + b I4 exp½V ðrÞ=kB T dr=r 2 ð10Þ

where

V(r) is the interaction potential energy of the two double layers


(J m− 2)
Fig. 4. Crystallization free energy change as a function of nuclei size (reproduced from kB is the Boltzmann constant (=1.38 × 10− 23 J K− 1)
Mullin (1993)). r is the distance between centers of two spheres (m)
a,b are the particle radii (m)

Thus for compounds preceded by spontaneous polymerization, as is


Aggregation is favoured as W decreases, i.e. as V(r) decreases. This is
the case with ferric (oxy)hydroxides (nc N 100), Scr,homo is less than 10 and
equivalent to double layer compression. This is achieved via an increase
this explains the difficulty of obtaining crystalline and “well behaving”
of the ionic strength (I) of the solution. Specific ion adsorption can have
precipitates. On the other hand BaSO4 forms small polymers (nc~10) and
an equally positive effect on colloid aggregation. Aggregation is also
this results in a very high Scr,homo value well above 1000. In other words
favoured as the population density of colloid increases. The phenom-
homogeneous crystallization of BaSO4 occurs without a problem even at
enon of colloid aggregation due to electrical forces is also known as
ambient conditions and this explains the success of the gravimetric
coagulation. High shear forces (agitation) tend to promote the formation
method of sulphate determination that involves accurate precipitation
of compact aggregates. Batch reactors are associated with extensive
and separation of BaSO4 crystals.
aggregation. Large particles are less prone to aggregation than fine
products. Aggregation is the major mode of growth in precipitation
2.3.1.2. Heterogeneous and surface nucleation. Nucleation on an
systems dominated by homogeneous nucleation. Another type of
existing surface is favoured as the surface energy is lower than that of
aggregation is that induced by polymers, otherwise called flocculation.
a new nucleus and ΔGmax is lower too. The result is the critical
Natural and synthetic macromolecules have been used successfully as
saturation ratio to decrease as well. Thus we have the following order:
aggregation agents — surfactants in water and waste treatment
1bScr; surface bScr; hetero bScr; homo (Gregory, 1978). Non-ionic surfactants work best at pHIEP, i.e. pH 8–9
for Fe(OH)3 (IEP is the pH at which the zeta potential is zero). The dosage
The order of critical saturation ratios is indicative of the order of of surfactant is critical. For flocculation only fractional coverage is
the activation energy which controls the nucleation rate. Finally, the required. With complete coverage colloid stabilization occurs instead.
rate equation for surface or heterogeneous nucleation is of the same
form with that of homogeneous nucleation.

2.3.2. Growth

2.3.2.1. Mechanism of growth. The incorporation of a solute species


to a growing crystal goes through a series of stages (Dirksen and Ring,
1991):
(a) Interfacial transport of solute, (b) adsorption on the particle
surface, (c) surface diffusion, (d) integration into the crystal at a kink
site (surface reaction), and (e) desolvation and water diffusion away
from the surface.

2.3.2.2. Growth rate. The growth of a spherical particle of radius r is


characterized by a first order and more rarely second order (in terms
of S) equation (Sugimoto, 1987; Randolph and Larsen, 1971):

G = dr=dt = kg Ceq ðS−1Þ ð9Þ


−1
where G is the crystal growth rate (m s ); kg is the growth constant
(m4 mol− 1 s− 1) and Ceq the solubility of the precipitating solute species
(mol m− 3).
In contrast, the nucleation rate (J) is an exponential or high power
function in terms of S (J ∝ S6–12) (Mullin, 1993; Sohnel and Garside, Fig. 5. Variation of homogeneous nucleation rate with S (log–log plot) for different values
1992). of precipitating solid surface energy ((reproduced from Sohnel and Garside (1992)).
G.P. Demopoulos / Hydrometallurgy 96 (2009) 199–214 203

temperatures around 100 to 120 °C as opposed to gibbsite (γ-Al(OH)3)


that forms at temperatures below 100 °C; or production of hemi-
hydrate of calcium sulphate near the boiling point of water instead of
gypsum (CaSO4·2H2O) which forms at lower temperatures (Li and
Demopoulos, 2006).

2.4.2. Low supersaturation


Lowering of supersaturation has the following beneficial effects:
(i) the formation of large and immobile polymeric species is avoided;
(ii) homogeneous nucleation, which is responsible for the production
of amorphous particles (see Stranski's rule in the next section) and
extensive aggregation leading to solution entrainment is avoided;
(iii) the external mass transport process is decelerated and the solid/
liquid interface becomes less crowded, thus allowing the surface
integration reaction to proceed relatively uninhibited.
Thus maintenance of low S in combination with seed effectively results
in lowering of the crystallization temperature. While the effect of
Fig. 6. Definition of two distinct precipitation regimes with the aid of the rate vs. temperature is well known, that of low supersaturation is not. The author
saturation ratio curves (adapted from Demopoulos (1993)).
and co-workers have convincingly demonstrated the powerful effect
supersaturation control has on the attainment of crystallinity in the case of
In an uncontrolled precipitation system (very common in batch scorodite (FeAsO4·2H2O) precipitation first from chloride media (Demo-
precipitation systems) what usually happens is that upon build-up of the poulos et al., 1995a) and later from sulphate solutions (Droppert et al.,
solute concentration, Scr,homo is exceeded and homogeneous nucleation 1996; Filippou and Demopoulos, 1997; Dabekaussen et al., 2001). Thus
occurs. After some modest molecular growth the nuclei aggregate. With with supersaturation control scorodite was produced at 95 °C as opposed
the progress of the precipitation reaction the concentration drops below to at least 150 °C required without supersaturation control (Dutrizac and
Scr,homo and surface nucleation followed by molecular growth starts Jambor, 1988; Swash and Monhemius, 1994).
which leads to cementing of the aggregates.
2.5. Phase transformation of precipitates
2.3.2.5. Particle morphology. The polynuclear and the aggregation
modes of growth are those that are most common (due to lack of Many precipitates formed are often amorphous (due to either low
supersaturation control) in industrial precipitation systems. As a temperature or lack of supersaturation control or both), and upon
result of this type of growth the produced particles are polycrystalline ageing are converted to crystalline compounds.
with rough surfaces.
2.5.1. Stranski's rule
2.3.3. Precipitation regimes According to Stranski's or Ostwald Step Rule “the least stable phase
A plot of the linear (growth) and exponential (homogeneous nucleates first as long as homogeneous nucleation determines the
nucleation) rate equations against saturation ratio helps to define two rate” (Blesa and Matijevic, 1989). This is so because the interfacial
distinct regimes of precipitation (Fig. 6). Thus when S b Scr, homo energy requirements are less stringent and as such nucleation is
precipitation occurs only on seed (i.e. crystallites of the same kind) or easier. There is no other statement more true in precipitation reactions
foreign solids — this is the regime of growth-dominated precipitation. than Stranski's rule. A good proof of the validity of this rule is Fe(III)
On the other hand when S N Scr, homo homogeneous nucleation hydrolysis reactions which first yield an amorphous hydroxide which
dominates, thus leading to production of ultra-fine colloidal pre- upon ageing is converted to progressively more stable crystalline
cipitates (regime of nucleation-dominated precipitation). In hydro- phases (Blesa and Matijevic, 1989):
metallurgy, where in general production of well grown precipitates is
preferred, it is deduced from the diagram of Fig. 6 that the FeðIIIÞaq YFeOOHiH2 OðamÞYα−FeOOH or β−FeOOHYα−Fe2 O3 ð11Þ
precipitation process will have to be designed in such a way that it
The intermediate phases are simply metastable modifications.
will operate at a supersaturation level just below Scr,homo, together
with recycling of the precipitate to provide for the necessary seed. The
2.5.2. The dissolution–recrystallization mechanism
use of this diagram is demonstrated later in relation to ferrihydrite
As research in colloid chemistry has shown (Blesa and Matijevic,
and scorodite precipitation.
1989) the phase transformation reactions in aqueous media proceed
2.4. Attainment of crystallinity

The factors favouring the direct production of crystalline (as


opposed to amorphous) compounds are high temperature and low
supersaturation. The role of these two factors can be elucidated by
referring to the mechanism of crystal growth.

2.4.1. High temperature


Elevation of temperature has the following beneficial effects:
(i) dehydration is favoured; (ii) neutralization of the charge of the
precursor complex is favoured; (iii) surface diffusion-mobility of the
adsorbed solute is enhanced; and (iv) the crystallization rate (i.e. the
reactive incorporation of solute into crystal) is accelerated.
Different phases can be produced at different temperature ranges
due to easier desolvation, e.g. formation of boehmite (γ-AlOOH) at Fig. 7. Dissolution–recrystallization processes (adapted from Blesa and Matijevic (1989)).
204 G.P. Demopoulos / Hydrometallurgy 96 (2009) 199–214

mainly via the dissolution–recrystallization mechanism. According to


this mechanism (Fig. 7) the least stable phase (A) forms first (as a
consequence of Stranski's Rule) at a rate RA. When the concentration
of Fe(III) (aq) drops below the solubility of A(s), dissolution of the
latter starts at a rate R− A. Under the new low S environment created,
solid phase B forms (at a rate RB) which once more when the
conditions become favourable dissolves back slowly at a rate R− B. And
finally phase C forms out of the new very dilute and seeded aqueous
environment. It must be emphasized here that the time scale of these
phase transformation reactions might be very short, say minutes to
hours, (for example in-situ transformation of metastable phases at
elevated temperatures) or very long, say years (which is the case of
most of the natural ageing processes).

2.5.3. Hydrothermal conversion of ferrihydrite Fig. 9. The LaMer diagram (adapted from Sugimoto (1987)).
Consider as an example the hydrothermal conversion of ferrihydrite
(FeOOH·xH2O(am)) to crystalline goethite (α-FeOOH) or hematite (α- Ccr,homo according to the diagram of Fig. 9) followed by regulation of S
Fe2O3)(Norlund Christensen et al.,1980). The amorphous phase, through (in the region: 1 b S b Scr,homo (or equivalently Ceq b C b Ccr,hono where
dissolution–recrystallization in an aqueous environment converts to α- Ceq is the equilibrium concentration). Condition (2) is met by carrying
FeOOH or α-Fe2O3 depending on the temperature and pH (Fig. 8). The out precipitation away from the IEP and at low ionic strength or by
reaction sequence for goethite formation is shown below: using surfactant-based steric stabilization.
On the other hand production of well grown crystals, which is the case
FeOOHdxH2 OðamÞ + 3H + dissolution NFe3+ + ð2 + xÞH2 O ð12Þ of relevance to industrial hydrometallurgy, is achieved via the use of seed
and the maintenance of low supersaturation, i.e. SbScr,homo (see Fig. 6). It is
the latter condition, i.e. the maintenance of low S that has been often
Fe3+ + 2H2 ORecrystallization Nα−FeOOH + 3H + ð13Þ
neglected in aqueous precipitation process research and development.

The S-shaped kinetic curves of Fig. 8 are typical for nucleation- 2.7. Methods for the control of supersaturation
growth type reactions (Sarraf-Mamoory et al., 1996).
Other examples are (i) the conversion of freshly prepared From the discussion so far it has become clear that supersaturation
amorphous ferric arsenate upon equilibration in the acidic region control is of paramount importance if the quality of the precipitated
(pH b4) to the crystalline equivalent (scorodite) (Nishimura and compound is to be enhanced. The question is however, how super-
Robins, 1996; Le Berre et al., 2008), (ii) the transformation of saturation can be controlled in practice and moreover how Scr,homo can
akageneite (β-FeOOH) to hematite in chloride media (Dutrizac and be detected and avoided if the production of well grown and
Riveros, 1999), and (iii) the hydrothermal conversion (at 220 °C) of crystalline precipitates is sought. Some of the methods available to
jarosite to hematite (Dutrizac, 1990). exercise supersaturation control are described below.

2.6. Particle size control 2.7.1. Via pH Control


For pH-dependent precipitation reactions, as are those of hydro-
For the production of very-fine monodispersed particles, two lysis, supersaturation can be controlled via proper adjustment-
conditions have to be met: (1) separation of homogeneous nucleation regulation of the pH of the solution. One of the methods that has
and growth stages via supersaturation control and (2) control of been practiced by colloid chemists is the use of a slow-releasing base,
aggregation via colloid stabilization. According to the well known like urea. In general slow neutralization is bound to provide
LaMer diagram (Fig. 9) (Sugimoto, 1987) condition (1) is met by improvement in precipitation due to the inherent slow supersatura-
causing an initial short-lived homogeneous nucleation event via the tion associated with it. In this respect, therefore, CaCO3 and CaO
sudden (but controlled) increase of supersaturation above Scr,homo (or (ignoring at this point any positive or negative interferences from the
formation of gypsum) would be expected to be better neutralizing
agents than NaOH. This has indeed been confirmed in the author's
laboratory in the cases, for example, of scorodite (Droppert et al., 1996)
and Ni(OH)2 precipitation (Sist and Demopoulos, 2003).
For full control of supersaturation, though, precise pH control is
needed and this can be accomplished with a step-wise neutralization
procedure as the one developed by the author (Demopoulos, 1993). This
step-wise neutralization procedure (Fig. 10) consists of the following
steps: First the solubility curve of the compound to be precipitated is
established in the form of C versus pH by equilibrating solids of the same
compound with solution at the temperature of interest.
Following that, supersaturation curves are drawn in parallel to the
solubility curve. These supersaturation curves are used to determine
alternate step-wise neutralization paths as shown in Fig. 10. Step-wise
neutralization involves neutralization of an acidic solution of initial
metal concentration Co and pH= pHo up to pH1. At this (or slightly lower)
pH, seed is added and heterogeneous precipitation is initiated during
Fig. 8. Hydrothermal conversion of amorphous ferrihydrite — curve I corresponds to
production of a mixture of α-FeOOH and α-Fe2O3 at 120 °C and pH 10 while curve II
which the pH is maintained constant (or at least within the
corresponds to production of α-Fe2O3 at 150 °C and pH 4.5. (data taken from Norlund heterogeneous precipitation zone) with base addition. When pH stops
Christensen et al. (1980)). changing — an indication that precipitation is more or less complete at
G.P. Demopoulos / Hydrometallurgy 96 (2009) 199–214 205

2.7.3. Via dilution


By dilution the concentration of the precipitating metal is
effectively kept low or equivalently the supersaturation level is kept
low and in this way the complications of homogeneous nucleation are
largely overcome. This approach is built-in in the operation of the
“crystallactor” – a specially designed fluid bed pellet reactor –
developed and used in Europe for the treatment of effluents (Van
Weert and Van Dijk, 1993). This reactor, illustrated in Fig. 11, via
recirculation and mixing of the treated liquor with the incoming
influent, causes dilution of the metal concentration to a level below
Scr,homo thus favouring the deposition of the precipitated compound
on sand or seed particles (pellets).
In the hydrometallurgical industry the goethite process developed
by EZ in Australia (Monhemius, 1980) is an example where in a
Fig. 10. Supersaturation-controlled precipitation strategy. practical way the supersaturation is controlled via dilution. This
approach to goethite production is illustrated in Fig. 12(a). If
precipitation of goethite is attempted from undiluted zinc process
solutions (typical composition 20 g/L Fe3+; T ∼ 95 °C) poorly settling
this pH — further base is added and the pH is raised to the second ferrihydrite (“paragoethite”) or hydronium jarosite is produced
destination: pH2 and the procedure is repeated. By testing various instead. Unpublished work done at McGill shows that there is
supersaturation levels the optimum one is defined for the production of considerable room for optimization of the goethite process involving
well grown (and crystalline if possible) precipitates. This semi-batch direct precipitation from a ferric sulphate solution via dilution. Thus
step-wise precipitation procedure can be implemented on a continuous with a combination of dilution and neutralization, goethite was
industrial circuit operation by using a series of tanks (refer to Section 3.1), produced from a 5.5 g/L Fe3+ sulphate solution at 95 °C in the presence
the number of which and their respective operating (steady-state) pH of seed. Iron concentration was reduced from 5.5 to 1 g/L in 60 min
being the same as the steps and pH values defined in Fig. 10. and the produced goethite contained only 1.2% sulphur as sulphate.
A critical feature during this stage of investigation is the
determination of the critical or otherwise called induction pH 2.7.4. Via a redox reaction
(Droppert et al., 1996; Dabekaussen et al., 2001). This pH is defined In this method, which is suitable for precipitation of a multi-valent
as the pH at which the onset of homogeneous nucleation is observed, metal ion, i.e. Fe2+/Fe3+, Mn2+/Mn4+ etc, controlled oxidation (or
i.e. this is the pH corresponding to Scr,homo. These concepts and reduction) is used to regulate the supersaturation level. The best
methods are further explained in connection to the description of two example of this approach is the goethite process developed by Vieille
iron (III) precipitation systems discussed in Section 3, i.e. precipitation Montagne, now Umicore (Belgium) (Monhemius, 1980). Thus upon
of ferrihydrite and precipitation of scorodite. reduction of Fe3+ by reaction with concentrate (ZnS) the ferrous
The approach outlined here may be adapted and used in non- sulphate solution (∼20 g/L Fe(II)) is oxidized with air at pH 3. Once
hydrolytic precipitation systems where instead of stepwise addition more, similar to the Australian method of goethite production by
of base other precipitating agents may be used. This is the case, for dilution, the Fe3+ concentration is effectively maintained at low level
example, of metal sulphide precipitation with controlled dosage of (∼2 g/L) (see Fig. 12 (b)), thus the complications of homogeneous
S2− (e.g. in the form of H2S). The controlled addition of precipitant as nucleation are largely avoided.
a means of controlling supersaturation has been successfully
demonstrated in the case of high grade nickel sulphate crystal- 2.7.5. Via a dissolution reaction
lization from spent electrolytes with isopropanol (Moldoveanu and In this method the metal to be precipitated is released (in a
Demopoulos, 2002). controlled manner) via the dissolution of a precursor compound. This

2.7.2. Via metal complexation and dissociation


In this method, which is better suited for the production of high
value compounds, the metal to be precipitated is complexed with
strong complexing agents like EDTA, NH3 etc. and then it is dissociated
in a controlled manner in order to allow its controlled release (and
thereby controlled supersaturation environment) and precipitation. A
good example of this approach is Sherritt's method for the production
of fine copper powders (Sutherland et al., 1990). This method consists
of the following steps:

1: Preparation of cuprous ammoniacal solution at around 80 °C:


CuB + CuðNH3 Þ2+ +
2 Y2CuðNH3 Þ2 ð14Þ

2: Controlled dissociation of the Cu(I)-ammine complex by acid


addition:
CuðNH3 Þ+2 + xH+Yð1−x=2ÞCuðNH3 Þ+2 + xNH+4 + x=2Cu+ ð15Þ

3: Spontaneous disproportionation of cuprous ions (simulta-


neously occurring with step 2)

 
x=4 2Cu + YCuB + Cu2 + ð16Þ Fig. 11. The crystallactor (reproduced from Van Weert and Van Dijk (1993)).
206 G.P. Demopoulos / Hydrometallurgy 96 (2009) 199–214

Fig. 12. Goethite precipitation by dilution (a) and by oxidation (b).

method imitates the mechanism of dissolution–recrystallization 2.8.1. Matrix anion effects


discussed in Section 2.5.2. This type of dissolution-controlled (i) The matrix anion determines the nature of the precursor
precipitation process usually happens in hydrometallurgy in situ complex-cluster. As the size, mobility and shape of the latter are
during pressure leaching operations. A good example of this is important in the crystal growth process it is understandable
oxidative pressure leaching of sulphides or the acid pressure leaching that different crystal phases of the same compound may form
of laterites; in connection with the latter Papangelakis and his co- in different complexing media. For example, hydrolytic pre-
workers at the University of Toronto have clearly demonstrated how cipitation of iron(III) oxyhydroxide under ambient pressure but
hydronium alunite (H3O)Al3(SO4)2(OH)6 forms via the dissolution of elevated (80–90 °C) temperatures yields α-FeOOH (goethite) in
gibbsite at 230 to 270 °C (Blakey and Papangelakis, 1994). sulphate/nitrate media but β-FeOOH (akaganeite) in chloride
At McGill some years ago, while studying the pressure oxidation of media.
arsenopyrite, the production of well grown scorodite crystals was (ii) The matrix anion may be incorporated in the solid phase due to
observed (Papangelakis and Demopoulos, 1990). The reactions incomplete exchange of the matrix anion ligand by the
involved were as follows: precipitating ligand.
Example:
Dissolution:
150BC FeSO4+ + 2OH− ðprecipitating ligandÞYFeðOHÞ+2 ð20Þ
FeAsSðsÞ + 13=4O2 + 3=2H2 OY Fe2+ + H3 AsO4 ðaqÞ + SO2−
4 ð17Þ
+ SO2− + 2−
4 ðmatrix anion ligandÞYFeOOHðsÞ + H + SO4
Oxidation: −
The above example shows complete exchange of SO2−
4 by OH .
+
2Fe2+ + 1=2O2 + 2H Y2Fe3+ + H2 O ð18Þ In the case of incomplete exchange, phases like Fe3(SO4)2

Precipitation:

Fe3+ + H3 AsO4ðaqÞ + 2H2 OYFeAsO4 d2H2 OðsÞ + 3H + ð19Þ

The leaching of arsenopyrite took place in a H2SO4 medium.


Attempts to precipitate scorodite from a similar medium but using
ferric sulphate as the source of iron failed to give crystalline solids —
another manifestation of Stanski's Rule.
Another example is the work of Van Weert and Droppert (1994)
who studied the conversion of As2O3 baghouse dust to crystalline
scorodite using a HNO3 medium and an autoclave (130–160 °C). When
iron was introduced as Fe(NO3)3 salt an extremely fine, difficult to
filter, scorodite material was produced; however when scrap iron was
used instead, coarse and easily filterable scorodite was produced. This
was so because iron release was regulated by the dissolution of scrap,
i.e. the supersaturation level was effectively kept low, thus favouring
the growth of crystals.

2.8. Matrix anion and impurity effects

The matrix anion of the solution plus many other minor


constituents of the solution (impurities) may exert significant effects Fig. 13. Composition of ferrihydrite precipitates produced by fast neutralization of ferric
to the quality of the produced compound(s). sulphate solutions (3 g/L Fe(III) at 50 °C (reproduced from Zinck (1993)).
G.P. Demopoulos / Hydrometallurgy 96 (2009) 199–214 207

(OH)5·2H2O may form. The net effect is to obtain a mixture of (ii) Substitution: In this case a foreign constituent is substituted in
precipitates. Since the ligand exchange process is a function of the crystalline lattice of the precipitated compound. Substitu-
pH, it is not surprising to find the content of a precipitate (in tion requires good “fitting” of the substituent group in the host
matrix anion) to depend on the precipitation pH. This is site and favourable bond energy. We may have cation
exemplified with the results of Fig. 13, taken from the work of substitution, such as Ca1 − x Mgx CO3 (Stumm and Morgan,
Zinck (1993). At least in part, the observed correlation of 1981), or anion substitution, such as NaFe3(SO4)2 − y (AsO4)y
sulphate incorporation to pH may be attributed to the existence (OH)6 − x (H2O)x (Dutrizac and Jambor, 1987). Vacancies or
of Fe(III)-sulphato complexes (FeSO+4 and Fe(SO4)−2) in the acidic distortions may be introduced in the crystal due to charge
region (Filippou et al., 1995). imbalance or non-size compatibility between the host site and
As, described later, similar to solution complexes, inner-sphere substituent group, hence altering properties such as solubility.
surface complexes have been observed to form. In this respect, Thus, for example, scorodite produced from sulphate media
therefore, the incorporation of SO2− 4 in the precipitated ferric tends to contain up to 4% SO4. This mode of impurity
oxyhydroxide can be attributed in part to surface complexation. incorporation is the most difficult to avoid. One possible way
The amount though of the matrix anion incorporated into the of avoiding cation substitution is complexing of the trouble-
precipitated compound will depend not only on the pH or the some cation and changing thus its size and charge. One such
anion concentration but on the properties of the compound (i.e. example is the complexation of Cd2+ (a very toxic element) with
amorphous vs. crystalline; particle size etc.) as well and in this Cl− to form CdCl−3 and thus avoid its incorporation in gypsum,
respect on the way precipitation is conducted. Thus, direct CaSO42H2O (Martynovich, 1994).
(without neutralization control) precipitation of ferrihydrite at (iii) The other major mechanism of impurity uptake is via adsorption/
pH 3 and 50 °C was found to yield precipitates with 18% SO4 surface complexation. Due to its wide occurrence this mode of
content (Fig. 13) while controlled precipitation of crystalline a- impurity uptake is discussed separately in the following section.
FeOOH in the presence of seed at pH 2 and 95 °C was found to
yield precipitates with only 3.5% SO4 content (unpublished 2.9. Impurity uptake by adsorption
research). Once more this is the manifestation of the impor-
tance of supersaturation control in precipitation processes. 2.9.1. Surface complexation reactions
(iii) The matrix anion interferes with the dehydration process and Solid surfaces are never smooth and chemically homogeneous.
thereby with the crystallization process. Thus, production of Surface heterogeneity is a fact of life and nothing is more true than for
crystalline phases from chloride-based solutions is in general solids precipitated out of solution. Let us examine in this respect the
easier than from sulphate-based solutions. This is so because surface of an oxide (Dzombak and Morel, 1990). Due to incomplete

the SO2−4 anion is highly aquaphilic as opposed to the Cl which coordination of the surface specific “anion” and “cation” sites develop
is not hydrated (Bates et al., 1970). This effect was observed by on it. The non-fully coordinated cations are hydrated and this is
Demopoulos and co-workers (Demopoulos et al., 1995a; followed by surface hydrolysis which gives rise to a hydroxylated
Droppert et al., 1996) who found crystalline scorodite to be surface: /XOHE. This hydroxylated surface is further hydrated and
produced more readily from chloride media than from the eventually is engaged in reactions with solute species which enter the
corresponding sulphate media. Similarly hematite produc- electrical double layer.
tion from chloride media can be achieved under atmospheric This group of reactions is called surface complexation reactions
pressure conditions (b100 °C)(Dutrizac and Riveros, 1999) as (Dzombak and Morel, 1990; Davis and Hayes, 1986). These reactions
opposed to the case of hematite production from sulphate are considered analogous to the formation of solution complexes. The
media that requires higher temperatures (N130 °C) (Zerella following discussion will focus on surface reactions occurring on
et al., 1983; Cheng, 2002). oxides. Cation or anion surface complexation are the two most
(iv) The matrix anion finally through surface adsorption might alter important types of surface reactions involved in impurity uptake:
the growth mode of the precipitate. For example, precipitation
of Al(OH)3 from acidic sulphate solutions yields spherical Cation complexation
particles but from chloride solutions yields rods/needles
(Sugimoto, 1987). =XOHB + M2+ Y=XOM + + H + ð21Þ

2.8.2. Impurity effects


Anion complexation
Many minor solution constituents tend to specifically chemisorb and
alter the particle morphology and growth mode of the precipitate. It
should be emphasized that it takes trace amounts of impurities (~10− 3M) =XOHB + A3− + H + Y=XA2− + H2 O ð22Þ
to cause significant effects on the crystal formation process (Dirksen and
Ring, 1991). Surface complexation of cations by hydrous oxides invokes the
But in addition to interfering with the crystal growth process, formation of covalent bonds with surface oxygen (or hydroxyl)
impurities tend to become incorporated into the main precipitated anions and the release of protons from the surface. Specific
phase, thus effecting its composition and properties (e.g. solubility– sorption of anions on hydrous oxides occurs via ligand exchange
stability, morphology etc.). Let us now consider in more detail the reactions in which hydroxyl surface groups are replaced by the
main mechanisms of impurity incorporation in precipitates: sorbing ions. In addition to these surface complexation reactions
simple surface ion-association reactions leading to the formation
(i) Coprecipitation as a district phase; for example the coprecipita- of ion-pairs via electrostatic attraction forces may occur. However,
tion of arsenic(V) in the form of scorodite, FeAsO4·2H2O during due to the weakness of the electrostatic forces involved in these
the removal of iron(III) from hot (95 °C) acidic sulphate reactions the retention of impurities is weak, hence washing is
solutions as jarosite, NaFe3(SO4)2(OH)6. This mechanism may effective in removing the latter. This is not the case with the
be suppressed as long as there is a reasonable difference in the inner-sphere-forming surface complexation reactions (reactions
solubility of the two coprecipitated compounds. This was (21) and (22)). The retention of impurities by the latter reactions
demonstrated by Demopoulos et al. (1995a) in the case of is governed by pH. This is clearly manifested in the general
scorodite/akageneite coprecipitation from chloride solutions. adsorption equilibrium curves illustrated in Fig. 14.
208 G.P. Demopoulos / Hydrometallurgy 96 (2009) 199–214

Let us explain why this happens by analyzing the relevant surface


complexation equation

h i  +  h 2+ i
K = M2+ H = M ð1−θÞ ð23Þ
ads aq

where concentrations are taken equal to activities, [M2+]ads is


equivalent to /XOM+, and θ is the fraction of surface that is covered.
Considering now the distribution coefficient D, where D = [M2+]ads/
[M2+]aq, Eq. (23) can be written in the following logarithmic form:

 
logD = logK + logð1−θÞ−log H + ð24Þ

For the iso-distribution point, i.e. 50% cation adsorbed, D = 1 (for


which log D = 0) and the last equation is reduced to

 
log H + 50 = logK + logð1−θÞ ð25Þ

where [H+]50 is the [H+] corresponding to D = 1.


For low metal ion concentration we have low coverage, i.e. (1 − θ)low
is higher than (1 − θ)high, where the latter corresponds to high metal
ion concentration.
By writing the log [H+]50 equation for “low” and “high” concentra-
tions and subtracting we obtain
Fig. 14. Typical equilibrium curves for cation and anion sorption on hydrous iron oxide.
Arrows indicate direction of increasing sorbate/sorbent ratio (reproduced from    
Dzombak and Morel (1990)).
log H + 50;high −log H + 50;low = logð1−θÞhigh =ð1−θÞlow ð26Þ

Since (1 − θ)high/(1 − θ)low b 1 we get, log [H+]50,high b log [H+]50,low or


2.9.2. Impurity uptake pH50,high N pH50,low.
An extensive compilation of cation (Pb2+, Zn2+, Cd2+, Hg2+, Cu2+, Ag+, It is clear, therefore, why with increased metal ion concentration
Ni , Co2+, Cr3+, Ca2+, Sr2+, and Ba2+) and anion (PO3−
2+ 3− 3−
4 , AsO4 , VO4 , the adsorption isotherms move to higher pH. By following exactly the
H2AsO−3, H2BO−3, SO2−
4 , SeO2−
3 , S O
2 3
2−
, and CrO 2−
4 ) adsorption isotherms on same type of analysis it can be shown why the adsorption isotherm
hydrous ferric oxides can be found in Surface Complexation Modelling moves to lower pH when the amount of sorbent, i.e. the amount of
(Chapters 6 and 7) (Dzombak and Morel, 1990). Here only some hydrous ferric oxide is increased.
selected adsorption data are presented. On the other hand, since for each surface reaction there exists a
characteristic adsorption equilibrium constant (reflecting the stability
2.9.2.1. Cation sorption. Some typical cation adsorption data on of the surface complex), it becomes obvious as to why we have the
hydrous ferric oxide are plotted in Fig. 15 (Benjamin and Leckie, 1981). different cation adsorption curves. In general high valence cations
As can be seen the adsorption isotherms move towards higher pH (for adsorb at lower pH than low valence cations. The chemical nature of
the same cation) as the concentration of the cation increases. the hydrous oxide substrate seems to have a rather modest effect on
the adsorption distribution curves. Thus, for example the order of
adsorption with increasing pH for the cations Pb2+, Cu2+, and Cd2+ on
amorphous silica (Stumm and Morgan, 1981) is the same with that on
hydrous ferric oxide (Benjamin and Leckie, 1981).

2.9.2.2. Anion sorption. Typical sorption data of As(V) on hydrous


ferric oxide are plotted in Fig.16. The position of the adsorption isotherm

Fig. 15. Cation adsorption equilibria on hydrous ferric oxide; conditions: 10− 3M FeT, .5 H Fig. 16. Adsorption of As(V) on hyrous ferric oxide; [Fe]/[As] = 15; conditions: 5 × 10− 5M
10− 7M MeT (reproduced from Benjamin and Leckie (1981)). FeT, 3.34 × 10− 6M As (reproduced from Dzombak and Morel (1990)).
G.P. Demopoulos / Hydrometallurgy 96 (2009) 199–214 209

Table 1
Solubilities of Cu(OH)2 in dilute sulphate solutions at 25 °C (initial concentrations 10− 3M
CuSO4; 10− 3M Fe2(SO4)3)

pH Solution copper concentration (mg/L)


CuSO4/alone CuSO4/Fe2(SO4)3
5.98 33.55 (30.18)⁎ 3.47
6.42 4.21 (4.43)⁎ 0.81
6.8
6.87 (0.56) 0.11
7.02 0.35 (0.28)⁎ 0.05
7.26 0.12 (0.09)⁎
7.59 0.10 0.02
8.28 0.05 0.006

* Numbers in () denote solubilities calculated according to Baes and Mesmer (1986).

depends on the molar ratio [sorbent]/[sorbate]. Thus with increasing


Fig. 17. Formation of local supersaturation pocket in a poorly mixed reactor.
anion concentration the anion adsorption isotherm shifts to lower
pH — the opposite of what happens with the cation adsorption
isotherms. On the other hand with increasing concentration of sorbent, from (i) pure adsorption (=surface complexation) to (ii) surface
i.e. hydrous iron oxide substrate, the adsorption isotherm shifts to higher precipitation to (iii) bulk precipitation (=coprecipitation). This is the
pH. case for example of arsenate adsorption on ferrihydrite from relatively
The removal of arsenic by adsorption on hydrous ferric oxide or concentrated solutions and the conversion of the surface complexed
ferrihydrite has been studied by Robins et al. (1988). Their “removal” arsenate to ferric arsenate surface precipitate over time (Jia et al.,
curves are the mirror images of the adsorption curves. At the high pH 2006, 2007).The surface precipitation mechanism is prevalent in
region the adsorption curves move to higher pH with increasing molar hydrolytic precipitation systems involving multi-metal ion solutions.
Fe/As ratio because in reality the amount of sorbent (i.e. Fe(OH)3) A consequence of adsorption occurring in co-precipitation systems
increases and this results in a pH shift. Let us consider the surface is the removal of a cation or anion at a level below the solubility of the
complexation reaction, compound in which it is a constituent. Thus it has been determined in
our laboratory (unpublished research) that the solubility of Cu2+,
=XOHB + HAsO2− + −
4 + H Y=XHAsO4 + H2 O ð27Þ when the latter is precipitated as hydroxide by neutralization, is
significantly lower when it is co-precipitated with Fe3+ (Fe(OH)3) than
   if it is precipitated alone. The relevant data are shown in Table 1.
K = ½Asads = ½Asaq H + ð1−θÞ ð28Þ
2.9.4. Impurity uptake control
D = ½Asads =½Asaq ð29Þ Adsorptive incorporation of impurities during precipitation can be
greatly suppressed via the following measures: a) by reducing the
By similar analysis as done for the cation sorption system we get specific surface area via the growth of large and smooth particles; b) by
enhancing the crystallinity of the precipitate — (a) and (b) are achieved
pH50 = logK + logð1−θÞ ð30Þ by keeping the saturation ratio, S, below Scr,homo; c) by suppressing
aggregation which is responsible for solution uptake; d) by modifying
Thus, as 1 − θ increases, i.e. as the amount of sorbent increases pH50 the pH or pX of the solution and thereby changing the character of the
increases i.e. the adsorption curves shift to higher pH and hence the troublesome impurity species (e.g. Zn2+ vs ZnSOo4 (aq) and the
enhanced stability of arsenic in an extended pH region. character of the surface charge; e) by complexing the troublesome
The stabilizing effect of cations (like Cu2+) on arsenic removal by cation (e.g. Cd2+ vs. CdCl−3); f) by using ionic surfactants which compete
high Fe/As precipitation reported by Harris and Monette (1988) and with impurities for preferential adsorption; g) by employing recycling
Emett and Khoe (1994) is postulated to be due to surface charge
neutralization, i.e. /HAsO−4d CuOH+.

2.9.3. Surface precipitation


The surface complexation reactions that were discussed above may
lead to surface precipitation if the sorbing ions (anions or cations) are
present at high concentrations (Dzombak and Morel, 1990). What
distinguishes surface precipitation from surface complexation is the
fact that in the former case we have the formation of a new surface
phase different from the substrate phase (e.g. formation of Cu(OH)2
layer on a Fe(OH)3 precipitate). In other words with increasing
concentration of sorbing–precipitating ions, we have a continuum

Table 2
Impurity removal by ferric hydroxide precipitation–adsorption (pH 4.5, 75 °C) (data
taken from Dutrizac (1993))

Element (ppm) Impure weak acid leach Purified neutral leach


Ge 1 0.1
As 0.2 0.02
Sb 1 0.1
Se – 0.01
Fig. 18. The mixed suspension classified product removal crystallizer (MSCPR).
210 G.P. Demopoulos / Hydrometallurgy 96 (2009) 199–214

Fig. 19. The three alternative neutralization paths followed to effect hydrolytic
precipitation of Fe(III) hydroxide.

Fig. 21. Particle size distribution of the iron hydroxide solids produced via neutralization
path II (see Fig. 19).

of the precipitate, (for example a reduction of the SO4 content of


amorphous FeOOH from 9% down to 4% was achieved upon recycling
2.10. Design of precipitation reactors
20 times the precipitates (Zinck, 1993) and finally h) by purifying the
solution. The above can prove very useful in industrial precipitation
Mixing of reagents and agitation regime are very important in
applications where a precipitate free of impurities is sought, as it is for
precipitation. It is imperative to ensure uniform mixing in order to
example the case of producing high grade nickel sulphate from de-
avoid local supersaturation pockets (Fig. 17). On the other hand
copperized electrolyte (Moldoveanu and Demopoulos, 2002) or clean
agitation speed has to be optimized as it seems to act in two opposing
gypsum from zinc plant waste waters (Verbaan et al., 1999).
ways. First, as it was mentioned in Section 2.3.2 shear forces help to
On the other hand, when impurity uptake via surface adsorption is
prepare compact aggregates; on the other hand excessive agitation
sought, then production of an amorphous precipitate, like ferric
might lead to scaling problems due to favouring of contact nucleation.
hydroxide (ferrihydrite), is the ideal impurity scavenger-carrier. This is
Contact nucleation involves the plastering of fresh homogeneously
the case of impurity removal during neutral leaching in zinc circuits —
nucleated colloidal particles on the reactor's walls.
refer to data in Table 2 (Dutrizac, 1993). It is worthy to note that all
The rate of contact nucleation (J) is a function of the following
impurities listed are suspected to be in the form of anions hence their
factors (Dirksen and Ring, 1991):
adsorption is favoured in acidic solutions (pH 4–4.5) as predicted by
the adsorption isotherms of Fig. 15. JαkMT ðrpmÞ2−4 S1−2 ð31Þ

where

MT is the mass concentration of solids (kg m− 3)


rpm is the agitation speed (rev min− 1)
S is the saturation ratio
k is the reactor size and geometry factor (dimensionless)

Fig. 20. The effect of recycling on ferric hydroxide densification for (A) simple
neutralization with NaOH; for (B) neutralization with NaOH and supersaturation
control; and for (C) neutralization with lime and supersaturation control (reproduced Fig. 22. SEM photo of the ferric hydroxide produced via staged-neutralization and
from Demopoulos et al. (1995b)). recycling (Neutralization path III in Fig. 19) (reproduced from Demopoulos et al. (1995b)).
G.P. Demopoulos / Hydrometallurgy 96 (2009) 199–214 211

Fig. 23. Particle size distribution of the iron hydroxide solids produced via staged
neutralization (neutralization path III (in Fig. 19)). Fig. 25. The solubility curves of scorodite and akageneite and the two domains of
precipitation (reproduced from Demopoulos et al. (1995a).

It is evident from this equation that high slurry density and excessive
agitation speed all contribute to contact nucleation or in other words to have different mean residence times. The solution flows continuously
scaling/wall plating. On the other hand, since the exponent of S is much and rather fast while the solids either are removed periodically at a much
lower than the corresponding one for homogeneous nucleation slower rate or are recycled. In this way the solids spend much longer
(typically ranging from 6 to 12) the latter is expected to dominate at time inside the reactor thus allowing for their growth and crystallization.
high levels of supersaturation. Operation finally of a series of CSTRs with solids recycling and
Batch reactors are characterized by variable supersaturation envir- distributed addition of the precipitant as described later in Fig. 24
onment. In this respect then batch reactors are not good from the appears to be the best configuration for a hydrometallurgical precipita-
standpoint of supersaturation control. To avoid this inherent deficiency tion process seeking to produce clean and well grown precipitated solids.
of the batch reactor, the precipitant will have to be added in a controlled
manner to ensure constant S (semi-batch configuration). This (for 3. Case studies
example) is the approach taken in developing the step-wise neutraliza-
tion procedure for supersaturation control discussed in Section 2.7.1. 3.1. Precipitation of Fe(III) (oxy)hydroxides
Continuous stirred tank reactors (CSTR), on the other hand, offer a
stable supersaturation and seeded environment due to their steady- Hydrolytic precipitation is one of the most important groups of
state operation. We can distinguish two types of continuous precipita- precipitation reactions. Prior to the formation of the hydroxide solid
tion reactors, the Mixed Suspension Mixed Product Removal Crystallizer the metal ions undergo partial hydrolysis and form soluble hydrolysed
(MSMPR) and the Mixed Suspension Classified Product Removal species. These hydrolysed species under high supersaturation
Crystallizer (MSCPR) (Fig. 18). The first one is nothing else but the undergo polymerisation and eventually condensation via partial loss
familiar CSTR-type reactor. This reactor in general results in fine particles of their solvation water. Due to the ease with which high valent
of wide size distribution (due to its inherent residence time distribution cations hydrolyse and polymerize, the products of their hydrolytic
function) and as such, while it is easy to operate, it fails nevertheless to precipitation reactions are highly amorphous. Meticulous control of S
produce dense and well grown precipitates. However the MSCPR-type at very low values is necessary to overcome this tendency of metal ion
reactor effectively operates in such a way that the liquid and solid phases hydrolytic polymerization. These phenomena are to be illustrated by

Fig. 24. Continuous staged-neutralization process circuit (reproduced from Demopoulos et al. (1995b)).
212 G.P. Demopoulos / Hydrometallurgy 96 (2009) 199–214

Fig. 26. SEM photos of scorodite deposited on gypsum (reproduced from Singhania et al. (2005)).

referring to the most common metal ion hydrolysis system: that of Fe Finally neutralization was effected in a step-wise fashion (neu-
(III). For an extensive review of the hydrolytic chemistry of cations the tralization path III. Fig. 19) as described in Section 2.7.1 (see Fig. 10).
reader is referred to Baes and Mesmer (1986). For the specific system Once more sludge recycling as in the HDS method was employed and
of Fe(III) hydrolysis extensive reviews have been published among the total duration of each test was again 4 h. This crystallization
others by Flynn (1984) and Dutrizac (1979). treatment proved indeed the best approach in yielding high density
Acid rock (mine) drainage solutions contain iron(III) sulphate and ferric hydroxide sludges. The sludge consisted of distinct particles
several other base metal sulphates. Neutralization of the effluents (Fig. 22) with uniform and coarse particle size distribution (Fig. 23)
with lime is the standard treatment method used around the world. (20 µm average particle size) and a 55% solids density after only 8
The problem with this method, as practised today, is that it produces a recycles (Fig. 20 — curve B). Moreover the sludge exhibited 3 to 6 times
large volume of amorphous sludge, which is characterized by poor faster settling rate (12 m/h).
settling properties and low solids content, (b5% without recycling and
10–20% with recycling).
To demonstrate the superiority of the supersaturation-controlled
approach to the treatment of such acidic effluents, an extensive series
of laboratory tests has been performed, details of which are described
elsewhere (Zinck, 1993; Demopoulos et al., 1995b). In this work
synthetic H2SO4/Fe2(SO4)3 containing effluents (pH 2.5; 1000 mg /L
Fe3+, 6000 mg /L SO2− 4 ) were neutralized with NaOH at 50 °C following
three different neutralization procedures. NaOH was used instead of
lime since it was our aim first to understand the crystallization of
“ferric hydroxide” (the actual product was 2-line ferrihydrite (Jambor
and Dutrizac, 1998)) and for that the formation of gypsum had to be
avoided.
The first procedure involved direct conventional neutralization in a
single step to final pH 9. This is depicted in Fig. 19 as neutralization
path I. Solution was agitated at that pH for 4 hours. The produced
sludge (ferrihydrite) was found to be totally amorphous with low
solids content (b5%) and poor settling rate (b2 m/h). This approach to
neutralization and the obtained results correspond to the most
common way neutralization is practised in industry.
The second neutralization approach tested corresponds to the High
Density Sludge Process (HDS), i.e. it involved sludge recycling and a
relatively slow neutralization rate (neutralization path II — Fig. 19) but
not strict supersaturation control. At the end of each test 90% of the
precipitate was recycled as seed for the next test. The duration of each
test (during which pH was gradually increased) was about 4 h.
Recycling was found to significantly improve the sludge quality,
yielding a solids content of 25 to 30% after 15 recycles (Fig. 20 — curve
A). In terms of settling, the product was slightly better than that
produced in the first experiment with typical settling rates varying
between 2 and 4 m/h. Size distribution analysis (Fig. 21) revealed that
precipitation occurred under both homogeneous nucleation-con-
trolled and growth-controlled regimes (see Fig. 6 — Section 2.3.3) Fig. 27. SEM photos of gypsum produced with (top) and without seed (bottom) by controlled
due to lack of precise supersaturation control. neutralization at low S level (reproduced from Omelon (1998) and Verbaan (2000)).
G.P. Demopoulos / Hydrometallurgy 96 (2009) 199–214 213

Having proved that supersaturation control results in the


production of super dense and compact sludges it was decided
next to repeat the step-wise neutralization/recycling experiment
(under the same conditions as before) but now using slaked lime
instead of NaOH. Not only did this series of tests confirm the
effectiveness of the supersaturation-controlled approach but it
provided even more dense sludges with solids content as high as
67% after 10 recycles (Fig. 20 — curve C). The sludge generated
consisted of distinct phases: gypsum (66.5%) and ferrihydrite
(33.5%). No unreacted lime was detected in the sludge which is an
indicator of the inherent high efficiency of lime utilization
associated with this supersaturation-controlled neutralization
procedure.
The implementation of the newly developed staged-neutralization
process on a continuously operating industrial circuit is described in
Fig. 24. The number of tanks and the operating pH in each tank has
been determined for a specific effluent according to the neutralization
procedure described in Fig. 10. Fig. 28. Impurity uptake during gypsum crystallization at pH 4 as a function of seeding
(Conditions 1 g/L cation concentration, except Na was 10 g/L, 50 g/L seed, 50 g/L CaO,
3.2. Crystallization of scorodite 10 g/L, H2SO4 (reproduced from Verbaan (2000)).

The production of scorodite (crystalline ferric arsenate dihydrate)


is advocated as the preferred route of immobilizing arsenic from 3.3. Crystallization of gypsum
arsenic-rich metallurgical effluents and dusts (Demopoulos et al.,
1994; Swash and Monhemius, 1998; Harris, 2000; Riveros et al., Formation of gypsum and precipitation is omnipresent in a variety
2001; Demopoulos, 2005). One method of production is autoclave of hydrometallurgical processes, being more prevalent in neutraliza-
precipitation at temperatures above 150°C (Swash and Monhemius, tion with lime or limestone of sulphuric acid-based process solutions —
1994, 1998). This approach is in agreement with the statement made as done for example at the Iijima Zinc Refinery in Japan (Yamada et al.,
earlier that elevated temperature favours the attainment of crystal- 1998) or for acidic waste waters (Verbaan et al., 1999). In the context of
linity. The other avenue for producing crystalline scorodite, i.e. the the latter case we have studied the crystallization of gypsum with the
maintenance of low S, was explored by the author and co-workers purpose of producing high density and clean gypsum (Omelon, 1998;
first from chloride media (Demopoulos et al., 1995a) and subse- Verbaan, 2000). This research has focused on elaborating a neutraliza-
quently from sulphate media (Droppert et al., 1996; Filippou and tion technique that promotes the production of well grown and clean
Demopoulos, 1997; Dabekaussen et al., 2001; Wang et al., 2000; gypsum material via controlled supersaturation (achieved via the step-
Demopoulos et al., 2003; Singhania et al., 2005, 2006). To this end wise elevation of pH — refer to Section 2.7.1) and product recycling. An
the controlled-supersaturation procedure of Fig. 10 combined with important finding of this work was that, in addition to controlling
the use of seed, described in Section 2.7.1 was employed at supersaturation and recycling, the choice of starting seed material is
temperatures below 100°C. More specifically crystalline scorodite critical. Thus, simple recycling of freshly prepared gypsum material
was produced first at 80 °C to 95 °C from chloride media proved ineffective even when all other controls were applied. It was
(Demopoulos et al., 1995a). Failure to control S b Scr,homo = 8) resulted only with the use of synthetic or naturally occurring gypsum as seed
in production of amorphous ferric arsenate. In other words the material (Omelon, 1998) that gypsum with the desired properties
maintenance of low S effectively reduced the crystallization could be produced. The characteristic differences in gypsum morphol-
temperature from N150 °C to b100°C. ogy obtained by using no seed or synthetic seed can be appreciated by
The crystallization of scorodite from sulphate media at 85–95 °C examining the SEM photos of Fig. 27. The unseeded system produced
proved successful as well when the controlled-supersaturation gypsum exhibiting the so-called “rosette” morphology characterized
procedure was applied, but there were some key differences by high specific surface area and thus prone to high impurity uptake
between the two systems. The critical supersaturation associated and poor settling behaviour. The material produced by simple
with the onset of homogeneous nucleation (i.e. Scr,homo) was found recycling of the freshly (homogeneously) prepared seed was very
to be at least 70, i.e. 10 times higher than that for the chloride fine in size that again gave poor settling behaviour. However, the
system. Equally surprising was the difference in behaviour when material produced with good quality seed (synthetic gypsum) and
scorodite is crystallized from solutions (chloride vs sulphate) with under controlled (low) supersaturation environment proved to be
high Fe(III)/As(V) molar ratio. Thus in the case of the chloride superior to the other ones.
system (Li and Demopoulos, 2006) it was possible by applying The influence of seeding under supersaturation control on impurity
strict supersaturation control to precipitate sequentially scorodite uptake is exemplified with the data of Fig. 28 (Verbaan, 2000). It can be
(at pH b1.5) followed by akageneite (β-FeOOH)(at pH N1.5) (refer to clearly seen that up to one order of magnitude lower impurity uptake
Fig. 25) from Fe/As = 4 solutions. However, the sulphate system occurs when neutralization is conducted in such a way favouring the
does not behave in the same manner. In the latter case the presence production of smooth and large gypsum crystals (Fig. 27a) (refer to
of excess iron, i.e. at Fe/As N 2, it was found to significantly retard Section 2.9.4).
the crystallization of scorodite — this most likely being the
manifestation of the difference in iron complexation prior to 4. Concluding remarks
precipitation (Singhania et al., 2006).
Finally, of interest to underline here, is that under proper By far supersaturation is the most important parameter that controls
supersaturation control it is possible to grow scorodite not only on the properties (particle size, morphology, crystallinity and cleanliness)
scorodite seed but as well on other types of solids like hematite and of a compound produced via aqueous precipitation. It would not be an
gypsum. The deposition of scorodite on gypsum is illustrated with the exaggeration to state that as long as supersaturation is controlled, it is
SEM photo of Fig. 26. almost certain that a product of superior quality will be obtained.
214 G.P. Demopoulos / Hydrometallurgy 96 (2009) 199–214

Development of supersaturation-controlled aqueous precipitation Harris, G.B., 2000. In: Young, C. (Ed.), Minor Elements 2000. SME, Littleton, Co, pp. 3–20.
Harris, G.B., Monette, S., 1988. In: Reddy, R.G., Hendrix, J.L., Queneau, P.B. (Eds.), Arsenic
methods is expected to grow in the near future both in the area of Metallurgy Fundamentals and Applications. TMS, Warrendale, Pa, pp. 469–488.
production of advanced materials such as nanosized powders (Li and Jambor, J.L., Dutrizac, J.E., 1998. Chem. Rev. 98, 2549–2585.
Demopoulos, 2008) but also in the area of industrial hydrometallurgy Jia, Y., Xu, L., Fang, Z., Demopoulos, G.P., 2006. Environ. Sci. Technol. 40, 3248–3253.
Jia, Y., Xu, L., Wang, X., Demopoulos, G.P., 2007. Geoch. Cosmochim. Acta 71, 1643–1654.
and treatment of effluents from chemical and metallurgical opera- Krause, E., Ettel, V., 1989. Hydrometallurgy 22, 311–337.
tions. A solid understanding of crystallization theory is a prerequisite Le Berre, J.F., Gauvin, R., Demopoulos, G.P., 2008. Colloids and Surfaces A Physicochem.
to successful development and implementation of advanced pre- Eng. Aspects 315, 117–129.
Li, Z., Demopoulos, G.P., 2006. Ind. Eng. Chem. Res. 45, 4517–4524.
cipitation technology. Li, Y., Demopoulos, G.P., 2008. Hydrometallurgy. 90, 26-33.
Martynovich, C.T.M.J., “Impurity Uptake in Calcium Sulphate during Phosphoric Acid
Acknowledgement Processing”, Ph.D. Thesis, 1994, Delft University of Technology, Delft, The
Netherlands.
Moldoveanu, G., Demopoulos, G.P., 2002. JOM 54 (1), 49–54.
The author is indebted to all past and current students and Monhemius, A.J., 1980. In: Burkin, A.R. (Ed.), Topics in Non-Ferrous Extractive
research associates of his, who through their hard work and ingenuity Metallurgy. Blackwell Scientific Publications, London, U.K., pp. 104–130.
contributed to the work reviewed in this article. Mullin, J.W., 1993. Crystallization, 3rd Edition. Butterworth Heinemann, Oxford, U.K.,
pp. 172–263.
Nishimura, T., Robins, R.G., 1996. In: Dutrizac, J.E., Harris, G.B. (Eds.), Iron Control and
References Disposal. CIM, Montreal, Canada, pp. 521–534.
Norlund Christensen, A., Convert, P., Lehmann, M.S., 1980. Acta Chem. Scand. A34,
Baes, C.F., Mesmer, R.E., 1986. The Hydrolysis of Cations. R.F Krieger Publishing Co., 771–776.
pp. 224–237. Omelon, S.J., “High Solids Density Gypsum Production Through An Improved
Bates, R.G., Staples, B.R., Robinson, K.J., 1970. Anal. Chem. 42, 867–871. Neutralization Process for Zinc Plant Effluent, M.Eng. Thesis, McGill University,
Benjamin, M.M., Leckie, J.O., 1981. J. Coll. Interf. Sci. 79, 209–221. Montreal, Canada, 1998.
Blakey, B.C., Papangelakis, V.G., 1994. In: Warren, G.W. (Ed.), EPD Congress 1994. TMS, Papangelakis, V.G., Demopoulos, G.P., 1990. Can. Metall, Q. 29, 1–12.
Warrendale, Pa, pp. 3–19. Posnjak, E., Merwin, H.E., 1922. J. Am. Chem. Soc. 44, 1965–1994.
Blesa, M.E., Matijevic, E., 1989. Adv. Coll. Interf. Sci. 29, 173–231. Ramalho, R.S., 1983. Introduction to Wastewater Treatment Processes, 2nd Edition.
Cheng, T.C.M., “Production of Hematite in Acidic Zinc Sulphate Media”, Ph.D. Thesis, Academic Press, San Diego, CA, p. 534.
2002, McGill University, Montreal, Canada. Randolph, A.D., Larsen, M., 1971. Theory of Particulate Processes. Academic Press,
Cheng, T.C.M., Demopoulos, G.P., 1997. In: Mishra, B. (Ed.), EPD Congress 1997. TMS, pp. 41–64.
Warrendale, PA, pp. 599–617. Riveros, P.A., Dutrizac, J.E., Spencer, P., 2001. Can. Metall. Quarterly 40 (4), 395–420.
Dabekaussen, R., Droppert, D., Demopoulos, G.P., 2001. CIM Bulletin 94 (1051), 116–122. Robins, R.G., Huang, J.C.Y., Nishimura, T., Khoe, G.H., 1988. In: Reddy, R.G., Hendrix, J.L.,
Davis, J.A., Hayes, K.F. (Eds.), 1986. Geochemical Processes at Mineral Surfaces. ACS Queneau, P.B. (Eds.), Arsenic Metallurgy: Fundamentals and Applications. TMS,
Symposium Series, 323. ACS, Washington, D.C. Warrendale, Pa, pp. 99–112.
Demopoulos, G.P., 1993. In: Henein, H., Oki, T. (Eds.), Processing Materials for Properties. Sarraf-Mamoory, R., Demopoulos, G.P., Drew, R.A.L., 1996. Metall. Materials Trans. B.
TMS, Warrendale, Pa, pp. 537–540. 27B, 585–594.
Demopoulos, G.P., 2005. In: Reddy, R.G., Ramachandran, V. (Eds.), Arsenic Metallurgy. Singhania, S., Wang, Q., Fillipou, D., Demopoulos, G.P., 2005. Metall. Mater. Trans. B. 36B,
TMS, Warrendale, PA, pp. 25–50. 327–333.
Demopoulos, G.P., Droppert, D.J., Van Weert, G., 1994. In: Harris, B., Krause, E. (Eds.), Singhania, S., Wang, Q., Filippou, D., Demopoulos, G.P., 2006. Metall. Mater. Trans. B 37B,
Impurity Control and Disposal in Hydrometallurgical Processes. CIM, Montreal, QC, 189–197.
pp. 57–70. Sist, C., Demopoulos, G.P., 2003. JOM 55 (8), 42–46.
Demopoulos, G.P., Droppert, D., Van Weert, G., 1995a. Hydrometallurgy 38, 245–262. Sohnel, O., Garside, J., 1992. Precipitation: Basic Principles and Industrial Applications.
Demopoulos, G.P., Zinck, J., Kondos, P.R., 1995b. In: Rao, R., et al. (Ed.), 2nd Int'l Symp. Butterworth Heinemann, Oxford, U.K., pp. 1–193.
Waste Processing and Recycling. CIM, Montreal, pp. 401–411. Stumm, W., Morgan, J.J., 1981. Aquatic Chemistry, 2nd Edition. John Wiley and Sons.
Demopoulos, G.P., Lagno, F., Wang, Q., Singhania, S., 2003. In: Riveros, P.A., et al. (Ed.), Sugimoto, T., 1987. Adv. Coll. Interf. Sci. 128, 65–108.
Copper 2003, vol. IV. CIM, Montreal, pp. 597–616. Sutherland, B.R., Rumsden, J.D., Bolton, G.L., 1990. In: Lakshmanan, V.I. (Ed.), Advanced
Dirksen, J.A., Ring, T.A., 1991. Chem. Eng. Sci. 46, 2389–2427. Materials — Application of Mineral and Metallurgical Processing Principles. SME,
Droppert, D., Demopoulos, G.P., Harris, G.B., 1996. In: Warren, G.W. (Ed.), EPD Congress Littleton, Co, pp. 109–115.
1996. TMS, Warrendale, Pa, pp. 227–239. Swash, P.M., Monhemius, A.J., 1994. Hydrometallurgy 94. Chapman and Hall, London, U.K.,
Dutrizac, J.E., 1979. In: Cigan, J.M., Machey, T.S., O'Keefe, T.J. (Eds.), Lead–Zinc–Tin '80. pp. 177–190.
TMS, Warrendale, Pa, pp. 532–564. Swash, P.M., Monhemius, A.J., 1998. In: Castro, S.H., Vergara, V., Sanchez, M.A. (Eds.),
Dutrizac, J.E., 1990. JOM 42, 36–39. Effluent Treatment in the Mining Industry. University of Concepcion, Concepction,
Dutrizac, J.M., 1993. Aqueous Chemical Processing of Metals, a Workshop organized Chile, pp. 119–162.
by V.G. Papangelakis and held in Kingston, Ontario, June 1993 as part of the 5th Van Weert, G., Van Dijk, J.L., 1993. In: Reddy, R.G., Weizenbach, R.N. (Eds.), The Paul E.
Can. Mat. Sci. Conf. Ch. 5, 43p. Queneau Int'l Symp. Extractive Metallurgy of Copper, Nickel and Cobalt. Funda-
Dutrizac, J.E., Monhemius, A.J. (Eds.), 1986. Iron Control in Hydrometallurgy. Ellis mental Aspects, vol. 1. TMS, Warrendale, Pa, pp. 1133–1144.
Horwood, Chichester, U.K. Van Weert, G., Droppert, D., 1994. JOM 36–40.
Dutrizac, J.E., Jambor, J., 1987. Can. Metal. Q. 26, 91–101. Verbaan, C.M., Omelon, S.J., Demopoulos, G.P., 1999. In: Harris, G.B., Omelon, S.J. (Eds.),
Dutrizac, J.E., Jambor, J.C., 1988. Hydrometallurgy 19, 372–384. Solid/Liquid Separation. CIM Montreal, Canada, pp. 251–266.
Dutrizac, J.E., Harris, G.B. (Eds.), 1996. Iron Control and Disposal. CIM, Montreal, Canada. Verbaan, C.M., “Impurity Uptake During Gypsum Crystallization in Wastewater
Dutrizac, J.E., Riveros, P.A., 1999. Metallurgical and Materials Transactions B 30B, Treatment”, M.Eng. Thesis, 2000, McGill University, Montreal, Canada.
992–1001. Voigt, B., Göbler, A., 1986. Cryst. Res. Technol. 21 (9), 1177–1183.
Dutrizac, J.E., Riveros, P. (Eds.), 2006. Iron Control Technologies. CIM, Montreal, Canada. Wang, Q., Demopoulos, G.P., Harris, G.B., 2000. In: Young, C. (Ed.), Minor Elements 2000.
Dzombak, D.A., Morel, F.M.M., 1990. Surface Complexation Modelling: Hydrous Ferric SME, Littleton, Co, pp. 225–238.
Oxide. J. Wiley and Sons, New York. Yamada, T., Kanumocho, S., Sato, S., Shibachi, Y.,1998. In: Dutrizac, J.E., Gonzales, J.A., Bolton,
Emett, M.T., Khoe, G.H., 1994. In: Warren, G.W. (Ed.), EPD Congress 1994. TMS, G.L., Hancock, P. (Eds.), Zinc and Lead Processing. CIM, Montreal, Canada, pp. 627–638.
Warrendale, Pa, pp. 153–166. Zerella, P.J., Randolph, A.D., Headington, T.A., 1983. In: Osseo-Asare, K., Miller, J.D. (Eds.),
Filippou, D., Demopoulos, G.P., 1997. JOM 49 (12), 52–55. Hydrometallurgy Research, Development and Plant Practice. TMS, Warrendale, Pa,
Filippou, D., Papangelakis, V.G., Demopoulos, G.P., 1995. AIChEJ 41 (1), 171–184. pp. 807–821.
Flynn, C.M., 1984. Chem. Rev. 84, 31–41. Zinck, J., qAn Investigation into the Hydrolytic Precipitation of Fe(III) from Sulphate-
Garside, J., 1985. Chem. Eng. Science 40 (1), 3–26. Bearing Effluentsq, M.Eng. Thesis, 1993, McGill University, Montreal, Canada.
Gregory, J., 1978. In: Ives, K.J. (Ed.), The Scientific Basis of Flocculation. Sigholt and
Noordhoff, The Netherlands, pp. 101–130.

You might also like